article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
a phase ordering system , such as a ferromagnet quenched below the critical point , offers the simplest example of slow relaxation with many of the interesting features observed also in more complex glassy systems@xcite . at the core of this phenomenology
is the separation of the time scales of fast and slow variables . in a domain forming system equilibrium
is rapidly reached in the interior of domains while the interfacial degrees of freedom remain out of equilibrium for a time which diverges with the size of the system . in these conditions
the order parameter autocorrelation function splits into the sum g(t , t_w ) = g_(t - t_w ) + g_(t / t_w ) [ i1 ] where @xmath3 is the stationary time translation invariant ( tti ) contribution due to fluctuations in the bulk of domains and @xmath4 is the aging , or scaling contribution@xcite originating in the off equilibrium fluctuations . in the time scale over
which @xmath3 decays to zero @xmath4 stays practically constant at @xmath5 , where @xmath6 is the equilibrium value of the order parameter , while the decay of @xmath4 takes place for much larger time separations .
using @xmath7 for spin variables , this implies @xmath8 . a structure similar to ( [ i1 ] ) is observed also in the linear response at the time @xmath9 to a random external field switched on at the earlier time @xmath1 ( t , t_w ) = _ ( t - t_w ) + _ ( t , t_w ) [ i2 ] where the stationary contributions @xmath10 and @xmath3 are related by the equilibrium fluctuation dissipation theorem ( fdt ) t_(t - t_w)=g_(0)-g_(t - t_w ) [ fdt ] and @xmath11 is the off equilibrium extra response .
notice that ( [ fdt ] ) may be read as the statement that @xmath10 depends on time through @xmath3 .
mean field theory for glassy systems predicts@xcite that in the asymptotic time region this holds also off equilibrium with @xmath12 . if this is the case and if @xmath13 then static and dynamic properties are connected@xcite by the relation .
-t ) _ g = q = p_(q ) [ i1.2 ] where @xmath14 and @xmath15 are the linear susceptibility and the overlap probability distribution@xcite in the equilibrium state .
applying the scheme to phase ordering we should find .
-t ) _ g = q = ( q - m^2 ) [ i1.2.bis ] since the overlap function in the equilibrium state is given by the delta function@xcite .
on the other hand ( [ fdt ] ) may be rewritten as @xmath16 $ ] and , if @xmath1 is sufficiently large , parameterizing time through the full autocorrelation function we have t_(g)=\ { ll 1-g & + 1-m^2 & .
[ i2.2 ] which gives @xmath17 .
hence , for ( [ i1.2 ] ) to hold one must necessarily have _ t_w ( t , t_w)= _ ( g ) [ i2.2.1 ] and @xmath18 . numerical simulations for the ising model@xcite in @xmath19 and @xmath20 as well as analytical results for the spherical model@xcite indeed show evidence that ( [ i2.2.1 ] ) is asymptotically satisfied .
more precisely , one expects@xcite the off equilibrium contribution to the response function to scale as @xmath21 with @xmath22 on the basis of the argument that @xmath11 is proportional to the interface density @xmath23 where @xmath24 is the typical domain size and @xmath25 for non conserved order parameter dynamics@xcite .
motivated by analytical results@xcite for the one dimensional ising model which , instead , give @xmath26 , recently a detailed study of the behavior of the response function has been undertaken@xcite finding an interesting dependence of the exponent @xmath27 on space dimensionality .
this is best explained in terms of the effective response @xmath28 due to a single interface and defined by _
( t , t_w ) = _ i(t)_(t , t_w ) .
[ i5 ] in this form the behavior of @xmath11 appears to be determined by the balance between the interface loss due to coarsening and the response associated to a single interface .
if one requires @xmath29 clearly @xmath28 must be a constant .
in ref.@xcite it was found that for an ising system this is the case only for @xmath30 , while for @xmath31 there is the power law growth _
( t , t_w ) ~(t - t_w)^ [ i6 ] with numerical values for the exponent at different dimensionalities compatible with @xmath32 . at @xmath20 the power law
is replaced by logarithmic growth .
it has been conjectured that this dimensionality dependence of @xmath33 is the outcome of the competition between the curvature of interfaces and the external perturbing field in the drive of interface motion . according to this picture @xmath20
is the dimensionality above which interface motion is dominated by the curvature , while below @xmath20 the external field competes effectively with the curvature .
the more so the lower is the dimensionality , until in @xmath0 interfaces reduce to pointlike objects driven only by the external field .
when this happens , the rate of growth of the single interface response matches exactly the rate of loss of the interface density , with @xmath34 and @xmath26 .
then , @xmath11 does not vanish as @xmath35 and the second term in ( [ i2 ] ) contributes to @xmath36 producing the violation of ( [ i1.2 ] ) . in short , dimensionality acts as the control parameter which allows to modulate the competition between the two opposing mechanisms driving interface motion and summarized by a=\ { ll & + & .
[ i7 ] in this paper we investigate a generalization of this phenomenon occurring in the framework of the random field ising model ( rfim ) , which further clarifies how the overall behavior of @xmath11 originates in the interplay between the rate of growth of the single interface response function and the rate of loss of defect density . here
we specialize to the @xmath0 rfim . in this
case domain walls perform random walks in a random potential of the sinai type@xcite .
the crucial feature of this walk , from our perspective , is the crossover between the preasymptotic regime , characterized by free diffusion as in the pure system , and the asymptotic regime characterized by sinai diffusion .
measuring @xmath28 and @xmath37 , in the preasymptotic regime we shall find _
( t , t_w ) _
i^-1(t ) [ i8 ] as in the pure system .
namely , as long as diffusion takes place in a flat landscape the rate of growth of @xmath28 does not go below the rate of loss of interfaces . however , going over to the asymptotic regime the landscape becomes rugged and activated processes do play a role .
when this happens @xmath28 slows down with respect to @xmath37 eventually reaching _
( t , t_w ) ~_i^-1/2(t ) [ i8.1 ] which leads to the vanishing of @xmath11 . through this mechanism
the validity of ( [ i1.2 ] ) is restored in the @xmath0 rfim .
in the following we consider the @xmath0 ising model with hamiltonian [ _ i ] = -j _ i=1^n _ i _ i+1 -_i=1^n h_i _ i [ 2.1 ] where @xmath38 is the ferromagnetic coupling and @xmath39 is an uncorrelated random field with expectations \ { ll e_h(h_i ) = 0 + e_h(h_ih_j)= h_0 ^ 2_ij . .
[ 2.2 ] here @xmath40 is not a perturbing field , so @xmath41 does not have to be small .
we will consider the two different dynamical evolutions taking place when the system is quenched to the final temperature @xmath42 starting from the two initial states : _ i _ ) the system is prepared into a spin configuration containing a single interface _
i = ( i ) [ i01 ] _ ii _ ) the system is in equilibrium at infinite temperature with the uniform measure _ 0[_i]= 2 ^ -n .
[ i02 ] in order to have a phase ordering process the equilibrium state at the final temperature of the quench must be ordered . in the @xmath0 rfim
this is not possible even at @xmath43 , since the size of the ordered domains is limited by the imry ma length @xmath44 .
so , we take the limit@xcite of an infinite ferromagnetic coupling @xmath45 in order to have @xmath46 . in this case
the equilibrium state is the mixture of the two ordered states _ [ _ i]= e^-[_i]= _ i ( _ i-1)+_i ( _ i+1 ) [ 2.3 ] for any finite temperature @xmath42 . notice that in the corresponding pure states @xmath47 which implies @xmath48 and @xmath49 .
furthermore , with @xmath50 thermal fluctuations within domains are suppressed and dynamics is reduced to interface diffusion . with the initial condition ( [ i01 ] ) we have the diffusion of the single interface and with the initial condition ( [ i02 ] ) the diffusion - annihilation process of the set of interfaces seeded by the initial condition .
characteristic lengths of the process are in the first case the root mean square displacement ( rmsd ) @xmath51 of the single interface and in the second case the average distance between interfaces ( or average domain size ) @xmath52 .
furthermore , since the typical size of the potential barrier that a walker must overcome after traveling a distance @xmath53 is @xmath54 , there remains defined another characteristic length l_g = ( ) ^2 [ 2.4 ] as the distance over which potential barriers are of the order of magnitude of thermal energy .
this length is important because separates the distances smaller than @xmath55 , over which diffusion takes place as in the pure system with @xmath51 or @xmath56 , from the distances larger than @xmath55 , over which diffusion is dominated by the random potential and it is of the sinai type@xcite with @xmath51 or @xmath57 .
clearly , the limit of the pure system corresponds to @xmath58 .
considering different temperatures @xmath42 in the presence of random fields of different strength @xmath41 , we have let the system evolve with initial conditions ( [ i01 ] ) and ( [ i02 ] ) .
one time unit is defined equal to one flip attempt per spin , on the average ( spins are updated in random order ) .
we use metropolis transition rates , that is the probability to flip the spin is @xmath59 . in the case of initial condition ( [ i01 ] )
, we have found that the rmsd of the single interface depends on @xmath42 and @xmath41 through @xmath55 satisfying ( fig .
[ fig:1 ] ) the scaling relation r(t , l_g)=l_g r ( ) [ 2.7 ] with ( x ) ~\ { ll x^ & + ( x)^2 & .
[ 2.8 ] which displays the crossover from the pure regime to the sinai regime .
a completely analogous behavior is obtained in the second case for the typical domain size @xmath52 , measured as the inverse density of interfaces @xmath60 .
we find ( fig .
[ fig:2 ] ) l(t , l_g)=l_g l ( ) [ 2.8.1 ] where the scaling function @xmath61 obeys the same limiting behaviors as in ( [ 2.8 ] ) . next ,
let us consider the autocorrelation function @xmath62 $ ] where the angular brackets denote the average over thermal noise for a given realization of the field @xmath63 $ ] .
due to the absence of thermal fluctuations within domains @xmath64 and the autocorrelation function is entirely given by the aging component which satisfies ( fig .
[ fig:3 ] ) the scaling relation g_(t , t_w , l_g)= f(,z ) [ 2.9 ] where @xmath65 .
the shape of the scaling function @xmath66 is known in the limits of the pure system @xmath67 and of the sinai regime @xmath68 . in the first case from ( [ 2.8.1 ] )
follows @xmath69 and from the exact solution@xcite of the glauber dynamics for the ising chain f(x , z=0)= ( ) . [ 2.11 ] the second case is obtained taking @xmath1 so large that @xmath70 can be neglected with respect to @xmath71 .
this yields @xmath72 and@xcite f(x , z= ) = - .
[ 2.12 ] in the intermediate cases with finite values of @xmath73 one expects f(x , z ) = \ { ll f(x,0 ) & + f(x , ) & . [ 2.23 ] where the condition on @xmath74 can be rewritten more transparently as @xmath75 or @xmath76 .
while our data in fig .
[ fig:3 ] show clearly that ( [ 2.23 ] ) is obeyed for @xmath77 , the check of the crossover to the sinai regime requires simulation times too long for what we can achieve . in any case , the behavior of the data under variation of @xmath73 in fig .
[ fig:3 ] shows clearly the onset of the crossover . notice that ( [ 2.23 ] ) means that even if the shortest time @xmath1 is chosen inside the sinai regime , namely after the potential has developed a rough landscape , for displacements up to @xmath55 in the bottom of the potential valleys interface diffusion takes place as in the pure system .
finally , let us consider the behavior of the staggered magnetization m(t , l_g)= e_h .
[ 2.10.1 ] essentially , this quantity gives the behavior of ( minus ) the magnetic energy per spin .
at @xmath78 the spin and field configurations are uncorrelated and @xmath79 .
as time evolves one expects the spins to correlate with the field producing growth in @xmath80 . indeed , @xmath80 displays ( fig .
[ fig:4 ] ) growth in the preasymptotic regime @xmath81 , which however is followed by an intermediate regime with a constant plateau for @xmath82 and then by a decrease toward zero for @xmath83 . in the intermediate and asymptotic regimes
one has m(t , l_g)= m ( ) [ 2.10.2 ] with @xmath84 for @xmath85 . in order to understand how this comes about it is useful to look at @xmath80 as made up by separate contributions associated to the single interfaces whose number decreases as time goes on
this is formalized by writing m(t , l_g)=_i(t ) m_(t , l_g ) [ 2.10.3 ] which defines @xmath86 as the effective staggered magnetization associated to a single interface . using ( [ 2.10.2 ] )
then we have m_(t , l_g ) = l_gm _ ( ) [ 2.10.4 ] with ( fig . [ fig:5 ] ) the function @xmath87 obeying _ ( x ) ~\ { ll x & + & .
[ 2.15 ] the interpretation of @xmath86 as the single interface contribution to the build up of magnetization can be substantiated by measuring @xmath80 in the process with the initial condition ( [ i01 ] ) . in this case , defining @xmath88 , we find ( fig .
[ fig:6 ] ) m_(t , l_g)= l_gm _ ( ) [ 2.10.5 ] where @xmath89 displays the behavior ( [ 2.15 ] ) . from this we may draw the following conclusions : _ i _ ) when in the system there is only one interface , the spin - field correlation grows with time .
therefore , this is not a local effect involving only the alignment of the pair of spins at the interface with the local field .
rather , it is a large scale effect involving the optimization of the interface position with respect to the entire field configuration . _
ii _ ) the magnetization growth takes place with different time laws in the preasymptotic and asymptotic regimes .
_ iii _ ) the total magnetization behavior of fig .
[ fig:4 ] , when multiple interfaces are present , then is just due to the fact that in the preasymptotic regime the rate of growth of the single interface magnetization balances the rate of loss of interfaces , while in the asymptotic regime interfaces do disappear faster than the growth of the single interface contribution to the magnetization .
let us now consider what happens if at the time @xmath90 an additional random field @xmath91 uncorrelated with @xmath40 and with expectations \
{ ll e_(_i)=0 + e_(_i _
j)=_0 ^ 2_ij .
[ 3.1 ] is switched on .
we take @xmath92 and we are interested in the linear response function with respect to the @xmath93-perturbation given by t ( t , t_w , l_g ) = e_he _ [ 3.6 ] where the external field @xmath94 acts from @xmath78 and undergoes the change from @xmath94 to @xmath95 at @xmath1 . due to the suppression of thermal fluctuations within domains enforced through @xmath50 , the stationary contribution @xmath10 in ( [ i2 ] )
vanishes identically .
therefore , the above response function is entirely constituted by the aging term @xmath11 .
the interest , then , is focused on the scaling properties . as in the case of the autocorrelation function ,
we find ( fig .
[ fig:7 ] ) that this quantity obeys the scaling form t ( t , t_w , l_g)= ( , z ) .
[ 3.2.1 ] for @xmath96 the pure system response function@xcite is recovered ( x , z=0 ) = [ 3.8 ] which is responsible of the violation of ( [ i1.2 ] ) in the pure ising chain@xcite .
in fact , from the equilibrium state ( [ 2.3 ] ) we ought to have @xmath49 which is consistent with ( [ i1.2 ] ) , as explained in the introduction , if only @xmath97 given by ( [ i2.2 ] ) enters into @xmath36 and the limit @xmath98 is taken after differentiation .
instead , eliminating @xmath74 between ( [ 3.8 ] ) and ( [ 2.11 ] ) one finds ( g , z=0 ) = [ 3.9 ] which spoils ( [ i1.2 ] ) . with @xmath99
there is a crossover . for values of @xmath74 up to @xmath100 fig .
[ fig:7 ] shows that @xmath101 behaves as in the pure case , on the basis of the same argument used for the autocorrelation function .
for larger values of @xmath74 the data show that @xmath101 levels off and then decreases .
this is clearly displayed also in the plot ( fig .
[ fig:8 ] ) against the autocorrelation function .
the mechanism responsible of this behavior is the same discussed in the previous section for the staggered magnetization .
let us look at the effective response of a single interface @xmath28 defined in ( [ i5 ] ) .
we find t_(t , t_w , l_g ) = l(t_w ) _ ( , z ) [ 3.9.1 ] with the scaling function displaying ( fig .
[ fig:9 ] ) the behavior _ ( x , z ) ~\ { ll _ ( x , z=0 ) & + & . [ 3.9.2 ] where @xmath102 .
> from this follows ( [ i8 ] ) in the preasymptotic regime and ( [ i8.1 ] ) in the asymptotic regime , which account for the crossover of the response function in fig .
[ fig:7 ] in terms of the balance between the rate of growth of the single interface response and the rate of loss of interfaces .
hence , for @xmath99 eventually @xmath101 vanishes and in the limit @xmath103 one expects ( x , z= ) 0 . [ 3.9.3 ] therefore , for any finite @xmath41 the validity of ( [ i1.2 ] ) is restored asymptotically .
lastly , it remains to make sure that @xmath104 may be identified with the response function of a single interface .
this we have done by measuring the response function @xmath105 in the process with only one interface in the initial condition and finding ( fig .
[ fig:10 ] ) a behavior quite close to the one of fig .
[ fig:9 ] .
one of the hypothesis for the validity of ( [ i1.2 ] ) relating static and dynamic properties is that the large time limit of @xmath106 reaches the equilibrium value @xmath14 .
it is then clear from ( [ i2 ] ) that for this to be true in the phase ordering process @xmath11 must vanish , since this is an intrinsically out of equilibrium contribution . in other words ,
the existence of the interfacial degrees of freedom which do not equilibrate must not play a role at the level of the response function .
indeed , @xmath11 normally does to vanish .
however , there are exceptions in special cases and the study of these is quite instructive since it allows to gain insight into the properties of @xmath11 .
one dimensional systems with a scalar order parameter and frozen bulk thermal fluctuations do make such a special case . in the @xmath0 pure ising model , due to the pointlike structure of domain walls ,
the minimization of magnetic energy introduces a bias in the diffusion of interfaces which leads to a nonvanishing @xmath11 through ( [ i8 ] ) .
elsewhere@xcite we have analyzed how ( [ i8 ] ) ceases to hold in going from @xmath0 to @xmath107 , due to the extended nature of interfaces . in that case
the minimization of magnetic energy is hindered by the competing need to minimize the curvature of interfaces and @xmath11 asymptotically disappears . in this paper
we have investigated a different mechanism altering the delicate balance ( [ i8 ] ) between the gain and loss of contributions to the response function , which does not require the passage from pointlike to extended defects . in the @xmath0 rfim
it is the gradual roughening of the landscape that slows down the minimization of the magnetic energy with respect to the growth law of the domain size and which eventually yields an asymptotically vanishing @xmath11 .
therefore , the overall picture which comes out is that a coarsening system is always out of equilibrium in the sense that there are always interfaces around , each one producing the out of equilibrium contribution @xmath28 to the response function , independently from dimensionality or the presence of quenched disorder .
these are elements which affect the finer details , like whether @xmath28 is a constant or how fast it grows .
where these properties come into play is in putting together the contributions of all the separate interfaces through @xmath108 which , now depending on dimensionality or quenched disorder , may then produce either a vanishing or an asymptotically persistent net result .
finally , a comment should be made about the non commutativity of the order of the limits @xmath109 and @xmath35 . if the limit @xmath109 is taken first , the linear response function of the pure ising model is obtained and , as discussed above , ( [ i1.2 ] ) is violated . instead , if @xmath35 is taken first one finds the asymptotic linear response function ( [ 3.9.3 ] ) of the rfim which is consistent with ( [ i1.2 ] ) .
however , if the limit @xmath109 is taken next , the response function sticks to @xmath110 , which is not the linear response function of the pure model . | a thorough numerical investigation of the slow dynamics in the @xmath0 random field ising model in the limit of an infinite ferromagnetic coupling is presented .
crossovers from the preasymptotic pure regime to the asymptotic sinai regime are investigated for the average domain size , the autocorrelation function and staggered magnetization . by switching on an additional small random field at the time
@xmath1 the linear off equilibrium response function is obtained , which displays as well the crossover from the nontrivial behavior of the @xmath0 pure ising model to the asymptotic behavior where it vanishes identically .
@xmath2 [email protected] [email protected]
[email protected]
pacs : 05.70.ln , 64.60.cn , 05.40.-a , 05.50.+q |
a _ geometric graph _ ( gg ) is a graph whose vertices are points in general position in the plane , and whose edges are segments connecting pairs of vertices .
if the vertices are in convex position ( i.e. , no vertex lies in the convex hull of the remaining vertices ) , the graph is called a _
convex geometric graph _ ( cgg ) .
one of the oldest turn - type questions for geometric graphs , mentioned already by erds @xcite in 1946 , is : what is the maximal number of edges in a geometric graph on @xmath0 vertices that does not contain @xmath1 _ pairwise disjoint
this question , along with the `` dual '' question concerning geometric graphs that do not contain @xmath1 _ pairwise crossing _
edges , became central research topics in geometric combinatorics , with dozens of papers obtaining upper and lower bounds and resolving special cases ( see , e.g. , @xcite ) .
for the sake of brevity , we call a gg that does not contain @xmath1 pairwise disjoint edges _ an @xmath12-free graph _ , and a gg that does not contain @xmath1 pairwise crossing edges
_ an @xmath13-free graph_. for general ggs , the questions of determining the maximal possible number of edges in an @xmath12-free graph and in an @xmath13-free graph are still widely open . for @xmath12-free
ggs , the best currently known upper bound is @xmath14 , obtained by felsner @xcite , and the best lower bound is @xmath15 , obtained by tth and valtr @xcite . for @xmath13-free
ggs , the best known upper bound is @xmath16 , obtained by fox , pach , and suk @xcite . for both questions ,
it is conjectured that the correct answer is of order @xmath17 ( see @xcite ) .
much more is known in the convex case .
for @xmath13-free cggs , capoyleas and pach @xcite showed in 1992 that the maximal possible number of edges is @xmath18 , for any @xmath19 .
it was also shown that any @xmath13-free cgg that is maximal with respect to inclusion ( i.e. , addition of any edge gives rise to @xmath1 pairwise crossing edges ) contains exactly @xmath18 edges @xcite , and the exact number of extremal examples was shown to be equal to the determinant of a certain matrix of catalan numbers @xcite .
strong upper bounds were obtained also for related questions , such as determining the maximal possible number of edges in a cggs that does not contain a @xmath6-grid ( i.e. , two families of @xmath6 edges each such that each edge in the first family crosses each edge in the other family ) , see @xcite . for @xmath12-free cggs ,
kupitz @xcite showed in 1982 that the maximal possible number of edges is @xmath2 ( for @xmath19 ) or @xmath20 ( for @xmath21 ) . in the cases
@xmath4 ( for all @xmath0 ) and @xmath5 ( for even @xmath0 ) , the exact characterization of all _ maximal _ @xmath12-free cggs ( i.e. , all @xmath12-free cggs with exactly @xmath2 edges ) is known . for @xmath4 , woodall @xcite showed that the maximal cggs are all self - intersecting circuits of odd order with inward pointing leaves .
this characterization holds also for general ( i.e. , not necessarily convex ) geometric graphs , and is related to conway s _ thrackle conjecture _
( see @xcite , p. 306 ) . for @xmath5 ,
a characterization was given recently in @xcite .
the extremal graphs turn out to be complements of trees of a special structure called _ caterpillar trees _ ( see @xcite ) . as for characterization of the maximal @xmath12-free cggs for @xmath22 ,
the knowledge is very scarce . unlike the case of
maximal @xmath13-free cggs , the number of maximal @xmath12-free cggs is not known , and only a few examples of such cggs were given ( see , e.g. , @xcite ) .
moreover , all known examples are of a very specific kind , containing roughly the @xmath2 edges of maximal order in @xmath7 , the complete cgg on @xmath0 vertices ( see section [ sec : notations ] for the formal definition of `` order '' ) . if @xmath7 is represented by a regular polygon @xmath23 of order @xmath0 with all its diagonals , such graphs are almost fully symmetric about the center of the polygon , and they do not contain an independent set of more than @xmath24 consecutive vertices on the boundary of @xmath23 . in this paper , we seek to broaden our acquaintance with the family of maximal @xmath12-free cggs . to this end
, we consider examples that are as far as possible from being centrally symmetric .
formally , we ask how large can be @xmath8 , such that there exists a maximal @xmath12-free graph that contains an independent set of @xmath8 consecutive vertices on the boundary of @xmath23 .
such a restriction forces the graph to be far from symmetric , as demonstrated in figure [ fig:1new ] .
we show by an explicit class of examples that @xmath8 can be as large as @xmath25 ( i.e. , twice as large as the value of @xmath8 for the previously known examples ) , and prove that this value is maximal .
we obtain this result as a special case of a more general result concerning @xmath12-free graphs with a large independent set of consecutive boundary vertices .
let @xmath26 be a cgg .
a set @xmath27 is a _ free boundary arc of order @xmath8 _ if : * @xmath28 consists of @xmath8 consecutive vertices on the boundary of @xmath29 , * @xmath28 is independent , i.e. , no edge of @xmath26 connects two vertices of @xmath28 .
we often say that @xmath26 _ avoids _
@xmath28 , when @xmath28 is an independent set of vertices of @xmath26 .
the general question we address can be formulated as follows : what is the maximal number of edges in an @xmath12-free cgg on @xmath0 vertices that admits a free boundary arc of order @xmath8 ?
we provide a complete answer in the following theorem : [ thm : main ] let @xmath30 and @xmath8 be integers such that @xmath31 and @xmath32
. denote by @xmath10 the maximal number of edges in an @xmath12-free cgg on @xmath0 vertices that admits a free boundary arc of order @xmath8 .
then : 1 . if @xmath9 , then @xmath33 .
in particular , there exists a _
@xmath12-free cgg that admits a free boundary arc of order @xmath8 .
2 . if @xmath34 for @xmath35 , then @xmath36 .
3 . if @xmath37 , then @xmath38 .
the proof that the stated values are upper bounds for @xmath10 is quite straightforward .
the proof of the equalities is somewhat more complex , and involves a construction of an asymmetric maximal @xmath12-free cgg . to illustrate the construction ,
two examples of cggs with @xmath2 edges that admit a free boundary arc of order @xmath25 are given in figure [ fig:1new ] .
this paper is organized as follows : in section [ sec : notations ] we introduce a few definitions and notations , and prove the upper bound of theorem [ thm : main ] . in section [ sec : basic - construction ]
we present a family of maximal @xmath12-free graphs @xmath39 that avoid @xmath40 consecutive vertices , thus proving theorem [ thm : main](1 ) . in section [ sec : extension ]
we modify the construction to produce @xmath12-free graphs @xmath41 that avoid @xmath42 consecutive vertices , and use this construction to complete the proof of theorem [ thm : main ] .
we start with a few definitions and notations , which are a bit unusual but will be very convenient for the proofs in the sequel .
let @xmath0 be a positive integer , @xmath43 .
in order to deal with a convex geometric graph @xmath44 of order @xmath0 with a free arc @xmath28 , we consider a regular polygon @xmath23 of order @xmath45 , whose vertices lie on a circle with center @xmath46 and are numbered cyclically from @xmath47 to @xmath0 , as shown in figure [ fig:2new ]
. we shall always insist that the free arc @xmath28 * is * symmetric with respect to the horizontal axis through @xmath46 and lie on the right .
thus , if @xmath48 is even , then we let @xmath49 be the set of odd - labelled vertices of @xmath23 , and if @xmath48 is odd , we let @xmath49 be the set of even - labelled vertices of @xmath23 .
the segment connecting the endpoints of @xmath28 will always be vertical .
given the set @xmath28 avoided by @xmath26 , we say that an edge @xmath50 is _ allowed _ if at least one of its endpoints lies in @xmath51 ( i.e. , @xmath52 ) .
otherwise , @xmath50 is called _
forbidden_. we also call the vertices of @xmath28 _ avoided _ or _ forbidden _ vertices , and the vertices of @xmath51 _ allowed _ vertices .
we define the _ order _ of an edge @xmath53 \in e(g)$ ] to be @xmath54)= \frac{1}{2 } \min(|i - j| , 2n - |i - j|)$ ] .
note that each edge @xmath55 divides the boundary of @xmath23 into two open arcs .
if @xmath56 , then these two arcs contain @xmath57 and @xmath58 vertices of @xmath26 , respectively .
we say that an edge @xmath59 \in e(g)$ ] _ emanates from _ its left vertex , according to the labelling in figure [ fig:2new ] .
formally , @xmath50 emanates from @xmath60 if @xmath61 . in the case
@xmath62 , we say that the edge @xmath50 emanates from the positive - labelled vertex . for each vertex @xmath63 of @xmath23
, we consider the set of edges on @xmath49 that are perpendicular to the radius @xmath64 .
the edges of this set are parallel
. if @xmath0 is odd , then their number is @xmath65 .
if @xmath0 is even , then their number is @xmath66 for @xmath67 and @xmath68 for @xmath69 .
we say that these are the edges _ in direction @xmath63 .
_ note that in this notation , the vertices labelled @xmath60 and @xmath70 ( for any @xmath71 ) correspond to the same direction .
see the right part of figure [ fig:2new ] for an illustration of the directions in a cgg on @xmath72 vertices .
a pair of parallel edges on @xmath49 is called _ consecutive _ if each vertex of one edge is adjacent on the boundary of @xmath49 to a vertex of the other edge . throughout the paper ,
we make use of sequences of consecutive parallel edges ( i.e. , sequences of the form @xmath73 , where @xmath74 are consecutive for each @xmath75 ) . of course , all edges in such a sequence belong to the same direction . for each such sequence @xmath76 belonging to the direction @xmath60 , the endpoints of the extremal edges of @xmath76 determine two disjoint open arcs on the boundary of @xmath23 that contain all the remaining vertices of @xmath26 .
we denote the arc that contains the vertex @xmath60 by @xmath77 and the other arc by @xmath78
. the following notation will be important in the sequel .
[ not : new ] for a direction @xmath60 , and for @xmath79 , let @xmath80 denote the set @xmath76 of @xmath6 consecutive edges on @xmath49 in direction @xmath60 , for which exactly @xmath81 vertices of @xmath26 lie in @xmath78 .
( this implies : @xmath82 , and therefore restricts the parity of @xmath81 : we have @xmath83 if and only if @xmath60 is * not * a label of a vertex of @xmath26 . )
for example , in the left part of figure [ fig:2new ] , the set of edges depicted in bold is @xmath84 , and the two corresponding arcs on the boundary of @xmath23 are @xmath85 ( containing the vertex @xmath86 ) and @xmath87 ( containing the vertices @xmath88 ) .
we note that , by its definition , the set @xmath80 depends also on @xmath0 and @xmath6 . in order to keep the notation concise
, we prefer to leave this dependence implicit , as the values of @xmath11 can always be understood from the context .
it is clear that all the edges in each direction are pairwise disjoint .
hence , if a set of edges does not contain @xmath1 pairwise disjoint edges , then it contains at most @xmath6 edges in each direction . as the number of distinct directions is @xmath0 ,
we get the following observation ( that was already used by kupitz @xcite ) : [ obs1 ] let @xmath26 be an @xmath12-free cgg of order @xmath0 .
. moreover , if @xmath90 , then @xmath91 contains exactly @xmath6 edges in each direction .
this implies the upper bound of theorem [ thm : main](1 ) .
assume now that , in addition , @xmath26 avoids a set @xmath28 of @xmath42 consecutive vertices , for @xmath93 .
in this case , the maximal possible number of edges is reduced , since , in some of the directions , only fewer than @xmath6 edges have at least one endpoint in @xmath51 . for @xmath94
, we say that direction @xmath81 _ loses _ @xmath95 edges , if it contains exactly @xmath96 edges with at least one endpoint in @xmath51 , and denote this situation by @xmath97 .
( if the number of edges in direction @xmath81 with at least one endpoint in @xmath51 is greater than or equal to @xmath6 , we say that @xmath98 . ) in the proof , we use the following fact . [ claim : triv ] let @xmath99 . then @xmath100 the proof , by induction on @xmath0 , is straightforward and is left to the reader .
let the allowed set @xmath51 be as described in the right part of figure [ fig:3new ] , where @xmath101 are its rightmost vertices .
( note that @xmath102 is a function of @xmath30 and @xmath103 . )
we compute the number of allowed edges in each direction .
the endpoints of each edge in direction @xmath104 are either both allowed or both forbidden .
hence , the number of allowed edges in direction @xmath104 is @xmath105 , or in other words , @xmath106 .
( note that if @xmath103 is odd , the leftmost vertex in @xmath51 is idle . ) in direction @xmath107 , @xmath108 , the edge @xmath109 $ ] is allowed , and so are all edges that are parallel to it and behind it ( i.e. , farther than @xmath50 from @xmath107 ) .
thus , effectively , the allowed arc is extended by @xmath107 vertices on its bottom side , and hence , the number of allowed edges in this direction is @xmath110 .
hence , @xmath111 .
when @xmath107 reaches @xmath103 , the effective length of the allowed arc becomes @xmath112 , and thus , the loss becomes zero . by symmetry ,
the situation for direction @xmath113 is the same as for direction @xmath107 .
therefore , the sum of the losses is @xmath114 where the equality holds by claim [ claim : triv ] .
the argument above is exemplified in the left drawing of figure [ fig:3new ] in the special case @xmath115 .
in direction @xmath104 there are only 2 allowed edges , which corresponds to @xmath116 .
the same holds for @xmath117 , and this corresponds to @xmath118 . for @xmath119 , there are already 3 allowed edges , which indeed corresponds to @xmath120 .
finally , we consider the case where @xmath26 avoids a set @xmath28 of @xmath121 consecutive vertices . in this case
, each edge of @xmath26 uses at least one of the vertices of @xmath51 , and since @xmath122 , this implies that @xmath26 is @xmath12-free .
hence , the only restriction on @xmath26 is the avoidance of @xmath28 , which leads to the upper bound @xmath123 on the number of edges ( as there are @xmath124 edges with both endpoints in @xmath28 ) .
on the other hand , the upper bound is clearly attained by the graph @xmath26 that contains all @xmath123 edges with at least one endpoint in @xmath51 , since this graph is @xmath12-free and avoids @xmath28 .
this proves theorem [ thm : main](3 ) .
in this section we consider @xmath12-free cggs on @xmath0 vertices that avoid a set @xmath28 of @xmath25 consecutive vertices . by observation [ obs1 ] , the number of edges in such a graph is at most @xmath2 . we show that this upper bound is attained by a sequence of graphs @xmath39 , thus proving theorem [ thm : main](1 ) . in sections [ sec :
sub : def - gnk ] and [ sec : sub : basic - proof ] we present the construction of @xmath39 for even values of @xmath0 and prove that these cggs satisfy the conditions of theorem [ thm : main](1 ) . in section [ sec : sub : basic - odd ] we sketch the modifications in the construction of @xmath39 and in the proof of the theorem required in the case of odd @xmath0 .
let @xmath0 be an even integer .
define @xmath125 , and let the free set be @xmath126 . recall that by the definition of _ direction _ given in section [ sec : sub : notations ] , directions @xmath60 and @xmath70 ( for any @xmath71 ) are the same . hence , in this section we restrict ourselves to the set of @xmath0 directions that correspond to the half - circle consisting of the consecutive vertices @xmath127 ( where direction @xmath128 is equal to direction @xmath129 ) .
the cgg @xmath39 that avoids @xmath28 is given by the following definition .
[ def : gnk ] for an even integer @xmath0 , and @xmath130 , we denote by @xmath39 the cgg on @xmath0 vertices whose edge set consists of the following sets of consecutive edges , arranged according to the directions : * for @xmath131 , the set of edges in direction @xmath81 is @xmath132 .
( see notation [ not : new ] . ) * for @xmath133 , @xmath134 , the set of edges in direction @xmath135 is @xmath136 , where @xmath137 note that the sets @xmath138 and @xmath139 are defined twice : @xmath138 appears as @xmath140 in both clauses , while @xmath139 appears as @xmath141 in the first clause and as @xmath142 in the second clause ( since @xmath129 and @xmath128 are the same direction ) . as the structure of @xmath39 is a bit complex
, we give an intuitive explanation of the construction before presenting the proof .
our explanation is illustrated by two examples of @xmath39 , presented in figure [ fig:5new ] .
* intuitive explanation of the construction .
* we consider the directions sequentially , in the order @xmath143 .
( note that direction @xmath129 is the same as direction @xmath128 , and thus , the order of directions we consider is logical . ) in each direction , we choose @xmath6 edges , aiming at choosing the `` most central '' ( i.e. , longest possible ) edges , subject to the restriction that @xmath39 must avoid @xmath28 .
the endpoints of the edges in each direction @xmath81 form two sets of @xmath6 consecutive vertices ( see figure [ fig:5new ] ) .
we call the set which is closer to the leftmost vertex `` the left set '' and denote it by @xmath144 , and call the other set `` the right set '' and denote it by @xmath145 .
is the set whose distance to the vertex labelled @xmath0 is smaller , where the distance @xmath146 of a set of vertices @xmath76 from a single vertex @xmath63 is defined as @xmath147)$ ] . a single exception is direction @xmath104 for which the distances of both sets from vertex @xmath0 are equal . for that direction ,
we choose for convenience @xmath148 and @xmath149 . ] in the first @xmath150 directions , the left set @xmath151 is fixed , while the right set @xmath145 is moved each step one place counterclockwise ( i.e. , starting with @xmath152 , then @xmath153 , etc . ) .
we continue in this fashion until @xmath154 , for which @xmath155 is central ( i.e. , each of the arcs @xmath156 contains exactly @xmath157 vertices ) . in the following directions , we try to maintain centrality : every second direction is central , and the directions in between are `` nearly central '' , i.e. , moved half a step downward from the central positions .
this is achieved by moving @xmath144 and @xmath145 alternately : in passing from @xmath158 to @xmath159 , @xmath160 is moved one step counterclockwise , while @xmath161 remains unchanged . in passing from @xmath159 to @xmath162 , @xmath163
is moved one step counterclockwise , while @xmath164 is retained , etc .
this continues until we reach direction @xmath128 ( again central ) , which is the same as direction @xmath129 .
the transition from here back to direction @xmath104 can be viewed as a mirror image ( in reverse ) of the transition from @xmath165 to @xmath138 . in the last step ( from @xmath166
back to @xmath165 ) @xmath167 is moved one step counterclockwise to @xmath168 , and then the roles of @xmath169 and @xmath170 are interchanged : @xmath171 and @xmath172 . in this section
we prove that @xmath39 satisfies the conditions of theorem [ thm : main](1 ) .
it is clear that @xmath39 contains exactly @xmath6 edges in each direction allowed edges . indeed , by the definition of @xmath28 , the set @xmath51 contains @xmath173 vertices .
hence , at least @xmath6 of the edges in each direction have an endpoint in @xmath51 . ] and avoids the set of vertices @xmath174 . hence , theorem [ thm : main](1 ) is implied by the following proposition . [ prop : basic - main ] for each @xmath0 and @xmath6 , the graph @xmath39 is @xmath12-free . in the proof of proposition
[ prop : basic - main ] , we use a lemma , which requires some additional notation .
[ not : arc ] the endpoints of any edge @xmath175 divide the boundary of @xmath23 into two open arcs .
denote the arc that does not contain the open boundary edge @xmath176n-1,-(n-1)[$ ] by @xmath177 , and the other arc by @xmath178 .
we say that an edge @xmath179 _ lies behind _ @xmath180 ( with respect to @xmath176n-1,-(n-1)[$ ] ) if @xmath181 .
( note that an edge is said to lie behind itself . )
note that the notations @xmath177 and @xmath178 differ from the notations @xmath182 and @xmath183 defined in section [ sec : sub : notations ] .
this difference is intentional , and both types of notation will be used in the sequel .
[ lemma : basic - main ] for any @xmath175 , the open arc @xmath177 contains at least @xmath184 vertices of @xmath26 .
furthermore , if @xmath50 emanates from a positive - labelled vertex , then @xmath177 contains at least @xmath157 vertices of @xmath26 .
we consider separately the sets of edges in @xmath185 for @xmath131 and for @xmath186 .
* case i : @xmath187 for @xmath131 .
* by definition [ def : gnk ] , we have @xmath132 .
recall that , as explained in section [ sec : sub : notations ] , @xmath132 divides the remaining vertices on the boundary of @xmath29 into two arcs : @xmath183 ( which includes the open edge @xmath176n-1,-(n-1)[$ ] ) and @xmath182 .
as by the definition of @xmath188 , @xmath183 contains @xmath189 vertices of @xmath26 , it follows that @xmath182 contains at least @xmath190 vertices of @xmath26 . since for each @xmath187 , the arc @xmath177 includes the arc @xmath182
, it follows that @xmath177 contains at least @xmath157 vertices of @xmath26 .
* case ii : @xmath187 for @xmath186 . * by definition [ def : gnk ] , @xmath191 for odd @xmath60 and @xmath192 for even @xmath60 .
for @xmath193 , each of the arcs @xmath194 and @xmath195 contains exactly @xmath157 vertices of @xmath26 , and thus , for any @xmath196 , the arc @xmath177 contains at least @xmath157 vertices of @xmath26 .
similarly , for @xmath197 , the arcs @xmath194 and @xmath195 contain @xmath150 and @xmath184 vertices of @xmath26 respectively , and thus , for any @xmath198 , the arc @xmath177 contains at least @xmath184 vertices of @xmath26 . when trying to show that , for some of the edges , the arc @xmath177 contains at least @xmath157 vertices of @xmath26 , we face a difficulty : unlike the case @xmath131 , where for any @xmath187 the arc @xmath177 includes the arc @xmath182 , here there is a variance between the edges . for some of the edges @xmath50 , @xmath177
includes @xmath194 ( which contains @xmath150 vertices of @xmath26 ) , while for the other edges , @xmath177 includes @xmath195 ( which contains only @xmath184 vertices of @xmath26 ) . however , we observe that the edges @xmath50 for which @xmath177 includes @xmath194 are exactly those which emanate from positive - labelled vertices ( see figure [ fig:7new ] ) . indeed ,
if @xmath199 emanates from a positive - labelled vertex , then the vertex @xmath200 and the open edge @xmath176-(n-1),n-1[$ ] lie on opposite sides of the edge @xmath50 .
thus , by the definitions of @xmath177 and of @xmath194 , it follows that @xmath201 .
on the other hand , if @xmath50 emanates from a negative - labelled vertex , then @xmath200 and @xmath176-(n-1),n-1[$ ] lie on the same side of @xmath50 , and thus , @xmath195 is included in @xmath202 .
hence , we conclude that for any @xmath175 that emanates from a positive - labelled vertex , @xmath177 contains at least @xmath157 vertices of @xmath26 .
this completes the proof of the lemma .
( of proposition [ prop : basic - main ] ) assume , on the contrary , that @xmath39 contains a set @xmath76 of @xmath1 pairwise disjoint edges . by the construction of @xmath39 ,
each edge of @xmath76 has at least one endpoint in the set @xmath203 that consists of @xmath6 positive - labelled vertices and @xmath6 negative - labelled vertices ( see figure [ fig:7.5 ] ) . as the edges of @xmath76 are disjoint , and there are @xmath1 edges in @xmath76 , there exist two edges @xmath204,[i_2,t_2 ] \in s$ ] , such that : * @xmath204 $ ] does not have a negative - labelled endpoint in @xmath51 . in the notations of figure [ fig:7.5 ] , @xmath204 $ ] does not have an endpoint in @xmath205 , so it must have an endpoint in @xmath206 .
moreover , @xmath207 } \subset d_1 \cup d_2 $ ] . * @xmath208 $ ] does not have a positive - labelled endpoint in @xmath51 .
in the notations of figure [ fig:7.5 ] , @xmath208 $ ] does not have an endpoint in @xmath206 , so it must have an endpoint in @xmath205 .
moreover , @xmath209 } \subset d_1 \cup d_3 $ ] .
it follows that none of the arcs @xmath207},arc_{[i_2,t_2]}$ ] is included in the other , and therefore they are disjoint ( see figure [ fig:7.5 ] ) .
we would like to show that the arc @xmath207}$ ] contains at least @xmath157 vertices of @xmath26 that are not used by any edge of @xmath76 , and the arc @xmath209}$ ] contains at least @xmath184 such vertices .
this will be a contradiction to the assumption @xmath210 , since by the assumption , @xmath76 uses @xmath211 vertices of @xmath26 , and thus , only @xmath212 vertices of @xmath26 are not used by @xmath76 .
we say that an edge @xmath213 is _ extremal _ if no other edge of @xmath76 lies behind @xmath50 ( see notation [ not : arc ] ) .
it is immediate that behind any edge of @xmath76 there is an extremal one .
note that since the edges of @xmath76 are pairwise disjoint , if @xmath214 is an extremal edge of @xmath76 then the vertices of @xmath215 are not used by any edge of @xmath76 .
let @xmath180 and @xmath179 be extremal edges that lie behind @xmath204 $ ] and @xmath208 $ ] , respectively .
( note that we may have @xmath216 $ ] and/or @xmath217 $ ] . ) by lemma [ lemma : basic - main ] , each of the arcs @xmath218 and @xmath219 contains at least @xmath184 vertices of @xmath26 , and these vertices are certainly not used by any edge of @xmath76 . moreover , since both endpoints of @xmath180 are contained in @xmath207 } \subset \{n-1,n-3,\ldots,1,-1,\ldots,-(n-2k-3),-(n-2k-1)\}$ ] , it follows that @xmath180 emanates from a positive - labelled vertex .
( this holds since @xmath180 must emanate from a vertex in @xmath51 , and @xmath207 } \cap a^c$ ] contains only positive - labelled vertices . )
hence , lemma [ lemma : basic - main ] actually implies that @xmath218 contains at least @xmath157 vertices of @xmath26 that can not be used by edges of @xmath76 . in total
, we have at least @xmath220 vertices of @xmath26 that can not be used by edges of @xmath76 ( and these vertices are distinct since @xmath221 } \cap arc_{[i_2,t_2 ] } = \emptyset$ ] ) .
this leads to a contradiction as explained above , and thus concludes the proof of proposition [ prop : basic - main ] .
the case of odd @xmath0 is very similar to the case of even @xmath0 , and is , in some respect , even easier .
thus , we only sketch briefly the required modifications in the notations , in the construction of @xmath39 , and in the proof of theorem [ thm : main](1 ) . * notations . *
the first slight change is in the notations .
let @xmath0 be odd , and let * @xmath222*. as @xmath223 is odd , we take the vertices of @xmath49 to be the even - labelled vertices of @xmath23 , as shown in figure [ fig:3old ] .
this is in contrast with the case of even @xmath0 , where @xmath48 is even , and the vertices of @xmath49 are taken to be the odd - labelled vertices of @xmath23 ( see section [ sec : sub : notations ] ) . * construction . *
the second change is in the definition of @xmath39 .
while the definition is very similar to the even case , there are a few changes .
the formal definition of @xmath39 in the odd case is as follows : [ def : gnk - odd ] for an odd integer @xmath0 , and @xmath224 , we denote by @xmath39 the cgg on @xmath0 vertices whose edge set is the union of the following sets of consecutive edges , arranged according to the directions : * for @xmath225 , the set of edges in direction @xmath81 is @xmath132 . * for @xmath133 , @xmath226 ,
the set of edges in direction @xmath135 is @xmath227 , where @xmath228 ( note that the edges in directions @xmath229 are defined twice .
of course , the definitions coincide . )
an example of @xmath39 , with @xmath230 and @xmath231 , is presented in figure [ fig:6new ] .
* main lemma . * the main lemma used in the proof of theorem [ thm : main](1 ) in the even case , lemma [ lemma : basic - main ] , is replaced by the following lemma , which turns out to be even stronger : [ lemma : basic - main - odd ] for any @xmath175
, the open arc @xmath177 contains at least @xmath157 vertices of @xmath26 . to prove lemma [ lemma : basic - main - odd ]
, we note that in the odd case , a set of the form @xmath232 or @xmath233 leaves @xmath157 and @xmath150 vertices of @xmath26 on its sides ( compared to @xmath157 and @xmath157 or @xmath184 and @xmath150 in the even case ) . as a result , since for any @xmath81 such that @xmath234 , @xmath185 is of one of the forms @xmath232 or @xmath233 , it follows immediately that for any edge @xmath187 ( for @xmath234 ) , @xmath177 contains at least @xmath157 vertices of @xmath26 .
the proof for @xmath235 is identical to the proof of the corresponding statement of lemma [ lemma : basic - main ] in the even case .
* proof of theorem [ thm : main](2 ) . * due to the stronger form of lemma [ lemma : basic - main - odd ] , the proof is even simpler than in the even case .
same as in the even case , we assume on the contrary that @xmath39 contains a set @xmath76 of @xmath1 pairwise disjoint edges , and show that @xmath76 contains two edges @xmath204,[i_2,t_2]$ ] such that @xmath204 $ ] does not have a negative - labelled endpoint in @xmath51 , and @xmath208 $ ] does not have a positive - labelled endpoint in @xmath51 .
furthermore , each of the closed arcs @xmath207},arc_{[i_2,t_2]}$ ] contains the endpoints of an extremal edge ( which may be one of the edges @xmath204,[i_2,t_2]$ ] ) .
we denote these extremal edges by @xmath180 and @xmath179 , respectively . by lemma [ lemma : basic - main - odd ] , each of the arcs @xmath236 contains at least @xmath157 vertices of @xmath26 that are not used by edges of @xmath76 .
hence , the number of vertices that are used by edges of @xmath76 is at most @xmath237 , contradicting the assumption that @xmath76 consists of @xmath1 _ disjoint _ edges .
this completes the proof of the theorem for odd @xmath0 .
in this section we consider @xmath12-free cggs on @xmath0 vertices that admit a free boundary arc of order @xmath239 , for @xmath240 . as shown in section [ sec : notations ] , the number of edges in such a graph is at most @xmath241 . we show that this upper bound is attained by a sequence of graphs @xmath41 , thus proving theorem [ thm : main](2 ) .
same as in section [ sec : basic - construction ] , we write @xmath242 ) , so that @xmath243 if @xmath0 is even , and @xmath244 if @xmath0 is odd . as indicated in section [ sec : sub : notations ] , we take @xmath49 to be the set of odd - labelled vertices of @xmath23 if @xmath48 is even , or the set of the even - labelled vertices of @xmath23 if @xmath48 is odd .
we define @xmath245 , and divide @xmath246 into three blocks : @xmath247 ( the uppermost @xmath248 vertices of @xmath246 ) , @xmath249 ( the lowest @xmath248 vertices of @xmath246 ) , and @xmath250 ( the @xmath103 vertices in the middle ) , see figure [ fig:9newnew ] . for ease of exposition , we consider the cases of even and odd @xmath0 separately . in sections [ sec :
sub : def - gnkl ] and [ sec : sub : advanced - proof ] we present the construction of @xmath41 for even values of @xmath0 and prove that these cggs satisfy the conditions of theorem [ thm : main](2 ) . in section [ sec : sub :
advanced - odd ] we sketch the modifications in the construction of @xmath41 and in the proof of the theorem required in the case of odd @xmath0 . for @xmath251
even , we describe the edges of @xmath41 by direction , as follows .
the directions range over the half - circle from @xmath252 up to @xmath253 .
the set of edges of @xmath41 in direction @xmath81 is denoted by @xmath185 .
we divide the directions into five sub - ranges : * * @xmath259 : * for @xmath259 , @xmath165 consists of the @xmath260 leftmost vertical edges . *
* @xmath261 : * we enlarge @xmath246 by adding @xmath81 vertices at the bottom . @xmath185 consists of the @xmath262 parallel edges that fit into this enlarged arc , with one free vertex in the middle if @xmath263 is odd . *
* @xmath264 : * same as above , with `` bottom '' replaced by `` top '' and @xmath81 replaced by @xmath265 . *
* @xmath257 : * we enlarge @xmath246 by adding @xmath81 vertices at the bottom . @xmath185
consists of the @xmath6 most central parallel edges that fit into this enlarged arc , leaving @xmath266 free vertices in its center .
( @xmath267 is already central . )
* * @xmath254 : * same as above , with `` bottom '' replaced by `` top '' , and @xmath81 by @xmath265 . * * @xmath268 : * @xmath185 is a @xmath6-block , central if @xmath269 , and near central if @xmath270 . in the latter case ,
there are @xmath150 unused vertices of @xmath26 above @xmath185 , and @xmath184 below .
the idea behind the construction is the following : for each direction @xmath81 , we consider the set of allowed edges ( i.e. , edges with at least one endpoint included in @xmath51 ) .
the edges of this set are consecutive . if their number is at most @xmath6 , we take all of them to @xmath41 . if their number exceeds @xmath6 , we take the @xmath6 edges which are the closest to the center , and if there are two @xmath6-blocks with the same distance from the center , we take the lower one among them . another way to view
the construction is to compare @xmath41 to the corresponding graph @xmath39 .
when moving from @xmath39 to @xmath41 , @xmath103 vertices are added to the forbidden set @xmath28 . as a result
, we lose part of the edges that emanate from these @xmath103 vertices . when defining @xmath41 , we retain all edges of @xmath39 that remain allowed , and add some edges to compensate for the edges that become forbidden .
specifically , in each direction @xmath81
in which all edges of @xmath39 remain allowed , we take @xmath185 to be the set of edges of @xmath39 in that direction . in the directions
in which some of the edges of @xmath39 become forbidden , we observe that these edges are those emanating from a set of consecutive vertices . for each such direction , we remove from the edges of @xmath39 the edges that become forbidden and compensate for them by adding a set of consecutive edges on the other side of the set of @xmath39 .
if the number of edges we can add is smaller than the number of edges that become forbidden ( since the direction `` ends '' , in a single vertex or in a boundary edge ) , we add the maximal possible number of edges . *
the set of edges of @xmath41 in direction
@xmath81 is consecutive . *
if at most @xmath6 edges in direction @xmath81 are allowed , then @xmath41 contains all allowed edges in direction @xmath81 .
otherwise , @xmath41 contains @xmath6 edges in direction @xmath81 .
an example of the construction @xmath41 , for @xmath271 , @xmath272 , and @xmath273 , is presented in the left part of figure [ fig:9newnew ] . in the figure , in directions
@xmath274 all edges remain allowed , in directions @xmath275 a single edge becomes forbidden , and in directions @xmath276 two edges become forbidden . on the other hand , a single compensating edge is added to each of directions @xmath277 . as a result
, @xmath278 contains @xmath279 edges in directions @xmath280 , and @xmath281 edges in each other direction .
* item ( 2 ) : * the number of edges of @xmath41 in direction @xmath81 is : @xmath260 for @xmath259 , @xmath286 for @xmath287 , and @xmath6 for all other @xmath288 directions .
it follows that @xmath289 by claim [ claim : triv ] .
* item ( 1 ) : * this is the main claim .
assume , on the contrary , that @xmath76 is a set of @xmath1 pairwise disjoint edges of @xmath41 .
since @xmath290 , there is an edge @xmath291 that does not use any vertex of @xmath292 .
it follows that @xmath293 is not vertical , that the left vertex of @xmath294 belongs to @xmath247 , and that its right vertex belongs to @xmath295 or to @xmath28 . among all edges in @xmath76
whose left endpoint is in @xmath247 , choose the one whose left endpoint is as far to the right as possible and call it @xmath296 . by the construction of @xmath41 , there are at least @xmath157 vertices above @xmath296 that are not used by any edge in @xmath76 . similarly , there is an edge @xmath297 that does not use any vertex in @xmath298 .
choose such an edge @xmath299 whose left endpoint is as far to the right as possible . by the construction of @xmath41 , there are at least @xmath184 vertices below @xmath299 that are not used by any edge of @xmath76 .
it follows that @xmath76 uses altogether at most @xmath300 vertices , a contradiction .
the construction of @xmath41 in the odd case is almost exactly the same as in the even case .
the only difference in the definition is in the last sub - range of directions , namely , @xmath301 . in the odd case ,
the directions @xmath302 are added to that sub - range , and ( unlike the even case ) , @xmath185 is always an almost central @xmath6-block , such that there are @xmath157 unused vertices of @xmath26 above it if @xmath303 , and @xmath150 unused vertices above it if @xmath269 .
an example of @xmath41 , with @xmath304 and @xmath273 , is presented in the right part of figure [ fig:9newnew ] . in the figure , in directions
@xmath305 all edges remain allowed , in directions @xmath306 a single edge becomes forbidden , and in directions @xmath307 two edges become forbidden . on the other hand , a single compensating edge is added to each of directions @xmath308 , and two compensating edges are added to direction @xmath279 . as a result , @xmath309 contains @xmath279 edges in directions @xmath280 , and @xmath281 edges in each other direction .
the proof of theorem [ thm : main](2 ) in the odd case is almost identical to the proof in the even case .
the only difference is that in the odd case , both @xmath296 and @xmath299 leave at least @xmath157 unused vertices of @xmath26 behind them .
this implies that at most @xmath237 vertices of @xmath26 are used by edges of @xmath310 , a contradiction to the assumption that the @xmath1 edges in @xmath310 are disjoint .
this completes the proof of theorem [ thm : main](2 ) for odd @xmath0 .
o. aichholzer , s. cabello , r. fabila - monroy , d. flores - pealoza , t. hackl , c. huemer , f. hurtado , and d. r. wood , edge - removal and non - crossing configurations in geometric graphs , _ disc .
, _ * 12(1 ) * ( 2010 ) , pp .
7586 . p. brass , g. krolyi , and p. valtr , a turn - type extremal theory of convex geometric graphs , in : discrete and computational geometry the goodman - pollack festschrift , b. aronov et al .
eds , springer - verlag , 2003 , pp . 275300 . | a well - known result of kupitz from 1982 asserts that the maximal number of edges in a convex geometric graph ( cgg ) on @xmath0 vertices that does not contain @xmath1 pairwise disjoint edges is @xmath2 ( provided @xmath3 ) . for @xmath4 and @xmath5 ,
the extremal examples are completely characterized . for all other values of @xmath6 ,
the structure of the extremal examples is far from known : their total number is unknown , and only a few classes of examples were presented , that are almost symmetric , consisting roughly of the @xmath2 `` longest possible '' edges of @xmath7 , the complete cgg of order @xmath0 . in order to understand further the structure of the extremal examples , we present a class of extremal examples that lie at the other end of the spectrum .
namely , we break the symmetry by requiring that , in addition , the graph admit an independent set that consists of @xmath8 consecutive vertices on the boundary of the convex hull .
we show that such graphs exist as long as @xmath9 and that this value of @xmath8 is optimal .
we generalize our discussion to the following question : what is the maximal possible number @xmath10 of edges in a cgg on @xmath0 vertices that does not contain @xmath1 pairwise disjoint edges , and , in addition , admits an independent set that consists of @xmath8 consecutive vertices on the boundary of the convex hull ? we provide a complete answer to this question , determining @xmath10 for all relevant values of @xmath11 and @xmath8 . |
axions were postulated to solve the strong - cp problem in qcd in the 1970s @xcite , and are still the most compelling explanation for it . moreover
, they are also appealing for astrophysicists , since they are valid dark matter candidates to constitute a portion or the totality of the non - barionic cold dark matter content predicted to exist in the universe .
but probably the most interesting property of axions , or in a more generic way , axion - like particles ( alps ) , where the mass @xmath0 and the coupling constant are not related to each other , is that they are expected to oscillate into photons ( and viceversa ) in the presence of magnetic fields @xcite .
this oscillation of photons to alps are the main vehicle used at present in axion searches , like those carried out by cast @xcite or admx @xcite , but they could also have important implications for astronomical observations .
for example , they could distort the spectra of gamma - ray sources , such as active galactic nuclei ( agns ) @xcite or galactic sources in the tev range @xcite .
these distortions may be detected by current gamma - ray experiments , such as imaging atmospheric cherenkov telescopes ( iacts ) like magic @xcite , hess @xcite , veritas @xcite or cangaroo - iii @xcite , covering energies in the range 0.1 - 20 tev , and by the fermi satellite @xcite , which operates at energies in the range 0.02 - 300 gev . given a domain of length @xmath1 , where there is a roughly constant magnetic field and plasma frequency , the probability of a photon of energy @xmath2 to be converted into an alp after traveling through it can be written as @xcite : @xmath3 here @xmath4 is the oscillation wave number : @xmath5 @xmath6 that gives us an idea of how effective is the mixing , i.e. @xmath7 where @xmath8 the magnetic field component along the polarization vector of the photon and @xmath9 the inverse of the coupling constant .
@xmath10 is the vacuum cotton - mouton term , i.e. @xmath11 where @xmath12 g the critical magnetic field strength ( @xmath13 is the electron charge ) .
@xmath14 is the plasma term : @xmath15 where @xmath16 the plasma frequency , @xmath17 the electron mass and @xmath18 the electron density .
finally , @xmath19 is the alp mass term : @xmath20 note that in eqs.([eq : deltab]-[eq : deltaa ] ) we have introduced the dimensionless quantities @xmath21 g , @xmath22 gev and @xmath23 ev .
since we expect to have not only one coherence domain but several domains with magnetic fields different from zero and subsequently with a potential photon / axion mixing in each of them , we can derive a total conversion probability @xcite as follows : @xmath24 \label{eq : totalprob}\ ] ] where @xmath25 is given by eq.([eq : p0 ] ) and @xmath26 represents the number of domains .
note that in the limit where @xmath27 , the total probability saturates to 1/3 , i.e. one third of the photons will convert into alps .
it is useful here to rewrite eq .
( [ eq : p0 ] ) following ref .
@xcite , i.e. @xmath28\ ] ] so that we can define a characteristic energy , @xmath29 , given by : @xmath30 or in more convenient units : @xmath31 where the subindices refer again to dimensionless quantities : @xmath32 , @xmath33 gev and @xmath34 b / gauss ; @xmath35 is the effective alp mass @xmath36 .
recent results from the cast experiment @xcite give a value of @xmath37 for axion mass @xmath38 ev .
although there are other limits derived with other methods or experiments , the cast bound is the most general and stringent limit in the range @xmath39 ev @xmath40 ev . in order to have an efficient conversion , we need @xcite : @xmath41 where s@xmath42 s / parsec . some astrophysical environments fulfill the above mixing requirement and the m@xmath43 constraints imposed by cast . indeed , when using @xmath37 in eq .
( [ eq : effconv ] ) , we can deduce that astrophysical sources with b@xmath44 0.01 will be valid .
this product b@xmath45 also determines the maximum energy e@xmath46 to which sources can accelerate cosmic rays , and is given by e@xmath47 b@xmath48 ev ( hillas criterium ) .
since we observe cosmic rays up to 3 @xmath49 10@xmath50 ev , b@xmath51 s@xmath52 can be as high as 0.3 , which also means that sources with b@xmath53 0.01 are completely plausible and should exist .
therefore , photon / axion mixing may have important implications for astronomical observations ( agns , pulsars , grbs ... ) .
note , however , that an efficient mixing can be expected to occur not only in compact sources : the mixing will be also present in the intergalactic magnetic fields ( igmfs ) , with typical values of @xmath541 ng for the b field , provided that the source is located at cosmological distances ( s@xmath55 ) in order to ensure that b @xmath56 is still larger than 0.01 .
therefore , in order to correctly evaluate the photon / axion mixing effect for distant sources , it will be necessary to handle under the same consistent framework the mixing that takes place inside or near the gamma - ray sources together with that one expected to occur in the igmf . in the literature , both effects have been considered separately .
we neglect , however , the mixing that may happen inside the milky way due to galactic magnetic fields.3% of the initial photon flux emitted by the source @xcite . ] to illustrate how the photon / axion mixing inside the source works , we show in figure [ fig : sourcemix ] an example for an agn modeled by the parameters listed in table [ tab:3c279 ] .
the only difference is the use of an alp mass of 1 @xmath57ev instead of the value that appears in that table , so that we can obtain a critical energy that lies in the gev energy range ; indeed , we get @xmath58 gev according to eq .
( [ eq : ecrit ] ) .
note that the main effect just above @xmath29 is an _ attenuation _ of the total expected flux coming from the source.10 gev ) gradually , the reason being the crucial role of the cotton - mouton term , which makes the efficiency of the source mixing to decrease as the energy increases . see ref .
@xcite for details . ] as already discussed , despite the low magnetic field * b * , the photon / axion oscillation can take place in the igmfs due to the large distances .
however , the strength of the igmfs is expected to be many orders of magnitude weaker ( @xmath54ng ) than that of the source and its surroundings ( @xmath54 g ) . consequently , as described by eq .
( [ eq : ecrit ] ) , the energy at which photon / axion oscillation occurs in this case is many orders of magnitude larger than that at which oscillation can occur in the source and its vicinity . assuming a mid - value of [email protected] ng , and @xmath59 ( cast upper limit )
, the effect could be observationally detectable by current iacts ( @xmath60 tev ) only if the alp mass is m@xmath61 ev , i.e. we need ultra - light alps . for example , we get @xmath62 gev when using m@xmath63 ev , which is the value given in table [ tab:3c279 ] as our reference one . .[tab:3c279 ] [ cols="<,<,<,<",options="header " , ] in our model , we assume that the photon beam propagates over n domains of a given length , the modulus of the magnetic field roughly constant in each of them .
we take , however , randomly chosen orientations , which in practice is also equivalent to a variation in the strength of the component of the magnetic field involved in the photon / axion mixing .
moreover , it will be necessary to include the effect of the extragalactic background light ( ebl ) in the equations as well , its main effect being an additional attenuation of the photon flux , especially at energies above 100 gev .
indeed , the ebl plays a crucial role to correctly evaluate and understand the importance of the intergalactic mixing .
the induced effect can be an _ attenuation _ or an _ enhancement _ of the photon flux at earth , depending on distance , magnetic field and the ebl model considered ( see next section ) . in conclusion , agns located at cosmological distances will be affected by both mixing in the source and in the igmf .
as already mentioned , up to now previous works have focused only in studying the photon / axion mixing either inside the source or in the igmfs . instead , for the first time we carried out a detailed study of the mixing in both regimes under the same consistent framework .
in order to observe both effects in the gamma - ray band , we need ultra - light alps . that is the reason why in our fiducial model , presented in table [ tab:3c279 ] , we use a value of m@xmath63 ev , which implies @xmath64 30 gev for the igmf mixing ( for [email protected] ng ) and @xmath64 1 ev within the source and its vicinity ( b@xmath541 g ) .
consequently , both effects need to be taken into account .
we show in fig .
[ fig : sketch ] a diagram that outlines our formalism .
very squematically , the diagram shows the travel of a photon from the source to the earth in a scenario with alps . in the same figure
, we show the main physical cases that one could identify inside our formalism : mixing in both the source and the igmf , mixing in only one of these environments , the effect of the ebl , etc .
since we expect the intergalactic mixing to be more important for larger distances , due to the more prominent role of the ebl , we chose two distant astrophysical sources as our benchmark agns : the radio quasar 3c 279 ( z=0.536 ) and the bl lac pks 2155 - 304 at z=0.117 .
we summarize in table [ tab:3c279 ] the parameters we have considered in order to calculate the total photon / axion conversion in both the source and in the intergalactic medium . as for the ebl model
, we chose the primack @xcite and kneiske best - fit @xcite ones .
they represent respectively one of the most transparent and one of the most opaque models for gamma - rays , but still within the limits imposed by the observations . + in order to quantitatively study the effect of alps on the total photon intensity , we plot in figure [ fig : boosts ] the difference between the predicted arriving photon intensity without including alps and that one obtained when including the photon / axion oscillations ( called here the
_ axion boost factor _ ) .
this was done for our fiducial model ( table [ tab:3c279 ] ) and for the two ebl models considered .
the plots show differences in the axion boost factors obtained for 3c 279 and pks 2155 - 304 due mostly to the redshift difference .
the inferred critical energies for the mixing in the source are @xmath65 ev for 3c 279 and @xmath66 ev for pks 2155 - 304 , while for the mixing in the intergalactic medium we obtain @xmath67 gev ( the same for both objects ) . for [email protected]
ng , and in the case of 3c 279 , the axion boost is an attenuation of about 16% below the critical energy ( due to mixing inside the source ) . above this critical energy and below 200 - 300 gev , where the ebl attenuation is still small
, there is an extra attenuation of about 30% ( mixing in the igmf ) . above 200 - 300
gev the axion boost reaches very high values : at 1 tev , a factor of @xmath547 for the primack ebl model and @xmath54340 for the kneiske best - fit model .
we find that the more attenuating the ebl model considered , the more relevant the effect of photon / axion oscillations in the igmf , since any alp to photon reconversion might substantially enhance the intensity arriving at earth . in the case of pks 2155 - 304 , the situation is different from that of 3c 279 due to the very low photon - attenuation at the source and , mostly , due to the smaller source distance .
furthermore , a very interesting result is found when varying the modulus of the intergalactic magnetic field .
higher @xmath69 values do not necessarily translate into higher photon flux enhancements .
there is always a @xmath69 value that maximizes the axion boost factors ; this value is sensitive to the source distance , the considered energy and the ebl adopted model ( see ref .
@xcite for a detailed discussion on this issue ) .
if we accurately knew the intrinsic spectrum of the sources and/or the density of the ebl , we should be able to observationally detect axion signatures or to exclude some portions of the parameter space .
we lack this knowledge , so detection is challenging but we believe that still possible .
the combination of the fermi / lat instrument and the iacts , which cover 6 decades in energy ( from 20 mev to 20 tev ) is very well suited to study the photon / axion mixing effect . nevertheless , and before assuming an scenario with axions to interpret the observational data , we should try to describe the observational data ( preferably several agns located at different redshifts , as well as the same agn undergoing different flaring states , from radio to tev ) with `` conventional '' theoretical models for the agn emission and for the ebl . if these `` conventional '' models for the source emission and ebl fail ( i.e. if we have important residuals for the best - fit model )
, then the axion scenario should be explored .
[ fig : obs ] summarizes what could be a good observational strategy to look for alps with fermi and iacts .
recent gamma observations might already pose substantial challenges to the conventional models to explain the observed source spectra and/or ebl density , e.g. : * the veritas collaboration recently claimed a detection above 0.1 tev coming from 3c66a ( [email protected] ) .
the ebl - corrected spectrum seems to be harder than 1.5 @xcite .
* tev photons coming from 3c 66a ?
( see refs .
if so , difficult to explain with conventional ebl models and physics . * the lower limit on the ebl at 3.6
@xmath57 m was recently revised upwards by a factor @xmath542 , suggesting a more opaque universe @xcite .
* some sources at z @xmath68 0.1 - 0.2 seem to have harder intrinsic energy spectra than previously anticipated @xcite . while it is still possible to explain these observations with conventional physics , the axion / photon oscillation would naturally explain these puzzles , since we get more high energy photons than expected as well as a softer intrinsic spectrum in a scenario with alps .
an example is given in fig .
[ fig : axionsfriends ] ( from ref . ) , where the effect of the existence of alps in the intrinsic spectrum of 3c 279 is shown assuming the fiducial model presented in table [ tab:3c279 ] and a kneiske best - fit ebl model .
if axions exist , then they could distort the spectra of astrophysical sources importantly .
in addition , if photon / axion mixing occurs in the igmfs , then a mixing in the source will be at work as well .
in particular , for axion masses of the order or 10@xmath70 ev , the induced effect is expected to be present in the gamma ray energy range . since photon / axion mixing in both the source and the igm
are expected to be at work over several decades in energy , it is clear that a joint effort of fermi and current iacts is needed .
the fermi / lat instrument is expected to play a key role , since it will detect thousands of agns ( up to z@xmath545 ) , at energies where the ebl is not important . by the other hand , iacts will be specially important at higher energies ( @xmath71300 gev ) , where the ebl is present .
main caveats : the effect of photon / axion oscillations could be attributed to conventional physics in the source and/or propagation of the gamma - rays towards the earth .
we believe that detailed observations of agns at different redshifts and different flaring states could be successfully used to identify the signature of an effective photon / axion mixing . 99 snchez - conde m. a. , paneque d. , bloom e. , prada f. and domnguez a. , 2009 , _ phys .
d _ , 79 , 123511 peccei r. d. and quinn h. r. , 1977 , _ phys .
_ , * 38 * , 1440 dicus d. a. , kolb e. w. , teplitz v. l. and wagoner r. v. , 1978 , _ phys .
d _ , * 18 * , 1829 sikivie p. , 1983
_ , * 51 * , 1415 [ erratum ibid . , 1984 , _ phys
_ , * 52 * , 695 andriamonje s. et al , 2007 , _ jcap _ , * 0704 * , 010 duffy l. d. et al . , 2006 , _ phys .
d _ , * 74 * , 012006 hooper d. and serpico p. , 2007
_ , * 99 * , 231102 de angelis a. , roncadelli m. and mansutti o. , 2007 , _ phys .
d _ , * 76 * , 121301 hochmuth k. a. , sigl g. , 2007 , _ phys .
d _ , * 76 * , 123011 simet m. , hooper d. and serpico p. , 2008
d _ , * 77 * , 063001 mirizzi a. , raffelt g. g. and serpico p. , 2007
d _ , * 76 * , 023001 lorentz e. et al . , 2004 , _ new astron .
_ , * 48 * , 339 hinton j. a. , 2004 , _ new astron
_ , * 48 * , 331 weekes t. c. et al . , 2002 , _ astropart .
_ , * 17 * , 221 enomoto r. et al . , 2002 , _ astropart .
_ , * 16 * , 235 gehrels n. and michelson p. , 1999 , _ astropart
_ , * 11 * , 277 hartmann et al . , 2001 , _
apj _ , * 553 * , 683 kusunose m. and takahara f. , 2008 , _ apj _ , * 682 * primack j.r . , 2005 , _ proceedings of the gamma 2004 symposium on high energy gamma ray astronomy _ , 26 - 30 july 2004 , heidelberg , germany , astro - ph/0502177 kneiske t. m. , bretz t. , mannheim k. and hartmann d. h. , 2004 , _ a&a _ ,
* 413 * , 807 acciari v. a. , 2009 , _ apj _ , * 693 * , l104 aliu e. et al . , 2009 , _
apj _ , * 692 * , l29 neshpor y. i. , stepanyan a. a. , kalekin o. p. , fomin v. p. , chalenko n. n. & shitov v. g. 1998 , _ astron .
_ , * 24 * , 134 stepanyan a. a. , neshpor y. i. , andreeva n. a. , kalekin o. p. , zhogolev n. a. , fomin v. p. & shitov v. g. , 2002 , _ astron . rep .
_ , * 46 * , 634 levenson l. r. & wright e. l. , 2008 , _ apj _ , * 683 * , 585 krennrich , dwek and imran , 2008 , _ apj _ , * 689 * , l93 snchez - conde m. a. & domnguez a. , in prep .
albert et al . , 2008 , _
, 320 , 1752 | axion like particles ( alps ) , postulated to solve the strong - cp problem , are predicted to couple with photons in the presence of magnetic fields , which may lead to a significant change in the observed spectra of gamma - ray sources such as agns . here
we simultaneously consider in the same framework both the photon / axion mixing that takes place in the gamma - ray source and that one expected to occur in the intergalactic magnetic fields .
we show that photon / axion mixing could explain recent puzzles regarding the observed spectra of distant gamma - ray sources as well as the recently published lower limit to the ebl intensity .
we finally summarize the different signatures expected and discuss the best strategy to search for alps with the fermi satellite and current cherenkov telescopes like cangaroo , hess , magic and veritas . |
the standard model for cosmology can be described by a small number of parameters , and relatively accurate determinations of many of those parameters are now available . recently
, cosmic microwave background experiments , most notably wmap , have determined the primordial spectrum of density fluctuations down to small angular scales with excellent agreement with galaxy and cluster surveys ( bennett 2003 ; spergel 2003 ) .
the overall curvature of the universe is determined to be @xmath4 within a few percent , where @xmath5 .
the evolution of the universe ( past , present , and future ) also depends on the composition of matter in the universe : radiation , baryons , cold dark mater , a cosmological constant or dark energy .
agreement between cmb measurements , supernovae type i data ( riess 1999 ; perlmutter 1999 ; tonry 2003 ) , and clustering data ( percival 2002 ; tegmark 2003 ) are concordant with @xmath6 approximately 2:1 . here , we are interested in the redundancies in the determination of the baryon density , @xmath7 , using both the cmb and big bang nucleosynthesis ( bbn ) ( walker 1991 ; sarkar 1996 ; olive , steigman , & walker 2000 ; fields & sarkar 2002 ) . the baryon density ( or the baryon - to - photon ratio , @xmath8 ) is the sole parameter in the standard model of bbn . prior to the recent high accuracy measurements of the microwave background power spectrum , the best available method for determining the baryon density of the universe was the concordance of the bbn predictions and the observations of the light element abundances of d , 3 , 4 , and 7 .
a high - confidence upper limit to the baryon density has long been available ( reeves 1976 ) from observations of local d / h abundance determinations ( giving roughly @xmath9 ) , but a reliable lower bound to @xmath3 , much less a precise value , has been problematic .
observations of each of the light elements d , 4 , and 7 can be used individually to determine the value of @xmath3 .
confidence in any such determination , however , relies on the concordance of the three light isotopes .
likelihood analyses ( fields & olive 1996 ; fields 1996 ; hata 1997 ; fiorentini 1998 ; esposito 2000 ; cyburt , fields , & olive 2001 ; burles , nollett , & turner 2001 ) using the combined 4 , 7 and d / h observations enable one to determine a 95 % cl range of @xmath10 with a most likely value of @xmath11 corresponding to @xmath12 . however , one concern regarding the likelihood method is the relatively poor agreement between 4 and 7 on the one hand and d on the other .
the former two taken alone indicate that the most likely value for @xmath13 is 2.4 , while d / h alone implies a best value of 6.1 .
this discrepancy may point to new physics , but could well be due to underestimated systematic errors in the observations .
recently , more weight has been given to the d / h determinations because of their excellent agreement with the cmb experiments . in the past few years , balloon and ground - based observations have made the first observations at multipoles @xmath14 , where the sensitivity to @xmath3 lies , and determinations of @xmath3 at the 20 % level became possible ( netterfield 2002 ; abroe 2002 ; pryke 2002 ; sievers 2002 ; rubio - martin 2003 ; goldstein 2002 ; benoit 2002 ) .
with wmap ( bennett 2003 ; spergel 2003 ) , the cmb - based inference of the baryon - to - photon ratio is @xmath15 , or @xmath16 a precision of 4% .
this estimate is the best - fit wmap value , which is sensitive mostly to wmap alone ( primarily the first and second acoustic peaks ) but does include cbi ( sievers 2002 ) and acbar ( goldstein 2002 ) data on smaller angular scales .
note , however , that there is a small and marginally significant difference between the values of @xmath3 derived by using only the wmap data or using the wmap data in combination with other observations ( spergel et al .
2003 ) as shown here in figure 1 .
[ fig1 ] one can use @xmath17 as an input to bbn calculations and predict the primordial abundances of the light elements .
this yields relatively precise abundances ( cyburt 2003 ) : @xmath18 because there are no measurements of 3 at very low metallicity ( i.e. , significantly below solar ) at this time , a higher burden is placed on the remaining three elements .
the d / h prediction is in excellent agreement with the average of the five best determined quasar absorption system abundances ( burles & tytler 1998a , b ; omeara 2001 ; pettini & bowen 2001 ; kirkman 2003 ) which give @xmath19 ( see figure 1 ) .
in contrast , the 7 prediction is rather high .
the results of ryan , norris , & beers ( 1999 ) , give 7/h = @xmath20 which is a factor of 3 below the wmap value , and almost a factor of 2 below even when systematics are stretched to maximize the 7 abundance ( ryan 2000 ) .
even a recent study ( bonifacio 2002 ) with temperatures based on h@xmath21 lines ( considered to give systematically high temperatures ) yields 7/h = @xmath22 .
recent attempts ( coc 2004 , cyburt , fields , & olive 2004 ) to ascribe this discrepancy to systematic uncertainties in the relevant nuclear rates for 7 show that this is unlikely .
of course , some of the discrepancy may be due to li depletion in the stellar atmosphere ( vauclair & charbonnel 1998 , pinsonneault 1999 , 2002 ) .
finally , the predicted 4 abundance is also rather high when compared to the large body of data on 4 in extragalactic hii regions . given the concordance of the wmap determinations with the average of the d / h determinations , is the value of @xmath3 secure , and thus , should the topic be considered closed ?
there are a number of reasons that we feel that it is important to resolve the differences indicated in figure 1 .
first , it is important to remember that individual cosmological parameters can not be derived independently from microwave background measurements ; the microwave background measurements constrain combinations of cosmological parameters .
thus , independent constraints on individual parameters ( or , in this case , the combination of just two parameters - @xmath23 ) continue to be very important .
second , the concordance of the primordial abundances of the light elements has long been regarded as a triumph of modern cosmology , and thus , regaining concordance remains an important goal . finally , there are some aspects of the @xmath3 determinations that bear further inspection .
one example is the previously mentioned small difference between the value derived from wmap alone and that derived from combining in other observations .
another example , also shown in figure 1 , is the fact that the error bars on the individual d / h @xmath3 determinations , in many cases , are not overlapping , and averaging the five best determined absorption systems may not be the correct way to treat the data
. the dispersion in the d / h data may be a result of chemical evolution or underlying systematic uncertainties in the observations ( fields 2001 ) .
thus , we feel that the problem of the concordance of the various determinations of @xmath3 is an important one , and here we will concentrate on the determination of the value of the primordial helium abundance . unlike the other light element abundances , in order to be a useful cosmological constraint , 4 needs to be measured with a precision at the few percent level .
thus , the determination of the 4 abundance has become limited by systematic uncertainties . to date
, the most useful 4 abundance determinations are made by observing helium emission lines in hii regions of metal - poor dwarf galaxies .
izotov & thuan ( 1998 , hereafter it98 ) assembled a sample of 45 low metallicity hii regions , observed and analyzed in a uniform manner , and derived a value of @xmath24 @xmath25 0.244 @xmath26 0.002 and 0.245 @xmath26 0.001 ( with regressions against o / h and n / h respectively ) .
this value is significantly higher than the value of @xmath24 @xmath25 0.228 @xmath26 0.005 derived by pagel ( 1992 , hereafter pste ) using nearly identical analysis techniques .
peimbert , peimbert , & ruiz ( 2000 , hereafter ppr00 ) have derived a very accurate helium abundance for the hii region ngc 346 in the small magellanic cloud , and from this they infer a value of @xmath24 @xmath25 0.2345 @xmath26 0.0026 .
ppr00 take a fundamentally different approach to determining the electron temperature ( compared to pste and it98 ) .
specifically , they use the he i emission lines to solve for the electron temperature ( resulting in a lower temperature than indicated by the [ o iii ] lines , in line with predictions from photoionization models ) and include the effects of estimating the amplitude of temperature fluctuations .
thus , these two different results depend not only on different observations , but also on differences in the analyses of the observations . recently ,
luridiana et al .
( 2003 , hereafter lppc ) have analyzed spectra of five metal poor hii regions and , after considering the effects of additional physical processes ( e.g. , collisional excitation of the balmer lines ) , have produced a higher ( compared to ppr00 ) determination of y@xmath27 .
as shown in figure 1 , this higher value is still significantly lower than the value derived by it98 or from the wmap observations .
we are left with the conundrum in that the studies that appear to have the higher quality observations and the more complete physical analysis ( ppr00 and lppc ) arrive at a answer which is further from concordance with wmap results ( relative to it98 ) .
one approach to the problem would be to assume that the cmb measurements are providing us with the correct value of @xmath7 , and then bbn allows us to calculate the corresponding value of y@xmath2 ( cyburt , fields , & olive 2002 , 2003 ) . assuming that this is correct
, we may be better able to understand the systematic effects in the determination of y@xmath2 from nebular spectroscopy . in an earlier paper , ( olive & skillman 2001 , hereafter os ) we critically examined the `` self - consistent '' approach of determining the 4 abundance used by it98 , and concluded that uncertainties were systematically underestimated , especially with regard to the treatment of underlying stellar absorption ( see also skillman , terlevich , & terlevich 1998 ) .
this effect alone could be sufficient to explain the discrepancy between the it98 result and the wmap result , but it is unlikely to resolve the difference between the it98 and ppr00 studies .
additionally , benjamin , skillman , & smits ( 2002 ) have presented new calculations of hei radiative transfer effects and have shown that some of the fitting formulae introduced by it98 had significant errors . in this paper , we will attempt to better quantify the true uncertainties in the individual helium determinations in extragalactic hii regions and hopefully produce a more secure estimate of y@xmath2 .
we present an update of our minimization procedure , adding in the ability to solve self - consistently for the electron temperature .
the new calculations of radiative transfer effects are also incorporated into this new code .
the goal of this paper is to explore different analysis methodologies and to promote particular analysis techniques in order to carefully assess the true systematic uncertainty in 4 abundance determinations .
in our previous work ( os ) , in order to make calculations directly comparable to those of it98 , we used the h i emissivities calculated by hummer & storey ( 1987 ) , the he i emissivities calculated by smits ( 1996 ) , the collisional rates of sawey & berrington ( 1993 ) , the resulting collisional corrections calculated by kingdon & ferland ( 1995 ) , and the he i radiative transfer model of it98 based on robbins ( 1968 ) . here ,
these are replaced and combined by new calculations from benjamin , skillman , & smits ( 1999 ; 2002 ) . while the emissivity and collisional corrections presented by benjamin , skillman , & smits ( 1999 ) did not result in large changes relative to previous values ,
the new radiative transfer model presented in benjamin , skillman , & smits ( 2002 ) is substantially different from that of it98 .
as detailed in our previous paper , we also allow for the possibility of underlying stellar absorption .
the details used in computing the helium abundance are given in the appendix .
the self consistent method we employ makes use of six he emission lines : @xmath283389 , @xmath284026 , @xmath284471 , @xmath285876 , @xmath286678 , @xmath287026 .
the ratio of each he i emission line intensity to h@xmath29 is compared to the theoretical ratio and corrected for the effects of collisional excitation , fluorescence , and underlying he i absorption . from these line ratios ,
we need to determine three physical parameters , the density , @xmath30 , the optical depth , @xmath31 , and the equivalent width for underlying helium absorption , @xmath32 . in os
, we had assumed an input temperature ( typically derived from [ o iii ] emission lines ) .
as pointed out in ppr00 , the temperature dependences of the he i emissivities are exponential and can be both positive and negative , so that the relative he i lines can provide a strong temperature diagnostic ( which will be more appropriate to the he@xmath33 zone )
. thus , our new procedure allows us to solve for the temperature as we do for the other physical parameters .
as described in detail in os , we use the derived he abundances @xmath34 to compute the average helium abundance , @xmath35 , |y = _ y^+ ( ) ( ) ^2 / _ 1 ( ) ^2 this is a weighted average , where the uncertainty @xmath36 is found by propagating the uncertainties in the observational quantities stemming from the observed line fluxes ( which already contains the uncertainty due to c(h@xmath29 ) ) , and the equivalent widths . from @xmath37
, we can define a @xmath38 as the deviation of the individual he abundances @xmath34 from the average , ^2 = _ ( y^+ ( ) - |y)^2 ( ) ^2 [ chi4 ] we then minimize @xmath38 , to determine @xmath39 , and @xmath31 and the temperature , @xmath40 .
uncertainties in the output parameters are determined by varying the outputs until @xmath41 .
the real challenge for the new code is determining the correct combination of temperature and density .
while it is true , in principle , that the he i lines can provide a strong temperature diagnostic , in practice , this requires very precise emission line fluxes to pin down the temperature to a narrow range .
the difficulty comes from the fact that the three strong lines which form the backbone of the solution ( @xmath285876 , @xmath284471 , and @xmath286678 ) all have the same sense of temperature and density dependence , i.e. , f_hei = a t^b - c n_e [ tndep ] where a , b , and c are positive constants .
because the three main lines ( and @xmath284026 ) all behave in similar ways , there is a natural tendency towards solutions combining either high temperatures and low densities or low temperatures and high densities .
thus , if temperatures are systematically overestimated ( as is probably true if one assumes the temperature derived from the [ o iii ] lines ) , then the solutions are forced systematically to low values of density .
the fact that two of the lines ( @xmath283889 and @xmath287065 ) have negative exponents in the temperature term does not provide as much leverage on the temperature because these two lines are also strongly affected by optical depth and collisional excitation effects .
finally , for a more robust determination of the uncertainties in the physical parameters and ultimately in @xmath37 , we perform a monte - carlo simulation of the data . starting with the observational inputs and their stated uncertainties , we have generated a data set which is gaussian distributed for the 6 observed he emission lines . from each distribution
, we randomly select a set of input values and run the @xmath38 minimization .
the selection of data is repeated 1000 times .
we thus obtain a distribution of solutions for @xmath39 , @xmath31 , and @xmath40 and we compare the mean and dispersion of these distributions with the initial solution for these quantities . in simple photoionization models ,
hii regions divide into two different ionization zones where the cooling is dominated by either o@xmath33 or o@xmath42 .
typical abundance analyses usually take this into account and derive different electron temperatures for the two different zones ( and use the different electron temperatures for deriving ionic abundances , see garnett 1992 ) . in our new code we have the option of using two different temperatures in determining the he / h abundances ( as done in ppr00 and lppc , but not in it98 ) .
we have experimented with this option , and , for this study have decided to concentrate on single temperature models .
given the number of physical parameters that one is solving for and the limited number of emission lines measured with high precision , satisfactory results were only achieved in cases where a relationship was assumed between the temperatures in the low and high ionization zones ( as is usually done when calculating oxygen abundances ) .
since we are advocating non - parametric solutions , we felt it best to stick to the simpler case of single temperature solutions .
the study of ngc 346 by ppr00 emphasized the advantages ( relative to studying more distant blue compact galaxies ) of ( 1 ) no underlying absorption correction for the helium lines , ( 2 ) the opportunity to observe multiple lines of sight , and ( 3 ) due to a lower electron temperature , the effects of collisional excitation on the permitted lines are small .
they also identified the disadvantage of a larger correction for chemical evolution for the study of the primordial helium abundance .
this study introduced the technique of deriving the electron temperature directly from the helium lines .
given the high quality of observations presented there , it is a good starting point to compare our own analysis techniques . here
we will concentrate on their region `` a '' , which provides the highest quality spectrum . the relevant he i emission lines , equivalent widths , and errors , taken from ppr00 , are presented in table 1 .
here we have taken the reddening corrected he i emission line fluxes and their associated errors directly from ppr00 . in analyzing these data ,
our main goal is to see if our analysis is consistent with that of ppr00 . in os we made specific suggestions about treating the errors in the reddening correction , and we thought it would be interesting to revisit that question here .
thus , we carry out parallel analyses on a second set of data .
the second set of data listed in table 1 ( labeled `` re - analyzed '' ) have been derived from the original flux ratio observations of ppr00 and then de - reddened using the prescription for solving for reddening and underlying h absorption as described in os . for this second set , we have adopted the average extinction law of cardelli , clayton , & mathis ( 1989 ; hereafter ccm ) .
this is the most commonly used astrophysical extinction law , but it differs slightly from that of the normal extinction law of whitford ( 1958 ) used by ppr00 . since the extinction law is normally assumed without any associated errors , we were interested in the effect of simply using a different ( although generally acceptable ) extinction law . figure 2 shows the differences between the ccm extinction law and that used by ppr00 .
we also assign errors of 10% to the ew measurements ( as some value of error is needed for our minimization technique , and none is reported in ppr00 ) .
the error in the ew is a conservative one , but the final results do not change significantly if the errors are varied between 10% and 2% .
[ fig2 ] in what follows we refer to a simple minimization based on the input data as a `` direct '' analysis ( labeled `` d '' in the tables ) , and an analysis over 1000 representations which are consistent with the errors in the input data as a `` monte carlo ( `` mc '' ) analysis . '' in our first analysis , we assume a temperature of 11,920 k , the average he i temperature found by ppr00 . holding this temperature fixed , and assuming no underlying absorption , we derive values of he@xmath43/h@xmath43 @xmath25 0.0795 @xmath26 0.0005 and n@xmath44 @xmath25 164@xmath45 which compare quite well with 0.0793 @xmath26 0.0006 and 146 @xmath26 50 reported by ppr00 ( see table 2 ) .
dropping the assumption of zero underlying absorption gives a very similar result in the direct analysis , supporting ppr00 s claim that there is no underlying absorption ( see table 2 ) . using a monte carlo analysis
, we derive he@xmath43/h@xmath43 @xmath25 0.0801 @xmath26 0.0011 and n@xmath44 @xmath25 159 @xmath26 69 , where we see that the central values are only slightly changed , but the error bars on he@xmath43/h@xmath43 increase by about a factor of two ( see table 2 ) .
the close similarity of the two results indicates that our minimization techniques produce similar results with high quality data and similar assumptions .
although the ppr00 could not use the radiative transfer results of benjamin skillman , & smits ( 2002 ) , the spectrum for ngc 346 indicates very low values of @xmath31(3889 ) , so good agreement is expected . in figure
[ fig3 ] , we show the results of our analysis for the case where we have fixed the temperature and allow for positive values of the underlying absorption coefficient , @xmath32 . for this model
we have used a two zone temperature model and assumed that the temperatures in those two zones are identical to the values derived by ppr00 .
thus , the main difference between these calculations and those of ppr00 are dropping the assumptions of zero underlying absorption and zero optical depth in the triplet helium lines .
note the very good agreement between the our results ( shown by the position of the solid circle ) and those of ppr00 ( shown by the intersection of the vertical and horizontal lines ) .
the small dots show the possible variations in solutions due to the monte carlo generation of data .
the solid square shows the position of the mean value of the 1000 sets of generated data .
while the mean value is consistent with both our direct solution and ppr00 , the uncertainty in @xmath46 is nearly doubled .
[ fig3 ] when we allow the code to solve for the electron temperature , we get a slightly higher value of 12,510 k ( see table 2 ) .
this temperature lies between the values of t(o iii ) ( 13,070 k ) and t(o ii ) ( 11,810 k ) , which would normally be expected ( peimbert , peimbert , & luridiana 2002 ) .
this new solution drops the @xmath38 value from 3.0 to 2.4 , which an _ ftest _ shows is significant only at the 45% confidence level .
the solution with the higher temperature also favors a lower density and a small amount of underlying absorption ( although consistent with zero ) with a resulting he@xmath43/h@xmath43 ratio which is 2% higher than the ppr00 value .
note that , although 2% seems like a small difference , the value obtained from the direct analysis allowing for underlying absorption and deriving the electron temperature from the helium line ratios is 3@xmath47 higher than the ppr00 value .
the mc value is slightly higher still , but this could be mainly due to the fact that the code requires the underlying absorption to be positive , so that , when the value is close to zero , there is a bias in the mean value to be larger .
the results of the mc analysis are shown in figure 4 .
in addition to the value of @xmath46 being larger , the uncertainty in @xmath46 is now over a factor of 2 larger than quoted in ppr00 .
[ fig4 ] as a check on the assumption of positive underlying he absorption , we then ran the direct and monte carlo codes allowing both positive and negative values of underlying he absorption .
the result of the direct analysis is almost identical to that obtained with the absorption held positive ( see table 2 ) . in the table , ` solved + ' refers to the restriction that only positive values of the parameter were permitted , ` solved free ' indicates that both positive and negative values are allowed , and ` fixed ' means that the parameter was not solved for , but rather an input value was used and held fixed .
thus , the detection of a small amount of underlying he absorption is not due to a bias from the code only allowing positive values .
the higher values of the he abundance ( roughly 2% higher for the direct method and 3% higher for the monte carlo method ) are probably due to a small , but real , amount of underlying he absorption .
finally we ran the direct and monte carlo codes allowing both positive and negative values of all three physical parameters : density , underlying absorption , and optical depth while still solving for electron temperature ( last lines in table 2 ) .
note that negative values of density and optical depth are non - physical .
now the favored value for the underlying absorption is almost very close to zero .
however , it is very interesting that the free solution has favored negative ( non - physical ) values of the optical depth . in this solution , because of the negative value for the optical depth , the density has jumped up to 427 ( mainly because of the inverse dependences on density and optical depth for the @xmath287065 line ) , resulting in a he@xmath43/h@xmath43 value which is even lower than the ppr00 value .
the favored negative values of the optical depth indicate that either our prescription for the treatment of radiative transfer on the helium line strengths is not appropriate for ngc 346 or that the value for @xmath283889 ( the main driver of the optical depth solution ) is somewhat in error ( or both ) . from this exercise
we conclude that the error bars on the derivation of individual hii region parameters are underestimated for several reasons .
first , as concluded in os , unless a monte carlo analysis of the data is performed , the resulting errors of a single minimization exercise are likely to be underestimates . in the case for which we solve for temperature and absorption , our mc uncertainty is over a factor of 2 larger than that quoted by ppr00 .
second , by making different reasonable assumptions , different central values of the helium abundance are obtained ( and these values differ by several standard deviations if the errors are calculated in the direct manner ) . for ngc 346a
these differences are only a few percent because the quality of the data constrains the results to a relatively narrow range . however , typical hii region spectra used for deriving helium abundances are not even close to the quality of the ngc 346a spectrum from ppr00 .
third , by allowing the code to examine non - physical values of certain parameters there is an indication that perhaps the prescription for treating the radiative transfer is not completely appropriate .
this implies a systematic uncertainty which has not been accounted for in the errors analysis . in table 2
we also report the results of re - analyzing the ngc 346a spectrum using our favored prescriptions for the solution for reddening and underlying hydrogen absorption and the appropriate error propagation ( we label these entries `` re - an '' ) .
our analysis finds a similar value for c(h@xmath29 ) @xmath25 0.174 @xmath26 0.008 ( compared to 0.15 @xmath26 0.01 for ppr00 ) , and no evidence of underlying hydrogen absorption ( formally @xmath480.6 @xmath26 0.5 ) .
as seen in table 1 , there are small differences between the ppr00 relative line fluxes and those resulting from our re - analysis , which are mostly due to the differences in the adopted reddening curve ( and some small differences might be expected simply due to round - off errors ) .
as expected , only small differences are found when we analyze these helium ratios ( as shown in table 2 ) .
there is , however , one interesting difference .
when we solve for all of the parameters including temperature , the temperature is higher ( although not as high as the [ o iii ] temperature ) with the result that the density falls and the helium abundance rises by 2% relative to that found in the similar analysis with the ppr00 data .
both the original data and the re - analyzed data favor negative values of the optical depth .
it appears that even with excellent quality spectra , one is never far from unsatisfactory results , and that one needs exquisite quality data to insure that the best minimization values favor physically meaningful values .
a detailed inspection of table 2 indicates that the values derived for the physical parameters and the helium abundance depend on the assumptions that are made in the analysis . while the variations are of order 2
3% for most of the helium abundance determinations , this is very significant for determination of @xmath3 , and of order the discrepancies found in the literature .
for example , the ppr00 value of @xmath49 corresponds to a 4 mass fraction of 0.2403 , from which they inferred a primordial abundance of 0.2345 which in turn corresponds to @xmath50 .
our value ( using the re - analyzed data ) of @xmath51 corresponds to a 4 mass fraction of 0.2483 . using the same slope ( dy / dz ) yields a primordial abundance of 0.2425 which in turn corresponds to @xmath52 .
thus , one sees a difference of nearly a factor of two in the derived value of @xmath13 simply from using a different set of assumptions .
figure 5 underscores this point .
here we have plotted the he@xmath43/h@xmath43 ratios and errors derived for each line as a result of our direct method which solves simultaneously for helium abundance , temperature , density , underlying absorption and optical depth under the assumption that all are positive .
the solid and dotted lines across the bottom indicate the solution from ppr00 .
the significant difference is due primarily to assumptions made in analyzing the observations . in other words ,
an independent analysis , using reasonable assumptions , produces a value of he@xmath43/h@xmath43 @xmath25 0.0828 @xmath26 0.0008 , which is over 4% ( or 6@xmath47 ) higher than the ppr00 value .
we are not claiming here that this solution is superior to that produced by ppr00 , only that it is a completely satisfactory solution but different .
these systematic differences need to be accounted for in estimating errors in the final helium abundance .
[ fig5 ] what is the `` best '' value for he@xmath43/h@xmath43 in ngc 346a ?
we favor an estimation which is as close to a non - parametric solution as possible .
thus , we feel that solving for electron temperature from the helium lines and allowing only for non - negative values of underlying absorption is preferred ( as well as the other parameters ) is allowed to be negative . ] .
this is not a truly non - parametric solution as we are still dependent on assumptions concerning the relative importance of underlying helium absorption and optical depth on the different helium lines . for our analysis of the ppr00 data for ngc 346a
, this means that we favor 0.0828 @xmath26 0.0008 , which , compared with the result of ppr00 ( 0.0793 @xmath26 0.0006 ) appears extremely high , and would even appear to rule out their value . at this point
we are not ready to interpret the results in that way .
our main conclusion is that the systematic error of analysis technique is clearly dominant , and that this error , when reported , has been grossly underestimated in the past .
in is now useful to revisit selected targets from it98 to better assess the robustness of their results . many of the targets in it98 are not of the quality necessary for high accuracy helium abundance determinations .
for example , the majority are not of sufficient quality to detect the @xmath284026 line which provides a strong constraint on the underlying he i absorption .
thus , we have chosen to re - analyze only a select few targets which meet three criteria .
first we select only the it98 targets with ew(h@xmath29 ) @xmath53 200 .
this cut is made in order to minimize the correction due to underlying absorption as well as to improve the signal - to - noise ratio in faint lines such as @xmath284026 ( the ppr00 value of ew(h@xmath29 ) for ngc 346a is 250 ) .
the value of 200 is identified by izotov & thuan ( 2004 ; it04 ) as a desirable cut - off . this selection screens out 35 of the 45 it98 spectra . next
we select targets with oxygen abundances less than 20% of the solar value ( 4.6 @xmath54 10@xmath55 ; asplund 2004 ) .
this cut minimizes the uncertainty induced by our lack of knowledge of the functional form of the relationship between 4 and o / h ( c.f . , olive , steigman , & walker 1991 ; pilyugin 1993 ; carigi et al .
1995 ; bradamante , matteucci , & dercole 1998 ; fields & olive 1998 ) .
this selection screens out 17 of the 45 it98 spectra .
finally we select out targets with radial velocities inside of the ranges 728 @xmath26 100 km s@xmath56 and 1032 @xmath26 100 km s@xmath56 ( to avoid contamination of he i @xmath285876 by na i absorption ; cf.dinerstein & shields 1986 ) .
this selection screens out 10 of the 45 it98 spectra .
clearly several of the targets fail more than one of the screens . in the end , we find there are seven targets which pass all three screens , and hereafter we will refer to this subsample as the `` high quality '' sample .
we proceed in the same manner as our analysis of ngc 346a , i.e. , we analyze the reported reddening corrected he i line strengths directly and also re - analyze the data starting from the raw flux ratio measurements . for our analysis
, we require the ew of the he i lines . in most cases these
are not reported , so we have estimated the strength of the continuum ( from published spectra ) and used the ew(h@xmath29 ) to derive these ews .
this is , of course , less than desirable , and , at first , we hesitated to carry this out .
however , since the primary effect is on the value of the underlying absorption , and since it is the ratio of the lines strengths that matters here , it turns out that the answers have a relatively low sensitivity to these estimated values .
we assigned uncertainties to the estimated ew of 10% .
in fact , changes of these values of up to 20% , in most cases , have little effect on the derived helium abundances .
in cases where we have re - analyzed the data by deriving our own reddening and underlying absorption corrections , we have also estimated the ew of the balmer lines in the same way . a spectrum for sbs 0335 - 052 was reported in it98 ( and therefore we do not need to make any adjustments to be consistent with the it98 analysis as will be the case for other targets ) . when we re - analyze the it98 spectrum for sbs 0335 - 052 using our favored prescriptions for the solution for reddening and underlying hydrogen absorption , we obtain a value of c(h@xmath29 ) @xmath25 0.121 @xmath26 0.014 , in good agreement with it98 value of 0.13 , and we find a higher value for the underlying hydrogen absorption of 1.3 @xmath26 0.9 ( to be compared with it98 value of 0.4 ) .
the resulting helium line strengths are reported in table 3 .
there are only small differences between the it98 values and the values resulting from our re - analysis .
the one noticeable difference is for @xmath283889 , where our value is 9% higher due primarily to the correction for underlying h absorption .
table 4 shows the the results of the full set of analyses of the it98 spectrum for sbs 0335 - 052 . our constrained analyses with the assumption of no
underlying he absorption find much higher values of the optical depth in the in helium lines ( @xmath57 5 vs. 1.7 ) and this results in lower densities but almost no difference in the value of the he abundance .
however , when we allow for underlying absorption , there is significant evidence for this and this results in higher helium abundances .
the @xmath38 drops significantly with the allowance for underlying absorption and the helium abundance rises significantly .
given the large variation in helium abundance values for the different degrees of constraint , it would appear that this spectrum of sbs0335 - 052 is not suitable for constraining the primordial helium abundance . by solving for the temperature , significantly lower temperatures ( with significantly larger uncertainties ) and higher densities
are found , with the resultant much lower he abundances .
the derived temperatures are so much lower than the [ o iii ] temperatures such that these solutions are most likely not to be believed .
the very high values of @xmath31(3889 ) are robust , and it is unlikely that our prescription for correcting for optical depth effects is valid in this regime .
although sbs 0335 - 052 is an ideal target from the viewpoint of its exceptionally low o / h abundance , the very high value of optical depth means that this is probably a less than suitable object ( at least in the highest surface brightness knot ) for determining helium abundance .
this is reflected in the fact that the error in he@xmath43/h@xmath43 in sbs 0335 - 052 is the largest ( in both absolute terms and relative terms ) of the seven objects from it98 that we have re - analyzed . the spectrum for ngc 2363a analyzed in it98 was first reported in izotov , thuan , & lipovetsky ( 1997 ; itl97 ) . in itl97 , an electron temperature of 15,100 @xmath26 100 k , an oxygen abundance of 0.78 @xmath26 0.01 @xmath54 10@xmath55 , and
a helium abundance of y @xmath25 0.245 @xmath26 0.005 were reported .
it98 re - analyzed this spectrum , and , primarily because of a difference in the estimated electron temperature , the oxygen abundance changed to 0.71 @xmath26 0.01 @xmath54 10@xmath55 ( a 7 @xmath47 decrease ) , and the helium abundance changed to y @xmath25 0.2456 @xmath26 0.0008 ( an 84% decrease in the uncertainty ) .
that such different values can be reported from the same spectrum by the same researchers gives one an impression of the size of the systematic errors . for our comparison analysis , we will assume an [ o iii ] electron temperature of 15,800 @xmath26 100 k , which would correspond to the scale used in it98 .
when we re - analyze the it98 spectrum for ngc 2363a using our favored prescriptions for the solution for reddening and underlying hydrogen absorption , we obtain a value of c(h@xmath29 ) @xmath25 0.117 @xmath26 0.003 , in good agreement with the itl97 value of 0.11 , and we find a slightly lower value for the underlying hydrogen absorption of 1.4 @xmath26 0.3 ( to be compared with itl97 value of 1.8 ) .
the resulting helium line strengths are reported in table 3 .
there are only small differences between the it98 values and the values resulting from our re - analysis ( all consistent with round - off errors ) .
table 5 shows the results of our analyses of the hei line ratios . in all analyses where their presence is allowed ,
we find evidence for significant amounts of underlying absorption and optical depth effects ( which were both assumed to be absent in the it98 analysis ) . holding the electron temperature fixed at the [ o iii ] temperature and assuming no underlying absorption reveals values of @xmath31(3889 ) @xmath57 2 .
when underlying absorption is allowed for , the solutions favor significant amounts ( the @xmath58 drops by a factor of 6 ) . finally , solving for the electron temperature yields a lower temperature .
note that the value of the derived helium abundance varies by 8% depending on the set of assumptions used in the derivation .
perhaps the most remarkable thing about this re - analysis is that our value derived by allowing for underlying absorption and optical depth effects and solving for the temperature produces a value nearly identical to the it98 value although our uncertainty ( from the mc analysis ) is 4.6 times larger .
the spectrum for sbs 0940 + 544n analyzed in it98 was reported in itl97 . in itl97 , an electron temperature of 19,000 @xmath26 300 k ,
an oxygen abundance of 0.30 @xmath26 0.01 @xmath54 10@xmath55 , and a helium abundance of y @xmath25 0.246 @xmath26 0.008 were reported .
it98 re - analyzed this spectrum , and , primarily because of a difference in the estimated electron temperature , the oxygen abundance changed to 0.27 @xmath26 0.01 @xmath54 10@xmath55 , and the helium abundance changed to y @xmath25 0.2500 @xmath26 0.0047 . for our comparison analysis , we will assume an [ o iii ] electron temperature of 20,200 @xmath26 300 k , which would correspond to the scale used in it98 .
when we re - analyze the it98 spectrum for sbs 0940 + 544n using our favored prescriptions for the solution for reddening and underlying hydrogen absorption , we obtain a value of c(h@xmath29 ) @xmath25 0.048 @xmath26 0.023 , in good agreement with the itl97 value of 0.05 , but , formally , we find a negative value ( though consistent with 0 ) for the underlying hydrogen absorption of @xmath481.4 @xmath26 1.4 to be compared with itl97 value of 0.6 .
this presents a problem concerning how to proceed with the analysis .
since the solution is formally consistent with no underlying absorption , we assume a value of 0 @xmath26 1.4 in our analyses .
the resulting helium line strengths are reported in table 3 .
there are only small differences between the it98 values and the values resulting from our re - analysis ( all consistent with round - off errors ) .
table 6 shows the results of our analyses of the hei line ratios .
our analyses are generally in agreement with it98 in that low values of underlying absorption , optical depth , and density are found .
however , perhaps the most interesting part of our re - analysis can be seen when non - physical values are allowed for .
specifically , negative values of underlying absorption and optical depth are favored .
these non - physical values indicate that there must be some problem with either the spectrum or our simple assumptions used when analyzing it . the spectrum for mrk 193 analyzed in it98 was first reported in itl94 . in itl94 , an electron temperature of 15,500 ,
an oxygen abundance of 0.70 @xmath26 0.02 @xmath54 10@xmath55 , and a helium abundance of y @xmath25 0.253 @xmath26 0.007 were reported .
it98 re - analyzed this spectrum , and , primarily because of a difference in the estimated electron temperature , the oxygen abundance changed to 0.64 @xmath26 0.01 @xmath54 10@xmath55 , and the helium abundance changed to y @xmath25 0.2478 @xmath26 0.0037 . for our comparison analysis
, we will assume an [ o iii ] electron temperature of 16,200 @xmath26 200 k , which would correspond to the scale used in it98 .
when we re - analyze the it98 spectrum for mrk 193 using our favored prescriptions for the solution for reddening and underlying hydrogen absorption , we obtain a value of c(h@xmath29 ) @xmath25 0.292 @xmath26 0.020 , in good agreement with the itl94 value of 0.260 , but , formally , we find a negative value for the underlying hydrogen absorption of @xmath481.7 @xmath26 1.0 ( to be compared with itl94 value of 1.9 ) . as in the case of sbs 0940 + 544n
, this again presents a problem concerning how to proceed with the analysis . in this case
, the solution is not formally consistent with zero underlying absorption , but we are forced to assume a value of 0 @xmath26 1.0 in our analyses .
the resulting helium line strengths are reported in table 3 .
there are only small differences between the it98 values and the values resulting from our re - analysis , except for the difference in i(@xmath283889 ) .
table 7 shows the results of our analyses of the hei line ratios .
our analyses are generally in agreement with it98 in that low values are found for underlying absorption and relatively high densities of density
. however , non - zero values of optical depth are favored , and strongly so in the re - analyzed data .
when the temperature is solved for , lower values are favored .
overall , once again the impression is that the assumptions of the analysis strongly affect the derived value of y. the spectrum for sbs 1159 + 545 analyzed in it98 was first reported in itl94 .
it98 re - analyzed this spectrum upon discovering errors made in the data reduction reported in itl94 . since the full spectrum and analysis for sbs 1159 + 545 was reported in it98 we do not need to make any adjustments for consistency .
when we re - analyze the it98 spectrum for sbs 1159 + 545 using our favored prescriptions for the solution for reddening and underlying hydrogen absorption , we obtain a value of c(h@xmath29 ) @xmath25 0.065 @xmath26 0.016 , in good agreement with the it98 value of 0.06 , and we find a value for the underlying hydrogen absorption of 0.2 @xmath26 1.1 ( smaller than but consistent with the it98 value of 0.6 ) .
the resulting helium line strengths are reported in table 3 , and , in general , there is very good agreement .
table 8 shows the results of several analyses of the it98 spectrum for sbs 1159 + 545 .
when we fix the temperature to the it98 value and assume no underlying he absorption , we find a value of the helium abundance consistent with that of it98 and also an optical depth of zero , consistent with it98 .
interestingly , when we re - analyze the spectrum , the helium line strengths change only slightly , but even in the constrained analysis , the favored density has fallen , the optical depth is non - zero , and the favored helium abundance has risen by 2% ( while the @xmath38 is almost identical ) . in the constrained solutions , there is never any evidence of underlying he absorption in the sbs 1159 + 545 spectrum .
the less constrained solutions show why . there ,
negative values of underlying absorption are strongly favored .
indeed , in the unconstrained solutions , negative values for both underlying he absorption and optical depth are favored , resulting in very high values of density , relatively low values of temperature , and thus , unrealistically low values of the he@xmath43/h@xmath43 ratio .
although the sbs 1159 + 545 spectrum is fairly high signal / noise , it would appear that a robust measurement of the helium abundance may be suspect for this spectrum .
the spectrum for haro 29 analyzed in it98 was reported in itl97 . in itl97 , an electron temperature of 15,400 @xmath26 100 k ,
an oxygen abundance of 0.65 @xmath26 0.01 @xmath54 10@xmath55 , and a helium abundance of y @xmath25 0.246 @xmath26 0.005 were reported .
it98 re - analyzed this spectrum , and , primarily because of a difference in the estimated electron temperature , the oxygen abundance changed to 0.59 @xmath26 0.01 @xmath54 10@xmath55 , and the helium abundance changed to y @xmath25 0.2509 @xmath26 0.0012 . for our comparison analysis , we will assume an [ o iii ] electron temperature of 16,180 @xmath26 100 k , which would correspond to the scale used in it98 .
when we re - analyze the it98 spectrum for haro 29 using our favored prescriptions for the solution for reddening and underlying hydrogen absorption , we obtain a value of c(h@xmath29 ) @xmath25 0.001 @xmath26 0.004 , in good agreement with itl97 value of 0.0 , and we find a value for the underlying hydrogen absorption of 2.7 @xmath26 0.3 ( to be compared with it98 value of 2.5 ) .
the resulting helium line strengths are reported in table 3 , and , in general , there is very good agreement .
table 9 shows the the results of several analyses of the it98 spectrum for haro 29 .
our constrained analyses with the assumption of no underlying he absorption find evidence for non - zero optical depth in the in helium lines and this results in higher densities and thus , lower values of the he abundance .
however , when we allow for underlying absorption , there is very significant evidence ( the @xmath38 drops roughly a factor of 6 with the allowance for underlying absorption ) and this results in higher helium abundances .
when we solve for the temperature , the value is very close to that obtained from the [ o iii ] lines , which is reassuring , and the solutions do not change significantly when we allow for positive or negative values for the underlying absorption .
however , for our least constrained solutions , we find the very discouraging result that very different solutions are found with higher values of the optical depth , higher values of the underlying he absorption , negative densities , higher electron temperatures and very high helium abundances .
the slightly lower @xmath38 for these solutions are probably only a warning of how loosely constrained even the best solutions are .
lppc03 have analyzed the it98 spectrum of haro 29 and derive a value of he@xmath43/h@xmath43 @xmath25 0.078 .
there are two large differences between their treatment and the one presented here .
first , a much larger value of the reddening is adopted and second , underlying he absorption is assumed to be zero . in all of our solutions , we find this to be a poor assumption .
a spectrum for sbs 1420 + 544 was reported in it98 ( and therefore we do not need to make any adjustments to be consistent with the it98 analysis as for most of the targets ) .
when we re - analyze the it98 spectrum for sbs 1420 + 544 using our favored prescriptions for the solution for reddening and underlying hydrogen absorption , we obtain a value of c(h@xmath29 ) @xmath25 0.167 @xmath26 0.013 , in good agreement with it98 value of 0.16 , and we find a value for the underlying hydrogen absorption of 0.4 @xmath26 0.9 ( to be compared with it98 value of 0.0 ) .
the resulting helium line strengths are reported in table 3 , and , in general , there is very good agreement with the it98 values .
table 10 shows the the results of several analyses of the it98 spectrum for sbs 1420 + 544 .
sbs 1420 + 544 is interesting in that it is one of the few hii regions for which it98 find definite signs of significant optical depth . our constrained solutions with underlying absorption fixed at zero find higher values for the optical depth , and this results in larger values of the density and lower values of the he@xmath43/h@xmath43 ratio .
when we allow for underlying he absorption , there is only weak or no evidence of this .
however , when we allow for the temperature to be solved for , significantly larger temperatures result , giving lower densities and higher helium abundances . using our re - analyzed spectrum and solving for temperature , the resulting helium abundance is significantly higher ( by 4% ) than it98 . given the evidence for the large optical depth in the helium lines , and the large uncertainty in our prescription for correcting for this , it would seem that again , a robust measurement of the helium abundance is difficult for this spectrum . a summary of our re - analysis of the it98 spectra is given in table 11 .
several differences between the results of the it98 analysis and our re - analysis are immediately apparent .
first , the physical parameters t , n , @xmath31(3889 ) , and he(abs ) are all significantly less well constrained in our re - analysis . in some cases , this is because some of these values were assumed to be zero with no associated errors , and in other cases we believe that the errors were underestimated . the larger uncertainties in the physical parameters naturally translate into larger uncertainties in the helium abundances .
table 11 shows a larger dispersion in the re - analyzed he abundances , with a slight bias toward higher values .
note that in the discussions of the individual objects we raised reservations concerning the suitability of several of these spectra for deriving very high accuracy helium abundances . specifically , sbs 0940 + 544n and mrk 193
both favored negative ( non - physical ) values for underlying hydrogen absorption .
sbs 0335 - 052 and sbs 1420 + 544 both have relatively large values of @xmath31(3889 ) , and we doubt that the simple prescription for correcting for optical depth effects will be accurate in this regime . although we specifically selected objects with high emission line equivalent width , two objects ( ngc 2363 a and haro 29 ) show significant evidence for underlying helium absorption , and thus , are dependent on our assumption of the behavior of the absorption equivalent widths on a line - by - line basis . in sum ,
even the spectra that were selected specifically to be the highest quality provide only nominal constraints on the physical conditions in the hii regions , and thus , carry significant errors on their helium abundances compared to what is necessary to constrain the value of the primordial helium abundance .
recently , izotov & thuan ( 2004 ; it04 ) have presented new observations of 33 hii regions in order to provide a better metallicity baseline and therefore stronger constraints on @xmath59 .
they combine these new observations with those of it98 to conduct an analysis of 82 hii regions .
we have studied the new spectra presented by it04 to determine if there are new observations which satisfy our criteria for inclusion in a `` high quality '' data sample , and , in fact , there are seven targets that do satisfy the criteria enumerated at the beginning of 4 .
these seven targets , in order of increasing metallicity are : j 0519 + 0007 , hs 2236 + 1344 , hs 0122 + 0743 , hs 0837 + 4717 , cgcg 007 - 025 , hs 0134 + 3415 , and hs 1028 + 3843 .
we performed a complete analysis of these seven targets following the method used in 4 .
[ fig6 ] unfortunately , we discovered a serious problem with many of the observations reported in it04 , which render them questionable for use in our analyses .
when we solve for the reddening and underlying balmer absorption in the spectra from it04 , we often encounter significantly negative values of underlying balmer absorption .
this is found in 4 of the 7 cases identified as potential `` high quality '' sample targets .
figure 6 shows the solutions for these four sources .
the solutions for underlying absorption are found to be negative ranging from 1.4 @xmath47 for cgcg 007 - 025 to 4.1 @xmath47 for j 0519 + 0007 . as can be seen from figure 6 , in all cases it04 derived values of 0 for the underlying balmer ratios .
this may be due to limiting the solution to physically meaningful , i.e. , positive values .
as noted before , it was problematic that two of the targets in the it98 `` high quality '' data sample showed negative values for underlying balmer absorption ( sbs 0940 + 544n at the 1.0 @xmath47 level and mrk 193 at 1.7 @xmath47 ) , but the prevalence in the new sample can not be ignored .
in fact , roughly half of the new observations reported in it04 indicate zero underlying balmer absorption , indicating that this problem exists throughout the dataset . if we look at the object in the `` high quality '' data sample with the most negative value of underlying balmer absorption ( j 0519 + 0007 ) we find that the reddening corrected values for the blue balmer lines are all significantly too high relative to the theoretical values ( i.e. , h@xmath60/h@xmath29 , h@xmath61/h@xmath29 , h9/h@xmath29 , and h10/h@xmath29 are @xmath624.4 @xmath47 , @xmath625.1 @xmath47 , @xmath620.4 @xmath47 , and @xmath621.8 @xmath47 respectively ) .
in fact , the values for the blue balmer lines before reddening correction are all close to their theoretical values , indicating very low reddening to this target .
brief inspection of the other targets reveals a similar pattern , indicating that the problem may be solely with the h@xmath21 line ( in the sense that the h@xmath21 line is too strong ) and this could point to a problem with the sensitivity calibration in that wavelength range . on the other hand ,
j 0519 + 0007 is at a galactic latitude of @xmath4820@xmath63 and according to schlegel et al .
( 1998 ) should have an extinction of a@xmath64 @xmath65 0.4 mag .
if there is no problem with the sensitivity calibration , then the effect could be due to atmospheric differential refraction ( cf .
, filippenko 1982 ) even thought the object was observed at an airmass of 1.2 .
an alternative explanation could be that these spectra indicate the collisional enhancement of the h@xmath21 ( see discussion in skillman & kennicutt 1993 ) , but it seems unlikely that we would see it in so many sources and not find a strong correlation with electron temperature .
until this problem is solved , we have reservations concerning the inclusion of the spectra from it04 in our study of the primordial helium abundance . in table 12
we have reported our re - analysis of the `` high quality '' sample of it04 .
unfortunately , it04 do not report their derived values for @xmath30 and @xmath31(3889 ) so direct comparisons are not possible , but it appears that many of the characteristics from table 11 are also found in table 12 . in our analysis , several of the targets show relatively large values of @xmath31(3889 ) and significant evidence for underlying hei absorption .
given the different possible assumptions one can make regarding the treatment of the underlying absorption , optical depth , or temperature , one can justifiably wonder whether it is indeed possible to extract a reliable value of the primordial abundance of 4 from these observations .
it is our hope that by striving to free our solutions from as many parametric constraints as possible and by considering all identifiable sources of error that we can now produce a value of the primordial helium abundance with reasonable error estimates .
[ fig7 ] if we take our preferred solutions based on the re - analyzed data ( where we use the monte carlo technique and solve for temperature and restrict all parameters to take positive values ) for the seven `` high quality '' targets from it98 , then a regression of y@xmath43 on o / h yields an intercept of : y_p = 0.2491 0.0091 the corresponding value for the baryon - to - photon ratio is _ 10 = 6.64^+11.1_-3.82 easily in good agreement with the cmb determination of @xmath3 .
figure 7 shows a comparison between the original it98 values and our re - analysis for the seven `` high quality '' targets . it also compares the results of the above regression with the original value obtained by it98 for the full set of 45 targets .
the error on y@xmath2 derived above and shown in figure 7 is 4.5 times larger than derived by it98 .
one could argue that we should drop the two observations which yielded negative balmer absorption . in this case , our regression for five points would yield , @xmath66 .
while the intercept has decreased slightly , the uncertainty has increased making the 1 @xmath47 upper limit virtually identical .
obviously , some of the increase in the error is due to the decrease in the sample size from 45 to 7 , and some of the increase in the error is due to a larger error on the slope due to a smaller metallicity baseline .
however , a good part of the increase in the error is due to the larger error on the individual points .
additionally , some of the decrease in the intercept error gained by extending the metallicity baseline to higher metallicities is misleading because it is dependent on the questionable assumption of a perfectly linear relationship between y and o / h .
we can investigate the effect of limiting the sample size by including some of the it04 data points . if we include the three `` high quality '' targets from it04 which do not show evidence of negative underlying h i absorption , using all eight points which show no sign of negative h i absorption we obtain : y_p = 0.2502 0.0093 the corresponding value for the baryon - to - photon ratio is _
10 = 7.48^+12.9_-4.45 again , easily in good agreement with the cmb determination of @xmath3 .
finally , if we include all seven `` high quality '' targets from it04 together with all seven from it98 , we obtain : y_p = 0.2504 0.0073 the corresponding value for the baryon - to - photon ratio is _ 10 = 7.63^+9.1_-3.95 also easily in good agreement with the cmb determination of @xmath3 . note that although the error on y@xmath2 is declining with the increase in number of targets , it is not decreasing as the square root of the number of targets .
it appears that there is a floor in the uncertainty which is likely greater than 0.005 . in figure 1
we summarize our view of the present situation .
clearly the true uncertainty on the primordial helium abundance is much larger than indicated in previous studies .
we can not rule out values as high as the those achieved with our re - analysis of the it98 observations .
neither can we rule out values as low as those derived by ppr00 . thus , perhaps the best we can say is that the primordial 4 abundance lies within the `` conservative allowable range '' indicated in figure 1 .
0.232 y_p 0.258 this extended range in @xmath24 corresponds to 1.8 _ 10 18 thus , at present , there is no conflict with the cmb inferred value of @xmath24 .
one of the key results of the preceding exercise is that despite having apparently high quality data , the final 4 abundance for any given hii region will depend critically on assumptions concerning the appropriate electron temperature , the presence of underlying absorption , and the treatment of the optical depth effects .
many of the systems analyzed here show significant evidence for underlying absorption which leads to an increase in the derived 4 abundance .
additionally , many of the systems show significant evidence for non - zero values of @xmath31(3889 ) which relies on an untested formula for correction .
furthermore , as one can see from the results for individual hii regions shown in the tables , the spread in the derived 4 abundances can be far greater than the derived error for a given solution .
this is a clear sign that systematic uncertainties are dominating the error budget . while some of the systematics such as optical depth and underlying absorption move the 4 abundance in opposite directions , it makes no sense to suppose that these systematic uncertainties cancel each other out and therefore can be ignored as argued in it04 .
nor does it make sense to continuously average literature data to beat down the statistical errors , once again in complete neglect of the systematic uncertainties as attempted in dmitriev , flambaum & webb ( 2004 ) . in this context
, we stress that there is no evidence for a discrepancy between bbn predictions of 4 using the wmap baryon density and observational determinations of the 4 abundance as frequently reported in the literature .
inherent degeneracies between the parameters allow for very different solutions with equally acceptable @xmath38 .
this was the main point of our previous work ( os ) .
perhaps most troubling is the degeneracy in the parameter space identified here between temperature and density .
as we noted throughout our analysis , it98 have fixed the temperature to the oiii temperature which is systematically high ( peimbert , peimbert , & luridiana 2002 ) . however , as one can see from eqs .
( a2 ) , changes in temperature can be compensated by changes in the density allowing for the possibility of very different results due to various corrections .
this may explain the puzzling result that the dispersion in the it98 helium abundances is actually smaller than their formal errors . by assuming that the oiii temperatures are appropriate to the entire he@xmath33 emission zone , the densities have been biased to unrealistically low values resulting in higher values of the helium abundance . on the other hand , by assuming that underlying he absorption is negligible , the derived helium abundances have been biased to lower values .
the narrow range of derived helium abundances may be a result of these biases in the analysis .
there are other potentially important sources for systematic uncertainty such as temperature fluctuations ( steigman , viegas , & gruenwald 1997 ; sauer & jedamzik 2002 ) , ionization corrections ( ballantyne , ferland , & martin 2000 ; viegas , gruenwald , & steigman 2000 ; gruenwald , steigman & viegas 2002 ; sauer & jedamzik 2002 ) , and the collisional excitation of hydrogen emission lines ( skillman & kennicutt 1993 ; stasi ' nska & izotov 2001 , lppc ) .
however , it is our impression that these are small compared to the uncertainties that we have focused on here . from our analyses
, it appears that the high ew regions may be more prone to high values of @xmath31(3889 ) , which makes them less suitable candidates .
this could be a real problem for identifying new candidates for high precision helium abundances .
if derived helium abundances depend on our simple model for helium optical depth effects , we have a systematic uncertainty which , at present , is not testable . at this point , the prudent thing to do is to exclude all targets with relatively high values of @xmath31(3889 ) from any analysis sample ( as opposed to trying to correct for such effects ) .
we suggest that the following efforts will result in future progress on obtaining a more accurate value for the primordial helium abundance . on the observation side : \(1 ) significantly higher quality spectra are needed for almost all of the regions studied to date .
concentrating on those objects which appear to have higher ew(h@xmath29 ) and low @xmath31(3889 ) , in addition to low o / h abundance , will be important .
\(2 ) it is still important to search for new targets . obviously the targets which satisfy the above criteria are rare ( terlevich , skillman , & terlevich 1996 )
\(3 ) observations along multiple lines of sight to individual targets will allow testing of our analysis techniques as the physical conditions are changing while ( hopefully ) the helium abundances are not ( e.g. , skillman et al . 1994 ) .
\(4 ) observations are needed at higher spectral resolution and over a larger wavelength range .
the higher resolution ( better than 1 ) will allow the profiles of the underlying hei absorption to be observed , and thus allow for better corrections .
the larger wavelength range will allow measurements of additional physical parameters ( i.e. , the temperature in the lower ionization zone from [ o ii ] lines ) which will place stronger constraints on the derived physical parameters .
long - slit double spectrographs provide the best tool for this work . on the theoretical side ,
work remains in order to better characterize the uncertainties in the current atomic data and the prescriptions by which the observations are converted into abundances ( e.g. , porter , bauman , & ferland 2003 ) . in sum ,
echoing our introduction , pinning down the primordial helium abundance to a higher accuracy remains a worthwhile goal . to significantly decrease
the uncertainty in the primordial helium abundance will require spectra of metal poor hii regions of a quality exceeding most of those present in the literature today and analyses that take into account the several sources of systematic errors which have been identified so far .
we would like to thank r. benjamin , d. garnett , r. kennicutt , d. kunth , h. lee , v. luridiana , b. pagel , a. peimbert , m. peimbert , g. shields , j. shields , g. steigman , s. viegas , e. terlevich , and r. terlevich , for informative and valuable discussions .
the work of kao is supported in part by doe grant de - fg02 - 94er-40823 .
eds is grateful for partial support from a nasa ltsarp grant no .
nag5 - 9221 and the university of minnesota .
eds is also grateful to the institute of astronomy at the university of cambridge where much of this work was done while on sabbatical leave from the university of minnesota .
we again start with a set of observed quantities : line intensities @xmath67 which include the reddening correction previously determined in os and its associated uncertainty and the equivalent width @xmath68 .
the helium line intensities are scaled to @xmath69 and the singly ionized helium abundance is given by y^+ ( ) = i ( ) i(h ) f_(n_e , t ) f_(n_e , t , ) ( w ( ) + a_hei w ( ) ) [ y+ ] where @xmath70 is the theoretical emissivity scaled to @xmath69 and the optical depth function @xmath71 contains the collisional correction . the expression ( [ y+ ] ) , also contains a correction factor for underlying stellar absorption , parameterized now by @xmath32 thus @xmath46 implicitly depends on four unknowns , the electron density , @xmath30 , @xmath32 , @xmath31 , and @xmath40 . to be definite ,
we list here the necessary components in expression ( [ y+ ] ) .
the theoretical emissivities scaled to @xmath69 are taken from benjamin , skillman , & smits ( 1999 ) : @xmath72 our expressions for the optical depth function which includes the collisional correction is also taken from benjamin , skillman , & smits ( 1999 ) .
we list them here for completeness .
they are : @xmath73 \nonumber \\
f(4026 ) & = & 1 + ( \tau/2 ) \left [ 0.00143 + ( 4.05 \times 10^{-4 } + 3.63 \times 10^{-8 } n_e ) t ) \right ] \nonumber\\ f(4471 ) & = & 1 + ( \tau/2 ) \left [ 0.00274 + ( 8.81 \times 10^{-4 } - 1.21 \times 10^{-6 } n_e ) t ) \right ] \nonumber \\
f(5876 ) & = & 1 + ( \tau/2 ) \left [ 0.00470 + ( 2.23 \times 10^{-3 } - 2.51 \times 10^{-6 } n_e ) t ) \right ] \nonumber \\
f(6678 ) & = & 1 \nonumber \\
f(7065 ) & = & 1 + ( \tau/2 ) \left [ 0.359 + ( - 3.46 \times 10^{-2 } - 1.84 \times 10^{-4 } n_e + 3.039 \times 10^{-7 }
n_e^2)t ) \right]\end{aligned}\ ] ] finally we note that we discovered two sign errors in os concerning the theoretical balmer line ratios . in eq .
a.3 of that paper , the last two expressions each contain an error .
the correct expressions should be : @xmath74 and @xmath75 .
lcc & from ppr00 & + @xmath285876 & 0.1064 @xmath26 0.0012 & 46 @xmath26 4.6 + @xmath284471 & 0.0384 @xmath26 0.0006 & 8.5 @xmath26 0.85 + @xmath286678 & 0.0296 @xmath26 0.0003 & 14.7 @xmath26 1.47 + @xmath287065 & 0.0211 @xmath26 0.0002 & 11.1 @xmath26 1.11 + @xmath283889 & 0.0940 @xmath26 0.0017 & 16.1 @xmath26 1.61 + @xmath284026 & 0.0185 @xmath26 0.0006 & 3.16 @xmath26 0.32 + & ppr00 ( re - analyzed ) & + @xmath285876 & 0.1053 @xmath26 0.0013 & 46 @xmath26 4.6 + @xmath284471 & 0.0388 @xmath26 0.0005 & 8.5 @xmath26 0.85 + @xmath286678 & 0.0296 @xmath26 0.0002 & 14.7 @xmath26 1.47 + @xmath287065 & 0.0210 @xmath26 0.0002 & 11.2 @xmath26 1.12 + @xmath283889 & 0.0987 @xmath26 0.0022 & 16.5 @xmath26 1.65 + @xmath284026 & 0.0189 @xmath26 0.0007 & 3.10
@xmath26 0.31 + lcccccc ppr00 & 0.0793 @xmath26 0.0006 & 11,920 @xmath26 370 & 146 @xmath26 50 & 0 & 0 & ... + & ( solved ) & ( solved ) & ( solved @xmath62 ) & ( fixed ) & ( fixed ) + d & 0.0795 @xmath26 0.0005 & 11,920 @xmath26 370 & 164 @xmath45 & 0 & 0.00 @xmath76 & 3.1 + d ( re - an ) & 0.0797 @xmath26 0.0007 & 11,920 @xmath26 370 & 149 @xmath77 & 0 & 0.00 @xmath78 & 10.0 + & ( solved ) & ( fixed ) & ( solved @xmath62 ) & ( fixed ) & ( solved @xmath62 ) + d & 0.0797 @xmath26 0.0006 & 11,920 @xmath26 370 & 163 @xmath79 & 0.03 @xmath80 & 0.00 @xmath76 & 3.0 + mc & 0.0801 @xmath26 0.0011 & 11,920 @xmath26 370 & 159 @xmath26 69 & 0.07 @xmath81 & 0.03 @xmath82 & ... + d ( re - an ) & 0.0797 @xmath26 0.0007 & 11,920 @xmath26 370 & 148 @xmath83 & 0.00 @xmath84 & 0.00 @xmath78 & 10.0 + mc ( re - an ) & 0.0800 @xmath26 0.0009 & 11,920 @xmath26 370 & 152 @xmath26 63 & 0.03 @xmath85 & 0.00 @xmath86 & ... + & ( solved ) & ( fixed ) & ( solved @xmath62 ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0810 @xmath26 0.0005 & 12,510 @xmath26 190 & 89 @xmath87 & 0.08 @xmath88 & 0.00 @xmath84 & 2.4 + mc & 0.0816 @xmath26 0.0014 & 12,650 @xmath26 580 & 67 @xmath89 & 0.12 @xmath26 0.11 & 0.06 @xmath90 & ... + d ( re - an ) & 0.0827 @xmath26 0.0005 & 13,460 @xmath26 230 & 0.1
@xmath91 & 0.07 @xmath92 & 0.00 @xmath93 & 3.7 + mc ( re - an ) & 0.0828 @xmath26 0.0008 & 13,420 @xmath26 280 & 2 @xmath94 & 0.09 @xmath26 0.09 & 0.01 @xmath95 & ... + & ( solved ) & ( solved ) & ( solved @xmath62 ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0810 @xmath26 0.0005 & 12,510 @xmath26 190 & 89 @xmath96 & 0.08 @xmath97 & 0.00 @xmath84 & 2.4 + mc & 0.0820 @xmath26 0.0014 & 12,810 @xmath26 540 & 36
@xmath98 & 0.13 @xmath26 0.13 & 0.13 @xmath26 0.13 & ... + d ( re - an ) & 0.0827 @xmath26 0.0005 & 13,460 @xmath26 230 & 0.3 @xmath99 & 0.07 @xmath26 0.12 & 0.00 @xmath93 & 3.7 + mc ( re - an ) & 0.0827 @xmath26 0.0009 & 13,430 @xmath26 260 & 1 @xmath100 & 0.07 @xmath101 & 0.01 @xmath102 & ... + & ( solved ) & ( solved ) & ( solved @xmath62 ) & ( solved free ) & ( solved @xmath62 ) + d & 0.0777 @xmath26 0.0005 & 11,530 @xmath26 150 & 427 @xmath103 & @xmath480.05 @xmath26 0.11 & @xmath480.47 @xmath26 0.07 & 1.9 + mc & 0.0813 @xmath26 0.0084 & 13,000 @xmath26 3,580 & 376 @xmath26 415 & 0.10
@xmath26 0.36 & 0.07 @xmath26 1.14 & ... + d ( re - an ) & 0.0829 @xmath26 0.0005 & 13,820 @xmath26 240 & 22 @xmath26 16 & 0.06 @xmath26 0.12 & @xmath480.24 @xmath26
0.07 & 2.3 + mc ( re - an ) & 0.0836 @xmath26 0.0051 & 14,080 @xmath26 1,770 & 72 @xmath26 223 & 0.09
@xmath26 0.24 & @xmath480.16 @xmath26 0.56 & ... + & ( solved ) & ( solved ) & ( solved free ) & ( solved free ) & ( solved free ) + lcccccccc & & & & from it98 & + @xmath285876 & 0.098 @xmath26 0.002 & 35.3 @xmath26 3.5 & 0.106 @xmath26 0.001 & 63.0 @xmath26 6.3 & 0.099 @xmath26 0.003 & 36.7 @xmath26 3.67 & 0.111 @xmath26 0.002 & 38.1 @xmath26 3.81 + @xmath284471 & 0.037 @xmath26 0.002 & 6.1 @xmath26 0.6 & 0.039 @xmath26 0.001 & 10.9 @xmath26 1.1 & 0.037 @xmath26 0.002 & 6.6 @xmath26 0.66 & 0.039 @xmath26 0.002 & 6.7 @xmath26 0.67 + @xmath286678 & 0.026 @xmath26 0.001 & 12.5 @xmath26 1.3 & 0.029 @xmath26 0.001 & 24.0 @xmath26 2.4 & 0.026 @xmath26 0.001 & 12.8 @xmath26 1.28 & 0.028 @xmath26 0.001 & 14.5 @xmath26 1.45 + @xmath287065 & 0.039 @xmath26 0.001 & 22.9 @xmath26 2.3 & 0.029 @xmath26 0.001 & 29.7 @xmath26 3.0 & 0.026 @xmath26 0.002 & 12.3 @xmath26 1.23 & 0.032 @xmath26 0.002 & 19.1 @xmath26 1.91 + @xmath283889 & 0.064 @xmath26 0.004 & 7.5 @xmath26 0.8 & 0.088 @xmath26 0.001 & 16.8 @xmath26 1.7 & 0.109 @xmath26 0.007 & 13.1 @xmath26 1.31 & 0.100 @xmath26 0.005 & 9.1 @xmath26 0.91 + @xmath284026 & 0.016 @xmath26 0.002 & 2.0 @xmath26 0.2 & 0.016 @xmath26 0.001 & 3.3 @xmath26 0.3 & 0.021 @xmath26 0.002 & 2.7 @xmath26 0.27 & 0.018 @xmath26 0.002 & 2.0 @xmath26 0.20 + & & & & ( re - analyzed ) & + @xmath285876 & 0.0977 @xmath26 0.0020 & 35.3 @xmath26 3.5 & 0.1056 @xmath26 0.0010 & 63.0 @xmath26 6.3 & 0.0997 @xmath26 0.0031 & 36.7 @xmath26 3.67 & 0.1122 @xmath26 0.0028 & 38.1 @xmath26 3.81 + @xmath284471 & 0.0370 @xmath26 0.0021 & 6.1 @xmath26 0.6 & 0.0391 @xmath26 0.0010 & 10.9 @xmath26 1.1 & 0.0375 @xmath26 0.0020 & 6.6 @xmath26 0.72 & 0.0399 @xmath26 0.0022 & 6.7 @xmath26 0.73 + @xmath286678 & 0.0264 @xmath26 0.0010 & 12.5 @xmath26 1.3 & 0.0284 @xmath26 0.0009 & 24.0 @xmath26 2.4 & 0.0270 @xmath26 0.0020 & 12.8 @xmath26 1.88 & 0.0286 @xmath26 0.0017 & 14.5 @xmath26 1.71 + @xmath287065 & 0.0395 @xmath26 0.0010 & 22.9 @xmath26 2.3 & 0.0289 @xmath26 0.0009 & 29.7 @xmath26 3.0 & 0.0259 @xmath26 0.0020 & 12.3 @xmath26 1.88 & 0.0324 @xmath26 0.0017 & 19.1 @xmath26 1.97 + @xmath283889 & 0.0696 @xmath26 0.0037 & 7.9 @xmath26 0.8 & 0.0884 @xmath26 0.0011 & 16.9 @xmath26 1.7 & 0.1046 @xmath26 0.0070 & 12.8 @xmath26 1.70 & 0.0877 @xmath26 0.0055 & 8.5 @xmath26 1.05 + @xmath284026 & 0.0160 @xmath26 0.0021 & 2.0 @xmath26 0.5 & 0.0160 @xmath26 0.0011 & 3.3 @xmath26 0.4 & 0.0206 @xmath26 0.0021 & 2.7 @xmath26 0.54 & 0.0188 @xmath26 0.0012 & 2.0 @xmath26 0.25 + lcccccc & & & from it98 & + @xmath285876 & 0.101 @xmath26 0.002 & 44.3 @xmath26 4.43 & 0.102 @xmath26 0.001 & 40.8 @xmath26 4.08 & 0.101 @xmath26 0.001 & 35.5 @xmath26 3.55 + @xmath284471 & 0.039 @xmath26 0.001 & 7.2 @xmath26 0.72 & 0.036 @xmath26 0.001 & 7.48 @xmath26 0.75 & 0.037 @xmath26 0.001 & 6.9 @xmath26 0.69 + @xmath286678 & 0.026 @xmath26 0.001 & 14.3 @xmath26 1.43 & 0.029 @xmath26 0.001 & 14.2 @xmath26 1.42 & 0.028 @xmath26 0.001 & 12.8 @xmath26 1.28 + @xmath287065 & 0.027 @xmath26 0.001 & 16.6 @xmath26 1.66 & 0.024 @xmath26 0.001 & 14.3 @xmath26 1.43 & 0.037 @xmath26 0.001 & 20.4 @xmath26 2.04 + @xmath283889 & 0.104 @xmath26 0.005 & 12.5 @xmath26 1.25 & 0.097 @xmath26 0.001 & 12.4 @xmath26 1.24 & 0.085 @xmath26 0.004 & 11.4 @xmath26 1.14 + @xmath284026 & 0.019 @xmath26 0.001 & 2.60 @xmath26 0.26 & 0.016 @xmath26 0.001 & 2.42 @xmath26 0.24 & 0.018 @xmath26 0.001 & 2.60 @xmath26 0.26 + & & & ( re - analyzed ) & + @xmath285876 & 0.1018 @xmath26 0.0021 & 44.3 @xmath26 4.43 & 0.1017 @xmath26 0.0010 & 40.8 @xmath26 4.08 & 0.1006 @xmath26 0.0019 & 35.5 @xmath26 3.55 + @xmath284471 & 0.0386 @xmath26 0.0010 & 7.2 @xmath26 0.72 & 0.0366 @xmath26 0.0010 & 7.5 @xmath26 0.75 & 0.0376 @xmath26 0.0011 & 6.9 @xmath26 0.69 + @xmath286678 & 0.0267 @xmath26 0.0010 & 14.3 @xmath26 1.43 & 0.0286 @xmath26 0.0010 & 14.2 @xmath26 1.42 & 0.0283 @xmath26 0.0009 & 12.8 @xmath26 1.28 + @xmath287065 & 0.0274 @xmath26 0.0010 & 16.6 @xmath26 1.66 & 0.0247 @xmath26 0.0010 & 14.3 @xmath26 1.43 & 0.0373 @xmath26 0.0010 & 20.4 @xmath26 2.04 + @xmath283889 & 0.1033 @xmath26 0.0047 & 12.3 @xmath26 1.23 & 0.0969 @xmath26 0.0011 & 12.3 @xmath26 1.24 & 0.0893 @xmath26 0.0037 & 11.4 @xmath26 1.14 + @xmath284026 & 0.0197
@xmath26 0.0011 & 2.60 @xmath26 0.28 & 0.0158 @xmath26 0.0010 & 2.42 @xmath26 0.30 & 0.0187 @xmath26 0.0011 & 2.60 @xmath26 0.31 + lcccccc it98 & 0.080 @xmath26 0.001 & 20,300 @xmath26 300 & 67 @xmath26 3 & 0 & 1.7 @xmath26 0.3 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( fixed ) & ( fixed ) + d & 0.0798 @xmath26 0.0016 & 20,300 @xmath26 300 & 0.01 @xmath104 & 0 & 5.06 @xmath105 & 9.9 + d ( re - an ) & 0.0808 @xmath26 0.0011 & 20,530 @xmath26 390 & 0.01 @xmath106 & 0 & 4.98 @xmath107 & 4.4 + & ( solved ) & ( fixed ) & ( solved + ) & ( fixed ) & ( solved @xmath62 ) + d & 0.0825 @xmath26 0.0016 & 20,300 @xmath26 300 & 0.02 @xmath108 & 0.48 @xmath109 & 4.86 @xmath105 & 6.8 + mc & 0.0827 @xmath26 0.0022 & 20,300 @xmath26 300 & 0.3 @xmath110 & 0.52 @xmath26 0.34 & 4.85 @xmath26 0.33 & ... + d ( re - an ) & 0.0829 @xmath26 0.0014 & 20,530 @xmath26 390 & 0.0 @xmath111 & 0.38 @xmath112 & 4.82 @xmath26
0.30 & 2.3 + mc ( re - an ) & 0.0832 @xmath26 0.0022 & 20,530 @xmath26 390 & 3 @xmath113 & 0.44 @xmath26 0.34 & 4.78 @xmath260.33 & ... + & ( solved ) & ( fixed ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0720 @xmath26 0.0013 & 13,990 @xmath26 660 & 283 @xmath114 & 0.00 @xmath115 & 5.22 @xmath116 & 0.3 + mc & 0.0735 @xmath26 0.0042 & 14,090 @xmath26 2,210 & 520 @xmath117 & 0.11 @xmath118 & 5.04 @xmath26 0.83 & ... + d ( re - an ) & 0.0780 @xmath26 0.0016 & 16,800 @xmath26 1,120 & 63 @xmath119 & 0.13 @xmath120 & 4.96 @xmath26 0.31 & 0.1 + mc ( re - an ) & 0.0763 @xmath26 0.0049 & 15,940 @xmath26 2,710 & 347 @xmath121 & 0.14 @xmath122 & 4.62 @xmath260.86 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0914 @xmath26 0.0015 & 32,910 @xmath26 500 & 0.02 @xmath123 & 0.42@xmath124 & 8.59 @xmath26 0.39 & 0.3 + mc & 0.0712 @xmath26 0.0055 & 13,290 @xmath26 2,220 & 874 @xmath125 & @[email protected] & 4.89 @xmath26 0.90 & ... + d ( re - an ) & 0.0780 @xmath26 0.0016 & 16,800 @xmath26 1,120 & 63 @xmath119 & 0.13@xmath126 & 4.96 @xmath26 0.31 & 0.1 + mc ( re - an ) & 0.0755 @xmath26 0.0065 & 15,390 @xmath26 3,190 & 581 @xmath127 & [email protected] & 4.50 @xmath26 1.03 & ... + & ( solved ) & ( solved ) & ( solved @xmath62 ) & ( solved free ) & ( solved @xmath62 ) + d & 0.0716 @xmath26 0.0013 & 13,850 @xmath26 650 & 307 @xmath128 & @xmath480.02@xmath129 & 5.20 @xmath26 0.35 & 0.3 + mc & 0.0742 @xmath26 0.0072 & 14,450 @xmath26 3,170 & 348 @xmath26 438 & [email protected] & 5.28 @xmath26 1.00 & ... + d ( re - an ) & 0.0780 @xmath26 0.0016 & 16,800 @xmath26 1,120 & 63 @xmath130 & 0.13@xmath131 & 4.95 @xmath26 0.31 & 0.1 + mc ( re - an ) & 0.0762 @xmath26 0.0098 & 15,730 @xmath26 4,700 & 771 @xmath26 1,040 & [email protected] & 4.57 @xmath26 1.38 & ... + & ( solved ) & ( solved ) & ( solved free ) & ( solved free ) & ( solved free ) + lcccccc it98 & 0.081 @xmath26 0.001 & 15,800 @xmath26 100 & 253 @xmath26 10 & 0 & 0.0 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( fixed ) & ( fixed ) + d & 0.0840 @xmath26 0.0010 & 15,800 @xmath26 100 & 0.4 @xmath132 & 0 & 2.19 @xmath133 & 16.5 + d ( re - an ) & 0.0840 @xmath26 0.0010 & 15,870 @xmath26 60 & 0.1 @xmath134 & 0 & 2.06 @xmath135 & 15.2 + & ( solved ) & ( fixed ) & ( solved + ) & ( fixed ) & ( solved @xmath62 ) + d & 0.0865 @xmath26 0.0007 & 15,800 @xmath26 100 & 0.01 @xmath136 & 0.75 @xmath137 & 1.98 @xmath138 & 2.5 + mc & 0.0865 @xmath26 0.0010 & 15,800 @xmath26 100 & 2.5 @xmath139 & 0.77 @xmath26 0.24 & 1.97 @xmath26 0.19 & ... + d ( re - an ) & 0.0862 @xmath26 0.0008 & 15,870 @xmath26 60 & 0.02 @xmath140 & 0.73 @xmath141 & 1.87 @xmath138 & 2.7 + mc ( re - an ) & 0.0861 @xmath26 0.0011 & 15,870 @xmath26 60 & 6 @xmath142 & 0.75 @xmath26 0.25 & 1.85 @xmath26 0.19 & ... + & ( solved ) & ( fixed ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0848 @xmath26 0.0009 & 14,720 @xmath26 510 & 28 @xmath143 & 0.57 @xmath144 & 1.84 @xmath26 0.17 & 0.4 + mc & 0.0824 @xmath26 0.0042 & 14,260 @xmath26 1,030 & 176 @xmath145 & 0.46 @xmath26 0.28 & 1.66 @xmath26 0.38 & ... + d ( re - an ) & 0.0803 @xmath26 0.0009 & 14,010 @xmath26 460 & 205
@xmath146 & 0.35 @xmath144 & 1.45 @xmath26 0.18 & 0.8 + mc ( re - an ) & 0.0801 @xmath26 0.0046 & 14,040 @xmath26 1,060 & 285 @xmath147 & 0.35 @xmath26 0.28 & 1.44 @xmath26 0.40 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0847 @xmath26 0.0009 & 14,720 @xmath26 510 & 28 @xmath148 & 0.57 @xmath144 & 1.84 @xmath26 0.17 & 0.4 + mc & 0.0823 @xmath26 0.0046 & 14,240 @xmath26 1,080 & 202 @xmath149 & 0.45 @xmath26 0.31 & 1.65 @xmath26 0.40 & ... + d ( re - an ) & 0.0804 @xmath26 0.0009 & 14,150 @xmath26 460 & 204 @xmath150 & 0.35@xmath144 & 1.45 @xmath26 0.18 & 0.8 + mc ( re - an ) & 0.0797 @xmath26 0.0054 & 13,970 @xmath26 1,160 & 346 @xmath151 & [email protected] & 1.41 @xmath26 0.43 & ... + & ( solved ) & ( solved ) & ( solved @xmath62 ) & ( solved free ) & ( solved @xmath62 ) + d & 0.0847 @xmath26 0.0009 & 14,710 @xmath26 510 & 29 @xmath152 & 0.56 @xmath144 & 1.84 @xmath26 0.17 & 0.4 + mc & 0.0885 @xmath26 0.0105 & 16,440 @xmath26 3,690 & 91 @xmath26 406 & 0.72 @xmath260.52 & 2.29 @xmath26 1.09 & ... + d ( re - an ) & 0.0804 @xmath26 0.0009 & 14,020 @xmath26 460 & 203 @xmath153 & 0.35@xmath144 & 1.45 @xmath26 0.18 & 0.8 + mc ( re - an ) & 0.0819 @xmath26 0.0089 & 14,910 @xmath26 3,300 & 311 @xmath26 448 & [email protected] & 1.67 @xmath26 0.94 & ... + & ( solved ) & ( solved ) & ( solved free ) & ( solved free ) & ( solved free ) + lcccccc it98 & 0.083 @xmath26 0.002 & 20,200 @xmath26 300 & 10 & 0 & 0.0 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( fixed ) & ( fixed ) + d & 0.0818 @xmath26 0.0016 & 20,200 @xmath26 300 & 72 @xmath154 & 0 & 0.0 @xmath155 & 4.0 + d ( re - an ) & 0.0852 @xmath26 0.0019 & 20,200 @xmath26 400 & 0.3 @xmath156 & 0 & 0.35 @xmath157 & 1.5 + & ( solved ) & ( fixed ) & ( solved + ) & ( fixed ) & ( solved @xmath62 ) + d & 0.0817 @xmath26 0.0018 & 20,200 @xmath26 300 & 75 @xmath158 & 0.00 @xmath159 & 0.00 @xmath160 & 4.0 + mc & 0.0824 @xmath26 0.0026 & 20,200 @xmath26 300 & 67 @xmath26 63 & 0.05 @xmath161 & 0.12 @xmath162 & ... + d ( re - an ) & 0.0852 @xmath26 0.0023 & 20,200 @xmath26 400 & 0.1 @xmath163 & 0.00 @xmath164 & 0.35 @xmath157 & 1.5 + mc ( re - an ) & 0.0852 @xmath26 0.0031 & 20,200 @xmath26 400 & 22 @xmath165 & 0.09 @xmath166 & 0.31 @xmath167 & ... + & ( solved ) & ( fixed ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0843 @xmath26 0.0024 & 22,920 @xmath26 4,110 & 42 @xmath168 & 0.00 @xmath169 & 0.00 @xmath170 & 3.5 + mc & 0.0824 @xmath26 0.0043 & 20,330 @xmath26 2,510 & 83 @xmath171 & 0.05 @xmath172 & 0.22 @xmath173 & ... + d ( re - an ) & 0.0852 @xmath26 0.0027 & 20,190 @xmath26 3,770 & 0.5
@xmath174 & 0.00 @xmath164 & 0.35 @xmath157 & 1.5 + mc ( re - an ) & 0.0841 @xmath26 0.0035 & 19,260 @xmath26 2,480 & 35 @xmath175 & 0.07 @xmath176 & 0.36 @xmath177 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0814 @xmath26 0.0025 & 21,640 @xmath26
3,880 & 64 @xmath178 & @xmath480.21 @xmath179 & 0.00 @xmath155 & 3.1 + mc & 0.0735 @xmath26 0.0105 & 18,300 @xmath26 3.510 & 456 @xmath180 & @xmath480.46 @xmath26 0.45 & 0.12 @xmath162 & ... + d ( re - an ) & 0.0799 @xmath26 0.0026 & 18,330 @xmath26 2,510 & 93 @xmath181 & @xmath480.25@xmath182 & 0.00 @xmath183 & 1.2 + mc ( re - an ) & 0.0761 @xmath26 0.0111 & 17,970 @xmath26 3,030 & 354 @xmath184 & @[email protected] & 0.23 @xmath185 & ... + & ( solved ) & ( solved ) & ( solved @xmath62 ) & ( solved free ) & ( solved @xmath62 ) + d & 0.0705 @xmath26 0.0021 & 17,960 @xmath26 1,490 & 394 @xmath186 & @xmath480.58 @xmath141 & @xmath481.05 @xmath26 0.49 & 1.6 + mc & 0.0711 @xmath26 0.0070 & 18,490 @xmath26 2,560 & 434 @xmath26 296 & @xmath480.55 @xmath260.33 & @xmath480.73 @xmath26 0.69 & ... + d ( re - an ) & 0.0732 @xmath26 0.0025 & 16,870 @xmath26 1,620 & 318 @xmath187 & @xmath480.45@xmath188 & @xmath480.62 @xmath189 & 0.9 + mc ( re - an ) & 0.0740 @xmath26 0.0110 & 17,840 @xmath26 2,990 & 420 @xmath26 454 & @[email protected] & @xmath480.06 @xmath26 0.89 & ... + & ( solved ) & ( solved ) & ( solved free ) & ( solved free ) & ( solved free ) + lcccccc it98 & 0.081 @xmath26 0.001 & 16,600 @xmath26 200 & 326 @xmath26 65 & 0 & 0.0 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( fixed ) & ( fixed ) + d & 0.0784 @xmath26 0.0011 & 16,600 @xmath26 200 & 448 @xmath190 & 0 & 0.56 @xmath26 0.44 & 1.6 + d ( re - an ) & 0.0866 @xmath26 0.0016 & 16,560 @xmath26 200 & 54 @xmath191 & 0 & 2.16 @xmath192 & 5.2 + & ( solved ) & ( fixed ) & ( solved + ) & ( fixed ) & ( solved @xmath62 ) + d & 0.0804 @xmath26 0.0018 & 16,600 @xmath26 200 & 387 @xmath193 & 0.15 @xmath194 & 0.62 @xmath195 & 1.4 + mc & 0.0813 @xmath26 0.0049 & 16,600 @xmath26 200 & 369 @xmath26 188 & 0.22 @xmath196 & 0.74 @xmath26 0.53 & ... + d ( re - an ) & 0.0896 @xmath26 0.0019 & 16,560 @xmath26 200 & 0.04 @xmath197 & 0.20 @xmath198 & 2.30 @xmath199 & 3.8 + mc ( re - an ) & 0.0888 @xmath26 0.0035 & 16,560 @xmath26 200 & 38 @xmath200 & 0.20 @xmath26 0.18 & 2.16 @xmath260.51 & ... + & ( solved ) & ( fixed ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0759 @xmath26 0.0017 & 14,890 @xmath26 990 & 637 @xmath201 & 0.00 @xmath169 & 0.61 @xmath202 & 0.4 + mc & 0.0781 @xmath26 0.0054 & 15,500 @xmath26 1,860 & 607 @xmath26 372 & 0.11 @xmath203 & 0.72 @xmath26 0.49 & ... + d ( re - an ) & 0.0795 @xmath26 0.0016 & 12,600 @xmath26 850 & 524 @xmath26 164 & 0.00 @xmath204 & 1.81 @xmath205 & 1.0 + mc ( re - an ) & 0.0802 @xmath26 0.0046 & 12,790 @xmath26 1,760 & 820 @xmath206 & 0.02 @xmath207 & 1.89 @xmath260.70 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0746 @xmath26 0.0017 & 14,640 @xmath26 970 & 716 @xmath208 & @xmath480.08 @xmath209 & 0.62 @xmath26 0.37 & 0.4 + mc & 0.0758 @xmath26 0.0071 & 15,040 @xmath26 2,010 & 790 @xmath26 513 & @xmath480.02 @xmath260.33 & 0.76 @xmath26 0.45 & ... + d ( re - an ) & 0.0703 @xmath26 0.0015 & 11,240 @xmath26 460 & 2,413
@xmath210 & @xmath480.27 @xmath26 0.13 & 1.55 @xmath26 0.15 & 0.1 + mc ( re - an ) & 0.0717 @xmath26 0.0070 & 11,900 @xmath26 1,300 & 1,925 @xmath26 866 & @[email protected] & 1.86 @xmath26 0.80 & ... + & ( solved ) & ( solved ) & ( solved @xmath62 ) & ( solved free ) & ( solved @xmath62 ) + d & 0.0746 @xmath26 0.0017 & 14,630 @xmath26 970 & 718 @xmath211 & @xmath480.08@xmath209 & 0.62 @xmath26 0.37 & 0.4 + mc & 0.0752 @xmath26 0.0068 & 14,980 @xmath26 1,870 & 822 @xmath26 511 & @[email protected] & 0.73 @xmath26 0.46 & ... + d ( re - an ) & 0.0703 @xmath26 0.0015 & 11,240 @xmath26 460 & 2,413 @xmath210 & @xmath480.27 @xmath26 0.13 & 1.55 @xmath26 0.15 & 0.1 + mc ( re - an ) & 0.0719 @xmath26 0.0075 & 11,940 @xmath26 1,440 & 1,916 @xmath26 885 & @[email protected] & 1.89 @xmath26 0.85 & ... + & ( solved ) & ( solved ) & ( solved free ) & ( solved free ) & ( solved free ) + lcccccc it98 & 0.081 @xmath26 0.002 & 18,400 @xmath26 200 & 110 @xmath26 58 & 0 & 0 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( fixed ) & ( fixed ) + d & 0.0811 @xmath26 0.0010 & 18,400 @xmath26 200 & 130 @xmath212 & 0 & 0.00 @xmath213 & 3.5 + d ( re - an ) & 0.0828 @xmath26 0.0011 & 18,570 @xmath26 210 & 83 @xmath214 & 0 & 0.31 @xmath215 & 3.3 + & ( solved ) & ( fixed ) & ( solved + ) & fixed & ( solved @xmath62 ) + d & 0.0811 @xmath26 0.0011 & 18,400 @xmath26 200 & 129 @xmath216 & 0.00 @xmath78 & 0.00 @xmath213 & 3.5 + mc & 0.0820 @xmath26 0.0020 & 18,400 @xmath26 200 & 100 @xmath26 62 & 0.01@xmath217 & 0.22 @xmath218 & ... + d ( re - an ) & 0.0828 @xmath26 0.0012 & 18,570 @xmath26 210 & 83 @xmath219 & 0.00 @xmath84 & 0.32 @xmath220 & 3.3 + mc ( re - an ) & 0.0832 @xmath26 0.0021 & 18,570 @xmath26 210 & 76 @xmath26 62 & 0.01 @xmath221 & 0.39 @xmath26 0.37 & ... + & ( solved ) & ( fixed ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0825 @xmath26 0.0013 & 19,460 @xmath26 1,480 & 93 @xmath222 & 0.00 @xmath223 & 0.05 @xmath224 & 3.3 + mc & 0.0830 @xmath26 0.0033 & 19,420 @xmath26 2,100 & 89 @xmath225 & 0.02 @xmath226 & 0.27 @xmath227 & ... + d ( re - an ) & 0.0840 @xmath26 0.0014 & 19,390 @xmath26 1,670 & 55 @xmath228 & 0.00 @xmath76 & 0.38 @xmath229 & 3.2 + mc ( re - an ) & 0.0838 @xmath26 0.0031 & 19,330 @xmath26 2,100 & 75 @xmath230 & 0.01 @xmath217 & 0.45 @xmath26 0.38 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0748 @xmath26 0.0013 & 16,610 @xmath26 860 & 297 @xmath231 & @xmath480.32 @xmath232 & 0.00 @xmath233 & 1.7 + mc & 0.0725 @xmath26 0.0111 & 17,550 @xmath26 2,720 & 529 @xmath26 484 & @xmath480.47 @xmath26 0.46 & 0.15 @xmath234 & ... + d ( re - an ) & 0.0741 @xmath26 0.0013 & 16,280 @xmath26 800 & 342 @xmath235 & @xmath480.38 @xmath232 & 0.00 @xmath236 & 1.0 + mc ( re - an ) & 0.0742 @xmath26 0.0080 & 16,980 @xmath26 2,330 & 434 @xmath26 378 & @xmath480.39 @xmath26 0.27 & 0.17 @xmath237 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved free ) & ( solved @xmath62 ) + d & 0.0697 @xmath26 0.0013 & 16,570 @xmath26 780 & 537 @xmath238 & @xmath480.48 @xmath26 0.13 & @xmath480.52 @xmath26 0.24 & 0.01 + mc & 0.0698 @xmath26 0.0058 & 16,970 @xmath26 1,650 & 564 @xmath26 276 & @xmath480.48 @xmath26 0.22 & @xmath480.38 @xmath26 0.44 & ... + d ( re - an ) & 0.0716 @xmath26 0.0014 & 16,280 @xmath26 760 & 466 @xmath239 & @xmath480.46 @xmath232 & @xmath480.31 @xmath26 0.26 & 0.45 + mc ( re - an ) & 0.0720 @xmath26 0.0062 & 16,710 @xmath26 1,840 & 493 @xmath26 287 & @xmath480.45 @xmath26 0.22 & @xmath480.13 @xmath26 0.47 & ... + & ( solved ) & ( solved ) & ( solved free ) & ( solved free ) & ( solved free ) + lcccccc it98 & 0.083 @xmath26 0.001 & 16,180 @xmath26 100 & 11 @xmath26 2 & 0 & 0 & ... + & ( solved ) & ( solved ) & ( solved ) & ( fixed ) & ( fixed ) + d & 0.0801 @xmath26 0.0009 & 16,180 @xmath26 100 & 92 @xmath240 & 0 & 0.20 @xmath26 0.18 & 13.4 + d ( re - an ) & 0.0789 @xmath26 0.0007 & 16,150 @xmath26 60 & 144 @xmath119 & 0 & 0.16 @xmath241 & 9.2 + & ( solved ) & ( fixed ) & ( solved + ) & fixed & ( solved @xmath62 ) +
d & 0.0841 @xmath26 0.0009 & 16,180 @xmath26 100 & 26 @xmath242 & 0.50 @xmath243 & 0.20 @xmath26 0.18 & 2.4 + mc & 0.0839 @xmath26 0.0017 & 16,180 @xmath26 100 & 38 @xmath244 & 0.50 @xmath26 0.17 & 0.20 @xmath26 0.17 & ... + d ( re - an ) & 0.0827 @xmath26 0.0009 & 16,150 @xmath26 60 & 69 @xmath245 & 0.42 @xmath246 & 0.20
@xmath26 0.19 & 1.2 + mc ( re - an ) & 0.0827 @xmath26 0.0020 & 16,150 @xmath26 60 & 71 @xmath26 60 & 0.42 @xmath26 0.16 & 0.22 @xmath26 0.19 & ... + & ( solved ) & ( fixed ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0853 @xmath26 0.0010 & 16,560 @xmath26 600 & 0.02 @xmath247 & 0.55 @xmath248 & 0.31 @xmath26 0.18 & 2.3 + mc & 0.0847 @xmath26 0.0019 & 16,470 @xmath26 780 & 19 @xmath249 & 0.53 @xmath26 0.20 & 0.27 @xmath26 0.21 & ... + d ( re - an ) & 0.0854 @xmath26 0.0010 & 16,960 @xmath26 640 & 0.1 @xmath250 & 0.53 @xmath248 & 0.44 @xmath251 & 0.8 + mc ( re - an ) & 0.0844 @xmath26 0.0024 & 16,760 @xmath26 880 & 31 @xmath252 & 0.50 @xmath26 0.20 & 0.37 @xmath26 0.24 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0853 @xmath26 0.0010 & 16,560 @xmath26 600 & 0.1 @xmath250 & 0.55 @xmath248 & 0.31 @xmath26 0.18 & 2.3 + mc & 0.0848 @xmath26 0.0018 & 16,510 @xmath26 760 & 14 @xmath253 & 0.54 @xmath26 0.20 & 0.28 @xmath26 0.21 & ... + d ( re - an ) & 0.0854 @xmath26 0.0010 & 16,960 @xmath26 640 & 0.1 @xmath254 & 0.53 @xmath248 & 0.44 @xmath26 0.19 & 0.8 + mc ( re - an ) & 0.0846 @xmath26 0.0022 & 16,810 @xmath26 840 & 25 @xmath255 & 0.50 @xmath26 0.19 & 0.38 @xmath26 0.24 & ... + & ( solved ) & ( solved ) & ( solved @xmath62 ) & ( solved free ) & ( solved @xmath62 ) + d & 0.0922 @xmath26 0.0011 & 18,650 @xmath26 700 & @xmath48125 @xmath256 & 0.81 @xmath257 & 0.98 @xmath26 0.18 & 1.9 + mc & 0.0905 @xmath26 0.0068 & 18,430 @xmath26 2,190 & @xmath4860 @xmath26 151 & 0.74 @xmath26 0.32 & 0.88 @xmath26 0.69 & ... + d ( re - an ) & 0.0909 @xmath26 0.0011 & 18,620 @xmath26 760 & @xmath48100 @xmath258 & 0.73 @xmath259 & 0.98
@xmath26 0.18 & 0.6 + mc ( re - an ) & 0.0896 @xmath26 0.0075 & 18,680 @xmath26 2,440 & @xmath4832 @xmath26 175 & 0.67 @xmath26 0.33 & 0.96 @xmath26 0.75 & ... + & ( solved ) & ( solved ) & ( solved free ) & ( solved free ) & ( solved free ) + lcccccc it98 & 0.082 @xmath26 0.002 & 17,600 @xmath26 100 & 26 @xmath26 7 & 0 & 1.8 @xmath26 0.3 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( fixed ) & ( solved ) + d & 0.0797 @xmath26 0.0007 & 17,600 @xmath26 100 & 128 @xmath260 & 0 & 3.17 @xmath26 0.28 & 2.3 + d ( re - an ) & 0.0808 @xmath26 0.0009 & 17,850 @xmath26 120 & 119 @xmath261 & 0 & 3.04 @xmath26 0.27 & 3.6 + & ( solved ) & ( fixed ) & ( solved + ) & fixed & ( solved @xmath62 ) + d & 0.0807 @xmath26 0.0010 & 17,600 @xmath26 100 & 98 @xmath260 & 0.07 @xmath262 & 3.28 @xmath26 0.28 & 2.2 + mc & 0.0811 @xmath26 0.0025 & 17,600 @xmath26 100 & 92 @xmath26 83 & 0.11 @xmath26 0.12 & 3.32 @xmath26 0.44 & ... + d ( re - an ) & 0.0808 @xmath26 0.0011 & 17,850 @xmath26 120 & 120 @xmath263 & 0.00 @xmath264 & 3.04 @xmath26 0.27 & 3.6 + mc ( re - an ) & 0.0815 @xmath26 0.0024 & 17,850 @xmath26 120 & 110 @xmath26 79 & 0.04 @xmath265 & 3.08 @xmath26 0.43 & ... + & ( solved ) & ( fixed ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0852 @xmath26 0.0013 & 19,160 @xmath26 1,710 & 2 @xmath266 & 0.23 @xmath243 & 3.54 @xmath26 0.28 & 2.0 + mc & 0.0831 @xmath26 0.0035 & 18,680 @xmath26 1,700 & 56 @xmath267 & 0.16 @xmath26 0.17 & 3.43 @xmath26 0.39 & ... + d ( re - an ) & 0.0866 @xmath26 0.0015 & 20,940 @xmath26 2,290 & 0.3 @xmath268 & 0.13 @xmath269 & 3.41 @xmath26 0.27 & 2.3 + mc ( re - an ) & 0.0854 @xmath26 0.0026 & 20,370 @xmath26 1,740 & 27 @xmath270 & 0.12 @xmath271 & 3.32 @xmath26 0.35 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved @xmath62 ) & ( solved @xmath62 ) + d & 0.0852 @xmath26 0.0013 & 19,130 @xmath26 1,710 & 4 @xmath272 & 0.22 @xmath246 & 3.52 @xmath26 0.28 & 2.0 + mc & 0.0804 @xmath26 0.0077 & 17,280 @xmath26 3,640 & 419 @xmath273 & 0.06 @xmath26 0.30 & 3.11 @xmath26 0.74 & ... + d ( re - an ) & 0.0866 @xmath26 0.0015 & 20,950 @xmath26 2,290 & 0.3 @xmath268 & 0.13 @xmath198 & 3.41 @xmath26 0.27 & 2.3 + mc ( re - an ) & 0.0860 @xmath26 0.0030 & 20,600 @xmath26 2,070 & 25 @xmath274 & 0.11 @xmath26 0.20 & 3.44 @xmath26 0.34 & ... + & ( solved ) & ( solved ) & ( solved + ) & ( solved free ) & ( solved @xmath62 ) + d & 0.0854 @xmath26 0.0012 & 19,200 @xmath26 1,720 & @xmath481 @xmath275 & 0.23 @xmath246 & 3.55 @xmath26 0.28 & 1.96 + mc ( re - an ) & 0.0802 @xmath26 0.0091 & 17,340 @xmath26 3,920 & 459 @xmath26 746 & 0.06 @xmath26 0.34 & 3.14 @xmath26 0.92 & ... + d ( re - an ) & 0.0870 @xmath26 0.0016 & 21,150 @xmath26 2,240 & @xmath486 @xmath276 & 0.14 @xmath198 & 3.46 @xmath277 & 2.3 + mc ( re - an ) & 0.0855 @xmath26 0.0057 & 20,450 @xmath26 2,400 & 72 @xmath26 309 & 0.10 @xmath26 0.26 & 3.41 @xmath26 0.58 & ... + & ( solved ) & ( solved ) & ( solved free ) & ( solved free ) & ( solved free ) + lcccccccc & & & & it98 + sbs 0335 - 052 & 20,300 @xmath26 300 & 67 @xmath26 3 & 1.7 @xmath26 0.3 & 0 & 0.080 @xmath26 0.001 & 0.002 @xmath26 0.001 & 0.2488 @xmath26 0.0044 & 0.19 @xmath26 0.01 + ngc 2363a & 15,800 @xmath26 100 & 253 @xmath26 10 & 0.0 & 0 & 0.081 @xmath26 0.001 & 0.001 @xmath26 0.001 & 0.2456 @xmath26 0.0008 & 0.71 @xmath26 0.01 + sbs 0940 + 544n & 20,200 @xmath26 300 & 10 & 0.0 & 0 & 0.082 @xmath26 0.002 & 0.000 & 0.2500 @xmath26 0.0057 & 0.27 @xmath26 0.01 + mrk 193 & 16,600 @xmath26 200 & 326 @xmath26 65 & 0.0 & 0 & 0.081 @xmath26 0.001 & 0.001 @xmath26 0.001 & 0.2478 @xmath26 0.0037 & 0.64 @xmath26 0.01 + sbs 1159 + 545 & 18,400 @xmath26 200 & 110 @xmath26 58 & 0.0 & 0 & 0.081 @xmath26 0.002 & 0.001 @xmath26 0.001 & 0.2456 @xmath26 0.0049 & 0.31 @xmath26 0.01 + haro 29 & 16,180 @xmath26 100 & 11 @xmath26 2 & 0.0 & 0 & 0.083 @xmath26 0.001 & 0.001 @xmath26 0.001 & 0.2509 @xmath26 0.0012 & 0.59 @xmath26 0.01 + sbs 1420 + 544 & 17,600 @xmath26 100 & 26 @xmath26 7 & 1.8 @xmath26 0.3 & 0 & 0.082 @xmath26 0.001 & 0.001 @xmath26 0.001 & 0.2497 @xmath26 0.0030 & 0.56 @xmath26 0.01 + & & & & our re - analysis + sbs
0335 - 052 & 15,940 @xmath26 2,710 & 347 @xmath278 & 4.6 @xmath26 0.9 & 0.1 @xmath279 & 0.0763 @xmath26 0.0049 & 0.0023 @xmath26 0.0023 & 0.2391 @xmath26 0.0116 & 0.19 @xmath26 0.01 + ngc 2363a & 14,040 @xmath26 1,060 & 285 @xmath147 & 1.4 @xmath26 0.4 & 0.35 @xmath26 0.28 & 0.0801 @xmath26 0.0046 & 0.0008 @xmath26 0.0008 & 0.2441 @xmath26 0.0107 & 0.69 @xmath26 0.01 + sbs 0940 + 544n & 19,260 @xmath26 2,480 & 35 @xmath175 & 0.4 @xmath26 0.4 & 0.1 @xmath280 & 0.0841 @xmath26 0.0035 & 0.000 & 0.2516 @xmath26 0.0078 & 0.26 @xmath26 0.01 + mrk 193 & 12,790 @xmath26 1,760 & 820 @xmath206 & 1.9 @xmath26 0.7 & 0.0 @xmath281 & 0.0802 @xmath26 0.0046 & 0.0011 @xmath26 0.0011 & 0.2451 @xmath26 0.0107 & 0.61 @xmath26 0.02 + sbs 1159 + 545 & 19,330 @xmath26 2,100 & 75 @xmath230 & 0.5 @xmath26 0.4 & 0.0 @xmath281 & 0.0838 @xmath26 0.0031 & 0.0010 @xmath26 0.0010 & 0.2531 @xmath26 0.0073 & 0.30 @xmath26 0.01 + haro 29 & 16,680 @xmath26 890 & 33 @xmath282 & 0.4 @xmath26 0.2 & 0.5 @xmath26 0.2 & 0.0844 @xmath26 0.0024 & 0.0010 @xmath26 0.0010 & 0.2543 @xmath26 0.0058 & 0.58 @xmath26 0.01 + sbs 1420 + 544 & 20,370 @xmath26 1,740 & 27 @xmath270 & 3.3 @xmath26 0.4 & 0.1 @xmath279 & 0.0854 @xmath26 0.0026 & 0.0011 @xmath26 0.0011 & 0.2568 @xmath26 0.0062 & 0.54 @xmath26 0.01 + lcccccccc & & & & it04 + j 0519 + 0007 & 20,740 @xmath26 340 & ... & ... & 0 & 0.0783 @xmath26 0.0015 & 0.0023 @xmath26 0.0002 & 0.2437 @xmath26 0.0047 & 2.70 @xmath26 0.09 + hs 2236 + 1344 & 21,120 @xmath26 290 & ... & ... & 0 & 0.0789 @xmath26 0.0013 & 0.0010 @xmath26 0.0001 & 0.2421 @xmath26 0.0040 & 2.96 @xmath26 0.08 + hs 0122 + 0743 & 17,740 @xmath26 230 & ... & ... & 0 & 0.0826 @xmath26 0.0013 & 0.0007 @xmath26 0.0001 & 0.2497 @xmath26 0.0042 & 3.97 @xmath26 0.11 + hs 0837 + 4717 & 19,510 @xmath26 240 & ... & ... & 0 & 0.0763 @xmath26 0.0011 & 0.0021 @xmath26 0.0001 & 0.2386 @xmath26 0.0034 & 3.98 @xmath26 0.10 + cgcg 007 - 025 & 16,470 @xmath26 170 & ... & ... & 0 & 0.0786 @xmath26 0.0009 & 0.0012 @xmath26 0.0001 & 0.2417 @xmath26 0.0029 & 5.96 @xmath26 0.14 + hs 0134 + 3415 & 16,390 @xmath26 180 & ... & ... & 0 & 0.0812 @xmath26 0.0013 & 0.0005 @xmath26 0.0002 & 0.2459 @xmath26 0.0042 & 7.20 @xmath26 0.19 + hs 1028 + 3843 & 15,820 @xmath26 160 & ... & ... & 0 & 0.0800 @xmath26 0.0011 & 0.0012 @xmath26 0.0001 & 0.2449 @xmath26 0.0035 & 7.81 @xmath26 0.20 + & & & & our re - analysis + j 0519 + 0007 & 22,050 @xmath26 1,910 & 335 @xmath26 189 & 2.09 @xmath26 0.86 & 0.18 @xmath26 0.27 & 0.0799 @xmath26 0.0070 & 0.0022 @xmath26 0.0022 & 0.2471 @xmath26 0.0166 & 2.46 @xmath26 0.10 + hs 2236 + 1344 & 22,740 @xmath26 2,230 & 139 @xmath26 226 & 4.52 @xmath26 0.71 & 0.38 @xmath26 0.40 & 0.0852 @xmath26 0.0057 & 0.0009 @xmath26 0.0009 & 0.2560 @xmath26 0.0129 & 2.76 @xmath26 0.09 + hs 0122 + 0743 & 18,860 @xmath26 2,210 & 31 @xmath26 107 & 0.72 @xmath26 0.43 & 1.19 @xmath26 0.32 & 0.0887 @xmath26 0.0041 & 0.0007 @xmath26 0.0007 & 0.2632 @xmath26 0.0089 & 3.81 @xmath26 0.12 + hs 0837 + 4717 & 21,850 @xmath26 1,760 & 303 @xmath26 112 & 4.36 @xmath26 0.61 & 0.01 @xmath26 0.03 & 0.0835 @xmath26 0.0038 & 0.0020 @xmath26 0.0020 & 0.2547 @xmath26 0.0095 & 3.64 @xmath26 0.13 + cgcg 007 - 025 & 20,180 @xmath26 2,210 & 85 @xmath26 116 & 0.87 @xmath26 0.49 & 0.43 @xmath26 0.30 & 0.0894 @xmath26 0.0051 & 0.0012 @xmath26 0.0012 & 0.2657 @xmath26 0.0114 & 5.69 @xmath26 0.19 + hs 0134 + 3415 & 18,160 @xmath26 2.370 & 108 @xmath26 205 & 1.06 @xmath26 0.58 & 0.25 @xmath26 0.31 & 0.0840 @xmath26 0.0051 & 0.0005 @xmath26 0.0005 & 0.2523 @xmath26 0.0115 & 6.86 @xmath26 0.20 + hs 1028 + 3843 & 17,530 @xmath26 2,940 & 461 @xmath26 448 & 5.35 @xmath26 0.88 & 0.03 @xmath26 0.14 & 0.0880 @xmath26 0.0050 & 0.0012 @xmath26 0.0012 & 0.2626 @xmath26 0.0113 & 7.55 @xmath26 0.23 + | -4.5 in umn - th-2311/04 + ftpi - minn-04/22 + astro - ph/0405588 + june 2004 3.5 in using the wmap determination of the baryon density , the standard model of big bang nucleosynthesis yields relatively precise predictions of the primordial light element abundances .
currently there are two significantly different observational determinations of the primordial helium abundance , and , if only statistical errors in 4 abundance determinations are considered , the discrepancies between the observational determinations and the value favored by the wmap results are significant . here
we examine in detail some likely sources of systematic uncertainties which may resolve the differences between the two determinations .
we conclude that the observational determination of the primordial helium abundance is completely limited by systematic errors and that these systematic errors have not been fully accounted for in any published observational determination of the primordial helium abundance . in principle , the observed metal - poor hii region spectra should be analyzed in a non - parametric way , such that the hii region physical conditions and the helium abundance are derived solely from the relative flux ratios of the helium and hydrogen emission lines . in practice , there are very few hii region spectra with the quality that allow this , so that most analyses depend on assumed ranges or relationships between physical parameters , resulting in parametric solutions with underestimated error bars .
a representative result of our analysis yields @xmath0 . however , given that most of the spectra analyzed to date do not significantly constrain the primordial helium abundance , we argue in favor a range of allowed values of 0.232 @xmath1 y@xmath2 @xmath1 0.258 .
this easily allows for concordance between measurements of the baryon - to - photon ratio ( @xmath3 ) from wmap , deuterium abundances , and helium abundance ( although the discrepancy with lithium remains ) . |
the parent compounds of the high - transition - temperature ( high-@xmath8 ) copper - oxide superconductors are antiferromagnetic ( af ) insulators characterized by a simple doubling of the crystallographic unit cell in the cuo@xmath3 planes @xcite .
when holes are doped into these planes , the long - range af - ordered phase is destroyed and the lamellar copper - oxide materials become metallic and superconducting .
this raises the fundamental question of the role of the magnetism in the superconductivity of these materials . over the past decade ,
experimental probes such as nuclear magnetic resonance ( nmr ) , muon spin resonance ( @xmath22sr ) , and neutron scattering have been used to address this fundamental question . in particular ,
neutron scattering experiments have unambiguously shown the presence of short - range af spin correlations ( fluctuations ) in cuprate superconductors such as la@xmath15sr@xmath16cuo@xmath17 @xcite and yba@xmath3cu@xmath4o@xmath5 @xcite at all doping levels , @xmath6 .
even so , the role of such fluctuations in the pairing mechanism and superconductivity is still a subject of controversy @xcite . if spin fluctuations are important for the mechanism of high-@xmath8 superconductivity , they should be universal for all copper oxide systems .
for the la@xmath3cuo@xmath17 family of materials , which has a single cuo@xmath3 layer per unit cell , superconductivity can be induced by the substitution of sr , ba , or ca for la or by intercalation of excess oxygen .
the low - energy spin fluctuations have been observed at a quartet of incommensurate wave vectors ( fig .
1 ) away from the af bragg positions , often referred to as @xmath23 , in the @xmath24 compounds @xcite .
these modulated low - frequency spin fluctuations persist in both the normal and superconducting states with a suppression of their intensity below @xmath8 @xcite . as a function of the doping ,
the incommensurability @xmath18 increases linearly with @xmath6 until saturating at @xmath25 for @xmath26 @xcite .
furthermore , the incommensurability depends only on the doping level and the unbound charge introduced by the dopants , and not on the method for introducing the charge @xcite . as a function of increasing frequency ,
the incommensurability does not change @xcite , and no particularly sharp features have been identified @xcite . for yba@xmath3cu@xmath4o@xmath5 , which has two cuo@xmath3 planes per unit cell ( bilayer ) ,
the situation is more complex .
early measurements @xcite have shown that the most prominent feature in the spin fluctuations spectra in the highly doped yba@xmath3cu@xmath4o@xmath5 ( @xmath27 ) is a sharp peak which appears below @xmath8 at an energy of @xmath2841 mev . when scanned at fixed energy as a function of wave vector , the sharp peak
is centered at @xmath29 and generally referred to as a resonance .
for underdoped yba@xmath3cu@xmath4o@xmath5 , the resonance also peaks at @xmath29 but with a reduced energy @xcite . in yba@xmath3cu@xmath4o@xmath30
, it occurs at 34 mev and is superposed on a continuum which is gapped at low energies @xcite .
below the resonance frequency , the continuum is actually strongest at a quartet of incommensurate positions consistent with those in la@xmath15sr@xmath16cuo@xmath17 of the same hole doping ( fig .
1 ) @xcite .
this discovery along with similar observations in yba@xmath3cu@xmath4o@xmath31 @xcite suggests that incommensurate spin fluctuations may be universal to all high-@xmath8 superconductors . in this article ,
we report systematic unpolarized and polarized inelastic neutron scattering studies of spin fluctuations in yba@xmath3cu@xmath4o@xmath5 . by carefully separating the magnetic signal from phonons and other spurious effects
, we determine the wave vector and energy dependence of the spin dynamical susceptibility , @xmath0 , as a function of temperature and doping concentration .
for yba@xmath3cu@xmath4o@xmath5 with oxygen concentrations at all doping levels addressed in this article , we observe a spin gap in the superconducting state , a commensurate resonance , and incommensurate spin fluctuations at energies below the resonance .
the magnitude of the spin gap in the superconducting state ( @xmath7 ) is proportional to @xmath13 with @xmath9 .
the resonance appears above @xmath8 for underdoped compounds but occurs exclusively below @xmath8 for optimal and overdoped materials .
the incommensurability of the excitations below the resonance decreases with increasing energy until merging into the resonance . for yba@xmath3cu@xmath4o@xmath5 with @xmath21 ,
the incommensurability is approximately proportional to @xmath8 . for oxygen concentrations @xmath6 above 0.6 ,
the incommensurability is independent of hole - doping and saturates at @xmath32 . in the case of yba@xmath3cu@xmath4o@xmath33 , the commensurate low - frequency normal state magnetic excitations centered around @xmath29
are suppressed with the opening of a large spin gap in the superconducting state . for yba@xmath3cu@xmath4o@xmath34 and yba@xmath3cu@xmath4o@xmath35 ,
the normal state spin fluctuations are below the detection limit of current triple - axis spectroscopy , and the magnetic fluctuations in the superconducting state are dominated by the well - discussed resonance at @xmath36 mev @xcite and incommensurate spin fluctuations below the resonance .
the rest of the paper is organized as follows .
the next section describes the experimental procedures , including details of sample characterization and spectrometer setup . in sec .
iiia , polarized and unpolarized neutron scattering data for yba@xmath3cu@xmath4o@xmath5 with oxygen concentrations , @xmath6 , of @xmath37 , 0.5 , 0.6 , 0.7 , 0.8 , 0.93 , and 0.95 are presented with a detailed description of the procedures used to extract magnetic scattering . in sec .
iiib , a comparison of the results with those for la@xmath15sr@xmath16cuo@xmath17 is presented .
the issues of the doping dependence of the incommensurability , the spin gap in the superconducting state , the resonance in the normal state of underdoped materials , and resonance energy are addressed . finally , the results presented in this paper and recent magnetic field dependence measurements of the spin fluctuations are discussed with the predictions and consequences of various theoretical models .
section iv gives a brief summary of the major conclusions of this work .
the experiments were performed on large twined crystals of yba@xmath3cu@xmath4o@xmath5 that were prepared by the melt - grown technique at the university of washington . as emphasized in previous publications @xcite ,
these samples contain considerable amounts of y@xmath3bacuo@xmath38 @xcite as an impurity .
the structural and superconducting properties of these melt - grown yba@xmath3cu@xmath4o@xmath5 samples with different y@xmath3bacuo@xmath38 concentrations have been studied in detail by gopalan _
these authors found that the yba@xmath3cu@xmath4o@xmath5 platelet width and crack width decrease with increasing y@xmath3bacuo@xmath38 content . the reduced average yba@xmath3cu@xmath4o@xmath5 grain width may facilitate the formation of equilibrium crystal structures at various oxygen concentrations . in a recent study of oxygen equilibration and chemical diffusion in massive yba@xmath3cu@xmath4o@xmath5 crystals , lindemer @xcite found that melt - grown samples containing a significant amount of y@xmath3bacuo@xmath38 reached their equilibrium values of oxygen concentration in a dynamic flow of oxygen considerably faster than samples without this impurity .
these results are consistent with the notion that oxygen invariant y@xmath3bacuo@xmath38 , micro - cracks , and the boundaries between twin grains are important in providing oxygen pathways through these large samples of yba@xmath3cu@xmath4o@xmath5 . for nonstoichiometric pure phase yba@xmath3cu@xmath4o@xmath5 , lindemer and coworkers @xcite have established the thermodynamic equilibrium phase diagram of oxygen concentration versus temperature and oxygen pressure ( @xmath39-@xmath40 ) .
we oxygenate yba@xmath3cu@xmath4o@xmath5 in a vacuum thermogravimetric apparatus ( tga ) by annealing to the equilibrium weight ( @xmath41 mg ) , equivalent to a sensitivity in @xmath42 of 0.0003 per mol , at @xmath39-@xmath40[o@xmath3 ] conditions for the desired oxygen doping following previously established models @xcite . instead of quenching the sample after annealing ,
the partial pressure of oxygen in the tga was adjusted during the cooling process to maintain a constant sample mass , thus ensuring uniformity in oxygen content and an equilibrium crystal structure .
seven single crystals of yba@xmath3cu@xmath4o@xmath5 were oxygenated in this way to obtain @xmath6 of 0.45 , 0.5 , 0.6 , 0.7 , 0.8 , 0.93 , and 0.95 .
the superconducting properties of these yba@xmath3cu@xmath4o@xmath5 crystals were characterized by a squid magnetometer .
susceptibility measurements made on small pieces cut from these samples show superconducting transition temperatures of @xmath44 k , @xmath45 k , @xmath46 k , @xmath47 k , @xmath48 k , and @xmath49 k , for yba@xmath3cu@xmath4o@xmath50 , yba@xmath3cu@xmath4o@xmath51 , yba@xmath3cu@xmath4o@xmath30 , yba@xmath3cu@xmath4o@xmath31 , yba@xmath3cu@xmath4o@xmath33 , and yba@xmath3cu@xmath4o@xmath34 , respectively .
the transition temperature for yba@xmath3cu@xmath4o@xmath35 of 92 k is estimated from the temperature dependence of the resonance @xcite .
although the widths of superconducting transition temperatures for yba@xmath3cu@xmath4o@xmath5 with @xmath52 are usually sharp ( @xmath53 k ) , the transition widths for lower doping samples are considerably broader .
this appears to be the intrinsic property of the underdoped yba@xmath3cu@xmath4o@xmath5 @xcite . to quantitatively determine the effectiveness of this oxygenation technique and the composition of the melt - grown crystal
, two pieces were cut from a 96-g sample of yba@xmath3cu@xmath4o@xmath34 .
sample @xmath54 weighed approximately 15-g and was crushed into fine powder .
sample @xmath55 weighed 14.98-g .
powder diffraction measurements were performed on sample @xmath54 at room temperature using the hb-4 high - resolution powder diffractometer with an incident beam wavelength of 1.0314 at the high - flux isotope reactor ( hfir ) at oak ridge national laboratory @xcite .
rietveld analysis of the powder data ( @xmath56 , @xmath57 ) revealed that the sample contains 83.6 mol% yba@xmath3cu@xmath4o@xmath5 with the partial oxygen occupancies [ o(1 ) ] at @xmath58 and 16.4 mol% y@xmath3bacuo@xmath38 .
the lattice parameters for the yba@xmath3cu@xmath4o@xmath34 [ @xmath59 , @xmath60 , and @xmath61 ] are slightly different from that of the yba@xmath3cu@xmath4o@xmath34 sample reported by jorgensen _
@xcite , but they agree well with those of the yba@xmath3cu@xmath4o@xmath5 ( @xmath62 ) crystals of altendorf _ et al .
_ @xcite . to further verify the partial oxygen content ,
sample @xmath55 was subjected to independent thermogravimetric analysis .
the 14.98-g specimen ( @xmath55 ) was subjected to the de - oxygenation study in the tga .
the weight change of @xmath55 relative to the room - temperature weight was measured after equilibration at 573 k and at successively higher temperatures .
the results were converted to @xmath42 values and compared to the expected behavior @xcite .
this analysis demonstrated that the initial oxygen content of @xmath55 was @xmath63 , thus confirming the effectiveness of our oxygenation technique . in an unpublished work , chakoumakos and lindemer @xcite
have systematically studied the lattice parameters of the powder yba@xmath3cu@xmath4o@xmath5 oxygenated using this technique .
they performed rietveld analysis of the powder neutron diffraction data obtained on hb-4 for a series yba@xmath3cu@xmath4o@xmath5 samples and established the relationship between oxygen concentration and lattice parameters ( fig .
high resolution measurements on the single crystals oxygenated in our study indicated that the lattice parameters have virtually the same value as powders . as a consequence
, we used the lattice parameters established in their powder work for our single - crystal inelastic neutron scattering measurements .
inelastic neutron scattering measurements were made at hfir using the hb-1 and hb-3 triple - axis spectrometers .
the momentum transfer @xmath64 is measured in units of @xmath65 and reciprocal space positions are specified in reciprocal lattice units ( rlu ) @xmath66 , where @xmath67 , @xmath68 ( @xmath69 ) , and @xmath70 are the lattice parameters of the orthorhombic unit cell of yba@xmath3cu@xmath4o@xmath5 ( see figs . 1 and 2 ) .
the magnetic neutron scattering directly measures the imaginary part of the generalized spin susceptibility @xmath0 for momentum transfer @xmath71 and energy transfer @xmath1 , where @xmath72 and @xmath73 are the incident and final neutron wave vectors , respectively .
the scattering cross - section for an isotropic system @xcite is @xmath74 where @xmath75 is the lande factor ( @xmath76 ) , @xmath77 is @xmath78 cm , @xmath79 is the magnetic form factor , @xmath80=1/[1-exp(-\hbar\omega / k_bt)]$ ] is the bose population factor , and @xmath81 is the bohr magneton .
the major difficulty in studying spin fluctuations in high-@xmath8 materials is to separate the magnetic scattering from ( single- and multi- ) phonon scattering and other spurious processes .
while many spurious events such as accidental bragg scattering can be identified by performing the desired inelastic scan in the two - axis mode @xcite , two approaches are used to separate magnetic from phonon scattering .
we first describe the use of neutron polarization analysis @xcite which , in principle , can unambiguously separate magnetic from non magnetic processes . for the polarized experimental setup , a crystal of @xmath82fe(si ) [ @xmath83 ] and crystals of heusler alloy cu@xmath3mnal [ @xmath84 ]
were used as a monochromator and analyzer , respectively .
the polarization state of the neutron beam can be changed by a spin - flipping device ( mezei coil ) that is placed in the incident beam _ before _ the sample .
the combination of a @xmath82fe(si ) ( 110 ) monochromator and a heusler ( 111 ) analyzer permits spin - flip ( sf ) measurements without using the mezei coil because the polarization of the neutrons scattered by the @xmath82fe(si ) is along the applied field direction while it is opposite for the heusler crystal . for the measurements ,
the final neutron energy was fixed at 30.5 mev , and a pyrolytic graphite ( pg ) filter was used before the analyzer .
the polarization is usually determined by measuring the flipping ratio @xmath85 , defined as the ratio of neutrons reflected by the analyzer with the flipper ( mezei coil ) `` on '' [ which corresponds to neutron nonspin flip ( nsf ) , or @xmath86 and `` off '' [ neutron spin - flip ( sf ) or @xmath87 . assuming the flipping ratios for guide field parallel ( horizontal field or hf ) and perpendicular ( vertical field or vf ) to the wave vector @xmath88 are @xmath89 and @xmath90 , respectively , the observed neutron cross sections for a paramagnetic system are given by @xcite : @xmath91 where @xmath92 is the magnetic , @xmath93 the nuclear , @xmath94 the nuclear spin incoherent neutron cross section , and @xmath95 is the background .
inspection of eq .
[ 2.2 ] reveals that the most effective way to detect a magnetic signal is to measure sf scattering with hf .
although such method has been successfully used to identify the magnetic origin of resonance peaks in yba@xmath3cu@xmath4o@xmath5 @xcite , the advantage of the technique comes at a considerable cost in intensity which makes it impractical for observing small magnetic signals . as a consequence ,
most of our experiments were performed with unpolarized neutrons with be ( 002 ) or pg ( 002 ) as monochromators as specified in the figure captions . in these experiments
, we also used pg(002 ) as analyzers .
fast neutrons were removed from the incident beam by an in - pile sapphire filter , and pg filters were used to remove high - order neutron contaminations where appropriate .
the horizontal collimations were controlled with soller slits and are specified in the figures containing the experimental data . to separate the magnetic from phonon scattering , we utilize the differences in their temperature and wave vector dependence of the cross sections . while the phonon scattering gains intensity on warming due to the thermal population factor , the magnetic signal usually becomes weaker because it spreads throughout the energy and momentum space at high temperatures .
in addition , magnetic intensity should decrease at large wave vectors due to the reduction in the magnetic form factor while phonon and multi - phonon scattering are expected to increase with increasing wave vector .
therefore , in an unpolarized neutron scattering measurement the net intensity gain above the multi - phonon background on cooling at appropriate wave vectors is likely to be magnetic in origin .
previous experiments have utilized this method to discover incommensurate spin fluctuations in yba@xmath3cu@xmath4o@xmath30 @xcite .
figure 1 depicts the reciprocal lattice of the yba@xmath3cu@xmath4o@xmath5 @xmath96 ( @xmath97 ) , @xmath98 ( @xmath99 ) directions shown in square lattice notation .
our crystals are all highly twined so we can not distinguish @xmath96 from @xmath98 .
because the cuo@xmath3 planes in yba@xmath3cu@xmath4o@xmath5 actually appear in coupled bilayers that are separated by a much larger distance between the bilayer , spin fluctuations that are in - phase ( acoustic or @xmath100 ) or out - of - phase ( optical or @xmath101 ) with respect to the neighboring plane will have different spectra @xcite .
it is therefore convenient to separate the magnetic response of yba@xmath3cu@xmath4o@xmath5 into two parts @xmath102 , where @xmath103 ( 3.342 ) is the spacing between the nearest - neighbor cuo@xmath3 planes along @xmath70 . for the af insulating parent compound yba@xmath3cu@xmath4o@xmath104 , @xmath100 and @xmath101
correspond to the acoustic and optical spin - wave excitations , respectively @xcite .
because the interesting features such as the resonance and incommensurate fluctuations @xcite are only observed in the acoustic channel of spin susceptibility @xmath105 and the optical fluctuations occur at higher energies than the acoustic excitations , we focused on the temperature , frequency , and composition dependence of spin excitations in the acoustic mode which has a modulation @xmath106 along the @xmath70-axis . for this purpose
, we used several crystal orientations and scattering geometries .
previous work by different groups @xcite were focused on the behavior of magnetic excitations for @xmath2 in the @xmath107 and/or @xmath108 zone . in view of the fact that the low frequency spin fluctuations in yba@xmath3cu@xmath4o@xmath5 peak at a quartet of incommensurate positions @xmath18
rlu away from @xmath109 , @xmath29 , as depicted in fig
. 1 @xcite , we scanned along the arrow direction in fig . 1 from @xmath110 to @xmath111 with fixed @xmath112 .
to perform these scans , we first aligned the crystal with its [ 1,0,0 ] and [ 0,0,1 ] axes along the rotational axes of the lower ( @xmath113-axis ) and upper ( @xmath114-axis ) arcs of the spectrometer goniometer , respectively .
the crystal was then rotated @xmath115 degrees such that after the rotation the @xmath116 $ ] and @xmath117 $ ] axes were in the horizontal scattering plane . to access the maxima of acoustic spin fluctuations ,
we chose @xmath118 or 5 rlu so the @xmath119 rotation angle would be approximately 37 or 17 degrees , respectively .
one advantage of performing scans in this manner is that the scan direction is along the direction of incommensurate peaks and the narrow part of the resolution ellipse . as shown in fig .
1 , the narrow part of the spectrometer resolution matches the separation of the incommensurate peaks . therefore , this is a favorable geometry to study the evolution of the incommensurate magnetic excitations .
in addition to the new experimental geometry , we also used the conventional @xmath108 zone as specified in the figure captions .
we begin this section by noting that inelastic neutron scattering experiments over the last decade @xcite have provided valuable information about the evolution of the spin dynamics in yba@xmath3cu@xmath4o@xmath5 .
the key features of the spin excitations spectra include a gap in the superconducting state , a pseudogap in the normal state , the resonance peak , and the incommensurate fluctuations below the resonance .
the observation of remarkable similarities in spin incommensurate structure for the bilayer yba@xmath3cu@xmath4o@xmath30 @xcite and the single - layer la@xmath15sr@xmath16cuo@xmath17 of the same hole doping density @xcite suggests that incommensurate magnetic fluctuations may be a universal feature of the cuprate superconductors . for the la@xmath15sr@xmath16cuo@xmath17 system ,
yamada and coworkers established the sr - doping dependence of the incommensurability @xmath18 of the spin fluctuations and found a remarkable linear relationship between @xmath8 and @xmath18 @xcite . in the case of yba@xmath3cu@xmath4o@xmath5 , balatsky and bourges @xcite suggested that a similar linear relationship exists between @xmath8 and the width in momentum space ( @xmath120 ) of the `` normal '' spin fluctuations , i.e. , fluctuations at frequencies away from the resonance .
the authors argued that spin excitations of yba@xmath3cu@xmath4o@xmath5 have two components and the low - energy incommensurate fluctuations are different than the commensurate resonance .
the conclusions of @xcite were reached based mostly on data where explicit incommensurability in the low - frequency spin fluctuations was not observed .
therefore , it is still not clear whether the linear relationship between @xmath8 and @xmath18 is a universal feature of the cuprates .
below , we present our systematic investigation of spin excitations in yba@xmath3cu@xmath4o@xmath5 . as stressed in sec .
iib , it is not a trivial matter to separate the magnetic scattering from phonon and other spurious processes in metallic crystals of yba@xmath3cu@xmath4o@xmath5 .
although the use of polarized neutrons can provide an unambiguous measure of the magnetic and phonon scattering @xcite , the technique usually suffers from the limited flux available with polarized neutrons . nevertheless , polarization analysis has distinctive advantages in separating the magnetic signal from phonon and other spurious processes .
as an example , we show in figure 3 the polarization analysis of the spin excitations for yba@xmath3cu@xmath4o@xmath50 at temperatures above and below @xmath8 . because magnetic fluctuations in yba@xmath3cu@xmath4o@xmath5 are centered around @xmath29 with a sinusoidal modulation along the @xmath121 direction
, we oriented the crystal in the @xmath108 zone and searched for the magnetic signal with scans along @xmath122 which corresponds to the maximum intensity of the acoustic modulation . figures 3(a - c ) show the raw data of sf scattering for @xmath123 , 15 , and 24 mev at 60 and 11 k. the nsf scattering of the same scans are shown in figs .
3(d - f ) . while sf scattering clearly peaks around @xmath29 , there are no similar features in the nsf scattering at the same position .
therefore , it is clear that the scattering around @xmath29 in the sf scattering are magnetic in origin . to quantitatively parameterize the sf scattering data
, we used a gaussian line shape on a linear background to fit the data as shown in solid lines in figs .
3(a - c ) @xcite .
the scattering function @xmath124 for such line shape is related to the in - plane dynamical spin susceptibility @xmath105 by @xmath125 } , \label{eq : three}\end{aligned}\ ] ] where @xmath126 , @xmath127 is the peak susceptibility and @xmath128 is related to the half width at half maximum ( hwhm ) of the excitations via @xmath129 .
although polarization analysis is excellent in identifying the magnetic nature of gaussian peaks around @xmath29 , the limited flux available with the technique means that detailed wave vector and energy dependence of the scattering need to be obtained with unpolarized neutrons . for this purpose
, we realigned the sample in the @xmath116 $ ] and @xmath117 $ ] zone .
figure 4 shows the constant-@xmath120 scans at various temperatures at @xmath29 .
the spectra consist of two major peaks at @xmath130 and 31 mev .
while the intensity of the @xmath130 mev mode follows the bose population factor @xmath80 $ ] consistent with phonons , the scattering at @xmath131 mev decreases with warming which is indicative of its magnetic origin . as we shall demonstrate below , phonons are also present at @xmath132 mev .
however , phonon scattering in this energy range is expected to change negligibly with temperature for @xmath133 k. therefore , the difference spectra in fig .
4 can be regarded as changes in the dynamical susceptibility .
clearly , there are spectral weight enhancements around 30 mev with decreasing temperature above and below @xmath8 . if we define the susceptibility gain on cooling above @xmath8 at the resonance energy as the precursor of the resonance , we find that the spectral weight of the resonance increases slightly from the normal to the superconducting state [ fig .
4(b ) and ( c ) ] and is only partially compensated by the loss at energies above it [ fig .
4(c ) ] .
figure 5 shows the momentum dependence of the response at energies above and below the magnetic resonance . at @xmath134 mev ,
the scattering [ fig .
5(a ) ] is featureless around @xmath29 , suggesting the presence of a spin gap in the superconducting state @xcite . on increasing the energy to 10 mev
, the scattering shows a broad peak that can be well described by a single gaussian centered at @xmath29 . at @xmath135 mev ,
the profile around @xmath29 develops a flattish top similar to previous observations @xcite .
although detailed analysis suggests that the profile is better described by a pair of peaks rather than a single gaussian , well defined incommensurate peaks at @xmath136 rlu from @xmath29 are only observed at @xmath137 mev [ fig .
5(d ) ] . at the resonance energy of @xmath131 mev ,
commensurate profiles are observed above and below @xmath8 [ fig .
5(e ) ] . for an energy above the resonance ( at 38 mev ) ,
the peak appears to be commensurate in the normal state , but changes to a flat top ( incommensurate ) profile in the low - temperature superconducting state [ fig .
5(f ) ] . in this section , we discuss results on three samples with superconducting transition temperatures of 52 k , 62.7 k , and 74 k. the fractional hole concentration per cu atom in the cuo@xmath3 sheet , @xmath40 , for yba@xmath3cu@xmath4o@xmath5 around the 60-k plateau phase is close to @xmath138 @xcite . for the single - layer superconducting la@xmath139sr@xmath140cuo@xmath17 which also has @xmath141 ,
the incommensurability of the spin fluctuations is at @xmath142 @xcite .
the discovery of the same incommensurate structure and @xmath18 ( @xmath143 ) for the equivalent hole concentration ( @xmath144 ) of the bilayer yba@xmath3cu@xmath4o@xmath30 @xcite has generated considerable interests because such observation suggests that incommensurate spin fluctuations may be the common feature of these two most studied families of cuprates . for la@xmath15sr@xmath145cuo@xmath17
, it is generally believed that the presence of dynamic stripes is the microscopic origin of the observed incommensurate spin fluctuations @xcite .
thus , if the incommensurate spin fluctuations in yba@xmath3cu@xmath4o@xmath5 have similar behavior as la@xmath15sr@xmath145cuo@xmath17 at all doping concentrations , it is likely that dynamic stripes should also be present in the bilayer cuprates .
although the explicit incommensurate structure @xcite and the one - dimensional nature of the incommensurate spin fluctuations in yba@xmath3cu@xmath4o@xmath30 @xcite favor the dynamic striped - phase interpretation , the detailed energy and doping dependence of the incommensurate spin fluctuations is still lacking . in the discussions below
, we attempt to remedy this situation .
we first describe measurements on yba@xmath3cu@xmath4o@xmath51 .
although the rocking curves of all other crystals of yba@xmath3cu@xmath4o@xmath5 show single gaussian peaks with mosaic spreads of @xmath2812.5 degrees , the rocking curve around @xmath146 bragg reflection for yba@xmath3cu@xmath4o@xmath51 displays two distinct peaks separated by @xmath283 degrees .
this means that the yba@xmath3cu@xmath4o@xmath51 sample is composed of two major grains @xmath283 degrees apart .
figure 6 summarizes constant-@xmath120 scans at various temperatures at @xmath29 for yba@xmath3cu@xmath4o@xmath51 . while the @xmath2820 mev phonons change negligibly from those for yba@xmath3cu@xmath4o@xmath50 ( fig .
4 ) , there are two features in the raw spectra of fig .
6(a ) worth noting .
first , the resonance energy shifted from @xmath2830.5 mev for yba@xmath3cu@xmath4o@xmath50 to @xmath2832 mev for yba@xmath3cu@xmath4o@xmath51 . in addition , the spectral weight of the resonance relative to that of the 20 mev phonons is clearly enhanced in yba@xmath3cu@xmath4o@xmath51 .
this observation indicates that the intensity gain of the resonance in this material below @xmath8 is considerably bigger than that for yba@xmath3cu@xmath4o@xmath50 .
similar to yba@xmath3cu@xmath4o@xmath50 , there are magnetic spectral weight enhancements around the resonance energy above and below @xmath8 .
high intensity , low - resolution measurements in the @xmath147 $ ] zone confirm this conclusion ( fig .
7 ) . to determine the magnitude of the spin gap and
search for incommensurate spin fluctuations , we performed constant - energy scans at various energies along the @xmath148 $ ] direction above and below @xmath8 as shown in fig .
8 . at 10 and 16 mev ,
the scattering shows weak features around @xmath29 .
this is different from yba@xmath3cu@xmath4o@xmath50 where spin fluctuations centered around @xmath29 are found at energies above 10 mev ( figs . 3 and 5 ) . at 24 mev [ fig .
8(c ) ] , the scattering displays a flattish top in both the normal and superconducting states @xcite . at energies near [ 27 mev , fig .
8(d ) ] and at the resonance [ 31.5 mev , fig .
8(e ) ] , the scattering narrows in width and increases in intensity . for an energy above the resonance , flattish top profiles indicative of incommensurate fluctuations
are observed at 40 mev as shown in fig .
8(f ) , which is similar to fig .
5(f ) of yba@xmath3cu@xmath4o@xmath50 .
next we describe measurements on yba@xmath3cu@xmath4o@xmath30 where the incommensurate spin fluctuations at 24 mev and the resonance at 34 mev have been extensively studied @xcite . for completeness , we plot in fig .
9 the difference spectra of constant-@xmath120 scans showing the presence of the 34 mev resonance above and below @xmath8 .
detailed temperature evolution of the resonance and the magnitude of the spin gap has already been reported @xcite . to establish the energy dependence of the incommensurate spin fluctuations
, we carried out high - resolution measurements using be(002 ) as the monochromator and fixed the final neutron energy at 14.78 mev .
figure 10 shows a series of constant - energy scans at energies below and above the 34 mev resonance .
consistent with previous results @xcite , constant - energy scans at 24 mev [ fig .
10(a ) ] show a pair of well - defined peaks separated by @xmath149 rlu in the low - temperature superconducting state and a flattish top profile just above @xmath8 .
although the same scans at 27 mev and 30 mev also exhibit the same double peak structure below @xmath8 [ figs .
10(b ) and ( c ) ] , the separation between the peaks or the incommensurabilities clearly decrease with increasing energy .
in addition , the normal state flat top profile at 24 mev is replaced by a single gaussian peak centered around @xmath29 at 30 mev . as energy increases , the scattering profile continues to sharpen , with the narrowest width and highest intensity profile occuring at the resonance energy [ fig .
10(e ) ] . when moving to energies above the resonance [ figs .
10(f - h ) ] , the profiles increase in width and decrease in intensity . at 41 mev ,
the scattering becomes incommensurate in both the normal and superconducting states [ fig .
10(h ) ] .
finally , we discuss measurements for yba@xmath3cu@xmath4o@xmath31 .
figure 11 shows the result of constant-@xmath120 scans above and below @xmath8 . as the energy of the resonance moves up to 37 mev , the phonon scattering at 31 mev becomes more prominent .
constant - energy scans at several energies below the resonance are shown in fig .
in contrast to the well - defined double peaks for yba@xmath3cu@xmath4o@xmath30 [ fig .
10(a ) ] , the scattering at 24 mev and 27 mev [ figs .
12(a ) and ( b ) ] shows a very weak structure above and below @xmath8 .
for yba@xmath3cu@xmath4o@xmath30 , the susceptibility increases on cooling from the normal to the superconducting state at the incommensurate positions , accompanied by a suppression of fluctuations at the commensurate point @xcite . in an attempt to determine whether the same is true for yba@xmath3cu@xmath4o@xmath31
, we plot the difference spectra between 11 k and 80 k. as emphasized in an earlier study @xcite , these difference spectra can be simply regarded as changes in the dynamical susceptibility , i.e. , @xmath150 .
figures 12(d ) and ( e ) show the outcome of this subtraction for 24 mev and 27 mev , respectively . for @xmath137 mev
, the difference spectrum shows negative scattering centered around @xmath29 with no evidence of susceptibility gain at the expected incommensurate positions .
this is consistent with the opening of a 24 mev gap in the spin fluctuations spectra below @xmath8 . at @xmath151 mev ,
both the raw data [ fig .
12(b ) ] and the difference spectrum show evidence of incommensurate spin fluctuations at @xmath152 rlu , consistent with previous measurements using an integrated technique that gives a incommensurability of @xmath153 @xcite . however , the clarity of incommensurate spin fluctuations is weak as compared to that for yba@xmath3cu@xmath4o@xmath30 .
experiments on a different crystal of yba@xmath3cu@xmath4o@xmath31 by arai _
@xcite show clear incommensurate spin fluctuations at @xmath153 rlu . at energies close to the resonance ( @xmath154 mev and 37 mev ) , the scattering [ figs .
12(c ) and ( f ) ] shows conventional gaussian profiles .
much of the discussion in this section will be focused on our analysis of separating phonon and spurious events from magnetic scattering for energies below the resonance .
while the magnetic signal usually spreads throughout the energy and momentum space at high temperatures and becomes unobservable at any particular @xmath1-@xmath155 point , the intensity of phonons should follow the bose statistics with increasing temperature . therefore
, careful measurements as a function of temperature and wave vector should allow the distinction of magnetic scattering from phonons .
as an example , we show in fig . 13 raw constant-@xmath155 scans at the wave vector @xmath156 rlu at various temperatures . the comparison of the results for temperatures above and below @xmath8 [ fig .
13(b ) ] indicates a clear enhancement of the scattering at @xmath157 mev , consistent with the presence of the resonance .
in addition to the enhancement of the resonance , there are general intensity reductions in the scattering below @xmath131 mev from the normal to the superconducting state .
such intensity reductions at low energies are likely due to the drop in the single- and multi- phonon scattering although a simultaneous reduction in magnetic signal can not be ruled out . nevertheless , the strong enhancement of the scattering at @xmath158 @xmath159 , @xmath132 , and @xmath160 mev at 200 k [ fig .
13(a ) ] indicates that the peaks in the constant-@xmath155 scans at these energies are mostly phonons in origin . to demonstrate the enhancement of the magnetic scattering on cooling above and below @xmath8
, we show the difference spectra between low and high temperatures in two experimental geometries in figs .
14 and 15 .
the data reveal two important features : first , the resonance in this underdoped compounds occurs at @xmath36 mev , almost identical to the optimally and overdoped doped compound @xcite ; second , there is an enhancement of the magnetic signal around the resonance energy above @xmath8 , consistent with the presence of a precursor for the resonance in the underdoped yba@xmath3cu@xmath4o@xmath5 @xcite . to search for incommensurate spin fluctuations in yba@xmath3cu@xmath4o@xmath33 , we scanned along the wave vector @xmath148 $ ] direction at energies below and above the resonance .
figure 16 summarizes the low - resolution raw data at temperatures just above and well below @xmath8 . at @xmath135 mev [ fig .
16(a ) ] , the scattering shows a sloped background with minimum intensity at around @xmath29 in both the normal and superconducting states .
this suggests the existence of a normal state spin gap of 16 mev . on increasing the energy to @xmath137 mev [ fig .
16(b ) ] , the intensity of the broad peak around @xmath29 in the normal state is somewhat reduced below @xmath8 with no clear evidence of an incommensurate structure .
a similar behavior is also observed at @xmath161 mev [ figs .
the scattering profiles between the normal and superconducting states are almost indistinguishable at @xmath162 mev and 52 mev [ figs .
16(d ) and ( f ) ] . near the resonance energy [ fig .
16(e ) ] , the scattering can be well described by gaussians on linear backgrounds and shows a clear increase in intensity on entering the superconducting state from the normal state .
in addition , the scattering at 200 k ( open squares ) shows a broad peak around @xmath29 , which indicates weak magnetic scattering at this temperature . since the scattering at 200 k is mostly nonmagnetic for energies below 39 mev [ fig .
16(e ) ] , we systematically subtracted 200 k data from that obtained at lower temperatures .
figures 17(c - f ) show the outcome of such procedure , and the net intensity gain around @xmath29 above the general reduction in the multi - phonon background should represent the enhancement of magnetic scattering at low temperatures . for @xmath163 mev [ fig .
17(c ) ] , the broad peak in the normal state vanishes below @xmath8 , which suggests the enlargement of the normal state spin gap in the superconducting state . on increasing the energy to @xmath162 mev
, the scattering shows little change from the normal state to the superconducting state [ figs .
17(d ) ] . in addition
, the maximum scattering intensity no longer seems to peak at @xmath29 , but shows weak evidence for incommensurate scattering .
unfortunately , the poor resolution of the measurements does not allow a conclusive identification of incommensurate spin fluctuations . at @xmath164 mev
17(e ) ] , the enhancement of resonance in fig .
16(e ) is confirmed .
finally for an energy above the resonance ( @xmath165 mev ) , the scattering shows no observable change above and below @xmath8 [ fig .
17(f ) ] .
because the intensity of low energy phonons is more sensitive to the increasing temperature and the bose population factor , the subtraction procedure described above does not work for low energy excitations .
however , we can compare the difference spectra at temperatures just above and below @xmath8 . at @xmath135 mev [ fig .
17(a ) ] , the difference spectra between 95 k ( @xmath166 k ) and 10 k ( @xmath167 ) show no identifiable feature which is consistent with the presence of a normal state spin gap .
similar subtraction at @xmath137 mev shows a clear peak around @xmath29 [ fig .
17(b ) ] which indicates the suppression of 24 mev spin fluctuations in the superconducting state .
therefore , the normal state spin fluctuations in yba@xmath3cu@xmath4o@xmath33 consist of a gap ( @xmath168 mev ) and broad excitations centered around @xmath29 . in the superconducting state , the normal state spin gap increases in magnitude , and the intensity of the low energy fluctuations is suppressed . to further determine the magnitude of the spin gap in the superconducting state
, we carried out high - resolution measurements with the neutron final energy fixed at @xmath169 mev .
figures 18(a - c ) show constant - energy scans at @xmath137 , 30 , and 39 mev .
the difference spectra at @xmath137 and 30 mev between low and high temperatures are shown in figs .
18(d - f ) .
careful temperature dependent measurements suggest that the weak structure marked by arrows in the low temperature raw data of figs .
18(a ) and ( b ) is not magnetic in origin .
the opening of the spin gap at @xmath170 mev in the superconducting state is demonstrated by the difference spectra of figs .
18(e ) and ( f ) , thus confirming the result of fig .
17 . finally , we present results on the incommensurate spin fluctuations at energies just below the resonance in yba@xmath3cu@xmath4o@xmath33 .
these high - resolution measurements were carried out on the hb-1 triple - axis spectrometer with @xmath171 mev . following the procedures established above , we show in fig .
19 the temperature difference spectra at @xmath162 , 34 , 37 , and 39 mev above and below @xmath8 .
incommensurate spin fluctuations were observed below @xmath8 at @xmath162 , 34 , and 37 mev . as in the case of yba@xmath3cu@xmath4o@xmath30 ,
the incommensurability of these fluctuations decreases with increasing energy and becomes commensurate at the resonance energy .
in addition , the incommensurability of yba@xmath3cu@xmath4o@xmath33 at @xmath172 mev is essentially the same as in the more underdoped yba@xmath3cu@xmath4o@xmath30 at 24 mev [ fig .
however , the normal state spin fluctuations of yba@xmath3cu@xmath4o@xmath33 are clearly commensurate and centered around @xmath29 whereas the fluctuations in yba@xmath3cu@xmath4o@xmath30 show a flattish top indicative of the incommensurability at temperatures above @xmath8 .
therefore , the spin fluctuation spectrum in yba@xmath3cu@xmath4o@xmath33 has a gap in the normal state ( @xmath173 mev ) , a superconducting gap ( @xmath174 mev ) , a resonance at @xmath175 mev , and commensurate - to - incommensurate transition for spin fluctuations below the resonance on entering the superconducting state . in the studies of
ideally doped and overdoped yba@xmath3cu@xmath4o@xmath5 , the important issues are the nature of the normal state magnetic scattering and the evolution of the incommensurate spin fluctuations .
to determine whether there is detectable magnetic scattering around @xmath176 mev in the normal state of yba@xmath3cu@xmath4o@xmath34 and yba@xmath3cu@xmath4o@xmath35 , we performed a series of constant-@xmath155 scans at @xmath29 at various temperatures below and above @xmath8 .
figures 20 and 21 display the low temperature minus high temperature difference spectra for yba@xmath3cu@xmath4o@xmath34 and yba@xmath3cu@xmath4o@xmath35 , respectively . in both compounds , we find the resonance peak at @xmath157 mev below @xmath8 and no trace or reminiscence of the resonance above @xmath8 as seen in all of the underdoped materials .
careful constant - energy scans for yba@xmath3cu@xmath4o@xmath35 at @xmath161 , 30 , 32 and 40 mev are shown in figs .
22(a - c ) .
the temperature difference spectra in figs .
22(d - f ) yield a spin gap value of @xmath162 mev in the superconducting state and show no discernible normal state magnetic scattering @xcite . to further elucidate the nature of the incommensurate spin fluctuations
, we performed a series of constant - energy scans at energies just below the 40 mev resonance for highly doped yba@xmath3cu@xmath4o@xmath35 .
figure 23 summarizes the outcome below and above @xmath8 at @xmath154 , 35.5 , 37 and 40 mev .
clear incommensurate spin fluctuations were found at @xmath154 , 35.5 , and 37 mev in the low temperature superconducting state .
surprisingly , the incommensurability of the fluctuations at @xmath154 mev is @xmath177 rlu , which is indistinguishable from all the other underdoped compounds with @xmath52 .
the incommensurability changes to @xmath178 and @xmath179 rlu for @xmath180 mev and 37 mev , respectively .
although there are weak features in the constant - energy scans just above @xmath8 [ figs .
23(a - d ) ] , careful temperature dependent studies of the constant - energy profiles indicate that these features are nonmagnetic in origin [ fig .
23(e ) , open circles ] . thus
, incommensurate fluctuations in highly doped yba@xmath3cu@xmath4o@xmath5 appear only in the low temperature superconducting state at energies close to the resonance .
furthermore , the incommensurabilities of the low energy spin fluctuations are insensitive to the doping for yba@xmath3cu@xmath4o@xmath5 with @xmath52 and saturate at @xmath32 . in contrast , incommensurate spin fluctuations in the single - layer la@xmath15sr@xmath16cuo@xmath17 near optimal doping survive to temperatures well above @xmath8 @xcite , and the incommensurabilities of the low energy spin fluctuations increase smoothly to the saturation value of @xmath181 for materials with optimal doping @xcite .
the existence of incommensurate spin fluctuations now appears to be a common feature of the yba@xmath3cu@xmath4o@xmath5 family of superconductors @xcite .
such fluctuations have also been observed in sr@xmath182- and oxygen- doped samples of la@xmath183cuo@xmath17 @xcite .
thus , it becomes clear that incommensurability is the common feature between these two classes of the most studied superconductors .
however , before attempting any quantitative comparison of the similarities between yba@xmath3cu@xmath4o@xmath5 and la@xmath15sr@xmath145cuo@xmath17 , we must first consider the number of holes ( @xmath40 ) in the cuo@xmath3 plane in these two systems .
for la@xmath15sr@xmath145cuo@xmath17 , yamada and coworkers @xcite found an excellent linear relationship between @xmath18 and the effective hole concentration up to around @xmath184 .
moreover , the incommensurability @xmath18 is energy independent for @xmath185 mev @xcite .
the situation for yba@xmath3cu@xmath4o@xmath5 is more subtle . using the established approximate parabolic relationship @xmath186 , where @xmath187 is the maximum transition temperature of the system , and @xmath40 is the hole - doping @xcite
, we can calculate the effective hole doping @xmath40 of yba@xmath3cu@xmath4o@xmath5 from their @xmath8 values .
however , as shown in figs .
5 , 10 , 19 , and 23 , the incommensurability in yba@xmath3cu@xmath4o@xmath5 decreases with increasing energy close to the resonance . as a consequence ,
it is difficult to compare directly the doping dependence of the incommensurability in these two families of materials as @xmath18 is not a well defined quantity for yba@xmath3cu@xmath4o@xmath5 . in order to make such comparison , we chose to systematically use the @xmath18 for the incommensurate low frequency spin fluctuations furthest away in energy from the resonace . in figure 24
, we plot the incommensurability @xmath18 obtained in such procedure as a function of @xmath40 . for completeness , we also included the data of arai _ et al . _
( open square ) for yba@xmath3cu@xmath4o@xmath31 @xcite and very recent data of bourges _ et al . _
( filled square ) for yba@xmath3cu@xmath4o@xmath188 @xcite . as we can see from the figure , the incommensurability increases initially with doping , but it saturates quickly at @xmath152 for hole doping @xmath189 .
this result is different from la@xmath15sr@xmath145cuo@xmath17 where the incommensurability is found to be linear with @xmath18 until reaching @xmath190 at large doping @xcite . for la@xmath15sr@xmath145cuo@xmath17
, the saturation of incommensurability above @xmath191 appears to be related to the saturation of the density of o(@xmath192)-type holes in the cuo@xmath3 plane as suggested from optical measurements @xcite . in the case of yba@xmath3cu@xmath4o@xmath5 , the incommensurability saturates for @xmath52 whereas the hole concentration @xmath40 in the cuo@xmath3 plane continues to increase with increasing oxygen doping and @xmath8 @xcite .
therefore , the saturation of the incommensurability in yba@xmath3cu@xmath4o@xmath5 may have a different microscopic origin from that of la@xmath15sr@xmath145cuo@xmath17 .
although it is interesting to compare the incommensurability as a function of effective hole concentration in the cuo@xmath3 plane of yba@xmath3cu@xmath4o@xmath5 and la@xmath15sr@xmath145cuo@xmath17 , a more robust test of the similarities in these two families of materials can be made by comparing the relation between @xmath8 and @xmath18 because these two quantities are free from any uncertainties associated with the doping level or oxygen stoichiometry .
for la@xmath15sr@xmath145cuo@xmath17 , yamada _
et al . _ found a linear relationship between @xmath8 and @xmath18 up to the optimal doping regime @xcite . using @xmath193 , balatsky and
bourges @xcite extracted a velocity @xmath194 mev . for yba@xmath3cu@xmath4o@xmath5 ,
these authors suggested a similar linear relation between @xmath8 and the @xmath120 width in momentum space , @xmath195 ( hwhm ) , for the whole doping range : @xmath196 , with @xmath197 mev .
they then claimed strong similarities in the magnetic states of yba@xmath3cu@xmath4o@xmath5 and la@xmath15sr@xmath145cuo@xmath17 @xcite . in fig .
25 , we plot the relation between @xmath8 and the measured incommensurability @xmath18 for yba@xmath3cu@xmath4o@xmath5 in the whole doping range . for superconducting transition temperatures below @xmath2860 k ,
our data are consistent with the proposed linear relation between @xmath8 and @xmath18 with @xmath198 mev @xcite .
however , the incommensurability is essentially unchanged for all samples with @xmath8 larger than 60 k. therefore , if there were excitations associated with such a velocity , its magnitude would approach infinity for samples with transition temperatures above 60 k. one of the salient features of the spin excitations spectra in yba@xmath3cu@xmath4o@xmath5 is the presence of spin gaps in the normal and superconducting states .
for example , the low - energy spin fluctuations of yba@xmath3cu@xmath4o@xmath33 show well defined gaussian peaks centered around @xmath29 at energies above 24 mev in the normal state .
on cooling the system to below @xmath8 , the normal state scattering is suppressed which indicates the opening of a gap in the spin fluctuation spectrum ( figs .
17 and 18 ) . in figure 26
, we plot the magnitude of the spin gap in the low temperature superconducting state as a function of @xmath8 for the whole doping range .
the solid line shows the linear relation between @xmath8 and the magnitude of the spin gap ( @xmath7 ) up to the optimal doping with @xmath9 .
this empirical result connects directly the spin gap and the superconductivity .
the value of @xmath9 is not far away from the @xmath199 for the weak - coupling limit of the bcs theory of superconductivity , where @xmath200 is the superconducting gap . for samples with low oxygen doping and low @xmath8 ,
the linear relation between @xmath8 and @xmath7 breaks down , possibly due to the oxygen doping inhomogeneity and broad superconducting transitions in weakly - doped materials . among the many interesting features observed in the spin excitations spectra of yba@xmath3cu@xmath4o@xmath5 ,
perhaps the most - discussed is the collective acoustic spin excitation named the `` resonance '' .
first reported by rossat - mignod _ et al . _
@xcite for yba@xmath3cu@xmath4o@xmath201 , the resonance at @xmath2840 mev in the highly doped yba@xmath3cu@xmath4o@xmath5 was shown to be magnetic in origin , narrow in energy @xcite , and to appear exclusively below @xmath8 @xcite . in subsequent experiments @xcite ,
the resonance was also found in underdoped yba@xmath3cu@xmath4o@xmath5 with the resonance energy ( @xmath12 ) approximately scaling with @xmath8 : @xmath202 .
consistent with previous results @xcite , the coherence length of the resonance ( @xmath203 ) is weakly doping dependent except for yba@xmath3cu@xmath4o@xmath30 which has considerably larger @xmath203 than for all the other dopings ( fig .
the coherence length of the incommensurate spin fluctuations ( fig .
28 ) also appears to be weakly doping dependent .
in addition , we note that on approaching the optimal oxygen doping for yba@xmath3cu@xmath4o@xmath5 , the resonance energy ( @xmath12 ) is insensitive to the changing @xmath8 ( fig .
therefore , the linear relationship between @xmath12 and @xmath8 appears to break down for materials with oxygen concentrations close to its optimal doping . finally , we compared the peak intensity of the lowest - energy observable incommensurate fluctuations with that of the resonance ( fig .
while the ratio clearly decreases with increasing doping for @xmath204 , it saturates to a value of @xmath28 0.25 for @xmath189 .
one interesting aspect of the resonance peak is its unusual temperature dependence . for optimal ( @xmath10 ) and over doped ( @xmath205 ) samples ,
the onset of the resonance occurs at a temperature @xmath206 which almost coincides with @xmath8 @xcite . for underdoped @xmath207 and 0.6
, @xmath206 increases to temperatures well above @xmath8 while both @xmath8 and the resonance energy itself are reduced @xcite . by empirically comparing the cross - over temperature @xmath206 and the pseudogap temperature obtained by other techniques @xcite , we associate the initial occurrence of the resonance with the pseudogap temperature @xcite . in
recent published @xcite and unpublished @xcite papers , bourges _
_ @xcite claimed that there is no justification for a separation of the normal state spin excitations spectrum into resonant and nonresonant parts .
although these authors agreed that the broad maximum of the spin susceptibility in the normal state occurs at the same energy as the resonance peak in some underdoped materials @xcite , they argued that `` the apparent equivalence of the normal state energy and the resonance peak energy breaks down in underdoped samples closer to optimal doping '' @xcite .
here we point out that the claims by bourges _
are incorrect . as can be seen from our results in sec .
iiia13 , there are clear enhancements in the spin susceptibility around the resonance energy at temperatures above @xmath8 for all the underdoped compounds including yba@xmath3cu@xmath4o@xmath33 .
these observations are in sharp contrast to the claims of bourges _ et al .
_ @xcite . moreover , from the temperature dependence of the resonance intensity as a function of doping @xcite , we find that the intensity of the resonance changes smoothly and shows progressively larger pretransitional regime above @xmath8 as a function of decreasing doping . as a consequence ,
it is natural to associate the enhancement of the spin susceptibility above @xmath8 around the resonance energy in underdoped materials as the precursor of the resonance .
we also emphasize that the temperature dependence of the resonance @xcite is closely related to the thermodynamics of yba@xmath3cu@xmath4o@xmath5 near @xmath8 @xcite and hence the superconducting condensation energy @xcite .
if the electronic entropy changes around @xmath8 are mostly due to spin excitations @xcite , they must respond to an external magnetic field in qualitatively the same way @xcite . in a conventional type - ii
bcs superconductor @xcite , the effect of a field is to simply shift the superconducting specific heat anomaly ( @xmath208 ) to lower temperatures and to reduce its amplitude until the upper critical field , @xmath209 , is reached @xcite . for high-@xmath8 superconductors such as yba@xmath3cu@xmath4o@xmath5 , the electronic specific heat anomaly near @xmath8
is drastically suppressed by a moderate field along the @xmath70-axis while it is little affected by the same field parallel to the @xmath210-plane @xcite . in a recent experiment on yba@xmath3cu@xmath4o@xmath30 @xcite , we discovered that a moderate magnetic field ( @xmath211 t ) along the direction approximately perpendicular to the cuo@xmath3 planes ( @xmath70-axis ) can drastically suppress the intensity of the resonance at 34 mev while leaving other scattering relatively unaltered .
similar , but much smaller effect also occurs for the same field parallel to the cuo@xmath3 planes . our results thus provide further support for the intimate connection between the magnetic excitation spectrum and thermodynamics of high-@xmath8 superconductors @xcite .
the persistence of a field effect above @xmath8 is consistent with the precursor of the resonance in the normal state of the underdoped yba@xmath3cu@xmath4o@xmath5 .
fermi - liquid theory , based on the premise that electron correlation effects in the metallic state of matter correspond to the noninteracting `` quasiparticles '' on a fermi surface , has been successful in solving many of the important problems in condensed - matter physics , including conventional superconductivity . in the case of copper - oxides , where high-@xmath8 superconductivity is obtained by hole - doping of the af insulating parent compounds , an important class of models @xcite argues that electrons in these materials can not be described by quasiparticles and fermi - liquid theory .
rather , the electrons have been incorporated in complex , inhomogeneous ( fluctuating ) patterns of the charge and spin confined to separate linear regions in the crystal resembling `` stripes '' @xcite . since their discovery @xcite
, the incommensurate spin fluctuations in the superconducting la@xmath15sr@xmath16cuo@xmath17 have been interpreted as due to the presence of dynamic spin stripes @xcite or as the consequence of spin - flip scattering across a nested fermi surface @xcite . in 1995 ,
tranquada and coworkers @xcite found evidence supporting the former interpretation in a closely related material la@xmath212nd@xmath213sr@xmath214cuo@xmath17 .
for nd substitution of @xmath215 , the distortion of the crystal structure causes the condensation of the incommensurate spin fluctuations in nd - free la@xmath15sr@xmath16cuo@xmath17 into a spin density wave ( sdw ) at the same wave vector @xcite .
the analysis of the elastic magnetic sdw peaks and their corresponding charge - ordering peaks associated with the modulation of the underlying lattice indicate the presence of a static striped phase in la@xmath212nd@xmath213sr@xmath214cuo@xmath17 .
this strongly suggests that the incommensurate fluctuations in superconducting la@xmath15sr@xmath16cuo@xmath17 are a fluctuating version ( dynamic ) of the stripes .
however , if the stripe picture is universal for all high-@xmath8 superconductors , it should also be able to explain the incommensurate spin fluctuations of yba@xmath3cu@xmath4o@xmath5 . in the case of la@xmath15sr@xmath16cuo@xmath17 ,
the incommensurability saturates at `` 1/8 '' . whereas the incommensurability of the low - energy spin fluctuations in yba@xmath3cu@xmath4o@xmath5 follows the expected linear behavior as a function of @xmath8 for low doping materials @xcite but abruptly saturates at `` 1/10 '' for @xmath52
this could mean that the minimum stripe spacing in yba@xmath3cu@xmath4o@xmath5 is five unit cells compared to four in la@xmath15sr@xmath16cuo@xmath17 @xcite . however , the reduction of the incommensurability with increasing energy close to the resonance in yba@xmath3cu@xmath4o@xmath5 , first reported by arai _ et al . _ @xcite and
subsequently confirmed by bourges _
@xcite in underdoped compounds , is different from the energy independent incommensurability of the low - frequency spin fluctuations in la@xmath15sr@xmath16cuo@xmath17 @xcite .
in addition , the incommensurate fluctuations in la@xmath15sr@xmath16cuo@xmath17 are observed at temperatures well above @xmath8 @xcite .
such fluctuations in yba@xmath3cu@xmath4o@xmath5 are present above @xmath8 for underdoped compounds , but for highly doped materials they are only found close to the resonance in the superconducting state . at present
, it remains unclear how to explain the observed behavior differences in these two families of materials from a stripe model .
in particular , the saturation of the incommensurability at @xmath32 for effective hole concentration @xmath189 ( fig .
24 ) is puzzling . assuming that the incommensurate spin fluctuations in yba@xmath3cu@xmath4o@xmath5 at all doping levels are due to dynamic stripes composed of the hole - rich metallic regions separated by the antiphase insulating domains @xcite
, one would expect the growth of the metallic stripes with increasing doping at the expense of the insulating domains @xcite . as a consequence ,
the incommensurability should increase with increasing hole doping which is in contrast to the observation of a saturating @xmath18 for yba@xmath3cu@xmath4o@xmath5 with @xmath20 .
although we have emphasized the differences , the incommensurate spin fluctuations for these two classes of materials also have remarkable similarities in their structure and doping dependence for oxygen concentrations @xmath21 ( @xmath204 ) . these similarities and the one - dimensional nature of the incommensurate spin fluctuations in underdoped yba@xmath3cu@xmath4o@xmath30 @xcite and the associated phonon anomaly
@xcite are consistent with the stripe formation in yba@xmath3cu@xmath4o@xmath5 for effective hole doping below 0.1 . to truly understand the microscopic origin of the incommensurate spin fluctuations in different cuprates using a stripe model ,
perhaps one needs to first explain the presence of resonance in the bilayer compounds , which takes much of the weight of the total magnetic scattering , and its absence in the single - layer materials @xcite . for highly doped materials ,
the incommensurate spin fluctuations in yba@xmath3cu@xmath4o@xmath5 appears to be intimately related to the resonance as they occur only at energies very close to the resonance . from the magnetic field dependence of the spin excitations @xcite
, it is clear that the resonance measures the pairing and long - range phase coherence .
thus , if a stripe model can explain the resonance , it may also be able to account for the differences in the incommensurability in la@xmath15sr@xmath16cuo@xmath17 and yba@xmath3cu@xmath4o@xmath5 . alternatively , assuming that the observed incommensurate fluctuations in yba@xmath3cu@xmath4o@xmath5 come from a nested fermi surface @xcite
, one may ask whether such models can predict the observed intensity and doping dependent behavior .
unfortunately , there are no explicit predictions about the doping dependence of the incommensurability and its relationship with the resonance from a nested fermi surface model that can be directly compared with our experiments .
therefore , it is still not clear whether the observed incommensurate fluctuations in yba@xmath3cu@xmath4o@xmath5 are consistent with the fermi - liquid description at all doping levels . in particular , the abrupt saturation of the incommensurability for yba@xmath3cu@xmath4o@xmath5 with @xmath52
suggests that the physics accounting for this behavior may have a different microscopic origin from that of the lower doping compounds .
since the discovery of the resonance in highly doped and underdoped yba@xmath3cu@xmath4o@xmath5 @xcite , many fundamentally different microscopic mechanisms have been proposed to explain its origin . in the earlier theoretical studies @xcite ,
the resonance is interpreted as the consequence of the @xmath103-wave gap symmetry of the cuprate superconductors .
subsequent workers have noticed that @xmath103-wave superconductivity alone is insufficient to explain the resonance , and that very strong coulomb correlations among electrons must also play an important role . in some models
, the band structure anomaly effects alone can produce a peak in spin susceptibility that grows with the opening of the superconducting gap @xcite , while other workers have applied a random - phase approximation ( rpa ) to treat the effects of strong coulomb correlations with the lindhard function appropriate for either a fermi liquid @xcite or a spin - charge separated metal @xcite .
in particular , the resonance has been treated as a disordered magnon - like collective mode in the particle - hole channel whose energy is bound by the superconducting energy gap @xcite . using a completely different approach @xcite , the resonance is viewed as a pseudo - goldstone boson mode in the particle - particle channel instead of the more conventional particle - hole channel . in the framework of this theory , antiferromagnetism and @xmath103-wave superconductivity in copper oxides
are treated in equal footing @xcite , and resonance is directly responsible for the superconducting condensation energy @xcite . while the purpose of this article is not to review the similarities and differences among various models of the resonance , we wish to make several general remarks in light of the results described here and in recent field - dependent studies of the resonance @xcite .
first , it is now clear that the incommensurability and the resonance are inseparable parts of the general features of the spin dynamics in yba@xmath3cu@xmath4o@xmath5 at all dopings .
thus any microscopic model used to describe the resonance must also be able to explain the doping and energy dependence of the incommensurate spin fluctuations .
while current theoretical models do not provide explicit predictions that can be directly compared our experiments @xcite , it is hoped that the work described here will stimulate future theoretical efforts in this direction .
the remarkable field dependence of the resonance @xcite indicates that it measures the superconducting pairing and phase coherence . in particular , the larger @xmath70-axis field effect is difficult to understand within the framework of the interlayer tunneling theory , according to which the resonance should be strongly affected by the in - plane field that disrupts the coherent josephson coupling along @xmath70-axis @xcite . on the other hand ,
the field data @xcite are consistent with mechanism where the dominant loss of entropy on entering the superconducting state is due to the growth of magnetic correlations in the cuo@xmath3 planes @xcite .
finally , if the resonance is a magnon - like collective mode @xcite , it must not be a spin - wave in the conventional sense as the intensity of spin - wave excitations in ferromagnets and antiferromagnets should be magnetic field independent for the relatively small fields used @xcite .
in this article , we have described in detail the extensive polarized and unpolarized neutron measurements of the dynamical magnetic susceptibility @xmath0 of superconducting yba@xmath3cu@xmath4o@xmath5 above and below the transition temperature @xmath8 . using a careful polarization and temperature dependence analysis
, we find that the resonance and incommensurate spin fluctuations are general features of the spin dynamical behavior in yba@xmath3cu@xmath4o@xmath5 for all oxygen doping levels .
we establish the doping dependence of the incommensurability , the resonance , and the superconducting spin - gap energy . in the superconducting state , we find that the spin - gap energy is proportional to @xmath13 , but the resonance energy itself deviates from the linear behavior with @xmath13 for oxygen doping close to the optimal values .
similarly , we show that the doping dependence of the incommensurability saturates to @xmath19 for oxygen doping @xmath52 . we compare the observed behaviors in the two most studied superconducting copper - oxides la@xmath15sr@xmath16cuo@xmath17 and yba@xmath3cu@xmath4o@xmath5 and discuss their similarities and differences .
we stress that any comprehensive theoretical understanding of the spin dynamical behavior in these two families of materials must take into account the fact that the resonance so clearly seen in the bilayer yba@xmath3cu@xmath4o@xmath5 is absent in the single - layer la@xmath15sr@xmath16cuo@xmath17 . for underdoped yba@xmath3cu@xmath4o@xmath5 , the evolution of the resonance as a function of doping
is illustrated .
in particular , we find the enhancement of the susceptibility at the resonance energy above @xmath8 for all underdoped compounds including yba@xmath3cu@xmath4o@xmath33 . if we define the `` resonance '' as the enhancement of the susceptibility in a small energy range of the magnetic excitations spectra as a function of decreasing temperature , it is natural to associate the susceptibility gain around the resonance energy above @xmath8 in the underdoped compounds as the precursor of the resonance .
this assertion is supported by the systematic studies of the evolution of the temperature dependence of the resonance that show the progressively larger pretransitional regime above @xmath8 for more underdoped yba@xmath3cu@xmath4o@xmath5 @xcite .
furthermore , the resonance and incommensurate fluctuations appear to be intimately connected .
thus , any microscopic model for the magnetic fluctuations in yba@xmath3cu@xmath4o@xmath5 must be able to account for both the incommensurate spin fluctuations and the resonance .
we are grateful to g. aeppli , s. chakravarty , t. egami , v. j. emery , s. m. hayden , s. a. kivelson , d. k. morr , d. pines , d. j. scalapino , j. zaanen , and s. c. zhang for many enlightening discussions .
we also wish to thank j. a. fernandez - baca , jiandi zhang , and s. c. zhang for a critical reading of the manuscript , to thank j. r. thompson and k. j. song for measuring the superconducting transition temperatures for all the samples discussed in this article , and to thank b. c. chakoumakos and t. b. lindemer for allowing the use of their unpublished data . finally , we acknowledge much technical support over the years by r. g. maple , s. a. moore , g. b. taylor , and c. m. redmon .
this work was supported by u.s .
doe under contract de - ac05 - 00or22725 with ut - battelle , llc .
d. vaknin , s. k. sinha , d. e. moncton , d. c. johnston , j. m. newsam , c. r. safinya , and h. e. king , jr .
* 58 * , 2802 ( 1987 ) .
j. m. tranquada , d. e. cox , w. kunnmann , h. moudden , g. shirane , m. suenaga , p. zolliker , d. vaknin , s. k. sinha , m. s. alvarez , a. j. jacobson , and d. c. jonhston , phys .
lett . * 60 * , 156 ( 1988 ) .
h. yoshizawa , s. mitsuda , h. kitazawa , and k. katsumata , j. phys .
jpn , * 57 * , 3686 ( 1988 ) .
r. j. birgeneau , y. endoh , y. hidaka , k. kakurai , m. a. kastner , t. murakami , g. shirane , t. r. thurston , and k. yamada , phys .
b * 39 * , 2868 ( 1989 ) .
cheong , g. aeppli , t. e. mason , h. a. mook , s. m. hayden , p. c. canfield , z. fisk , k. n. clausen , and j. l. martinez , phys .
* 67 * , 1791 ( 1991 ) .
t. e. mason , g. aeppli , and h. a. mook , phys .
lett . * 68 * , 1414 ( 1992 ) .
t. r. thurston , p.
m. gehring , g. shirane , r. j. birgeneau , m. a. kastner , y. endoh , m. matsuda , k. yamada , h. kojima , and i. tanaka , phys .
b * 46 * , 9128 ( 1992 ) .
k. yamada , c. h. lee , k. kurahashi , j. wada , s. wakimoto , s. ueki , h. kimura , y. endoh , s. hosoya , g. shirane , r. j. birgeneau , m. greven , m. a. kastner , and y. j. kim , phys .
b * 57 * , 6165 ( 1998 ) .
k. yamada , s. wakimoto , g. shirane , c. h. lee , m. a. kastner , s. hosoya , m. greven , y. endoh , and r. j. birgeneau , phys .
lett . * 75 * , 1626 ( 1995 ) .
b. lake , g. aeppli , t. e. mason , a. schroder , d. f. mcmorrow , k. lefmann , m. isshiki , m. nohara , h. takagi , and s. m. hayden , nature ( london ) * 400 * , 43 ( 1999 ) .
for a review , see m. a. kastner , r. j. birgeneau , g. shirane , and y. endoh , rev .
phys . * 70 * , 897 ( 1998 ) .
j. rossat - mignod , l. p. regnault , c. vettier , p. bourges , p. burlet , j. bossy , j. y. henry , and g. lapertot , physica ( amsterdam ) * 185c * , 86 ( 1991 )
. j. m. tranquada , p. m. gehring , g. shirane , s. shamoto , and m. sato , phys .
b * 46 * , 5561 ( 1992 ) .
h. a. mook , m. yethiraj , g. aeppli , t. e. mason , and t. armstrong , phys .
. lett . * 70 * , 3490 ( 1993 ) .
b. j. sternlieb , j. m. tranquada , g. shirane , m. sato , and s. shamoto , phys .
b * 50 * , 12915 ( 1994 ) .
h. f. fong , b. keimer , p. w. anderson , d. reznik , f. do@xmath216an , and i. a. aksay , phys .
lett . * 75 * , 316 ( 1995 ) .
p. bourges , l. p. regnault , y. sidis , and c. vettier , phys .
b * 53 * , 876 ( 1996 ) .
h. f. fong , b. keimer , d. reznik , d. m. milius , and i. a. aksay , phys . rev .
b * 54 * 6708 ( 1996 ) .
p. dai , m. yethiraj , h. a. mook , t. b. lindemer , and f. do@xmath216an , phys .
* 77 * 5425 ( 1996 ) . h.
f. fong , b. keimer , f. do@xmath216an , and i. a. aksay , phys .
lett . * 78 * , 713 ( 1997 ) .
p. bourges , h. f. fong , l. p. regnault , j. bossy , c. vettier , d. l. milius , i. a. aksay , and b. keimer , phys
b * 56 * , r11439 ( 1997 ) .
p. dai , h. a. mook , and f. do@xmath216an , phys .
. lett . * 80 * , 1738 ( 1998 ) .
h. a. mook , p. dai , r. d. hunt , and f. do@xmath216an , j. phys .
solids * 59 * , 2140 ( 1998 ) .
h. a. mook , p. dai , s. m. hayden , g. aeppli , t. g. perring , and f. do@xmath216an , nature ( london ) * 395 * , 580 ( 1998 ) . m. arai , t. nishijima , y. endoh , t. egami , s. tajima , k. tomimoto , y. shiohara , m. takahashi , a. garret , and s. m. bennington , phys . rev
. lett . * 83 * , 608 ( 1999 ) .
s. m. hayden , g. aeppli , p. dai , h. a. mook , t. g. perring , s .- w .
cheong , z. fisk , f. do@xmath216an , and t. e. mason .
physica b * 241 - 243 * , 765 ( 1998 ) .
p. dai , h. a. mook , s. m. hayden , g. aeppli , t. g. perring , r. d. hunt , and f. do@xmath216an , science * 284 * , 1344 ( 1999 ) .
d. j. scalapino , science * 284 * , 1282 ( 1999 ) . to facilitate the comparison of the results between la@xmath15sr@xmath16cuo@xmath17 and yba@xmath3cu@xmath4o@xmath5
, we adopted the definition of incommensurability @xmath18 used in [ 9 ] .
note that this definition of incommensurability @xmath18 is different ( by a factor of 2 ) from that in [ 24 ] . c. michel and b. raveau , j. solid state chem .
* 43 * , 73 ( 1982 ) .
r. gopalan , t. rajasekharan , t. roy , g. rangarajan , v. ganesan , and r. srinivasan , j. mater . res . *
11 * , 2406 ( 1996 ) . t. b. lindemer , ( unpublished ) .
t. b. lindemer , j. f. hunley , j. e. gates , a. l. sutton , j. brynestard , c. r. hubbard , p. k. gallagher , j. am .
* 72 * , 1775 ( 1989 ) .
p. dai , j. zhang , h. a. mook , s .- h .
liou , p. a. dowben , and e. w. plummer , phys .
b * 54 * , r3694 ( 1996 )
. j. d. jorgensen , b. w. veal , a. p. paulikas , l. j. nowicki , g. w. grabtree , h. claus , and w. k. kwok , phys .
b * 41 * , 1863 ( 1990 ) .
e. altendorf , x. k. chen , j. c. irwin , r. liang , and w. n. hardy , phys .
b * 47 * , 8140 ( 1993 ) .
b. c. chakoumakos and t. b. lindermer , ( unpulished work ) .
s. w. lovesey , in _ theory of thermal neutron scattering from condensed matter _
( clarendon , oxford , 1984 ) , vol .
2 . r. m. moon , t. riste , and w. c. koehler , phys . rev . * 181 * , 920 ( 1969 ) . strictly speaking ,
this equation is only true if the efficiency of the mezei coil is 100%
. s. m. hayden , g. aeppli , t. g. perring , h. a. mook , f. do@xmath216an , phys .
b * 54 * , r6905 ( 1996 ) d. reznik , p. bourges , h. f. fong , l. p. regnault , j. bossy , c. vettier , d. l. milius , i. a. aksay , and b. keimer , phys . rev .
b * 53 * , r14741 ( 1996 ) . a. v. balatsky and p. bourges , phys
lett . * 82 * , 5337 ( 1999 ) .
although there is also a spin pseudogap in the normal state of yba@xmath3cu@xmath4o@xmath5 , we did not systematically investigate its temperature and doping dependence because the difficulty in separating the low - energy magnetic scattering from phonons and other spurious processes . j. l. tallon , c. bernhard , h. shaked , r. l. hitterman , and j. d. jorgensen , phys .
b * 51 * , 12911 ( 1995 ) .
j. m. tranquada , b. j. sternlieb , j. d. axe , y. nakamura , and s. uchida , nature ( london ) * 375 * , 561 ( 1995 ) ; j. m. tranquada , j. d. axe , n. ichikawa , y. nakamura , s. uchida , and b. nachumi , phys .
b * 54 * , 7489 ( 1996 ) . h. a. mook , p. dai , f. do@xmath216an , and r. d. hunt , nature ( london ) * 404 * , 729 ( 2000 ) .
although one expects incommensurate spin fluctuations at this doping level , detection of explicit incommensurability may require samples to have single grain with narrow mosaic spread .
the mosaic of yba@xmath3cu@xmath4o@xmath51 is considerably larger than other crystals in this study . as a consequence ,
we only an observe broad flattish top peak around @xmath29 at @xmath137 mev [ fig .
8(c ) ] . the incommensurability was determined by fitting the 24 mev profile with a pair of gaussian peaks separated by @xmath18 rlu from @xmath29 .
g. aeppli , t. e. mason , s. m. hayden , h. a. mook , and j. kulda , science * 278 * , 1432 ( 1997 ) .
although we have demonstrated the clear presence incommensurate spin fluctuations for yba@xmath3cu@xmath4o@xmath5 with @xmath217 , 0.6 , 0.7 , 0.8 , and 0.95 , no explicit incommensurability is observed for our yba@xmath3cu@xmath4o@xmath51 sample and the clarity of incommensurate spin fluctuations appears to be sample dependent . at present , we do not understand the microscopic origin for such behavior .
perhaps the detailed incommensurate structure is very sensitive to the oxygen uniformity of the material .
p. bourges , y. sidis , h. f. fong , l. p. regnault , j. bossy , a. ivanov , and b. keimer , science * 288 * , 1234 ( 2000 ) .
s. uchida , t. ido , h. takai , t. arima , y. tokura , and s. tajima , phys .
b * 43 * , 7942 ( 1991 ) .
t. ito , k. takenaka , s. uchida , phys .
lett . * 70 * , 3995 ( 1993 ) .
b. wuyts , v. v. moshchalkov , y. bruynseraede , phys .
b * 53 * , 9418 ( 1996 ) . m. takigawa , a. p. reyes , p. c. hammel , j. d. thompson , r. h. heffner , z. fisk , and k. c. ott , phys . rev .
b * 43 * , 247 ( 1991 ) ; h. alloul , a. mahajan , h. casalta , and o. klein , phys . rev . lett . *
70 * , 1171 ( 1993 ) ; m. horvati , t. auler , c. berthier , y. berthier , p. butaud , w. g. clark , j. a. gillet , p. sgransan , and j. y. henry , phys . rev .
b * 47 * , 3461 ( 1993 ) .
h. f. fong , p. bourges , y. sidis , l. p. regnault , j. bossy , a. ivanov , d. l. milius , i. a. aksay , and b. keimer , phys . rev .
b * 61 * , 14733 ( 2000 ) .
p. bourges , b. keimer , l. p. regnault , and y. sidis , cond - mat/0006085 ( unpublished ) .
p. bourges , in _ the gap symmetry and fluctuations in high temperature superconductors , _ edited by j. bok , g. deutscher , d. pavuna , and s. a. wolf ( plenum press , 1998 ) p349 .
j. w. loram , k. a. mirza , j. r. cooper , and w. y. liang , phys .
lett . * 71 * , 1740 ( 1993 ) .
d. j. scalapino and s. r. white , phys . rev .
b * 58 * , 8222 ( 1998 ) . e. demler and s. c. zhang , nature ( london ) * 396 * , 733 ( 1998 ) .
s. chakravarty and h. y. kee , phys .
b * 61 * , 14 821 ( 2000 ) .
m. r. norman , m. randeria , b. jank , and j. c. campuzano , phys .
b * 61 * , 14742 ( 2000 ) .
b. jank , cond - mat/9912073 ( unpublished ) .
j. r. shrieffer , _ theory of superconductivity _ , ( benjamin , new york , 1964 ) .
a. junod , in _ studies of high temperature superconductors _ , vol .
a. v. narlikar , nova science publishers , n. y. 1996 , page 1 .
a. junod , a. erb , and ch .
renner , physica c * 317 * , 333 ( 1999 ) .
p. dai , h. a. mook , g. aeppli , s. m. hayden , and f. do@xmath216an , nature ( london ) * 406 * , 965 ( 2000 ) . j. zaanen and o. gunnarson , phys .
b * 40 * , 7391 ( 1989 ) .
j. schulz , j. phys .
( paris ) * 50 * , 2833 ( 1989 ) .
d. poilblanc , and t. m. rice , phys .
b * 39 * , 9749 ( 1989 )
. s. a. kivelson and v. j. emery , in _ strongly correlated electronic materials : the los alamos symposium 1993 _ ( eds k. s. bedell _ et al .
_ , addison - wesley , maine , 1994 ) , p 619 .
for a recent review on advances in the physics of high-@xmath8 superconductivity , see j. orenstein and a. j. millis , science * 288 * , 468 ( 2000 ) .
j. zaanen , j. phys .
solids * 59 * , 1769 ( 1998 ) .
v. j. emery , s. a. kivelson , and o. zahar , phys .
b * 56 * , 6120 ( 1997 ) .
s. r. white and d. j. scalapino , phys .
* 80 * , 1272 ( 1998 ) . g. b. martins , c. gazza , j. c. xavier , a. feiguin , and e. dagotto , phys .
lett . * 84 * , 5844 ( 2000 ) .
b. bulut , d. hone , d. j. scalapino , and n. e. bickers , phys .
lett . * 64 * , 2466 ( 1990 ) .
q. m. si , y. zha , k. levin , and j. p. lu , phys .
b * 47 * , 9055 ( 1993 ) ; y. zha , k. levin , and q. m. si , phys . rev .
b * 47 * , 9124 ( 1993 )
. p. b. littlewood , j. zaanen , g. aeppli , and h. monien , phys .
b * 48 * , 487 ( 1993 ) .
j. brinckmann , and p. a. lee , phys .
lett . * 82 * , 2915 ( 1999 ) .
d. k. morr and d. pines , phys .
81 * , 1086 ( 1998 ) ; phys .
b * 61 * , 6483 ( 2000 ) .
y. j. kao , q. si , and k. levin , phys .
b * 61 * , 11898 ( 2000 ) .
m. r. norman , phys .
b * 61 * , 14751 ( 2000 ) . h. a. mook and f. do@xmath216an , nature ( london ) * 401 * , 145 ( 1999 ) .
k. maki and h. won , phys .
lett . * 72 * , 1758 ( 1994 ) .
m. j. lercher and j. m. weatley , phys .
b * 49 * , 736 ( 1994 ) . m. lavagna and g. stemmann , phys . rev .
b * 49 * , 4235 ( 1994 ) ; g. stemmann , c. ppin , and m. lavagna , phys . rev .
b * 50 * , 4075 ( 1994 ) .
f. onufrieva and j. rossat - mignod , phys .
b * 52 * , 7572 ( 1995 ) .
g. blumberg , b. p. stojkovic , and m. v. klein , phys .
b * 52 * , 15741 ( 1995 ) .
a. a. abrikosov , phys .
b * 57 * , 8656 ( 1998 ) .
lan yin , s. chakravarty , and p. w. anderson , phys .
lett . * 78 * , 3559 ( 1997 ) .
n. bulut and d. j. scalapino , phys .
b * 53 * , 5149 ( 1996 ) .
d. z. liu , y. zha , and k. levin , phys .
lett . * 75 * , 4130 ( 1995 ) .
i. i. mazin and v. m. yakovenko , phys .
rev . lett.*75 * , 4134 ( 1995 ) .
a. j. millis and h. monien , phys .
b * 54 * , 16172 ( 1996 ) .
t. tanamoto , h. kohno , and h. fukuyama , j. phys .
. jpn . * 61 * 1886 ( 1992 ) .
b. normand , h. kohno , and h. fukuyama , j. phys .
* 64 * , 3903 ( 1995 ) .
a. abanov and a. v. chubukov , phys .
lett . * 83 * , 1652 ( 1999 ) .
s. sachdev , c. buragohain , and m. vojta , science * 286 * , 2479 ( 1999 ) .
e. demler and s. c. zhang , phys .
lett . * 75 * , 4126 ( 1995 ) ; e. demler , h. kohno , and s. c. zhang , phys .
b * 58 * , 5719 ( 1998 ) . s. c. zhang , science * 275 * , 1089 ( 1997 ) .
after the completion of this article , we noted a theoretical attempt to reconcile the differences between spin fluctuations in la@xmath15sr@xmath16cuo@xmath17 and yba@xmath3cu@xmath4o@xmath5 .
k. k. voo , h. y. chen , and w. c. wu , cond - mat/0006312 ( unpublished ) . | polarized and unpolarized neutron triple - axis spectrometry was used to study the dynamical magnetic susceptibility @xmath0 as a function of energy ( @xmath1 ) and wave vector ( @xmath2 ) in a wide temperature range for the bilayer superconductor yba@xmath3cu@xmath4o@xmath5 with oxygen concentrations , @xmath6 , of 0.45 , 0.5 , 0.6 , 0.7 , 0.8 , 0.93 , and 0.95 .
the most prominent features in the magnetic spectra include a spin gap in the superconducting state , a pseudogap in the normal state , the much - discussed resonance , and incommensurate spin fluctuations below the resonance .
we establish the doping dependence of the spin gap in the superconducting state , the resonance energy , and the incommensurability of the spin fluctuations .
the magnitude of the spin gap ( @xmath7 ) up to the optimal doping is proportional to the superconducting transition temperature @xmath8 with @xmath9 .
the resonance , which exists exclusively below @xmath8 for highly doped yba@xmath3cu@xmath4o@xmath5 with @xmath10 and 0.95 , appears above @xmath8 for underdoped compounds with @xmath11 .
the resonance energy ( @xmath12 ) also scales with @xmath13 , but saturates at @xmath14 mev for @xmath6 close to 0.93 .
the incommensurate spin fluctuations at energies below the resonance have structures similar to that of the single - layer superconducting la@xmath15sr@xmath16cuo@xmath17 . however , there are also important differences .
while the incommensurability ( @xmath18 ) of the spin fluctuations in la@xmath15sr@xmath16cuo@xmath17 is proportional to @xmath8 for the entire hole - doping range up to the optimal value , the incommensurability in yba@xmath3cu@xmath4o@xmath5 increases with @xmath8 for low oxygen doping and saturates to @xmath19 for @xmath20 .
in addition , the incommensurability decreases with increasing energy close to the resonance .
finally , the incommensurate spin fluctuations appear above @xmath8 in underdoped compounds with @xmath21 but for highly doped materials they are only observed below @xmath8 .
we discuss in detail the procedure used for separating the magnetic scattering from phonon and other spurious effects . in the comparison of our experimental results with various microscopic theoretical models
, particular emphasis was made to address the similarities and differences in the spin fluctuations of the two most studied superconductors .
finally , we briefly mention recent magnetic field dependent studies of the spin fluctuations and discuss their relevance in understanding the microscopic origin of the resonance . |
99 j. marro and r. dickman , _ nonequilibrium phase transitions in lattice models _ ( cambridge university press , cambridge , 1999 ) . h. hinrichsen , advances in physics * 49 * , 815 ( 2000 ) ; m. henkel , h. hinrichsen , and s. lbeck , _ non - equilibrium phase transitions _ , berlin : springer ( 2008 ) .
p. grassberger , z. phys .
b * 47 * , 365 ( 1982 ) ; h. k. janssen , z. phys .
b * 42 * , 151 ( 1981 ) . j. l. cardy and r. l. sugar , j. phys .
a : math . gen . * 13 * l423 ( 1980 ) .
a. lipowski and m. lopata , phys .
e * 60 * , 1516(1999 ) .
m. basu , u. gayen , and p. k. mohanty , _ arxiv:1102.1631 _ ; a. ghosh , u. basu , a. chakraborti , and b. k. chakrabarti , phys .
e * 83 * , 061130 ( 2011 ) .
i. jensen , phys .
lett . * 70 * , 1465 , ( 1993 ) ; i. jensen and r. dickman , phys .
e * 48 * , 1710 ( 1993 ) ; j. f. f. mendes , r. dickman , m. henkel , and m. c. marques , j. phys .
a * 27 * , 3019 ( 1994 ) ; d. ben- avraham and j. kohler , j. stat . phys . * 65 * , 839 ( 1992 ) ; e. v. albano , j. phys .
a * 25 * , 2557 ( 1992 ) .
p. k. mohanty and d. dhar , phys .
* 89 * , 104303 ( 2002 ) ; physica a * 384 * , 34(2007 ) .
t. nagatani , jnl .
japan * 61 * , 3196 ( 1992 ) ; j. d. noh and h. park , phys .
lett . * 94 * , 145702 ( 2005 ) ; s. kwon and y. kim , phys .
e * 72 * , 066122 ( 2006 ) .
s. cornell , m. droz , r. dickman and m. c. marques , j. phys .
a : math . gen . * 24 * 5605 ( 1991 ) ; h. k. janssen , phys .
lett . * 78 * , 2890 ( 1997 ) ; j. stat .
phys.*103 * , 801 ( 2001 ) ; u. c. tuber , m. j. howard , and h. hinrichsen , phys .
. lett . * 80 * , 2165 ( 1998 ) ; j. l. cardy and u. c. tuber , phys .
lett . * 77 * , 4780 ( 1996 ) ; j. stat .
phys . * 90 * , 1 ( 1998 ) .
e. arashiro and t. tom , j. phys .
a : math . theor . *
40 * , 887 ( 2007 ) ; a. lipowski and d. lipowska , physica a * 276 * , 456 ( 2000 ) . | in a recent article [ arxiv:1108.5127 ] park has shown that the four - state predator - prey model studied earlier in _ j. stat .
mech , l05001 ( 2011 ) _ belongs to directed percolation ( dp ) universality class .
it was claimed that predator density is not a reasonable order parameter , as there are many absorbing states ; a suitably chosen order parameter shows dp critical behaviour . in this article
, we argue that the configuration that does not have any predator is the only dynamically accessible absorbing configuration , and the predator density too settles to dp critical exponents after a long transient .
^s__a systems having absorbing configurations may undergo a non - equilibrium phase transition @xcite from an active to an absorbing state .
the critical behavior of these absorbing state phase transitions ( apts ) @xcite depends on the the symmetry of the order parameter and presence of additional conservation laws .
it has been conjectured @xcite that in absence of any special symmetry the apt belongs to the directed percolation ( dp ) universality class as long as the system has a single absorbing state .
since the coarse grained microscopic theory of dp , which is a birth - death - diffusion process , is based on a single component reggeon field theory@xcite , critical behavior in presence of additional field is expected to alter the critical behavior .
the additional field may bring in multiple absorbing states and/or additional conservation laws .
presence of multiple absorbing states may @xcite or may not @xcite affect the universality .
coupling of order parameter to a conserved field too lead to dp @xcite or non - dp @xcite critical behavior .
the models of directed percolation with more than one species @xcite , which brings in additional coarse grained fields , has also been studied @xcite .
the predator - prey cellular automaton models @xcite in higher dimension too shows an apt to an absorbing ( extinct ) state which belongs to dp - class .
the role of additional fields in these models are not quite well understood .
recently we studied a predator - prey model@xcite on a @xmath0-dimensional lattice , where each lattice site is either vacant , occupied by a predator @xmath1 , a prey @xmath2 or both ( one @xmath1 and one @xmath2 ) . in these four state predator - prey ( 4spp ) model growth of preys and death of predators occurs independently , whereas death of a prey is always followed by instant birth of a predator .
based on the numerical simulations and estimated critical exponents , we have suggested the possibility of a new universality class . in particular , the decay of clusters at the critical point was found to be distinctly different from those of dp .
however , in a recent article park @xcite has suggested a different scenario .
it was claimed that the predator density @xmath3 can not be taken as a order parameter as there are infinitely many absorbing states .
the transition is found to be in dp class , when order parameter is chosen suitably . in this article , we show that although there are many absorbing states , only one of them @xmath4 is dynamically accessible . in fact
the order parameter @xmath3 which was showing an apparently new critical behavior , slowly crosses over to dp . for completeness , first let us define the model . on a one dimensional periodic lattice ,
each site is either vacant , or occupied by a single particle @xmath1 ( prey ) , or occupied by a single particle @xmath2 ( predator ) or by both particles ( co - existing @xmath1 and @xmath2 ) ; correspondingly at each site @xmath5 we have @xmath6 alternatively one may describe the 4-state predator - prey ( 4spp ) model considering two separate branches , one for @xmath1 and the other for @xmath2 particles , where particles living in one branch can not move to the other .
we consider only the asymmetric case of the model which evolves according to the following dynamics , where @xmath7 denotes both presence and absence of particles in respective branches .
( 100 , 35 ) ( -10,30)(-11,28.5)@xmath8(-16,30)(-18,28.5)@xmath1 ( -10,24)(-12,22.5)@xmath7(-16,24)(-18,22.5)@xmath7 ( -7,25)@xmath9(-5,28)@xmath10 ( 10,30)(8,28.5)@xmath1(4,30)(2,28.5)@xmath1 ( 10,24)(8,22.5)@xmath7(4,24)(2,22.5)@xmath7 ( 40,30)(38,28.5)@xmath7(34,30)(32,28.5)@xmath7 ( 40,24)(38,22.5)@xmath2(34,24)(32,22.5)@xmath2 ( 43,25)@xmath9(45,28)@xmath11 ( 60,30)(58,28.5)@xmath7(54,30)(52,28.5)@xmath7 ( 60,24)(59,22.5)@xmath8(54,24)(53,22.5)@xmath8 ( 90,30)(88,28.5)@xmath1(84,30)(82,28.5)@xmath7 ( 90,24)(89,22.5)@xmath8(84,24)(82,22.5)@xmath2 ( 93,25)@xmath9(95,28)@xmath12 ( 110,30)(109,28.5)@xmath8(104,30)(102,28.5)@xmath7 ( 110,24)(108,22.5)@xmath2(104,24)(102,22.5)@xmath2 although , it was not explicitly mentioned in our earlier report @xcite , numerical simulations of the model was carried out in using exactly the same state variables @xmath13 mentioned in @xcite .
clearly these bond variables @xmath14 follow a dynamical rules , @xmath15 the neighbors are updated along with @xmath16 as follows .
for dynamics ( [ eq:1 ] ) @xmath17 is increased by @xmath18 whereas for dynamics ( [ eq:2 ] ) @xmath17 and @xmath19 are decreased by @xmath20 and @xmath21 respectively . clearly a bond @xmath5 is active when @xmath16 is either @xmath18 or @xmath22 , or it is greater than @xmath23 it was argued by park @xcite that the density of active bonds @xmath24 , the correct order parameter ( as there are infinitely many absorbing states ) of the system , vanishes at the critical point @xmath25 when @xmath26 and @xmath27 for the same parameter values , we had estimated earlier that predator density @xmath3 vanishes at @xmath28 .
this raises a question that possibly in the region @xmath29 , the system falls into an absorbing state which has isolated @xmath2s .
however a configuration with one or more isolated @xmath2 can be absorbing only when there are no @xmath1s .
thus , at @xmath30 we must have @xmath31 it was evident from fig .
2(a ) of ref . @xcite that @xmath32 this indicates that @xmath3 must vanish at the same value of @xmath33 possibly at @xmath34 as estimated by park @xcite .
vs @xmath35 for @xmath36 with @xmath37 , @xmath38 and @xmath39 . here
we use dp critical exponent @xmath40 ( b ) both @xmath41 and @xmath42 asymptotically decay as @xmath43 at @xmath44.,width=529 ] to check this we redo the monte - carlo simulation for larger system size @xmath39 and measured @xmath42 up to @xmath45 mcs for different values of @xmath33 while keeping @xmath26 and @xmath27 from fig .
[ fig : pc](a ) it is evident that @xmath46 shows a saturation at @xmath47 for @xmath48 figure [ fig : pc](b ) shows log - scale plot of @xmath41 and @xmath42 at @xmath49 it is evident that after a long transient @xmath42 decays as @xmath50 with @xmath51 , similar to @xmath52 such a change in @xmath53 to a lower value was appearing as a saturation in @xmath3 , leading to a lower estimate of the critical point .
vs @xmath35 at criticality ( @xmath54 @xmath37 and @xmath38)for @xmath55 and @xmath56 . the initial slope @xmath57 crosses over to @xmath58 as system size is increased .
( b ) the same data are collapsed according to eq .
( [ eq : rhot_l ] ) by using dp critical exponents @xmath59 and @xmath60,width=491 ] with this correct estimation of critical point , we proceed to calculate other critical exponents taking @xmath3 as the order parameter . as for finite system of size
@xmath61 , starting from a high density of predators , @xmath62 decays as @xmath50 , indicating a scaling form @xmath63 at the critical point , where @xmath64 is the dynamical critical exponent .
thus , @xmath65 for different values of @xmath61 are expected to collapse to a single function when plotted against @xmath66 .
figure [ fig : z_nu](a ) shows decay of @xmath67 for different @xmath61 starting from the configuration @xmath68 ; clearly small systems show an effective exponent @xmath57 which crosses over to @xmath69 as the system size is increased .
true finite size effect of the critical point sets in at a reasonably large @xmath70 the data collapse according to eq .
( [ eq : rhot_l ] ) is observed in fig .
[ fig : z_nu](b ) where we use the dp exponents @xmath71 and @xmath72 in summary , although there are many absorbing configurations in 4spp model , the numerical simulations suggests that only one of them @xmath73 is dynamically accessible .
the critical behavior of the absorbing transition can be well described by taking @xmath3 as an order parameter .
the earlier estimated values of the critical exponents vary slowly with system size and settle to dp values .
these studies truly emphasize the complications , difficulties and danger associated in numerical determination of universality class , when the possibility of a long transient is not ruled out .
_ acknowledgement : _ pkm would like to acknowledge m. basu and u. basu for doing independent monte - carlo simulations to confirm the results . |
we propose here a conjecture ( conjecture [ theconjecture ] below ) refining the stark conjecture @xmath2 ( tate s formulation @xcite ) in the function field case , and we verify our conjecture in the genus zero case .
of course @xmath2 in the function field case is a theorem due to deligne @xcite and independently to hayes @xcite .
the main novelty of our conjecture is that it involves _
two- _ rather than one - variable algebraic functions .
the conjecture is formulated in the powerful language and framework of tate s thesis .
ultimately this paper is inspired by remarkable examples of coleman @xcite and thakur @xcite . to motivate the paper we devote most of this introduction to the discussion of an example similar to the ones originally considered by coleman and by thakur .
we first discuss the example in `` raw form '' and then we explain how to look at it from the adelic point of view cultivated in this paper .
consider the smooth projective model @xmath3 of the affine plane curve @xmath4 over @xmath5 . on the surface @xmath6 ( product formed over @xmath5 )
consider the meromorphic function @xmath7 following the lead of coleman , it can easily be verified that @xmath8 by the theory of correspondences on curves it follows that the analogously formed sum of endomorphisms of the jacobian of @xmath9 vanishes .
thus we reprove following coleman by a strikingly elementary argument that the `` universal gauss sum '' @xmath10\ ] ] acts on the jacobian of @xmath9 in the same way as does the @xmath11 power frobenius endomorphism .
now view @xmath9 as a covering of the @xmath12-line over @xmath5 by setting @xmath13 still following coleman , notice that by specializing the second variable to @xmath14 in the example @xmath15 , we obtain a `` stark - like '' function in the extension @xmath16 .
we say `` stark - like '' because the functions arising this way do not have precisely the right properties to qualify as stark units in the sense of @xmath2 ; nonetheless , these functions would appear to be promising `` raw material '' for building stark units since their divisors and galois properties are evident .
what is yet more intriguing is that this approach gives insight into the variation of the stark unit as we vary the choice of completely split place .
it is coleman s brilliant idea to link two - variable algebraic functions to stark s conjecture in this way .
we turn to follow thakur s lead .
we have @xmath17\\ \deg a < n \end{subarray } } \frac{a+\beta t^n+t^{n+1}+\alpha / t}{a+t^n},\ ] ] for all @xmath18 , @xmath19 and integers @xmath20 , as can be verified by manipulating moore determinants . by definition ,
the left side of the equation is the pull - back of @xmath15 via the map @xmath21 the system of identities ( [ equation : thakurinterpolation ] ) uniquely characterizes our example @xmath15 , and illustrates an extremely valuable heuristic which we credit to thakur , namely : `` frobenius interpolation formulas '' with `` @xmath22-partial - products '' on the right side automatically have `` functions of coleman type '' on the left side .
we pause to remark that a fairly extensive body of theory and applications of phenomena of coleman / thakur type has already been developed .
for example , see @xcite , @xcite , @xcite , @xcite , @xcite ; the applications to transcendence theory have been particularly notable . for a survey with a wealth of examples , see ( * ? ? ?
* chaps . 4 and 8) .
while the main emphasis in the theory to this point has been on the case of the basefield @xmath23 , significant efforts to generalize the base have been made ( see @xcite and @xcite ) .
but a general framework for understanding and classifying all these examples has been lacking .
the main point of this paper is to rebuild the theory on the foundation of tate s thesis so that a clear ( albeit conjectural ) picture of the situation over a general basefield emerges .
we return to the analysis of our example @xmath15 .
we explain how to look at the example from the point of view peculiar to this paper .
put @xmath24 the group @xmath25 is a copy of @xmath26 and in particular is abelian . notice that @xmath15 is invariant under the diagonal action @xmath27 of @xmath25 on @xmath28 .
the example @xmath15 lives in a setting with somewhat subtle commutative algebra properties .
notice that ( i ) the ring extension @xmath29 is a finite etale extension of dedekind domains , ( ii ) the diagonal prime @xmath30 lies above the diagonal prime @xmath31 and ( iii ) the prime of the diagonally @xmath25-invariant subring of @xmath28 below @xmath32 has the same residue field as @xmath33 , namely @xmath34 .
the upshot is that the diagonal completion of @xmath35 coincides with the diagonal completion of the diagonally @xmath25-invariant subring of @xmath28 .
we now simply identify the latter with the former . under this identification the strange - looking formula @xmath36 makes sense .
now let s put global class field theory into the picture .
let @xmath37 be the adele ring of @xmath34 and for each idele @xmath38 , let @xmath39 be the corresponding idele norm .
let @xmath40 be the reciprocity law homomorphism of global class field theory , redefined the modern way , as in tate s article @xcite , so that uniformizers map to geometric frobenius elements and hence , for all @xmath41 algebraic over @xmath5 and @xmath38 , we have @xmath42 .
( we follow this rule throughout the paper . )
we make @xmath43 act on the perfect ring @xmath44{k}\otimes_{{{\mathbb{f}}}_q}\sqrt[q^\infty]{k}$ ] by the rule @xmath45 which makes sense since @xmath46{k}/ \sqrt[q^\infty]{k})$ ] . since the @xmath43-action defined above commutes with the diagonal action of @xmath25
, we may naturally view all the functions @xmath47 as elements of the perfection of the diagonal completion of @xmath35 .
we need a few more definitions before getting to the `` punchline '' .
let @xmath48 be the normalized additive valuation of @xmath35 giving the order of vanishing at the diagonal ideal , and let this additive valuation be extended to the perfection of the completion in evident fashion .
let @xmath49 be the maximal compact subring .
let @xmath50 be the unique place of @xmath34 at which @xmath12 has a pole .
let @xmath51 be the idele whose component at @xmath50 is @xmath52 and whose components elsewhere are @xmath53 . consider the @xmath54-valued schwartz function @xmath55 on @xmath37 , where @xmath56 is probabilist s notation for the function taking the value @xmath53 on the set @xmath1 and @xmath57 elsewhere .
we now get to the heart of the matter .
the divisor formula ( [ equation : colemandivisor ] ) can be rewritten as the formula @xmath58 holding for all @xmath38 , plus a further assertion to the effect that no primes of the dedekind domain @xmath28 other than those of the form @xmath59{k}\right)\;\;\;(a\in { { \mathbb{a}}}^\times)\ ] ] enter into the prime factorization of @xmath15 in @xmath28 . the interpolation formula ( [ equation : thakurinterpolation ] ) can be rewritten as the formula @xmath60 holding for all @xmath38 such that @xmath61 .
formulas ( [ equation : approximateconjecture1 ] ) and ( [ equation : approximateconjecture2 ] ) very strongly suggest that there s more to the coleman / thakur phenomenon than just a few isolated examples .
we now draw our discussion of the example @xmath15 to a close .
we turn to the task of describing our conjecture , in order at least to convey some its flavor .
we continue with similar notation .
let @xmath34 be any global field with constant field @xmath5 , let @xmath37 be the adele ring of @xmath34 , and let @xmath62 be the reciprocity law homomorphism of global class field theory .
we define a _ coleman unit _ to be an element of the fraction field of the `` big ring '' @xmath44{k^{{\mathrm{ab}}}}\otimes_{{{\mathbb{f}}}_q^{{\mathrm{ab}}}}\sqrt[q^\infty]{k^{{\mathrm{ab}}}}$ ] which is ( i ) invariant under the diagonal action @xmath63 of @xmath64 and ( ii ) a unit at every maximal ideal of the big ring not transformable under the action @xmath65 of @xmath43 to the diagonal ideal @xmath66 .
our conjecture associates to every @xmath67$]-valued schwartz - function @xmath68 on @xmath37 such that @xmath69 a coleman unit @xmath15 for which analogues of ( [ equation : approximateconjecture1 ] ) and ( [ equation : approximateconjecture2 ] ) hold , the latter in a more general formulation ( not subject to any restriction on @xmath70 ) relating the `` leading taylor coefficient '' to a gadget we call the _ catalan symbol_. the values of the catalan symbol are function field elements roughly analogous in structure to the catalan numbers @xmath71 .
the catalan symbol is defined in terms of a further gadget we call the _ rational fourier transform_. the characteristic feature of the theory of rational fourier transforms is that the function @xmath72 is assigned the role usually played by a nonconstant complex - valued character of @xmath5 . as mentioned above , we prove our conjecture in genus zero .
we believe that ideas from the author s papers @xcite and @xcite suitably combined have a fighting chance to prove our conjecture in general .
perhaps the rather different set of ideas from @xcite could also yield a proof .
we hope for a proof of the conjecture in the general case not so heavily computational as the proof we give here in the genus zero case . to make the point that our conjecture refines @xmath2 , we provide a deduction of the latter from the former in [ section : starkrecovery ] below .
in fact a large fraction of the paper is devoted to developing the machinery needed to make the deduction of @xmath2 from our conjecture run smoothly .
in particular , the technical tool we call the _
adelic stirling formula _
( theorem [ theorem : adelicstirling ] below ) is crucial for proving our `` recipe '' ( theorem [ theorem : recipe ] below ) for the stark unit conditional on our conjecture .
we conclude the introduction by remarking on the organization of the paper . since we have provided a table of contents , we need nt make comments section - by - section .
we just offer the following advice . after glancing at [ subsection : generalnotation ] to be apprised of general notation and terminology ,
the reader could very well start in [ section : globalstirling ] , because that s where the main story - line of the paper begins ; the preceding sections , which are more or less independent of each other , could be treated as references .
but on the other hand , if the reader would take the trouble to work patiently through the sections before [ section : globalstirling ] , he / she would be prepared to hear the main story undistracted by technical issues of an essentially nonarithmetical nature .
[ [ section ] ] we denote the cardinality of a set @xmath1 by @xmath73 . [
[ section-1 ] ] given sets @xmath74 , let @xmath56 denote the real - valued function taking the value @xmath53 on @xmath1 and @xmath57 on @xmath75 ; we omit reference to @xmath76 in the notation because it can always be inferred from context .
we borrow this notation from the probabilists .
[ [ section-2 ] ] _ rings _ are always commutative with unit .
let @xmath77 denote the group of units of a ring @xmath78 .
let @xmath5 denote a field of @xmath79 elements .
[ [ section-3 ] ] a function @xmath80 on a ring @xmath78 taking values in @xmath81 is called an _ additive valuation _ if @xmath82 , @xmath83 and @xmath84 for all @xmath85 .
if @xmath78 is a field and @xmath86 , we say that @xmath80 is _
normalized_. additive valuations are said to be _ equivalent _ if proportional .
place _ of a field @xmath34 is an equivalence class of nontrivial additive valuations of @xmath34 .
[ [ section-4 ] ] given an integral domain @xmath78 of characteristic @xmath87 , let @xmath88 be the closure of @xmath78 under the extraction of @xmath89 roots , i. e. , the direct limit of the system @xmath90 of rings and homomorphisms . if @xmath91 , then we say that @xmath78 is _
perfect_. [ [ section-5 ] ] given a locally compact totally disconnected topological space @xmath76 , we denote by @xmath92 the _ schwartz space _ of locally constant compactly supported complex - valued functions on @xmath76 .
( in the literature this is sometimes called instead the _ schwartz - bruhat space _ , cf .
@xcite . ) given an @xmath5-algebra @xmath78 and ring elements @xmath93 , we define the _ moore determinant _
@xmath94 for this we have the well - known _
moore identity _ ( see ( * ? ? ? * chap . 1 ) ) @xmath95 for example , working over the field of rational functions in a variable @xmath12 with coefficients in @xmath5 , we have @xmath96\\ \deg a = n\\ a:\,\mbox{\tiny monic in}\,t \end{subarray}}a= \frac{\operatorname{{\mathrm{moore}}}(t^n,\dots,1)}{\operatorname{{\mathrm{moore}}}(t^{n-1},\dots,1)}=\prod_{i=0}^{n-1}(t^{q^n}-t^{q^i})\ ] ] for all nonnegative integers @xmath97 , where @xmath98 denotes the degree of @xmath99 as a polynomial in @xmath12 .
the last equality is justified by the vandermonde identity .
the following variant of the moore determinant will also be needed .
given an @xmath5-algebra @xmath78 , ring elements @xmath93 , and an @xmath5-linear functional @xmath100 defined on the @xmath5-span of @xmath101 , put @xmath102 we call @xmath103 the _ ore determinant _ of @xmath104 .
for this we have a variant of the moore identity which , for simplicity , we state under some special assumptions always fulfilled in practice . we suppose now that @xmath78 is a field , and that @xmath101 are @xmath5-linearly independent , in which case @xmath105 .
let @xmath106 be the @xmath5-linear span of @xmath101 .
we suppose further that @xmath107 .
then , so we claim , @xmath108 we call this relation the _ ore identity_. to prove the claim , first note that the ratio on the left side does not change if we replace @xmath101 by any @xmath5-basis of @xmath106
. we may therefore assume without loss of generality that @xmath109 for @xmath110 , in which case the claim follows from the moore identity ( [ equation : mooreidentity ] ) .
the claim is proved .
relation ( [ equation : oreidentity ] ) is a convenient way to restate what for our purposes is the main point of ore - elkies - poonen duality ( see @xcite ) .
we discuss in `` toy form '' some simple algebraic notions later to be applied in the adelic context .
we fix vector spaces @xmath106 and @xmath111 over @xmath5 of the same finite dimension .
we fix a perfect @xmath5-bilinear pairing @xmath112 we fix a nonconstant homomorphism @xmath113 where @xmath114 is the group of complex numbers of absolute value @xmath53 .
we also fix a field @xmath115 containing a copy of @xmath5 .
let @xmath116 be the set of @xmath117-valued functions on @xmath106 .
given @xmath118 , put @xmath119(v^*)=\sum_{v\in v}\phi(v)\lambda(-\langle v , v^*\rangle)\ ] ] for all @xmath120 , thus defining the _ toy fourier transform _
@xmath121\in { { \mathrm{sch}}}(v^*).\ ] ]
perhaps it is wasteful to define two notations for the fourier transform , but it is quite convenient , and we follow this practice throughout the paper for all the versions of the fourier transform that we consider .
for example , given @xmath122 and an @xmath5-subspace @xmath123 , we have @xmath124(v^*)={{\#}}w\cdot { { \mathbf{1}}}_{w^\perp}(v^*)\cdot\lambda(-\langle v_0,v^*\rangle)\ ] ] for all @xmath120 , where @xmath125 is the subspace of @xmath111 annihilated by @xmath126 .
further , we have @xmath127 for all @xmath118 and @xmath128 ; thus we can invert the toy fourier transform explicitly .
so far all this is completely familiar .
we begin to warp things a bit .
put @xmath130 then we have trivial identities @xmath131 holding for all @xmath132 . roughly speaking
, these identities say that although @xmath129 is not a character of @xmath5 ( save in the case @xmath133 ) , it behaves sufficiently like a character that we can base an alternative theory of fourier transforms on it . given @xmath118 , put @xmath134(v^*)=\sum_{v\in v}\phi(v)\lambda_0(\langle v , v^*\rangle)\ ] ] for all @xmath120 , thus defining the _ toy rational fourier transform _
@xmath135\in { { \mathrm{sch}}}(v^*).\ ] ] via ( [ equation : lambdaidentities ] ) we have @xmath136 for all @xmath137 ; thus we can express toy fourier transforms in terms of their rational analogues and _ vice versa_. for example , given @xmath122 and an @xmath5-subspace @xmath123 , we have @xmath138={{\#}}w\cdot \left({{\mathbf{1}}}_{w^\perp\cap[\langle v_0,\cdot\rangle=0]}-{{\mathbf{1}}}_{w^\perp\cap[\langle v_0,\cdot\rangle=1]}\right)\ ] ] by combining ( [ equation : toyfourierexample ] ) , ( [ equation : lambdaidentities ] ) and ( [ equation : toyrelationoffts ] ) .
the notation @xmath139 $ ] is probabilist - style shorthand for @xmath140 .
we use similar notation below without further comment . by combining the toy fourier inversion formula with ( [ equation : lambdaidentities ] ) and ( [ equation :
toyrelationoffts ] ) we have @xmath141 for all @xmath118 and @xmath128 ; thus we can invert the toy rational fourier transform explicitly . let @xmath67 $ ] be the ring obtained from @xmath54 by inverting @xmath142 .
let @xmath143 be the group of @xmath67$]-valued functions on @xmath106 .
notice that we have @xmath144={{\mathrm{sch}}}_0(v^*)$ ] .
it is because of the latter relation that @xmath145 deserves to be called rational .
recall that @xmath115 is a field containing @xmath5 .
let @xmath146 be an injective @xmath5-linear map . for all @xmath147 ,
put @xmath148 the definition makes sense because @xmath149 is a uniquely @xmath142-divisible group .
we call @xmath150 the _ toy catalan symbol_. it is clear that @xmath151 depends @xmath67$]-linearly on @xmath68 .
we see the values of the toy catalan symbol as being very roughly analogous in structure to the _ catalan numbers _
@xmath152}-{{\mathbf{1}}}_{(0,n+1]}-{{\mathbf{1}}}_{(0,n]})(x)},\ ] ] whence the terminology .
[ proposition : easymooreapp ] let @xmath146 be an injective @xmath5-linear map .
let @xmath123 be an @xmath5-subspace .
then the map @xmath153 is @xmath5-linear .
choose a basis @xmath154 for @xmath126 over @xmath5 .
the @xmath5-linear map @xmath155 coincides with the map in question . let @xmath146 be an injective @xmath5-linear map .
let @xmath123 be an @xmath5-subspace .
then the map @xmath156 } \end{array}\right)&\mbox{if $ v^*\not\in w^\perp$}\\ 0&\mbox{if $ v^*\in w^\perp$ } \end{array}\right.\right):v^*\rightarrow k\ ] ] is @xmath5-linear .
we may assume without loss of generality that @xmath157 .
choose an @xmath5-basis @xmath158 .
the @xmath5-linear map @xmath159 coincides with the map in question . [ proposition : dualityclincher ] let @xmath160 be an injective @xmath5-linear map .
let @xmath123 be an @xmath5-subspace .
fix @xmath161 .
then the map @xmath162 \end{array}\right)&\mbox{if $ v\not\in w$}\\ 0&\mbox{if $ v\in w$ } \end{array}\right.\right):v\rightarrow k\ ] ] is @xmath5-linear .
this delicate property of the toy rational fourier transform is the main formal justification for defining it as we have . in the adelic context
there will be a similar phenomenon ( see theorem [ theorem : genericbehavior ] below ) .
for @xmath163 we have by ( [ equation : toyrationalfourierexample ] ) and the definitions that @xmath164 \end{array}\right)^{1/{{\#}}w}\\\\ = & \left(\begin{array}{c } \beta\\ { { \mathbf{1}}}_{w^\perp\cap[\langle v,\cdot \rangle=0 ] } -{{\mathbf{1}}}_{w^\perp\cap[\langle v,\cdot \rangle=1 ] } -{{\mathbf{1}}}_{w^\perp\cap[\langle v_0,\cdot \rangle=0 ] } + { { \mathbf{1}}}_{w^\perp\cap[\langle v_0,\cdot \rangle=1 ] } \end{array}\right)\\\\ \stackrel{\star}{=}&\left(\begin{array}{c } \beta\\ -{{\mathbf{1}}}_{w^\perp\cap[\langle v,\cdot \rangle\neq 0 ] } -{{\mathbf{1}}}_{w^\perp\cap[\langle v,\cdot \rangle=1 ] } + { { \mathbf{1}}}_{w^\perp\cap[\langle v_0,\cdot \rangle\neq 0 ] } + { { \mathbf{1}}}_{w^\perp\cap[\langle v_0,\cdot \rangle=1 ] } \end{array}\right)\\\\ = & \left(\begin{array}{c } \beta\\ -{{\mathbf{1}}}_{w^\perp\cap[\langle v,\cdot \rangle=1 ] } + { { \mathbf{1}}}_{w^\perp\cap[\langle v_0,\cdot \rangle=1 ] } \end{array}\right)^q , \end{array}\ ] ] which by the preceding lemma proves the result .
the star marks the main trick of the toy theory .
it is nothing more than the fact that for any three subsets @xmath78 , @xmath165 and @xmath9 of a set @xmath76 we have @xmath166 .
we shall use the trick again to make explicit calculations in [ section : genuszero ] .
we work out identities in which on the left side appear functions on a certain commutative affine group scheme over @xmath5 evaluated at certain special points , and in which on the right side appear expressions interpretable as values of naturally occurring toy catalan symbols .
these identities are used in the sequel for no purpose other than as an inputs to the proof in [ section : genuszero ] of the genus zero case of the conjecture to which we allude in the title of the paper .
some similar identities were studied and applied in the papers @xcite , @xcite and @xcite , but there are two novel features here .
namely , ( i ) we work out connections with duality and ( ii ) we allow for the possibility of infinite ramification at infinity . [ [ subsubsection : residueintroduction ] ] let @xmath168 be a variable . given an @xmath5-algebra @xmath169 , let @xmath170 be the ring consisting of all power series @xmath171 with coefficients @xmath172 vanishing for @xmath173 , and put @xmath174 .
let @xmath175\subset r((1/t))$ ] ( resp .
, @xmath176\subset r((1/t))$ ] ) be the subring consisting of power series @xmath171 such that @xmath177 unless @xmath178 ( resp . , @xmath179 ) .
when @xmath169 is a field , we denote by @xmath180 the field of rational functions in @xmath168 with coefficients in @xmath169 , and we identify @xmath170 with the completion of @xmath180 at the infinite place .
[ [ section-6 ] ] we define a representable group - valued functor @xmath167 of @xmath5-algebras @xmath169 by the rule @xmath181/m(t))^\times\times ( 1+(1/t)r[[1/t]]),\ ] ] where the inverse limit is extended over monic @xmath182 $ ] ordered by the divisibility relation , and the group law is induced by multiplication .
the commutative affine group scheme @xmath167 is the natural one to consider in connection with the problem of constructing the maximal abelian extension of a field of rational functions in one variable with coefficients in @xmath5 .
later , in [ section : genuszero ] , we discuss and apply the arithmetical properties of @xmath167 .
but now we focus on properties more in the line with classical symmetric function theory .
[ [ section-7 ] ] by definition , to give an @xmath169-valued point @xmath183 is to give a power series @xmath184\ ] ] and for each monic @xmath185 $ ] a congruence class @xmath186/m(t))^\times,\ ] ] subject to the constraints that @xmath187 for all monic @xmath188 $ ] . for definiteness
, we always choose the representative @xmath189 $ ] to be of least possible degree in its congruence class modulo @xmath190 .
[ [ section-8 ] ] we produce useful examples of @xmath169-valued points of @xmath167 as follows .
let @xmath191 $ ] be a monic polynomial of degree @xmath192 such that for all monic @xmath182 $ ] , the resultant of @xmath193 and @xmath190 is a unit of @xmath169 .
we then define @xmath194\in { { \mathcal{h}}}(r)$ ] by setting @xmath194_\infty = m(t)/t^d$ ] and @xmath195_m=\mbox{remainder of $ m(t)$ upon division by $ m(t)$}\ ] ] for each monic @xmath185 $ ] .
given for @xmath196 a polynomial @xmath197 $ ] for which @xmath198\in { { \mathcal{h}}}(r)$ ] is defined , note that @xmath199[m_2(t)]=[m_1(t)m_2(t)]$ ] .
[ [ subsubsection : brackets ] ] given @xmath200 and @xmath183 , put @xmath201 where @xmath185 $ ] is any monic polynomial such that @xmath202 $ ] .
it is easy to verify that the right side of ( [ equation : hthetadef ] ) is independent of @xmath203 and hence @xmath204 is well - defined .
clearly , @xmath204 depends @xmath5-linearly on @xmath205 .
now write @xmath206,\;\;\lfloor x\rfloor \in { { \mathbb{f}}}_q[t]).\ ] ] as the notation suggests , we think of @xmath207 as the `` fractional part '' and @xmath208 as the `` integer part '' of @xmath205 . with @xmath209 , @xmath205 and @xmath203 as above , we have @xmath210 after throwing away terms which do not contribute to the residue .
note that the right side of ( [ equation : hthetadefbis ] ) remains unchanged if we replaced @xmath211 by any member of its congruence class in @xmath175 $ ] modulo @xmath190 .
if moreover @xmath212 $ ] , where @xmath191 $ ] is a monic polynomial of degree @xmath192 whose resultant with each monic element of @xmath213 $ ] is a unit of @xmath169 , then we have @xmath214)= \operatorname{{\mathrm{res}}}_{t=\infty } ( \langle x\rangle(t)- \lfloor x\rfloor(t ) t^{-d-1})m(t)dt,\ ] ] after replacing @xmath215_m$ ] in ( [ equation : hthetadefbis ] ) by @xmath193 . in the next three propositions we study the values of the natural transformations @xmath216 at certain special points of @xmath167 .
[ proposition : jacobitrudiredux ] let @xmath12 be a variable independent of @xmath168 .
let @xmath97 be a nonnegative integer .
fix nonzero @xmath217 and put @xmath218 consider the point @xmath219^{-1}\in { { \mathcal{h}}}({{\mathbb{f}}}_q(t)).\ ] ] then we have @xmath220\\ \deg a < n \end{subarray } } \frac{x(t)+a}{t^n+a}\in { { \mathbb{f}}}_q(t)^\times.\ ] ] to simplify writing , put @xmath221 and @xmath222 .
let @xmath223 be independent variables , independent also of @xmath168 and @xmath12 , and put @xmath224.\ ] ] choose monic @xmath185 $ ] such that @xmath202 $ ] and write @xmath225).\ ] ] then we have , so we claim , @xmath226 this chain of equalities is justified as follows .
we get the first equality by plugging into version ( [ equation : hthetadefbis ] ) of the definition of @xmath227 .
we get the second equality after throwing away the term @xmath228 , which does not contribute to the residue , and then rearranging in evident fashion .
since we have @xmath229 we get the penultimate equality by `` sum of residues equals zero '' for meromorphic differentials on the @xmath168-line over @xmath115 .
the last equality follows directly from the definitions .
the claim is proved . via the moore determinant identity ( [ equation : mooreidentity ] ) , the result follows .
[ proposition : dualjacobitrudiredux ] let @xmath12 and @xmath205 be as in the preceding proposition . let @xmath230 be a nonnegative integer
. put @xmath231 consider the point @xmath232\in { { \mathcal{h}}}({{\mathbb{f}}}_q(t)).\ ] ] consider the @xmath5-linear functionals @xmath233 where @xmath234
then we have @xmath235 moreover , we have @xmath236+{{\mathbb{f}}}_q[t]\ ] ] and @xmath237 write @xmath238 then @xmath239 for @xmath240 is the @xmath241 elementary symmetric function of @xmath242 , and @xmath243 . by specializing the well - known representation of the @xmath241 elementary symmetric function as a quotient of determinants ( see ( * ? ? ? * i,3 ) ) to the present case we have @xmath244 for @xmath245 ; here the term bearing the `` hat '' is to be omitted , and if @xmath246 , the denominator ( an empty determinant ) is put to @xmath53 . by means of the ore identity ( [ equation : oreidentity ] ) we can express the right side of the claimed identity as the ratio of ore determinants @xmath247 and the latter ( note that @xmath248 for @xmath245 ) can be brought to the form @xmath249 by straightforward manipulation of determinants .
but this last equals @xmath204 by ( [ equation : simplesthhhexample ] ) .
thus ( [ equation : elementarysymmetric ] ) is proved .
it follows that @xmath250 .
moreover , @xmath251 if and only if the condition on the right side of ( [ equation : thetavanishing1 ] ) is fulfilled . therefore ( [ equation : thetavanishing1 ] ) holds .
we turn , finally , to the proof of ( [ equation : thetavanishing2 ] ) .
consider the laurent expansion @xmath252 of @xmath205 at @xmath253 , where @xmath254 and @xmath255 for @xmath173 .
since @xmath256 , not all the coefficients @xmath257 vanish .
moreover , by ( [ equation : thetavanishing1 ] ) , we have @xmath258 for @xmath245 .
finally , by definition of @xmath76 , the laurent expansion of @xmath76 at @xmath253 is @xmath259 .
the latter laurent expansion does not vanish identically since to form it we have merely suppressed some repetitions of digits in the laurent expansion @xmath252 . thus relation ( [ equation : thetavanishing2 ] ) holds .
[ proposition : peerintothehole ] let @xmath12 , @xmath205 , @xmath230 , @xmath76 and @xmath209 be as in the preceding proposition . assume now that @xmath260 ( and hence @xmath261 ) .
put @xmath262 ( note that the ring on the right is a principal ideal domain of which @xmath263 is a prime element . )
consider the point @xmath264^{-1 } \left[\prod_{i=0}^\nu(t-1\otimes t^{q^i})\right]\in { { \mathcal{h}}}({{\mathbb{f}}}_q(t)\otimes_{{{\mathbb{f}}}_q } { { \mathbb{f}}}_q(t))\ ] ] ( which reduces modulo @xmath263 to @xmath209 ) .
we have @xmath265 where @xmath266 put @xmath221 to simplify writing .
fix monic @xmath185 $ ] such that @xmath267 $ ] .
put @xmath268^{-1}\in { { \mathcal{h}}}(k),\ ] ] noting that @xmath269,\ ] ] and write @xmath270).\ ] ] let @xmath271,\;\;\;1\otimes q_\infty(t)\in 1+(1/t)(k\otimes_{{{\mathbb{f}}}_q}k)[[1/t]]\ ] ] be the results of applying the homomorphism @xmath272 coefficient by coefficient to @xmath273 and @xmath274 , respectively .
then , so we claim , we have congruences @xmath275 to verify the first congruence , multiply both sides by @xmath276 and see that both sides reduce to @xmath277 modulo @xmath278 . to verify the second congruence multiply both sides by @xmath279 and see that both sides reduce to @xmath280 modulo @xmath281 .
thus the claim is proved .
continuing our calculation , we now plug into version ( [ equation : hthetadefbis ] ) of the definition of @xmath227 , taking into account our hypothesis that @xmath260 .
we find that @xmath282 where @xmath283 is the unique coefficient for which ( [ equation : thetalead2 ] ) holds . by ( [ equation : thetavanishing1 ] ) and
another application of our hypothesis that @xmath260 , we have @xmath284 finally , we get the claimed value for @xmath9 by applying `` sum of residues equals zero '' for meromorphic differentials on the @xmath168-line over @xmath115 .
we prove a couple of technical results which in the sequel are used for no purpose other than as inputs to the proof of theorem [ theorem : recipe ] .
we `` quarantine '' the results here since they are general group - theoretical facts whose proofs would be a distraction from the main story .
[ lemma : idealtheoryingrouprings ] let @xmath285 be an integer .
let @xmath22 be an abelian group equipped with a nonconstant homomorphism @xmath286 .
let @xmath287 $ ] be the integral group ring of @xmath22 and let @xmath288 $ ] be the kernel of the ring homomorphism @xmath287\rightarrow { { \mathbb{r}}}$ ] induced by @xmath289 . for every subgroup @xmath290 , let @xmath291 $ ]
be the ideal generated by differences of elements of @xmath292 .
let @xmath293 be a subgroup of finite index and put @xmath294 .
then we have @xmath295 . since @xmath296
, we may pass to the quotient @xmath297={{\mathbb z}}[\gamma]/{{\mathcal{i}}}(\pi_1)$ ] in order to carry out our analysis of ideals .
after replacing @xmath22 by @xmath298 , we may simply assume that @xmath299 , in which case the function @xmath289 maps @xmath300 isomorphically to a subgroup of @xmath301 .
we can not have @xmath302 lest @xmath303=\infty$ ] , and hence @xmath300 is a free abelian group of rank @xmath53 . let @xmath304 be a generator .
then @xmath305 is the principal ideal of @xmath287 $ ] generated by @xmath306 .
suppose now that we are given @xmath307 .
then we can write @xmath308 for some @xmath309 $ ] and since @xmath310 , we must have @xmath311 .
therefore we have @xmath312 .
[ lemma : abelianrootredux ] let @xmath34 be a field . fix an algebraic closure @xmath313 .
let @xmath314 be the abelian closure of @xmath34 in @xmath315 .
let @xmath0 be a finite subextension of @xmath316 .
let @xmath317 be the group of roots of unity in @xmath115 and assume that @xmath318 .
put @xmath319 and let @xmath320 $ ] be the ideal of the integral group ring annihilating @xmath317 .
let @xmath321\rightarrow \bar{k}^\times$ ] be a homomorphism of abelian groups such that @xmath322 .
then @xmath323 takes all its values in @xmath324 .
the lemma is a variant of ( * ? ? ?
* lemma 6 ) and ( * ? ? ?
* , prop .
let @xmath325 be the separable algebraic closure of @xmath34 in @xmath315 .
put @xmath326:j]={{\#}}\mu(k)$ ] .
since @xmath327\subset j$ ] and @xmath328 is prime to the characteristic of @xmath315 , the homomorphism @xmath323 takes all its values in @xmath329 , and moreover the restriction map @xmath330,(k^{{\mathrm{sep}}})^\times)\rightarrow \operatorname{{\mathrm{hom}}}(j,(k^{{\mathrm{sep}}})^\times)\ ] ] is surjective .
put @xmath331 .
regard @xmath332 $ ] and @xmath333 as ( left ) @xmath334-modules by inflation . given two ( left ) @xmath334-modules @xmath78 and @xmath165 ,
let the group @xmath335 of homomorphisms of abelian groups be equipped with ( left ) @xmath334-module structure by the rule @xmath336 .
fix a generator @xmath337 . from the exact sequence of @xmath25-modules
@xmath338\xrightarrow{\zeta\mapsto \zeta^{{{\mathbf{a}}}}}\mu(k)\rightarrow 0\ ] ] we deduce an exact sequence @xmath339,(k^{{\mathrm{sep}}})^\times ) \rightarrow\operatorname{{\mathrm{hom}}}(j,(k^{{\mathrm{sep}}})^\times)\rightarrow 0\ ] ] of @xmath334-modules , where we make the evident identification @xmath340 since @xmath334 acts trivially on @xmath341 , we can extract an exact sequence @xmath342,(k^{{\mathrm{sep}}})^\times ) \rightarrow \operatorname{{\mathrm{hom}}}_{{{\mathfrak{g}}}}(j,(k^{{\mathrm{sep}}})^\times)\xrightarrow{\delta } \operatorname{{\mathrm{hom}}}_{\mbox{\tiny loc.\ const.}}({{\mathfrak{g}}},{{\mathbb z}}/e{{\mathbb z}})\ ] ] of abelian groups from the long exact sequence in galois cohomology .
the group on the right is the group of locally constant homomorphisms from @xmath334 to @xmath341 .
when we make the boundary map @xmath343 explicit by a diagram - chase , we find that @xmath344(\sigma)}\epsilon({{\mathbf{a}}})\ ] ] for all @xmath345 $ ] and @xmath346 .
( in this nonabelian setting , even though we employ exponential notation , we remain steadfastly leftist we follow the rule @xmath347 . )
it follows that @xmath323 commutes with every @xmath348 , and since every such @xmath349 acts trivially on @xmath332 $ ] , it follows finally that @xmath323 takes all its values in @xmath324 .
in this section we work in the setting of harmonic analysis on a nonarchimedean local field .
the main result here is the _ local stirling formula _ ( theorem [ theorem : localstirling ] ) , which is a technical result used in the sequel for no purpose other than as an input to the proof of the adelic stirling formula ( theorem [ theorem : adelicstirling ] ) . to motivate the local stirling formula ,
we prove a corollary ( corollary [ corollary : localstirling ] , which is not needed in the sequel ) which begins to explain the relationship with the classical stirling formula .
see @xcite for background on local harmonic analysis .
* let @xmath34 be a nonarchimedean local field .
( it is not necessary to assume that @xmath34 is of positive characteristic . ) * let @xmath350 be a nonconstant continuous homomorphism from the additive group of @xmath34 to the group of complex numbers of absolute value @xmath53 .
all the constructions in [ section : localstirling ] proceed naturally from these choices .
* let @xmath351 be the maximal compact subring of @xmath34 . * fix @xmath352 such that @xmath353 .
* let @xmath354 be haar measure on @xmath34 , normalized by @xmath355 .
* let @xmath356 be haar measure on @xmath357 , normalized by @xmath358 .
* let @xmath142 be the cardinality of the residue field of @xmath351 .
* for each @xmath359 , put @xmath360 and @xmath361 .
* let @xmath362 be the schwartz space of functions on @xmath34 . to help reconcile the present system of notation to whatever system
the reader might be familiar with , we recall a few facts routinely used below .
* @xmath363 for @xmath354-integrable @xmath364 and @xmath365 .
* @xmath366 and @xmath367 .
* @xmath289 is an absolute value of @xmath34 with respect to which @xmath34 is complete .
* @xmath368 is a normalized additive valuation of @xmath34 .
* @xmath369 for @xmath354-integrable @xmath364 .
[ [ section-9 ] ] given a complex - valued @xmath354-integrable function @xmath364 on @xmath34 , put @xmath370(\xi)=\hat{f}(\xi)=\int f(x){{\bf e}}(-x\xi)d\mu(x)\ ] ] for all @xmath371 , thus defining the _ fourier transform _ @xmath372 , denoted also by @xmath373 $ ] , which is again a complex - valued function on @xmath34 .
the fourier transform @xmath372 is continuous and tends to @xmath57 at infinity . if @xmath364 is compactly supported , then @xmath372 is locally constant .
if @xmath364 is locally constant , then @xmath372 is compactly supported .
for example , we have @xmath374(\xi)= { { \left\|\kappa\right\|}}^{1/2}{{\left\|b\right\|}}{{\mathbf{1}}}_{b^{-1}\kappa^{-1}{{\mathcal{o}}}}(\xi){{\bf e}}(-x\xi)\ ] ] for all @xmath375 and @xmath376 . [ [ section-10 ] ] the schwartz class @xmath362 is stable under fourier transform and the _ fourier inversion formula _ states that @xmath377 for all @xmath378 . our normalization of @xmath354 is chosen to make the inversion formula hold in the latter particularly simple form ; in general there would be a positive constant depending on @xmath354 but independent of @xmath68 multiplying the right side .
the fourier inversion formula implies the _ squaring rule _
@xmath379(x)=\phi(-x)\ ] ] for all @xmath378 .
we mention also the very frequently used _ scaling rule _
@xmath380={{\left\|a\right\|}}\hat{f}^{(a^{-1})}\;\;\;(f^{(a)}(x)= f(a^{-1}x))\ ] ] which holds for all @xmath365 and @xmath354-integrable @xmath364 .
[ [ section-11 ] ] we define a _ lattice _
@xmath381 to be a cocompact discrete subgroup .
we remark that lattices exist in @xmath34 only if @xmath34 is of positive characteristic .
we remark also that the notion of lattice figures only in corollary [ corollary : localstirling ] below , and otherwise is unused in the sequel . for each lattice
@xmath382 put @xmath383 .
for all lattices @xmath382 again @xmath384 is a lattice and @xmath385 .
the _ poisson summation formula _ states that @xmath386 for all lattices @xmath381 and schwartz functions @xmath378 , where @xmath387 denotes the _ covolume _ of @xmath382 in @xmath34 with respect to @xmath354 .
[ [ subsubsection : mdef ] ] for each @xmath378 and @xmath365 put @xmath390=\int ( \phi(t)-{{\mathbf{1}}}_{a{{\mathcal{o}}}}(t)\phi(0))d\mu^\times(t).\ ] ] to see that @xmath391 $ ] is a well - defined complex number , let @xmath392 temporarily denote the integrand on the right side and note the following : * @xmath392 is defined and locally constant on @xmath357 . *
@xmath392 vanishes for @xmath393 sufficiently large .
therefore the integral on the right side of ( [ equation : mdef ] ) converges .
we have a scaling rule @xmath394={{\mathcal{m}}}^{(ab)}[\phi]={{\mathcal{m}}}^{(a)}[\phi]+\phi(0)\operatorname{{\mathrm{ord}}}b\ ] ] for all @xmath375 , cf . scaling rule ( [ equation : localscaling ] ) for the fourier transform . to motivate the definition of @xmath388 we remark that lemma [ lemma : lfeconsequence ] below can be reinterpreted as the assertion that the linear functional @xmath395\ ] ] on @xmath362 is the fourier transform of @xmath396 in the sense of the theory of distributions .
[ [ subsubsection : ldef ] ] for each @xmath378 and @xmath397 , put @xmath398(x)=\int \frac{h(t^{\mp 1 } ) \phi(xt)-\frac{1}{2 } { { \mathbf{1}}}_{{{\mathcal{o}}}^\times}(t)\phi(x)}{{{\left\|t-1\right\|}}}{{\left\|t\right\|}}d\mu^\times(t),\ ] ] where @xmath399 to see that @xmath400(x)$ ] is a well - defined complex number , let @xmath401 temporarily denote the integrand on the right side of ( [ equation : ldef ] ) , and note the following : * @xmath401 is defined and locally constant on @xmath402 . *
@xmath403 for @xmath404 sufficiently small . * @xmath403 for @xmath405 sufficiently large .
* @xmath406 for @xmath405 sufficiently small .
* @xmath407 for @xmath405 sufficiently small .
therefore the integral on the right side of ( [ equation : ldef ] ) converges .
it is not difficult to verify that @xmath408(x)= \phi(0)\int_{{{\left\|t\right\|}}<1}{{\left\|t\right\|}}d\mu^\times(t)=\frac{\phi(0)}{q-1}.\ ] ] we obtain another integral representation @xmath409(x)=\int \frac{h(t ) \phi(xt^{-1})-\frac{1}{2 } { { \mathbf{1}}}_{{{\mathcal{o}}}^\times}(t)\phi(x)}{{{\left\|1-t\right\|}}}d\mu^\times(t)\ ] ] for @xmath410 by substituting @xmath411 for @xmath168 in the integral on the right side of ( [ equation : ldef ] ) .
in contrast to the case of the linear functional @xmath388 we can not easily give the motivation for the definition of the operators @xmath389 .
the best we can say at present is that @xmath410 makes the following theorem hold , and that @xmath412 is indispensable to the proof of the theorem . a more conceptual characterization of the operators @xmath389 would be nice to have . [
theorem : localstirling ] there exists a unique linear operator @xmath413 such that @xmath414(x)= \left\{\begin{array}{cl } { -\phi(x)\left(\frac{1}{2}\operatorname{{\mathrm{ord}}}\kappa+\operatorname{{\mathrm{ord}}}x\right)+\operatorname{{\mathcal{l}}}^+[\phi](x ) } & \mbox{if $ x\neq 0$,}\\\\ -\frac{1}{2}\phi(0)\operatorname{{\mathrm{ord}}}\kappa+{{\mathcal{m}}}^{(1)}[\phi]&\mbox{if $ x=0$,}\end{array}\right.\ ] ] for all @xmath378 and @xmath415 .
moreover , we have @xmath416=-\operatorname{{\mathcal{f}}}[\operatorname{{\mathcal{k}}}[\phi]]\ ] ] for all @xmath378 . for the proof
we need three lemmas .
[ lemma : lfeconsequence ] we have @xmath417 } \end{array}\ ] ] for all @xmath418 and @xmath365 . by scaling rule ( [ equation : mscale1 ] ) the right side of ( [ equation : lfeconsequence ] ) is independent of @xmath99 . for every @xmath375 we have @xmath419 and hence by the scaling rules ( [ equation : localscaling ] ) and ( [ equation : mscale1 ] ) ,
the difference of left and right sides of ( [ equation : lfeconsequence ] ) remains unchanged if we replace @xmath68 by @xmath420 ( and hence @xmath421 by @xmath422 ) .
clearly , the left side of ( [ equation : lfeconsequence ] ) remains unchanged if we replace @xmath68 by @xmath423 for any @xmath424 .
the same holds for the right side by the scaling rules ( [ equation : localscaling ] ) and ( [ equation : mscale1 ] ) .
consequently both sides of ( [ equation : lfeconsequence ] ) remain unchanged if we replace @xmath68 by the averaged function @xmath425 , which is a finite linear combination of functions of the form @xmath426 with @xmath375 . taking into account all the preceding reductions
, we may now assume without loss of generality that @xmath427 and @xmath428
. then equation ( [ equation : lfeconsequence ] ) is easy to check by direct calculation .
we omit further details .
[ lemma : lfeconsequence2 ] we have @xmath429(\xi ) } \end{array}\ ] ] for all @xmath418 and @xmath430 .
we calculate as follows : + @xmath431(\xi)}\end{array}\ ] ] + we get the first equality by applying ( [ equation : lfeconsequence ] ) with @xmath99 replaced by @xmath100 and @xmath432 replaced by @xmath433 .
the rest of the calculation is routine .
[ lemma : mollificationestimate ] fix @xmath378 .
fix @xmath365 such that @xmath68 is supported in @xmath434 . fix @xmath435 such that @xmath68 is constant on cosets of @xmath436 . fix @xmath437
such that @xmath438 for all @xmath415 .
then we have @xmath439 for all @xmath397 and @xmath440 . for convenience in applying the lemma , note that @xmath441 for all @xmath442 .
by hypothesis we have @xmath443 and hence @xmath444 whence the result .
the right side of ( [ equation : kdef ] ) defines on @xmath34 a complex - valued function @xmath445 $ ] depending linearly on @xmath68 . by lemma [ lemma : mollificationestimate ]
the function @xmath446 $ ] is @xmath354-integrable and compactly supported , and hence so is the function @xmath445 $ ] .
similarly , @xmath447 $ ] is @xmath354-integrable and compactly supported . since @xmath445 $ ] is @xmath354-integrable and compactly supported , it follows that the fourier transform @xmath448 $ ] is well - defined and locally constant .
we have @xmath449(\xi)=\operatorname{{\mathcal{l}}}^-[\operatorname{{\mathcal{f}}}[\phi]](\xi)\;\;\;(\xi\in k^\times),\ ] ] since by another application of lemma [ lemma : mollificationestimate ] it is justified to reverse the order of integration , and by limit formula ( [ equation : eulermascheroni ] ) it follows that @xmath450(0)=\frac{\hat{\phi}(0)}{q-1}.\ ] ] by ( [ equation : mollificationestimateter ] ) above , lemma [ lemma : lfeconsequence ] , and the definitions we have @xmath451(0)=-\operatorname{{\mathcal{k}}}[\operatorname{{\mathcal{f}}}[\phi]](0).\ ] ] by ( [ equation : mollificationestimatebis ] ) above , lemma [ lemma : lfeconsequence2 ] , and the definitions we have @xmath451(\xi)=-\operatorname{{\mathcal{k}}}[\operatorname{{\mathcal{f}}}[\phi]](\xi)\;\;(\xi\in k^\times).\ ] ] therefore we have @xmath448=-\operatorname{{\mathcal{k}}}[\operatorname{{\mathcal{f}}}[\phi]]$ ] , hence @xmath447 $ ] is both compactly supported and locally constant , and hence @xmath447\in { { \mathrm{sch}}}(k)$ ] . since @xmath378 was arbitrarily specified and the operator @xmath452 is invertible , it follows that the operator @xmath453 stabilizes the schwartz space @xmath362 and anticommutes with the fourier transform @xmath452 , as claimed .
the following corollary will not be needed in the sequel we include it just to motivate the theorem .
[ corollary : localstirling ] fix a schwartz function @xmath418 and a lattice @xmath381 .
put @xmath454 then the following hold : @xmath455 @xmath456 @xmath457 @xmath458 we have @xmath459 for all @xmath365 . functional equation ( [ equation :
localstirling1 ] ) follows from scaling formula ( [ equation : localscaling ] ) and the poisson summation formula ( [ equation : poissonsummation ] ) .
statements ( [ equation : localstirling2],[equation : localstirling3],[equation : localstirling4 ] ) are easy to verify .
we omit the details .
statements ( [ equation : localstirling2],[equation : localstirling3],[equation : localstirling4 ] ) granted , it is clear that the integral on the right side of ( [ equation : localstirling ] ) converges .
we turn to the proof of equation ( [ equation : localstirling ] ) .
first , we reduce to the case @xmath427 by observing that under replacement of the pair @xmath460 by the pair @xmath461 , neither the left side nor the right side of ( [ equation : localstirling ] ) change .
next , we note that @xmath462(x)+\mu(k / l)^{-1}\sum_{\xi\in l^\perp}\operatorname{{\mathcal{k}}}[\hat{\phi}](\xi)=0\ ] ] by the poisson summation formula ( [ equation : poissonsummation ] ) and the anticommutation relation ( [ equation : anticommutation ] ) . clearly ,
it is possible to rearrange the terms above to put the left side in coincidence with the left side of ( [ equation : localstirling ] ) in the case @xmath427 , leaving a sum @xmath463 on the right side , where @xmath464(x),\;\;\ ; a^*=\sum_{0\neq \xi\in l^\perp } \operatorname{{\mathcal{l}}}^+[\hat{\phi}](\xi).\ ] ] using the presentation ( [ equation : ldefbis ] ) of the operator @xmath410 and carrying the sum under the integral , we have @xmath465 exchange of sum and integral is justified by lemma [ lemma : mollificationestimate ] and the remark ( [ equation : integrationofestimate ] ) immediately following .
we have a similar representation for @xmath466 , which after the substitution of @xmath411 for @xmath168 takes the form @xmath467 finally , by exploiting functional equation ( [ equation : localstirling1 ] ) we can bring @xmath463 to the form of the right side of ( [ equation : localstirling ] ) in the case @xmath427 .
we omit the remaining details of bookkeeping . because the sum @xmath468 vanishes for @xmath469 , equation ( [ equation : localstirling ] )
provides a precise asymptotic description of the sum @xmath470 as @xmath471 . in the simple special case @xmath472 ,
the latter sum is analogous to the sum @xmath473}(n^{-1}x)\log |x|\\\\=&\log n!= \displaystyle n\log n -n+\frac{1}{2}\log n+\frac{1}{2}\log(2\pi)+o_{n\rightarrow\infty}(1 ) \end{array}\ ] ] on the real line .
so it is reasonable to view equation ( [ equation : localstirling ] ) as an analogue and generalization of stirling s formula in the nonarchimedean setting .
given the key role played by theorem [ theorem : localstirling ] in deriving ( [ equation : localstirling ] ) , we choose to regard theorem [ theorem : localstirling ] itself as an analogue and generalization of the classical stirling formula .
this explains our terminology .
but the analogy of theorem [ theorem : localstirling ] with stirling s formula is rather imperfect .
a closer look reveals some complications .
as it happens , for theorem [ theorem : localstirling ] and corollary [ corollary : localstirling ] we can derive direct analogues on the real line , using standard methods of the theory of tempered distributions , and these results in turn lead to a precise description of the asymptotic behavior as @xmath474 of sums of the form @xmath475 where @xmath476 is a schwartz function on the real line . for example , we can derive in this way the asymptotic formula @xmath477 where @xmath478 is the euler - mascheroni constant .
from the point of view of harmonic analysis , the sum ( [ equation : stirlingdiscussiontwo ] ) in the simple special case @xmath472 is more closely analogous to the `` soft - edged '' sum ( [ equation : stirlingdiscussionfive ] ) than to the `` hard - edged '' sum ( [ equation : stirlingdiscussionthree ] ) .
these complications taken into account , we may only say that a `` soft '' analogy exists between stirling s formula and theorem [ theorem : localstirling ] .
we shall discuss analogues of theorem [ theorem : localstirling ] for @xmath479 and @xmath117 and their applications in detail on another occasion .
from now on in this paper we work in the setting of harmonic analysis on the adele ring of a global field of positive characteristic , a setting for which the ( unfortunately ) rather elaborate notation is set out in detail in tables immediately below . the main result of this section is the _ adelic stirling formula _ ( theorem [ theorem : adelicstirling ] ) , which in form resembles corollary [ corollary : localstirling ] .
see @xcite for background on global harmonic analysis .
* let @xmath34 be a global field of positive characteristic , of genus @xmath480 , and with constant field @xmath5 .
* fix a nonzero khler differential @xmath481 .
* fix a nonconstant character @xmath482 , as in [ section : duality ] .
all constructions below proceed naturally from these choices .
let @xmath80 be any place of @xmath34 .
* let @xmath483 be the completion of @xmath34 at @xmath80 .
* let @xmath484 be the maximal compact subring of @xmath483 .
* let @xmath485 be the residue field of @xmath484 .
* let @xmath486 be the residue map at @xmath80 .
* put @xmath487 .
* let @xmath488 be haar measure on @xmath483 .
* normalize @xmath488 by the rule @xmath489 .
* let @xmath490 be the cardinality of @xmath485 .
* for each @xmath491 , put @xmath492 and @xmath493 .
* let @xmath494 be haar measure on @xmath495 .
* normalize @xmath494 by the rule @xmath496 .
* let @xmath497 be the order of vanishing of @xmath498 at @xmath80 .
* choose @xmath499 such that @xmath500 , and hence @xmath501 .
* let @xmath502 be the schwartz class of functions on @xmath483 .
* let @xmath37 be the adele ring of @xmath34 .
* let @xmath503 be the maximal compact subring .
* recall that each adele @xmath504 is a family @xmath505 $ ] indexed by places @xmath80 such that @xmath506 for all @xmath80 and @xmath507 for all but finitely many @xmath80 .
moreover , we have @xmath508 . * as usual
, we regard @xmath34 as diagonally embedded in @xmath37 ; thus @xmath34 becomes a cocompact discrete subgroup of @xmath37 .
* let @xmath354 be a haar measure on @xmath37 .
* normalize @xmath354 by the rule @xmath509 .
equivalently : @xmath510 . *
recall that each idele @xmath511 is a family @xmath512 $ ] indexed by places @xmath80 such that @xmath513 for all @xmath80 and @xmath514 for all but finitely many @xmath80 .
moreover , we have @xmath515 . * for each @xmath511 , put @xmath516 .
* let @xmath356 be a haar measure on @xmath43 .
* normalize @xmath356 by the rule @xmath517 .
* put @xmath518\mapsto
\sum_v \operatorname{{\mathrm{tr}}}_{{{\mathbb{f}}}_v/{{\mathbb{f}}}_q}\operatorname{{\mathrm{res}}}_v x_v\omega\right):{{\mathbb{a}}}\rightarrow { { \mathbb{f}}}_q$ ] . * for all @xmath519 , put @xmath520 .
* put @xmath521\mapsto \prod_v { { \bf e}}_v(x_v ) ) : { { \mathbb{a}}}\rightarrow u(1)$ ] . * put @xmath522\in { { \mathbb{a}}}^\times$ ] .
* we write @xmath523 for all places @xmath80 .
* for each place @xmath80 let @xmath524 be the unique maps such that @xmath525 for all @xmath526 , @xmath527 and places @xmath528 .
* let @xmath529 be the schwartz class of functions on @xmath37 . to help reconcile the present system of notation to whatever system
the reader might be familiar with , we recall a few facts routinely used below .
* @xmath363 all @xmath354-integrable @xmath364 and @xmath38 .
* @xmath530 and @xmath531 . * @xmath532 for all @xmath512\in { { \mathbb{a}}}^\times$ ] .
* @xmath533 for all @xmath397 .
( artin s product formula . )
* @xmath534 for all @xmath415 .
( sum of residues equals zero . ) * @xmath535 . *
* for all @xmath537 such that @xmath538 , the spaces @xmath539 and @xmath540 are paired perfectly by @xmath541 .
moreover we have @xmath542 with respect to @xmath543 .
( serre duality . )
[ [ section-12 ] ] given a @xmath354-integrable complex - valued function @xmath364 on @xmath37 , put @xmath370(\xi)=\hat{f}(\xi)= \int f(x)\lambda(-\langle x,\xi\rangle)d\mu(x)=\int f(x){{\bf e}}(-x\xi)d\mu(x)\ ] ] for all @xmath544 , thereby defining the _ fourier transform _ @xmath372 , also denoted by @xmath373 $ ] , which is a complex - valued continuous function on @xmath37 tending to @xmath57 at infinity .
if @xmath364 is compactly supported , then @xmath372 is locally constant .
if @xmath364 is locally constant , then @xmath372 is compactly supported .
for example , we have @xmath545(\xi)= q^{1-g}{{\left\|b\right\|}}{{\mathbf{1}}}_{b^{-1}\kappa^{-1}{{\mathcal{o}}}}(\xi ) \lambda(-\langle x,\xi\rangle)\ ] ] for all @xmath546 and @xmath547 . [ [ section-13 ] ]
the schwartz class @xmath529 is stable under the fourier transform and the _ fourier inversion formula _ states that @xmath548 for all @xmath549 .
our normalization of @xmath354 is chosen to make the fourier inversion formula hold in the simple form stated above ; in general there would be a positive constant depending on @xmath354 but independent of @xmath68 multiplying the right side .
the fourier inversion formula implies the _ squaring rule _
@xmath550(x)=\phi(-x)\ ] ] for all @xmath549 .
we mention also the _ scaling rule _ @xmath551={{\left\|a\right\|}}\hat{f}^{(a^{-1})}\;\;\ ; ( f^{(a)}(x)=f(a^{-1}x))\ ] ] holding for all @xmath38 and @xmath354-integrable @xmath364 . [ [ section-14 ] ] the _ poisson summation formula _ states that @xmath552 for all @xmath38 and @xmath549 .
note that since both @xmath68 and @xmath421 are compactly supported , only finitely many nonzero terms occur in the sums on either side of the formula .
the normalization @xmath553 of haar measure @xmath354 on @xmath37 was chosen to make the poisson summation formula hold in the particularly simple form above ; in general the right side would be multiplied by a positive constant depending on @xmath354 but independent of @xmath68 . [ [ subsubsection : thetasymboldef ] ] for all @xmath38 and @xmath549 put @xmath554 thereby defining the _ theta symbol _
we have @xmath556 for all @xmath397 .
in other words , @xmath557 depends only on the image of @xmath99 in the _ idele class group _ @xmath558 .
clearly , we have a _ scaling rule _
@xmath559 for all @xmath546 .
we have a _ functional equation _
@xmath560 which is just a rewrite of the poisson summation formula . [ proposition : thetaasymptotics ] fix @xmath549 .
fix @xmath546 and @xmath561 such that @xmath68 is constant on cosets of @xmath436 and supported in @xmath562 . then : ( i ) we have @xmath563 for all @xmath38 .
( ii ) moreover , @xmath557 depends in locally constant fashion on @xmath99 . from example ( [ equation : globalfourierexample ] ) and serre duality it follows that the fourier transform @xmath421 is constant on cosets of @xmath564 and supported in @xmath565 .
we therefore have @xmath566 via ( [ equation : thetapoisson ] ) . statement ( i ) follows via the artin product formula .
statement ( ii ) follows from the observation that the map @xmath567 from @xmath43 to finite subsets of @xmath34 is locally constant . [
theorem : adelicstirling ] for all ideles @xmath38 , schwartz functions @xmath568 and places @xmath80 of @xmath34 we have @xmath569 note that proposition [ proposition : thetaasymptotics ] guarantees convergence of the integral involving the theta symbol on the right side of ( [ equation : adelicstirling ] ) .
the pattern set by the proof of corollary [ corollary : localstirling ] will guide us in proving ( [ equation : adelicstirling ] ) .
the proof will be completed in [ equation : endofadelicstirlingproof ] after some preparation .
the rationale for the stirling formula terminology here is essentially the same as that offered in connection with theorem [ theorem : localstirling ] and corollary [ corollary : localstirling ] .
namely , to the extent that sums of the form @xmath570 may be regarded as analogues of @xmath571 , formula ( [ equation : adelicstirling ] ) may be regarded as an analogue of the classical stirling formula .
it is an interesting problem to devise and interpret a version of the adelic stirling formula for number fields .
we hope to have progress to report on this problem in the near future .
[ [ section-15 ] ] given any family of vector spaces @xmath572 , where each @xmath573 is equipped with a _ neutral element _ @xmath574 , there is a natural way to form the ( possibly infinite ) tensor product @xmath575 .
namely , the latter is by definition spanned by symbols of the form @xmath576 subject to the obvious relations .
moreover , given for each index @xmath577 a linear endomorphism @xmath578 of @xmath573 such that @xmath579 for all but finitely many @xmath577 , there is a natural tensor product @xmath580 of operators , namely , that sending each symbol @xmath581 to @xmath582 .
[ [ section-16 ] ] in the general sense above , as is well known , @xmath529 can naturally be identified with @xmath583 , where for each place @xmath80 the space @xmath502 is equipped with the neutral element @xmath584 .
( see , for example , @xcite .
but note that in the definition presented there the condition `` @xmath585 '' should be strengthened to `` @xmath586 '' .
otherwise the function @xmath364 thus defined might fail to be compactly supported . )
[ [ section-17 ] ] we have at least one natural example of a linear operator on @xmath529 that factors as an infinite tensor product , namely the fourier transform @xmath452 . more precisely ,
for each place @xmath80 , let @xmath587 be the local fourier transform defined by the rule @xmath588(\xi)=\int \phi(x){{\bf e}}_v(-x\xi)d\mu_v(x).\ ] ] since @xmath589={{\mathbf{1}}}_{{{\mathcal{o}}}_v}$ ] for all but finitely many @xmath80 , it makes sense to form the tensor product of the operators @xmath590 over all places @xmath80 .
one can easily verify that this tensor product does indeed coincide with @xmath452 .
[ [ section-18 ] ] let a place @xmath80 and a linear endomorphism of @xmath382 of @xmath502 be given .
let @xmath591 be the `` coordinate hyperplane '' of @xmath37 consisting of families @xmath592 $ ] such that @xmath593 .
given @xmath594 and @xmath549 , let @xmath595 be defined by the rule @xmath596 for all @xmath526 .
then , so we claim , there exists a unique linear endomorphism @xmath597 of @xmath529 such that @xmath598(i_v(x)+y)=l[\phi_{y , v}](x)\ ] ] for all @xmath549 , @xmath526 and @xmath594 .
to prove the claim one just has to check that @xmath597 is the tensor product of the family @xmath599 extended over all places @xmath528 .
we omit these details .
we call @xmath597 the _ canonical prolongation _ of @xmath382 .
fix a place @xmath80 of @xmath34 .
we keep the notation introduced in the preceding discussion of the tensor decomposition of @xmath529 .
[ [ section-19 ] ] given @xmath549 , we define a function @xmath603 $ ] on @xmath591 by the rule @xmath604(y)= \int(\phi_{y , v}(t)-{{\mathbf{1}}}_{{{\mathcal{o}}}_v}(t)\phi_{y , v}(0))d\mu_v^\times(t)\ ] ] for all @xmath594 . the function
@xmath603 $ ] is well - defined by a repetition of the argument given in
[ subsubsection : mdef ] . note in particular that @xmath605(0)=\int ( \phi(i_v(t))-{{\mathbf{1}}}_{{{\mathcal{o}}}_v}(t)\phi(0))d\mu_v^\times(t)\ ] ] for all @xmath549 .
[ [ section-20 ] ] given @xmath549 , we define a function @xmath606 $ ] on @xmath607 by the rule @xmath608(i_v(x)+y)=\displaystyle\int\frac{h_v(t)\phi_{y , v}(t^{-1}x)-\frac{1}{2 } { { \mathbf{1}}}_{{{\mathcal{o}}}_v^\times}(t)\phi_{y , v}(x ) } { { { \left\|1-t\right\|}}_v}d\mu_v^\times(t)\ ] ] for all @xmath609 and @xmath610 , where @xmath611
the function @xmath606 $ ] is well - defined by a repetition of the argument given in
[ subsubsection : ldef ] .
equivalently , we have @xmath612(z)=\int\frac{h_v(t)\phi(i_v^\times(t^{-1})z)-\frac{1}{2}{{\mathbf{1}}}_{{{\mathcal{o}}}_v^\times}(t)\phi(z ) } { { { \left\|1-t\right\|}}_v}d\mu_v^\times(t)\ ] ] for all @xmath613 .
note that the operator @xmath601 preserves supports in the sense that for all @xmath614 and @xmath549 , if @xmath68 vanishes outside @xmath434 , then @xmath606 $ ] vanishes outside @xmath615 .
[ [ section-21 ] ] theorem [ theorem : localstirling ] and the canonical prolongation process now yield a unique linear operator @xmath616 such that @xmath617(i_v(x)+y)\\\\ = & \displaystyle\left\{\begin{array}{ll } -\phi(i_v(x)+y)(\frac{1}{2}\operatorname{{\mathrm{ord}}}_v\omega+\operatorname{{\mathrm{ord}}}_vx)+\operatorname{{\mathcal{l}}}_v[\phi](i_v(x)+y ) & \mbox{if $ x\neq 0$,}\\ -\frac{1}{2}\phi(y)\operatorname{{\mathrm{ord}}}_v\omega+ { { \mathcal{m}}}_v[\phi](y)&\mbox{if $ x=0 $ , } \end{array}\right . \end{array}\ ] ] for all @xmath549 , @xmath526 and @xmath594 . moreover , since @xmath452 factors as the tensor product of local fourier transforms , and the operator on @xmath502 defined by theorem [ theorem : localstirling ] anticommutes with @xmath590 , we necessarily have @xmath618=-\operatorname{{\mathcal{f}}}[\operatorname{{\mathcal{k}}}_v[\phi]]\ ] ] for all @xmath549 , i. e. , @xmath602 anticommutes with the adelic fourier transform @xmath452 .
[ lemma : integralandsums ] fix @xmath549 .
fix a place @xmath80 of @xmath34 . put @xmath619 for all @xmath527 .
then we have @xmath620(x)+ \sum_{\xi\in k^\times}\operatorname{{\mathcal{l}}}_v[\hat{\phi}](\xi)\\\\ = & \displaystyle \int\frac{\vartheta(t)-\left\{\begin{array}{cl } \phi(0)&\mbox{if $ { { \left\|t\right\|}}_v<1$}\\ \vartheta(1)&\mbox{if $ { { \left\|t\right\|}}_v=1$}\\ { { \left\|t\right\|}}_v\hat{\phi}(0)&\mbox{if $ { { \left\|t\right\|}}_v>1 $ } \end{array}\right.}{{{\left\|1-t\right\|}}_v}d\mu_v^\times(t ) . \end{array}\ ] ] convergence of the sums on the left and the integral on the right will be established in the course of the proof .
the integrand on the right side of ( [ equation : keyidentity ] ) is locally constant and is supported on a compact subset of @xmath621 by proposition [ proposition : thetaasymptotics ] .
so the integral on the right side of ( [ equation : keyidentity ] ) converges .
now fix @xmath614 such that both @xmath68 and @xmath421 are supported in @xmath434 , hence both @xmath606 $ ] and @xmath622 $ ] are supported in @xmath615 , and hence the quantities @xmath623(x),\;\;\;\operatorname{{\mathcal{l}}}_v[\hat{\phi}](\xi)\ ] ] vanish for all @xmath624 and @xmath625
. convergence of the sums on the left side of ( [ equation : keyidentity ] ) follows , and moreover it is justified to carry the summations under the integral ( [ equation : lsubvalternativedef ] ) representing the operator @xmath601 .
we find that the left side of ( [ equation : keyidentity ] ) equals @xmath626 @xmath627 whence the result after making a substitution of @xmath411 for @xmath168 in the second integral and then collecting like terms .
neither side of ( [ equation : adelicstirling ] ) changes under the replacement of @xmath628 by @xmath629 .
we may therefore assume without loss of generality that @xmath427 . since @xmath602 anti - commutes with @xmath452 , the poisson summation formula gives the identity @xmath630(x)+\sum_{\xi\in k}\operatorname{{\mathcal{k}}}_v[\hat{\phi}](\xi)=0.\ ] ] after rearranging the terms in this last identity and applying ( [ equation : thetapoisson ] ) , ( [ equation : msubvalternativedef ] ) and ( [ equation : keyidentity ] ) ,
we obtain ( [ equation : adelicstirling ] ) in the case @xmath427 .
the proof of the theorem is complete .
we pick up again the ideas introduced in [ section : duality ] .
[ [ section-22 ] ] with @xmath631 as in [ subsection : trivialidentities ] , put @xmath632 and then given any @xmath549 , put @xmath633(\xi)=\tilde{\phi}(\xi)=\int \phi(x){{\bf e}}_0(x\xi)d\mu(x)= \int \phi(x)\lambda_0(\langle x,\xi\rangle)d\mu(x)\ ] ] for all @xmath544 , thereby defining the _ rational fourier transform _ @xmath634 of @xmath68 , also denoted by @xmath635 $ ] . via the trivial identities ( [ equation : lambdaidentities ] ) we have @xmath636 for all @xmath544 .
thus @xmath421 and @xmath634 are expressible each in terms of the other .
it follows that @xmath145 preserves the schwartz space @xmath529 , just as does the standard fourier transform @xmath452 .
it follows from the trivial identities ( [ equation : lambdaidentities ] ) and the example ( [ equation : globalfourierexample ] ) that @xmath637= q^{1-g}{{\left\|b\right\|}}\left({{\mathbf{1}}}_{b^{-1}\kappa^{-1}{{\mathcal{o}}}\cap[\langle x,\cdot\rangle=0]}-{{\mathbf{1}}}_{b^{-1}\kappa^{-1}{{\mathcal{o}}}\cap[\langle x,\cdot\rangle=1]}\right)\ ] ] for all @xmath546 and @xmath547 .
it follows in turn via serre duality that for all @xmath638 , @xmath639 and @xmath568 such that @xmath68 is constant on cosets of @xmath436 and supported in @xmath562 , the rational fourier transform @xmath634 is constant on cosets of @xmath564 and supported in @xmath565 .
[ [ section-23 ] ] fix @xmath549 . by equations ( [ equation : relationoffts ] ) relating standard and rational fourier transforms , the inversion formula ( [ equation : adelicfourierinversion ] ) for the standard fourier transform , and the trivial identities ( [ equation : lambdaidentities ] )
, we have @xmath640 for all @xmath641 ; thus we can invert the rational fourier transform . [
[ section-24 ] ] given @xmath549 and @xmath641 we put @xmath642(x)=\sum_{c\in { { \mathbb{f}}}_q^\times}\phi(cx),\ ] ] thereby defining @xmath642\in { { \mathrm{sch}}}({{\mathbb{a}}}).\ ] ] then the inversion formula above for the rational fourier transform can be rewritten more compactly in the form @xmath643 the related operator identities @xmath644 and the _ squaring rule _
@xmath645 are easy to verify .
we omit the proofs .
the last formula should be compared with the squaring rule ( [ equation : adelicfouriersquared ] ) for the adelic fourier transform .
( if @xmath133 , then the two formulas are the same . )
[ [ section-25 ] ] we have for the rational fourier transform a _ scaling rule _ of exactly the same form as that obeyed by the usual fourier transform , namely @xmath646={{\left\|a\right\|}}\tilde{\phi}^{(a^{-1})}\ ] ] for all @xmath549 and @xmath38 .
moreover , since @xmath647 we have @xmath648(\xi ) = { { \left\|a\right\|}}\sum_{\xi\in k}\tilde{\phi}(a\xi)\ ] ] for all @xmath38 and @xmath549 , i. e. , the poisson summation formula continues to hold with the rational fourier transform replacing the standard one .
consequently we have a _ functional equation _
@xmath649 for all @xmath38 and @xmath549 .
[ [ subsubsection : catalandef ] ] recall that @xmath67 $ ] ( as in the discussion of toy rational fourier transforms and toy catalan symbols ) is the ring consisting of those rational numbers @xmath205 such that @xmath650 for @xmath651 .
let @xmath652 denote the space of @xmath67$]-valued schwartz functions on @xmath37 , and let @xmath653 denote the subgroup consisting of functions @xmath68 such that @xmath654 .
clearly , the groups @xmath655 are stable under the action of the operators @xmath145 and @xmath656 .
note that @xmath657 $ ] for all @xmath38 and @xmath658 .
note that @xmath659 makes sense for all @xmath660 and @xmath661 $ ] .
note that @xmath39 for all @xmath614 and hence @xmath662 makes sense for all @xmath663 .
for all places @xmath80 of @xmath34 and @xmath664 we define @xmath665 $ ] for all sufficiently large @xmath666 . [
[ subsubsection : catalanformalproperties ] ] for all @xmath667 and @xmath38 , put @xmath668 in the infinite products only finitely many terms differing from @xmath53 occur , so these objects are well - defined .
we call @xmath150 the _ catalan symbol _ , and @xmath669 the _ partial catalan symbol_. the rationale for the terminology is the same as the one offered in the case of the toy catalan symbol , namely the values of the catalan and partial catalan symbol bear a certain structural resemblance to the catalan numbers @xmath71 .
[ proposition : genericcatalanbehavior ] fix @xmath670 .
fix @xmath546 and @xmath671 such that @xmath68 is constant on cosets of @xmath436 and supported in @xmath562 .
then @xmath672 for all @xmath38 . in parallel to formula ( [ equation : thetaasymptotics ] )
, we have @xmath673 whence the result by artin s product formula and the definitions .
[ [ section-26 ] ] given @xmath537 , @xmath397 , and @xmath658 , we claim that @xmath674 @xmath675 and @xmath676 note in particular that if @xmath677 , then @xmath678 depends only on the image of @xmath99 in the idele class group @xmath558 .
the last two relations we call the _ scaling rule _ and _ functional equation _ , respectively , obeyed by the catalan symbol . [ [ subsubsection : routinecatalancalculations ] ] here are the proofs of the claims .
the calculation @xmath679 proves ( [ equation : ideleclassinvariancebis ] ) . at the last equality we applied artin s product formula and functional equation ( [ equation : thetarationalfe ] ) .
the calculation @xmath680(a\xi){{\left\|a\right\|}}}=\left(\begin{array}{c } a\\ \phi^{(b ) } \end{array}\right ) \end{array}\ ] ] proves ( [ equation : catalanscaling ] ) . at the second equality we applied scaling rule ( [ equation : adelicrationalscaling ] )
finally , the calculation @xmath681(a^{-1}\xi)}\\\\ & = & \displaystyle \prod_{x\in k^\times}x^{\tilde{\phi}(ax){{\left\|a\right\|}}q } \cdot \prod_{\xi\in k^\times } \xi^{(1+\operatorname{{\mathcal{n}}})[\phi](a^{-1}\xi)}\\\\ & = & \displaystyle\left(\begin{array}{c } a\\ \phi \end{array}\right)^q \prod_{c\in { { \mathbb{f}}}_q^\times}\prod_{\xi\in k^\times } c^{\phi(a^{-1}\xi)}\\\\&= & \left(\begin{array}{c } a\\ \phi \end{array}\right)^q(-1)^{\theta(a,\phi)-\phi(0 ) } \end{array}\ ] ] proves ( [ equation : gfe2 ] ) . at the second equality we applied formula ( [ equation : relatedopidentitiesbis ] ) for @xmath682 .
[ [ section-27 ] ] for all @xmath38 , @xmath397 and @xmath677 we have @xmath683 we note this for convenient reference .
the proof is more or less the same as for ( [ equation : ideleclassinvariancebis ] ) .
[ [ section-28 ] ] the rational fourier transform and the catalan symbol do in fact depend on the choice of differential @xmath498 , even though we have suppressed reference to @xmath498 in the notation . to clarify this dependence
let us temporarily write @xmath684 and @xmath685 .
then for all @xmath686 , @xmath549 and @xmath544 we have @xmath687(\xi)=\operatorname{{\mathcal{f}}}_{0,\omega}[\phi](t\xi)=\operatorname{{\mathcal{f}}}_{0,\omega}[\phi]^{(t^{-1})}(\xi).\ ] ] in turn , for all @xmath38 , @xmath677 and @xmath686 , we have @xmath688 by ( [ equation : partialcatalanscaling ] ) above , functional equation ( [ equation : thetarationalfe ] ) obeyed by the theta symbol , and the definitions .
[ [ section-29 ] ] rewritten in terms of the catalan symbol and the rational fourier transform , with attention restricted to schwartz functions taken from the group @xmath653 , the adelic stirling formula ( [ equation : adelicstirling ] ) takes a greatly simplified form , namely @xmath689 for all @xmath38 , @xmath677 and places @xmath80 of @xmath34 .
clearly , the right side of ( [ equation : adelicstirlingbis ] ) is independent of the differential @xmath498 . as a consistency check ,
note that ( [ equation : catalanchangeofdifferential ] ) forces the left side of ( [ equation : adelicstirlingbis ] ) to be independent of @xmath498 .
[ theorem : genericbehavior ] fix @xmath690 such that @xmath691 then the map @xmath692 is @xmath5-linear .
thus , besides obvious @xmath67$]-linearity ( @xmath652 to @xmath693 ) , the catalan symbol has a `` hidden '' @xmath5-linearity . as the proof shows , the theorem is essentially just a rehash of propositions [ proposition : easymooreapp ] and [ proposition : dualityclincher ] . by scaling rule ( [ equation : catalanscaling ] )
, we may assume without loss of generality that @xmath427 .
let @xmath694 denote the map in question ( with @xmath427 ) .
our task is to prove that @xmath694 is @xmath5-linear .
to do so , we distinguish two cases , namely : ( i ) @xmath695 and ( ii ) @xmath696 .
we turn to case ( i ) . put @xmath697 for all @xmath698 , the quantity @xmath699 is positive and independent of @xmath205 by proposition [ proposition : thetaasymptotics ] .
it follows that the inclusion - induced natural map @xmath700 is bijective .
it suffices to prove the @xmath5-linearity of the map @xmath701 obtained by following the isomorphism @xmath700 by @xmath694 .
choose @xmath702 such that @xmath703 .
by proposition [ proposition : genericcatalanbehavior ] and the definition of the toy catalan symbol we have @xmath704 for all @xmath705 , where @xmath706 is the inclusion @xmath707 .
therefore the map @xmath701 is @xmath5-linear by proposition [ proposition : easymooreapp ] .
thus case ( i ) is proved .
we turn to case ( ii ) . in the obvious way
let us now identify the space @xmath708 of complex - valued functions on the finite set @xmath709 with the subspace of @xmath529 consisting of functions constant on cosets of @xmath436 and supported in @xmath562 . by example
( [ equation : rationalfourierexample ] ) and the remark following we have @xmath710 \subset{{\mathrm{sch}}}(b^{-1}\kappa^{-1}{{\mathcal{o}}}/ fb^{-1}\kappa^{-1}{{\mathcal{o}}}).\ ] ] but the spaces @xmath709 and @xmath711 are serre dual with respect to the pairing @xmath712 and so we also have at our disposal a toy rational fourier transform @xmath713 by comparing the examples ( [ equation : toyrationalfourierexample ] ) and ( [ equation : rationalfourierexample ] ) it can be seen that there exists an integer @xmath714 such that @xmath715=\operatorname{{\mathcal{f}}}_0[\phi]\ ] ] for all @xmath716 .
fix @xmath717 arbitrarily . by case ( i )
already proved , the map @xmath718 is @xmath5-linear .
let @xmath719 be the @xmath5-linear map obtained by multiplying @xmath720 by the factor @xmath721 . by proposition [ proposition : genericcatalanbehavior ]
, the map @xmath719 takes the form @xmath722 note that @xmath719 is injective . by proposition [ proposition : genericcatalanbehavior ] and the definition of the toy catalan symbol
we have @xmath723 \end{array}\right)_+\\\\ & = & \left(\begin{array}{c } \beta\\ \operatorname{{\mathcal{f}}}_0^{{\mathrm{toy}}}[{{\mathbf{1}}}_{x+b{{\mathcal{o}}}}-{{\mathbf{1}}}_{x_0+b{{\mathcal{o } } } } ] \end{array}\right)^{q^\ell } \end{array}\ ] ] for all @xmath724 .
therefore the map @xmath694 is @xmath5-linear by proposition [ proposition : dualityclincher ] .
thus case ( ii ) is proved .
in this section we state the conjecture ( conjecture [ theconjecture ] ) to which we allude in the title of the paper . the conjecture links theta and catalan symbols to two - variable algebraic functions of a certain special type . following the conjecture
we make some amplifying remarks .
[ [ section-30 ] ] put @xmath726 it is not difficult to verify that the ring @xmath727 is a noetherian integrally closed domain of dimension one , and hence a dedekind domain . clearly , the ideal @xmath728 is maximal in @xmath727 , and by construction has residue field canonically identified with @xmath34 .
let @xmath729 be the completion of @xmath727 with respect to @xmath728 , and let @xmath730 be the closure of @xmath728 in @xmath729 .
then @xmath729 is a discrete valuation ring with residue field canonically identified with @xmath34 and with maximal ideal @xmath730 .
put @xmath731{\hat{\delta}_0}.\ ] ] then @xmath725 is a ( nondiscrete ) valuation ring with residue field canonically identified with @xmath732 and with maximal ideal @xmath33 .
[ [ section-31 ] ] we identify the universal derivation @xmath733 with the map @xmath734 thus fixing an identification @xmath735 we fix a generator @xmath263 of the principal ideal @xmath730 such that @xmath736 we call @xmath263 a _ lifting _ of @xmath498 . [ [ section-32 ] ] every nonzero element @xmath15 of the fraction field of @xmath729 has a unique @xmath730-adic expansion of the form @xmath737 and in terms of this expansion we define @xmath738 also put @xmath739 . more generally , for all nonzero @xmath15 in the fraction field of @xmath725 , we define @xmath740,\;\;\ ; \operatorname{{\mathrm{lead}}}_\delta \varphi=(\operatorname{{\mathrm{lead}}}_{\delta}\varphi^{q^n})^{q^{-n}}\in k_{{\mathrm{perf}}}^\times\ ] ] for all sufficiently large @xmath666 .
the function @xmath48 is an additive valuation of the fraction field of @xmath725 independent of the choice of differential @xmath498 and lifting @xmath263 .
the function @xmath741 is a homomorphism from the multiplicative group of the fraction field of @xmath725 to @xmath693 depending only on the differential @xmath498 , not on the lifting @xmath263 .
[ [ section-33 ] ] we clarify the dependence of @xmath741 on the choice of differential @xmath498 as follows .
let us temporarily write @xmath742 to stress the @xmath498-dependence .
then we have @xmath743 for all @xmath686 and nonzero elements @xmath15 of the fraction field of @xmath725 .
this ought to be compared with equation ( [ equation : catalanchangeofdifferential ] ) above . [
[ subsubsection : dagger1 ] ] the exchange - of - factors automorphism @xmath744 preserves @xmath728 , hence has a unique @xmath730-adically continuous extension to an automorphism of @xmath729 , and hence has a unique extension to an automorphism of the fraction field of @xmath725 .
the result of applying the latter automorphism to an element @xmath15 of the fraction field of @xmath725 we denote by @xmath745 .
we have @xmath746 for all nonzero @xmath15 in the fraction field of @xmath725 .
[ [ subsubsection : renormalization ] ] for any abelian extension @xmath0 , let @xmath748 be the reciprocity law homomorphism of global class field theory defined according to the nowadays commonly followed convention of tate s paper @xcite .
it is important to bear in mind two facts concerning this convention , which differs from the older convention of , say , artin - tate @xcite .
firstly , if @xmath80 is a place of @xmath34 unramified in @xmath115 , and @xmath749 is a uniformizer at @xmath80 , then @xmath750 is a geometric frobenius element ( inverse of the usual artin symbol ) at @xmath80 .
secondly , we have @xmath751 for all @xmath38 and constants @xmath283 .
[ [ section-34 ] ] let @xmath314 be the abelian closure of @xmath34 in the algebraic closure @xmath315 .
let @xmath752 be the abelian ( also the algebraic ) closure of @xmath5 in @xmath315 .
note that @xmath753 and that the natural action of @xmath64 on @xmath754 commutes with the @xmath11 power automorphism of @xmath754 .
the group @xmath755\in\operatorname{{\mathrm{gal}}}(k^{{\mathrm{ab}}}/k)\times\operatorname{{\mathrm{gal}}}(k^{{\mathrm{ab}}}/k)\left| \sigma\vert_{{{\mathbb{f}}}_q^{{\mathrm{ab}}}}= \tau\vert_{{{\mathbb{f}}}_q^{{\mathrm{ab}}}}\right.\right\}\ ] ] acts naturally on the integral domains @xmath756 by the rule @xmath757}=x^\sigma\otimes y^\tau,\ ] ] with fixed rings @xmath727 and @xmath725 , respectively .
put @xmath758\vert \sigma\in \operatorname{{\mathrm{gal}}}(k^{{\mathrm{ab}}}/k)\},\ ] ] @xmath759 note that @xmath760 .
note that @xmath761 may be identified with the galois groups of the etale ring extensions @xmath762 and @xmath763 under the map @xmath764 $ ] .
[ [ subsubsection : dagger2 ] ] let @xmath38 be given .
the automorphism @xmath765 commutes with the action of the group @xmath766 , hence descends to an automorphism of @xmath747 and hence has a unique extension to an automorphism of the fraction field of @xmath747 .
the result of applying the latter automorphism to an element @xmath15 of the fraction field of @xmath747 we denote by @xmath767 .
thus we equip the fraction field of @xmath747 with an action of @xmath43 factoring through the idele class group @xmath558 . [
[ section-35 ] ] the exchange - of - factors automorphism @xmath768 commutes with the action of the group @xmath766 , hence descends to an automorphism of @xmath747 , and hence has a unique extension to an automorphism of the fraction field of @xmath747 .
the result of applying the latter automorphism to an element @xmath15 of the fraction field of @xmath747 we denote by @xmath745 .
we have @xmath769 for all @xmath15 in the fraction field of @xmath747 and @xmath38 .
\(i ) there exists a unique @xmath727-algebra embedding @xmath770 such that @xmath771 ( ii ) moreover , @xmath772 commutes with @xmath773 .
after the proof of this proposition , in order to keep notation simple , we just identify @xmath747 with its image in @xmath725 under @xmath772 , and dispense with the notation @xmath772 altogether . in this way @xmath774 and @xmath775
are defined for all nonzero @xmath15 in the fraction field of @xmath747 , and moreover both ( [ equation : daggersymmetrybis ] ) and ( [ equation : daggersymmetry ] ) hold . part ( i )
granted , the @xmath727-algebra embedding @xmath776 has the property uniquely characterizing @xmath772 and hence coincides with @xmath772 .
it is enough to prove part ( i ) . in turn
, it is enough to prove that there exists a unique @xmath727-algebra embedding @xmath777 such that @xmath778 because once this is proved , the induced homomorphism @xmath779 is the only possibility for @xmath772
. now let @xmath0 be any finite abelian extension and consider the following naturally associated objects : * let @xmath319 .
* let @xmath780 be the constant field of @xmath115 .
* let @xmath781\in g\times g\vert\sigma\vert_{{{\mathbb{f}}}}= \tau\vert_{{\mathbb{f}}}\}$ ] .
* let @xmath782 act on @xmath783 by the rule @xmath784}=x^\sigma\otimes y^\tau$ ] .
* let @xmath785\vert \sigma \in g\}\subset g'$ ] .
* put @xmath786 . *
put @xmath787 .
it is enough to prove that there exists a unique @xmath727-algebra homomorphism @xmath788 such that @xmath789 , because once this is proved the limit @xmath790 extended over all finite subextensions @xmath0 of @xmath316 is the only possibility for @xmath791 . now
the ring @xmath783 is a dedekind domain finite etale and abelian over @xmath727 .
it follows that the @xmath792-fixed subring @xmath169 has these same properties by descent .
the prime @xmath793 lies above the prime @xmath794 and we have @xmath795 . since @xmath169 is etale over @xmath727 , existence and uniqueness of @xmath796
follow now by the infinitesimal lifting criterion .
we declare a nonzero element @xmath15 of the fraction field of @xmath747 to be a _ coleman unit _ if for every maximal ideal @xmath797 not of the form @xmath798 for some @xmath38 , we have @xmath799 , where @xmath800 is the local ring of @xmath747 at @xmath801 .
the coleman units form a group under multiplication . for every coleman unit @xmath15 and @xmath38 , again @xmath767 is a coleman unit , and so also is @xmath745 .
the group of coleman units is closed under the extraction of @xmath11 roots .
the inspiration for this definition comes from coleman s paper @xcite .
[ theconjecture ] for every @xmath677 , there exists a unique coleman unit @xmath15 such that @xmath802 for all @xmath38 .
[ [ section-36 ] ] we have , so we claim , the following _ uniqueness principle _ : for every @xmath803 in the fraction field of @xmath747 , @xmath804 in any case , we can find a dedekind domain @xmath169 between @xmath747 and @xmath727 which is finite over @xmath727 and to which @xmath803 belongs .
since the maximal ideals of @xmath727 of the form @xmath805 are distinct , the hypothesis implies that @xmath803 has nonzero valuation at infinitely many maximal ideals of @xmath169 and hence vanishes identically .
the claim is proved . by the uniqueness principle ,
for each @xmath806 , there can be at most one element @xmath15 of the fraction field of @xmath747 satisfying ( [ equation : conjecturalattachment ] ) for all @xmath614 .
so the uniqueness asserted in the conjecture is clear .
only existence is at issue .
[ [ subsubsection : simnotation ] ] let nonzero @xmath15 in the fraction field of @xmath747 , @xmath807 , and @xmath38 be given .
we write @xmath808 $ ] if ( [ equation : conjecturalattachment ] ) holds for the data @xmath15 , @xmath99 and @xmath68 .
we write @xmath809 $ ] if @xmath808 $ ] for all @xmath38 and @xmath15 is a coleman unit . in the situation
@xmath809 $ ] , as the notation is meant to suggest , we think of @xmath15 as the image of @xmath68 under a transformation @xmath810 . from the latter point of view
, the conjecture asserts the well - definedness of this transformation @xmath810 on the whole of the space @xmath653 .
[ [ section-37 ] ] let @xmath15 in the fraction field of @xmath747 and @xmath677 be given .
comparison with ( [ equation : conjecturalattachment ] ) of the formula ( [ equation : catalanchangeofdifferential ] ) explaining the dependence of @xmath811 on @xmath498 and the formula ( [ equation : leadchangeofdifferential ] ) explaining the dependence of @xmath741 on @xmath498 shows that if @xmath809 $ ] , then no matter what nonzero differential @xmath812 we choose to define rational fourier transforms , catalan symbols , and `` leading taylor coefficients at @xmath33 '' , it remains the case that @xmath809 $ ] . in short
, @xmath810 is independent of @xmath498 .
so if the conjecture is true for one choice of @xmath498 , it is true for all .
[ [ section-38 ] ] grant the conjecture for this paragraph , so that the transformation @xmath810 is defined .
clearly , we have @xmath813={{\mathcal{c}}}[\phi_1]^{n_1}{{\mathcal{c}}}[\phi_2]^{n_2}\ ] ] for all @xmath814 and @xmath815 $ ] .
comparison with ( [ equation : conjecturalattachment ] ) of the scaling rule ( [ equation : thetascaling ] ) for the theta symbol and the scaling rule ( [ equation : catalanscaling ] ) for the catalan symbol shows that @xmath816={{\mathcal{c}}}[\phi]^{(a)}\ ] ] for all @xmath38 and @xmath677 .
comparison with ( [ equation : conjecturalattachment ] ) of the functional equation ( [ equation : thetarationalfe ] ) for the theta symbol , the functional equation ( [ equation : gfe2 ] ) for the catalan symbol , and the properties ( [ equation : daggersymmetrybis ] ) and ( [ equation : daggersymmetry ] ) of the dagger operation on the fraction field of @xmath747 shows that @xmath817={{\mathcal{c}}}[\phi]^\dagger\ ] ] for all @xmath677 . for any fixed @xmath546 and @xmath818 the map @xmath819&\mbox{if $ x\not\in b{{\mathcal{o}}}$}\\ 0&\mbox{if $ x\in b{{\mathcal{o}}}$ } \end{array}\right.\right):{{\mathbb{a}}}/b{{\mathcal{o}}}\rightarrow { { \bf k}}\ ] ] is @xmath5-linear by the `` hidden '' @xmath5-linearity of the catalan symbol ( theorem [ theorem : genericbehavior ] ) and the uniqueness principle ( [ equation : uniquenessprinciple ] ) .
[ [ section-39 ] ] we return now to the general discussion of the conjecture , no longer assuming that it holds .
it is easy to verify that the set of @xmath677 for which @xmath820 $ ] is defined is a @xmath67[{{\mathbb{a}}}^\times]$]-submodule of @xmath653 .
it follows that to prove the conjecture it is enough to fix a set of generators for @xmath653 as a @xmath67[{{\mathbb{a}}}^\times]$]-module and for each generator @xmath68 to prove the existence of @xmath820 $ ] .
for example , for any fixed @xmath638 and @xmath818 , the family @xmath821 is large enough to generate @xmath653 as a @xmath67[{{\mathbb{a}}}^\times]$]-module .
[ [ section-40 ] ] here are some trivial but handy cases in which we can easily prove that the transformation @xmath810 is defined .
fix @xmath822 .
fix @xmath397 .
note that @xmath823 .
we claim that @xmath824=x^{-\phi(0)}\otimes x^{\tilde{\phi}(0)}.\ ] ] in any case , the right side is a unit of @xmath747 and _ a fortiori _ a coleman unit ; further , we have @xmath825 for all @xmath38 by ( [ equation : thetaideleclassdependence ] ) and ( [ equation : thetascaling ] ) ; and finally we have @xmath826 for all @xmath38 by ( [ equation : ideleclassinvariancebis ] ) and ( [ equation : catalanscaling ] ) .
this is enough to prove the claim .
we assume in [ section : genuszero ] that @xmath827 and under this additional assumption we are going to prove conjecture [ theconjecture ] .
every function field of genus zero with field of constants @xmath5 has a place with residue field @xmath5 , and as we noted in [ subsection : ccalculus ] , the conjecture is invariant under change of differential .
we therefore may assume without loss of generality that @xmath828 where @xmath12 is transcendental over @xmath5 .
we take @xmath829 as a lifting of @xmath498 .
note that @xmath263 is already a prime element of the principal ideal domain @xmath727 .
note also that @xmath263 is a coleman unit .
let @xmath50 be the unique place of @xmath34 at which @xmath12 has a pole and put @xmath830 noting that @xmath831 put @xmath832\in { { \mathcal{o}}}^\times \left|\ ; { { \left\|a_\infty-1\right\|}}_\infty<1\right.\right\}.\ ] ] then every @xmath38 factors uniquely in the form @xmath833 we use this factorization repeatedly below .
for each @xmath415 , write @xmath834,\;\;{{\left\|\langle x\rangle\right\|}}_\infty<1)\ ] ] in the unique possible way , in parallel with the definitions made in
[ subsubsection : brackets ] . put @xmath835 for @xmath836 $ ] . for each @xmath397 , put @xmath837 the set @xmath838 generates @xmath653 as a @xmath67[{{\mathbb{a}}}^\times]$]-module , and so to prove the conjecture in the case at hand it suffices to prove that @xmath839 $ ] exists for every @xmath397 .
let @xmath840 and @xmath397 be given .
we calculate @xmath841 and @xmath842 .
[ [ section-41 ] ] assume at first that @xmath20 , which is the relatively easy case .
we shall handle the case @xmath843 presently .
we have @xmath844\vert\deg a < n\}. \end{array}\ ] ] therefore we have @xmath845 @xmath846\\ \deg a < n \end{subarray } } \frac{a+\langle x\rangle+t^n\lfloor x\rfloor}{a+t^n},\ ] ] by proposition [ proposition : genericcatalanbehavior ] at the first equality of ( [ equation : positivecatalancalc ] ) , and at the other two equalities by direct appeal to the definitions . [ [ section-42 ] ] we now take a close look at the rational fourier transform @xmath847 in order to prepare for handling the case @xmath843 . by specializing the example ( [ equation : rationalfourierexample ] ) of an adelic rational fourier transform to the present case ( we may take @xmath848 because @xmath849 has a double pole at @xmath50 and no other poles or zeroes ) , we have @xmath850= { { \mathbf{1}}}_{\tau{{\mathcal{o}}}\cap[\langle x,\cdot\rangle=0 ] } -{{\mathbf{1}}}_{\tau{{\mathcal{o}}}\cap[\langle x,\cdot\rangle=1]}\ ] ] and hence @xmath851\;\;\;(\psi^\star_x=-{{\mathbf{1}}}_{\tau{{\mathcal{o}}}\cap[\langle x,\cdot\rangle=1]}+{{\mathbf{1}}}_{\tau{{\mathcal{o}}}\cap[\langle 1,\cdot\rangle=1]}),\ ] ] where to get ( [ equation : tildepsi ] ) from the preceding formula we reused the `` starred '' trick from the proof of proposition [ proposition : dualityclincher ] .
it follows ( without any restriction on @xmath97 ) that @xmath852 by functional equation ( [ equation : thetarationalfe ] ) at the first equality , and scaling rule ( [ equation : thetascaling ] ) at the second .
it follows in turn ( again without restriction on @xmath97 ) that @xmath853 by scaling rule ( [ equation : partialcatalanscaling ] ) followed by an application of ( [ equation : rationalpsistar ] ) .
[ [ section-43 ] ] we now handle the remaining case @xmath843 of our calculation . put @xmath854 noting that @xmath855 .
\deg a\leq \nu\}= \tau^{n+1}{{\mathcal{o}}}\cap k,\ ] ] @xmath857 noting that @xmath858 the latter relations are verified by applying `` sum of residues equals zero '' and then discarding terms not contributing to the residue at @xmath859 . since @xmath860 ) = \{a\in v\vert \xi(a)=-1\},\ ] ] we have via ( [ equation : rationalpsistar ] ) that @xmath861+{{\mathbb{f}}}_q[t]$,}\\ 1&\mbox{if $ \langle x\rangle - t^{-\nu-1}\lfloor x\rfloor\in t^{-\nu-2}{{\mathbb{f}}}_q[[1/t]]+{{\mathbb{f}}}_q[t]$. } \end{array}\right . \end{array}\ ] ] it follows that @xmath862 finally , it follows via ( [ equation : carlitzrationalfourierter ] , [ equation : intermediate1 ] , [ equation : negativethetacalc ] , [ equation : intermediate2 ] ) and the definitions that @xmath863 where @xmath864 cf . equation ( [ equation : dn ] ) .
the calculation is complete .
we study a certain @xmath866-valued point of the group - scheme @xmath167 introduced in
[ section : functorsandidentities ] .
[ [ section-44 ] ] given @xmath867 , the set @xmath868 consists of a single element @xmath836 $ ] such that @xmath869 .
that noted , it is clear that there is a unique group isomorphism @xmath870):{{\mathcal{u}}}{{\stackrel{\sim}{\rightarrow}}}{{\mathcal{h}}}({{\mathbb{f}}}_q)\ ] ] such that @xmath871_\infty(t)=u_\infty(t)\;\;\;(u_\infty = u_\infty(t)\in 1+(1/t){{\mathbb{f}}}_q[[1/t]]\subset{{\mathcal{o}}}_\infty^\times)\ ] ] and @xmath871_f(t)=a(t)\;\;\;(a = a(t)\in k\cap ( u+f\tau{{\mathcal{o}}}))\ ] ] for every monic @xmath872 $ ] .
let @xmath873 be a separable algebraic closure of @xmath34 of which @xmath316 is a subextension .
let @xmath874 be the @xmath11 power frobenius endomorphism .
according to geometric class field theory ( see @xcite ) the set of solutions @xmath875 of the equation @xmath876x\ ] ] forms an @xmath877-torsor , and for any solution @xmath76 we have an explicit reciprocity law @xmath878x\;\;\;(u\in { { \mathcal{u}}},\;z\in k^\times , \;n\in { { \mathbb z}}).\ ] ] in particular , we always have @xmath879 .
[ [ section-45 ] ] put @xmath880^{-1 } [ t-1\otimes t^{q^i}]\right)\in { { \mathcal{h}}}(\hat{{{\bf k}}}_0).\ ] ] the product is convergent because @xmath881^{-1 } [ t-1\otimes t^{q^i}]\equiv 1\bmod{\varpi^{q^i}}\ ] ] for all @xmath577 . since the lang isogeny @xmath882 is etale , the functional equation and congruence @xmath883[t-1\otimes t]^{-1}{{\mathbf{p}}},\;\;\;{{\mathbf{p}}}\equiv 1\bmod{\varpi}\ ] ] characterize @xmath865 uniquely in @xmath884 .
[ [ section-46 ] ] let @xmath885 be any solution of ( [ equation : langtorsor ] ) , let @xmath886 be the image of @xmath76 under the map induced by the ring homomorphism @xmath887
let @xmath888 be defined analogously , and put @xmath889 then , so we claim , we have @xmath890 the equality holds because ( i ) @xmath891 satisfies the conditions ( [ equation : bigpcharacterization ] ) uniquely characterizing @xmath865 , and ( ii ) @xmath891 is @xmath766-invariant by the explicit reciprocity law ( [ equation : explicitreciprocity ] ) .
the claim is proved .
[ [ section-47 ] ] by combining the observations of the preceding three paragraphs , for any @xmath38 , we can calculate the image @xmath892 of @xmath865 under the map @xmath893 induced by the ring homomorphism @xmath894 .
we have @xmath895{{\mathbf{p}}}\left\{\begin{array}{ll } \displaystyle\prod_{i=0}^{n}[t-1\otimes t^{q^i}]^{-1}&\mbox{if $ n\geq 0$,}\\\\ \displaystyle\prod_{i=1}^{\nu}[t-1\otimes t^{q^{-i}}]&\mbox{if $ n<0 $ , } \end{array}\right.\ ] ] for all @xmath867 , @xmath896 and @xmath897 , where @xmath898 .
[ lemma : straighttheta ] for all @xmath899 and @xmath38 we have @xmath900.\ ] ] recall that @xmath227 was defined in
[ subsubsection : brackets ] and @xmath901 in [ subsubsection : simnotation ] .
write @xmath902 in the unique possible way . without loss of generality
we may assume that @xmath903 .
fix monic @xmath872 $ ] such that @xmath904 $ ] and let @xmath905 be uniquely characterized by the relations @xmath871_f(t)\langle x(t)\rangle \equiv \langle\tilde{x}(t)\rangle\bmod{{{\mathbb{f}}}_q[t]},\ ] ] @xmath871_\infty(t)\lfloor x(t)\rfloor\equiv\lfloor \tilde{x}(t)\rfloor\bmod{(1/t){{\mathbb{f}}}_q[[1/t]]},\ ] ] in which case ( equivalently ) @xmath906 let @xmath907 be defined by the expression on the left side of ( [ equation : straighttheta ] ) .
then we have @xmath908{{\mathbf{p}}}^{(\tau^{-n-1})})= \theta_{\tilde{x}(t)}({{\mathbf{p}}}^{(\tau^{-n-1})})= \psi_{\tilde{x}}^{(\tau^{-n})}\ ] ] and @xmath909 after replacing @xmath205 by @xmath910 , we may simply assume that @xmath911 .
all the preceding reductions taken into account , it will be enough to prove that @xmath912 if @xmath20 , then we have via ( [ equation : bigpcharacterization ] ) and ( [ equation : boldptwist ] ) that @xmath913^{-1}\right ) \bmod{\hat{\delta}_0},\ ] ] and this noted , ( [ equation : simclincher ] ) follows by comparing proposition [ proposition : jacobitrudiredux ] to formulas ( [ equation : positivethetacalc ] ) and ( [ equation : positivecatalancalc ] ) . if @xmath914 , then we have via ( [ equation : bigpcharacterization ] ) and ( [ equation : boldptwist ] ) , along with the definition of @xmath865 , that @xmath915\right)\\ & \equiv & \theta_{x(t)}\left([t - t^{q^\nu}\otimes 1]^{-1}\prod_{i=0}^{\nu } [ t-1\otimes t^{q^i}]\right ) \bmod{\hat{\delta}_0^{q^\nu+1 } } , \end{array}\ ] ] and this noted , ( [ equation : simclincher ] ) follows by comparing propositions [ proposition : dualjacobitrudiredux ] and [ proposition : peerintothehole ] to formulas ( [ equation : negativethetacalc ] ) and ( [ equation : negativecatalancalc ] ) .
[ lemma : usefulcolemanunits ] fix @xmath397 and an integer @xmath916 .
put @xmath917 then there exists a coleman unit @xmath918 such that @xmath919 for all @xmath920 . by example
( [ equation : trivialsolitoncalculation ] ) we may assume without loss of generality that @xmath921 .
now write @xmath922 to keep track of dependence on @xmath666 .
we have @xmath923 so the lemma holds trivially for @xmath924 , and holds for @xmath925 by another application of ( [ equation : trivialsolitoncalculation ] ) . for all @xmath926 we have @xmath927 where @xmath928 by induction on @xmath666 ( twice ) it suffices to find a coleman unit @xmath929 such that @xmath930 @xmath931 for @xmath932
clearly , @xmath933 has the desired properties .
fix @xmath397 . by lemma [ lemma : straighttheta ]
it remains only to prove that @xmath934 is a coleman unit .
fix monic @xmath935 $ ] and an integer @xmath936 .
note that @xmath937 and hence @xmath938 is a finite union of cosets of @xmath939 .
consider the product @xmath940 since all the factors in the product by construction are nonzero elements of @xmath747 , and for suitable @xmath364 and @xmath666 the function @xmath941 is found among the factors on the right , it suffices simply to show that @xmath15 is a coleman unit .
consider the function @xmath942 by lemma [ lemma : straighttheta ] we have @xmath808 $ ] for all @xmath38 .
it follows by lemma [ lemma : usefulcolemanunits ] and the uniqueness principle ( [ equation : uniquenessprinciple ] ) that @xmath15 is a coleman unit .
the proof of conjecture [ theconjecture ] in the case @xmath827 is complete .
[ [ section-48 ] ] the _ solitons _ defined in ( * ? ? ?
2 ) , and the _ coleman functions _ applied in @xcite , @xcite and @xcite are all characterized by `` interpolation formulas '' with righthand sides that can be put into the form of the righthand side of equation ( [ equation : positivecatalancalc ] ) above .
a similar remark applies to the formula ( [ equation : thakurinterpolation ] ) stated in the introduction .
for example , by substituting @xmath943 for @xmath205 in the righthand side of ( [ equation : positivecatalancalc ] ) , one recovers the righthand side of the interpolation formula stated in ( * ? ? ? * thm . 2 ) .
for another example , by substituting @xmath944 in the righthand side of ( [ equation : positivecatalancalc ] ) , one recovers the righthand side of formula ( [ equation : thakurinterpolation ] ) .
it follows via the uniqueness principle ( [ equation : uniquenessprinciple ] ) that solitons and coleman functions belong to the class of two - variable algebraic functions engendered by conjecture [ theconjecture ] in the genus zero case .
[ [ section-49 ] ] what about the higher genus case ?
we have four general methods at our disposal for attacking the problem , namely the methods of `` rational fourier analysis '' discussed in this paper , along with the methods discussed in the author s papers @xcite , @xcite , and @xcite .
we are currently working with all these methods in an effort to find a proof of the conjecture .
we return to our studies at the full level of generality of conjecture [ theconjecture ] .
we study embeddings of certain subrings of @xmath747 into certain power series rings .
then we draw some general conclusions concerning the operation of `` setting the second variable equal to a constant '' in a coleman unit , expressing the results in the form of an easy - to - use calculus . only at the very end of this section
do we draw any conclusions conditional on conjecture [ theconjecture ] .
fix a field @xmath780 .
[ [ section-50 ] ] let @xmath168 , @xmath945 and @xmath946 be independent variables .
let @xmath947 be the field obtained from the one - variable power series ring @xmath948 $ ] by inverting @xmath168 .
let @xmath949 be the principal ideal domain obtained from the two - variable power series ring @xmath950 $ ] by inverting @xmath945 and @xmath946 .
given @xmath951 , put @xmath952 and let the involutive automorphism @xmath953 of @xmath949 be extended in the unique possible way to the fraction field of @xmath949 .
[ [ section-51 ] ] let @xmath954 be given .
there is a unique factorization @xmath955 where @xmath956 and @xmath957 $ ] is divisible by neither @xmath945 nor @xmath946 . in turn
, there is by the weierstrass preparation theorem a unique factorization @xmath958 where @xmath959^\times$ ] , @xmath203 is a nonnegative integer and @xmath960 $ ] is a polynomial in @xmath945 of degree @xmath961 .
we now define @xmath962=\epsilon[f](t)=t^{\ell_1+m}u(t,0)\in { { \mathbb{f}}}((t))^\times.\ ] ] also put @xmath963 .
we have @xmath964\in { { \mathbb{f}}}[[t]]^\times\leftrightarrow m=0\leftrightarrow f\in { { \mathbb{f}}}((t_1,t_2))^\times.\ ] ] the function @xmath965 is an additive valuation of @xmath949 .
we extend @xmath965 in the unique possible way to a normalized additive valuation of the fraction field of @xmath949 .
the @xmath966-valued function @xmath323 defined on @xmath967 is multiplicative .
we extend @xmath323 in the unique possible way to a homomorphism from the multiplicative group of the fraction field of @xmath949 to @xmath966 .
fix a finite subextension @xmath0 of @xmath968 .
let @xmath780 be the field of constants of @xmath115 . put @xmath969}=k\otimes_{{{\mathbb{f}}}}k$ ] , which is a dedekind domain .
[ [ section-52 ] ] given a place @xmath528 of @xmath115 with residue field equal to @xmath780 , a uniformizer @xmath970 in the completion @xmath971 , and @xmath972 , let @xmath973 be the unique laurent series such that the equality @xmath974 holds in @xmath971 .
put @xmath975}\rightarrow k((t)),\ ] ] thus defining an embedding of the ring @xmath969}$ ] into the field @xmath976 .
we extend @xmath977 in the unique possible way to an embedding of the fraction field of @xmath969}$ ] into @xmath976 .
[ [ section-53 ] ] with @xmath528 and @xmath970 as above , and given nonzero @xmath15 in the fraction field of @xmath969}$ ] , write @xmath978 and put @xmath979 also put @xmath980 .
the function @xmath981 is independent of @xmath970 .
we write @xmath982 hereafter .
the function @xmath982 is a normalized additive valuation of the fraction field of @xmath969}$ ] .
the function @xmath983 is a homomorphism from the multiplicative group of the fraction field of @xmath969}$ ] to @xmath984 .
while @xmath983 does depend on @xmath970 , the dependence is rather mild : given another uniformizer @xmath985 at @xmath528 , the ratio @xmath986 is a homomorphism from the fraction field of @xmath969}$ ] to @xmath987 of the form @xmath988 for some @xmath989 .
[ [ section-54 ] ] let @xmath528 , @xmath970 and @xmath15 be as above , but assume now that @xmath15 belongs to the fraction field of @xmath969}\cap{{\bf k}}$ ] . fix @xmath38 such that @xmath990 . then @xmath767 is defined and belongs again to the fraction field of @xmath969}\cap { { \bf k}}$ ] . we claim that @xmath991 in any case , notation as in ( [ equation : prewrite ] ) , we must have @xmath992 on account of the rule @xmath993 by which @xmath767 is defined
. the claim follows .
[ [ section-55 ] ] given for @xmath196 a place @xmath994 of @xmath115 with residue field @xmath780 and a uniformizer @xmath995 in @xmath996 , put @xmath997}\rightarrow { { \mathbb{f}}}((t_1,t_2)),\ ] ] thus defining an embedding of the ring @xmath969}$ ] into @xmath949 .
we extend @xmath998 in the unique possible way to an embedding of the fraction field of @xmath969}$ ] into the fraction field of @xmath949 . clearly , we have @xmath999 it is not difficult to verify that the equality @xmath1000(\pi_1)\ ] ] holds in @xmath1001 . if @xmath15 belongs to the fraction field of @xmath1002}$ ] , then @xmath745 is defined and belongs again to the fraction field of @xmath1002}$ ] , and we have @xmath1003 on account of the rule @xmath1004 defining the `` dagger '' operation .
let @xmath115 , @xmath780 and @xmath969}$ ] be as in
[ subsection : embeddings ] .
for @xmath196 let @xmath994 be a place of @xmath115 with residue field equal to @xmath780 , let @xmath995 be a uniformizer in @xmath996 , and let @xmath1005 be the unique place of @xmath34 below @xmath994 .
assume that @xmath1006 .
fix @xmath38 and put @xmath1007}\rightarrow k_{{\mathrm{perf}}}\right).\ ] ] then there exists @xmath1008 such that @xmath1009^\times.\ ] ] for any @xmath972 we have @xmath1010 since the additive valuations @xmath1011 are inequivalent ( they already have inequivalent restrictions to @xmath34 by hypothesis ) , we can find @xmath972 by the artin - whaples approximation theorem such that @xmath1012 then @xmath1013 has the desired property .
[ proposition : offdiagonal ] let @xmath115 , @xmath780 and @xmath969}$ ] be as in [ subsection : embeddings ] . for @xmath196 let @xmath994 be a place of @xmath115 with residue field equal to @xmath780 ,
let @xmath995 be a uniformizer in @xmath996 , and let @xmath1005 be the unique place of @xmath34 below @xmath994 .
assume that @xmath1006 .
for all coleman units @xmath15 belonging to the fraction field of @xmath969}$ ] , we have @xmath1014 . by the preceding lemma and the definition of a coleman unit
, we can find @xmath1015}$ ] such that @xmath1016},\;\;\iota_{k , w_1,w_2,\pi_1,\pi_2}(\psi)\in { { \mathbb{f}}}[[t_1,t_2]]^\times,\ ] ] in which case it follows that @xmath1017 we then have @xmath1018(\pi_1)\\ & = & { { \mathrm{wt}}}\,\left(\iota_{k , w_1,w_2,\pi_1,\pi_2}\varphi\right)^*\\ & = & { { \mathrm{wt}}}\ , \iota_{k , w_2,w_1,\pi_2,\pi_1}(\varphi^\dagger)\\ & = & { { \mathrm{wt}}}_{k , w_1,\pi_1}\varphi^\dagger \end{array}\ ] ] at the first equality by ( [ equation : powerseriesinvariants1 ] ) , at the second by ( [ equation : numericalcriterion ] ) , at the third by ( [ equation : powerseriesinvariants2 ] ) , and at the fourth by ( [ equation : powerseriesinvariants0 ] ) .
[ proposition : grosskoblitzredux ] let @xmath115 , @xmath780 and @xmath969}$ ] be as in
[ subsection : embeddings ] .
let @xmath528 be a place of @xmath115 with residue field equal to @xmath780 and let @xmath970 be a uniformizer in @xmath971 .
let @xmath80 be the place of @xmath34 below @xmath528 and let @xmath1019 be a uniformizer in @xmath483 .
for every nonzero @xmath15 belonging to the fraction field of @xmath969}\cap{{\bf k}}$ ] , we have @xmath1020 where @xmath1021 $ ] , and @xmath1022 is the normalized additive valuation of @xmath115 associated to @xmath528 .
we are indebted to d. thakur for teaching us about this sort of convergence phenomenon .
( see , for example , ( * ? ? ? * thm .
4.8.1 ) . ) in general , @xmath15 is a quotient of elements of @xmath969}\cap{{\bf k}}$ ] .
but since we get the result if we can prove it for numerator and denominator of @xmath15 separately , we may assume without loss of generality that @xmath1023}\cap{{\bf k}}$ ] .
put @xmath1024 write @xmath1025 where @xmath957 $ ] is divisible neither by @xmath945 nor @xmath946 . by the weierstrass preparation theorem , write @xmath1026 where @xmath203 is a nonnegative integer , @xmath960 $ ] is a polynomial in @xmath945 of degree @xmath961 , and @xmath959^\times$ ] .
we have @xmath1027 by ( [ equation : powerseriesinvariants0 ] ) , ( [ equation : powerseriesinvariants1 ] ) and the definitions .
write @xmath1028 note that for @xmath916 we have @xmath1029 also we have @xmath1030 , and hence equality @xmath1031 @xmath1032 holds in the completion @xmath971 . the result follows .
we boil the preceding somewhat complicated considerations down to a few easy - to - apply rules .
[ [ section-56 ] ] let @xmath80 be a place of @xmath34 , let @xmath1035 be a place of @xmath754 above @xmath80 and let @xmath1036 be the unique additive valuation of @xmath754 belonging to the place @xmath1035 which extends the normalized additive valuation @xmath1037 of @xmath34 . for every nonzero @xmath15 in the fraction field of @xmath747 we define @xmath1038 where : * @xmath0 is any finite subextension of @xmath968 such that * * @xmath15 belongs to the fraction field of @xmath969}$ ] , * * @xmath528 is the place of @xmath115 below @xmath1035 , * * @xmath970 is a uniformizer at @xmath528 , and * * the constant field @xmath780 of @xmath115 equals the residue field of @xmath528 . we are obliged to check that @xmath1039 and @xmath1034 are well - defined . in any case , for every @xmath15 we can find suitable @xmath115 as above .
moreover , for fixed @xmath15 , the expression on the right side of the definition of @xmath1034 is easily verified to depend only on @xmath1035 , and the same is true of the expression on the right side of the definition of @xmath1040 .
finally , the latter depends only on @xmath80 by proposition [ proposition : grosskoblitzredux ] .
thus @xmath1039 and @xmath1034 are indeed well - defined . for convenience
put @xmath1041 .
then the function @xmath1039 is an additive valuation of the fraction field of @xmath747 .
the function @xmath1034 is a homomorphism from the multiplicative group of the fraction field of @xmath747 to @xmath1042 which on the subgroup @xmath1043 is refined to a homomorphism to @xmath1044 .
[ [ section-57 ] ] we claim that the following relations hold : @xmath1045 @xmath1046 @xmath1047 @xmath1048 here @xmath1035 ( resp . ,
@xmath1049 , @xmath1050 ) are places of @xmath754 above places @xmath80 ( resp . , @xmath1051 , @xmath1052 ) of @xmath34 , @xmath15 is a coleman unit , @xmath38 , and @xmath1053 is a uniformizer at @xmath80 .
relation ( [ equation : twistfunctoriality ] ) justifies ( [ equation : twisttransformationbis ] ) .
proposition [ proposition : grosskoblitzredux ] justifies relations ( [ equation : grosskoblitzreduxbis ] ) and ( [ equation : grosskoblitzreduxter ] ) .
proposition [ proposition : offdiagonal ] justifies ( [ equation : offdiagonal ] ) .
[ [ section-58 ] ] keeping the notation of the preceding paragraph , suppose further that @xmath809 $ ] for some @xmath807 .
( notice that not until now have we invoked conjecture [ theconjecture ] . )
we claim that @xmath1054 @xmath1055 @xmath1056 equations ( [ equation : weightintegral ] ) and ( [ equation : barv2 ] ) are proved by comparing ( [ equation : grosskoblitzreduxbis ] ) to the simplified version ( [ equation : adelicstirlingbis ] ) of the adelic stirling formula .
equation ( [ equation : daggerweightintegral ] ) follows from ( [ equation : weightintegral ] ) since , as remarked in
[ subsection : ccalculus ] , we have @xmath1057={{\mathcal{c}}}[\phi]^\dagger$ ] .
we continue in the setting of conjecture [ theconjecture ] .
let @xmath0 be a finite subextension of @xmath316 .
let @xmath1 be a finite set of places of @xmath34 .
assume the following : we are going to show that tate s formulation @xmath2 ( * ? ? ?
* , conj.2.2 ) of the stark conjecture is a consequence of conjecture [ theconjecture ] . after introducing suitable notation and making a convenient reduction , we recall the statement of @xmath2 in detail below .
we must of course remark that since we are working in the function field situation , @xmath2 is already a theorem due to deligne @xcite and ( independently ) to hayes @xcite .
the point of deriving @xmath2 from our conjecture is to establish that the latter does in fact refine the former .
* let @xmath1061 be the image of a fixed choice of uniformizer of @xmath483 under the map @xmath1062 .
* put @xmath1063 .
* fix a place @xmath1035 of @xmath754 above @xmath80 . *
let @xmath1036 be the unique additive valuation of @xmath754 belonging to @xmath1035 and extending the normalized additive valuation @xmath1037 of @xmath34 . according to tate ( * ? ? ?
3.5 , p. 92 ) , @xmath2 implies @xmath1064 for any subextension @xmath1065 of @xmath0 .
accordingly , after choosing a suitable finite subextension @xmath1066 of @xmath1067 and replacing @xmath1068 by @xmath1069 , we may assume without loss of generality that the data @xmath1068 satisfy the following further condition : for @xmath1076 or @xmath1077 , consider the euler product @xmath1078\;\;\;(\re(s)>1)\ ] ] extended over places @xmath80 of @xmath34 not in @xmath12 , where @xmath1079 is the geometric frobenius at @xmath80 ( cf .
* , prop .
if @xmath1076 , we drop the subscript and write simply @xmath1080 and @xmath1081 .
it is well - known that @xmath1082 continues meromorphically to the entire @xmath1083-plane with no singularity other than a pole at @xmath1084 .
note that we have @xmath1085 ( cf .
* , cor . 1.7 ) ) since @xmath50 splits completely in @xmath0 .
the _ stark unit _
@xmath1094 predicted to exist by conjecture @xmath1095{tate}}),\ ] ] given the choice @xmath1096 of a place of @xmath115 above @xmath50 , is uniquely determined up to a factor in @xmath987 by the formulas @xmath1097 or , equivalently , the formula @xmath1098 the rest of our work in [ section : starkrecovery ] is devoted to working out a `` recipe '' for the stark unit in terms of the transformation @xmath810 defined by conjecture [ theconjecture ] .
let @xmath1099 be the set of places of @xmath34 not belonging to @xmath1 .
let @xmath22 be the subgroup of @xmath43 generated by @xmath1100 .
let @xmath287 $ ] be the group ring of @xmath22 over the integers .
let @xmath1101 be the kernel of the ring homomorphism @xmath287\rightarrow { { \mathbb z}}[1/q]$ ] induced by the restriction of the idele norm function @xmath289 to @xmath22 .
since @xmath22 is a free abelian group with basis @xmath1102 , the ring @xmath287 $ ] may be viewed as the ring of laurent polynomials with integral coefficients in independent variables @xmath1103 indexed by @xmath1104 . from the latter point of view
it is clear that the set @xmath1105\ ] ] generates @xmath1101 as an ideal of @xmath287 $ ] . here ,
and in similar contexts below , the expression @xmath1106 is to be viewed as a formal @xmath54-linear combination of elements of @xmath22 .
put @xmath1107\in { { \mathcal{u}}}\vert { { \left\|a_\infty-1\right\|}}_\infty<1\}.\ ] ] let @xmath1108 be the unique subextension of @xmath316 such that @xmath1109 and put @xmath1110 crucially , the constant field @xmath780 of @xmath115 is also the constant field of @xmath1111
. put @xmath1112 extend the group homomorphism @xmath1113 to a ring homomorphism @xmath287\xrightarrow{\rho_{k / k } } { { \mathbb z}}[g]$ ] , and define @xmath287\xrightarrow{\rho_{k(\infty)/k}}{{\mathbb z}}[g(\infty)]$ ] analogously .
let @xmath1114 $ ] be the annihilator of @xmath987 .
for any subgroup @xmath290 let @xmath1115 $ ] be the ideal generated by differences of elements of @xmath292 . from the exactness of the sequence @xmath1116 which is well - known , we deduce an exact sequence @xmath1117\xrightarrow{\rho_{k / k}}{{\mathbb z}}[g]\rightarrow 0.\ ] ] analogously we have an exact sequence @xmath1118\xrightarrow{\rho_{k(\infty)/k}}{{\mathbb z}}[g(\infty)]\rightarrow 0.\ ] ] we claim that we have an exact sequence @xmath1119 exactness at @xmath1101 follows from lemma [ lemma : idealtheoryingrouprings ] .
the set @xmath1120 by ( * ? ? ? * , lemme 1.1 ) generates @xmath1121 as an ideal of @xmath1122 $ ] , whence exactness at @xmath1121 .
the claim is proved .
put @xmath1125\in { { \mathcal{o}}}\left|v\in s_0\rightarrow a_v\neq 0,\;\;\;\prod_{v\in s_0}i_v^{\times}(a_v)\in { { \mathcal{u}}}\right.\right\}\subset { { \mathcal{o}}},\ ] ] which is an open compact subset of @xmath37 .
note that @xmath1126 is stable under the action of the group @xmath1127 by multiplication .
extend the map @xmath1128 @xmath54-linearly to a homomorphism @xmath1129\rightarrow { { \mathrm{sch}}}_0({{\mathbb{a}}})\ ] ] of abelian groups , noting that @xmath1130 for all @xmath1131 $ ] and @xmath1132 .
we claim that for all @xmath1133 we have @xmath1134(i_\infty(t))d\mu_\infty^\times(t)=0.\ ] ] in any case , the set of @xmath1135 satisfying the equation above forms an ideal of @xmath287 $ ] , and so to prove the claim we may assume without loss of generality that @xmath1136 for some @xmath1104 , in which case already by the scaling rule ( [ equation : adelicrationalscaling ] ) the integrand above vanishes identically as a function of @xmath1137 .
the claim is proved .
so far all our constructions and definitions make sense unconditionally . but
from this point onward we must assume that conjecture [ theconjecture ] holds so that the transformation @xmath810 is defined .
from ( [ equation : weightintegral ] ) and ( [ equation : weightinfinity ] ) it follows that for every @xmath1135 we have @xmath1138=0\ ] ] and hence @xmath1139 \in ( k^{{\mathrm{ab}}}_{{\mathrm{perf}}})^\times\ ] ] is well - defined .
note that by ( [ equation : twisttransformationbis ] ) and ( [ equation : preliminaryabold ] ) the homomorphism @xmath1140 is @xmath22-equivariant , where we view the target in the natural way as a @xmath22-module via the reciprocity law @xmath1141 . for all @xmath1144 we have @xmath1145 by limit formula ( [ equation : grosskoblitzreduxter ] ) and the definitions .
moreover , the catalan symbol in question takes values in @xmath357 for all @xmath651 by proposition [ proposition : genericcatalanbehavior ] .
the result follows .
fix @xmath1147 arbitrarily and write @xmath1148 with @xmath867 , @xmath1149 and @xmath840 , noting that since @xmath1073 , this factorization is unique . for all @xmath1144 we have
@xmath1150&=&\epsilon_{\overline{\infty}}\,{{\mathcal{c}}}[{{\mathbf{1}}}_{{\mathcal{v}}}^{({{\mathbf{a}}}x\tau_\infty^n)}]\\ & = & \epsilon_{\overline{\infty}}\,{{\mathcal{c}}}[({{\mathbf{1}}}_{{\mathcal{v}}}^{({{\mathbf{a}}})})^{(x\tau_\infty^n)}]\\ & = & \epsilon_{\overline{\infty}}\,({{\mathcal{c}}}[{{\mathbf{1}}}_{{\mathcal{v}}}^{({{\mathbf{a}}})}]^{(\tau_\infty^n)})=\epsilon({{\mathbf{a}}})^{\rho(\tau_\infty^n)}=\epsilon({{\mathbf{a } } } ) \end{array}\ ] ] where the fourth equality is justified by the @xmath43-equivariance of @xmath810 noted in
[ subsection : ccalculus ] , the fifth by ( [ equation : twisttransformationbis ] ) , and the last by the preceding lemma .
therefore we have @xmath1151 .
now suppose that @xmath1152 .
then necessarily @xmath1153 and @xmath1154 , hence @xmath1155 , and hence @xmath1156=\epsilon_{\overline{\infty}}(1\otimes x^{\mu({{\mathcal{v}}})})=1,\ ] ] where the middle equality is justified by example ( [ equation : trivialsolitoncalculation ] )
. therefore we have @xmath1157 . by exactness of the sequence ( [ equation : curlyjexact0 ] ) and the previous two lemmas , @xmath323 takes values in @xmath984 .
we claim that @xmath1159 for all @xmath1135 , @xmath1132 and places @xmath80 of @xmath34 .
if @xmath1160 , then we have @xmath1161 for all @xmath546 and _ a fortiori _ ( [ equation : weightofspecialepsilon ] ) holds ; otherwise , if @xmath1162 , we get ( [ equation : weightofspecialepsilon ] ) by an evident manipulation of integrals .
the claim is proved .
it follows by formula ( [ equation : daggerweightintegral ] ) of [ subsection : weightepsiloncalculus ] that @xmath1163^\dagger=\left\{\begin{array}{rl } \displaystyle\int { { \mathbf{1}}}_{{\mathcal{v}}}^{({{\mathbf{a}}})}(i_v(t))d\mu_v^\times(t ) & \mbox{if $ s_0=\{v\}$}\\ 0&\mbox{if $ s_0\neq\{v\}$ } \end{array}\right.\ ] ] for all @xmath1135 , @xmath1132 and places @xmath80 of @xmath34 . in turn
it follows by formula ( [ equation : offdiagonal ] ) of [ subsection : weightepsiloncalculus ] and the @xmath22-equivariance of @xmath323 that @xmath1164 for all @xmath1135 .
finally , since by the preceding lemma and exactness of sequences ( [ equation : curlyjexact1 ] ) and ( [ equation : curlyjexact2 ] ) we may view @xmath323 as a @xmath1165-equivariant homomorphism @xmath1166 , in fact @xmath323 takes values in @xmath1167 by lemma [ lemma : abelianrootredux ] . for all @xmath1135 , by ( [ equation : barv2 ] ) and ( [ equation : weightofspecialepsilon ] ) in the case
@xmath1169 , we have @xmath1170 and so with @xmath1171 as above , the proof boils down to verifying the analytic identity @xmath1172 relating partial zeta values to the theta symbol .
since the set of @xmath1135 satisfying the identity in question forms an ideal of @xmath287 $ ] , we may assume without loss of generality that @xmath1173 for some @xmath1174 .
let @xmath1175 be any character , and extend @xmath1176 in @xmath117-linear fashion to a ring homomorphism @xmath1177\rightarrow{{\mathbb c}}$ ] .
it is enough to prove the @xmath1176-version of ( [ equation : thetarecipefortatetheta ] ) , i. e. , the relation obtained by applying @xmath1176 to both sides of ( [ equation : thetarecipefortatetheta ] ) .
now we adapt to the present case the method presented in tate s thesis for meromorphically continuing abelian @xmath382-functions . at least for @xmath1178 , when we have absolute convergence and can freely exchange limit processes , we have @xmath1179 the last integrand is constant on cosets of @xmath1180 and by proposition [ proposition : thetaasymptotics ]
has compact support in @xmath558 .
so the last integral defines an entire function of @xmath1083 .
now since @xmath1073 we have @xmath1181 .
so by plugging in @xmath1182 at beginning and end of the calculation above and breaking the last integral down as a sum of integrals over cosets of @xmath1180 in @xmath558 , we recover the @xmath1176-version of ( [ equation : thetarecipefortatetheta ] ) in the case @xmath1173 , as desired .
g. w. anderson , _ a two - dimensional analogue of stickelberger s theorem , _ in : the arithmetic of function fields ( columbus , oh , 1991 ) , 5173 , ohio state univ
. math .
publ . , 2 , de gruyter , berlin , 1992 . h. stark , _
@xmath382-functions at @xmath1084 .
first derivatives at @xmath1182 .
_ adv . in math .
* 35*(1980 ) , 197 - 235 .
j. tate , _ number - theoretic background _
, proc . of symp . in pure math .
* 33*(1979 ) , part 2 , pp . 3 - 26 .
j. tate , _ les conjectures de stark sur les fonctions @xmath382 dartin en @xmath1182 , _ prog . in math . * 47 * , birkhuser boston 1984 . | we propose a conjecture refining the stark conjecture st(@xmath0,@xmath1 ) in the function field case . of course st(@xmath0,@xmath1 ) in this case is a theorem due to deligne and independently to hayes .
the novel feature of our conjecture is that it involves _
two_- rather than one - variable algebraic functions over finite fields .
because the conjecture is formulated in the language and framework of tate s thesis we have powerful standard techniques at our disposal to study it .
we build a case for our conjecture by ( i ) proving a parallel result in the framework of adelic harmonic analysis which we dub the _ adelic stirling formula _ , ( ii ) proving the conjecture in the genus zero case and ( iii ) explaining in detail how to deduce st(@xmath0,@xmath1 ) from our conjecture . in the genus
zero case the class of two - variable algebraic functions thus arising includes all the _ solitons _ over a genus zero global field previously studied and applied by the author , collaborators and others .
ultimately the inspiration for this paper comes from striking examples given some years ago by r. coleman and d.thakur . |
`` _ follow me _ '' ( fm ) is probably the most natural strategy to solve the guidance problem , that is , to guide something or somebody along a given trajectory or towards a specific target , whether physically along geographical paths ( streets , roads , buildings ) , conceptually ( language learning , politics ) or even spiritually ( religion , social networks ) . the success of the fm strategy is based on the effect of the attraction that the guide ( the leader , the driver ) exerts on the guided ( followers , driven ) .
also successfull is the `` _ do as i do _ '' strategy , based on imitation ( of behavior ) or alignment ( of opinion ) , and used in the above mentioned fields , among others ( social learning ) .
less expected is the effectiveness of the opposite strategy , namely , `` _ move away from me _ ''
( ma ) , based on repulsive interactions .
that repulsion can serve to guide something or somebody is shown by nature itself , not only by means of gradient fields ( electromagnetic , temperature or chemical fields ) , where attraction towards high densities can be viewed as repulsion from low densities , but also by specifically repelling targets or agents . in neural development , axonogenesis takes place by combining attractive and repulsive guidance , so that the axon growth follows guidance cues presented by chemoattractant and chemorepellent molecules located in the environment of the cell @xcite . in animal herding ,
sheepdogs are used to guide sheep flocks through a repulsive force that dogs exert on sheep @xcite .
guidance cues can also be magnetic , as in drug targeting @xcite , thermotactic , as in sperm guidance @xcite , cognitive , as in crowd motion and traffic flow @xcite or in opinion formation @xcite , acoustic signals , as in animal alarm calls or instrumental conditioning @xcite , food trail pheromones in ants @xcite ( chemical at a scale larger than the cell ) , etc .
attractive and alignment guidance problems are being studied for a long time by means of agent based models @xcite , with attention to guidance by leadership @xcite , and optimal strategies to minimize time guidance ( optimal evacuation times ) or distance travelled by the agents have been found for several systems @xcite . very recently , in 2015 ,
the optimal strategy for a flocking model to reach a target point or to follow a given trajectory through attractive and alignment guidance has been presented @xcite . in this model ,
individuals from the flock interact through attractive - repulsive and alignment forces with the rest of individuals .
interactions are symmetric , except for one specific individual the leader which exerts on each other individual an extra attractive force .
the result is that the leader is followed by the flock , so that , by controlling the behavior of the leader , a fm strategy can be used to make the flock reach a given target or move along a given trajectory .
repulsive forces in attractive - repulsive models have mostly been considered for collision avoidance with obstacles or with other agents .
guidance by repulsion has received much less attention , which is reduced , to our knowledge , to the above mentioned axonogenesis and animal herding . also very recently , in 2015 , an agent based model has been introduced to describe a so - called defender - intruder interaction , where repulsion is used by a defender to expel an intruder away as far as possible from a protected target @xcite .
the authors in @xcite find an optimal ma strategy , which not always consists in approaching the intruder as close as possible , but to simply drive the intruder away beyond a short distance of security .
repulsion in the intruder - defender interaction is qualitatively different than in collision avoidance or in interception @xcite , where the attractive and/or alignment forces determine ( most of ) the behavior .
defender - intruder problems fall into the category of `` conflicting interactions '' @xcite , which are well described by the classical pursuit - evasion ( pe ) framework @xcite .
the simplest scenario for a pe interaction consists of a single pursuer that follows and tries to capture a single evader that tries to escape to infinite from the pursuer @xcite .
although our interest does not focus on conflicting interactions , we adopt here the pe framework .
the guidance by repulsion can indeed be described with the simple two - agents pe framework , provided two considerations are taken into account : first , the guide is not exactly a pursuer , as it often separates from the direction towards the evader and the guide can not be arbitrarily close to the evader , and second , the agent to be guided is not exactly an evader , as it does nt try necessarily to escape to infinite but simply moves away a short distance from the repelling guide .
we present here a guidance by repulsion model based on the two - agents pe framework .
we will refer to the guiding agent as _ the driver _ , which tries to drive the _
evader_. the driver thus follows the evader but can not be arbitrarily close to it .
this is especially interesting if the driver can not approach the evader or contact between agents should be avoided ( because of chemical reactions , animal conflict , etc ) .
the evader moves away from the driver but does nt try to escape beyond a not so large distance .
the driver is of course faster than the evader . at a critical short distance
, the driver can display a circumvention maneuver around the evader that forces the evader to change the direction of its motion .
thus , by adjusting the onset and offset of the circumvention maneuver , the evader can be driven towards a desired target or along a given trajectory .
our goal is to find optimal strategies to drive the evader in the most efficient way .
we use an inertial model where interactions between agents take place through asymmetric newtonian forces .
the asymmetry consists in that one agent is attracted and repulsed , while the other is simply repulsed .
this kind of interaction has been coined `` anti - newtonian '' by sprott @xcite and others @xcite .
thus , velocities are not constant ( they depend on the state of the system ) , and no alignment forces are considered . we denote by @xmath0 and @xmath1 ( and indexes @xmath2 and @xmath3 ) the driver and the evader agents respectively .
both agents obeys the newton s second law , that is , @xmath4 and @xmath5 , where @xmath6 , @xmath7 , @xmath8 and @xmath9 denote the position vector , the velocity vector and the mass of agent @xmath10 , and the resultant force to which agent @xmath10 is subject , respectively , for @xmath11 .
the force acting on the evader @xmath12 has only one component , which is in the direction of escape from the pursuer , @xmath13 . the force acting on the pursuer @xmath14 has a component collinear to @xmath12 and a lateral perpendicular component @xmath15 which allows the pursuer to surround the evader , therefore forcing the evader to change the direction of escape . here
@xmath16 .
the perpendicular component of the force acting on the pursuer @xmath14 can be switched on and off by means of the control parameter @xmath17 which takes values on @xmath18 .
the control parameter @xmath17 is the key ingredient of the model , as it determines the behavior of the pursuer , which in turn determines the behavior of the evader .
the resulting dynamical system can be considered as a driver - evader system with two operating modes controlled by a single parameter .
[ [ contents - of - the - paper - and - sketch - of - the - results ] ] contents of the paper and sketch of the results + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + section titles are self - explanatory . in sec . 2
, we introduce the model equations and parameters , describing in detail the interactions between agents and with the environment .
we show that the driver - evader system can be viewed as having two operating modes controlled by a single parameter , so that the system can be moved from one state to another in order to make the driver to guide the evader from any point to any other point .
we study the controllability of the system , and prove that , although the system is not fully controllable , agents positions remain asymptotically close to each other ( _ i.e. _ , an agent can not go to infinity ) . in sec . 3
, we consider what are the optimal strategies which allow the driver to guide the evader to a desired target .
activating a system has a cost , as well as keeping it in the active mode . as an illustrating example
, the driver can be viewed as a spacecraft with two lateral propellers whose ignition process and fuel consumption are very high with respect to the consumption of the back propeller .
our interest is in reducing the cost by minimizing 1 ) the number of activation and 2 ) the time of use of the lateral propellers . to do that
, we formulate an optimal control problem @xmath19 for a cost functional including these two costs .
whe then find , in sec .
3.1 , the ( unique ) optimal strategy allowing the reach the target in an initially active system ( @xmath20 , and the ( unique ) optimal strategy which reduces the length of the time interval @xmath21 $ ] during which the control parameter has a non - zero value .
this second strategy also reduces the number of manipulations of the control to two ( one _ switch on _ and one _ switch off _ ) . in this minimization process ,
the cost of back propellers ( which are always switched on , the driver is a self - propelled agent which is always attracted by the evader ) is neglected compared to the cost of using the lateral propellers , so that there is not ( too much ) concern by long trajectories or long execution times where lateral propellers are switched off .
these open - loop controls are subject to the reproducibility of the initial and environmental conditions and to uncertainty about the model @xcite .
in fact , we show here in detail that the system is highly sensitive to small variations of the values of the control , thus suggesting the appropriateness of the use of closed - loop controls that can afford for the random perturbations arising in real systems .
we then present in sec .
3.2 a feedback control law allowing to drive the evader to any desired target with an arbitrary accuracy .
this feedback law is especially advantageous as it yields a similar cost than the open - loop controls , in real conditions ( _ i.e. _ , under perturbations ) .
moreover , the feedback control law provides an excellent insight for the search of a control function yielding a substantially lower cost ; as an example , we report numerical simulations of a case in which the resulting cost is reduced almost a 60% of the cost provided by the open - loop controls . finally , in sec . 4
, we present our conclusions and we comment on future directions . in the last section
the system consists of 8 ordinary differential equations ( odes ) and 8 initial conditions for the components @xmath22 , @xmath23 , @xmath24 and @xmath25 , for @xmath11 . expanding the resultant force acting on the @xmath10th - agent @xmath26
, the system can be written as the following 4 vectorial odes with 4 initial conditions : @xmath27 , \label{eq - vp } \\
\dot{\vec{v}}_e(t ) & = { 1 \over m_e } \left [ c^d_e { \vec{u}_e(t ) - \vec{u}_d(t ) \over \|\vec{u}_e(t)-\vec{u}_d(t)\|^2 } - \nu_e \vec{v}_e(t ) \right ] , \label{eq - vs }
\\ \vec{u}_d(t_0 ) & = \vec{u}_d^0 , \quad \vec{u}_e(t_0 ) = \vec{u}_e^0 , \quad \vec{v}_d(t_0 ) = 0 \quad \mbox{and } \quad \vec{v}_e(t_0 ) = 0 . \label{condinis}\end{aligned}\ ] ]
interactions between agents are as follows .
the expression between brackets in eq .
( [ eq - vp ] ) consists of two terms with respective coefficients @xmath28 and @xmath29 . the first term corresponds to the long - range attraction and short - range repulsion force that the evader exerts on the driver . here
@xmath30 is the distance at which the attraction balances the repulsion : if @xmath31 , then the evader attracts the driver ( and therefore the driver accelerates towards the evader ) , while if @xmath32 , then the driver is repulsed by the evader ( and the driver decelerates in the direction opposed to the evader ) .
the term with coefficient @xmath29 corresponds to the circumvention force , which as a component collinear to the attraction - repulsion interaction with the evader , given by @xmath13 , and a component perpendicular to this direction , denoted by @xmath15 , where @xmath16 .
the circumvention force is thus a force that attracts the driver towards one of the two sides of the evader , where the side is determined by the sign of the parameter @xmath17 .
the model considers two other critical distances , @xmath33 , which is the ( short ) distance at which the intensity of the circumvention force is effective , and @xmath34 , which denotes the distance between agents left by the driver during the circumvention maneuver . in
( [ eq - vs ] ) , the evader is simply subject to the repulsion from the driver , which has the same expression than the force that the evader exerts on the driver but with a different ( and smaller ) coefficient @xmath35 .
finally , both agents are subject to friction forces with the ground , which have the same form , in the opposite direction of the motion : @xmath36 , for @xmath11 .
the model can be formulated in different ways . the formulation presented in ( [ eq - up])([condinis ] )
is based on the general expressions of the attraction - repulsion forces introduced by gazi & passino @xcite , later widely used in realistic ( biological ) models @xcite .
agents are prevented from collisions by means of strong short - range repulsive forces ( large exponents of @xmath37 in the denominator ) , so that , although it is theoretically possible , in practice , the denominators are never ( too close to ) zero , and the model is in this sense well - posed .
another advantage of this formulation is that each ingredient of the model , especially the circumvention behavior , appears explicitly in the equations , thus facilitating the realistic interpretation of the contribution of each component of the model .
for example , the exponent @xmath38 of @xmath39 in the circumvention force ( which is large compared to the exponent @xmath40 in the attracting force ) shows that the circumvention maneuver will only be effective when both agents are sufficiently close to each other . with this formulation ,
the control parameter @xmath17 appears as a factor of the perpendicular component of the force acting on the evader , and , by the simple choice of a value in @xmath18 , practically determines the behavior of the system .
let us define the instantaneous distance between agents @xmath41 . then , omitting time - dependence to lighten notation , eqs .
( [ eq - vp])([eq - vs ] ) can be rewritten as @xmath42 + \kappa { \delta_1 ^ 4 \delta_2 \over r^3 } { c_{\rm r } \over m_d } \ , { ( \vec{u}_d - \vec{u}_e)^\bot \over r^2 } - { \nu_d \over m_d } \vec{v}_d , \label{eq - vp - d } \\ \dot{\vec{v}}_e & = - { c^d_e \over m_e } { \vec{u}_d - \vec{u}_e \over r^2 } - { \nu_e \over m_e } \vec{v}_e .
\label{eq - ve - d}\end{aligned}\ ] ] we have solved the system ( [ eq - up])-([condinis ] ) numerically with several classical methods ( runge - kutta , adaptive or not , crank - nicolson ) , finding that a simple explicit euler method with time - step of order @xmath43 is sufficiently effective and provides a sufficiently accurate solution for our purposes .
our results are based on exhaustive numerical simulations for a wide range of the parameters preserving the significance of the model .
the behavior of the model is stable and coherent under reasonable variations of the parameters .
the results we describe here are not dataset dependent .
the values we have chosen are those providing the more illustrative figures .
( [ eq - vp - d ] ) shows that a necessary condition to have an effective short - range repulsion acting on the driver is therefore that @xmath44 is negative , as it is the case for the dataset considered in our study .
similarly , in order to have a faster and more reactive behavior of the driver , we assume that @xmath45 and that @xmath46 and @xmath47 are sufficiently smaller than @xmath48 and @xmath49 respectively . for the numerical simulations and graphical descriptions presented here , we have considered the following dataset for the typical values of the constant parameters of the model : first
, @xmath50 and @xmath51 are the mass and the friction coefficients of agent @xmath10 respectively , with @xmath52 , @xmath53 , @xmath54 and @xmath55 ; then , @xmath56 is the coefficient of the attractive - repulsive force that the evader exerts on the driver , @xmath57 is the coefficient of the repulsive force that the driver exerts on the evader , and @xmath58 is the coefficient of the circumvention component of the force exerted on the driver when @xmath59 . finally , the critical distances are as follows : @xmath60 , @xmath61 and @xmath62 .
parameter values are dimensionless ( of order one ) and chosen so that their relative value allow the system to reproduce the behavior of a realistic driving phenomenon .
the numerical simulations allow us to establish that the system has two operating modes , depending on the value of @xmath63 : * _ pursuit mode _ ( @xmath64 ) : the driver @xmath0 pursues the evader @xmath1 , which moves away from @xmath0 .
both agents tend to move in the same direction , given by their acceleration vectors , along the straight line @xmath65 . if @xmath66 continuously during a sufficiently long interval of time , both agents velocities converge asymptotically to the same vector of norm @xmath67 ( due to the friction ) , and the driver stays at a constant distance from the evader , @xmath68 .
see appendix a for an analytical estimate of these values , and the left panels in fig .
[ modes ] : the lower panel shows that , before reaching the constant value @xmath69 , the pursuer is faster than the evader . for the values we have used , @xmath70 * _ circumvention mode _ ( @xmath71 ) : the driver @xmath0 separates from the straight line @xmath65 and starts a circumvention maneuver around the evader @xmath1 ( clockwise or counterclockwise , depending on the sign of @xmath63 ) .
the response of the evader is to move away from the driver . if @xmath72 continuously during a sufficiently long interval of time , as the driver is faster than the evader , the asymptotic behavior of the system is a circular motion of the evader around a fixed point and a circular motion of the driver around the circle described by the evader .
see fig .
[ modes ] , where the right panels show that , for @xmath73 $ ] , the system tends to a periodic configuration where both agents are separated by a constant distance @xmath74 ( not shown in the figure ) and have the same angular velocity of norm @xmath75 ( the period @xmath76 of the oscillations is @xmath77 ; bottom - right panel of fig .
[ modes ] shows that @xmath78 ) .
= 0.95 when the driver is sufficiently close to the evader , the circumvention mode is effective and triggers the circular behavior of the agents .
see the right panels of fig .
[ modes ] , where the driver initial position at @xmath79 is so far from the initial position of the evader @xmath80 that , although the control is set to @xmath81 from @xmath82 to @xmath83 , the first part of the trajectory is almost a straight line . until @xmath84
, both agents are almost still in the horizontal axis : @xmath85 , @xmath86 . at @xmath87 , @xmath88 and @xmath89 ,
the driver is close enough to the evader and the circular behavior becomes perceptible .
see the oscillations of @xmath90 and @xmath91 , of period @xmath92 , in the right - bottom panel of fig .
[ modes ] .
[ remark1 ] * 1 : * when the driver is far from the evader ( _ i.e. _ , @xmath93 ) , the term between brackets in eq .
( [ eq - vp - d ] ) is such that @xmath94 , so the first term of this equation ( which has coefficient @xmath28 ) is of order @xmath95 , while the term with @xmath17 is of order @xmath96 .
then , eq . ( [ eq - vp - d ] ) can be reduced to @xmath97 [ remark2 ] * 2 : * in particular , it can be observed that , when @xmath98 , then @xmath99 so that the term deviating the driver from the pure pursuit trajectory is negligible with respect to the term corresponding to the attracting force exerted by the evader .
therefore , when the driver is sufficiently far from the evader , the value of @xmath17 has no influence on the behavior of the driver , meaning that @xmath63 can be set to zero .
the circumvention mode can be viewed as the _ active _ state of the system , where the control @xmath63 is set to on , while the pursuit mode is the _ rest _
state of the system , where @xmath63 is set to off . with the appropriate combination of both modes ,
the driver is able to make the evader reach any given target point or move along any ( relatively smooth ) given trajectory .
the resulting behavior of such combination is what we call a _ driving _ behavior .
see fig . [ tray ] .
= 0.95 this strongly suggests that the system can be controlled with a single control parameter , @xmath17 . in control
theory , a system is said to be _ fully controllable _ when , starting from any arbitrary initial state , every possible state of the system can be reached by appropriately adjusting the control parameters ( see , _
e.g. _ , @xcite ) . in this sense
, our numerical simulations have shown that agents can not be separated an arbitrarily long distance , so that the driver - evader system can not be labelled as fully controllable .
in fact , we prove analytically in appendix a that , for any initial state , the separation between agents is bounded by above and tends to a distance ( of order 1 ) , showing that the evader can not escape from the driver to infinity , and the driver can not move away from the evader to infinity .
moreover , numerical simulations show that agents velocities are also bounded and that , when @xmath17 remains unchanged for sufficiently long time intervals , both velocities tend asymptotically to the same constant value ( @xmath67 or @xmath100 ) . finally , agents are prevented from occupying the same place at the same time ( _ i.e. _ , @xmath101 ) by the strong short - range repulsion force that the driver exerts on the evader , as our numerical simulations confirm .
we will say instead that the driver - evader system is _ partially controllable _ , in the sense that each agent can be controlled separately : the driver can force the evader to reach any point in the plane ( this is shown in the successive sections ) , and vice - versa , a series of targets for the evader can be selected so that the driver is driven to reach any point in the plane .
the question arises now as how the driver - evader system behavior can be optimized to minimize a given cost functional accounting for the use of the lateral propellers .
denote by @xmath102 a ball of radius @xmath103 centered in @xmath104 , assume that the evader is initially far from the driver ( _ e.g. _ , @xmath105 ) , and assume also that @xmath0 , @xmath1 and the ball @xmath102 are not aligned , that is , @xmath106 .
consider now the objective of driving the evader @xmath1 into the target ball @xmath102 at a final time @xmath107 by controlling the driver @xmath0 with an appropriate strategy @xmath17 .
as @xmath0 , @xmath1 and @xmath102 are not aligned , the circumvention mode has to be activated and/or has to remain active a suitable time to modify the trajectory of the evader and guide it towards the target ball @xmath102 .
such an objective must appraise the cost of 1 ) forcing the system to leave its resting state , and/or 2 ) keeping the system in an active state .
this cost is given by the functional @xmath108 where @xmath109 is the number of times that the system is forced to leave its resting state in the time interval @xmath110 $ ] ( that is , the number of ignition processes , where @xmath17 changes from 0 to @xmath111 ) , and @xmath112 is the total time during which the system is active in @xmath110 $ ] ( _ i.e. _ , the time spent with a lateral propeller in active mode , where @xmath17 has a non zero value ) , @xmath113 and @xmath114 are nonnegative weights fixed to balance the contribution of each partial cost .
an optimal control problem @xmath19 can then be formulated as @xmath115 \to \{-1,0,1\ } \mbox { such that } \vec{u}_e(t_f ) \in b_\rho(t ) \big\ } \end{array } , \right.\end{aligned}\ ] ] where @xmath116 is the set of admissible controls . here
we solve @xmath19 for @xmath117 and @xmath118 in sections 3.1 and 3.2 respectively , finding the corresponding optimal open - loop controls .
our results show that the system is highly sensitive to small variations of the conditions of the problem , so that a general open - loop control for @xmath119 would not be of practical interest .
instead , we provide in section 3.3 a feedback control law for @xmath119 that substantially reduces the cost of the open - loop controls for @xmath120 , and preserves the number of ignitions at a relatively low value ( @xmath121 ; see later )
. solving @xmath19 for @xmath117 consists in minimizing @xmath122 for @xmath123 , with @xmath124 . indeed , if @xmath125 , then @xmath126 , because @xmath0 and @xmath1 are initially not aligned with the target ball and no inition process can be used to modify the trajectory of the evader , so @xmath127 for all @xmath128 .
similarly , solving @xmath19 for @xmath118 consists in minimizing @xmath122 for @xmath17 of the form @xmath129 in a time interval @xmath130 $ ] and @xmath131 outside .
we thus seek two control function profiles like those sketched in fig .
[ shapes ] .
= 0.95 [ shapes ] the following result holds : let @xmath133 : @xmath134 be the following step - function in the time interval @xmath110 $ ] , @xmath135 where @xmath124 is the initial value at time @xmath136 : @xmath137 .
then , if @xmath107 is sufficiently large , there exists an interval @xmath138 \subset ( t_0,t_f)$ ] such that , for all @xmath139 $ ] , there exists a time @xmath140 for which the evader is in the interior of the ball of radius @xmath141 centered in the target @xmath104 .
that is : @xmath142 , \ ; \exists \ ; t \in ( t_0,t_f ) \mbox { such that } \| \vec{u}_e(t ) -
\vec{u}_t \| < r.\end{aligned}\ ] ] moreover , if @xmath143 , then the interval @xmath138 $ ] shrinks to a single point @xmath144 such that there exists a time @xmath140 for which @xmath145 .
see @xmath144 in fig .
[ shapes](a ) . the proof is based on a continuity argument applied to a ( numerical ) shooting method .
[ shoot-1change ] shows the trajectories of the evader for different values of @xmath146 . by a continuity argument , it is possible to find the values of @xmath147 , @xmath148 and @xmath144 with a simple shooting method based on comparing the direction of the velocity vector of the evader @xmath149 with respect to the direction towards the target when the control is switched off .
see appendix b for a more detailed description of the decision test of the shooting method .
= 0.85 in this case , the cost of a strategy @xmath133 given by ( [ estecost ] ) is @xmath150 for all @xmath139 $ ] , so that the optimal strategy is @xmath151 . for a sufficiently small value of @xmath141 ( we used @xmath152 in fig . [ shoot-1change ] ) , the interval @xmath138 $ ] colapses to the value @xmath144 and the cost of the optimal control is @xmath153 .
noticeably , fig . [ shoot-1change ] reveals that the system is highly sensitive to small variations of @xmath146 around the optimal value @xmath144 : the same variation of @xmath146 produces a larger variation of the deviation of @xmath149 with respect to a reference line ( _ e.g. _ , the horizontal line ) when @xmath146 is close to @xmath144 than when @xmath146 is far from @xmath144 : for @xmath154 , the angular variation @xmath155 from @xmath156 to 42 is larger than @xmath157 ( actually , @xmath158 ) , and is 10 times smaller from @xmath159 to 36 ( @xmath160 ) . in the previous case ,
the optimal control is set to 1 in the whole interval @xmath162 .
the possibility of using one ignition process allows us to consider controls where the system is at rest while the driver is approaching the evader , taking advantage of the fact that when the driver is far from the evader the circumvention term has ( practically ) no influence on the trajectory of the driver ( and therefore , on the behavior of the system ) ; see remark 2 .
this delay in the ignition process substantially reduces the cost by allowing the use of a control function profile like the one depicted in fig .
[ shapes](b ) , defined by @xmath163 where @xmath164 and @xmath165 are the instants of time in which the control is switched on ( @xmath166 ) and off ( @xmath64 ) respectively . for these control functions , the cost is given by @xmath167 so that the @xmath19 problem is reduced to find the values of @xmath164 and @xmath165 minimizing @xmath168 .
these values are found by means of two successive ( numerical ) shooting methods .
let us consider the extreme case where @xmath169 ( numerically we used @xmath170 ) , so that the solution of the shooting method described in the previous section for @xmath117 is unique .
the first shooting method consists in finding , for each value of @xmath164 , the ( unique ) value @xmath171 for which there exists a time @xmath140 such that @xmath145 .
this shooting method is the one presented in the previous case where @xmath117 .
in fact , in that case , @xmath172 . this procedure allows us to build a function @xmath173 in such a way that the function @xmath174 is _ admissible _ : @xmath175 .
then , the second shooting method consists in finding the value of @xmath164 for which the characteristic function of the interval @xmath176 $ ] minimizes @xmath177 .
= 0.92 fig .
[ shoot - double ] shows the cost function @xmath178 as a function of @xmath164 : @xmath179 .
when @xmath107 is sufficiently large , the cost @xmath180 tends to a constant value @xmath181 which is the time is takes to the driver to make the evader turn back towards the target . see fig .
[ turnbacks ] , where two examples with different ignition times @xmath164 yield ( approximately ) the same cost @xmath182 .
this value @xmath182 constitutes a substantial reduction of the optimal cost found for @xmath117 .
in fact , fig . [ shoot - double ] shows that there is a wide range of values of @xmath164 yielding a similar value of the cost , so that , for a tolerance larger than @xmath183 , the optimal control would not be unique .
this would allow the system to accept another criteria or another restriction to determine the optimal strategy , which can be for instance to minimize the total time or the total distance travelled by the agents ( and which would be equivalent to consider the cost of back - propellers ) .
= 0.95 however , if the variations of @xmath182 of order @xmath184 are relevant , a closer analisys shows that the first linear decreasing range in fig .
[ shoot - double ] , which lasts until @xmath185 , is due to the fact that during this interval of time @xmath186 $ ] the driver is far from the evader , that is , @xmath187 , so that the term with @xmath63 in eq
. ( [ eq - vp ] ) can be neglected ( it is of order @xmath96 ) with respect to the terms corresponding to the attracting force ( of order @xmath188 ) .
this means that switching on the control before @xmath189 does nt contribute effectively to reach the target and is a waste of resources .
in fact , the optimal value of the cost is reached at @xmath189 : @xmath190 .
when @xmath107 is so short that the turning back maneuver can not be completed , then @xmath191 , so that the cost function @xmath180 decreases ( again linearly with @xmath164 ) to zero ( but of course the target is not reached ) .
note that , for all @xmath164 , the value @xmath171 will exist provided @xmath107 is sufficiently large .
interestingly , @xmath180 exhibits an abrupt nonlinearity of small amplitude at @xmath189 ; see the right panel of fig .
[ shoot - double ] .
in a pure pursuit regime , the velocity of the agents both converge to a constant value @xmath67 due to the friction with the ground .
see the lower left panel in fig .
[ modes ] . the nonlinearity is located precisely at the time where the evader reaches this constant velocity ( @xmath69 ) .
the curve of @xmath180 reaches a ( local ) maximum at a value of @xmath164 slightly larger than @xmath189 , due to that the driver is close to the evader and both agents velocities have converged to @xmath67 . for larger values of @xmath164 in the horizontal plateau of fig .
[ shoot - double ] , the turning back maneuver is practically identical in space and time ; see fig .
[ turnbacks ] , especially the right panel , where we show that two different values of @xmath164 in the plateau yield two turning back maneuvers that practically overlap .
the slight decrease of the plateau shown in the right panel of fig .
[ shoot - double ] is due to the fact that turning back maneuvers are less consuming the farther from the target they takes place , because the angle at which the control is switched off is smaller .
this situation is reversed ( that is , the plateau increases ) when the target point is above the turning back region _e.g .
_ , at @xmath192 .
although the relative amplitude of the nonlinearity of @xmath180 is very small ( 0.024 with respect to 3.52 ) , it unexpectedly adds an important complexity to the study of the system : the minimum located at @xmath189 can be global or local depending on @xmath107 , therefore complicating the search for optimal directions of descent in numerical minimization methods . moreover , the high sensitivity of the system detected in the previous section is also in action if control functions like @xmath174 are used when small variations of @xmath193 with respect to @xmath164 can occur .
this section shows that , despite a quite predictive general behavior of the system ( no signs of chaos have been detected ) , its high sensitivity strongly suggests the use of closed - loop controls . for systems that are subject to conditions of high sensitivity like those described in section 3.1 ,
closed - loop or feedback controls offer the possibility of correcting instantaneously the state of the system for deviations from the desired behavior @xcite .
moreover , the explicit form of the control as a function of time need not to be known _ a priori _ in the whole time interval @xmath110 $ ] . in turn , feedback laws have to pay the cost of continuously monitoring the position and velocity of the agents .
we present here a feedback control law based on the following observations : 1 . in real situations , the orientation of the vector @xmath194 used in the shooting method can be difficult to observe with the accuracy required by the high sensitivity of the system . + instead , the alignment @xmath195 of the driver @xmath0 and the evader @xmath1 with the target point @xmath104 is easier to observe and is a good approximation of @xmath194 .
when the driver is sufficiently far from the evader , @xmath17 can be set to zero ( remarks 1 & 2 ) .
the instantaneous information about the state of the system is processed in real time to determine the distance separating both agents @xmath196 and the alignment @xmath195 .
the alignment @xmath195 can be charaterized by the following scalar product ( time dependence is omitted to lighten notation ) : @xmath197 the sign of @xmath195 reveals in which half - plane the target @xmath104 is with respect to the line @xmath65 , and can be used to determine the sign of @xmath17 .
moreover , @xmath198 is an instantaneous measure of how urgently the control must be set to on .
let us consider a maximal tolerance of deviation @xmath199 .
the feedback control law is based on the idea that when @xmath198 is smaller than @xmath199 , it is possible to consider that @xmath104 is practically on the line @xmath65 , so that @xmath17 can be set to off , thus saving cost , and when @xmath200 , the deviation is excessive and the control must be set to on .
the tolerance of deviation @xmath199 is an effective bound for both the angle and the intensities of the velocities ( @xmath201 ) , so @xmath202 restricts also the velocities of the agents : a slightly deviated evader at a high speed can miss the target as well as a largely deviated evader at a lower speed .
note also that when @xmath203 , the control can be switched off provided the evader and the target are at the same side with respect to the driver , in order to prevent the driver from driving the evader away from the target ; that is , @xmath17 can be set to zero only if the scalar product @xmath204 is negative .
finally , remark 2 is introduced into the feedback law by means of the characteristic function @xmath205 which serves to switch off the control when the driver is far ( @xmath206 ) from the evader .
the feedback control law can then be written as follows : @xmath207 we have solved the system ( [ eq - up])([condinis ] ) numerically using the condition @xmath208 to have @xmath209 in expression ( [ fidbaclaw ] ) .
we have considered an alignment tolerance @xmath210 .
the rest of values are as in previous sections .
let us refer to this case as case ( a ) .
the result is that the feedback law reduces substantially the cost obtained with the open - loop control in sec . 3.1 : @xmath211 , so an improvement of 60% with respect to @xmath212 , with however a slight increase of the number of ignition processes , from @xmath213 to @xmath214 .
see figs .
[ feedbk-1 ] , [ feedbk-2 - 3 ] and [ feedbacks](a1),(a2 ) .
= 0.98 figs .
[ feedbk-1 ] shows the trajectories of the agents ( the whole trajectories in the left panel , and a zoom of @xmath215 $ ] in the right panel ) .
surprisingly , the trajectory of the driver exhibits an oscillatory behavior around the circular trajectory of the evader , allowing the driver to remain closer to the evader than in the previous cases ( see , _
e.g. _ , fig .
[ turnbacks ] ) . moreover , the time spent with the control in active mode ( @xmath216 ) is surprisingly short compared to the time spent in this state in the open - loop control @xmath217 .
= 0.95 the improvement consists in that the circumvention mode is interrupted ( @xmath218 ) during the surrounding motion of the driver around the evader , reducing the time spent with @xmath216 to four small intervals @xmath219 , @xmath220 , of total length @xmath211 , and , accordingly , @xmath221 .
see the wide solid segments in the trajectories of the agents in fig .
[ feedbk-1 ] and figs [ feedbacks](a1 ) and ( a2 ) , which show the resulting control function @xmath222 for @xmath223 $ ] , with @xmath82 and @xmath224 .
the solution found in fig .
[ feedbk-1 ] with the feedback law ( [ fidbaclaw ] ) can indeed be considered a good solution of the control problem . however , this is not a general situation , as shown by the wide range of cases analysed in our numerical simulations , because of the high sensitivity exhibited by the system .
let us illustrate this observation here by showing the results for two slightly different external conditions ; see the cases ( b ) and ( c ) in figs .
[ feedbacks ] and [ feedbk-2 - 3 ] .
\(b ) ( c ) = 0.9 case ( b ) and ( c ) use an alignment tolerance @xmath225 .
[ feedbk-2 - 3 ] shows that the evader follows almost the same trajectory than in case ( a ) ( depicted in the figure to facilitate the comparison ) and reaches the target with a more accurate orientation .
however , such a small deviation requires an enormous increase of the use of the lateral propellers , as shown in fig .
[ feedbacks](b2 ) .
not only the cost of having the control set to 1 is larger , @xmath226 ( and larger than with the open - loop controls ) , but a much greater number of ignition processes is required ( @xmath227 ) , moreover involving both the right and the left propellers alternatively ( _ i.e. _ , @xmath228 ) .
on the other hand , case ( c ) shows that small variations of the switching times of the feedback control can produce huge qualitative differences in the behavior of the agents .
we have introduced a small perturbation of the first interval of activation @xmath229 $ ] , by keeping @xmath230 until 39.6 instead of until 39.55 ( a perturbation of relative size @xmath231 ) . for @xmath232 , we use again the feedback law ( [ fidbaclaw ] ) , so that a different control function profile arises . the resulting behavior of the agents is depicted in fig .
[ feedbk-2 - 3](c ) .
[ feedbk-2 - 3](c ) shows that after the perturbation , the agents describe a widely deviated trajectory , especially in the case of the driver , with respect to the one described with the unperturbed feedback law shown in fig .
[ feedbk-1 ] .
accordingly , the perturbed control function profile , depicted in fig . [ feedbacks](c ) , is significantly different from the unperturbed feedback law .
the driver requires more time to reach the target ( @xmath233 ) than in cases ( a ) and ( b ) .
compared with case ( b ) , the cost is higher , although not excessively , @xmath234 ( with no additional cost for the extra time because the evader moves in the right direction ) , and the number of ignition process is much smaller : @xmath235 .
perturbations of the rest of switching times produce similar results , that is , qualitative deviations of the trajectories and a larger number of ignition processes , with respect to the unperturbed case ( a ) , meaning that , in cases where delays can appear in the instants of time in which the control has to be manipulated , important changes of the behavior of the system may arise .
the oscillatory behavior arisen in case ( c ) can appear to be less convenient but , according to our numerical simulations , it is not necessarily worse than the smooth evader - following trajectory ; oscillations can contribute to reduce the number of ignition processes , with a low increase of the cost ; see left panels ( a ) in fig .
[ multitarget ] .
= 0.95 the feedback control law is able to drive the evader along a given trajectory , provided the trajectory is sufficiently smooth , that is , the trajectory can be described by a series of targets ; see fig . [
multitarget](b ) for a series of random targets , and ( c ) for a sinusoidal trajectory described by a large number of closely spaced targets .
the study of the behavior of the system when the path to follow has a very high curvature or describes very acute corners is matter for future work .
we have presented an agent based model for the guidance by repulsion problem , consisting of a system of equations corresponding to newton s second law .
the system can adopt two operating modes that can be controlled by a single parameter @xmath17 : for @xmath66 , the system is in the pure pursuit mode ( control in mode @xmath236 , _
i.e. _ , resting state ) , and for @xmath72 , the system is in the circumvention mode ( control in mode @xmath237 , _
i.e. _ , active state ) .
we have shown that by appropriately defining the function @xmath238 \to \{-1,0,1\}$ ] , the driver can guide the evader to any desired target or along any ( relatively smooth ) path .
we have then formulated an optimal control problem @xmath19 to find the optimal guidance strategy minimizing the cost in terms of the number of times the system is activated from its resting state and the time the system spends in the active mode . by means of ( numerical ) shooting methods ,
we have obtained the optimal open - loop strategies for the case where the number of activations is equal or smaller than one , finding that the system is highly sensitive to small variations of the activation / deactivation times .
these results show that open - loop controls would not be of practical interest in real problems due to the presence of external perturbations , thus suggesting the use of a feedback law . taking advantage of the information provided in the study of the behavior of the system under open - loop controls for @xmath120
, we have designed a feedback law for the case where @xmath119 , which allows to correct in real time for deviations from the desired trajectory .
we have found that the feedback law is also highly sensitive to small variations of the conditions of the problem , in this case , of the accuracy with which the target is reached ( _ i.e. _ , the radius of the target ball ) . moreover
, our results show that the feedback law , and therefore the resulting behavior of the system , are highly sensitive to possible delays in the switching times of the control .
this means that , in systems or devices where the manipulation of the control can not be carried out at arbitrarily close instants of time , the behavior of the driver can exhibit large oscillations that can produce an increase of the cost .
this may happen in situations where time delays exist in collecting and interpreting the data about the state of the system or in the reaction time of the system once the control is changed , especially when two consecutive changes are very close in time .
the main direction for the immediate future work consists in taking into account the cost for the driver to get close to the evader , both in time and travelled distance .
this would correspond to add the cost of back propellers to the cost functional : @xmath239 and find a feedback control law for the two controls @xmath17 and @xmath107 .
the interest of guidance by repulsion could also be extended to the case where the evader s behavior has a stochastic component and when multiple agents ( evaders and/or drivers ) are considered .
the feedback law will be especially relevant when noise is considered in both the behavior of the agents and in the manipulation of the data .
we thank the anonymous reviewers for their interesting suggestions and for pointing out the interest of open problem which include the cost of back propellers in the cost functional .
this material is based upon work that has received funding from the european union s horizon 2020 research and innovation programme under the marie skodowska - curie grant agreement no 655235 , entitled `` smartmass '' , the advanced grant numeri - waves / fp7 - 246775 of the european research council executive agency , the berc 2014 - 2017 program and the pre - doctoral grant pre-2014 - 1 - 461 of the basque government , the fa9550 - 15 - 1 - 0027 of afosr , the mtm2014 - 52347 and mtm2011 - 29306 grants and the severo ochoa program sev-2013 - 0323 of the mineco , and a humboldt award at the university of erlangen - nrnberg .
numerical simulations show that when @xmath131 continuously for a sufficiently large time , both agent s velocities converge asymptotically to the same constant velocity @xmath240 . in that state , @xmath241 and , from eqs .
( [ eq - vp - d])([eq - ve - d ] ) , we have @xmath242 = - { c^d_e \over \nu_e } { \vec{u}_d - \vec{u}_e \over \delta_{\rm as}^2 } = \vec{v}_{\rm as}.\end{aligned}\ ] ] thus , comparing norms , we obtain @xmath243 , and extracting @xmath68 , we obtain @xmath244 provided @xmath245 , as it is the case for the values we are considering . note also that a necessary condition to have an effective short - range repulsion acting on the driver is that the factor between large parentheses in eq .
( [ eq - vp - d ] ) is positive , so @xmath246 and the radicand in expression ( [ radis ] ) is positive .
we show here that driver and evader are prevented from separating infinitely from each other and tend asymptotically to be separated a distance of order one .
the proof follows the idea of `` free agents '' used in @xcite ( see also @xcite for a more similar model ) . a driver agent @xmath0 is said to be a _ free agent _ at time @xmath247 if its distance to the evader @xmath1 is greater than an arbitrarily large positive constant @xmath248 ; that is , @xmath249 . if the driver @xmath0 is a free agent , then the system ( [ eq - up])([eq - vs ] ) can be reduced as follows : @xmath250 using remark 1 .
if @xmath251 , then , @xmath252 , @xmath253 [ lema1 ] let @xmath254 and @xmath255 : @xmath256 be the following potential functional : @xmath257 then , for free agents ( _ i.e. _ , @xmath258 ) and under the hypotheses of lemma [ lema1 ] , @xmath255 is positive .
if the driver @xmath0 is a free agent , then @xmath255 is bounded from below and @xmath259 is negative along the agents trajectories defined by the system ( [ eq - up])([eq - vs ] ) .
consequently , @xmath260 converges in time to a minimum which is reached when the distance between both agents is @xmath261 .
[ teolyapunov ] the time - derivative of @xmath255 along the agents trajectories is given by : @xmath262 then : @xmath263 which , under the conditions of lemma [ lema1 ] , is negative .
then @xmath255 decreases and is bounded from below , so @xmath255 has a minimum , which is reached when @xmath264 and @xmath265 ( which is the minimum value of @xmath196 for a free agent ) .
thus , agents are prevented from separating infinitely from each other because as soon as @xmath258 , the driver becomes a free agent and is forced to move back towards the evader , provided the balance between the mass and the friction of the agents verifies lemma [ lema1 ] .
once the evader is turning back towards the target , that is , @xmath266 , we check the direction of the velocity vector of the evader with the line @xmath267 described by the evader and the target .
then , if for some time @xmath140 the vector @xmath194 points towards a point located below @xmath104 , the tentative value of @xmath165 must be reduced ; if instead , at the final time @xmath107 , the vector @xmath194 points towards a point located above @xmath104 , then the tentative value of @xmath165 must be augmented . 1 .
solve the system ( [ eq - up])([condinis ] ) with @xmath217 for @xmath140 and anotate the value of @xmath268 , which is the first time such that @xmath269 .
if no such time is reached , this means that @xmath270 , so take a larger value of @xmath165 and shoot again ( goto 0 ) .
2 . for each time @xmath271 , evaluate the instantaneous alignment @xmath272 of the velocity vector of the evader @xmath194 with respect to the target point @xmath273 : @xmath274 then , if @xmath275 , take a smaller value of @xmath165 and shoot again ( goto 0 ) .
3 . if at time @xmath107 the velocity vector of the evader is still pointing above the target , that is , @xmath276 , then take a larger value of @xmath165 and shoot again ( goto 0 ) .
the new value of @xmath165 for the next shoot can be selected with a simple method ( _ e.g. _ , bisection ) .
+ stop when @xmath277 , for a small value of the tolerance @xmath278 ; the value of @xmath279 has been found , proceed to the next value of @xmath164 .
coppinger rp , feinstein m. _ how dogs work . _ university of chicago press ( 2015 ) .
see also coppinger rp , coppinger l. _ dogs : a startling new understanding of canine origin , behavior and evolution . _
scribner , new york ( 2001 ) .
garnier s , combe m , jost c , theraulaz g. _ do ants need to estimate the geometrical properties of trail bifurcations to find an efficient route ?
a swarm robotics test bed .
_ plos comput biol 9 ( 2013 ) e1002903 .
gazi v , passino km .
_ stability analysis of swarms_. ieee trans on automatic control 48 ( 2003 ) 6927 .
gazi v , passino km .
_ a class of attraction / repulsion functions for stable swarm aggregations_. int j control 77 ( 2004 ) 156779 .
gazi v , passino km .
_ swarm stability and optimization_. springer science & business media ( 2011 ) .
escobedo r , muro c , spector l , coppinger rp . _
group size , individual role differentiation and effectiveness of cooperation in a homogeneous group of hunters_. j roy soc interface 11 ( 2014 ) 20140204 . | we present a _ guidance by repulsion _ model based on a driver - evader interaction where the driver , assumed to be faster than the evader , follows the evader but can not be arbitrarily close to it , and the evader tries to move away from the driver beyond a short distance .
the key ingredient allowing the driver to guide the evader is that the driver is able to display a circumvention maneuver around the evader , in such a way that the trajectory of the evader is modified in the direction of the repulsion that the driver exerts on the evader .
the evader can thus be driven towards any given target or along a sufficiently smooth path by controlling a single discrete parameter acting on driver s behavior .
the control parameter serves both to activate / deactivate the circumvention mode and to select the clockwise / counterclockwise direction of the circumvention maneuver . assuming that the circumvention mode is more expensive than the pursuit mode , and that the activation of the circumvention mode has a high cost
, we formulate an optimal control problem for the optimal strategy to drive the evader to a given target . by means of numerical shooting methods , we find the optimal open - loop control which reduces the number of activations of the circumvention mode to one and which minimizes the time spent in the active mode .
our numerical simulations show that the system is highly sensitive to small variations of the control function , and that the cost function has a nonlinear regime which contributes to the complexity of the behavior of the system , so that a general open - loop control would not be of practical interest .
we then propose a feedback control law that corrects from deviations while preventing from an excesive use of the circumvention mode , finding numerically that the feedback law significantly reduces the cost obtained with the open - loop control .
* keywords : * guidance by repulsion , driver - evader agents , optimal strategies , feedback control law , nonlinear dynamics , numerical simulations |
variational inequalities in probability , ergodic theory and harmonic analysis have been the subject of many recent research papers . one important character of these inequalities is the fact that they can be used to measure the speed of convergence for the family of operators in consideration . to be more precise ,
consider , for instance , a measure space @xmath17 and an operator @xmath2 on @xmath18 .
form the ergodic averages of @xmath2 : @xmath19 a fundamental theorem in ergodic theory states that if @xmath2 is a contraction on @xmath20 for every @xmath21 , then the limit @xmath22 exists a.e .
for every @xmath9 .
one can naturally ask what is the speed of convergence of this limit .
a classical tool for measuring that speed is the following square function @xmath23 the problem being to estimate its norm @xmath24 .
this issue goes back to stein @xcite , who proved that if @xmath2 as above is positive on @xmath25 ( in the hilbertian sense ) , then a square function inequality @xmath26 holds for any @xmath1 .
stein s inequality is closely related to dunford - schwartz s maximal ergodic inequality , @xmath27 this maximal inequality and its weak type @xmath28 substitute for @xmath29 are key ingredients in the proof of the previous pointwise ergodic theorem .
the strong @xmath8-variation is another ( better ) tool to measure the speed of @xmath22 .
bourgain was the first to consider variational inequalities in ergodic theory .
to state his inequality we need to recall the definition of the strong @xmath8-variation . given a sequence @xmath30 of complex numbers and a number @xmath31 , the strong @xmath8-variation norm is defined as @xmath32 where the supremum runs over all increasing sequences @xmath33 of integers such that @xmath34 .
it is clear that the set @xmath7 of all sequences with a finite strong @xmath8-variation is a banach space for the norm @xmath35 .
bourgain @xcite proved that if @xmath2 is induced by a measure preserving transformation on @xmath17 , then for any @xmath36 , @xmath37 this inequality was then extended to @xmath20 for any @xmath1 by jones , kaufman , rosenblatt and wierdl @xcite .
the latter paper contains many other interesting results on the subject .
note that the predecessor of bourgain s inequality is lpingle s variational inequality for martingales @xcite .
the latter says that if @xmath38 is an increasing sequence of conditional expectations on a probability space @xmath39 , then we have an estimate @xmath40 ( see also @xcite for further results on that theme ) . since @xcite
, variational type inequalities have been extensively studied in ergodic theory and harmonic analysis .
many classical sequences of operators and semigroups have been proved to satisfy strong variational bounds , see in particular @xcite and references therein .
the main purpose of this paper is to exhibit a large class of operators @xmath2 on @xmath20 for a fixed @xmath1 with the following property : for any @xmath6 , there exists a constant @xmath41 ( which may depend on @xmath8 and @xmath2 ) such that for any @xmath9 , the sequence @xmath42 belongs to @xmath43 , and @xmath44 we show that this holds true provided that @xmath2 is a positive contraction ( more generally , a contractively regular operator ) and @xmath2 is analytic , in the sense that @xmath45 inequality implies , of course , a similar variational inequality for the ergodic averages @xmath46 .
however , in this latter case , the analytic assumption above can be removed .
namely , for a positive contraction @xmath2 on @xmath20 with @xmath1 we have an estimate @xmath47 for any @xmath6 .
this result extends those of @xcite and @xcite quoted previously . note that inequality for positive analytic contractions considerably improves our previous maximal ergodic inequality for such operators @xmath2 proved in @xcite . in this sense , this paper is a continuation of @xcite . on the other hand ,
our proof of heavily relies on the square function inequality of @xcite .
we also establish results similar to ( [ 1q1 ] ) and ( [ 1q3 ] ) for strongly continuous semigroups .
this requires the following continuous analog of @xmath7 .
given a complex family @xmath48 , define @xmath49 where the supremum runs over all increasing sequences @xmath50 of positive real numbers .
then we let @xmath51 be the resulting banach space of all @xmath48 such that @xmath52 .
consider a bounded analytic semigroup @xmath15 on @xmath20 , with @xmath1 . we show that if @xmath53 is a positive contraction ( more generally , a contractively regular operator ) for any @xmath54 , then for any @xmath6 and any @xmath9 , the family @xmath55(\lambda)\bigr)_{t>0}$ ] belongs to @xmath51 for almost every @xmath56 , and we have an estimate @xmath57(\lambda)\bigr)_{t>0}\bigr\vert}_{v^q}\bigr\vert}_p \,\leq\,c\,{\vertx\vert}_p,\qquad x\in l^p(\omega).\ ] ] we mention that jones and reinhold @xcite proved ( [ 1q2 ] ) for positive , unital symmetric diffusion semigroups and ( [ 1q1 ] ) for a certain class of convolution operators ( only in the case @xmath58 ) .
our results turn out to extend these contributions in various directions . as in the discrete case ,
we obtain similar results for the averages of the semigroup if we remove the analyticity assumption .
inequalities for ergodic averages , such as ( [ 1q3 ] ) , will be established in section 3 . then our main results leading to ( [ 1q1 ] ) and ( [ 1q2 ] ) will be established in section 4 , using the above mentioned results from section 3 , as well as square function estimates from @xcite .
section 5 is devoted to various complements . on the one hand ,
we establish individual (= pointwise ) ergodic theorems in our context .
for example , if @xmath0 is a positive analytic contraction , then @xmath59 exists a.e . for every @xmath9 . on the other hand , it is well - known that ( [ 1q3 ] ) can not be extended to the case @xmath60 .
then we show analogs of ( [ 1q3 ] ) and ( [ 1q1 ] ) when @xmath7 is replaced by the oscillation space @xmath61 .
we give examples and applications in section 6 .
we end this introduction with a few notation .
if @xmath62 is a banach space , we let @xmath63 denote the algebra of all bounded operators on @xmath62 and we let @xmath64 denote the identity operator on @xmath62 ( or simply @xmath65 if there is no ambiguity on @xmath62 ) . for any @xmath66 , we let @xmath67 denote the spectrum of @xmath2
. also we let @xmath68 be the open unit disc of @xmath69 . for a measurable function @xmath70 acting on a measure space @xmath17
, @xmath71 denotes the @xmath16-norm of @xmath72 .
we refer to @xcite and @xcite for background on strongly continuous and analytic semigroups on banach space .
the aim of this section is to provide some elementary background on the spaces @xmath7 and @xmath51 defined above .
we fix some @xmath31 .
let @xmath30 be an element of @xmath7 .
then the sequence @xmath30 is bounded , with @xmath73 and @xmath74 thus the space of eventually constant sequences is dense in @xmath7 .
consequently any element of @xmath7 is a converging sequence . for any integer @xmath75 ,
let @xmath76 be the space of @xmath77-tuples @xmath78 of complex numbers , equipped with the norm @xmath79 it follows from above that an infinite sequence @xmath30 belongs to @xmath7 if and only if there is a constant @xmath80 such that @xmath81 for any @xmath75 and in this case , @xmath82 let @xmath17 be a measure space , let @xmath1 and let @xmath43 denote the corresponding bochner space .
any element of that space can be naturally regarded as a sequence of @xmath20 .
the following is a direct consequence of the above approximation properties .
[ 6lem1 ] let @xmath83 be a sequence of @xmath20 , the following assertions are equivalent .
* the sequence @xmath83 belongs to @xmath43 . *
the sequence @xmath84 belongs to @xmath7 for almost every @xmath11 and the function @xmath85 belongs to @xmath20 .
* there is a constant @xmath80 such that @xmath86 for any @xmath75 . in this case , @xmath87 we now consider the continuous case .
we note for further use that any element @xmath48 of @xmath51 admits limits @xmath88 the following is a continuous analog of lemma [ 6lem1 ] .
note however that families satisfying the next statement do not necessarily belong to the bochner space @xmath89 .
[ 6lem2 ] let @xmath90 be a family of @xmath20 and assume that : * for a.e .
@xmath11 , the function @xmath91 is continuous on @xmath92 .
* there exists a constant @xmath80 such that whenever @xmath93 is a finite increasing sequence of positive real numbers , we have @xmath94 then @xmath95 belongs to @xmath51 for a.e .
@xmath11 , the function @xmath96 belongs to @xmath20 and @xmath97 for any integer @xmath98 , define @xmath99 it follows from ( [ 6approx ] ) that @xmath100 is measurable .
moreover the sequence @xmath101 is nondecreasing , and we may therefore define @xmath102 by construction , @xmath103 is measurable and by the monotone convergence theorem , its @xmath16-norm is equal to @xmath104 . according to the assumption ( 2 ) and the approximation property ( [ 6approx ] ) , the @xmath16-norm of @xmath100 is @xmath105 for any @xmath98 .
hence @xmath106 this implies that @xmath107 for a.e . @xmath11 . if @xmath108 is such that @xmath91 is continuous on @xmath92 , then @xmath109 according to the assumption ( 1 ) , this holds true almost everywhere .
hence the quantity @xmath110 is finite for a.e .
the lemma clearly follows from these properties .
throughout we let @xmath17 be a measure space and we let @xmath1 .
we first recall the notion of regular operators on @xmath20 and some of their basic properties which will be used in this paper .
we refer e.g. to ( * ? ? ?
* chap . 1 ) and to @xcite for more details and complements
. an operator @xmath0 is called regular if there is a constant @xmath80 such that @xmath111 for any finite sequence @xmath112 in @xmath20 . then we let @xmath113 denote the smallest @xmath114 for which this holds .
the set of all regular operators on @xmath20 is a vector space on which @xmath115 is a norm .
we say that @xmath2 is contractively regular if @xmath116 .
let @xmath117 be a banach space .
if an operator @xmath118 is regular , then the operator @xmath119 extends to a bounded operator on the bochner space @xmath120 , and @xmath121 indeed by definition this holds true when @xmath122 for any @xmath5 , and the general case follows from the fact that for any @xmath123 , any finite dimensional banach space is @xmath124-isomorphic to a subspace of @xmath125 for some large enough @xmath5 .
any positive operator @xmath2 ( in the lattice sense ) is regular and @xmath126 in this case .
thus all statements given for contractively regular operators apply to positive contractions .
it is well - known that conversely , @xmath2 is regular with @xmath127 if and only if there is a positive operator @xmath128 with @xmath129 , such that @xmath130 for any @xmath9 .
finally , following @xcite , we say that an operator @xmath131 is an absolute contraction if it induces two contractions @xmath132 then the resulting operator @xmath133 is contractively regular . the main result of this section is the following theorem , which might be known to experts .
its proof relies on the transference principle , already used in @xcite .
[ 2main ] let @xmath0 be a contractively regular operator , with @xmath1 , and let @xmath6 .
then we have @xmath134 for some constant @xmath135 only depending on @xmath136 and @xmath8 .
let @xmath137 denote the shift operator on @xmath138 . according to ( * ? ? ?
b ) , @xmath139 satisfies theorem [ 2main ] ( the crucial case @xmath58 going back to @xcite ) . thus for any @xmath1 and @xmath6
we have a constant @xmath140 the following lemma is a variant of the well - known coifman - weiss transference theorem ( * ? ? ?
2.4 ) and is closely related to @xcite and @xcite .
[ 2transfer ] let @xmath141 be an invertible operator such that @xmath142 is regular for any @xmath143 and suppose that @xmath144 .
let @xmath145 be a basis of a finite dimensional banach space @xmath117 , and let @xmath146 be @xmath147 elements of @xmath148 .
consider the two operators @xmath149 and @xmath150 then @xmath151 the operator @xmath152 extends to a bounded operator @xmath153 , whose norm is equal to @xmath154 .
( nothing special about @xmath155 is required for this tensor extension property . ) by fubini s theorem , @xmath156 isometrically .
further , under these identifications , the extension of @xmath152 corresponds to the operator @xmath157 thus we have @xmath158 . by approximation
, we may suppose that @xmath159 are finitely supported .
let @xmath75 be chosen such that @xmath160 for any @xmath161 and any @xmath162 .
let @xmath9 and let @xmath163 for any @xmath164 .
our aim is to estimate the norm of @xmath165 in @xmath166 . for any @xmath167
, we have @xmath168 hence applying ( [ 2reg ] ) to @xmath169 , we derive that @xmath170 let @xmath5 be an arbitrary integer and let @xmath171 be the characteristic function of the interval @xmath172 $ ] .
we deduce from the above estimate that @xmath173 since @xmath174\bigr\vert}^p_{\ell^p_{{\ensuremath{\mathbb{z}}}}(l^p(\omega ; e))}\ ] ] and @xmath158 , this yields @xmath175 letting @xmath176 , we get the result .
let @xmath0 be a contractively regular operator .
there exists another measure space @xmath177 , two positive contractions @xmath178 and @xmath179 and an isometric invertible operator @xmath180 such that @xmath181 in the case when @xmath2 is positive , this is akcoglu s famous dilation theorem ( see @xcite ) .
the extension to regular operators stated here is from @xcite or @xcite .
moreover @xmath182 can be chosen so that @xmath182 and @xmath183 are both contractively regular .
thus @xmath184 note that when @xmath185 , any isometry on @xmath16 is contractively regular , so the latter information is relevant only when @xmath58 .
we fix an integer @xmath75 and we consider @xmath186 as defined in section 2 . for any @xmath187 ,
we clearly have @xmath188 since @xmath189 , it follows from ( [ 2reg ] ) that @xmath190 for any @xmath191 , let @xmath192 be defined by letting @xmath193 if @xmath194 and @xmath195 otherwise .
then @xmath196 applying lemma [ 2transfer ] with @xmath197 and recalling ( [ 2cpq ] ) , we therefore deduce that @xmath198 for any @xmath199 . combining with the inequality ( [ 2dilation ] )
we obtain that @xmath200 for any @xmath9 .
then the result follows from lemma [ 6lem1 ] .
[ 2amenable ] the above lemma [ 2transfer ] extends without any difficulty to amenable groups , as follows .
let @xmath201 be a locally compact amenable group , with left haar measure @xmath202 , let @xmath203 be a strongly continuous representation valued in the space of regular operators on @xmath20 , and assume that @xmath204 next let @xmath205 be @xmath147 elements of @xmath206 and let @xmath207 be defined by letting @xmath208 for any @xmath209 . then for any @xmath9 , we have @xmath210 we conclude this section with a continuous version of theorem [ 2main ] . given a strongly continuous semigroup @xmath211 on @xmath20 , we let @xmath212 defined in the strong sense .
[ 2continuous ] let @xmath211 be a strongly continuous semigroup on @xmath20 and assume that @xmath213 is contractively regular for any @xmath54 .
let @xmath6 and let @xmath9 . then for a.e .
@xmath11 , the family @xmath214(\lambda)\bigr)_{t>0}$ ] belongs to @xmath51 and @xmath215(\lambda)\bigr)_{t > 0}\bigr\vert}_{v^q}\bigr\vert}_{p}\,\leq\ , c_{p , q}\,{\vertx\vert}_p.\ ] ] consider @xmath9 . according to ( * ? ? ?
* section viii.7 ) , the function @xmath216(\lambda)$ ] is continuous for a.e .
@xmath11 . let @xmath217 be positive real numbers and let @xmath123 .
it follows from the strong continuity of @xmath211 that there exist @xmath218 and integers @xmath219 such that @xmath220 hence applying theorem [ 2main ] and a limit argument , we deduce that @xmath221 the result therefore follows from lemma [ 6lem2 ] .
an alternative proof of ( [ 2key ] ) consists in using fendler s dilation theorem for semigroups ( see @xcite ) , and then arguing as in the proof theorem [ 2main ] .
this only requires knowing that the result of corollary [ 2continuous ] holds true for the translation group on @xmath222 , which follows from @xcite , and using remark [ 2amenable ] for @xmath223 to transfer that result to strongly continuous groups of contractively regular isometries .
let @xmath62 be an arbitrary banach space , and let @xmath15 be a strongly continuous semigroup on @xmath62 .
we call it a bounded analytic semigroup if there exists a positive angle @xmath224 and a bounded analytic family @xmath225 extending @xmath226 , where @xmath227 is the open sector of angle @xmath228 around @xmath92 .
we refer to @xcite for various characterizations and properties of bounded analytic semigroups .
we simply recall that if we let @xmath229 denote the infinitesimal generator of @xmath15 , then the latter is a bounded analytic semigroup if and only if @xmath53 maps @xmath62 into the domain of @xmath229 for any @xmath230 and there exist two constants @xmath231 such that @xmath232 the definition of analyticity for discrete semigroups parallels ( [ 3anal1 ] ) .
let @xmath66 .
we say that @xmath2 is power bounded if @xmath233 and that it is analytic if moreover , @xmath234 this notion goes back to @xcite and has been studied in various contexts so far .
we gather here a few spectral properties of these operators and refer to @xcite for proofs and complements .
the most important result is the following : an operator @xmath2 is power bounded and analytic if and only if @xmath235 this property is called the ` ritt condition ' .
the key argument for this characterization is due to o. nevanlinna @xcite . for any angle @xmath236 ,
let @xmath237 be the convex hull of @xmath238 and the closed disc @xmath239 . ]
( 0,0 ) ( -2,65)@xmath238 ( -68,65)@xmath240 ( -32,81)@xmath241 ( -55,85)@xmath237 then ( [ 3ritt ] ) implies that @xmath242 furthermore , ( [ 3stolz1 ] ) is equivalent to @xmath243 the aim of this section is to show that under an analyticity assumption , the ergodic averages can be replaced by the semigroup itself in either theorem [ 2main ] ( discrete case ) or corollary [ 2continuous ] ( continuous case ) . as in section 3 , we consider a measure space @xmath17 and a number @xmath1 .
we will consider an operator @xmath0 and we let @xmath244 for any integers @xmath245 .
note that @xmath246 is the @xmath247-difference sequence of @xmath248 .
we will need the following littlewood - paley type inequalities which were estabished in @xcite .
[ 3sfe ] let @xmath0 be a contractively regular operator , with @xmath1 , and assume that @xmath2 is analytic .
then for any integer @xmath249 , there is a constant @xmath250 such that @xmath251 we will also use the following elementary estimates , whose proofs are left to the reader .
[ 3sum ] for any integer @xmath249 , there exists a constant @xmath252 such that for any @xmath5 , @xmath253 [ 3mult ] for any sequences @xmath254 and @xmath255 , we have @xmath256 and @xmath257 in the next statements and their proofs , @xmath258 will stand for an inequality up to a constant which may depend on @xmath2 , @xmath8 and @xmath247 , but not on @xmath72 .
[ 3main ] let @xmath0 be a contractively regular operator , with @xmath1 , and assume that @xmath2 is analytic .
then for any @xmath6 , we have an estimate @xmath259 more generally , for any integer @xmath249 , we have an estimate @xmath260 it will be convenient to set @xmath261 then for any @xmath262 and any @xmath263 , we have @xmath264 with the above notation , ( [ 3delta ] ) holds true for @xmath265 , by theorem [ 2main ]
. we will proceed by induction .
we fix an integer @xmath249 and assume that ( [ 3delta ] ) holds true for @xmath266 . thus using lemma [ 3mult ] , we both have @xmath267 and @xmath268 next for any @xmath5 , we write @xmath269 using ( [ 31 ] ) in due places
. hence @xmath270 this identity suggests the introduction of the following two sequences of operators . for any @xmath5
, we set @xmath271 also for any @xmath9 , we set @xmath272 according to proposition [ 3sfe ] , this function is an element of @xmath20 .
let @xmath33 be an increasing sequence of integers , with @xmath273 .
for any @xmath274 , we set @xmath275 this yields a decomposition @xmath276 let @xmath9 .
if @xmath277 , we have , using cauchy - schwarz , @xmath278 similarly if @xmath279 , we have @xmath280 hence in both cases , we have @xmath281 by lemma [ 3sum ] . summing up
, we deduce that @xmath282 likewise we have @xmath283 we now turn to @xmath284 .
assume that @xmath277 .
then using again cauchy - schwarz and lemma [ 3sum ] , we have @xmath285 hence @xmath286 for any integer @xmath287 , define @xmath288 and set @xmath289 then it follows from the above calculation that @xmath290 let us now estimate the @xmath291 s .
observe that if @xmath292 is a non empty set , then it is a finite interval of integers .
thus it reads as @xmath293 where @xmath294 is the biggest element of @xmath292 and @xmath147 is its cardinal .
suppose that @xmath295 , so that the sequence @xmath296 is increasing .
then @xmath297 since @xmath298 , we have @xmath299 , hence @xmath300 on the other hand , we have @xmath301 for any @xmath302 , hence @xmath303 we immediatly deduce that @xmath304 in the case when @xmath305 , we have similarly @xmath306 hence @xmath307 in this case .
lastly , it is plain that if @xmath308 , we have @xmath309 .
this shows that in all cases , we have an estimate @xmath310 now recall ( [ 3decomp ] ) . combining the above estimate with ( [ 3alpha ] ) and ( [ 3beta ] ) ,
we obtain that @xmath311 since the upper bound does not depend on the sequence @xmath33 , this estimate and proposition [ 3sfe ] imply that the sequence @xmath312 belongs to @xmath313 , and that we have an estimate @xmath314 we will now apply a similar treatment to the sequence @xmath315 . according to ( [ 31 ] ) , we can write @xmath316 for any @xmath274 , we set @xmath317 if @xmath277 , and @xmath318 if @xmath319 . arguing as above , we obtain that for any @xmath9 , we have @xmath320 and then @xmath321 likewise we have @xmath322 as well as an estimate @xmath323 these three inequalities imply that the sequence @xmath324 belongs to @xmath313 , and that we have an estimate @xmath325 we can now conclude our proof . recall that @xmath326 , so that @xmath327
. then using ( [ 3induction1 ] ) , ( [ 3induction2 ] ) , ( [ 3a ] ) , ( [ 3b ] ) and lemma [ 3mult ] , it follows from the decomposition formula ( [ 32 ] )
that for any @xmath9 , @xmath328 belongs to @xmath43 and that we have an estimate @xmath329 since @xmath330 , a second application of ( [ 3b ] ) yields ( [ 3delta ] ) .
the following is an analog of theorem [ 3main ] for continuous semigroups .
[ 3continuous ] let @xmath15 be a bounded analytic semigroup on @xmath20 and assume that @xmath213 is contractively regular for any @xmath54 .
let @xmath6 and let @xmath9 .
then for a.e .
@xmath11 , the family @xmath331(\lambda)\bigr)_{t>0}$ ] belongs to @xmath51 and we have an estimate @xmath332(\lambda)\bigr)_{t > 0}\bigr\vert}_{v^q}\bigr\vert}_{p}\,\lesssim\,{\vertx\vert}_p.\ ] ] more generally , for any integer @xmath249 , the family @xmath333 belongs to @xmath51 for a.e .
@xmath11 and we have an estimate @xmath334 let @xmath249 be an integer .
it follows from ( * ? ? ?
* lemma , p. 72 ) that for any @xmath9 , the function @xmath335 is continuous for a.e . @xmath108 .
next it follows from the proof of ( * ? ? ?
4.2 ) that there exists a constant @xmath250 such that @xmath336 for any @xmath230 and any @xmath9 .
that is , the operators @xmath53 satisfy proposition [ 3sfe ] uniformly .
since they also satisfy theorem [ 2main ] uniformly , it follows from the proof of theorem [ 3main ] that they satisfy the estimate ( [ 3delta ] ) uniformly . hence using an approximation argument as in the proof of ( * ? ? ?
4.2 ) , we deduce that for any @xmath9 and for any @xmath337 , we have @xmath338 the result therefore follows from lemma [ 6lem2 ] .
we give here further properties of contractively regular operators and contractively regular semigroups , in connection with variational inequalities .
let @xmath17 be a measure space and let @xmath1 .
let @xmath0 be a contraction .
according to the mean ergodic theorem , we have a direct sum decomposition @xmath339 where @xmath340 and @xmath341 denote the kernel and the range , respectively .
moreover if we let @xmath342 denote the corresponding projection onto @xmath343 , then @xmath344 for any @xmath9 .
it is well - known that if @xmath2 is an absolute contraction , then @xmath345 almost everywhere ( see e.g. ( * ? ? ?
* section viii.6 ) ) .
we extend this classical result , as follows .
assume that @xmath0 is contractively regular .
then for any @xmath9 , @xmath346(\lambda)\,\longrightarrow\ , [ p_t(x)](\lambda)\qquad \hbox{for a.e.}\ \lambda\in\omega.\ ] ] let @xmath9 and let @xmath6 . according to theorem [ 2main ] , the sequence @xmath347(\lambda)\bigr)_{n\geq 0}$ ] belongs to @xmath7 for almost every @xmath11 .
hence @xmath347(\lambda)\bigr)_{n\geq 0}$ ] converges for almost every @xmath11 . combining with ( [ 7mean ] )
, we obtain the result .
if a contraction @xmath0 is analytic , then @xmath348 , hence @xmath349 for any @xmath350 . consequently , @xmath351 for any @xmath9 . using theorem [ 3main ] and arguing as above , we obtain the following . [ 72 ]
let @xmath0 be a contractively regular operator and assume that @xmath2 is analytic .
then for any @xmath9 , @xmath352(\lambda)\,\longrightarrow\ , [ p_t(x)](\lambda)\qquad \hbox{for a.e.}\ \lambda\in\omega.\ ] ] we now consider the continuous case .
the situation is essentially similar , except that we can also consider the behaviour when the parameter @xmath353 tends to @xmath354 .
let @xmath211 be a strongly continuous semigroup of contractions . by definition , for any @xmath9 , @xmath355 in the @xmath16-norm
when @xmath356 .
this implies that @xmath357 when @xmath356 .
let @xmath229 denote the infinitesimal generator of @xmath211 .
as in the discrete case , we have a direct sum decomposition @xmath358 moreover if we let @xmath359 denote the corresponding projection onto @xmath360 , then @xmath361 for any @xmath9 . if further @xmath15 is a bounded analytic semigroup , then @xmath362 for any @xmath9 .
now applying corollaries [ 2continuous ] and [ 3continuous ] , we deduce the following individual ergodic theorems . [ 74 ] let @xmath211 be a strongly continuous semigroup of contractively regular operators on @xmath20 , and let @xmath9 .
then for almost every @xmath11 , @xmath363(\lambda)\,\longrightarrow\ , [ p_a(x)](\lambda)\quad\hbox{when}\ t \to\infty\ ] ] and @xmath363(\lambda)\,\longrightarrow\ , x(\lambda)\quad\hbox{when}\ t\ , \to 0^+.\ ] ] [ 74 ] let @xmath211 be a bounded analytic semigroup and assume that @xmath53 is contractively regular for any @xmath54 .
let @xmath9 .
then for almost every @xmath11 , @xmath364(\lambda)\,\longrightarrow\ , [ p_a(x)](\lambda)\quad\hbox{when}\ t\to\infty\ ] ] and @xmath364(\lambda)\,\longrightarrow\ , x(\lambda)\quad\hbox{when}\ t\to 0^+.\ ] ] in this section , we fix a increasing sequence @xmath33 of integers , with @xmath365 . given any sequence @xmath30 of complex numbers , we define the so - called oscillation norm @xmath366 and we let @xmath61 denote the banach space of all sequences with a finite oscillation norm , equipped with @xmath367 .
this space ( whose definition depends on the sequence @xmath33 ) was used in @xcite as a substitute to @xmath7 in the case @xmath60 ( see also @xcite ) .
indeed , neither theorem [ 2main ] nor theorem [ 3main ] holds true for @xmath60 , see @xcite and ( * ? ? ?
* section 8) .
recall the shift operator @xmath368 for any @xmath1 ( see section 3 ) . according to (
a ) , there is a constant @xmath369 such that @xmath370 for any @xmath371 . hence arguing as in the proof of theorem [ 2main ] , we obtain the following @xmath61-version of the latter statement . [ 41 ] let @xmath0 be a contractively regular operator , with @xmath1
. then we have @xmath372 we also have an @xmath61-version of theorem [ 3main ] , as follows .
[ 42 ] let @xmath0 be a contractively regular operator , with @xmath1 , and assume that @xmath2 is analytic .
then we have an estimate @xmath373 more generally , for any integer @xmath249 , we have an estimate @xmath374 the proof is a variant of the one written for theorem [ 3main ] , let us explain this briefly .
we use the notation from section 4 . for any @xmath263 , consider the following three properties : * @xmath375 ; * @xmath376 ; * @xmath377 .
property ( i)@xmath378 is the result we wish to prove . our strategy is to show , by induction , that these three estimates hold true .
first , it is easy to deduce from proposition [ 3sfe ] that ( i)@xmath378 and ( ii)@xmath378 are equivalent .
second it follows from ( [ 3a ] ) , ( [ 3b ] ) and ( [ 32 ] ) that ( ii)@xmath379 and ( iii)@xmath379 imply ( iii)@xmath378 and ( i)@xmath378 .
indeed @xmath380 and @xmath381 .
hence it suffices to show ( i)@xmath382 and ( iii)@xmath382 . this is obtained by applying theorem [ 41 ] twice , the first time for the @xmath61-space associated with the sequence @xmath383 , the second time for @xmath61-space associated with the sequence @xmath384 .
there are also @xmath61-versions of corollary [ 2continuous ] and corollary [ 3continuous ] , whose statements are left to the reader .
it is well known that variational inequalities for a sequence of operators have consequences in terms of jump functions . for any @xmath385 and any sequence @xmath386 of complex numbers ,
let @xmath387 denote the number of @xmath388-jumps of @xmath389 , defined as the supremum of all integers @xmath390 for which there exist integers @xmath391 such that @xmath392 for each @xmath164 .
it is clear that for any @xmath31 , @xmath393 combining with theorem [ 2main ] and theorem [ 3main ] , we immediately obtain ( as in ( * ? ? ?
* thm . 3.15 ) ) the following jump estimates . [ 3jumps ] consider @xmath1 and @xmath6 .
let @xmath394 be a contractively regular operator .
* we have an estimate @xmath395(\lambda)\bigr)_{n\geq 0}\ , , \ , \tau\bigr)^{\frac{1}{q}}\bigr\vert}_p\,\lesssim \ , \frac{{\vertx\vert}_p}{\tau}\,,\ ] ] and for any @xmath396 , we also have @xmath397(\lambda)\bigr)_{n\geq 0}\ , , \ , \tau\bigr)\,>\ , k\bigr\}\,\lesssim\ , \frac{{\vertx\vert}_p^p}{\tau^p k^{\frac{p}{q}}}\,,\ ] ] * assume moreover that @xmath2 is analytic .
then we have similar estimates @xmath398(\lambda)\bigr)_{n\geq 0 } \
, , \ , \tau\bigr)^{\frac{1}{q}}\bigr\vert}_p\,\lesssim \ , \frac{{\vertx\vert}_p}{\tau}\,,\ ] ] and @xmath399(\lambda)\bigr)_{n\geq 0}\ , , \ , \tau\bigr)\,>\ , k\bigr\}\,\lesssim\ , \frac{{\vertx\vert}_p^p}{\tau^p k^{\frac{p}{q}}}\,,\ ] ] furthermore for any integer @xmath249
, similar results hold with @xmath400 instead of @xmath401 .
similar results for the continuous case can be deduced as well from corollary [ 2continuous ] and corollary [ 3continuous ] .
we will now exhibit various classes of operators or semigroups to which our results from section 4 and section 5 apply .
we focus on statements involving @xmath8-variation , although statements involving the oscillation norm from the subsection 5.2 are also possible .
we start with the continuous case .
let @xmath15 be a strongly continuous semigroup on @xmath25 and assume that each @xmath53 is an absolute contraction , that is , @xmath402 for any @xmath403 and any @xmath230 ( see section 3 ) .
thus for any @xmath1 , @xmath15 extends to a strongly continuous semigroup of contractively regular operators .
it is well - known that @xmath15 is a bounded analytic semigroup on @xmath20 for every @xmath1 if ( and only if ) it is a bounded analytic semigroup on @xmath20 for one @xmath1 .
applying corollary [ 3continuous ] and corollary [ 74 ] , we derive the following .
[ 3diff2 ] let @xmath15 be a bounded analytic semigroup on @xmath25 satisfying ( [ 3diff1 ] ) .
then it satisfies estimates ( [ 3cont1 ] ) and ( [ 3cont2 ] ) for every @xmath1 and every @xmath6 .
moreover for any @xmath9 , @xmath404 converges almost everywhere when @xmath356 and when @xmath405 .
we now consider the so - called subordinated semigroups .
let @xmath1 and let @xmath15 be a strongly continuous bounded semigroup on @xmath20 .
let @xmath229 denote its infinitesimal generator and let @xmath407 .
then @xmath408 generates a bounded analytic semigroup @xmath409 on @xmath20 , and for any @xmath230 , there exists a continuous function @xmath410 such that @xmath411 and @xmath412 see e.g. ( * ? ? ?
* ix.11 ) for details and complements .
[ 3subord ] let @xmath15 be a strongly continuous semigroup on @xmath20 , with @xmath1 , and assume that @xmath213 is contractively regular for any @xmath54 .
then for any @xmath413 and any @xmath6 , we have an estimate @xmath414(\lambda)\bigr)_{t > 0}\bigr\vert}_{v^q}\bigr\vert}_{p}\,\lesssim\,{\vertx\vert}_p\ ] ] for the subordinated semigroup @xmath415 .
moreover for any @xmath9 , @xmath416 converges almost everywhere when @xmath356 and when @xmath405 .
the ` only if ' part holds for any operator , as discussed at the beginning of section 4 .
conversely , assume that a normal operator @xmath421 satisfies ( [ 3stolz2 ] ) .
applying the spectral theorem , we deduce that @xmath2 is a contraction and that for some constant @xmath396 , we have @xmath422 } n(r^{n-1 } -r^n)\,=\,k \bigl(\frac{n-1}{n}\bigr)^{n-1}\end{aligned}\ ] ] for any @xmath5 .
hence the sequence @xmath423 is bounded .
[ 3normal2 ] let @xmath424 be a contractively regular normal operator satisfying ( [ 3stolz2 ] ) .
then it satisfies an estimate @xmath425 for any @xmath6 .
moreover for any @xmath426 , @xmath427 converges almost everywhere when @xmath176 .
[ 3diff3 ] let @xmath428 be an absolute contraction and assume that @xmath424 is analytic .
then @xmath2 satisfies estimates ( [ 3tn ] ) and ( [ 3delta ] ) for every @xmath1 and every @xmath6 .
moreover for any @xmath9 , @xmath427 converges almost everywhere when @xmath176 .
of course the above corollary applies if @xmath424 is a positive selfadjoint operator , more generally if @xmath2 is normal and satisfies ( [ 3stolz2 ] ) . as a consequence , we extend the main result of @xcite , as follows .
the next statement solves a problem raised in the latter paper .
let @xmath201 be a locally compact abelian group and let @xmath429 denote the corresponding @xmath16-spaces with respect to a haar measure .
let @xmath430 be a probability measure on @xmath201 and let @xmath431 be the associated convolution operator , @xmath432 assume that there exists a constant @xmath396 such that @xmath433 for any @xmath434 .
then we have an estimate @xmath435 for any @xmath1 and any @xmath6 .
regard @xmath2 as an @xmath418-operator . by
fourier analysis , its spectrum is equal to the essential range of @xmath436 .
it therefore follows from the assumption and lemma [ 3normal1 ] that the operator @xmath437 is analytic .
hence @xmath2 satisfies the assumptions of corollary [ 3diff3 ] , which yields the result . for an operator @xmath438 ,
let @xmath439 denote the numerical range .
we recall that @xmath440 is a compact convex set and that @xmath441 .
assume that there exists @xmath442 such that @xmath443 ( which is a stronger condition than ( [ 3stolz1 ] ) or ( [ 3stolz2 ] ) ) .
then according to @xcite , there exists a constant @xmath41 such that @xmath444 for any polynomial @xmath445 .
arguing as in the proof of lemma [ 3normal1 ] , we deduce that @xmath2 is an analytic power bounded operator .
m. akcoglu , and l. sucheston , _ dilations of positive contractions on @xmath446 spaces _ , canad .
math . bull . 20
( 1977 ) , 285 - 292 .
n. asmar , e. berkson , and t. a. gillespie , _ transference of strong type maximal inequalities by separation - preserving representations _
j. math . 113
( 1991 ) , 47 - 74 .
s. blunck , _ analyticity and discrete maximal regularity on @xmath446-spaces _ , j. funct .
183 ( 2001 ) , 211 - 230 .
j. bourgain , _ pointwise ergodic theorems for arithmetic sets _ ,
ihes 69 ( 1989 ) , 5 - 41
. j. t. campbell , r. l. jones , k. reinhold , and m. wierdl , _ oscillation and variation for the hilbert transform _ , duke math .
j. , 105 ( 2000 ) , 59 - 83 .
r. coifman , r. rochberg , and g. weiss , _ applications of transference : the @xmath446 version of von neumann s inequality and the littlewood - paley - stein theory _ , pp . 53 - 67 in linear spaces and
approximation " , birkhuser , basel , 1978 .
r. r. coifman , and g. weiss , _ transference methods in analysis _ , cbms 31 , amer .
1977 . t. coulhon , and l. saloff - coste , _
puissances dun oprateur rgularisant _ , ann .
inst . h. poincar probab .
statist . 26
( 1990 ) , no . 3 , 419 - 436 .
r. crescimbeni , r. a. macas , t. menrguez , j. l. torrea , and b. viviani , _ the @xmath447-variation as an operator between maximal operators and singular integrals _ , j. evol .
9 ( 2009 ) , 81 - 102 .
a. de la torre , _ a simple proof of the maximal ergodic theorem _ ,
j. math . 28
( 1976 ) , 1073 - 1075
. b. delyon , and f. delyon , _ generalization of von neumann s spectral sets and integral representation of operators _ , bull .
france 127 ( 1999 ) , 25 - 41 .
n. dunford , and j. t. schwartz , _ linear operators , part 1 _ , pure and applied mathematics , vol .
7 interscience publishers , inc . , new york ; interscience publishers , ltd .
, london 1958 xiv+858 pp
. g. fendler , _ dilations of one parameter semigroups of positive contractions on @xmath446-spaces _ , canad . j. math
. 49 ( 1997 ) , 736 - 748 .
gaposhkin , _ ergodic theorem for functions of normal operators ( russian ) _ , funktsional .
i prilozhen .
18 ( 1984 ) , 1 - 6 . j. a. goldstein , _ semigroups of linear operators and applications _ , oxford university press , new - york , 1985 .
r. l. jones , r. kaufman , j. m. rosenblatt , and m. wierdl , _ oscillation in ergodic theory _ , ergodic theory dynam .
systems 18 ( 1998 ) , 889 - 935 .
r. l. jones , and k. reinhold , _ oscillation and variation inequalities for convolution powers _
, ergodic theory dynam .
systems 21 ( 2001 ) , no . 6 , 1809 - 1829 .
r. l. jones , a. seeger , and j. wright , _ strong variational and jump inequalities in harmonic analysis _ , trans .
, 360 ( 2008 ) , 6711 - 6742 .
r. l. jones , and g. wang , _ variation inequalities for the fejr and poisson kernels _ , trans .
, 356 ( 2004 ) , 4493 - 4518 . c. le merdy , and q. xu , _ maximal theorems and square functions for analytic operators on @xmath16-spaces _ , preprint 2010 , arxiv:1011.1360v1 .
d. lpingle , _ la variation dordre @xmath136 des semi - martingales ( french ) _ z. wahrscheinlichkeitstheorie und verw .
gebiete 36 ( 1976 ) , 295 - 316 .
lyubich , _ spectral localization , power boundedness and invariant subspaces under ritt s type condition _ , studia math .
134 ( 1999 ) , 153 - 167 .
meyer - nieberg , _ banach lattices _ , springer , berlin - heidelberg - newyork , 1991 .
b. nagy , and j. zemanek , _ a resolvent condition implying power boundedness _ , studia math .
134 ( 1999 ) , 143 - 151 .
o. nevanlinna , _ convergence of iterations for linear equations _ ,
birkhaser , basel , 1993 .
r. oberlin , a. seeger , t. tao , c. thiele , j. wright , _ a variation norm carleson theorem _ ,
preprint 2010 , arxiv:1011.1360v1 .
a. pazy , _ semigroups of linear operators and applications to partial differential equations _ , springer , 1983 .
v. peller , _ an analogue of j. von neumann s inequality for the space @xmath448 ( russian ) _ , dokl .
nauk sssr 231 ( 1976 ) , no . 3 , 539 - 542 . g. pisier , _ complex interpolation and regular operators between banach lattices _ , arch .
( basel ) 62 ( 1994 ) , no . 3 , 261 - 269
. g. pisier , and q. xu , _ the strong @xmath136-variation of martingale and orthogonal series _ , probab .
fields 77 ( 1988 ) , 497 - 514 .
e. m. stein , _ topics in harmonic analysis related to the littlewood - paley theory _
studies , princeton , university press , 1970 .
k. yosida , _ functional analysis _ , springer verlag , 1968 . | let @xmath0 be a positive contraction , with @xmath1 .
assume that @xmath2 is analytic , that is , there exists a constant @xmath3 such that @xmath4 for any integer @xmath5 .
let @xmath6 and let @xmath7 be the space of all complex sequences with a finite strong @xmath8-variation .
we show that for any @xmath9 , the sequence @xmath10(\lambda)\bigr)_{n\geq 0}$ ] belongs to @xmath7 for almost every @xmath11 , with an estimate @xmath12 .
if we remove the analyticity assumption , we obtain an estimate @xmath13 , where @xmath14 denotes the ergodic averages of @xmath2 .
we also obtain similar results for strongly continuous semigroups @xmath15 of positive contractions on @xmath16-spaces . _
2000 mathematics subject classification : 47a35 , 37a99 , 47b38 _ |
models of quasistatic elasto - plasticity date back to prandtl s model of an elastic - ideal plastic element , which can be thought of as a cascade connection of an ideal hook s spring and a coulomb dry friction element .
this simple model accounts for two important effects , saturation of stress with increasing deformation ( strain ) and hysteresis in the stress - strain relationship .
hysteresis is a manifestation of the fact that stress at a moment @xmath0 is not a single - valued function of the concurrent deformation value , but rather a function of state of the elasto - plastic material , which depends on the history of variations of the deformation prior to the instant @xmath0 .
two parameters of prandtl s model are the stiffness @xmath1 of the spring and the maximal spring force @xmath2 ( which equals the friction force in the sliding regime for quasistatic deformations ) . in order to account for complex relationship between deformation and stress in real materials ,
prandtl proposed to model the constitutive law of the material with a parallel connection of elastic - ideal plastic elements .
a similar idea was developed by ishlinskii who modeled individual fibers of wire ropes by prandtl s elements . in the prandtl - ishlinskii phenomenological model , a finite or infinite set of prandtl s elements ( characterized by different values @xmath3 of parameters of stiffness and maximal stress )
are all subject to the same deformation @xmath4 , and the total force ( or stress ) @xmath5 is proportional to the sum of all spring forces .
the operator that maps the time series @xmath4 of the deformation ( input ) to the time series of stress @xmath5 ( output ) , given a set of initial stresses of all the springs ( initial state ) , is known as the prandtl - ishlinskii ( pi ) operator in one - dimensional elasto - plasticity .
thanks to the set of good mathematical properties of this operator ( see , for example , @xcite ) , its equivalent counterparts have been used in several other disciplines including tribology ( maxwell - slip friction model ) , damage counting and fatigue estimation ( the rain flow counting method ) , and , more recently , modeling constitutive laws of smart materials such as piezo - electric and magnetostrictive materials and shape memory alloys .
one useful property , called the composition rule , is that a composition of pi operators is also a pi operator and , as a consequence , the inverse operator for a pi operator is another pi operator .
furthermore , a pi operator and its inverse admit an efficient analytic and numerical implementation .
this property , in particular , underpins the design of compensation - based control schemes for microactuators and sensors , which use smart materials for energy conversion .
another fact that facilitates modeling various constitutive laws with the pi operator , and is also central to this paper , is stated by the representation theorem , which allows one to determine whether a set of input - output data can be modeled by a pi operator and , moreover , equips one with an algorithm for identifying parameters of a pi operator from a simple measurement procedure .
the representation theorem states that if ( a ) the input - output relationship between deformation and stress is _ rate - independent _ that maps the time series @xmath4 of deformation to the time series @xmath5 of stress commutes with any increasing transformation @xmath6 of time , @xmath7 . ] ; ( b ) hysteresis loops corresponding to periodic inputs are _ closed _ ; and , ( c ) every hysteresis loop is _ centrally symmetric _ , then @xmath8 , where @xmath9 is a pi operator .
furthermore , if properties ( a)(c ) are satisfied , then starting from the initial state in which all the springs are relaxed , applying an increasing input ( deformation ) @xmath10 , and measuring the corresponding increasing output @xmath11 , one obtains the so called _ loading curve _ @xmath12 , which completely defines the pi operator @xmath9 on the class of all continuously varying inputs @xmath13\to\mathbb{r}$ ] via an explicit formula involving @xmath14 and a sequence of local running extremum values of @xmath10 ( see below ) .
it is worth mentioning that the same measurement of the loading curve is used in nonlinear elasticity to identify the non - hysteretic relationship between @xmath10 and @xmath15 by the nemytskii operator @xmath16 .
the main premise of the of the phenomenological pi model is that prandtl s elements do not interact .
the model of moreau addresses a much more general setting , in which an arbitrary spatial configuration of nodes connected by prandtl s elastic - ideal plastic springs deforms quasistatically ( the balance of forces at each node is zero at all times ) in response to either ( i ) external forcing applied at a selected set of nodes or ( ii ) controlled variation of the distance between nodes for a selected set of pairs of nodes , or ( iii ) simultaneous application of the two types of inputs above . in this setting , the relationship between the vector of stresses of individual springs and the input is complicated and is described by the differential inclusion ( or , equivalently , a variational inequality ) , which is known as the moreau _ sweeping process_. this process has a nice kinematic interpretation , in which variations of the input induce the motion and deformation of the convex domain @xmath17 of admissible stresses , and the set @xmath18 drags a point representing the vector of stresses according to a natural kinematic rule . here
, we consider an ( arbitrary ) configuration of prandtl s elastic - ideal plastic springs controlled by one scalar - valued input - the distance @xmath19 between two selected nodes , @xmath20 and @xmath21 .
the corresponding sweeping process belongs to a special class known as a ( multidimensional ) play operator with a unidirectional input .
even for this setting , the relationship between the input @xmath19 and the stress @xmath22 of any given spring can be quite complicated . in particular , hysteresis loops corresponding to a periodically varying @xmath23 can be non - closed , thus violating property ( b ) of the pi operator stated above . on the other hand , there are multiple examples of topologies of the spring configuration for which the relationship between @xmath23 and @xmath24 is a pi operator @xmath25 for every @xmath26 ( as well as the total reaction force at the node a , and at the node b , is related to @xmath23 via a pi operator ) .
in particular , such examples can be constructed using the composition property of pi operators . in this paper , we consider the following question : under which assumptions the moreau sweeping process driven by a scalar - valued input @xmath19 is equivalent to a pi operator ?
the answer will be given in geometric terms involving a vector - valued analog of the loading curve , which we define for the sweeping process .
we show that if this curve and the domain @xmath18 of admissible stresses satisfy simple geometric conditions , then the reaction force of the system of coupled prandtl s elements responds to the input @xmath19 exactly in the same way as the reaction force of a system of decoupled prandtl s elements of a certain effective pi model .
furthermore , @xmath27 for each spring of the moreau model , where @xmath28 is an appropriate projection of the loading curve of the moreau model in the space of stresses .
in other words , we perform a reduction of the more complex model of moreau to a simpler affective pi model when such reduction is possible .
the paper is organized as follows . in the next section ,
we present an outline of models of moreau and prandtl - ishlinskii . in section
[ main ] , the main reduction theorem is proved .
section [ discussion ] contains some discussion .
in particular , we show that any moreau model obtained by a small perturbation of a pi model satisfies the conditions of the main theorem and hence is reducible to to an effective pi model .
the paper ends with a summary of results and a few concluding remarks .
following the model of moreau , let us consider an arrangement of @xmath29 nodes , some of which are connected by elastic - ideal plastic springs , see fig .
[ fig1 ] . for simplicity
, we consider a one - dimensional arrangement ( see remark [ ? ? ] ) .
that is , all the nodes lie on a straight line ( the @xmath30-axis ) and the springs , as well as spring forces , are elongated along this line at all times .
the coordinate of node @xmath26 will be denoted @xmath31 .
it is assumed that there is a configuration @xmath32 called the zero configuration , in which the springs experience zero stress , and we consider deformations of the springs relative to this configuration .
the deformation of the spring connecting nodes @xmath26 and @xmath33 is denoted by @xmath34 : @xmath35 and the force ( stress ) of this spring is denoted by @xmath36 .
this arrangement of springs can be associated with a non - directed graph @xmath37 with the nodes associated with graph s vertices and springs associated with graph s edges . without loss of generality ,
this graph is assumed to be connected .
by @xmath38 , we denote the set of indices @xmath39 such that @xmath40 whenever there is a spring connecting the nodes @xmath26 and @xmath33 . here and henceforth ,
we agree that @xmath41 and @xmath42 for @xmath40 .
the general moreau sweeping process models a quasistatic response of the system of springs to external controls ( inputs ) .
there are two types of admissible controls : ( a ) an external force @xmath43 applied to a node @xmath26 ; and , ( b ) a _ moving affine constraint _ that prescribes a distance @xmath44 between a pair of nodes @xmath26 and @xmath33 at all times .
multiple moving constraints and external forces may be applied simultaneously at several pairs of nodes and nodes , respectively . however , in this paper , we consider a system with one control . to be specific ,
we choose to consider a system of springs controlled by one moving constraint .
without loss of generality , we assume that this constraint defines the distance between nodes @xmath45 and @xmath29 and that these nodes are not connected by a spring : @xmath46 where @xmath47 is a given function of @xmath48 satisfying @xmath49 and @xmath50 .
we note that moreau s theory equally applies to configurations , where a pair of nodes can be connected by multiple springs . in this case , @xmath37 is a multigraph . in the models of moreau and prandtl - ishlinskii , the stress @xmath51 and deformation @xmath52 of each spring
are related by prandtl s nonlinear hysteretic constitutive law that combines an ideally elastic spring with a dry friction element , see fig .
[ fig2 ] . according to this constitutive law , @xmath53 where @xmath54 is the young s modulus of the spring in the elastic domain , and the internal variable @xmath55 called _
elastic deformation _ is related to the deformation @xmath34 by the so - called one - dimensional _ stop _ operator @xmath56 : @xmath57.\ ] ] here @xmath58 is the maximal value of the elastic deformation , which defines the maximal magnitude @xmath59 of stress for the spring , _
latexmath:[\[\label{adm } for piecewise monotone inputs @xmath34 , relationship between the time series of the deformation and the elastic deformation is given by the explicit formula@xmath61,\ ] ] where @xmath62 is a partition of the time domain into intervals of monotonicity of the input @xmath34 and @xmath63 see fig .
[ fig3 ] . furthermore
, as shown in @xcite , operator admits a continuous extension from the dense set of piecewise monotone inputs to the whole space in each of the spaces @xmath64 , @xmath65 , and @xmath66 . in the space @xmath65
, operator can be alternatively defined by the variational inequalities latexmath:[\[\label{var } which should be satisfied for a.e .
this law postulates the ideal elastic response of the spring as long as the stress remains in the range @xmath68 .
if the maximal admissible value of the stress is reached but the deformation continues to increase in absolute value , the stress remains at its maximal value , _
i.e. _ @xmath69 for @xmath70 and @xmath71 for @xmath72 , that is the spring _
yields_. everywhere below , we assume that all the deformations and forces are initially zero , that is @xmath73 , @xmath74 for all @xmath40 at the initial moment @xmath75 .
-10truemm of the ideal spring is called elastic deformation ; @xmath34 is the elongation of the distance between the nodes @xmath26 and @xmath33 ( with respect to the distance @xmath76 at zero configuration ) .
dry friction between the box @xmath21 and the surface @xmath77 produces the friction force , which is opposite to the ideal spring force at all times ( the quasistatic model ) .
the magnitude of the friction force is limited by the maximal value @xmath78 .
therefore , the ideal spring deforms but the box does not move with respect to the surface @xmath77 as long as @xmath79 .
when @xmath80 , the deformation @xmath55 of the ideal spring and the force remain constant , while the box moves with respect to the surface in the direction of the spring force . ] -10truemm of a nonlinear spring with @xmath81 and @xmath82 .
the time series of the deformation @xmath34 and the elastic deformation @xmath55 are related by the stop operator @xmath83 $ ] .
slanted lines have the slope @xmath45 ; horizontal lines are @xmath84 . ]
-10truemm define the vector @xmath10 of deformations @xmath34 ( where @xmath40 ) of all the springs and the vector @xmath15 of their corresponding stresses @xmath36 .
similarly , @xmath85 denotes the vector of elastic deformations @xmath55 , and we will also use the vector @xmath86 of _ plastic deformations _ @xmath87 .
the model of moreau defines dynamics in the configuration space @xmath88 , where @xmath89 is the total number of springs . in what follows
, the set of geometric constraints with the additional affine moving constraint will be expressed as the inclusion @xmath90 with an appropriate choice of the subspace @xmath91 of the configuration space @xmath92 and a vector @xmath93 .
we remark that geometric constraints in parametric form can be equivalently expressed in terms of the cycles @xmath94 of the graph @xmath37 with the added edge @xmath95 .
namely , the sum of deformations along the edges @xmath96 of every cycle must be zero : @xmath97 the main assumption of moreau model is that deformations are quasistatic .
therefore , the balance of forces at each node is zero at any moment : @xmath98 where the summation is over @xmath33 . the balance of forces at each of the nodes @xmath45 and @xmath29 connected by the moving constraint includes the ( unknown ) reaction @xmath99 of this constraint : @xmath100 relations follow from .
let us briefly summarize the results of moreau for the particular case of a system of springs with one control that we are considering .
for a closed convex set @xmath101 , we denote by @xmath102 the external normal cone to the set @xmath18 at a point @xmath103 : @xmath104 where the standard scalar product @xmath105 is used .
let us introduce the parallelepiped @xmath106 of admissible stresses defined by . as was shown by j. j. moreau , system ,
is equivalent to the system @xmath107 with time dependent variables @xmath108 .
here @xmath20 is the diagonal positive matrix with the entries @xmath109 such that is equivalent to ; recall that @xmath110 , hence is nothing else as the relationship , which is equivalent to the set of geometric constraints with the additional moving affine constraing ; the inclusion @xmath111 , where @xmath112 denotes the orthogonal complement of @xmath91 in the space @xmath92 , is equivalent to equations that state the balance of forces ; and , the differential inclusion @xmath113 , where dot denotes differentiation with respect to time and @xmath114 , is equivalent to the variational inequality that expresses the constitutive law of prandtl s spring .
following j. j. moreau , it is convenient to use the rescaled variables @xmath115 where the diagonal matrix @xmath116 is the positive square root of @xmath20 . introducing the orthogonal subspaces @xmath117 and the scaled parallelepiped @xmath118 and noticing that @xmath119 we see that equations are equivalent to the system @xmath120 now , introducing the orthogonal projection @xmath121 of the vector @xmath122 on the subspace @xmath123 , and using the new variables @xmath124 one can rewrite this system as @xmath125 finally , combining inclusion with @xmath126 and using the identity @xmath127 , we arrive at the differential inclusion @xmath128 which is known as the _ moreau sweeping process _ with the characteristic set ( input ) @xmath129 .
a few remarks are in order .
first , @xmath130 is a centrally symmetric convex polytope .
its central symmetry is important for the following discussion .
second , the characteristic set @xmath129 of the moreau process is obtained as time - dependent shift of this polytope .
the moreau process with a characteristic set of the form @xmath131 , where @xmath132 is a single - valued function @xmath133 and @xmath134 is a convex set , is known as the multi - dimensional play operator with input @xmath132 @xcite . in our case ,
the input @xmath135 has a fixed direction @xmath136 , _
i.e. _ we consider the moreau process of the play type , which in effect has a one - dimensional input .
in general , for a system with multiple inputs ( moving constraints and external forces ) , the shape of the characteristic set @xmath137 may change with time , _
i.e. _ @xmath138 is a more general regular set - valued convex - valued function .
is equivalent to the differential inclusion @xmath139 known as the multi - dimensional stop operator with the input @xmath140 .
this differential inclusion coupled with the initial condition @xmath141 has a unique solution @xmath142 for any input @xmath143 ; regularity properties of the solution operator that maps @xmath23 to @xmath144 in spaces @xmath65 , @xmath64 , and @xmath66 are well understood .
furthermore , system coupled with the initial conditions @xmath145 ( we also assume @xmath49 ) has a solution @xmath146 .
while forces of the springs ( the component @xmath147 ) are defined uniquely for a given input @xmath148 , simple examples show that the solution @xmath146 may be non - unique , that is deformations are not necessarily uniquely defined .
it should be noted that examples with multiple solutions are non - generic .
however , the author is not aware of results that would establish uniqueness of deformations under genericity assumptions .
consider the system of springs shown in fig .
the configuration space @xmath149 of this system consists of vectors @xmath150 ( where prime denotes the transposition ) , which satisfy the following set of geometric constraints : @xmath151 and @xmath152 ( cf . ) .
we assume that the system is subject to the additional moving affine constraint @xmath153 .
combining all the constraints , we obtain @xmath154 where @xmath155 this equation is equivalent to relation with @xmath156 the balance of forces at the nodes 2 , 3 , 4 reads @xmath157 or , equivalently , @xmath158 with @xmath159 .
the matrix @xmath20 has the form @xmath160 in this example , the set @xmath130 is is a 6-dimensional centrally symmetric convex polytope .
the prandtl - ishlinskii operator @xmath161 has scalar - valued inputs and outputs and is obtained as a weighted sum of a finite number of one - dimensional stop operators : @xmath162:=\sum_{n=1}^k \bar a_{n } \mathcal{s}_{\bar\rho_n}[g],\ ] ] where @xmath148 and @xmath163 are continuous input and output , respectively ; @xmath164 is the young s modulus of the @xmath165-th spring in the elastic domain ; @xmath166 is the maximal stress of the spring \,{\rm d\rho}$ ] .
however , in this work , only the finite ( discretized ) model is considered . ] ; and , without loss of generality we can assume that @xmath167 this operator is associated with the connection of prandtl s springs shown in fig .
[ fig4](a ) , i.e. @xmath37 consists of two vertices ( nodes ) @xmath45 and @xmath168 connected by a multiedge .
formula relates the elongation @xmath169 of the distance between the nodes ( relative to the zero configuration ) with the reaction force @xmath99 applied at node @xmath45 ; the reaction at node @xmath168 equals @xmath170 . -8truemm
\(a ) 5.5truecm ( b ) it is easy to see that the same relationship between the elongation @xmath23 of the moving constraint and the reaction force @xmath99 is valid for any simple connected graph @xmath37 of order @xmath29 in which every vertex @xmath171 has degree @xmath168 and the moving affine constraint is applied to the distance between the nodes @xmath45 and @xmath29 as in .
such connections will be called _ linear _ , see fig . [ fig5 ] . for linear connections ,
the parameters @xmath172 , @xmath164 , @xmath173 in are `` effective '' quantities that can be related to the parameters @xmath109 , @xmath58 of the springs ( young s moduli and yielding thresholds ) via simple formulas .
indeed , one can show that the stress @xmath174 $ ] of each spring is related to the controlled distance @xmath148 between the nodes @xmath45 and @xmath29 by another stop operator @xmath175 $ ] with appropriate effective @xmath176 and @xmath177 .
for example , for the connection shown in fig .
[ fig5 ] , @xmath178 @xmath179 and @xmath180 @xmath181 where @xmath182 , @xmath183 .
equations now imply that @xmath23 is mapped to the reaction force @xmath99 of the constraint by the prandtl - ishlinskii operator @xmath184(t)=\tilde a_{38 } { \mathcal s}_{\tilde\rho_{38}}[g ] + \tilde a_{48 } { \mathcal s}_{\tilde\rho_{48}}[g]+\tilde a_{78 } { \mathcal s}_{\tilde\rho_{78}}[g]$ ] .
the general linear connection is considered in section 4 .
-10truemm an important characterization of prandtl - ishlinskii operator is the so - called _ loading curve _ @xmath185 defined as the response to the linear input : @xmath186\ \quad { \rm with } \
\quad g_{id}(t):=t , \ \
t\ge 0\ ] ] ( where , in accordance with our agreement , we assume that @xmath187 for all the stop operators @xmath188 $ ] in ) . clearly , @xmath14 is a piecewise linear function , which satisfies @xmath189 and @xmath190 for sufficiently large @xmath191 .
equivalently , @xmath192 the loading curve uniquely identifies operator @xcite , therefore we will write @xmath193 when needed . furthermore , for any @xmath194 with @xmath49 , the output @xmath195 defined by can be expressed explicitly in terms of the function @xmath14 as follows : @xmath196 where the so - called _ running main extremum values _
@xmath197 of @xmath23 are defined by the relationships @xmath198\ } ; \\ \tau_1:=\max\{\tau\in [ 0,t]:\ |g(\tau)|=g_0\};\quad g_1:=g(\tau_1 ) \end{array}\ ] ] and , if @xmath199 , then @xmath200\ } ; \quad \tau_{2i}:=\max\{\tau\in [ 0,t]:\ g(\tau)=g_{2i}\};\\ g_{2i+1}(t):=\min \{g(\tau):\ t\in[\tau_{2i } , t]\ } ; \quad \tau_{2i+1}:=\max\{\tau\in[0,t]:\ g(\tau)=g_{2i+1}\ } ; \end{array}\ ] ] if @xmath201 , then @xmath202\ } ; \quad \tau_{2i}:=\max\{\tau\in [ 0,t]:\ g(\tau)=g_{2i}\};\\ g_{2i+1}(t):=\max \{g(\tau):\ t\in[\tau_{2i } , t]\ } ; \quad \tau_{2i+1}:=\max\{\tau\in[0,t]:\ g(\tau)=g_{2i+1}\ } \end{array } \end{array}\ ] ] for @xmath203 by definition , a linear combination of operators with non - negative coefficients is also a preisach - ishlinskii operator , @xmath204 .
remarkably , the class of the prandtl - ishlinskii operators is also closed with respect to composition due to the identity @xmath205 @xcite for the stop and play operators @xcite . ] . using these properties , one can associate a prandtl - ishlinskii operator with more complex topologies @xmath37 than a linear connection such as in fig .
in particular , for a multigraph with vertices @xmath206 , let us define an elementary graph operation that replaces a multiedge with a simple edge ; an elementary operation that replaces a vertex @xmath207 of degree @xmath168 and the two edges emanating from it with one edge ; and , an elementary operation that eliminates a vertex @xmath207 of degree @xmath45 and the edge emanating from it from the graph .
we call a connected graph @xmath208 of order @xmath29 _ reducible _ if a finite number of such operations can reduce it to the trivial graph that consists of two vertices @xmath45 and @xmath29 and an edge between them , see fig .
[ fig4](b ) .
one can show that if a connection of springs with a reducible graph @xmath208 is driven by one moving constraint , then the elongation @xmath23 of the constraint is mapped to the stress of every spring by an ( effective ) prandtl - ishlinskii operator , @xmath209 $ ] and the reaction force is given by @xcite .
consider a graph @xmath37 with edges @xmath40 and the corresponding system of @xmath89 connected prandtl s springs with parameters @xmath109 ( stiffness ) and @xmath210 ( maximal stress ) , see sections [ 1.1 ] , [ 1.2 ] .
assume that the system is driven by one moving affine constraint and the set of all the constraints is described by inclusion .
denote by @xmath211\to \pi \cap v$ ] the solution of the differential inclusion with the input @xmath212 and the initial value @xmath213 .
the trajectory @xmath214 of this solution is a polyline @xmath215 with one end @xmath216 at the origin .
each link of this polyline belongs to a different face of the polytope @xmath130 , that is @xmath217 for @xmath218 , where @xmath219 are ( closed ) faces of the polytope , which are all different , and @xmath220 .
denote by @xmath221 the minimal affine subspace of @xmath222 which contains the face @xmath219 and by @xmath223 the interior of the set @xmath219 in @xmath221 .
the parametrization @xmath224 of the polyline @xmath214 defines the partition @xmath225 of the interval @xmath226 $ ] by preimages of the points @xmath227 : @xmath228 2truemm * theorem 1 . *
_ suppose that @xmath229 and @xmath230 suppose that the parallelepiped @xmath231 belongs to the polytope @xmath232 .
assume that the map @xmath233\to\gamma$ ] is invertible .
then , for every input @xmath19 satisfying @xmath49 and @xmath234 for all @xmath48 , the stress of each spring relates to the variable @xmath23 via the preisach operator @xmath235(t),\qquad ( ij)\in m , \sigma_{ij}(t ) = \mathcal p_{\phi_{ij}}[g](t),\qquad ( ij)\in m,\ ] ] with the loading curve @xmath236 where @xmath237 is the @xmath39-th component of the vector - valued function @xmath233\to \pi$ ] . _
2truemm as we establish below , under the conditions of this theorem , the moreau sweeping process behaves as an analog of the prandtl - ishlinskii operator with vector - valued outputs and the vector - valued @xmath233\to \pi\cap v$ ] acts as a counterpart of the loading curve for this operator .
more precisely , the solution of with the zero initial condition @xmath141 is given by the counterpart of formula : @xmath238 where @xmath197 are running main extremum values of the scalar - valued input @xmath23 ; and , we extend the function @xmath239 to the interval @xmath240 $ ] by setting @xmath241.\ ] ] formulas , immediately follow from and the relationship @xmath242 .
given an input @xmath243;\mathbb{r})$ ] with @xmath244 , we need to show that the function defined by satisfies @xmath245 for all @xmath246 $ ] and the inclusion for almost every @xmath246 $ ] .
suppose for definiteness that @xmath247 .
since @xmath248 with @xmath249 defined in ( recall that @xmath250 ) , formula with @xmath247 implies the relation @xmath251 with @xmath252 where @xmath253 and @xmath254 . from the definition of the sequence @xmath255 given below eq .
, it follows that @xmath256 where only a finite number of differences @xmath257 are nonzero .
therefore , implies @xmath258 and from it follows that @xmath259 belongs to the parallelepiped @xmath260 , which by assumption belongs to @xmath232 .
this proves @xmath245 for the function in the case @xmath247 .
as the function @xmath239 in is odd , it is easy to see that the same inclusion is valid for @xmath261 .
it remains to prove .
as the solution operator of the moreau sweeping process subject to the zero initial condition @xmath141 is continuous in the space @xmath262;v)$ ] and so is the prandtl - ishlinskii operator in the space @xmath262;\mathbb{r})$ ] , it suffices to show that function satisfies for piecewise linear continuous inputs @xmath243;\mathbb{r})$ ] .
furthermore , given such an input , it suffices to establish that if the inclusion is valid for a.e .
@xmath0 from an interval @xmath263\subset [ 0,t)$ ] , then there is a @xmath264 such that this inclusion is also true almost everywhere in @xmath265 $ ] . in order to establish is a solution of on an initial small interval @xmath263 $ ] , we recall the _ rate - independence _ property of the moreau process and the prandtl - ishlinskii model , which means that the input - output operator of each model commutes with increasing transformations of time @xcite .
let us take a sufficiently small @xmath266 so that @xmath23 is linear on @xmath263 $ ] and the trajectory @xmath267 , which satisfies @xmath268 , remains in the interior of the polytope @xmath232 for all @xmath269 $ ] .
then , is equivalent to @xmath270 , which agrees with the rate - independence property of the moreau process for an increasing @xmath23 .
since the function @xmath239 and the solution operator of the moreau process are odd , defines a solution of on a sufficiently small interval @xmath263 $ ] for a decreasing @xmath23 too . assuming that satisfies for @xmath271 $ ] ,
let @xmath272 be the sequence of the main extremum values of @xmath23 at the moment @xmath273 .
due to the central symmetry of @xmath274 , we can assume without loss generality that @xmath275 .
since @xmath23 is piecewise linear , from the definition of @xmath276 it follows that there is an integer @xmath93 such that @xmath277 for all @xmath278 and , depending on the parity of @xmath93 , either @xmath279 with @xmath280 , or @xmath281 with @xmath282 .
the argument for both cases is similar , and we assume for definiteness that holds . since @xmath23 is piecewise linear , @xmath283 on some interval @xmath284 . using the rate - independence of the moreau process , it suffices to consider the cases @xmath285 , @xmath286 , and @xmath287 for @xmath288 .
the latter case is trivial as @xmath289 implies that @xmath290 and @xmath291 in for all @xmath292 and all @xmath293 $ ] .
the cases when @xmath23 increases ( @xmath285 ) and @xmath23 decreases ( @xmath286 ) on @xmath294 will be considered separately .
let us introduce some notation .
suppose that the polytope @xmath232 is the intersection of subspaces @xmath295 where each hyperplane @xmath296 contains one facet of @xmath232 and @xmath297 is the unit outward normal vector to this facet .
condition ensures that the face @xmath219 of @xmath232 belongs to the intersection of @xmath292 hyperplanes @xmath298 , hence we can number these hyperplanes in such a way that @xmath299 recall that @xmath300 the interior of the face @xmath219 in the affine subspace @xmath221 ( with the agreement that @xmath301 ) . with this notation , the outward normal cone to @xmath232 on @xmath300 coincides with the positive linear span of the vectors @xmath302 : @xmath303 in particular , since relations ensure that @xmath304 for @xmath305 , the inclusion for @xmath306 is equivalent to @xmath307 with @xmath308 . combining these relations with , we obtain @xmath309 one can also see that the inclusion for @xmath310 is equivalent to the equality @xmath311 where @xmath312 is the orthogonal projection of the vector @xmath313 onto the subspace @xmath314 , which is parallel to @xmath221 in @xmath222 . note that since @xmath315 , @xmath316 now , suppose that holds at the moment @xmath273 and @xmath286 for @xmath288 .
if @xmath264 is sufficiently small , then according to the definition of the main extremum values of @xmath23 , one has @xmath317 with @xmath318\ ] ] ( cf . ) . therefore , comparing formula at the moments @xmath273 and @xmath319 ( note that the sum in contains nonzero terms for @xmath320 at the moment @xmath273 and for @xmath321 at the moment @xmath0 ) , we obtain @xmath322 and using , and @xmath323 , we arrive at the relation @xmath324,\ ] ] which agrees with .
finally , suppose that @xmath285 for @xmath288 and , again , relations are valid at the moment @xmath273 . in this case
, assuming that @xmath264 is sufficiently small , @xmath325 with @xmath326.\ ] ] therefore , implies @xmath327 hence @xmath328 now , note that there is a @xmath292 , @xmath329 , such that @xmath330 and since @xmath331 , one also has @xmath332 for a sufficiently small @xmath333 . combining these relations with and
, we see that @xmath334 also , relations and imply @xmath335 where @xmath336 for @xmath337 and @xmath338 .
hence , for @xmath339 , becomes @xmath340 ( note that all terms with @xmath341 in the sum in are zero ) and implies @xmath342 since @xmath343 , it follows that @xmath344 and hence @xmath345 combining this relation with and , we conclude that with @xmath285 holds for @xmath346 .
this completes the proof .
a few remarks are in order .
* 1 . * under the assumptions theorem 1 , the complex connection of springs defined by the graph @xmath37 responds to arbitrary variations @xmath19 of the given moving affine constraint in the same way as a simple parallel connection of springs described by an effective prandtl - ishlinskii opearator .
indeed , the stress of each spring of the complex system is related to @xmath19 via the prandtl - ishlinskii operator and the total force applied to the system at node 1 relates to the displacement @xmath23 via the prandtl - ishlinskii operator ( cf . )
@xmath347(t).\ ] ] 2truemm * 2 . *
according to formula , under the conditions of theorem 1 , hysteresis loops of the moreau model and the loops of the prandtl - ishlinskii model have similar properties and can be constructed by the same simple manipulations with the graph of the loading curve , see @xcite for details .
the only difference is that this graph is two - dimensional for the prandtl - ishlinskii model , while the loading curve @xmath214 for the moreau model in theorem 1 is multi - dimensional .
2truemm * 3 . *
[ fig6 ] illustrates the role of conditions @xmath348 and @xmath349 of theorem 1 .
[ fig6](a ) violates condition because @xmath350 . in fig .
6(b ) , @xmath351 , hence conditions are violated . in both cases , we see that the trajectory of the moreau process does not have the shape prescribed by formula , which generalizes the prandtl - ishlinskii operator . .
the vector @xmath313 used in the definition of the moreau process points in the direction of the vector @xmath352 .
( a ) the prism @xmath260 does not belong to @xmath232 .
the polyline @xmath353 that contains the the polyline @xmath354 represents the trajectory of a solution @xmath144 to for the input @xmath355 where @xmath23 increases from zero to a maximum value @xmath356 and then decreases to the minimum value @xmath357 . formula defines a different polyline @xmath358 for the same input .
( b ) conditions are violated because @xmath351 .
the polylines @xmath359 and @xmath360 the trajectory of the inclusion and curve prescribed by formula in response to the same input as in panel ( a ) .
, title="fig : " ] . the vector @xmath313 used in the definition of the moreau process points in the direction of the vector @xmath352 .
( a ) the prism @xmath260 does not belong to @xmath232 .
the polyline @xmath353 that contains the the polyline @xmath354 represents the trajectory of a solution @xmath144 to for the input @xmath355 where @xmath23 increases from zero to a maximum value @xmath356 and then decreases to the minimum value @xmath357 .
formula defines a different polyline @xmath358 for the same input .
( b ) conditions are violated because @xmath351 .
the polylines @xmath359 and @xmath360 the trajectory of the inclusion and curve prescribed by formula in response to the same input as in panel ( a ) .
, title="fig : " ] -8truemm 1truecm ( a ) 5.5truecm ( b ) 2truemm * 4 . * the assumption that the map @xmath233\to\gamma$ ] is invertible is made for simplicity .
it is straightforward to extend the theorem to the case when it is not satisfied .
2truemm * 5 .
* examples in which the system of springs shown in fig . [ fig1 ] can not be reduced to a prandtl - ishlinskii operator can be found in @xcite . of a general _ linear _ connection of @xmath361 springs ( thick lines ) .
thin dashed lines represent possible additional edges as introduced in theorem 2 . ]
-11truemm * 6 .
* examples with complex topology that satisfy the conditions of theorem 1 can be created by a perturbation of simple systems considered in section [ sect ] . in particular , let us consider a _ linear _ connection of springs with the corresponding graph @xmath37 shown in fig .
[ fig7 ] ( the graph shown in fig .
[ fig5 ] is an example of such a graph ) . here
@xmath362 and the list of all edges @xmath39 with @xmath363 is as follows : node @xmath45 is connected with nodes @xmath364 ; node @xmath29 is connected with nodes @xmath365 ; and , each node @xmath26 with @xmath366 is connected with node @xmath367 for @xmath368 .
the total number of edges equals @xmath369 .
denote @xmath370 @xmath371 with @xmath372 ( see example in section 2.6 ) and set @xmath373 2truemm * theorem 2 .
* _ consider a linear connection of @xmath361 springs ( see fig .
[ fig7 ] ) .
assume that @xmath374 @xmath375 for @xmath376 .
let us extend this system with any set of additional connections of nodes by springs with thresholds @xmath210 and young s moduli @xmath109 .
then , given any @xmath377 there is a @xmath378 such that if for all the added connections @xmath379 and simultaneously @xmath380 , i.e. the young s moduli @xmath109 of the added springs are sufficiently small , then the extended system of springs satisfies all the conditions of theorem 1 . _
2truemm 2truemm _ proof of theorem 2 .
_ first , we consider the linear connection of @xmath361 springs . for this connection ,
the set of geometric and moving affine constraints has the form @xmath381 therefore , the subspace @xmath382 of @xmath383 is defined by the relations @xmath384 its orthogonal complement @xmath385 is given by @xmath386 and the components @xmath387 of the vector @xmath93 in have the form @xmath388 therefore , the orthogonal subspaces @xmath389 and @xmath390 ( see ) are defined by the systems @xmath391 @xmath392 with @xmath372 , respectively . from these relations
, it follows that the components of vector are given by @xmath393 and the polytope @xmath232 is an @xmath77-dimensional parallelepiped defined by relations and @xmath394 where @xmath395 , @xmath396 are defined in , . using , , without loss of generality , we can assume that @xmath397 and @xmath398 for each @xmath372 , i.e. @xmath399 ( because the edges can always be labeled so as to ensure these relationships ) .
now , from , and , it is easy to derive explicit expressions for the components of vectors @xmath400 and coordinates of points @xmath401 : @xmath402
@xmath403 with @xmath372 . these explicit formulas ensure that : * the point @xmath404 belongs to the open face @xmath405 of the parallelepiped @xmath274 , i.e. @xmath406 where @xmath407 denotes the facet @xmath408 of @xmath274 for @xmath409 .
* @xmath410 and conditions , are satisfied . * the subspace @xmath222 and the hyperplanes @xmath411 which are parallel to the facets @xmath407 of @xmath274 , are in general linear position in the sense that @xmath412 this implies @xmath413 in the notation of the previous section . *
@xmath414 for @xmath415 ( where @xmath416 ) .
the next step is to show that the same properties ( i)-(iv ) are satisfied for the extended system of @xmath417 springs . to this end
, we first consider a simple embedding of the geometric picture considered above into the space @xmath418 by identifying each vector @xmath419 with components @xmath420 , @xmath40 , with the vector @xmath421 with the components @xmath422 ( recall that @xmath37 is a subgraph of @xmath423 , hence @xmath424 ) . with this identification
, @xmath92 becomes a subspace @xmath425 and we denote by @xmath426 its orthogonal complement with respect to the scalar product @xmath427 ( cf . ) , i.e. @xmath428 .
also , define the sets @xmath429 and the diagonal @xmath430 matrix @xmath431 by @xmath432 then , it is easy to see that all statements ( i1)-(i4 ) , in which we replace the space @xmath92 with @xmath433 , the subspaces @xmath434 with @xmath435 , the parallelepipeds @xmath274 , @xmath232 and their faces @xmath407 , @xmath436 with the polyhedra @xmath437 , @xmath438 and their faces @xmath439 , @xmath440 , and the matrix @xmath20 with the matrix @xmath431 , remain valid .
that is , * @xmath441 for @xmath409 . * the faces @xmath442 of @xmath443 satisfy conditions , . * the subspace @xmath444 and the hyperplanes @xmath445 , @xmath446 are in general linear position . * the vectors @xmath447 satisfy @xmath448 for @xmath415 . finally , consider the moreaux process corresponding to the system of @xmath449 springs , in which the springs with @xmath450 have small young s moduli @xmath109 . for this system , we denote by @xmath451 , @xmath452 and @xmath453 the counterparts of the vector @xmath93 , the subspaces @xmath454 and the matrix @xmath20 ( cf
. section 2.4 ) .
furthermore , we denote by @xmath455 , @xmath456 the parallelepiped @xmath457 and its facets @xmath458 note that the subspace @xmath459 is a small perturbation of the subspace @xmath222 in the sense that the intersections of @xmath222 and @xmath459 with any ball centered at the origin can be made arbitrarily close by making the coefficients @xmath109 , @xmath460 , sufficiently small . similarly , the sets @xmath455 , @xmath456 are small perturbations of @xmath274 , @xmath407 , respectively . also , the matrix @xmath453 and hence the vector @xmath461 are small perturbations of the matrix @xmath431 and the vector @xmath462 , respectively , because @xmath463 for @xmath40 .
therefore , properties ( j3 ) , ( j4 ) imply that the subspace @xmath459 and the hyperplanes @xmath464 , @xmath465 are in general linear position ; each subspace @xmath466 is a small perturbation of the subspace @xmath467 ; and , each vector @xmath468 is a small perturbation of the vector @xmath469 , hence @xmath470 .
now , since @xmath471 is the intersection point of the ray @xmath472 with the parallelepiped @xmath473 for every @xmath409 , property ( j1 ) implies that the relation @xmath474 uniquely defines a point @xmath475 whenever a point @xmath476 is sufficiently close to @xmath477 and the vector @xmath478 is sufficiently close to @xmath479 .
therefore , for sufficiently small @xmath109 , @xmath460 , the points @xmath480 satisfy @xmath481 with @xmath482 , and the polyline @xmath483 is a small perturbation of the polyline @xmath484 . since @xmath485 for all @xmath33
, we also see that the vertex @xmath486 of @xmath487 belongs to the open face @xmath488 of the polytope @xmath489 and the faces @xmath490 satisfy conditions , .
it remains to show that the set @xmath491 ( cf .
) belongs to @xmath455 . to this end , denote by @xmath55 the standard basis of vectors @xmath492 in @xmath493 : @xmath494 for @xmath495 , @xmath372 , consider the estimate @xmath496 with @xmath497 .
here @xmath498 , hence @xmath499 for @xmath500 and therefore @xmath501 for @xmath502 and since @xmath478 is a small perturbation of the vector @xmath479 , @xmath503 combining this estimate with the second relation in results in the relation @xmath504 for @xmath505 , hence is equivalent to @xmath506 where the last equality follows from @xmath507 and .
now , consider @xmath508 , i.e. @xmath509 for some @xmath368 , @xmath510 .
formulas , imply that @xmath511 for all @xmath512 , @xmath513 and @xmath514 for @xmath515 .
therefore , taking into account that @xmath516 and arguing exactly in the same way as above , we see that the unperturbed polyline @xmath214 satisfies @xmath517 furthermore , formulas , and @xmath516 imply @xmath518 hence @xmath519 where , due to @xmath507 , @xmath520 therefore , implies @xmath521 since @xmath522 is a small perturbation of @xmath214 , a similar estimate is true for @xmath487 : @xmath523 , @xmath524 is small . combining , and , we obtain @xmath525 .
this completes the proof of theorem 2 .
we considered a set of prandtl s elastic - ideal plastic elements , which are arranged into a network and deform quasistatically when the distance between two nodes is varied according to a given law ( input ) .
as in the general setting of the model of moreau , no _ a priori _ constraint was imposed on the topology of the network .
we defined the loading curve @xmath14 for the corresponding sweeping process as a graph of the solution corresponding to the zero initial condition and an increasing input .
it was shown that if this curve satisfies simple geometric conditions , then the structure of hysteresis loops of the model in the space of stresses extended by one dimension representing the input is similar to the structure of hysteresis loops of the pi model .
furthermore , the relationship between the input and the varying stress of each prandtl s element in the network is given by a pi operator @xmath25 , and the loading curve @xmath28 of this operator is the corresponding projection of the loading curve @xmath14 of the sweeping process .
the question of how the geometric conditions of the main theorem can be related to the topology and parameters of the network in general remains open .
however , we showed that these conditions are satisfied for any moreau network obtained by a small perturbation of a pi model .
in other words , they are satisfied for any network with sufficiently small coupling .
this is in line with the results from @xcite obtained by a different approach based on the composition property of pi operators . in this work ,
we considered networks of springs aligned along a straight line and used the distance between two nodes a and b as the input .
however , the results can be easily generalized to networks of prandtl s elements connecting nodes in a two- or three - dimensional space .
it would be interesting to consider other types of scalar - valued inputs , for example , the force applied at the node a. in this case , the input @xmath137 of the sweeping process is a set with changing size and shape and hence is more complicated than the inputs we considered
. this will be the subject of future work .
the author thanks ivan gudoshnikov for multiple stimulating discussions of the moreau model .
p krejci , j p okane , a pokrovskii , d rachinskii , properties of solutions to a class of differential models incorporating preisach hysteresis operator , physica d : nonlinear phenomena 241 ( 22 ) , 2012 , 2010 - 2028 | the sweeping process was proposed by j. j. moreau as a general mathematical formalism for quasistatic processes in elastoplastic bodies .
this formalism deals with connected prandtl s elastic - ideal plastic springs , which can form a system with an arbitrarily complex topology .
the model describes the complex relationship between stresses and elongations of the springs . on the other hand ,
the prandtl - ishlinskii model assumes a very simple connection of springs .
this model results in an input - output operator , which has many good mathematical properties .
it turns out that the sweeping processes can be reducible to the prandtl - ishlinskii operator even if the topology of the system of springs is complex . in this work
, we analyze the conditions for such reducibility . * on geometric conditions for reduction of the moreau sweeping process to the prandtl - ishlinskii operator * d. rachinskii |
the main purpose of this paper is to introduce , motivate , and conduct a detailed study of , two new ` dimension spectra ' motivated by the assouad and lower dimensions . roughly speaking
, the assouad dimension of a metric space is the smallest exponent @xmath0 such that , for any pair of scales @xmath1 , any ball of radius @xmath2 may be covered by @xmath3 balls of radius @xmath4 .
this provides a very important , but very coarse , piece of geometric information about the set .
it quantifiably identifies the ` thickest ' part of the space , thus making it a vital concept in contexts where extremal behaviour is important , such as embedding theory .
what we propose is a programme designed to obtain more information : a _ spectrum _ of exponents , rather than a single one . given @xmath5
, we find the smallest exponent @xmath8 such that , for any pair of scales @xmath1 with _ fixed _
relationship @xmath9 , any ball of radius @xmath2 may be covered by @xmath10 balls of radius @xmath4 .
we also consider the corresponding spectrum derived from the _ lower _ dimension , which acts as a dual to the assouad dimension . instead of finding the ` thickest part ' of a set
, the lower dimension seeks out the ` thinnest ' . for completeness
, we also consider a third spectrum , based on the modified lower dimension , which is a simple modification of the lower dimension designed to be monotonic .
we begin by considering how these spectra behave as functions of @xmath7 for arbitrary sets .
some of the notable results we obtain in this direction include : 1 .
there are non - trivial ( and sharp ) bounds on the spectra in terms of familiar dimensions , see propositions [ bb ] and [ bbl ] , and also corollaries [ bdconverge ] and [ bd0 ] .
the spectra are continuous in @xmath7 , see corollaries [ con ] and [ con2 ] .
3 . the assouad spectrum interpolates between upper box dimension and assouad dimension . in particular , as @xmath11 the spectrum always approaches the upper box dimension , and as @xmath12 the spectrum always approaches its maximal value , which is often the assouad dimension . see corollary [ bdconverge ] , corollary [ staya ] , and proposition [ mono ] .
4 . the spectra are often , but not necessarily , monotonic , see proposition [ mono ] and sections [ non - mono ] and [ non - mono2 ] . 5 .
the spectra have good distortion properties under bi - hlder functions , which is in contrast to the assouad dimension , see proposition [ holderthm ] .
the assouad and lower spectra are measureable , and of baire class 2 , when viewed as functions on the set of compact subsets of a metric space , see theorems [ meas1 ] and [ meas2 ] .
we analyse how the spectra behave under standard geometric operations such as unions , closures and products , see propositions [ basic1 ] , [ basic2 ] and [ basic3 ] .
next we compute the spectra explicitly in several concrete and well - studied settings , including : 1 .
self - affine carpets of the type considered by bedford and mcmullen , see theorem [ carpetsspectra ] .
2 . self - similar and self - conformal sets with overlaps ,
see theorem [ ssmain ] and corollary [ sscor ] .
mandelbrot percolation , see theorem [ mp ] .
moran constructions , see theorem [ moranthm ] .
5 . decreasing sequences with decreasing gaps ,
see theorem [ sequencesmain ] .
. spirals with sub - exponential and monotonic winding , see theorem [ spiral ] .
although the main purpose of this article is to introduce , and conduct a thorough investigation of , our new dimension spectra , we also obtain several results as corollaries or bi - products of our work , which are not _ a priori _ related to the spectra .
we summarise some of these results here : 1 .
we prove that the assouad dimension of a decreasing sequence of points with decreasing gaps is either 0 or 1 .
this answers a question posed to us by chris miller ( ohio state university ) , see theorem [ seqdichotomy ] .
2 . we compute the modified lower dimension of self - affine carpets , which typically turns out to be strictly larger than the lower dimension , see corollary [ modifiedlower ] .
we provide new bi - lipschitz invariants , both in general and in specific situations , see , for example , proposition [ t2 ] .
we provide bi - hlder distortion results for assouad dimension , both positive and negative , see theorem [ holderassouadcor ] and proposition [ negativeholder ] .
in particular , if the assouad spectrum reaches the assouad dimension , then we can give a bound on how the assouad dimension distorts under a bi - hlder map .
if the assouad spectrum does not reach the assouad dimension , then we show that the assouad dimension can be arbitrarily distorted .
we prove that the lower dimension is a baire 2 function , which is sharp and improves upon a previous result of fraser ( see ( * ? ? ?
* question 4.6 ) ) where it was only shown to be baire 3 , see our theorem [ meas3 ] . 6 .
we prove that sub - exponential spirals can not be ` unwound ' to line segments via certain bi - hlder functions , see corollary [ spiralcor ] .
this provides a natural extension to work of fish and paunescu concerning bi - lipschitz unwinding @xcite , as well as classical unwinding theorems of katznelson , subhashis and sullivan @xcite .
we prove that a spiral with ` monotonic winding ' either has assouad dimension 1 or 2 , see theorem [ assspiral ] .
8 . for a set @xmath13
we introduce and study the ( spectra of ) @xmath14-_tail densities _ and attempt to motivate their use in other fields . just as the dimension spectra give finer information than the assouad and lower dimensions , the tail densities give finer information than familiar notions of density such as the _ banach densities _ or _
asymptotic densities_. see section [ moranexamples ] and theorem [ taildensity ] .
finally , we collect several open questions and discuss possible directions for future work .
the assouad dimension is a fundamental notion of dimension used to study fractal objects in a wide variety of contexts .
it was popularised by assouad in the 1970s @xcite and subsequently took on significant importance in embedding theory .
recall the famous _ assouad embedding theorem _ which states that if @xmath15 is a metric space with the doubling property ( equivalently , with finite assouad dimension ) , then @xmath16 admits a bi - lipschitz embedding into some finite dimensional euclidean space for any @xmath17 .
the notion we now call assouad dimension does go back further , however , to larman s work in the 1960s @xcite and even to bouligand s 1928 paper @xcite .
it is also worth noting that , due to its intimate relationship with tangents ( see @xcite ) , it is intimately related to pioneering work of furstenberg on micro - sets which goes back to the 1960s , see @xcite .
roughly speaking , the assouad dimension assigns a number to a given metric space which quantifies the most difficult location and scale at which to cover the space .
more precisely , it considers two scales @xmath1 and finds the maximal exponential growth rate of @xmath18 as @xmath2 and @xmath4 decrease , where @xmath19 is the minimal number of @xmath4-balls required to cover a set @xmath20 .
the assouad dimension has found important applications in a wide variety of contexts , including a sustained importance in embedding theory , see @xcite .
it is also central to quasi - conformal geometry , see @xcite , and has recently been gaining significant attention in the literature on fractal geometry and geometric measure theory , see for example @xcite .
however , its application and interest does not end there .
for example , in the study of fractional hardy inequalities , if the boundary of a domain in @xmath21 has assouad dimension less than or equal to @xmath22 , then the domain admits the fractional @xmath23-hardy inequality , see @xcite .
also , hieronymi and miller have recently used the assouad dimension to study nondefinability problems relating to expansions in the real number field , see @xcite . since it is an _
quantity , the assouad dimension gives rather coarse information about the space and is often very large ; larger than the other familiar notions of dimension such as hausdorff and box - counting dimension .
also , despite the fact that two scales are used ( @xmath4 and @xmath2 ) , the assouad dimension returns no information about which scales ` see ' the maximal exponential growth rate described above . in this paper
we propose a programme to tackle these problems : we fix the relationship between the scales @xmath2 and @xmath4 and then compute the corresponding _ restricted _ assouad dimension by only considering pairs of scales with this fixed relationship .
more precisely , for a fixed @xmath5 , we look for the maximal exponential growth rate of @xmath18 as @xmath2 decreases and @xmath4 is defined by @xmath6 .
one can then vary @xmath7 and obtain a spectrum of dimensions for the given metric space which can be viewed as providing finer geometric information about the ( lack of ) homogeneity present and a more complete picture of how the space scales .
one may also be able to pick out @xmath7s which ` see ' the assouad dimension , i.e. , values where this spectrum reaches the true assouad dimension .
if the assouad dimension is ` seen ' by the spectrum , then we are able to glean more information about the assouad dimension because the spectrum is generally better behaved than the assouad dimension , see for example theorem [ holderassouadcor ] .
another key motivation of this work is that the finer the information we are able glean concerning the scaling structure of the space , the better the applications should be . in particular , we believe that the notions we introduce and study here should bear fruit in other areas where the assouad dimension already plays a role ; such as embedding theory , quasi - conformal geometry , and geometric measure theory .
we begin by recalling the precise definition of the assouad dimension , which serves to motivate our new definition .
let @xmath24 where @xmath25 is a fixed metric space ; which will often be a compact subset of euclidean space .
the _ assouad dimension _ of @xmath26 is defined by @xmath27 as described above , we will modify this definition by taking the infimum over the less restrictive condition that the scaling property only holds for scales @xmath28 satisfying a particular relationship . for @xmath5 , we define @xmath29 we are particularly interested in the function @xmath30 which we refer to as the _ assouad spectrum _ ( of @xmath26 ) .
for convenience we extend @xmath31 to @xmath32 by setting @xmath33 for @xmath34 .
of course there are many other ways to fix the relationship between the scales @xmath4 and @xmath2 .
however , it turns out that if one wants to develop a rich theory , the most natural way to do this is what we propose here . see the discussion in section [ diffdirdisc ] at the end of the paper for more details on this point .
the lower dimension , introduced by larman @xcite , is the natural dual of the assouad dimension .
we refer the reader to @xcite for an in - depth discussion of the relationships and differences between these dimensions .
let @xmath24 be as above .
the _ lower dimension _ of @xmath26 is defined by @xmath35 due to the local nature of this definition , it has many strange properties which may not be seen as desirable for a ` dimension ' to satisfy .
for example , it is not monotone as the presence of a single isolated point renders the lower dimension 0 , and it may take the value 0 for an open subset of euclidean space , see ( * ? ? ?
* example 2.5 ) .
one can modify the definition to get rid of these ( perhaps ) strange properties by defining the _ modified lower dimension _ by @xmath36 for @xmath5 , we define @xmath37 again , we are particularly interested in the function @xmath38 which we refer to as the _ lower spectrum _ ( of @xmath26 ) .
as before , we extend @xmath39 to @xmath32 by setting @xmath40 for @xmath34 . we can also modify this definition to force it to be monotone and to take on the ambient spatial dimension for open sets . the _ modified lower spectrum _ ( of @xmath26 ) is defined by @xmath41
the key motivation behind these new definitions is that the geometric information provided by the assouad and lower dimensions is too coarse .
we gain more information by understanding how the inhomogeneity depends on the scales one is considering .
alternative approaches to getting more out of these dimensions are possible . for example , fraser and todd @xcite recently considered a quantitative analysis of the assouad dimension where they looked to understand how inhomogeneity varies in space , i.e. as one changes the point @xmath42 around which one is trying to cover the set .
they found that for some natural examples this inhomogeneity could be described by a _
large deviations principle_. in a certain sense our approach is dual to that of @xcite in that we put restrictions on _ scale _ but still maximise over _ space _ , whereas in @xcite restrictions were put on _ space _ , but the quantities were still maximised over all _
here we summarise some notation which we will use throughout the paper . for real - valued functions @xmath43 , we write @xmath44 to mean that there exists a universal constant @xmath45 , independent of @xmath46 , such that @xmath47
. some readers may be more familiar with the notation @xmath48 , which is sometimes more convenient and means the same thing .
similarly , @xmath49 means that @xmath50 with a universal constant independent of @xmath46 . if both @xmath44 and @xmath49 , then we write @xmath51
. generally one should think of @xmath46 as being the tuple consisting of all variables in the expression @xmath52 .
usually @xmath46 will be a length scale but could sometimes also incorporate points in the metric space in question or other independent length scales . for a real number @xmath53 , we write @xmath54 to denote a real number that is strictly larger than @xmath53 but can be chosen as close to @xmath53 as we wish .
similarly , we write @xmath55 to denote a real number that is strictly less than @xmath53 but can be chosen as close to @xmath53 as we wish . for real numbers
@xmath56 , we write @xmath57 for the minimum of the two numbers and @xmath58 for the maximum . also , for a non - negative real number @xmath59 , we write @xmath60 $ ] for the integer part of @xmath46 .
the assouad and lower dimensions are closely related to the upper and lower box dimension .
upper _ and _ lower box dimensions _ of a totally bounded set @xmath26 are defined by @xmath61 and @xmath62 and if the upper and lower box dimensions coincide we call the common value the _ box dimension _ of @xmath26 and denote it by @xmath63 .
we refer the reader to ( * ? ? ?
* chapter 3 ) for more details on the upper and lower box dimensions and their basic properties .
note that , similar to the assouad and lower dimensions , the covering function @xmath64 can be taken to be numerous related covering or packing functions and it is sometimes convenient to use a variety of different definitions in practice . in particular
we note the following general relationships which hold for any totally bounded set @xmath26 : @xmath65 see @xcite .
the upper and lower box dimensions will play an important role in our analysis .
unlike the assouad and lower dimensions , we can give simple explicit formulae for the dimension spectra .
indeed , it follows immediately from the definitions that @xmath66 and @xmath67 as with the definitions of assouad and lower dimension , as well as the upper and lower box dimensions , the definition of @xmath68 may be replaced with a number of related concepts without altering any of the definitions .
for example , if working in @xmath21 we could use the number of @xmath4-cubes in an @xmath4-mesh which intersect the given set .
another possibility is to let @xmath19 be the maximal cardinality of an @xmath4-packing of @xmath20 , where an @xmath4-packing is a collection of closed pairwise disjoint balls of radius @xmath4 with centres in @xmath20 .
also , using the explicit formulae given above , we see that letting @xmath69 through an exponential sequence of scales , such as @xmath70 ( @xmath71 ) , yields the same limits .
we leave it to the reader to show that these variations lead to the same dimensions and spectra and refer to ( * ? ? ?
* chapter 3 ) ) for more details .
it is often useful to adopt these different definitions of @xmath68 , in particular when considering measureability properties in section [ borelsect ] . in practice ,
one often obtains estimates for @xmath72 and the following simple lemma is useful in summarising what one can deduce from such estimates .
we will often use it without referring to it explicitly .
[ n2d ] let @xmath26 be a given set and suppose that for some @xmath73 we have @xmath74 for some @xmath5 and all sufficiently small @xmath75 then we have @xmath76 if for some @xmath5 we can find a sequence @xmath77 for @xmath78 such that @xmath79 then we have @xmath80 in the following arguments , @xmath81 and @xmath82 are the implied constants relating to the use of @xmath83 and @xmath84 in the hypothesis . in particular , they do not depend on @xmath2 . then
, directly from the explicit formula above , we get : @xmath85 in the other direction we have @xmath86 which completes the proof .
we also provide the dual version of lemma [ n2d ] , which we state without proof .
[ ln2d ] let @xmath26 be a given set and suppose that for some @xmath73 we have @xmath87 for some @xmath5 and all sufficiently small @xmath75 then we have @xmath88 if for some @xmath5 we can find a sequence @xmath77 for @xmath78 such that @xmath89 then we have @xmath90 our first proposition gives general bounds on the assouad spectrum in terms of the assouad and box dimensions .
[ bb ] let @xmath26 be a totally bounded set .
then for all @xmath5 we have @xmath91 we will write @xmath92 for the upper box dimension of @xmath26 .
first of all there is a clear upper bound holding for any @xmath93 and small enough @xmath2 :
@xmath94 this implies that @xmath95 whenever we have a covering of @xmath26 by @xmath2-balls , if we further cover each @xmath2-ball with @xmath96-balls then we get a cover of @xmath26 by @xmath96-balls and an upper bound for @xmath97 .
we can cover @xmath26 with @xmath98 @xmath2-balls , and all those @xmath2-balls can be covered by at most @xmath99 many @xmath100-balls , therefore @xmath101 and so @xmath102 since @xmath103 for all small enough @xmath2 and @xmath104 for infinitely many @xmath105 we have that @xmath106 holds for a sequence of @xmath107 .
it now follows from lemma [ n2d ] and ( [ estimatebb1 ] ) and ( [ estimatebb2 ] ) that @xmath108 finally , since @xmath109 is a trivial upper bound for @xmath110 for all @xmath111 , the desired conclusion follows .
the above estimates show that if the upper box and assouad dimensions of a set coincide , then the assouad spectrum is constantly equal to the common value for all @xmath5 .
such sets are highly homogeneous and therefore it is not surprising that the assouad spectrum yields no new information .
fortunately , sets with distinct upper box and assouad dimensions abound and we will focus on such examples . in this case , the above estimates show that , in some sense , the assouad spectrum _ must _ yield finer information than the upper box and assouad dimensions alone . indeed , the only way the spectrum can be constant is if it is constantly equal to the upper box dimension , but such behaviour would be quite striking since the definition is more similar to the assouad dimension than the upper box dimension . in such cases ,
the assouad dimension is not ` seen ' by any @xmath7 and this shows that to obtain the assouad dimension , one must use a complicated collection of pairs @xmath112 without any clear exponential relationship .
we also note that these basic bounds are sharp .
we show that for a natural family of decreasing sequences , the upper bound is always obtained , see section [ sequencessect ] , and for mandelbrot percolation the ( constant ) lower bound is always obtained almost surely conditioned on non - extinction , see section [ mandelbrotsect ] .
we also note that for many natural examples the spectrum lies strictly between these upper and lower bounds , for example for the self - affine carpets studied by bedford and mcmullen , see section [ carpetssect ] . letting @xmath113 in the previous result we obtain the following corollary : [ bdconverge ] for any totally bounded set @xmath26
, we have @xmath114 as @xmath11 .
we note that one can not say anything as succinct about the limit of @xmath115 as @xmath12 , but it is often the case that @xmath116 in some ( sometimes large ) interval to the left of @xmath117 .
proposition [ bb ] has the following immediate corollary , which at first sight looks surprising since the definition of the assouad spectrum does not appear to depend so sensitively on the upper box dimension .
[ bd0 ] for any totally bounded set @xmath26 with @xmath118 , we have @xmath119 for all @xmath5 .
we now move towards analytic properties of the spectra .
[ abounds ] for any set @xmath26 and @xmath120 we have @xmath121 following similar ideas as in the proofs above , for @xmath120 , we have for any @xmath122 @xmath123 within @xmath26 we can cover any @xmath124-ball with @xmath125 many @xmath126-balls .
then for small enough @xmath122 we have @xmath127 therefore , @xmath128 also notice that for a sequence of @xmath69 we have @xmath129 as well as for any sufficiently small @xmath122 we have @xmath130 the desired bounds then follow immediately from lemma [ n2d ]
. the bounds in proposition [ abounds ] have some very useful consequences .
firstly , they imply continuity of the spectrum .
[ con ] the function @xmath30 is continuous in @xmath5 . using the bounds in proposition [ abounds ] and letting @xmath131 establishes lower semicontinuity of @xmath30 at @xmath132 and letting @xmath133 establishes upper semicontinuity of @xmath30 at @xmath134 . since @xmath134 and @xmath132 are arbitrary the desired continuity follows .
continuity of the dimension spectra is a useful property , especially when dealing with random fractals such as mandelbrot percolation : a _ continuous _ function is determined by its values on a _
countable _ dense set .
we point out that the spectra are not any more regular than continuous , as for most of our examples the spectra exhibit phase transitions where they fail to be differentiable
. another useful consequence of proposition [ abounds ] is that ` if the spectrum reaches the assouad dimension of @xmath26 , then it stays there ' .
[ staya ] if for some @xmath5 , we have @xmath135 , then @xmath136 for all @xmath137 . starting with the right hand inequality from proposition [ abounds ] ,
assume that @xmath138 .
this immediately gives that @xmath139 which , together with proposition [ bb ] , proves the result .
another consequence of proposition [ abounds ] is that the assouad spectrum is lipschitz on any closed subinterval of @xmath140 . in most examples
the spectrum will turn out to be lipschitz on the whole domain @xmath140 .
[ spectralip ] for any @xmath141 we have @xmath142 in particular , for any @xmath143 , the function @xmath30 is lipschitz on the interval @xmath144 $ ] .
this follows immediately from the bounds presented in proposition [ abounds ] .
another natural question concerns monotonicity .
indeed , all of the ` natural ' examples we consider have monotone spectra , i.e. the spectrum is non - decreasing in @xmath7 .
surprisingly , this is not always the case .
we exhibit this by constructing an example in section [ non - mono ] .
the following result shows that one does have some sort of ` quasi - monotonicity ' , however .
[ mono ] for any @xmath26 and @xmath145 we have @xmath146 in particular , by setting @xmath147 , we have @xmath148 for any @xmath149 .
furthermore , this implies that for any @xmath150 , we can find @xmath151 arbitrarily close to @xmath152 such that @xmath153 .
fix @xmath120 and notice that @xmath154 . for sufficiently small @xmath122
, we have : @xmath155 for @xmath156 .
we can also find infinitely many @xmath69 such that : @xmath157 let @xmath158 , and observe that @xmath159 .
we can cover any @xmath2 ball with at most @xmath160 balls with radius @xmath161 for small enough @xmath2 .
then we need no more than @xmath162 balls with radius @xmath163 to cover any @xmath161-ball .
given an arbitrary @xmath2-ball , first cover it with @xmath161-balls , and then cover those @xmath161-balls by @xmath163-balls using optimal covers as indicated above .
this yields @xmath164 but since @xmath159 we have @xmath165 for all @xmath161 small enough and also @xmath166 for some arbitrarily small @xmath2 .
therefore for some arbitrarily small @xmath2 we get : @xmath167 and so replacing @xmath161 by @xmath168 , taking logs , and dividing through by @xmath169 yields @xmath170 this , in turn , yields @xmath171 as required . the strategy of the above proof is to cover a large ball with middle sized balls and further cover the middle sized balls with smaller balls .
this can be generalized to arbitrarily many levels of covering to obtain more general results .
we can cover an @xmath2-ball with @xmath172-balls , then each of these balls with @xmath173-balls and so on and then applying the same proof strategy as above we end up with the following inequality : for @xmath174 we have @xmath175 by setting @xmath176 we end up with : @xmath177{\theta_1 } } f\ ] ] for any @xmath149 and any natural number @xmath178 , which is a slightly stronger result . we will now discuss the analogous properties for the lower ( and modified lower ) spectrum .
[ bbl ] let @xmath26 be a totally bounded set .
then for all @xmath5 we have @xmath179 and @xmath180 first note that it follows immediately from the definitions that @xmath181 and therefore also @xmath182 .
we will now prove the upper bounds and during the proof we will write @xmath183 for the lower box dimension of @xmath26 . fix @xmath5 and @xmath184 .
let @xmath185 denote the largest possible cardinality of an @xmath4-packing of a set @xmath20 by closed balls of radius @xmath4 .
take an optimal @xmath186-packing of @xmath26 by closed balls and then inside each of these balls construct an optimal @xmath96-packing of the smaller ball centered at the same point but with radius @xmath2 .
the resulting @xmath96-balls are centered in @xmath26 and are pairwise disjoint and , therefore , one obtains an @xmath96-packing of @xmath26 by more than @xmath187 balls .
this yields @xmath188 for infinitely many @xmath107 .
it then follows from lemma [ ln2d ] that @xmath189 from which the desired upper bound follows .
this also passes to the modified lower spectrum , completing the proof .
since the lower dimension is bounded above by the modified lower dimension ( and the hausdorff dimension if the set is compact , see @xcite ) it is natural to ask if this is also ( uniformly ) true for the lower spectrum , i.e. , if the upper bounds in proposition [ bbl ] can be improved ?
perhaps surprisingly , this is not the case , which we demonstrate by example in section [ carpetssect ] .
[ con2 ] the functions @xmath190 and @xmath191 are continuous in @xmath5 .
moreover , they are lipschitz on any closed subinterval of @xmath140 . for any @xmath192 and @xmath120 we have @xmath193 , therefore it is clear that for any @xmath192 : @xmath194 now notice that , @xmath195 this is because we may cover at least one @xmath2-ball with @xmath196 balls of radius @xmath197 , and therefore this number is no smaller than @xmath198 .
the above two inequalities imply that for infinitely many @xmath69 we have @xmath199 and also for infinitely many @xmath69 we have @xmath200 it follows that @xmath201 dividing through by @xmath202 and then letting @xmath131 establishes lower semicontinuity of @xmath203 at @xmath132 and letting @xmath133 establishes upper semicontinuity of @xmath203 at @xmath134 . since @xmath134 and @xmath132 are arbitrary the desired continuity follows .
the above discussion holds for any metric space @xmath26 , and in particular for any subspace @xmath204 we have @xmath205 taking the supremum over all @xmath204 throughout , we get @xmath206 and therefore the modified lower spectrum is also continuous . finally , the fact that the lower spectrum and modified lower spectrum are lipschitz on any closed subinterval of @xmath140 also follows immediately by applying the above bounds . in this section
we investigate how the various dimension spectra are affected by standard geometric operations such as products , unions , and images under hlder continuous maps .
it is clear that the spectra satisfies the following properties and we leave the proofs to the reader .
[ basic1 ] for any set @xmath26 in a metric space and any @xmath5 , we have : @xmath207 @xmath208 and for any @xmath209 and any @xmath5 , we have : @xmath210 @xmath211 interestingly , the modified lower dimension and modified lower spectrum are not stable under taking closure as the following example illustrates .
let @xmath212 ^ 2\ ] ] and observe that every point @xmath213 is isolated and therefore any subset of @xmath25 has an isolated point .
this implies that for any @xmath5 @xmath214 however , @xmath215 \times \{0 \ }
\subseteq \overline{x}$ ] and so @xmath216 [ basic2 ] let @xmath217 be a finite collection of sets and let @xmath218 then , for all @xmath150 , @xmath219 the upper bound follows immediately since covering the @xmath220 individually and then taking the covers together yields a cover of @xmath26 and so for all @xmath184 we have @xmath221 and the result follows by applying lemma [ n2d ] the lower bound is a trivial consequence of monotonicity .
clearly the above result does not hold for countable unions .
for example @xmath222 $ ] is a countable union of point sets , all of which have assouad spectrum constantly equal to @xmath223 , but @xmath222 $ ] has assouad spectrum constantly equal to @xmath152 by the closure property .
also , the previous result fails ( even for finite unions ) , in the case of the lower spectrum : consider the union of @xmath215 \cup \{2\}$ ] and @xmath224 $ ] .
one can say more if the sets in the union are properly separated .
[ basic2sep ]
let @xmath225 be ` properly separated ' subsets of a metric space @xmath15 , i.e. sets such that @xmath226 then , @xmath227 and , for all @xmath150 , @xmath228 and @xmath229 moreover , these results extend to arbitrary finite unions of pairwise ` properly separated ' sets where the maximum / minimum is taken over all sets in the union .
the argument for the lower spectrum is similar to ( * ? ? ?
* theorem 2.2 ) and is omitted . for the modified lower dimension and modified lower spectrum , the proof is straightforward and we only briefly give the modified lower dimension argument .
the lower bound ( @xmath230 ) follows from monotonicity . for the upper bound
, we have @xmath231 as required .
note that we used the fact that the lower dimension of the union of two properly separated sets is given by the minimum of the individual dimensions , which is provided in ( * ? ? ?
* theorem 2.2 ) .
we also adopt the convention that @xmath232 .
there are many results in dimension theory related to how the dimension of a product space depends on the dimensions of the marginals .
a common phenomenon is that dimensions are best considered in pairs and the following standard formula has been verified for many ` dimension pairs ' @xmath233 and @xmath234 : @xmath235 such examples include hausdorff and packing dimension , see howroyd @xcite ; lower and upper box dimension ; and lower and assouad dimension . for recent works on such product formulae
see @xcite .
we show below that the assouad and lower spectra give rise to a continuum of ` dimension pairs ' .
there are many natural ` product metrics ' to impose on the product @xmath236 of metric spaces @xmath237 and @xmath238 , with a natural choice being the sup metric @xmath239 on @xmath236 defined by @xmath240 in particular , this metric is compatible with the product topology and bi - lipschitz equivalent with many other commonly used product metrics , such as those induced by @xmath23 norms . [ basic3 ]
let @xmath241 be metric spaces and equip the product @xmath242 with any suitable product metric .
for any @xmath5 we have @xmath243 @xmath244 and @xmath245 for the purposes of this proof we use the sup metric on the product space .
in particular this means that the product of two covering sets of diameter @xmath4 is a set of diameter @xmath4 and so covers of parts of @xmath20 and @xmath26 can be easily combined to provide covers of the corresponding parts of @xmath242 .
let @xmath246 and @xmath247 denote the projection on to @xmath20 and @xmath26 respectively .
then clearly for any @xmath1 and @xmath248 we have @xmath249 setting @xmath250 yields @xmath251 and then applying lemma [ n2d ] gives @xmath252 on the other hand for any @xmath253 : @xmath254 as @xmath255 can be chosen such that @xmath256 we have @xmath257 this implies that : @xmath258 which , similar to above , yields @xmath259 as required .
the second chain of inequalities ( which concern the lower spectrum ) follow by a similar argument , which we omit .
the third chain of inequalities ( which concern the modified lower spectrum ) follow easily from the second . in particular , for the lower bound choose nonempty subsets @xmath260 and @xmath261 such that @xmath262 and @xmath263 and then apply monotonicity and the result for the lower spectrum to obtain @xmath264 which proves the desired lower bound . for the upper bound ,
the upper bound concerning the lower spectrum implies that @xmath265 which is almost what we want , apart from that it is not _ a priori _ obvious that the quantity on the left is equal to @xmath266 .
however , this follows since for any @xmath267 we have for any @xmath268 and @xmath122 that @xmath269 which yields that @xmath270 completing the proof .
we also obtain a sharp result for ` self - products ' .
let @xmath26 be a metric space , @xmath271 , and equip the @xmath178-fold product @xmath272 with any suitable product metric .
for any @xmath5 we have @xmath273 @xmath274 and @xmath275 this proof is similar to the general case and we omit the details .
the key point is that , for a self - product , one may choose @xmath42 which witnesses the extremal behaviour at some scale and then consider the point @xmath276 .
the projection of this point onto every coordinate then witnesses extremal behaviour and this passes to @xmath277 .
we note that using a similar approach one may also obtain the following minor , but useful , improvement on ( * ? ? ?
* theorem 2.1 ) . for metric spaces @xmath241 we have @xmath278
another important aspect of a dimension is how it behaves under distortion by maps which are ` not too wild ' .
indeed , all of the standard notions of dimension , such as the hausdorff , box , packing , assouad and lower dimension , are stable under bi - lipschitz distortion , for example .
relaxing bi - lipschitz to simply lipschitz or even hlder , there are elementary bounds which show that under distortion by an @xmath279-hlder map the hausdorff or box dimensions can not increase by more than a factor of @xmath280 , see ( * ? ? ?
* chapter 2 - 3 ) .
assouad and lower dimension do not enjoy such stability and can wildly increase under distortion by even a lipschitz map , see @xcite .
the reason for this is that because one is trying to control two scales ( one in each direction ) , one needs bounds on the distortion of the map in _ both _ directions .
here we conduct a detailed analysis of how the dimension spectra distorts under bi - hlder maps , i.e. , hlder maps with hlder inverses .
it is noteworthy that one can not relate the value of the assouad spectrum of the set and its image at a particular value @xmath7 , but rather at two different values of @xmath7 which are related according to the hlder parameters .
this makes the theory of dimension distortion for our spectra rather more subtle than for a dimension which returns a single exponent .
[ holderthm ] let @xmath281 be a map between doubling metric spaces @xmath237 and @xmath238 such that for all @xmath282 with @xmath283 sufficiently small , @xmath284 for some fixed constants @xmath285 .
then , for any @xmath286 and @xmath5 , we have @xmath287 @xmath288 and @xmath289 first notice that @xmath290 is invertible and for all @xmath291 with @xmath292 sufficiently small we have @xmath293 by the assumptions on @xmath290 there are uniform constants @xmath294 such that for any sufficiently small @xmath295 and @xmath213 and @xmath296 we have @xmath297 and @xmath298 therefore , for @xmath1 with @xmath2 small enough ( recall that our metric space has the doubling property ) we have @xmath299 for any @xmath213 .
the left inequality holds because any @xmath4-cover of @xmath300 can be mapped under @xmath301 to yield an ( up to multiplicative constants ) @xmath302-cover of @xmath303 by the same number of sets ( up to another multiplicative constant depending on the doubling property of the space ) .
similarly , the right inequality holds because any @xmath304-cover of @xmath305 can be mapped under @xmath290 to yield an ( up to multiplicative constants ) @xmath4-cover of @xmath300 by the same number of sets ( up to another multiplicative constant depending on the doubling property of the space ) . setting @xmath306 from here
we notice that for any sufficiently small @xmath122 we have by definition @xmath307 and , similarly , for infinitely many @xmath107 we have @xmath308 recall that if @xmath309 , then @xmath310 . also by definition , for any sufficiently small @xmath122 we have @xmath311 and for infinitely many @xmath107 we have @xmath312 together these estimates yield that for infinitely many @xmath107 we have @xmath313 and @xmath314 which gives @xmath315 as required .
the argument for the lower spectrum is similar and omitted .
since subsets of @xmath26 are in one to one correspondence with subsets of @xmath316 through the map @xmath290 , we may take supremum over nonempty subsets of @xmath26 throughout , which yields the analogous estimates for the modified lower spectrum , completing the proof .
the lower bounds for the spectra of @xmath316 all become equal to 0 ( and thus trivial ) when @xmath317 and the upper bounds for the spectra of @xmath316 blow up as @xmath12 .
these are unfortunate properties but are indicative of the complex relations between the spectra at different values of @xmath7 once the set has been distorted by @xmath290 .
one can rectify this situation somewhat by combining our estimates with lemmas [ bb ] and [ bbl ] and the classical results that upper and lower box dimension can not increase by more than a factor of @xmath280 under distortion by an @xmath279-hlder map .
[ holdercor ] let @xmath281 be as in proposition [ holderthm ] . then for any @xmath24 and @xmath5 @xmath318 @xmath319 and @xmath320 note that in the above we could also bound @xmath321 from above by @xmath322 , but the point is to bound dimensions of @xmath316 by expressions involving only dimensions of @xmath26 and the bi - hlder restrictions on @xmath290 are not enough to yield bounds for the _ assouad dimension _ of @xmath316 in terms of @xmath26 .
also , we could have used lemma [ bb ] to apparently improve the upper bound for the assouad spectrum to include the bound @xmath323 but by virtue of lemma [ bb ] it follows that @xmath324 for all @xmath5 and so this estimate can not improve the one we already have .
it is important to comment on the sharpness of the estimates from corollary [ holdercor ] .
a first observation is that such estimates can not possibly be sharp in any precise sense , although letting @xmath325 shows that they are at least asymptotically sharp .
the reason for this is that they are based on knowledge of the extremal distortion of @xmath26 over the whole space and the spectra are only sensitive to the extremal properties of the set in question .
indeed , the thickest part of the set @xmath26 ( and @xmath316 ) , which determines the spectra , may occur at a location in the domain of @xmath290 where the distortion is less than the global extreme .
we will consider our estimates in detail for a natural family of sets and bi - hlder maps in section [ seq1 ] .
setting @xmath326 in proposition [ holderthm ] , we obtain bi - lipschitz stability as another immediate corollary .
[ bilipcor ] the assouad , lower , and modified lower , spectra are bi - lipschitz invariant .
being bi - lipschitz invariant is a useful property and one possible application is in classifying metric spaces up to bi - lipschitz equivalence .
there has been considerable interest in this problem since the seminal paper of falconer and marsh @xcite which sought to determine for which pairs of self - similar subsets of the line one can find a bi - lipschitz function taking one to the other . having the same hausdorff dimension is a necessary condition , since hausdorff dimension is a bi - lipschitz invariant , but it is not sufficient : there are self - similar subsets of the line which have the same hausdorff dimension but which are not bi - lipschitz equivalent . as such , it is useful to find other bi - lipschitz invariants , such as the other notions of dimension mentioned above .
corollary [ bilipcor ] provides a new continuum of bi - lipschitz invariants and so has potential applications in proving that certain metric spaces are not bi - lipschitz equivalent , even if their hausdorff , box , packing , assouad and lower dimensions are equal .
suppose @xmath290 is a map on @xmath26 such that @xmath327 uniformly as @xmath328 .
such maps are sometimes called _ quasi - lipschitz _ , see @xcite . rather than setting @xmath326 in proposition [ holderthm ] ,
if we just let @xmath329 we see that the spectra are also all invariant under _ quasi - lipschitz _ maps .
[ quasibilipcor ] the assouad , lower , and modified lower , spectra are quasi - lipschitz invariant .
in proposition [ holderthm ] we gave some estimates for the assouad spectrum of a set after distortion by a bi - hlder map .
similar , but simpler , estimates hold for the other standard notions of dimension , such as the hausdorff dimension , but nothing seems to be known in th assouad dimension case . in light of the other known results , one might expect that : @xmath330 for any bi - hlder map with parameters @xmath331 . in particular , these bounds hold if assouad dimension is replaced by hausdorff , packing , upper box or lower box dimension .
the situation turns out to be more subtle for assouad dimension .
first , we obtain the following partial result as a neat consequence of our results on the assouad spectrum .
[ holderassouadcor ] let @xmath281 be as in proposition [ holderthm ] and let @xmath332 : \dim_\mathrm{a}^{\theta } f=\dim_\mathrm{a } f \right\}\ ] ] assuming the set of suitable @xmath7s is non - empty .
then for any @xmath24 we have @xmath333 before we prove the result , note that if @xmath334 exists and @xmath335 , then proposition [ bb ] implies that @xmath336 we have for any @xmath337 $ ] that @xmath338 and proposition [ holderthm ] further implies that @xmath339 therefore , applying these inequalities with @xmath340 proves the desired lower bound .
observe that the smaller @xmath341 is , the better our lower bound for the assouad dimension of @xmath316 .
it is natural to consider an analogous upper bound , but this would require _ a priori _ knowledge of the set @xmath316 , i.e. , we have to know @xmath342 : \dim_\mathrm{a}^{\theta } s(f)=\dim_\mathrm{a } s(f ) \rbrace $ ] .
one may obtain a precise analogue by applying the above theorem with @xmath290 replaced by @xmath301 , but we do not pursue the details here . we have identified that if one assumes that the assouad spectrum reaches the assouad dimension , then one can give some non - trivial dimension bounds for distortion under bi - hlder maps . in particular , to obtain some bounds one needs additional assumptions about the set @xmath26 . in this section
we show that this really is the case and for general sets the situation is as bad as possible : the assouad dimension may be distorted by an absolute constant for bi - hlder maps with parameters arbitrarily close to 1 , i.e. , maps which are arbitrarily close to being bi - lipschitz .
this shows that there does not exist general bounds on the assouad dimension of @xmath316 in terms of the assouad dimension of @xmath26 and the hlder parameters @xmath279 and @xmath343 .
[ negativeholder ] there exists a compact set @xmath344 $ ] and a homeomorphism @xmath290 satisfying @xmath327 uniformly as @xmath328 such that @xmath345 in particular , @xmath290 is bi - hlder with parameters @xmath346 and @xmath347 for any @xmath143 , i.e. , for any @xmath143 we have @xmath348 note that the dimensions of @xmath316 and @xmath26 can be reversed in the above result by using the map @xmath301 . to construct @xmath26 , we start with a rooted infinite tree @xmath349 with the following properties : 1 .
each vertex has either 1 or @xmath350 sub - vertices 2 . at each level of the form @xmath351 for @xmath271
, there is a special vertex chosen to be the root of a full binary subtree with @xmath178 levels , embedded into @xmath349 between the levels @xmath352 3 .
all other vertices , i.e. those not involved in the family of binary subtrees with @xmath178 levels , have 1 sub - vertex the tree @xmath349 contains full binary trees of any length @xmath178 , and any full binary tree of length @xmath178 appears only once ( ignoring embeddings in larger trees appearing at later levels ) .
the binary tree of length @xmath178 is rooted at the level @xmath353 ; a sequence which increases very fast .
let us order the vertices of @xmath349 at each level such that on level @xmath354 the special vertex is the smallest .
note that the tree @xmath349 exists , and is a subtree of the full binary tree @xmath355 .
it is standard to represent infinite paths in the full binary tree @xmath355 as infinite sequences of @xmath223s and @xmath152s , i.e. we identify the ends of @xmath356 with the cantor space @xmath357 , equipped with the product topology .
define a bijective function from @xmath358 $ ] by @xmath359 noting that the image of @xmath360 is the familiar middle - third cantor set .
there is a closed subset @xmath361 corresponding to the infinite paths ( ends ) of @xmath349 and we set @xmath362 which is a countable compact subset of @xmath215 $ ] . observe that , due to our ordering of the vertices of @xmath349 , @xmath26 has one accumulation point at @xmath223 , and all other elements of @xmath26 are isolated points .
straightforward calculations yield that the hausdorff and box - counting dimensions of @xmath26 are @xmath223 and the assouad dimension is @xmath363 ; the same value as the middle - third cantor set .
indeed , for @xmath364 the set @xmath365 needs @xmath366 many @xmath4-balls to cover it and so @xmath367 .
the reverse inequality follows by observing that @xmath26 is actually a subset of the middle - third cantor set .
now let us take a closer look at @xmath26 , and decompose it as a countable union of sets of finite points @xmath368 , where the union is over numbers @xmath369 of the form @xmath370 and @xmath371 has @xmath372 points corresponding to the ends below the binary @xmath178-level boundary tree embedded in @xmath349 .
the points in @xmath371 are of the form @xmath373 where @xmath374 is any sequence over @xmath375 of length @xmath178 .
since @xmath376 grows very fast , the sets @xmath371 intersect each other only at @xmath223 , and in fact @xmath371 is contained fully inside the interval @xmath377 where @xmath378 is the left most point other than @xmath223 of @xmath379 .
consider the monotonic rearrangement of the points inside each @xmath371 which fixes 0 and spreads the remaining @xmath380 points out uniformly inside the interval @xmath381 $ ] .
call the rearranged set @xmath382 . even after this rearranging
, the @xmath382 are still disjoint except at @xmath223 , and we let @xmath383 . define the map @xmath384 to be the map which rearranges all of the @xmath371 simultaneously and we will now show that @xmath290 satisfies the desired regularity property . for @xmath385 , @xmath386 and @xmath387 with @xmath388 therefore , @xmath389 now @xmath351 and so as @xmath390 , we have both @xmath391 and @xmath392 . similarly , consider @xmath393 belonging to different @xmath371s , say , @xmath394 and suppose without loss of generality that @xmath395
. then we have @xmath396 and @xmath397 and so @xmath398 recalling the inequality @xmath399 it is enough to prove that : @xmath400 and @xmath401 since @xmath402 we have that @xmath403 and so whenever @xmath404 we have @xmath405 and so the desired convergence is clear .
all that remains is to prove that @xmath406 has full assouad dimension .
clearly for @xmath351 , there is a @xmath407-ball that needs at least @xmath408 many @xmath409-balls to cover it , and therefore @xmath410 which implies that @xmath411 we recently discovered that a similar result to our proposition [ negativeholder ] was also obtained by l and xi ( * ? ? ?
* proposition 1.2 ) , using a different proof .
the examples constructed by l and xi relied on a previous result of wang and xi @xcite which states that two quasi ahlfors - david @xmath412-regular sets with @xmath413 are quasi - lipschitz equivalent .
our proof has the advantage of being self - contained .
in this section we consider the borel measureability of the assouad and lower spectra . this question has previously been investigated in related contexts .
for example , mattila and mauldin @xcite proved that the hausdorff dimension and upper and lower box dimensions are borel measurable and , moreover , are of baire class 2 , but packing dimension is _ not _ borel measurable . in @xcite
it was shown that the assouad and lower dimensions are borel measureable .
more precisely , it was shown that assouad dimension is baire 2 , but it was only shown that lower dimension is baire 3 , leaving open the possibility that it is baire 2 ( * ? ? ?
* question 4.6 ) .
we solve this problem here by proving that the lower dimension is in fact ` precisely ' baire 2 .
the _ baire hierarchy _ is used to classify functions by their ` level of discontinuity ' and is formulated for functions between metric spaces @xmath414 and @xmath415 as follows . a function @xmath416 is baire 0 if it is continuous .
the latter classes are defined inductively by saying that a function @xmath417 is baire @xmath418 if it is a pointwise limit of a sequence of baire @xmath178 functions .
thus functions lying in higher baire classes ( and not in lower ones ) are set theoretically further away from being continuous . for more details on the baire hierarchy ,
see @xcite .
in particular , if a function belongs to any baire class , then it is borel measurable .
let @xmath419 denote the set of all non - empty compact subsets of a non - empty compact metric space @xmath25 and endow this space with the hausdorff metric , @xmath420 , defined by @xmath421 for @xmath422 and where @xmath423 denotes the @xmath424-neighbourhood of @xmath20 .
the metric space @xmath425 is compact . equip the product space @xmath426 with the product topology and any compatible metric .
[ meas1 ] the function @xmath427 defined by @xmath428 is of baire class 2 and , in particular , borel measurable .
moreover , it is not in general of baire class 1 .
[ meas2 ] the function @xmath429 defined by @xmath430 is of baire class 2 and , in particular , borel measurable .
moreover , it is not in general of baire class 1
. we also answer ( * ? ? ?
* question 4.6 ) in the affirmative : [ meas3 ] the function @xmath431 defined by @xmath432 is of baire class 2
. we will prove theorems [ meas1 ] , [ meas2 ] and [ meas3 ] in section [ borelproof ] .
finally we remark that , by using similar techniques , one may also prove that for a fixed @xmath5 the maps @xmath433 and @xmath434 are baire 2 ( and not baire 1 ) , but we omit the details .
recall that for a fixed @xmath435 the maps @xmath436 and @xmath437 are continuous ( baire 0 ) , see corollary [ con ] and theorem [ con2 ] .
we are unaware of any _ a priori _ way of relating baire classes of a function on a product space which is separately measureable to the baire classes of its fibre maps ( in this case 2 and 0 ) .
a classical result of lebesgue shows that if a function on a product space is separately continuous , then it is no worse than baire 1 , see @xcite .
we first establish some technical semicontinuity properties , from which the desired baire classifications will swiftly follow in the subsequent section .
for clarity we write @xmath438 for the _ closed _ ball centered at @xmath213 with radius @xmath122 and @xmath439 for the _ open _ ball centered at @xmath213 with radius @xmath122 . also , @xmath440 will denote the smallest number of _ open _ sets required for an @xmath4-cover of @xmath24 and @xmath441 will denote the maximum number of _ closed _ balls in an @xmath4-packing of @xmath286 , where an @xmath4-packing of @xmath26 is a pairwise disjoint collection of closed balls centered in @xmath26 of radius @xmath4 . for convenience we assume that @xmath442 for any @xmath295
. we will establish semicontinuity of six related covering or packing functions , which we define now . for fixed @xmath184 define four functions @xmath443 , @xmath444 , @xmath445 , @xmath446 by @xmath447 @xmath448 @xmath449 and @xmath450 also , for fixed @xmath184 and @xmath451 define @xmath452 , @xmath453 by @xmath454 and @xmath455 it is useful to note that all of the infimums and supremums in the above definitions are actually minimums and maximums , respectively .
this is because the packing and covering numbers only take positive integer values .
[ semicontinuity ] for fixed @xmath184 and @xmath451 the following semicontinuity properties hold : 1 .
@xmath443 is upper semicontinuous 2 .
@xmath444 is upper semicontinuous 3 .
@xmath445 is lower semicontinuous 4 .
@xmath456 is lower semicontinuous 5 .
@xmath452 is upper semicontinuous 6 .
@xmath457 is lower semicontinuous fix @xmath184 and @xmath451 . 1 . to prove that @xmath443 is upper semicontinuous , it suffices to show that for any @xmath458 , the set @xmath459 is closed .
fix @xmath458 and let @xmath460 be a convergent sequence of pairs from the above set , with limit @xmath461 .
thus for each @xmath462 we may find @xmath463 such that @xmath464 take a subsequence of the sets @xmath465 which converges in the hausdorff metric , which we may do by compactness .
denote the hausdorff limit of this sequence of sets by @xmath466 and note that @xmath467 .
let @xmath468 be any finite open @xmath96-cover of @xmath469 .
we may choose @xmath462 large enough to guarantee that @xmath468 is an open @xmath96-cover of @xmath465 and , moreover , we may simultaneously choose @xmath462 large enough such that we may find an open @xmath470-cover of @xmath465 by the same number of sets . if @xmath471 then @xmath468 suffices and if @xmath472 then this can be achieved by shrinking the sets @xmath468 slightly , using the fact that this is an open cover of a closed set and so there is room to do so .
this guarantees @xmath473 and thus the limit @xmath474 belongs to ( [ closedset1 ] ) , proving that it is closed .
2 . to prove that @xmath444 is upper semicontinuous , it suffices to show that for any @xmath458 , the set @xmath475 is open .
fix @xmath458 and let @xmath474 be a pair from the above set .
thus we may find @xmath42 such that @xmath476 let @xmath468 be an open @xmath96-cover of @xmath469 by @xmath477 sets .
if @xmath478 is sufficiently close to @xmath26 in the hausdorff metric , then we may find @xmath479 such that @xmath468 is also an open @xmath96-cover of @xmath480 .
moreover , if @xmath151 is sufficiently close to @xmath7 , then we may distort @xmath468 to obtain an open @xmath481-cover of @xmath480 by the same number of sets .
this guarantees that if @xmath482 is sufficiently close to @xmath474 in the product metric , then @xmath483 which proves that ( [ openset1 ] ) is open .
3 . to prove that @xmath445 is lower semicontinuous , it suffices to show that for any @xmath458 , the set @xmath484 is open .
fix @xmath458 and let @xmath474 be a pair from the above set .
thus we may find @xmath42 such that @xmath485 let @xmath486 be an @xmath96-packing of @xmath487 by @xmath488 closed balls . if @xmath478 is sufficiently close to @xmath26 in the hausdorff metric , then we may find @xmath479 close to @xmath46 and @xmath489 with @xmath490 close to @xmath491 such that @xmath489 is an @xmath96-packing of @xmath492 by the same number of closed balls .
moreover , if @xmath151 is sufficiently close to @xmath7 , then we may distort the @xmath493 to obtain an @xmath481-packing of @xmath492 .
this guarantees that if @xmath482 is sufficiently close to @xmath474 in the product metric , then @xmath494 which proves that ( [ openset2 ] ) is open .
4 . to prove that @xmath456 is lower semicontinuous , it suffices to show that for any @xmath458 , the set @xmath495 is closed .
fix @xmath458 and let @xmath460 be a convergent sequence of pairs from the above set , with limit @xmath461 .
thus for each @xmath462 we may find @xmath463 such that @xmath496 take a convergent subsequence from the sequence @xmath497 , which we may do by compactness .
denote the limit of this sequence by @xmath46 and note that @xmath42 .
we claim that @xmath498 which shows that @xmath474 belongs to ( [ closedset2 ] ) , proving that it is closed .
it remains to prove the claim , which we do by contradiction .
assume that @xmath486 is an @xmath96-packing of @xmath487 by strictly greater than @xmath499 closed balls .
following ( 3 ) above , if @xmath482 is sufficiently close to @xmath474 in the product metric , then we may find @xmath479 close to @xmath46 and @xmath489 with @xmath490 close to @xmath491 such that @xmath489 is an @xmath481-packing of @xmath492 by the same number of closed balls .
this contradicts ( [ contradictionbound ] ) and thus completes the proof .
we note that it was crucial to our argument that @xmath439 was an _
open _ ball .
upper semicontinuity of @xmath452 was already established in ( * ? ? ? * lemma 5.6 ) , where it was written as @xmath500 .
the proof is very similar to ( in fact slightly simpler than ) the proof of ( 2 ) above and is omitted .
it was hinted at in @xcite that @xmath457 may not be lower semicontinuous , but we show here that it is .
this is the key to lowering the baire class of lower dimension from 3 to the sharp value of 2 .
the proof is very similar to ( in fact slightly simpler than ) the proof of ( 4 ) above and is omitted . to prove that a function from a metric space @xmath501 to @xmath502 is baire 2 , it suffices to show that any open interval in @xmath502 pulls back to a @xmath503 set in @xmath501 , see ( *
* theorem 24.3 ) . recall that a set is @xmath503
if it can be expressed as a countable union of sets which are themselves expressible as countable intersections of open sets .
as such , throughout this section let @xmath504 with @xmath505 and we will consider various pullbacks of the interval @xmath506 .
we write @xmath507 to denote the strictly positive rationals .
we begin by considering the assouad spectrum .
straight from the definition of the assouad spectrum ( and the equivalent formulations ) , we obtain @xmath508 which is @xmath503 by the lower semicontinuity of @xmath509 and the upper semicontinuity of @xmath510 , see lemma [ semicontinuity ] parts ( 3 ) and ( 1 ) respectively .
this completes the proof that @xmath511 is baire 2 .
we now turn to the lower spectrum , which is similar .
straight from the definition of the lower dimension ( and the equivalent formulations ) , we obtain @xmath512 which is @xmath503 by the lower semicontinuity of @xmath513 and the upper semicontinuity of @xmath514 , see lemma [ semicontinuity ] parts ( 4 ) and ( 2 ) respectively .
this completes the proof that @xmath515 is baire 2 .
finally , it is straightforward to see that neither the assouad spectrum nor the lower spectrum are typically baire 1 .
this can be seen by recalling that the points of continuity for a baire 1 function form a dense @xmath516 set , see ( * ? ? ?
* theorem 24.14 ) , but it is evident that the assouad and lower spectra are discontinuous everywhere ( unless @xmath25 has a particularly simple form ) . in particular , every set in @xmath419 can be approximated by finite sets ( which all have dimension spectra equal to 0 ) or by their closed @xmath424-neighbourhoods ( which typically have dimension spectra larger than 0 ) .
if @xmath25 is the euclidean unit interval @xmath215 $ ] , for example , then the closed @xmath424-neighbourhood of any set in @xmath419 has lower spectrum equal to 1 .
of course , pathological examples are possible : if @xmath25 is a single point , for example , then all of the dimension spectra are trivially continuous .
we now turn to proving that the lower _ dimension _ is baire 2 , thus answering ( * ? ? ?
* question 4.6 ) .
straight from the definition of the lower dimension ( and the equivalent formulations ) , we obtain @xmath517 which is @xmath503 by the lower semicontinuity of @xmath518 and the upper semicontinuity of @xmath519 , see lemma [ semicontinuity ] parts ( 6 ) and ( 5 ) respectively .
this completes the proof that @xmath520 is baire 2 .
it was observed in @xcite that @xmath520 is not baire 1 , again since it is discontinuous everywhere .
in this section we will consider a simple family of countable compact fractal subsets of the line which allow explicit calculation of the assouad spectra . observe that any countable compact set has lower spectra and modified lower spectra equal to zero and so we omit discussion of these spectra for the duration of this section . despite being relatively simple ,
these fractal sequences have several useful properties .
first they provide us with a continuum of examples where the upper bound from proposition [ bb ] is sharp , see section [ seqmain ] .
they also allow us to analyse the bounds for dimension distortion under bi - hlder maps given in corollary [ holdercor ] in an explicit and representative way , see section [ seq1 ] .
finally , they provide our first example showing that the assouad spectrum need not peak at the assouad dimension , see section [ seq2 ] . more precisely , we study _ decreasing sequences with decreasing gaps _ , which we formulate as follows . let @xmath360 be a function from @xmath521 to @xmath215 $ ] such that @xmath52 and @xmath522 are both strictly decreasing functions and they converge to @xmath223 as @xmath523 .
we also assume for convenience that both @xmath360 and @xmath524 are smooth .
our set of interest is then @xmath525 we start by studying the assouad dimension .
let @xmath526 and consider the number @xmath527 for any @xmath295 , there is a smallest number @xmath528 such that @xmath529 .
notice that by definition @xmath530=g^{-1}(r)+o(1)$ ] . if @xmath531 then we will need approximately @xmath532 many @xmath4-balls to cover @xmath533 , and this is already of the largest order possible .
therefore we will focus the case where @xmath534 .
the following bound follows from the fact that the sequence has decreasing gaps : @xmath535 if @xmath536 we have the key formula : @xmath537 this is because for points smaller than @xmath538 the gaps are smaller than @xmath4 , therefore we need approximately @xmath539 many @xmath4-balls to cover , and for points between @xmath538 and @xmath2 which are approximately @xmath540 of them , we need one @xmath4-ball for each of them .
our first observation is that if the sequence decays sufficiently quickly , then the assouad dimension is 0 . in turn
this means that the box dimensions are 0 and the assouad spectrum is constantly equal to 0 . [ fastdecay ] if there is some @xmath541 such that @xmath542 for all @xmath271 ( or for all @xmath543 for some integer @xmath544 ) , then @xmath545 .
first note that @xmath546 therefore the smallest @xmath547 such that @xmath548 satisfies the following relation : @xmath549 and hence @xmath550 remember from the definition of @xmath528 that we have @xmath551 let us write @xmath552 to denote the smallest positive integer such that @xmath553 .
an argument similar to the discussion after ( [ key ] ) above shows that we need no more than a constant times @xmath554 many @xmath4-balls to cover @xmath555 a simple telescoping argument yields @xmath556 this implies that @xmath557 and so we need no more than @xmath558 many @xmath4-balls to cover @xmath533 .
this proves that @xmath26 has assouad dimension 0 .
the following result provides a dichotomy which at first sight may be surprising , since the box dimensions of such sequences can take any value in @xmath215 $ ] .
this result answers a question posed to us by chris miller ( ohio state university ) .
we recently discovered that this result was also obtained by garca , hare and mendivil ( * ? ? ?
* proposition 3.1 ) , using a different proof .
[ seqdichotomy ] let @xmath559 be a decreasing sequence with decreasing gaps as described above .
then the assouad dimension of @xmath26 is either @xmath223 or @xmath152 .
since @xmath522 , it is always true that @xmath560 and as both functions go to @xmath223 for large @xmath46 , we can conclude @xmath561 for @xmath4 small enough . for any @xmath122 small enough , @xmath562 and
since @xmath563 is decreasing and goes to @xmath564 as @xmath105 we can find @xmath451 such that @xmath565 .
for this choice of @xmath2 and @xmath4 we have @xmath566 and we can deduce that : @xmath567 therefore , if we can choose @xmath4 such that the ratio @xmath568 goes to @xmath564 as @xmath107 , then the assouad dimension has to be 1 .
therefore , if the assouad dimension is strictly less than 1 , it must be true that there exists a constant @xmath569 such that for all @xmath122 the @xmath4 chosen above satisfies @xmath570 .
this means that @xmath571 applying @xmath524 throughout yields @xmath572 which , by the definition of @xmath524 , gives @xmath573 replacing @xmath2 by @xmath574 , and rearranging yields @xmath575 which shows that the sequence decays exponentially fast and so , by lemma [ fastdecay ] , the assouad dimension 0 .
our main result on dimensions of decreasing sequences with decreasing gaps is as follows and shows that the assouad spectrum only depends on the upper box dimension of the set .
a pleasant ancillary benefit of the explicit formula we obtain is that for all such sets the upper bound in proposition [ bb ] is sharp . also , we note that the upper box dimensions of such sets are well studied and can be computed effectively . for example
, the upper ( and lower ) box dimensions can be estimated in terms of the exponential decay rate of the gap lengths @xmath576 .
see the ` cut out sets ' discussed in ( * ? ? ?
* chapter ) for more information .
[ sequencesmain ] let @xmath559 be a decreasing sequence with decreasing gaps as described above .
then for all @xmath5 we have @xmath577 if either @xmath578 or @xmath579 then the result follows immediately from proposition [ bb ] and therefore we may assume from now on that @xmath580 let @xmath581 and recall the key formula ( [ key ] ) : @xmath582 in particular , if @xmath122 is sufficiently small , then @xmath583 . if we can find infinitely many @xmath69 such that @xmath584 then @xmath585 and we get @xmath586 proving the result . therefore assume that for all sufficiently small @xmath122 we have @xmath587 this implies that for @xmath2 small enough we have @xmath588 which in turn implies that for @xmath178 large enough we have @xmath589 and @xmath590 this holds for all large enough @xmath178 and therefore we can assume it holds for all @xmath178 without loss of generality . for simplicity , we write @xmath591 .
we have @xmath592 iterating the above inequality yields @xmath593 and therefore @xmath594 this implies that for all large enough @xmath178 we have @xmath595 and for small enough @xmath46 we have @xmath596 recalling that @xmath597 is the upper box dimension , we can find a sequence @xmath598 such that : @xmath599 if for infinitely many @xmath462 we have @xmath600 then the situation is the same as in the beginning of the proof and we get our conclusion
. otherwise we have infinitely many @xmath462 such that @xmath601 it follows from the key formula ( [ key ] ) , that for infinitely many @xmath462 we have @xmath602 since @xmath603 we have @xmath604 and so @xmath605 for infinitely many @xmath462 .
it follows that for @xmath606 we have @xmath607 and proposition [ bb ] then implies that @xmath608 finally , continuity of the spectrum ( corollary [ con ] ) gives that for @xmath609 we have @xmath610 and corollary [ staya ] yields that this also holds for all @xmath611 completing the proof . in this section
provide our first concrete example where we can compute the assouad spectrum explicitly .
we specialise to a particular continuously parameterised family of decreasing sequences with prescribed polynomial decay . in particular , for a fixed @xmath612 we study the set @xmath613 and give an explicit formula for @xmath614 .
these sets @xmath615 are some of the first examples one considers when studying the box and assouad dimensions ( in particular @xmath616 ) and elementary calculations reveal that for any @xmath612 @xmath617 we therefore have the following immediate corollary of theorem [ sequencesmain ] . for all @xmath612 and @xmath5 we have
@xmath618 for @xmath619 going from left to right .
the bounds from proposition [ bb ] are shown as dashed lines , although the upper bound is obtained in each case .
[ sequences ] ] the family of sets @xmath620 studied in this section provide us with a simple continuum of sets any of which can be mapped onto any other by a bi - hlder map . since we have a very simple explicit formula for the assouad spectrum of each @xmath615 , this provides an excellent opportunity to test the bounds obtained in corollary [ holdercor ] . for @xmath73 ,
let @xmath621 \to [ 0,1]$ ] be defined by @xmath622 and observe that for any @xmath612 @xmath623 also note that if @xmath624 then @xmath290 is @xmath279-hlder with a lipschitz inverse , i.e. , for all @xmath625 $ ] @xmath626 and if @xmath627 then @xmath290 is lipschitz with a @xmath280-hlder inverse , i.e. , for all @xmath625 $ ] @xmath628 in the @xmath624 region , the bounds on the assouad spectrum from corollary [ holdercor ] yield that for @xmath5 @xmath629 and applying the explicit formulae for the assouad spectra derived above gives : @xmath630 under bi - hlder distortion by @xmath631 . on the left @xmath632 and @xmath633 and on the right @xmath634 and @xmath635 .
the actual spectrum of @xmath636 is shown as a solid line and the bounds are dashed.[holderpic1 ] ] in the @xmath627 region , the bounds on the assouad spectrum from corollary [ holdercor ] yield that for @xmath5 @xmath637 and applying the explicit formulae for the assouad spectra derived above gives : @xmath638 [ holderpic2 ] under bi - hlder distortion by @xmath631 . on the left @xmath639 and @xmath640 and on the right @xmath641 and @xmath642 .
the actual spectrum of @xmath636 is shown as a solid line and the bounds are dashed.,title="fig : " ] here we give a simple example where the assouad spectrum does not reach the assouad dimension .
our example is the set @xmath644 which falls into the context of this section .
explicit discussion of this example is pertinent because it clearly indicates one of the key differences between particular values of the assouad spectrum and the assouad dimension .
for this example , the sequence does not decay fast enough in the sense of theorem [ fastdecay ] , and so the assouad dimension is 1 .
now we compute its assouad spectrum . using the notation of this section , we have @xmath645 @xmath646 @xmath647 @xmath648 and @xmath649 for any @xmath5
, we have @xmath650 and therefore the assouad spectrum is controlled by @xmath651 and therefore @xmath652 for all @xmath653 .
of course this fact also follows by arguing that @xmath654 and applying corollary [ bd0 ] .
here the crucial point is that if we do not fix the relationship between @xmath2 and @xmath4 then for any @xmath2 we can choose @xmath4 to be small but increasingly close to @xmath2 as @xmath107 , such that the formula gives us a non - trivial @xmath568 term .
then only thing to be careful of is to ensure that @xmath568 is unbounded and we will achieve assouad dimension 1 .
however , for computing @xmath655 we do not have such freedom and are forced to choose @xmath4 consistently far from @xmath2 , i.e. so that @xmath656 at a fixed polynomial rate .
in this section we consider the problem of ` unwinding spirals ' or , more precisely , the question of whether a given spiral can be mapped to a unit line segment via a homeomorphism with certain ` metric ' restrictions ; see @xcite for an overview of recent results in this direction .
a classical positive result is that the ` logarithmic spiral ' can be ` unwound ' to a unit line segment by a bi - lipschitz map , see @xcite .
moreover , this is sharp in the sense that if the winding rate is slower than logarithmic , then a simple length estimate shows that the spiral can not be unwound by a bi - lipschitz map . in particular
, the spiral has infinite length and bi - lipschitz images of sets with infinite length must have infinite length .
we are interested in the more general question of whether a spiral can be unwound via a bi - hlder homeomorphism and what restrictions are there on the bi - hlder parameters ?
our main result is in the negative direction : we show that if the bi - hlder map is too close to being bi - lipschitz , then it can not unwind the spiral , where ` too close ' is precisely characterised by the upper box dimension of the spiral .
our result is a simple application of our work on how the assouad spectrum can change under bi - hlder maps and , moreover , we show that one gets strictly better information than if one considers the box or hausdorff dimensions directly . in general a _ spiral
_ @xmath290 is defined to be the set : @xmath657 where @xmath658 is any continuous decreasing real - valued function such that @xmath659 and for convenience we assume that @xmath660 .
we say a spiral is _ convex differentiable _ if @xmath658 is differentiable and its derivative is non - decreasing .
we also say that the spiral has _
monotonic winding _ if the function @xmath661 is decreasing in @xmath59 .
this is similar to the assumption that decreasing sequences have decreasing gaps .
indeed , monotonic winding guarantees that any ray starting at the origin intersects the spiral in a decreasing sequence with decreasing gaps .
if @xmath662 for some @xmath663 , then the resulting spiral is often referred to as the _ logarithmic spiral _
( mentioned above ) .
if @xmath664 as @xmath665 , then the winding is said to be _ sub - exponential_. this is the interesting case since then the spiral has infinite length and thus can not be unwound by a bi - lipschitz homeomorphism . ; and , on the right , a spiral with sub - exponential winding where @xmath666.[holderpic2 ] ] first we prove a dichotomy for spirals with monotonic winding . [ assspiral ] let @xmath290 be a spiral with monotonic winding .
if the winding is sub - exponential , then @xmath667 and otherwise @xmath668 . to prove this result we prove that if the assouad dimension is strictly less than 2 , then the winding must be exponential .
note that this is yet another proof of the fact that spirals with sub - exponential winding can not be unwound to a line segment by a bi - lipschitz homeomorphism , since such maps preserve assouad dimension .
[ spiral ] let @xmath290 be a convex differentiable spiral with monotonic winding and such that @xmath669 .
then for all @xmath5 we have @xmath670 note that such spirals must have sub - exponential winding .
clearly any spiral @xmath290 is homeomorphic to the unit line segment @xmath215 $ ] .
let @xmath360 be a homeomorphism between these two sets and suppose that @xmath360 is also bi - hlder with parameters @xmath285 , i.e. for all @xmath671 we have @xmath672 it follows immediately from the standard results for box dimension , that @xmath673 \ , = \ , 1\ ] ] and so , provided the spiral has upper box dimension larger than its topological dimension , @xmath360 must be quantitatively far away from being bi - lipschitz . in particular , we require @xmath674 observe that if we consider hausdorff dimension here , then we get no information on @xmath343 because the hausdorff dimension of _ any _ spiral is 1 .
fortunately , we can get more information if we consider the assouad dimension .
[ spiralcor ] let @xmath360 be a bi - hlder homeomorphism with parameters @xmath285 mapping a convex differentiable spiral @xmath290 with sub - exponential and monotonic winding to a line segment .
also , assume that @xmath675 .
then @xmath676 and , if @xmath677 , then @xmath678 it follows from theorem [ holderassouadcor ] that @xmath679 where @xmath332 : \dim_\mathrm{a}^{\theta } s=\dim_\mathrm{a } s \right\ } = 1 - \frac{\overline{\dim}_\mathrm{b}s}{2}\ ] ] by theorem [ spiral ] .
therefore , by theorem [ spiral ] , we have @xmath680 and solving for @xmath343 yields @xmath681 as required .
note that corollary [ spiralcor ] gives strictly better information than we get directly from the upper box dimension , provided the upper box dimension of @xmath290 is strictly less than 2 .
otherwise , both estimates reduce to @xmath682 . for the purpose of the following proofs , define @xmath683 and @xmath522 . by definition @xmath360 is decreasing and monotonic winding guarantees that @xmath524 is also decreasing .
clearly @xmath560 for any @xmath684
. we will refer to the @xmath685-neighbourhood of @xmath290 ( or part of @xmath290 ) as a @xmath685-_sausage _ for @xmath686 .
let @xmath687 and observe that there is a unique @xmath688 such that @xmath689 , and a unique @xmath690 such that @xmath691 . if @xmath692 , we see that the @xmath4-sausage of @xmath290 will completely cover the ball @xmath693 , thus the number of @xmath4-balls needed to cover @xmath694 is @xmath695 .
if @xmath696 , the @xmath4-sausage of @xmath290 will completely cover a ball smaller than @xmath693 .
it is easy to see that the smaller ball can be taken to be @xmath697 , thus we need @xmath698 many @xmath4-balls to cover @xmath694 .
since @xmath524 and @xmath360 are continuous decreasing functions and @xmath699 we can deduce that the inverse functions @xmath700 are also decreasing and @xmath701 ( on the appropriate domain ) .
therefore @xmath562 and there exists a unique @xmath702 such that @xmath565 .
we see that the number of @xmath4-balls needed to cover @xmath694 is @xmath703 therefore , if the assouad dimension of @xmath290 is strictly smaller than @xmath350 , then it must be true that for all small enough @xmath122 , the @xmath4 defined above must be such that @xmath568 is uniformly bounded from above .
this reasoning is similar to the case of decreasing sequences .
suppose there exists @xmath569 such that @xmath704 for all small enough @xmath2 and the @xmath4 chosen above .
then we conclude that @xmath705 and so , applying @xmath524 throughout , we get @xmath706 observe that @xmath707 and so @xmath708 and subtracting @xmath709 throughout and taking negatives yields @xmath710 in particular this holds for sufficiently small @xmath122 and so by continuity of @xmath360 we can deduce that @xmath711 for any large enough @xmath46 .
from here it is clear that @xmath712 because the limit holds for integral @xmath46 , and since @xmath52 is a decreasing function the limit holds in general .
therefore @xmath360 , and hence @xmath658 , is at least exponential and since spirals with exponential winding are bi - lipschitz equivalent to a line segment we deduce that @xmath668
. denote the upper box dimension of @xmath290 by @xmath92 and recall that by assumption and theorem [ assspiral ] we know @xmath713
. we will prove that @xmath714 for @xmath715 which , combined with proposition [ bb ] and corollary [ staya ] , proves the result .
fix such a @xmath7 .
if we can find a sequence of @xmath598 such that @xmath716 then we would have : @xmath717 which proves the result , using proposition [ n2d ]
. therefore assume that for all @xmath295 small enough we have the inequality @xmath718 which also implies that for all @xmath295 small enough @xmath719 since @xmath720 and @xmath721 is decreasing .
this also implies that @xmath722 for all large enough @xmath46 .
we assumed that @xmath52 is differentiable and convex and therefore for all large enough @xmath723 we have by the mean value theorem that @xmath724 where @xmath725 $ ] .
in fact it is true that for a suitable constant @xmath726 and all @xmath723 we have @xmath727 for simplicity we write @xmath728 .
we have @xmath729 which gives @xmath730 integrating both sides of this inequality yields @xmath731 and this implies for all large enough @xmath46 that @xmath732 and therefore @xmath733 we now move towards bounding @xmath734 .
first of all we need @xmath735 many @xmath4-balls to cover @xmath736 because the @xmath4-sausage will cover @xmath737 completely .
now consider the subset of @xmath290 given by @xmath738 and decompose part of this set into the disjoint union of the sets @xmath739 over integers @xmath740 + 1{\leqslant}m{\leqslant}[g^{-1}(r)]$ ] , where @xmath60 $ ] denotes the integer part of @xmath741 .
the projection of each @xmath742 onto the real axis contains an interval of length @xmath743 , so to cover @xmath742 we need at least @xmath744 many @xmath4-balls . since distinct sets @xmath742 are at least @xmath4 separated , in order to cover all the sets in the above union we need at least @xmath745-[f^{-1}(r^\theta)]}f([f^{-1}(r^\theta)]+k)\ ] ] many @xmath4-balls .
the sequence @xmath746+k)$ ] for @xmath747-[f^{-1}(r^\theta)]$ ] is decreasing and the minimum of the sequence is @xmath748){\geqslant}f(g^{-1}(r ) ) { \geqslant}r$ ] .
since @xmath524 is also decreasing we see that @xmath749+k)-f([f^{-1}(r^\theta)]+k+1 ) \
{ \geqslant}\ r\ ] ] for @xmath747-[f^{-1}(r^\theta)]-1 $ ] .
therefore we have @xmath749+[g^{-1}(r)]-[f^{-1}(r^\theta)]-k ) \ { \geqslant}\ ( k+1 ) \ , r\ ] ] and applying this inequality to the above sum yields @xmath750-[f^{-1}(r^\theta)]}f([f^{-1}(r^\theta)]+k ) \\ \\ & { \geqslant } & \sum_{k=1}^{[g^{-1}(r)]-[f^{-1}(r^\theta ) ] } k r \\ \\ & = & \frac{1}{2}([g^{-1}(r)]-[f^{-1}(r^\theta)])([g^{-1}(r)]-[f^{-1}(r^\theta)]+1 ) \ , r\\ \\ & \gtrsim & ( g^{-1}(r)-f^{-1}(r^\theta))^2 r\end{aligned}\ ] ] where , in particular , the last @xmath84 holds for all @xmath751 .
therefore we have @xmath752 now we will take into consideration the box dimension of the spiral .
we have @xmath753 where @xmath754 is the length of the rectifiable part of the spiral corresponding to angles @xmath223 to @xmath755 .
we may bound @xmath754 from above by employing the classical length formula : @xmath756 where @xmath757 , @xmath758 . by splitting the integral in this way we obtain @xmath759 }
f(k ) \ + \ \sqrt{2}\end{aligned}\ ] ] where the last inequality comes from the fact that @xmath360 is decreasing . therefore @xmath760 } f(k)/r\ + \ 1/r.\ ] ] using ( [ nicespiralbound ] ) we can bound the middle term above by @xmath761 } f(k)/r \
\lesssim \ \sum_{k=0}^{[g^{-1}(r)+1 ] } k^{-\frac{1-b^-/2}{b^-/2}}/r \ \lesssim
\ r^{-1}g^{-1}(r)^{1-\frac{1-b^-/2}{b^-/2}}.\ ] ] therefore @xmath762 since the ( upper ) box dimension of @xmath290 is @xmath92 we can find a sequence of @xmath598 such that @xmath763 since @xmath764 we can assume @xmath765 , and then either the first term or the second term is @xmath766 for infinitely many @xmath462 .
if this is true for the first term then by our initial observation we have @xmath767 .
if this is true for the second term , then we deduce that for infinitely many @xmath462 we have @xmath768 recalling the lower bound ( [ funnybound ] ) and the assumption that @xmath769 , we conclude that @xmath770 which completes the proof .
self - affine sets are an important class of fractal subsets of euclidean space .
they are attractors of finite iterated function systems consisting of affine contractions . due to the fact that affine maps can scale by different amounts in different directions ,
the dimension theory of self - affine sets is much more complicated than their self - similar counter parts ; where the defining maps are similarities .
self - affine sets have been studied intensively for over 30 years and have important connections with many other fields , such as non - conformal dynamical systems . an important contribution to this area was bedford and mcmullen s independent treatments of a restricted class of self - affine sets @xcite .
these ` bedford - mcmullen carpets ' are planar self - affine sets where the affine maps all have the same linear part ; which is a simple diagonal matrix .
these sets are much simpler than general self - affine sets , but still display many of the key features : the maps scale by different amounts in the two principle directions and so the constituent pieces become increasingly elongated as one iterates the construction . due to their simple and explicit form ,
the dimension theory of bedford - mcmullen carpets is rather well - developed .
indeed , bedford and mcmullen computed their hausdorff and box dimensions in the mid-1980s , mackay computed their assouad dimensions in 2011 @xcite , and fraser computed their lower dimensions in 2014 @xcite . in this section we give explicit formulae for the assouad and lower spectra .
we begin by recalling the construction .
fix @xmath771 with @xmath772 and divide the unit square @xmath215 ^ 2 $ ] into an @xmath773 regular grid .
let @xmath774 and @xmath775 and label the @xmath776 rectangles in the regular grid by @xmath777 counting from bottom left to top right .
choose a subset of the rectangles @xmath778 of size at least 2 and for each @xmath779 , associate a contraction @xmath780 ^ 2 \to [ 0,1]^2 $ ] defined by @xmath781 let @xmath782 be the set of infinite words over @xmath783 and let @xmath784 ^ 2 $ ] be the canonical map from the symbolic space to the geometric space defined by @xmath785 ^ 2\big).\ ] ] the set @xmath786 is the bedford - mcmullen carpet .
we thus obtain a symbolic coding of the set @xmath26 , but we note that some points may not have a unique code .
the set @xmath26 can be viewed in several other contexts .
for example , @xmath26 is the attractor of the _ iterated function system _ ( ifs ) @xmath787 , i.e. it is the unique non - empty compact set satisfying @xmath788 in the context of hyperbolic dynamical systems , bedford - mcmullen carpets are also of particular interest . identifying @xmath215 ^ 2 $ ] with the 2-torus in the natural way , @xmath26 is a ( forward and backward ) invariant repeller for the hyperbolic toral endomorphism @xmath789 and @xmath790 and on the right @xmath791 and @xmath792.[carpets ] ] in order to state the dimension results mentioned above we need some more notation .
let @xmath793 be the projection onto the first coordinate .
for @xmath794 , let @xmath795 be the number of chosen rectangles in the @xmath462th column , let @xmath796 and @xmath797 .
[ carpetdimensions ] the assouad , box , hausdorff and lower dimensions of a bedford - mcmullen carpet @xmath26 are given by @xmath798 @xmath799 @xmath800 and @xmath801 observe that the term @xmath802 appears in three of the dimension formulae .
this is the dimension of the self - similar projection of @xmath26 onto the first coordinate .
where this term appears , the second term relates to the dimension of fibres : for the assouad dimension , @xmath803 is the maximal fibre dimension ( also given by a self - similar set ) , for the lower dimension , @xmath804 is the minimal fibre dimension and , for the box dimension , the second term can be interpreted as an ` average fibre dimension ' , observing that @xmath805 is the arithmetic average of the @xmath491 .
the box , hausdorff , assouad and lower dimensions coincide if and only if @xmath491 is the same for all @xmath806 , commonly referred to as the _ uniform fibres case _ , and otherwise they are all distinct , see @xcite .
before we discuss our main results we provide a minor addition to theorem [ carpetdimensions ] .
the notable omission is the modified lower dimension and we point out that this can be computed as a direct consequence of results of ferguson - jordan - shmerkin @xcite and the formula for the lower dimension given by fraser @xcite .
perhaps surprisingly the modified lower dimension is equal to the _ hausdorff _ dimension and not the lower dimension .
[ modifiedlower ] the modified lower dimension of a bedford - mcmullen carpet @xmath26 is given by @xmath807 moreover , it is also true that if @xmath26 is a self - affine carpet of the type considered by baraski @xcite or lalley - gatzouras @xcite , then @xmath808 we refer the readers to @xcite for the precise formulae . since the modified lower dimension of a compact set
is always bounded above by the hausdorff dimension we need only prove the lower bound .
it suffices to prove that for any @xmath143 we can find a subset of @xmath26 which has lower dimension within @xmath424 of the hausdorff dimension . via an elegant application of stirling s formula , ferguson - jordan - shmerkin ( *
* lemma 4.3 ) showed that one can always find a subset of such a carpet generated by a subsystem of an iterate of the original ifs which both has hausdorff dimension within @xmath424 of the original hausdorff dimension and has the key additional property that the fibres are uniform .
then it follows from ( * ? ? ?
* theorem 2.13 ) that the lower dimension and the hausdorff dimension of the subset coincide , thus proving the result .
the result for more general carpets follows in exactly the same way and we omit the details .
our main result on self - affine carpets is the following theorem , which gives explicit expressions for the assouad and lower spectra .
[ carpetsspectra ] let @xmath26 be a bedford - mcmullen carpet .
then for all @xmath5 we have @xmath809 and @xmath810 moreover , both both spectra have only one phase transition which occurs at @xmath811 and both are analytic and monotonic in the interval @xmath812 and constant in @xmath813 .
an immediate consequence of theorem [ carpetsspectra ] is that for any bedford - mcmullen carpet with non - uniform fibres the assouad spectrum is _ strictly smaller _ than the general upper bound given by proposition [ bb ] for all @xmath814 .
also , the unique phase transition always occurs to the right of the phase transition in the general bound , i.e. @xmath815 these observations follow by simple manipulation of the various dimension formulae . .
on the left : @xmath816 , @xmath790 , @xmath817 , @xmath818 , @xmath819 and @xmath820 . on the right : @xmath791 , @xmath792 , @xmath821 , @xmath822 , @xmath823 and @xmath824 .
the bounds from proposition [ bb ] and and [ bbl ] are shown with dashed lines .
[ carpetsspectrum ] ] a , perhaps surprising , corollary of theorem [ carpetsspectra ] is that the lower spectra is not uniformly bounded above by the hausdorff or modified lower dimension .
this might be surprising because the lower spectrum is based on the lower dimension , which _ is _ bounded above by the hausdorff and modified lower dimension for compact sets .
the assouad spectrum does not violate any of the general inequalities between the assouad dimension and the other common dimensions since it is uniformly bounded below by the upper box dimension , see proposition [ bb ] .
in fact it follows that for any bedford - mcmullen carpet with non - uniform fibres there will be an open interval @xmath825 within which the lower spectrum is strictly larger than the hausdorff and modified lower dimension .
this follows since for such carpets the lower spectrum is continuous and converges to the box dimension as @xmath11 which is strictly larger than the hausdorff and modified lower dimension .
this goes someway to justifying the sharpness of proposition [ bbl ] .
theorem [ carpetsspectra ] proves that the ratio @xmath826 is a bi - lipschitz invariant within the class of bedford - mcmullen carpets . to the best of our knowledge
this has not been observed before . since
sets are bi - lipschitz equivalent to themselves this invariant can be useful in determining what kinds of affine ifss give rise to a particular carpet .
for example , let @xmath26 be a bedford - mcmullen carpet generated by an ifs using an @xmath773 grid for some @xmath827 .
assume that @xmath26 does not have uniform fibres and that it can also be viewed as a bedford - mcmullen carpet generated by an ifs using an @xmath828 grid for some @xmath829 .
then we may conclude that @xmath830 the question of when two bedford - mcmullen carpets are bi - lipschitz equivalent has been considered by li - li - miao @xcite , where they give positive results assuming some natural conditions , including that @xmath831 are constant . in some sense , our results compliment these results as we provide a mechanism for proving negative results .
we demonstrate this by the following example .
[ t2 ] we can find two topologically equivalent bedford - mcmullen carpets with the same box counting , lower and assouad dimensions , but different spectra . therefore @xmath20 and @xmath26 are not bi - lipschitz invariant despite this not being revealed by knowledge of the other dimensions .
we construct the first bedford - mcmullen carpet such that @xmath832 , @xmath833 , @xmath834 , @xmath835 , @xmath836 , @xmath820 and the second carpet such that @xmath832 , @xmath837 , @xmath838 , @xmath839 , @xmath840 , @xmath820 .
moreover , we may choose the rectangles satisfying these parameters so that they do not touch each other , rendering both carpets totally disconnected and thus homeomorphic .
it follows from theorem [ carpetdimensions ] that for both carpets @xmath841 @xmath842 and @xmath843 but , since the ratio @xmath844 is not the same for both constructions , they necessarily have different spectra . for the above examples , we guaranteed the spectra would be different by ensuring that @xmath845 .
this turns out to be necessary in finding such examples .
the following corollary expresses this precisely and follows immediately by rearranging the formulae given in theorems [ carpetdimensions ] and [ carpetsspectra ] .
the dimension spectra of bedford - mcmullen carpets are completely determined by the ratio @xmath846 and the box , lower and assouad dimensions of the carpet .
more precisely , we have @xmath847 and @xmath848 let @xmath849 and @xmath295 be small .
define @xmath850 to be the unique natural numbers satisfying @xmath851 and @xmath852 the approximate square ` centred ' at @xmath853 with ` radius ' @xmath295 is defined by @xmath854 and the geometric projection of this set , @xmath855 , is a subset of @xmath26 which contains @xmath856 and naturally sits inside a rectangle which is ` approximately a square ' in that it has a base with length @xmath857 $ ] and height @xmath858 $ ] .
one obtains equivalent definitions of the assouad and lower spectra if one replaces @xmath859 by @xmath860 , i.e. we may use approximate squares instead of balls .
our subsequent analysis breaks naturally into two cases , illustrated by the following lemma .
let @xmath861 and @xmath5
. then 1 .
if @xmath862 , then @xmath863 2 . if @xmath864 , then @xmath865 this follows immediately from the definitions of @xmath866 and @xmath867 .
[ key1 ] let @xmath868 $ ] .
then @xmath869 let @xmath870 and @xmath184 .
when one looks at the approximate square @xmath871 one finds that it is made up of several horizontal strips with base length the same as that of the approximate square , i.e. @xmath872 , and height @xmath873 which is considerably smaller , but still larger than @xmath874 since @xmath875 .
these strips are images of @xmath26 under @xmath876-fold compositions of maps from @xmath787 and so to keep track of how many there are , one counts rectangles in the appropriate columns for each map in the composition .
the total number of such strips is seen to be : @xmath877 we want to cover these strips by sets of diameter @xmath96 and so we iterate the construction inside each horizontal strip until the height is @xmath878 , which is approximately @xmath96 .
this takes @xmath879 iterations and , this time , for every iteration we pick up @xmath880 smaller rectangles , rather than just those inside a particular column . at this stage
we are left with a large collection of rectangles with height approximately @xmath96 and base @xmath881 , which is somewhat larger .
we now iterate inside each of these rectangles until we obtain a family of rectangles each with base of length @xmath882 .
this takes a further @xmath883 iterations .
we can then cover the resulting collection by small sets of diameter @xmath96 , observing that sets formed by this last stage of iteration can be covered simultaneously , provided they are in the same column .
therefore , for each of the last iterations we only require a factor of @xmath884 more covering sets .
putting all these estimates together , yields @xmath885 taking logs and dividing by @xmath886 yields @xmath887 letting @xmath107 yields the required upper bound , noting that these estimates are uniform in @xmath888 .
the required lower bound also follows by noting that if we choose @xmath889 , such that @xmath890 for all @xmath891 , then the only appearance of an inequality in the above argument is replaced by equality and observing that all of our covering estimates were optimal up to multiplicative constants .
[ key2 ] let @xmath892 .
then @xmath893 this proof is similar to the proof of proposition [ key1 ] .
let @xmath870 and @xmath184 .
we proceed as before , but this time one obtains a family of horizontal strips with height approximately @xmath96 after @xmath894 steps , which is before the bases become smaller than @xmath2 , since in this case @xmath895 .
the effect is that the ` middle term ' above ( concerning powers of @xmath880 ) is not required .
the horizontal strips are then covered as before yielding @xmath896 taking logs and dividing by @xmath897 yields @xmath898 letting @xmath107 yields the required upper bound , noting that these estimates are uniform in @xmath888 .
the required lower bound again follows by noting that if we choose @xmath888 , such that @xmath890 for all @xmath891 , then the only appearance of an inequality in the above argument is replaced by equality and observing that all of our covering estimates were optimal up to multiplicative constants .
we now turn to the lower spectrum , which can be handled similarly and so only the key points in the proofs are given .
[ key3 ] let @xmath868 $ ] .
then @xmath899 let @xmath853 and @xmath184 .
proceeding as above , one obtains @xmath900 but this time we continue by considering uniform lower bounds : @xmath901 taking logs and dividing by @xmath886 yields @xmath902 letting @xmath107 yields the required lower bound .
the required upper bound also follows by noting that if we choose @xmath889 , such that @xmath903 for all @xmath891 , then the only appearance of an inequality in the above argument is replaced by equality and observing that all of our covering estimates were optimal up to multiplicative constants .
[ key4 ] let @xmath892 .
then @xmath904 let @xmath853 and @xmath184 .
following the strategy of the proof of lemma [ key2 ] but this time considering uniform lower bounds we get : @xmath905 which yields @xmath906 letting @xmath107 yields the required lower bound .
the required upper bound again follows by noting that if we choose @xmath888 , such that @xmath903 for all @xmath891 , then the only appearance of an inequality in the above argument is replaced by equality and observing that all of our covering estimates were optimal .
theorem [ carpetsspectra ] follows immediately by combining lemmas [ key1 ] , [ key2 ] , [ key3 ] , [ key4 ] and the known formulae for the box , assouad and lower dimensions .
bedford - mcmullen carpets are the simplest , and probably most studied , family of self - affine sets .
however the assouad and lower dimensions are known for considerably more general families , in particular , self - affine classes where the matrices do not need to be constant , see for example the lalley - gatzouras class @xcite or baraski class @xcite . for more details on the assouad dimensions of these carpets ,
we refer the reader to @xcite .
also see @xcite for results on the assouad dimension of self - affine carpets with no underlying grid structure
. it would be interesting to also compute the assouad and lower spectra for these more general families , but this seems to be considerably more complicated .
we will attempt to explain this extra complication in this section via a simple heuristic , and also try to indicate what different phenomena we expect to find .
we will consider lalley - gatzouras carpets , which are similar to the bedford - mcmullen carpets in that they are generated by affine maps based on diagonal matrices and the matrices all contract more strongly in the vertical direction than in the horizontal direction .
moreover , the projection onto the first coordinate is a self - similar set satisfying the open set condition and again this set plays a key role in the dimension theory of the self - affine set .
the difference is that the family of matrices need not be constant and so the columns can have varying widths and within each column the rectangles can have varying heights and be distributed less rigidly . ]
the assouad dimension of a lalley - gatzouras carpet is given by a formula similar to that used in the bedford - mcmullen case .
indeed , it is equal to the dimension of the projection onto the first coordinate plus the largest dimension corresponding to a particular column . this is related to ` tangent sets ' having a ` product structure ' , see @xcite .
this is also seen when computing the assouad spectrum of bedford - mcmullen carpets .
when @xmath907 , the approximate squares have a ` discretised product structure ' viewed at the pair of scales @xmath2 and @xmath96 .
however , if @xmath908 , then one has to iterate the construction for a longer time before the height of the basic sets is roughly @xmath96 .
this means the approximate squares lose their discrete product structure and the box dimension starts to contribute to the formula for the spectrum .
we would expect a similar phenomena to occur for lalley - gatzouras sets , but different columns may have a different _ threshold _ for witnessing a product structure .
for example , suppose the column with the largest dimension , i.e. , the one which contributes to the assouad dimension consisted only of @xmath909 rectangles .
this would give rise to a discrete product structure ( and so the assouad spectrum would be equal to the assouad dimension ) for @xmath910 , but for smaller values of @xmath7 the spectrum would decay towards the box dimension . however , if there was another column which corresponded to a smaller dimension but which consisted only of @xmath911 rectangles , then this column would allow a discrete product structure to appear on small scales for @xmath912 .
this would mean that the assouad spectrum was at least equal to the dimension of the projection plus the dimension corresponding to the second column for @xmath913 .
we expect that this could lead to phase transitions at @xmath914 and @xmath915 ; recall that the assouad spectrum in the bedford - mcmullen case always has precisely one phase transition .
moreover , by carefully arranging many columns which have smaller dimension as the log - log eccentricity increases could lead to the same phenomena many times , perhaps leading to arbitrarily many phase transitions .
computing the dimension spectra in this setting seems like a delicate and interesting problem . even in the simpler subcase ( which we allude to here ) where the matrices are constant within each column seems challenging and probably a good place to start .
the advantage of this case is that each column has a well - defined log - log eccentricity .
self - similar and self - conformal sets are another important class of fractals generated by iterated function systems .
they are simpler than the self - affine sets considered in the previous section in that the defining maps locally scale by the same amount in every direction .
if the constituent pieces making up the set do not overlap too much , then the situation is quite straightforward and it follows from standard results that the lower and assouad dimensions coincide , see for example ( * ? ? ?
* corollary 2.11 ) , rendering the dimension spectra not interesting .
however , if there are overlaps in the construction then fraser proved that the assouad dimension can be strictly larger than the upper box dimension , but the lower dimension always equals the upper box dimension , see ( * ? ? ?
* section 3.1 and corollary 2.11 ) .
therefore , the interesting problem is to consider the assouad spectrum in the case of complicated overlaps .
this is not straightforward and we do not obtain precise results , but we show how to obtain non - trivial bounds in terms of the local dimensions of related gibbs measures .
this technique was used by fraser and jordan to study the assouad dimension of self - affine carpets where the self - similar set in the projection has overlaps , see @xcite .
let @xmath916 be a bounded simply connected non - empty open set .
a map @xmath917 is called a _ conformal map _ if it is differentiable and the jacobian @xmath918 satisfies @xmath919 for all @xmath920 and @xmath921 .
let @xmath922 be a finite collection of conformal maps on a common domain @xmath923 , and assume that each maps @xmath924 into itself . for convenience we will extend each of the maps to the boundary of @xmath924 by continuity .
furthermore , assume that each map is a bi - lipschitz contraction , i.e. @xmath925 there is a unique non - empty compact set @xmath26 satisfying @xmath926 which is called the _ self - conformal set _ associated with the iterated function system ( ifs ) @xmath922 .
see ( * ? ? ?
* chapter 9 ) for more details on the theory of attractors of ifss .
if for each map the jacobian @xmath927 is independent of @xmath46 , then the maps @xmath290 are _ similarities _ and the set @xmath26 is called _ self - similar_. here the situation is slightly simpler , but is subsumed into the following analysis of the self - conformal setting .
let @xmath928 be the set of infinite words over @xmath929 and let @xmath930 be the canonical map from the symbolic space to the geometric space defined by @xmath931 evidently @xmath932 and we thus obtain a symbolic coding of the set @xmath26 , but we note that some points may have many codes .
write @xmath933 for the set of all finite words over @xmath929 . for @xmath934 write @xmath935
\ = \ \ { \mathbf{i } \mathbf{j } : \mathbf{j } \in \mathcal i^\infty \}\ ] ] for the cylinder consisting of all infinite words beginning with the finite word @xmath936 and @xmath937 if @xmath938 , then it is well - known that conformal maps satisfy the _ bounded distortion principle_. this means that for @xmath939 and @xmath940 @xmath941 with the implied constants independent of @xmath393 and @xmath369 .
this means that for each @xmath939 , there is a constant @xmath942 such that @xmath943 where we abuse notation slightly by writing @xmath944 for the determinant of the jacobian @xmath945 and @xmath946 for the diameter of the set @xmath947 .
the constant @xmath948 could be the determinant of the jacobian derivative of @xmath949 at the fixed point of @xmath949 , or the upper ( or lower ) lipschitz constant of @xmath949 .
for definiteness let @xmath950 be the upper lipschitz constant of @xmath949 . to get the bounded distortion principle to hold in @xmath502 ( which we will need ) , we assume hlder continuity of the derivatives @xmath951 .
we emphasise that this extra assumption is not required for @xmath938 .
the _ topological pressure _ of the ifs is defined by @xmath952 where the limit is easily seen to exist by submultiplicativity of the upper lipschitz constant for example .
actually we have quasi - multiplicativity in the sense that for @xmath953 @xmath954 the topological pressure is a continuous decreasing function of @xmath955 and there is a unique zero , i.e. a unique @xmath956 such that @xmath957 .
from now on we will fix @xmath412 to be the unique zero of the topological pressure . a classical result in dimension theory
states that @xmath958 and if the _
open set condition _ is satisfied , then one has equality .
this dimension formula is often referred to as _
bowen s formula _
goes back to bowen @xcite and ruelle @xcite .
we will be focusing on the case where the open set condition fails , but the general upper bound given by @xmath412 is still significant .
we recall that the assouad dimension may exceed @xmath412 .
another classical result , this time from the thermodynamic formalism , states that there exists a unique borel probability measure @xmath959 on @xmath960 ( equipped with the product topology ) satisfying @xmath961 ) \asymp
\text{lip}(\mathbf{i})^s.\ ] ] this measure is the _ gibbs measure _ corresponding to the potential @xmath962 and is a fundamental object in the thermodynamical formalism and related problems from statistical physics , see @xcite . let @xmath963 be the push forward of @xmath959 onto @xmath26 .
the measure @xmath964 is a borel probability measure and is intimately related to the dimension theory of @xmath26 .
the number @xmath965 will play a central role in our analysis .
this number is related to the fine structure of @xmath964 and can be expressed in terms of more familiar dimensional quantities , such as the local dimension or @xmath966-spectrum .
we record these connections to put our results in context
. the lower local dimension of a borel probability measure @xmath967 at a point @xmath46 in its support is defined by @xmath968 with the upper local dimension defined similarly with @xmath969 in place of @xmath970 .
it is clear that @xmath971 the @xmath966-spectrum gives a coarse indication of the global fluctuations of a measure and are a standard tool in multifractal analysis and information theory . for @xmath972 and @xmath295
let @xmath973 the upper @xmath966-spectrum of @xmath964 is defined by @xmath974 it is straightforward to see that @xmath975 is non - decreasing and concave in @xmath976 . by following the argument in (
* lemma 2.1 ) , one may prove that @xmath499 is actually equal to the slope of the asymptote as @xmath977 , i.e. , @xmath978 moreover , if the _ multifractal formalism _ holds for a sequence of @xmath979 , then @xmath980 see @xcite for more details on the multifractal formalism .
if @xmath26 is self - similar , then the gibbs measure @xmath964 is a self - similar measure and in this case we know more .
for example , it was shown by peres and solomyak @xcite that the @xmath966-spectrum exists , and feng @xcite proved that the multifractal formalism holds whenever @xmath981 is differentiable .
this must happen for a sequence of @xmath979 since @xmath981 is non - decreasing and concave .
it seems highly likely to us that these results extend to gibbs measure on self - conformal sets , but we omit further discussion .
we can now state our main result , which gives a non - trivial upper bound for the assouad spectrum of @xmath26 .
[ ssmain ] let @xmath26 be a self - conformal set in @xmath21 satisfying the bounded distortion principle and let @xmath412 and @xmath499 be as above .
then @xmath982 in certain situations we can say more .
for simplicity we will specialise to @xmath983 and the self - similar setting , although some results also hold in higher dimensions and for self - conformal sets .
the main result of @xcite was that for a self - similar subset of the line , either the assouad and box dimensions coincide , or the assouad dimension is 1 .
this was generalised to the self - conformal setting in @xcite ; see also @xcite .
the precise condition determining this dichotomy is also known to be the _ weak separation condition _ : if the weak separation condition is satisfied , then the assouad , hausdorff and box dimensions of @xmath26 coincide , and otherwise the assouad dimension is 1 .
the weak separation condition was introduced by lau and ngai @xcite and developed by zerner @xcite . for the formal definition and basic properties
we refer the reader to @xcite but , roughly speaking , it means the overlaps are ` simple ' in the sense that either small cylinders overlap exactly or are somewhat separated .
a famous folklore conjecture on self - similar sets is that if the semigroup generated by the defining maps is free ( i.e. , there are no exact overlaps ) , then the hausdorff and box dimension of the associated self - similar set should be given by @xmath984 .
significant progress has been made on this conjecture in recent years , in particular through the work of hochman @xcite .
hochman proved that if the dimension is not given by the expected value , then there must be ` super - exponential concentration of cylinders ' .
the only known mechanism for such concentration is exact overlaps and if the parameters defining the maps in the ifs are algebraic then the two phenomena are equivalent .
shmerkin @xcite recently extended hochman s work to include the @xmath966-spectrum , which in turn allows us to obtain a precise result in an important and general setting . for self - similar sets the measure
@xmath964 is self - similar and it therefore follows from ( * ? ? ?
* theorem 6.6 ) that the @xmath966-spectrum of @xmath964 is affine and given by @xmath985 , provided the ifs does not have ` super - exponential concentration of cylinders ' .
in particular , this implies that @xmath986 . for the precise definition of ` super - exponential concentration of cylinders '
we refer the reader to @xcite .
roughly speaking it means that sequences of construction cylinders of level @xmath178 cluster towards each other super - exponentially fast as @xmath987 .
we can glean the following corollary by combining our theorem [ ssmain ] with the work of hochman @xcite and shmerkin @xcite and the main result of @xcite .
[ sscor ] let @xmath988 be a self - similar set . if the ifs does not have ` super - exponential concentration of cylinders ' , then for all @xmath5 @xmath989 in particular , this is satisfied if the semigroup generated by the defining maps is free and the parameters defining the maps are algebraic .
if the semigroup is not free , but the weak separation property is satisfied , then for all @xmath5 @xmath990 finally we note that if a self - similar measure has a convolution structure , then one may write it as the convolution of : a self - similar measure with no overlaps , and another measure .
then , using the fact that taking convolutions does not lower the value of @xmath499 , one can conclude that @xmath991 .
this means that we get a genuinely better upper bound than that given by proposition [ bb ] in the region : @xmath992 which is non - empty provided @xmath993 in particular , if @xmath994 then @xmath991 guarantees an improved bound . for more details of the convolution argument ,
see ( * ? ? ?
* lemma 2.4 ) . .
in both cases the weak separation property is assumed to fail , rendering the assouad dimension equal to 1 . on the left
we have chosen @xmath995 and on the right we have chosen @xmath996 .
the familiar upper and lower bounds on the assouad spectrum from proposition [ bb ] are shown with dashed lines.[selfsimeg ] ] let @xmath42 , @xmath184 and @xmath5 .
for @xmath997 , let @xmath998 be the @xmath685-_stopping _ , defined by @xmath999 where @xmath1000 is @xmath936 with the final letter removed .
let @xmath1001 all of the sets @xmath1002 contributing to @xmath1003 lie completely inside the ball @xmath1004 for some fixed constant @xmath1005 which is independent of @xmath2 and @xmath46 .
this follows from the bounded distortion principle .
furthermore , the sets @xmath1002 contributing to @xmath1003 all carry ( symbolic ) weight @xmath1006 ) \asymp
\text{lip}(\mathbf{i})^s \asymp r^s$ ] and so @xmath1007 it also follows from the definition of @xmath499 that @xmath1008 where , crucially , the implied constants are independent of @xmath46 , and therefore @xmath1009 we wish to cover @xmath859 by sets of diameter less than or equal to @xmath96 and to do this we iterate the construction inside each of the cylinders @xmath1002 contributing to @xmath1003 .
more precisely , we consider the stopping @xmath1010 and then decompose each set @xmath1002 individually into the sets @xmath1011 all of these sets have diameters @xmath1012 and so can themselves be covered by @xmath1013 many sets of diameter less than or equal to @xmath96 .
we claim that @xmath1014 to see this note that @xmath1015 the fact that the final term is finite follows since @xmath1016 by the definition of @xmath412 .
thus @xmath1017 and so by applying lemma [ n2d ] and then letting @xmath1018 and @xmath1019 we conclude that @xmath1020 as required .
mandelbrot percolation , first appearing in the works of mandelbrot in the 1970s as a model for intermittent turbulence @xcite , is one of the most well - studied and famous random fractal constructions and can be defined as follows .
fix @xmath1021 , @xmath1022 and @xmath1023 .
let @xmath1024^d$ ] and divide @xmath1025 into @xmath1026 identical closed smaller cubes by dividing each side into @xmath178 equal length closed intervals .
for each of the @xmath1026 small cubes either ` select it ' with probability @xmath23 or ` reject it ' with probability @xmath1027 .
do this independently for each cube and let the union of all the selected cubes be denoted by @xmath1028 .
assuming @xmath1028 is not empty , then we repeat the ` dividing and selecting ' process for each of the cubes making up @xmath1028 _
independently_. this process is repeated inductively and , assuming non - extinction , we obtain a nested sequence of compact sets @xmath1029 . the intersection of all @xmath1029 is denoted by @xmath1030 and is a randomly generated compact subset of the unit cube .
the random set @xmath26 is the set of interest and is the known as the limit set of mandelbrot percolation . and @xmath1031 and on the right @xmath1032 and @xmath1033.[percolation ] ]
it is well known that if @xmath1034 , then there is a positive probability of non - extinction and , conditioned on non - extinction , almost surely @xmath1035 it can be derived from the work of jrvenp - jrvenp - mauldin @xcite on upper and lower porosities that if @xmath1034 then , conditioned on non - extinction , almost surely @xmath1036 which are notably independent of @xmath23 .
see also fraser - miao - troscheit @xcite .
our main result in this section gives the almost sure assouad spectrum of mandelbrot percolation .
[ mp ] suppose @xmath1034 .
then , conditioned on non - extinction , almost surely @xmath1037 for all @xmath111 . surprisingly , the assouad spectrum _ does _ depend on @xmath23 and does not reach the assouad dimension . since the spectrum is constantly equal to the ( upper ) box dimension , mandelbrot percolation provides another family of examples showing that the lower bound in proposition [ bb ] is sharp .
the proof will be split up into several steps and first we need to discuss some basic properties of branching processes .
let @xmath25 be a non - negative integer - valued random variable with finite @xmath369-th moment for any @xmath1038 , that is , @xmath1039<\infty$ ] for any @xmath1038 .
we can define a stochastic process @xmath1040 @xmath1041 by letting @xmath1042 and for any @xmath1043 , we define @xmath1044 where @xmath1045 are independent identically distributed random variables with the same distribution as @xmath25 .
we adopt the convention that @xmath1046 .
the process @xmath1040 is known as a _ galton - watson process_. we will need the following result , which expresses the @xmath369th moments of @xmath1040 as an explicit polynomial . although the result for the first moment is well - known , we could not find a reference for the @xmath369th moment and so include the details for completeness .
[ br ] for any positive integers @xmath1047 we have : @xmath1048=\sum_{i=1}^{k}a_{ki}\mu^{in}\ ] ] where @xmath1049 $ ] and the coefficients @xmath1050 can be computed explicitly .
we will proceed by induction , first noting the well - known case when @xmath1051 . for any @xmath1052 we have @xmath1053=\mathbb{e}\left[\mathbb{e}\left[\sum_{i=1}^{y_n}x_i\bigg\vert y_n\right]\right]=\mathbb{e}[x]\mathbb{e}[y_n]=\mu \mathbb{e}[y_n]=\dots=\mu^{n+1}\mathbb{e}[y_0]=\mu^{n+1}.\ ] ] suppose that for any @xmath178 we have @xmath1054=\sum_{i=1}^{k}a_{ki}\mu^{in}\ ] ] for @xmath1055 with @xmath1056 an integer . then @xmath1057=\mathbb{e}\left[\mathbb{e}\left[\left(\sum_{i=1}^{y_n}x_i\right)^{k+1}\bigg|y_n\right]\right]\ ] ]
and we can expand the @xmath1058 into polynomial terms of the following form : @xmath1059 where @xmath1060 is an integer , @xmath1061 are @xmath412 different integers from @xmath152 to @xmath1040 , and @xmath1062 are positive integers which sum to @xmath1063 .
if we write @xmath1064=\mu_m$ ] , then by independence we get @xmath1065=\mu_{k_1}\mu_{k_2}\dots \mu_{k_s}\ ] ] and so the expectations do not really depend on the @xmath462-s but rather on the @xmath369-s .
the expansion of @xmath1066\ ] ] can be written as the sum of terms of the form : @xmath1067 where the binomial coefficients take the value @xmath223 when @xmath1068 .
we can arrange this sum with respect to the powers of @xmath1040 since the binomial coefficients are polynomials in @xmath1040 .
we obtain an expression of the form @xmath1066=\sum_{i=0}^{k+1}b_{i}y_n^i\ ] ] and it is clear that @xmath1069 since there is only one way to write @xmath1063 as the sum of @xmath1063 positive integers , and the corresponding value of @xmath1070 is @xmath1071 .
we then have @xmath1072=\sum_{i=0}^{n}b_{i}y_n^i = \sum_{i=0}^{n-1}b_{i}y_n^i+\mu^{n}y_n^{n}.\ ] ] we will write the expectation of the first term on the right side as @xmath1073=\sum_{i=1}^{n-1}b_{i}\mathbb{e}[y_n^i].\ ] ] then @xmath1074 & = & \mathbb{e}\left[\mathbb{e}\left[\left(\sum_{i=1}^{y_{n-1}}x_i\right)^{n}\bigg|y_{n-1}\right ] \right ] \\ \\ & = & \sum_{i=0}^{n}b_{i}\mathbb{e}[y_{n-1}^i]\\ \\ & = & k(n-1)+\mu^{n}\mathbb{e}[y_{n-1}^{n } ] \\
\\ & = & k(n-1)+\mu^{n}k(n-2)+\mu^{2n}k(n-3)+\dots + \mu^{n(n-1)}\mathbb{e}[y_1^n ] \\ \\ & = & k(n-1)+\mu^{n}k(n-2)+\mu^{2n}k(n-3)+\dots + \mu^{n(n-1)}\mu_n \\ \\ & = & \sum_{i=1}^{n-1}b_i\left(\mathbb{e}[y_{n-1}^i]+\mu^{n}\mathbb{e}[y_{n-2}^i]+\dots + \mu^{n(n-2)}\mathbb{e}[y_{1}^i]\right)+\mu^{n(n-1)}\mu_n \\ \\ & = & \sum_{i=1}^{n-1}\sum_{j=1}^{i}b_ia_{ij}\left(\mu^{j(n-1)}+\mu^{j(n-2)+n}+\dots + \mu^{j+n(n-2)}\right ) + \mu^{n(n-1)}\mu_n \\ \\ & = & \sum_{i=1}^{n-1}\sum_{j=1}^{i}b_ia_{ij}\frac{\mu^{j(n-1)}-\mu^{n(n-1)}}{1-\mu^{n - j}}+\mu^{n(n-1)}\mu_n \\ \\ & = & \sum_{j=1}^{n-1}\sum_{i = j}^{n-1}\frac{b_ia_{ij}}{1-\mu^{n - j}}\mu^{j(n-1)}+\left(\mu_n-\sum_{i=1}^{n-1}\sum_{j=1}^{i}\frac{b_ia_{ij}}{1-\mu^{n - j}}\right)\mu^{n(n-1)}\end{aligned}\ ] ] by our inductive assumption ( [ ind11 ] ) , all the coefficients of @xmath1075 do not depend on @xmath178 , therefore we can of course write @xmath1076=\sum_{i=1}^{n}a_{ni}\mu^{in}\ ] ] for @xmath1077 which do not depend on @xmath178 , completing the proof .
in particular , this theorem shows that @xmath1078 $ ] can be written as a polynomial in @xmath1079 with degree ( at most ) @xmath1080 and the coefficients of the polynomial depend on @xmath1080 but not @xmath178 .
write @xmath1081 to denote the unique value of the box dimension which occurs with positive probability , and fix @xmath5 .
let @xmath1082 be the number of cubes selected at step @xmath1083 .
we call the collection of selected cubes at level @xmath412 the _ @xmath412-layer cubes_. they are of side length @xmath1084 .
let @xmath1083 and @xmath1085 .
we refer to an @xmath412-layer cube as a _
@xmath1086-cube _ if there are at least @xmath1087 cubes selected at layer @xmath1088 $ ] which lie inside it .
the selected cubes can be considered independently and so the probability of at least one @xmath1086-cube existing is @xmath1089 where @xmath1090 is the probability that a particular @xmath412-layer cube is a @xmath1086-cube which , by the statistical self - similarity of the process , is the same as the probability that @xmath26 itself has more than @xmath1087 many @xmath369-layer cubes , i.e. @xmath1091 .
we can view the construction of the mandelbrot percolation in the context of the galton - watson process discussed in the previous section . in this case the branching process @xmath1040 is constructed from the binomial random variable @xmath1092 , where we note that @xmath1093 . by the markov inequality , @xmath1094}{\geqslant}m ) \ { \leqslant}\ \frac{\mathbb{e}[y_{[k]}^n]}{m^n}\ ] ] for any @xmath1095 .
therefore we have the following bound on the probability that there exists at least one @xmath1086-cube : @xmath1096}^n]}{m^n}.\ ] ] however , what we really want to estimate is the probability that there exists at least one @xmath1086-cube without knowledge of ( conditioning on ) the value of @xmath1082 .
this is easily overcome , however , and we obtain the probability @xmath1097}^n]}{m^n}p(n(s)=z ) \\ \\ & { \leqslant } & \frac{\mathbb{e}[y_{[k]}^n]}{m^n}\mathbb{e}[n(s ) ] \\ \\ & = & \frac{\mathbb{e}[y_{[k]}^n]}{m^n}n^{bs}.\end{aligned}\ ] ] therefore , for any layer @xmath412 , the probability that there exists at least one @xmath1098-cube can be bounded from above by @xmath1099}^n]}{n^{nb_1s(1/\theta-1)}}.\ ] ] here @xmath1100 is a non - specified constant . for large enough @xmath412 , we have from theorem [ br ] that @xmath1101}^n]\ ] ] is a polynomial in @xmath1102 with degree at most @xmath1080 .
therefore we can bound the whole term from above by @xmath1103 where @xmath754 is a positive number depending only on @xmath1080 .
we have @xmath1099}^n]}{n^{nb_1s(1/\theta-1 ) } } \ { \leqslant}\ ln^{s(b+bn(1/\theta-1)-nb_1(1/\theta-1))}\ ] ] which means that if @xmath1104 then @xmath1105}^n]}{n^{nb_1s(1/\theta-1 ) } } \
< \ \infty.\ ] ] the borel - cantelli lemma then implies that almost surely there are only finitely many layers @xmath412 such that there exists at least one @xmath1098-cube .
this implies that almost surely @xmath1106 , provided ( [ bc111 ] ) holds .
therefore , almost surely , we have @xmath1107 and since @xmath1080 can be chosen arbitrarily large , it follows that almost surely , @xmath1108 .
the reverse inequality holds almost surely , conditioned on non - extinction , due to proposition [ bb ] and the classical dimension result for box dimension .
therefore , for our fixed choice of @xmath7 , almost surely , conditioned on non - extinction , we have @xmath1109 finally , since @xmath5 was arbitrary we can obtain this result simultaneously for all @xmath1110 and since we know the assouad spectrum is continuous in @xmath7 ( corollary [ con ] ) , this is sufficient to move the quantifier on @xmath7 inside the almost sure result , which proves the theorem .
moran constructions are an important and well - studied class of fractals which can be thought of in a similar way to attractors of ifss but where the rigidity between construction levels has been relaxed in a number of different ways .
similar to ifs attractors , one defines a set as the intersection of a nested sequence of compact sets made up of increasingly small construction cylinders . however , the difference is that the relative position , size and number of ` child ' cylinders within their ` parent ' can vary .
this leads to a richer theory , less rigid construction rules and , as we shall see , more complicated dimension results .
a natural way to code the nested construction cylinders is via an infinite rooted tree .
let @xmath349 be such a tree with root vertex @xmath1111 on level 0 and such that every vertex on level @xmath1112 has at least one ` child ' vertex at level @xmath418 .
this guarantees that there are no ` degenerate ' paths , i.e. , paths starting from @xmath1111 and terminating at some finite level . in general , the ` level ' of a vertex is defined uniquely as the graph distance of the vertex to the root @xmath1111 .
let @xmath20 be the set of edges , @xmath1113 be the set of vertices , @xmath1114 be the level of vertex @xmath1115 , and @xmath1116 be the set of level @xmath369 vertices .
let @xmath1117 be a compact convex set which is equal to the closure of its interior , for example we could take @xmath1118 to be a ball or a cube .
for each @xmath1119 we assign @xmath1120 and a set @xmath1121 with diameter @xmath1122 such that 1 . for all @xmath1115 , @xmath1123 is similar to @xmath1118 , i.e. , the image of @xmath1118 under a similarity map .
if @xmath1124 is a child of @xmath1115 , then @xmath1125 3 .
if @xmath1124 is a child of @xmath1115 , then @xmath1126 4
. if @xmath1127 are distinct children of @xmath1115 , then the interiors of @xmath1128 and @xmath1129 are disjoint 5 .
there is a uniform constant @xmath663 such that @xmath1130 for all @xmath1131 . by construction
the sequence @xmath1132 is a decreasing sequence of non - empty compact sets and so we may define @xmath1133 which is itself a non - empty compact , typically fractal , set and is the main object of study in this section .
such sets are often called _ moran sets _ or _
moran constructions_. for convenience , we make a further homogeneity assumption that the contraction constants @xmath1134 are uniform across a given level .
more precisely , we assume that for each @xmath71 , there is a constant @xmath1135 such that for all @xmath1136 we have @xmath1137 .
in this situation we call the set @xmath1087 a _ homogeneous moran set_. before we can state our main result , we need some more notation . for @xmath1136 and @xmath1138 , let @xmath1139 be the number of level @xmath1140 children of @xmath1141 , i.e. , the number of vertices one can get to from @xmath1141 by going a further @xmath891 levels down the tree . given @xmath5 and @xmath71 let @xmath1142 be defined by @xmath1143 and observe that @xmath1144 .
[ moranthm ] let @xmath1087 be a homogeneous moran set . then @xmath1145 and @xmath1146 since the dimension spectra are local quantities , it is instructive to make the further simplifying assumption that the growth is also uniform across level .
more precisely , suppose that for each @xmath71 , there is a constant @xmath1147 such that for all @xmath1136 we have @xmath1148 . in this setting
we call @xmath1087 a _ uniformly homogeneous moran set_. this makes describing examples with particular properties more straightforward as one only has to specify the sequences @xmath1149 and @xmath1150 .
[ morancor ] let @xmath1087 be a uniformly homogeneous moran set .
then @xmath1151 and @xmath1152 moreover , if @xmath1153 for all @xmath1154 , then @xmath1155 and @xmath1156 let @xmath1157 and define @xmath71 by @xmath1158 observe that level @xmath369 cylinders @xmath1123 are images of @xmath1118 scaled down by similarities with contraction ratio @xmath1159 , where the implied constant depends only on the constant @xmath1160 appearing in condition ( 5 ) above .
this guarantees that any @xmath2-ball centered in @xmath1087 will intersect @xmath1013 many level @xmath369 cylinders .
this uses the fact that each level @xmath369 cylinder contains a ball with radius @xmath1161 where the implied constant depends only on @xmath1118 and by construction these balls may be taken to be disjoint .
the conclusion then follows from the doubling property of euclidean space .
furthermore , a particular level @xmath369 cylinder @xmath1123 may be broken down into precisely @xmath1162 many level @xmath1163 cylinders .
each of these cylinders is a scaled down copy of @xmath1118 by a factor of @xmath1164 and may therefore be covered by @xmath1013 many open balls of diameter @xmath96 .
thus we may cover any @xmath2-ball with @xmath1165 many @xmath96-balls .
this shows that @xmath1166 which is the desired upper bound . to prove the lower bound , choose a sequence of @xmath1167 and corresponding maximising @xmath1136 realising the @xmath969 above .
observe that the cover constructed above is optimal up to a constant since any ball of diameter @xmath96 used in the cover can cover at most @xmath1013 many level @xmath1163 cylinders .
this again uses the fact that each level @xmath1163 cylinder contains a ball with radius @xmath1168 where the implied constant depends only on @xmath1118 and by construction these balls may be taken to be disjoint . therefore choosing @xmath1169 we have
@xmath1170 this proves the lower bound and completes the proof for the assouad spectrum . the analogous result for the lower spectrum
is proved similarly and we omit the details .
let @xmath1171 $ ] , @xmath1172 and @xmath1173 for all @xmath1115 .
for example , we can think of the construction cylinders @xmath1123 as just the closed dyadic intervals .
let @xmath1174 it follows from corollary [ morancor ] that @xmath1175 and @xmath1176 this gives us an easy way of building interesting examples simply by specifying the sequence @xmath1177 .
note that the upper and lower box dimensions of @xmath1087 are given by the , more familiar , upper and lower _ asymptotic densities _ : @xmath1178 and @xmath1179 and the assouad and lower dimensions are given by the upper and lower _ banach densities _ : @xmath1180 and @xmath1181 the density functions we study are clearly related to the asymptotic and banach densities but seem not to be so well - studied .
generally , for @xmath1182 one might define the _ upper @xmath14-tail density _ of a set @xmath1183 as @xmath1184}{\lambda k - k}\ ] ] and the _ lower @xmath14-tail density _ as @xmath1185}{\lambda k - k}.\ ] ] in the same way that our spectra give finer information regarding the geometric scaling and homogeneity of the fractal set in question , the @xmath14-tail densities would give finer information ( via upper and lower spectra ) regarding the density of the set @xmath25 within @xmath1186 . due to the relationships between the @xmath14-tail densities and our dimension spectra , we can deduce several basic properties of the tail densities immediately from our results .
[ taildensity ] let @xmath13 and write @xmath1187 , @xmath1188 , @xmath1189 , @xmath1190 , for the lower and upper asymptotic densities and the lower and upper banach densities respectively . 1
. the upper and lower @xmath14-tail densities are continuous as functions of @xmath1182 2 .
for any @xmath1182 we have @xmath1191 3 . for
any @xmath1182 we have @xmath1192 \(1 ) and ( 2 ) follow immediately from propositions [ con ] , [ con2 ] and [ bb ] and setting @xmath1193 .
if either were false , there would exist a set @xmath25 exhibiting their falsehood and choosing the sequence @xmath1177 to be 2 on elements indexed by @xmath25 and 1 otherwise would yield a moran construction with spectra contradicting the basic results we have derived .
( 3 ) follows by observing that @xmath1194 and applying ( 2 ) .
similar to corollary [ bd0 ] , we see that if the upper asymptotic density is 0 then the upper @xmath14-tail density is identically 0 and , furthermore , if the lower asymptotic density is 1 then the lower @xmath14-tail density is identically 1 .
returning to the issue of examples , first we give an example illustrating the sharpness of propositions [ bb ] and [ bbl ] .
let @xmath1177 be a sequence of @xmath152s and @xmath350s where the @xmath350s are placed with @xmath14-tail density equal to 1/2 for all @xmath1182 , but that there are arbitrarily long sequences of 1s and 2s ( thus making the upper and lower banach densities 1 and 0 respectively ) .
one way to build such a sequence would be to begin with the sequence @xmath1195 and then for each @xmath271 insert a run of @xmath178 1s followed by @xmath178 2s starting at position @xmath1196 where @xmath1196 is some sequence which grows very fast , for example @xmath353 .
it follows that @xmath1197 and so this gives a simple demonstration that the lower ( respectively upper ) bounds from propositions [ bb ] and [ bbl ] are sharp . in a different direction , if we let @xmath1198 be the set produced by following the construction of @xmath1087 but with the placement of 1s and 2s reversed , then we get @xmath1199 and @xmath1200 with similar formulae relating the assouad and lower dimensions and the upper and lower box dimensions .
this is a very simple observation , but it allows one to build examples with lower spectrum equal to @xmath1201 which has not been seen yet and tends not to turn up as naturally as the analogous formula for the assouad spectrum .
also it means one can produce examples simultaneously for the assouad and lower spectrum whilst only worrying about the upper densities .
we end this section with a simple recipe for building moran constructions with interesting spectra , but we leave the details of the calculations to the interested reader .
let @xmath1202 $ ] and @xmath1182 and choose the sequence @xmath1177 as follows .
every entry is a 1 , apart from for a sequence of blocks @xmath1203 $ ] where @xmath1204 is a sequence of integers which grows very quickly . within each block
@xmath1205 $ ] , distribute 2s with density @xmath499 and as uniformly as possible , i.e. choose @xmath1206-f(k)+1 ) t ] $ ] many of the @xmath1177 to be 2 , rather than 1 .
simple calculations reveal that for the corresponding moran construction @xmath1087 we have @xmath1207 @xmath1208 @xmath1209 and for all @xmath5 @xmath1210 this simple construction allows us to build a variety of moran constructions with interesting spectra . for example
, we may vary the parameters @xmath1211 above to create finitely many moran constructions , @xmath1212 , with different spectra of the above form .
we can then build another moran construction , @xmath1087 , which is the disjoint union of scaled down copies of the @xmath1212 .
this can be done by choosing as many first level vertices as there are sets @xmath1212 and then mimicking the constructions of each @xmath1212 independently within each first level child . finally , using the fact that the assouad spectrum is stable under taking finite unions , see proposition [ basic2 ] , the assouad spectrum of @xmath1087 is given by the maximum of the assouad spectra of the @xmath1212 . finally , by ` inverting ' the construction of each @xmath1212 to @xmath1213 as above
, we get a corresponding example with lower spectrum mirroring the assouad spectrum of the initial example .
observe that this time the lower spectrum of @xmath1198 is given by _
minimum _ of the lower spectra of the @xmath1213 , see proposition [ basic2sep ] .
some examples of this type are given in figure [ moranfig ] , with the details below .
in particular , this recipe allows us to produce examples with the following properties , which we have not observed in any of our previous examples : 1 .
an arbitrarily large ( but finite ) number of phase transitions . 2 .
an arbitrarily large ( but finite ) number of disjoint intervals where the spectrum is constant ( with different constants ) .
3 . as @xmath7 approaches 1 ,
the spectra approach intermediate values , not equal to any of the familiar dimensions . are shown as dashed lines .
bottom row : the lower spectra of the corresponding moran construction @xmath1198 . the basic bounds from proposition [ bbl ]
are shown as dashed lines.[moranfig ] ] for completeness we provide the precise construction data for the above examples .
the first example is produced with four basic moran constructions with : @xmath1214 and @xmath1215 , @xmath1216 and @xmath1217 , @xmath1218 and @xmath1219 , and @xmath1220 and @xmath1221 , respectively .
the second example is produced with two basic moran constructions .
the first has @xmath1218 and @xmath1222 and the second one is produced by a sequence @xmath1177 which has all 1s apart from arbitrarily long runs of 2s inserted at positions which increase very rapidly .
this guarantees that the upper box dimension is 0 , but the assouad dimension is 1 .
the third example is produced with three basic moran constructions .
the first two have @xmath1216 and @xmath632 , and @xmath1223 and @xmath1219 , respectively , and the third one is again the construction with assouad dimension 1 and upper box dimension 0 .
proposition [ mono ] showed that the assouad spectrum was ` quasi - monotone ' in some quantitative sense . also ,
all of the ` natural ' examples we have considered thus far have had monotone spectra .
surprisingly , this is not always the case , although one has to work quite hard to break monotonicity .
we provide such an example here .
we start by defining a family of finite subsets of the line . for any @xmath1224
, we define a point set @xmath1225 as follows : @xmath1226 the idea of this construction is that , for some interval of length @xmath1227 , we put some small intervals of length @xmath1228 uniformly inside the larger interval .
then inside each small interval of length @xmath1228 we put some even smaller intervals of length @xmath1229 , and then we choose exactly one point from every interval of length @xmath1229 .
our set @xmath1230 is constructed by first choosing some interval of length @xmath1231 , and then placing @xmath178 sub - intervals of length @xmath1232 with equal distance , then further for each interval of length @xmath1232 we place @xmath178 sub - intervals of length @xmath1233 with equal distance .
then we simply pick every left most point of each interval of length @xmath1233 .
let @xmath1234 and finally let @xmath1235 we can define a metric @xmath1236 on @xmath92 by setting @xmath1237 if @xmath1238 belong to the same @xmath1239 for some @xmath178 and @xmath1240 if @xmath1238 belong to different @xmath1239 .
for now let us work with the ( non - compact ) metric space @xmath92 , but later on we will point out how this construction can be adapted to be a compact subset of euclidean space .
notice that the box - counting dimension of @xmath92 can not be defined because @xmath92 is not totally bounded .
observe that if @xmath1241 , then any @xmath2-ball inside @xmath92 intersects precisely one of the @xmath1239 , therefore in following discussion we only need to consider exactly one of the @xmath1239 at one time .
one consequence of this is that @xmath1242 since each @xmath1243 . for the assouad spectrum
, we need to count the number of @xmath96-balls required to cover a given @xmath2-ball , and the computation falls naturally into the following cases : a. @xmath1244 b. @xmath1245 c. @xmath1246 d. @xmath1247 the reason for this is simply that the order of the numbers @xmath1248 for @xmath1249 is crucial to the argument .
this will be clear in the subsequent steps .
+ * ( i ) the case : @xmath1244 . * for @xmath1227 ,
we have @xmath1250 let @xmath184 be small and consider an arbitrary ball @xmath859 centered at @xmath1251 which must intersect exactly one of the @xmath1239 . if our @xmath2-ball is inside a @xmath1239 with length @xmath1227 , then @xmath2 has to lie in one of the following intervals : 1 .
@xmath1252 $ ] . in this case
we need at most one ball with any radius to cover @xmath859 because any @xmath2-ball can contain at most one point .
2 . @xmath1253 $ ] . in this case
we need at most @xmath1254 many @xmath96-balls to cover @xmath859 , and so @xmath1255 3 .
@xmath1256 $ ] .
we still have @xmath1257 , and therefore we need at most @xmath1258 many @xmath96-balls to cover @xmath859 . as @xmath1259 , we see that @xmath1260 4 . @xmath1261 $ ] . in this case ,
our @xmath859 can cover more than one cluster of points that lie within @xmath1228 length , therefore we have at most @xmath1262 points being covered .
again we have @xmath1257 , and so we have @xmath1263 5 . @xmath1264 $ ] . in this case
@xmath2 is larger than @xmath754 , and we can cover all the points , but still we have @xmath1257 , so we need at most ( actually exactly ) @xmath1265 many @xmath96-balls , which gives @xmath1266 6 . @xmath1267 $ ] . in this case @xmath1268 , but @xmath1269 , therefore , we will need at most @xmath1270 many @xmath96-balls , which gives @xmath1271 7 .
@xmath1272 $ ] . in this case
@xmath1273 , therefore we need at most @xmath1274 many @xmath96-balls , which gives @xmath1275 8 .
@xmath1276 $ ] . in this case
we need at most @xmath1277 many @xmath96-balls , which gives @xmath1278 9 . @xmath1279 . in this case
@xmath1280 , therefore we need at most @xmath152 @xmath96-ball . by combining the estimates obtained in cases ( 1)-(9 ) above
we conclude that for all @xmath1281 we have @xmath1282 and then lemma [ n2d ] gives @xmath1283 moreover , we can see that all of the above estimates are optimal and can be achieved for arbitrarily small @xmath2 and therefore @xmath1284 for all @xmath1281 .
specifically , choose @xmath1285 for @xmath987 and consider case ( 5 ) above .
note that once we embed @xmath92 inside euclidean space ( which we do at the end of the section ) , the box dimension of the embedded set will be @xmath1286 and so the assouad spectrum attains the upper bound from proposition [ bb ] for @xmath1281 . at first sight
it might seem that we should have just focused on case ( 5 ) say , to get the lower bound and then applied the general upper bound , but at this point @xmath92 is not totally bounded and so proposition [ bb ] does not apply . + * ( ii ) the case : @xmath1287 .
* we have for @xmath1227 : @xmath1288 proceeding as before we consider the different possible relationships between @xmath2 and the length @xmath754 of @xmath1239 containing a particular @xmath2-ball : 1 . @xmath1252 $ ] . in this case
we need at most one ball with any radius to cover any @xmath2-ball because any @xmath2-ball can contain at most one point .
2 . @xmath1253 $ ] . in this case , each @xmath96-ball can contain at most @xmath152 point because : @xmath1289 and therefore @xmath1290 3 . @xmath1256 $ ] .
since we still have @xmath1257 , we need at most @xmath1291 @xmath96-balls to cover any such @xmath2-ball . as @xmath1259
, we see that @xmath1292 4 . @xmath1293 $ ] . in this case ,
our @xmath2-ball can cover more than one cluster of points that lie within @xmath1228 length , therefore we have at most @xmath1262 points being covered , and again we have @xmath1257 , for those points we need @xmath1262 many @xmath96-balls to cover the @xmath2-ball .
so we have @xmath1294 5 . @xmath1295 $ ] . in this case
we need at most @xmath1296 many @xmath96-balls and this gives @xmath1297 6 .
@xmath1298 $ ] . in this case
we need at most @xmath1270 many @xmath96-balls and this gives @xmath1299 7 .
@xmath1272 $ ] . in this case
@xmath1273 , therefore we need at most @xmath1300 many @xmath96-balls and this gives @xmath1301 8 .
@xmath1276 $ ] . in this case
we need at most @xmath1277 many @xmath96-balls and this gives @xmath1302 9 .
@xmath1303 . in this case
@xmath1280 , and therefore we need at most one @xmath96-ball . by combining the estimates obtained in cases ( 1)-(9 ) above we conclude that for @xmath1304 we have @xmath1305 and then lemma [ n2d ] gives @xmath1306 moreover , we can see that all of the above estimates are optimal and can be achieved for arbitrarily small @xmath2 and therefore @xmath1307 for all @xmath1304 .
+ * ( iii ) the case : @xmath1308 . *
this time we have we have : @xmath1309 proceeding as before we consider the different possible relationships between @xmath2 and the length @xmath754 of @xmath1239 containing a particular @xmath2-ball : 1 .
@xmath1252 $ ] . in this case
we need at most one ball with any radius to cover any @xmath2-ball because any @xmath2-ball can contain at most one point .
2 . @xmath1253 $ ] . in this case , each @xmath96-ball can contain at most @xmath152 point because : @xmath1310 therefore we have the bound @xmath1290 3 .
@xmath1311 $ ] .
since we still have @xmath1257 , we need at most @xmath1258 @xmath96-balls to cover any such @xmath2-ball . as @xmath1312
, we see that @xmath1313 4 . @xmath1314 $ ] . in this case ,
@xmath1315 , but since @xmath1316 , we have that an @xmath2-ball can cover at most one @xmath1228 interval , and for that @xmath1228 interval we need at most @xmath1317 many @xmath96-balls , so that we have : @xmath1318 5 . @xmath1319 $ ] . in this case
we need at most @xmath1296 many @xmath96-balls and this gives @xmath1320 6 .
@xmath1321 $ ] . in this case
@xmath1273 , therefore we need at most @xmath1322 many @xmath96-balls and this gives @xmath1323 7 .
@xmath1324 $ ] . in this case
we need at most @xmath1300 many @xmath96-balls and this gives @xmath1301 8 .
@xmath1276 $ ] . in this case
we need at most @xmath1277 many @xmath96-balls and this gives @xmath1302 9 .
we need in this case at most one @xmath96 ball . by combining the estimates obtained in cases ( 1)-(9 ) above
we conclude that for @xmath1304 we have @xmath1325 and then lemma [ n2d ] gives @xmath1326 moreover , we can see that all of the above estimates are optimal and can be achieved for arbitrarily small @xmath2 and therefore @xmath1327 for all @xmath1328 .
+ * ( iv ) the case : @xmath1329 .
* this case follows easily from the previous case .
since the spectra reaches the assouad dimension of @xmath92 ( which is 1 ) at @xmath1330 , it follows from corollary [ staya ] that it remains constantly equal to 1 for all @xmath1331 .
+ combining the results in the various ranges , we get an explicit piecewise formula for the spectrum of @xmath92 .
let @xmath92 be defined as above .
the the assouad spectrum of @xmath92 is non - monotonic and given by @xmath1332 \\
\frac{5/6-\theta}{1-\theta } & \theta\in ( 1/3,1/2 ] \\
\frac{1/3}{1-\theta } & \theta\in ( 1/2,2/3 ] \\ 1 & \theta\in ( 2/3,1 ) \end{cases}\ ] ] we conclude by describing how such an example can be constructed in euclidean space . to do this
we can first choose a sequence @xmath1333 of integers than increases very rapidly , for example @xmath1334 is sufficient , and then we choose @xmath1335 .
now arrange the sets @xmath1336 in @xmath1337 by first placing @xmath1338 along @xmath215\times \{0 \}$ ] with the left most point exactly @xmath1339 .
then we put @xmath1340 along @xmath215\times \{\frac{1}{2 } \}$ ] with the left most point exactly @xmath140 .
continuing , we put @xmath1336 along @xmath215\times \{1-\frac{1}{2^n } \}$ ] with the left most point exactly @xmath1341 . finally we get a bounded subset of the euclidean plane , which we can make compact by taking the closure if we wish .
this subset @xmath1342 will have the same assouad spectrum as the set @xmath92 defined before .
this is because the lengths @xmath1336 shrink much faster than @xmath1343 such that for any @xmath424-ball , if we would have more than one piece of the point sets involved , the length of the interval where the points lie will be so small such that @xmath1344 will be larger than them for any fixed @xmath653 and small enough @xmath424 .
then the situation is actually the same as for @xmath92 , since we only need to consider one of the @xmath1336 at any given point in the argument .
it follows from corollary [ bdconverge ] that the upper box dimension of @xmath81 is @xmath1286 . ,
defined in this section as a function of @xmath7 .
the basic bounds from proposition [ bb ] are shown as dashed lines.[nonmono],width=302 ] note that for this example @xmath1345 which is a demonstration of the sharpness of proposition [ mono ] .
the example in the previous section was relatively simple ( although it demonstrated an important property ) and explicit , i.e. we had an explicit formula for the spectra over the whole range . in this section
we give a more complicated family of examples , for which we unfortunately do not have an explicit formula for the spectra over the whole range , but which demonstrates some more important properties .
in particular , we prove that the lower spectra is not necessarily monotone ; monotonicity for both the assouad and lower spectra can be broken infinitely many times ; and , both the assouad and lower spectra can have infinitely many phase transitions , i.e. points where they fail to be differentiable .
the basic idea of construction is similar to section [ non - mono ] . given an interval @xmath1346 of length @xmath1227 and numbers @xmath1347 , we construct a set @xmath1348 via the following inductive procedure : * we pack closed intervals of length @xmath1349 with gaps of length @xmath1350 inside @xmath754
. the particular way of packing does not matter as long as it is by an optimal number .
we let the union of the closed intervals of length @xmath1349 be denoted by @xmath1351 . * for each interval of length @xmath1349 appearing at the first step , we optimally pack intervals of length @xmath1352 with gaps of length @xmath1353 .
we let the union of the closed intervals of length @xmath1352 be denoted by @xmath1354 . * for each interval of length @xmath1355 appearing at the @xmath1356th step , we optimally pack intervals of length @xmath1357with gaps of length @xmath1358 .
we let the union of the closed intervals of length @xmath1357 be denoted by @xmath1359 . we obtain a nested sequence of compact sets @xmath1360 and
finally we let @xmath1361 which is a non - empty compact set .
we can compute the dimensions of @xmath1362 , but we leave the details to the reader . in particular , @xmath1363 the exact computation of the spectrum is complicated , but nevertheless we can get some information of the spectrum without going through too much calculation .
[ l1 ] whenever @xmath1364 is an integer , we have @xmath1365 and @xmath1366 in particular , the assouad / lower spectrum is equal to the upper / lower box dimension infinitely often .
this is the general lower / upper bound from proposition [ bb]/[bbl ] .
suppose @xmath1367 is an integer .
for any @xmath1368 , there is a unique integer @xmath1369 such that @xmath1370 in particular , this means that @xmath1371 there are two different cases to consider : 1 .
@xmath1372 and @xmath1373 .
+ any @xmath859 for @xmath1374 contains one @xmath1357 interval , and any @xmath96-ball can cover at most one @xmath1375 interval because the distance between two disjoint intervals in the construction is @xmath1376 . by construction , in each @xmath1357 interval the number of intervals of length @xmath1377 is : @xmath1378[l^{\alpha^{k+1}(1-\beta)}][l^{\alpha^{k+2}(1-\beta)}] ... [l^{\alpha^{k+m}(1-\beta)}]\ ] ] where the @xmath1379 $ ] denotes the integer part .
therefore , the above argument shows that @xmath1380 we now a lower bound for @xmath1381 .
recall we only need the behaver of @xmath1381 when @xmath2 is sufficiently small .
choose @xmath2 so small such that @xmath369 will satisfy @xmath1382>cl^{\alpha^{k}(1-\beta)}\ ] ] where @xmath1383 .
this is possible because @xmath60{\geqslant}x(1 - 1/x)$ ] for any positive @xmath46 , and for @xmath46 large enough we have @xmath1384 .
then for all small enough @xmath122 we have @xmath1385 in summary , for all sufficiently small @xmath2 satisfying the conditions of case ( 1 ) we have : @xmath1386 + 2 .
@xmath1387 and @xmath1388 + in this case any @xmath859 contains @xmath1389\pm 1\ ] ] many intervals of length @xmath1357 . also to cover any interval of length @xmath1390
we need @xmath1391\pm 1\ ] ] many @xmath96-balls . using the same tricks as in case ( 1 ) , we may get rid of the integer part and the @xmath1392 . in summary , for all sufficiently small @xmath2 satisfying the conditions of case ( 2 ) we get : @xmath1393 finally , applying lemmas [ n2d ] and [ ln2d ] and the definition of @xmath369 yields the desired result .
[ l2 ] for every @xmath7 such that @xmath1367 is an integer , there exists @xmath1394 such that @xmath1395 and @xmath1396 are not constant in the interval @xmath1397 $ ] .
again , for any @xmath122 , there is an integer such that : @xmath1398 all the computation steps are the same as in the proof of lemma [ l1 ] , unfortunately because of the small fractional part of @xmath1399 , write @xmath1400 where @xmath959 is the integer part and @xmath1160 is the fractional part which is assumed to be small , we have now four cases to compute : inside each case , we omit the integer part symbols @xmath1379 $ ] because we can always use the argument in the proof of lemma [ l1 ] to get rid of the @xmath1379 $ ] symbols and obtain the asymptotic behaver of @xmath1381 . 1 . @xmath1401 and @xmath1402 any @xmath859 for @xmath1374 will contain one interval of length @xmath1357 and , for any @xmath1377 interval , we need approximately @xmath1403 many @xmath96-ball to cover it . therefore : @xmath1404 2 . @xmath1405 and @xmath1406 any @xmath859 for @xmath1374 will contain one @xmath1357 interval and , on the other hand , every @xmath1377 interval contained inside this @xmath1357 interval needs one @xmath96-ball to cover it .
therefore : @xmath1407 3 .
@xmath1408 and @xmath1406 in this case any @xmath859 for @xmath1374 contains approximately @xmath1409 many intervals of length @xmath1357 , but for each interval of length @xmath1377 we need one @xmath96-ball to cover it .
therefore : @xmath1410 4 .
@xmath1411 and @xmath1412 in this case any @xmath859 for @xmath1374 contains approximately @xmath1409 many intervals of length @xmath1357 .
also for any @xmath1390 interval we need approximately @xmath1413 many @xmath96-ball to cover it .
therefore @xmath1414 then we can apply lemmas [ n2d ] and [ ln2d ] to deduce that for @xmath1160 sufficiently small we have the following formula for spectra : @xmath1415 and @xmath1416 in particular , they are not constant .
the above formulae are obtained by considering the 4 cases above along with estimates derived from the definition of @xmath369 .
this gives an expression for the asymptotic behaviour of @xmath1381 which allows one to apply lemmas [ n2d ] and [ ln2d ] .
then one takes the @xmath969 of the largest asymptotic rate to obtain the assouad spectrum and the @xmath970 of the smallest asymptotic rate to obtain the lower spectrum .
note that these formulae only hold for @xmath1160 smaller than some constant @xmath1417 ( independent of @xmath959 ) .
this means they they are valid for @xmath1418 for all positive integers @xmath959 .
the following immediate corollary shows that the sets constructed in this section exhibit some new phenomena . for the sets
@xmath1362 constructed in this section , the assouad and lower spectra have the following properties : 1 .
they are not monotonic and , moreover , there are infinitely many disjoint intervals within which they fail to be monotonic .
2 . they have infinitely many points of non - differentiability .
this follows immediately by combining lemmas [ l1 ] and [ l2 ] with the fact that the spectra are continuous , see propositions [ con ] and [ con2 ] .
in this section we collect several open questions concerning the work we have presented in this paper .
corollary [ con ] and theorem [ con2 ] show that all of the spectra we consider are continuous in @xmath7 , but many of our examples exhibit phase transitions , preventing the spectra from being any more regular globally .
however , in all of our examples the spectra are piecewise analytic and we wonder if this is always the case , either with finitely or countably many phase transitions . in corollary
[ spectralip ] and theorem [ con2 ] we proved that the assouad and lower spectra are lipschitz when restricted to any closed subinterval of @xmath140 .
however , we have not ruled out the possibility of examples where the spectra exhibit less regularity on the whole domain .
in proposition [ basic2sep ] we proved that the modified lower dimension is stable under finite unions , provided the sets are ` properly separated ' .
note that this property does not hold for the lower dimension .
we were unable to determine if the ` properly separated ' condition can be dropped .
theorems [ meas1]-[meas3 ] proved that the assouad and lower spectra are of baire class 2 and thus borel measureable .
we have not been able to determine if the modified lower dimension or modified lower spectrum are borel measureable .
recall that in our study of spirals with sub - exponential and monotonic winding we needed to make the additional assumption that the upper box dimension was strictly larger than 1 . at first this might seem like a strange assumption , but it is the analogue of assuming that a decreasing sequence has positive box dimension .
indeed , if a set has box dimension 0 , then corollary [ bd0 ] tells us that the assouad spectrum is constantly equal to 0 , thus hiding any strange properties which may occur in that case .
there is no analogous result here and it remains an interesting problem to investigate what can happen when the box dimension of a spiral is 1 .
we suspect that theorem [ spiral ] no longer holds and that other phenomena are possible . also on the topic of spirals
, we proved that if a straight line segment is mapped to a spiral with sub - exponential and monotonic winding , then the hlder exponent of that map must satisfy certain restrictions based on the upper box dimension of the spiral
. it would be interesting to investigate the sharpness of this result .
a first step in this investigation could be the following question .
can a spiral with sub - exponential and monotonic winding and with upper box dimension strictly larger than 1 be mapped to line segment by a bi - hlder map ?
if so , what are the sharp bounds on the hlder exponents ?
one of our main examples involved self - affine carpets of the type introduced by bedford and mcmullen .
these results could be generalised in several interesting directions .
firstly , to the planar carpets of lalley - gatzouras @xcite and baraski @xcite . here
the matrices associated with the affine maps need not be the same for each map and this presents serious complications .
we expect different types of spectra to be possible , for example with more than one phase transition , see section [ lalley - gatzs ]
. another possibility would be to consider the higher dimensional analogue of the bedford - mcmullen carpets ; the so - called _ self - affine sponges_. these were first considered by kenyon and peres @xcite and recently the assouad and lower dimensions were computed by fraser and howroyd @xcite . here
the maps act on @xmath1236-dimensional space ( instead of the plane ) and the matrices associated to the affine maps are constantly equal to @xmath1420 for some fixed integers @xmath1421 .
again we believe that more complicated spectra are possible , with phase transitions possibly occurring at the @xmath1422 ( possibly distinct ) ratios @xmath1423 ( @xmath1424 ) . recall that in the ( non - uniform fibres ) planar case precisely one phase transition occurs at the ratio @xmath1425 .
self - similar and self - conformal sets with complicated overlaps provide an important class of sets with distinct box and assouad dimensions , recall ( * ? ? ? * section 3.1)and subsequent developments @xcite . in section [ self - similarsect ]
we gave non - trivial upper bounds on the assouad spectrum in terms of local dimensions of certain gibbs measures , but it would be interesting to take this work further
. it would clearly be valuable to find non - trivial _ lower _ bounds ( if they exist ) or to compute the spectrum explicitly for specific examples .
two specific examples come to mind : the example from ( * ? ? ?
* section 3.1 ) where the overlaps are complicated by the fact that two of the maps share a fixed point ; and the example considered by bandt and graf ( * ? ? ?
* section 2 ( 5 ) ) where the translations are chosen carefully to ensure that the weak separation condition fails despite all the contraction ratios being the same . given corollary [ sscor ] and the various folklore conjectures on self - similar sets and measures with overlaps
, we conjecture that in fact for any self - similar and self - conformal set the assouad spectrum coincides with the upper box dimension .
in section [ mandelbrotsect ] we considered an important random mode : mandelbrot percolation . here
the assouad spectrum behaved rather differently from many of the deterministic models we investigated : the spectrum was constantly equal to the box dimension despite the assouad dimension generally being much larger .
there are many other natural and important methods for producing random fractals and it would be interesting to see if randomness tends to lead to this phenomenon more generally .
one possible interpretation of this is as follows .
once you introduce randomness the assouad dimension is generically very large because almost surely you eventually see extremal behaviour .
however , to the assouad spectrum to be large , one needs to control ( in some precise sense ) how long you have to wait to see the extremal behaviour . in section [ moranexamples ]
we introduced the @xmath14-_tail densities _ as potential tools for studying density properties of subsets of @xmath1186 .
the definitions seem very natural to us , but we have been unable to find reference to them in the literature .
it would be interesting to find applications in , or connections to , other areas , or to study these finer density functions in their own right .
one might ask for their general properties as functions of @xmath14 or to consider examples of sets @xmath13 which produce interesting tail densities . once one has a reasonable notion of metric dimension
, one may wish to consider how this dimension behaves under canonical geometric operations , such as orthogonal projections or sections ( intersections with hyperplanes ) .
this theory is very well - developed for the hausdorff dimension , starting with the classical paper of marstrand @xcite , see also @xcite for the higher dimensional analogue and the survey papers @xcite for more details and up - to - date references .
roughly speaking , the philosophy behind marstrand s theorem and later developments is that if @xmath1426 has ` dimension ' @xmath1427 $ ] , then the ` dimension ' of the projection of @xmath26 onto hyperplanes of dimension @xmath1428 should be almost surely constant , with respect to the natural measure on the grassmanian manifold . in the case of the hausdorff dimension ,
the almost sure value is the largest possible , namely @xmath1429 .
however , for box dimension the almost sure constant is more subtle and given by a _ dimension profile _
, introduced by falconer and howroyd see@xcite .
recently fraser and orponen @xcite proved that the assouad dimension does not follow in the spirit of marstrand s theorem in that it can attain multiple values with positive probability ( under projection )
. it would be interesting to consider these results for the assouad spectrum , since it can be viewed as an interpolation between the assouad and upper box dimension .
the key theme of this paper has been what happens when one fixes the relationship between the two scales @xmath2 and @xmath4 used in the definition of the assouad dimension .
of course , there are many ways to fix this relationship .
indeed , let @xmath1430 \to [ 0,1]$ ] be a decreasing continuous function such that for all @xmath1431 for all @xmath1432 $ ] .
then one may define the @xmath658-_assouad dimension _ to be the analogue where the relationship between the two scales is fixed by always choosing @xmath1433 .
we have studied the continuously paramterised family of functions @xmath1434 and it turns out that this really is the ` correct ' family to consider in order to develop a rich theory .
indeed it follows from our results that if @xmath1435 then the @xmath658-assouad dimension coincides with the upper box dimension for any totally bounded set . moreover , if @xmath1436 then the @xmath658-assouad dimension coincides with the assouad dimension for any set where the assouad dimension is ` witnessed ' by the assouad spectrum ( i.e. the spectrum reaches the assouad dimension for some @xmath5 ) .
therefore one will ( usually ) only obtain a rich theory for functions @xmath658 which have an intermediate behaviour , which leads one directly to our functions @xmath1434 .
however , sets for which the assouad dimension is _ not _ ` witnessed ' by the assouad spectrum fall through the net in some sense . we propose the following programme to deal with such examples . for functions @xmath658 defined above , let @xmath1437 notice that this is not quite the definition we alluded to above because we only require @xmath1438 , and not @xmath1433
. however , this seems more natural for what follows .
one now asks the question : how difficult is it to witness the assouad dimension ?
more precisely , the problem is to classify for which functions @xmath658 we have @xmath1439 .
k. j. falconer , j. m. fraser and x. jin .
sixty years of fractal projections , _ fractal geometry and stochastics v , ( eds . c. bandt , k. j. falconer and m. zhle ) _ , birkhuser , progress in probability , 2015 .
p. mattila .
recent progress on dimensions of projections , in _ geometry and analysis of fractals _ , d .- j .
feng and k .- s .
lau ( eds . ) , pp 283301 , _ springer proceedings in mathematics & statistics .
_ * 88 * , springer - verlag , berlin heidelberg , 2014 .
m. rams and k. simon . the geometry of fractal percolation , in _ geometry and analysis of fractals _ d .- j . feng and k .- s .
lau ( eds . ) , pp 303324 , _ springer proceedings in mathematics & statistics . _
* 88 * , springer - verlag , berlin heidelberg , 2014 . | we introduce a new dimension spectrum motivated by the assouad dimension ; a familiar notion of dimension which , for a given metric space , returns the minimal exponent @xmath0 such that for any pair of scales @xmath1 , any ball of radius @xmath2 may be covered by a constant times @xmath3 balls of radius @xmath4 .
to each @xmath5 , we associate the appropriate analogue of the assouad dimension with the restriction that the two scales @xmath4 and @xmath2 used in the definition satisfy @xmath6 .
the resulting ` dimension spectrum ' ( as a function of @xmath7 ) thus gives finer geometric information regarding the scaling structure of the space and , in some precise sense , interpolates between the upper box dimension and the assouad dimension .
this latter point is particularly useful because the spectrum is generally better behaved than the assouad dimension .
we also consider the corresponding ` lower spectrum ' , motivated by the lower dimension , which acts as a dual to the assouad spectrum .
we conduct a detailed study of these dimension spectra ; including analytic , geometric , and measureability properties .
we also compute the spectra explicitly for a range of examples , exhibiting several different types of behaviour .
our examples include : self - affine carpets , self - similar and self - conformal sets with overlaps , mandelbrot percolation , moran constructions , decreasing sequences with decreasing gaps , and spirals with sub - exponential and monotonic winding .
we also give numerous applications of our results , including : dimension distortion estimates under bi - hlder maps for assouad dimension and the provision of new bi - lipschitz invariants .
[ multiblock footnote omitted ] |
statistical mechanics provides a complete explanation of the thermodynamic ( that is , macroscopic ) properties of systems in terms of their microscopic laws when the so - called thermodynamic limit is considered the limit of infinite system size @xcite .
however , there is a natural and increased interest in the properties of small systems @xcite , i.e. , systems whose size can not be considered infinite .
what is finite and what is infinite is a relative matter , as systems displaying a continuous or second - order phase transition illustrate .
the key issue is that the size of the system needs to be measured in terms of its correlation length . for these systems
a useful tool to deal with finite - size effects near the critical point of the transition is finite - size scaling @xcite .
let us consider a simple ferromagnetic system , whose thermodynamic variables are : the magnetic moment per particle @xmath0 ( proportional to magnetization ) , the absolute temperature @xmath1 , and the magnetic field @xmath2 .
it is convenient to rescale ( and center ) @xmath1 by means of the critical temperature @xmath3 , yielding the reduced temperature @xmath4 , and to rescale @xmath2 by @xmath5 ( with @xmath6 boltzmann constant ) , yielding the reduced magnetic field @xmath7 .
additionally , one may consider a system of units in which @xmath0 and @xmath8 are dimensionless .
the former , @xmath0 , will be the order parameter , whereas @xmath8 and @xmath9 are control parameters .
`` near '' the critical point of the transition , defined by @xmath10 , the equation of state fulfills a scaling law , which gives @xmath0 as a function of @xmath9 and @xmath8 as @xmath11 where @xmath12 and @xmath13 are critical exponents , and @xmath14 represents two scaling functions , one ( @xmath15 ) for @xmath16 and another one @xmath17 for @xmath18 .
the scaling law ( [ scaling_muth ] ) indicates the invariance of the equation of state under appropriate scale transformations ( which are linear transformations of the axes @xmath19 and @xmath8 ) . by the universality property
many different systems share the same values of the critical exponents and the same scaling functions , and then the scaling law ( [ scaling_muth ] ) constitutes a law of corresponding states @xcite .
for instance , for the mean - field theory or the landau theory of the ising model @xcite , @xmath20 , @xmath21 , and the scaling function @xmath14 is given by the two real solutions of @xmath22 , ref .
this yields @xmath23{3 } & \mbox { for } x=0\\ \pm 1/x & \mbox { for } x\rightarrow \infty\\ \end{array}\right.\ ] ] and substituting into the scaling law ( [ scaling_muth ] ) , one gets @xmath24{3h } & \mbox { for } \tau=0 \\ h/\tau & \mbox { for } \tau>0 \mbox { and } h\rightarrow 0\\ \end{array}\right . \label{threelaws}\ ] ] leading to the equation of the spontaneous magnetization , the critical - isotherm equation , and the curie - weiss law , respectively @xcite .
as @xmath25 is a smooth function , it is only at the critical point that a sharp transition emerges .
it is important that the correlation length @xmath26 fulfills a scaling law analogous to eq .
( [ scaling_muth ] ) , @xmath27 with @xmath28 another critical exponent and @xmath29 another pair of scaling functions .
then the main fact of critical phenomena is that @xmath26 diverges ( goes to @xmath30 ) right at the critical point ( as @xmath28 and @xmath13 are positive ) .
for instance , at the critical isotherm , @xmath31 , one has @xmath32 , whereas at zero field , @xmath33 . strictly , all these equations are only valid in the thermodynamic limit . for a system of finite size @xmath34 ( in all dimensions @xcite )
the correlation length can not be infinite .
when @xmath34 is much larger than the correlation length one does not expect that the finiteness of the system has any influence on the behavior of the system ; however , this is not the case when @xmath34 becomes smaller than the correlation length of the corresponding infinite system @xcite .
so , one can introduce a phenomenological additional dependence on @xmath35 in the equation of state @xcite , as @xmath36 which , substituting the equation for @xmath26 [ eq . ( [ xi ] ) ] , can be written as @xmath37 or , equivalently , @xmath38 where the terms @xmath39 and @xmath40 have been transformed to @xmath41 and @xmath42 , respectively .
the previous equation constitutes a finite - size scaling law or ansatz , where now @xmath14 , @xmath43 and @xmath44 become bivariate scaling functions , with the latter unifying the positive and negative values of @xmath8
. the finite - size - scaling ansatz can be verified by plotting @xmath45 versus @xmath46 and @xmath47 ; if a data collapse emerges , this gives the shape of the scaling function @xmath44 . in this way
, finite - size behavior is determined from the critical exponents of the infinite system @xcite .
note that for a finite system with @xmath48 the system size @xmath34 plays a role similar to that of the inverse of the magnetic field in an infinite system , or more precisely , @xmath49 acts as @xmath50 , and in this way , one expects that the first argument of the scaling function @xmath44 in eq .
( [ fssone ] ) behaves , qualitatively , as the scaling function @xmath14 in eq .
( [ scaling_muth ] ) .
this implies that a sharp transition can only take place for @xmath51 , i.e. , in the thermodynamic limit .
the galton - watson process @xcite provides the simplest model for the growth ( and degrowth ) of a biological population @xcite , but it is equally applicable to the growth of a nuclear reaction @xcite , an earthquake @xcite , or mean - field self - organized critical processes in general @xcite .
it belongs to a more general class of models known as branching processes .
the galton - watson process starts with one single element that replicates , producing more elements , called offsprings , which also replicate , producing more elements and so on .
the model is stochastic , as the ( total ) number of offsprings produced by each element is random , characterized by a distribution that is the same for all elements and also independent of the number of offsprings of the other elements . in mathematical terms , the probability that the number of offsprings @xmath52 of one element takes the value @xmath53 is given by @xmath54 $ ] , with @xmath53 taking discrete values from 0 to @xmath30 . in this paper
we will consider that @xmath54 $ ] is given by the geometric distribution , or by the generalized geometric distribution , but the model is totally general .
the distribution @xmath54 $ ] completely defines the model , as , we insist , the number of offsprings of each element are identically distributed and independent .
the initial element defines the @xmath55th generation , its offsprings are the first generation , and so on .
an index @xmath56 labels each generation .
the model does not incorporate time , but one can interpret @xmath56 as a discrete time .
an important auxiliary variable is @xmath57 , which counts the number of elements in each generation , starting with @xmath58 ( one single original element ) . the key question to ask is if the process gets extinct , i.e. , @xmath59 at some @xmath60 , or not ( where it goes on forever ) .
a fundamental result in the theory of branching processes @xcite is that the probability of extinction @xmath61 can be obtained from @xmath62 = \lim_{t\rightarrow \infty } f^t(0),\ ] ] where @xmath63 is the @xmath64th composition of the probability generating function @xmath65 of the random variable @xmath52 , i.e. , @xmath66 ( composed @xmath56 times ) , with @xmath67 s^k.\ ] ] as we iterate successive compositions of @xmath65 starting from @xmath68 , the limit is given by the smallest fixed point @xmath69 of @xmath65 in the interval @xmath70 $ ] ; so , @xmath69 necessarily satisfies @xmath71 , but it is the smallest value in @xmath70 $ ] verifying such relation . introducing the probability of survival , or probability of non extinction @xmath72 ,
fulfilling @xmath73 , the fixed - point condition becomes @xmath74 ( 1-\rho)^k.\ ] ] from here , it is clear by normalization that @xmath75 is a possible solution .
expanding the equation up to second order in @xmath72 using the binomial theorem one gets @xmath76
\left(1-k\rho + \frac{k(k-1 ) } 2\rho^2\right)\ ] ] @xmath77 the solutions , in terms of the mean number of offsprings , @xmath78 , and close to @xmath79 , are then @xmath80 where we have used that when @xmath72 is close to zero ( from above ) @xmath81 is close to one , and therefore @xmath82 , where @xmath83 is the variance of @xmath52 when its mean is one .
it can be proved that there are no other fixed points than the two above @xcite .
it is clear that the case in which the offspring distribution verifies @xmath79 is critical , in the sense that it separates two very different `` phases '' of the system : extinction for sure if @xmath84 and non - sure extinction ( and the possibility of a `` demographic '' explosion ) for @xmath85
. even more , this phase diagram is analogous to the spontaneous ( zero - field ) behavior of a magnetic system , eq .
( [ threelaws ] ) , if we identify @xmath86 with the control parameter @xmath9 and @xmath72 with the order parameter @xmath0 , and so we can talk about a phase transition in the galton - watson model @xcite with critical point at @xmath87 .
note then that @xmath83 becomes the variance of the number of offsprings in the critical case .
there are , though , two quantitative differences : @xmath88 ( in contrast to @xmath20 in the magnetic example above ) and that the ordered phase ( non - zero order parameter ) is above the critical point now . equation ( [ nosecual ] ) also tell us that when the distance to the critical point , @xmath86 , is rescaled by @xmath83 the behavior of the transition is universal , i.e. , independent on the underlying distribution of the number of offsprings @xmath52 . in this paper
we investigate this phase transition for a finite number of generations , i.e. , when the number of generations is limited by @xmath89 . in a previous paper @xcite
we expanded @xmath90 around the critical point @xmath69 to obtain a general finite - size - scaling law for the probability of survival @xmath72 . here
we follow a different , more direct approach , particularized for a geometric distribution in the number of offsprings , which will allow us to obtain also the corrections to scaling . after the introduction to finite - size scaling in critical phenomena in the previous section and the introduction to branching processes in this section , in sec .
3 we analyze the finite - size effects in the critical properties of the galton - watson process when the offspring distribution is given by the geometric distribution .
two different order parameters are explored , [ @xmath72 and @xmath91 , and the corrections to scaling and the range of validitity of the scaling law are obtained as well .
we generalize the finite - size scaling law for the so - called generalized geometric distribution in sec .
5 . previously , in sec . 4
, we establish that our scaling law also describes the escape probability of a simple one - dimensional random walk .
an appendix gives some details of the calculations of secs .
we consider the galton - watson model with a finite number of generations @xmath34 , which means that the process is stopped when it reaches the @xmath92th generation , i.e. , the elements of this generation are not allowed to replicate . viewing the process as a branched tree
, @xmath34 becomes the height of the tree and is therefore a measure of system size ( more precisely , the height of the tree is @xmath93 , counting the @xmath55th generation ) .
the extinction of this process is given by the event @xmath94 , as extinction at any generation @xmath95 is included in the case @xmath94 ( extinction is forever , as it is an absorbing state ) .
in the same way as for an unbounded system , the probability of extinction will be @xmath96 = f^l(0)\ ] ] ( we only make explicit the dependence on @xmath34 , but a hidden dependence exists in the parameters of the distribution of @xmath52 , in particular on @xmath81 ) .
the probability of extinction is obtained then as the @xmath92th composition of the probability generating function of the distribution of the number of offsprings , but note that as @xmath34 is not infinite , @xmath97 will not reach the fixed point @xmath69 .
although formally the problem is solved by the calculation of @xmath97 , in general it is not feasible to arrive to an explicit expression for the composition , even for small values of @xmath34 .
a remarkable exception is the case when @xmath52 follows the geometric distribution , given by @xmath98 = p q^k , \label{geodist}\ ] ] for @xmath99 ( and zero otherwise ) and with @xmath100 .
the only parameter of the distribution is @xmath101 , which is called the success probability .
the geometric distribution has a straightforward interpretation in terms of biological populations . for instance , consider that the elements that replicate are female individuals , and each female has a probability @xmath102 to produce another female and a probability @xmath101 of producing a male .
each female reproduces until it gets a male , and when the male is obtained the mother does not reproduce anymore .
although getting a male is considered a `` success '' ( this is just a name ) , it is the female individuals what are counted as offsprings , so @xmath52 counts the number of females disregading the male .
note that another variant of the geometric distribution counts also the male , this would be for us a shifted geometric distribution and is not considered here .
the probability generating function of the geometric distribution turns out to be @xmath103 from which the mean is obtained as @xmath104 and the variance as @xmath105 , see ref .
note that the critical point of the corresponding galton - watson process is at @xmath106 and so @xmath107 , with a critical variance @xmath108 .
the fundamental property ( for our problem ) of the geometric distribution comes from the fact that its probability generating function is a fractional linear function @xcite , also called a linear fractional function @xcite . in this case
the successive compositions of @xmath65 can be computed for any @xmath34 , yielding @xmath109 see ref .
@xcite or eq .
( [ ladelapendice ] ) at our appendix .
the constant @xmath110 is a fixed point of @xmath65 different from 1 ( this fixed point , @xmath110 , always exists except for @xmath79 ) , and the constant @xmath111 is given in the appendix .
then , the probability of survival will be @xmath112 which contains the solution to our problem . for the geometric distribution
the fixed point @xmath110 is at @xmath113 , and then @xmath114 ( see appendix ) ; therefore , substituting into eq .
( [ labuena ] ) we get @xmath115 this exact equation provides the order parameter @xmath72 as a function of the control parameter @xmath81 for any system size @xmath34 ( in the case of the geometric distribution ) . in order to verify
if a scaling law is fulfilled it is convenient to introduce the rescaled distance to the critical point , @xmath116 where the `` distance '' @xmath86 is rescaled ( divided ) by the term @xmath117 , with the value of the exponent @xmath28 unknown . substituting @xmath118 and @xmath119 into eq .
( [ scaling_geo ] ) we observe that the rescaled survival probability @xmath120 in the limit @xmath51 either tends to zero or infinite ( depending on the sign of @xmath121 and on whether @xmath122 or @xmath123 ) , except in the case @xmath124 . for @xmath124 and close to the critical point , the limit of @xmath120 is a positive value that only depends on @xmath121 , which is the signature of a scaling law , @xmath125 with @xmath126 the scaling function . indeed , rewritting eq .
( [ scaling_geo ] ) in terms of @xmath121 , using that @xmath127 for @xmath124 leads to @xmath128 up to the lowest order in @xmath129 . taking into account that the variance at the critical point is @xmath108 , the scaling law can be written as @xmath130 with scaling function @xmath131 in total agreement with ref .
the reason to introduce the value of @xmath83 will become more clear when we consider the generalized geometric case , in sec .
it is important that the scaling function ( [ scalingfunction ] ) fulfills @xmath132 although our calculation does not include the critical case , @xmath133 , the appendix shows that indeed the critical case is also described by the value of the scaling function @xmath44 at @xmath133 .
therefore , there is a removable singularity at @xmath133 .
the limit behavior of @xmath44 , substituted into the scaling law , leads to @xmath134 we see that the infinite - size case , eq .
( [ nosecual ] ) , is recovered when @xmath34 is infinite , and that it is only in this case that a sharp transition exists .
comparison with eq .
( [ threelaws ] ) allows one to see which is the equivalent of the `` critical isotherm '' and `` spontaneous magnetization '' laws for the galton - watson process . for the latter case
we see that @xmath88 .
the curie - weiss law is not fulfilled as @xmath72 does not decay as a power law in @xmath34 but exponentially for @xmath135 .
we may also obtain the corrections to scaling , taking care of terms beyond the leading one .
going back to eq .
( [ scaling_geo ] ) , we substitute there the exact expression @xmath136 , with @xmath137 , @xmath138 , etc .
, then , @xmath139 @xmath140\ ] ] @xmath141,\ ] ] with @xmath142 and @xmath143 .
the first terms of the different sums are @xmath144 @xmath145 @xmath146 @xmath147 @xmath148 @xmath149 let us study the behavior as far from the critical point as possible .
below it ( @xmath150 ) , we take @xmath151 and then @xmath152 ( exponentially in @xmath121 ) ; therefore , only @xmath153 contributes and we get @xmath154 so , the first correction - to - scaling term goes as @xmath155 .
this means that if this term is of order @xmath156 ( i.e. , @xmath157 ) all other terms are of higher order in @xmath156 , in the limit @xmath51 .
this is so because the rest of terms are of the form @xmath158 and @xmath159 above the critical point ( @xmath160 ) we consider @xmath161 , then , @xmath162 and the sums lead to the cancellation of all terms that are not powers of @xmath163 , so @xmath164 the first correction to scaling is given by the term @xmath165 .
if we impose this to be of order @xmath156 , ( i.e. @xmath166 ) , we will obtain the limit of validity of the scaling law above the critical point . in summary
, the scaling law will hold in the range @xmath167 with @xmath168 .
for instance , for a 5 % error [ defined as the ratio between the approximation given by the scaling law and the exact @xmath169 , eq .
( [ scaling_geo ] ) ] , @xmath170 and then @xmath171 .
figure [ figerror ] shows that this is valid for @xmath92values above 40 for @xmath172 and above 160 for @xmath85 . note that the range of validity that we obtain , eq . ( [ wepropose ] ) , is much larger than the one implicit in ref .
@xcite , @xmath173 , with @xmath174 a constant .
if we do not take the limits @xmath175 , we have , keeping terms up to first order in @xmath176 , @xmath177 which is also shown in figs . [ figerror ] a and b. a scaling law with a broader range of validity
is obtained taking as an order parameter not @xmath72 but @xmath178 .
this is just the ratio between the number of realizations that survive at @xmath179 and the number that are extinct at @xmath179 .
( [ scaling_geo ] ) we obtain @xmath180 and proceeding as in the preceding case , we get @xmath181 @xmath182.\ ] ] the factors @xmath183 go to zero exponentially fast when @xmath175 , except the first one ( @xmath184 ) when @xmath185 , for which @xmath162 .
this is the only contribution away from the critical point , and so ( below the critical point ) the correction to scaling goes as @xmath186 .
the range of validity of the scaling law is then given by @xmath187 i.e. , the scaling law is valid arbitrarily far from the fixed point in the supercritical region , as the correction term there decays exponentially fast in @xmath121 .
if we keep @xmath121 finite and terms up to first order in @xmath176 we arrive at @xmath188 this can be verified in fig .
[ dos ] , where the scaling law describes system sizes as small as @xmath189 arbitrarily far from the critical point in the supercritical region .
thanks to a well - known mapping between branching processes and random walks @xcite , our finite - size scaling law is also applicable to the latter system . in concrete , a one - dimensional random walk can be obtained from the geometric galton - watson branching process by following the branches sequentially . instead of considering that each generation @xmath56 of the process is generated in parallel from the previous one ( as the identification of the index @xmath56 with time suggests )
one changes the order in which offsprings appear .
the position of a walker in the tree associated to the branching process determines which element ( which node of the tree ) replicates .
the walker is initially located at the root ( the element at the @xmath55th generation ) , and moves to one of the elements in the first generation ( it does not matter which one ) .
if this element has its own offsprings , the walker moves to one of this , and so on .
a branch is followed sequentially until the branch gets extinct ( the last element has no offsprings ) , and then the walker moves back to the parent of the last element ( from generation @xmath56 to @xmath190 ) ; if this parent has more offsprings the walker follows the branch of one of the remaining offsprings ; if not , the walker moves back to the previous parent ( at generation @xmath191 ) and so on .
note then that the walker passes twice through each link or edge between parent and offspring .
if , arbitrarly , we consider that the root is at the bottom of the tree ( as in real , biological trees ! ) and each new generation is one level above the previous one , the walker travels up and down through all the tree .
the one - dimensional random walk is obtained from the projection of the position of the walker on the axis counting the number of generations , so , the @xmath64axis of the branching process becomes the spatial axis of the random walk .
then , the walker moves up with probability @xmath102 and down with probability @xmath101 ( the parameters of the geometrical distribution ) .
notice that the mapping is possible and exact because the number of offprings follows the geometric distribution , eq .
( [ geodist ] ) . the finite - size condition imposed to
the branching process translates into the existence of a reflecting boundary at @xmath179 for the random walk , and then , the probability of survival @xmath72 of the branching process turns out to be the probability of hitting the reflecting boundary , @xmath192 , for the random walk .
this also has an absorbing boundary at @xmath193 , where the walk dies ( after a duration equal to twice the number of elements , minus one ) .
after all these considerations , the mapping is established , and we can write a finite - size scaling relation for the hitting probability , @xmath194 with @xmath195 given by eq .
( [ scalingfunction ] ) and @xmath196 remember that this is valid for large @xmath34 and close to the critical point @xmath197
. in particular , the corrections to scaling of the previous section also hold when the relationships are written in terms of @xmath81 or @xmath198 .
in fact , the previous scaling law describes the probability that a random walk starting next to the absorbing boundary hits the other boundary , independently of the nature of the latter ( reflecting or not ) , as it is only the first - passage time what matters . in this way ,
the one - dimensional random walk , the simplest system in statistical physics , displays a continuous phase transition with finite size scaling , for which the corrections to scaling can be easily obtained as well .
the previous analysis of the geometric galton - watson process in terms of fractional linear functions ( see appendix ) suggests a generalization of the problem . we may consider the generalized geometric distribution , in which the zero - offspring probability , @xmath199 $ ] , is released from following the geometric distribution and instead it takes a free value @xmath200 , which is a new parameter .
the rest of values of @xmath52 follow the geometric distribution , but rescaled by @xmath201 ( because of normalization ) . in a formula ,
@xmath98 = \left\ { \begin{array}{ll } p_0 & \mbox { for } k=0\\ ( 1-p_0 ) p q^{k-1 } & \mbox { for } k=1,2 , \dots\\ \end{array } \right.\ ] ] and zero otherwise .
we recover the usual geometric distribution for @xmath202 .
the generating function is indeed a fractional linear function , @xmath203 which yieds @xmath204 and @xmath205 .
the critical point turns out to be at @xmath206 the analysis of sec .
3 is fully applicable in this case , in particular eq .
( [ labuena ] ) .
we need to know that @xmath207 and @xmath208 ( see appendix ) ; in fact , we write @xmath110 as a function of @xmath81 and @xmath200 , which is @xmath209 , with @xmath210 .
notice that we study the transition keeping fixed @xmath200 .
substituting into the formula for the order parameter @xmath169 , eq .
( [ labuena ] ) , we arrive at @xmath211 @xmath212 introducing again the rescaled variable @xmath213 , and taking the limit @xmath51 , the only non trivial limit arises for @xmath124 . in this case , up to first order in @xmath176 and introducing the critical variance @xmath214 , we get @xmath215 which is the same scaling law as for the geometric case , with the scaling function @xmath195 given again by eq .
( [ scalingfunction ] ) .
we have presented here direct analogies between branching processes and thermodynamic phase transitions .
we have considered the classical galton - watson model of branching processes when the number of offsprings @xmath52 per element is given by the geometric distribution .
this process has as natural control and order parameters the mean value of @xmath52 and the probability of survival @xmath72 , respectively .
we study finite - size effects by imposing an upper limit @xmath34 to the number of generations . after obtaining the exact expression for the equation of state , that is , the dependence of the order parameter with the control parameter , eq .
( [ scaling_geo ] ) , we introduce the rescaled distance to the critical point , @xmath216 .
when @xmath124 we demonstrate that a finite - size scaling law , eq .
( [ scalingdeverdad ] ) , emerges in the limit @xmath51 . in general ,
the theory of critical phenomena does `` not explain why in some systems scaling holds for only 1 - 2 % away from the critical point and in other systems it holds for 30 - 40 % away '' @xcite .
in particular , finite - size scaling should work when the system size tends to infinite and the control parameter approaches the critical point ; nevertheless , in practice , finite - size scaling predictions turn out to apply to rather small systems at a non - negligible distance from the critical point @xcite .
we provide a quantitative derivation of these limits for the finite - size scaling behavior of the galton - watson process , eq .
( [ wepropose ] ) , thanks to the calculation of the corrections to scaling , eqs .
( [ suspiro ] ) and ( [ massuspiro ] ) , or eq . ( [ scalingpluscorrection ] ) . if we define an alternative order parameter as @xmath178 , the same scaling law holds , but with a larger range of validity , given by eq .
( [ largerrange ] ) . in this case
the corrections to scaling are given by eq .
( [ suspiro ] ) , below the critical point or by eq .
( [ nuevasl ] ) , in general .
a straightforward mapping between branching processes and random walks allows one to establish that all our results for the survival probability of a geometric galton - watson process are equally valid for the probability that a one - dimensional random walk , starting above but close to an absorbing origin and evolving through @xmath217 increments , reaches a distance to the origin equal to @xmath34 . in this way
, a subcritical galton - watson process corresponds to a random walk with a bias to the negative ( @xmath218 ) increment , for which the hitting probability becomes zero as @xmath51 . on the other hand , the supercritical case corresponds to a random walk with a positive bias in the increment , for which there exists a non - zero probability that never returns to the origin in the limit @xmath51 .
obviously then , the critical case is the one of a fair random walk .
to the best of our knowledge , the one - dimensional random walk provides the simplest example of a system exhibiting a finite - size scaling law .
therefore , the analogies between branching processes and equilibrium phase transitions are totally applicable to the one - dimensional random walk .
g. pruessner encouraged us to study the limits of applicability of the finite - size scaling law .
expansions have been revised with the help of the online taylor series calculator of wolframalpha .
r. garcia - millan has enjoyed a stay at the centre de recerca matemtica through its internship program , as well as a scholarship from the london mathematical laboratory .
the rest of authors acknowledge support from projects fis2012 - 31324 , from spanish mineco , and 2014sgr-1307 , from agaur .
a fractional linear function is defined by @xmath219 with @xmath220 and @xmath221 constants fulfilling @xmath222 ( to avoid that the numerator and the denominator are proportional ) .
note that for the geometric distribution , eq .
( [ fgeo ] ) , @xmath223 , and @xmath224 , although in the next paragraphs we will keep generality .
the advantge of fractional linear functions is that their compositions are very manageable . to see this
we follow the calculation of karlin and taylor @xcite .
let us consider any point @xmath225 , then , it is direct to see that @xmath226 and for two points @xmath110 and @xmath227 one has @xmath228 for fractional linear functions representing probability generating functions there exist just two fixed points that , by definition , verify @xmath229 , so one can identify the previous @xmath110 and @xmath227 with these fixed points .
it can be also verified that it is only at the critical point ( @xmath79 ) that the two fixed points take the same value , @xmath230 .
note that the fixed point @xmath69 corresponding to the probability of extinction in the infinite system ( mentioned in sec .
2 ) is defined as @xmath231 .
so , using the defining property of fixed points ( @xmath229 ) and defining @xmath232 and @xmath233 one gets @xmath234 in order to calculate @xmath235 one can iterate the same argument for the left - hand side of the equation , and in general , by induction , @xmath236 with @xmath237 . isolating @xmath238
one arrives at the desired formula for the compositions of @xmath65 , @xmath239 which holds for any values of the parameters of the offspring distribution , except at the critical point ( @xmath79 ) . in the case
in which @xmath65 is a probability generating function , one of the fixed points is equal to one , by normalization .
so , one can take , without loss of generality @xmath240 . substituting the form of a fractional linear function , eq .
( [ flf ] ) , into @xmath241 one gets a relation between the parameters @xmath220 and @xmath221 .
one can also verify that the other fixed point is @xmath242 .
finally , the constant @xmath111 turns out to be , substituting @xmath110 , @xmath243 which happens to be identical to the inverse of the mean , i.e. , @xmath244 for the generalized geometric distribution , from its probability generating function , eq .
( [ ggeopgf ] ) , and from the definition of fractional linear functions , eq .
( [ flf ] ) , one establishes that @xmath245 , @xmath246 , @xmath247 , and @xmath224 , and the fixed point @xmath110 turns out to be @xmath248 which , for the particular case of the geometric distribution , defined by @xmath202 , turns into @xmath249 the knowledge of the value of the fixed point @xmath110 leads to the explicit form for @xmath63 . at the critical point , given by @xmath79 ,
it is necessary to follow a separate approach . for the generalized geometric distribution , the critical point
is given by @xmath250 , which , substituting into the probability generating function , eq . ( [ ggeopgf ] ) , leads to @xmath251 induction leads directly to @xmath252s\}}{1+(t-2)p_0-(t-1)p_0 ^ 2 -(t
p_0-t p_0 ^ 2 ) s } \ ] ] @xmath253s}{1+(t-1)p_0-t p_0s},\ ] ] from where the order parameter of the transition turns out to be @xmath254 taking the limit of large @xmath34 and using the expression above for @xmath83 .
this is in perfect agreement with the results obtained for @xmath255 .
note that the results for the geometric distribution are a particular case corresponding to @xmath256 at @xmath79 .
a. corral and f. font - clos .
criticality and self - organization in branching processes : application to natural hazards . in m.
aschwanden , editor , _ self - organized criticality systems _ , pages 183228 .
open academic press , berlin , 2013 . | the theory of finite - size scaling explains how the singular behavior of thermodynamic quantities in the critical point of a phase transition emerges when the size of the system becomes infinite .
usually , this theory is presented in a phenomenological way . here
, we exactly demonstrate the existence of a finite - size scaling law for the galton - watson branching processes when the number of offsprings of each individual follows either a geometric distribution or a generalized geometric distribution .
we also derive the corrections to scaling and the limits of validity of the finite - size scaling law away the critical point .
a mapping between branching processes and random walks allows us to establish that these results also hold for the latter case , for which the order parameter turns out to be the probability of hitting a distant boundary . |
observational data collected over the last two decades from supernovae type i a ( sn ia ) confirmed that the universe is undergoing an accelerated phase of expansion .
the first evidences for such a behavior have been discovered by two independent collaborations , and reported in ref .
@xcite , by the high - redshift supernova search team , and in ref .
@xcite , by supernova cosmology project team .
analyses from combined sn ia dataset ( see _ e.g. _ the one reported in ref .
@xcite ) , cosmic microwave background radiation ( cmb ) , for which we refer for instance to refs .
@xcite , and from large scale structure @xcite , have further provided consolidated evidence for the current acceleration of the universe .
the source of late - time cosmic acceleration has been dubbed as `` dark energy '' ( de ) , an exhaustive theoretical characterization of which is still lacking .
there have been many attempts so far to individuate the origin of de , but a consensus in the literature has not been reached yet .
a review of the models so far advocated within the literature of de is beyond the purpose of this study , and we prefer to address reader to a sizable and rich literature that exists on this subject ( see _ e.g. _ refs .
@xcite ) . in
what follows , we will focus on a rather simple idea , which address the problem of de from the perspective of condensation of yang - mills fields .
cosmic acceleration as a source of cosmological inflation has been first proposed by yang zhang in ref .
@xcite , and then further developed by the same author and collaborators in the framework of current cosmic acceleration , as a source of de , in refs .
@xcite and @xcite , respectively in the perturbative two - loops and three - loops analysis of the effective action of yang - mills theory . to be more specific , in ref .
@xcite the author has considered a yang - mills gauge boson condensate as described by the renormalization - group - improvement ( rgi ) action within a homogenous and isotropic flrw background . following refs .
@xcite , the action for su(n ) yang - mills fields has been reshuffled in terms of an effective running coupling constant @xmath5 , namely @xmath6 in which @xmath7 stands for the background metric and @xmath8 is the square of the renormalization mass - scale . from now on , for simplicity of notation we define the contraction of the field - strength tensors @xmath9 which plays the role of an order - parameter for the ymc , and allows us to write the effective lagrangian in a more compact way @xmath10 . a further simplification within the studies so far developed in the literature concerns the use of a su(n ) gauge symmetry group , in which the number of colors @xmath11 is not fixed _ a priori_. this choice connects the physics under investigation to the constituents gauge - groups of the standard model of particles . moving from the two - loops perturbative ( ultraviolet ) expansion at large @xmath12 of the effective coupling constant , which from the analyses reported in refs .
@xcite reads @xmath13 the analysis within ref .
@xcite has first focused on a one - loop expansion .
this latter entails considering the effective action @xmath14 the constants @xmath15 and @xmath16 are of order one , and depend on the number of colors @xmath11 of the su(n ) gauge - group . relying on the form of the effective lagrangian that has been derived for the asymptotically free regime , it has been shown that when the minimum is attained , namely when @xmath17 , @xmath18 , the energy density becomes @xmath19 the equation of state then reads like the one of dark energy , namely @xmath20 the issue of proving that the dark energy behavior of ymc is stable with respect to higher order loop - corrections has been addressed in @xcite and @xcite . in ref .
@xcite , the analysis has resorted to the two - loops effective lagrangian , @xmath21\,,\end{aligned}\ ] ] derived within the asymptotically - free regime and then extended to the infrared confining regime , in the range of values for @xmath1 in which @xmath22 has a minimum .
again , the coefficients @xmath23 and @xmath24 depend on the number of color @xmath11 of the gauge - symmetry group , and specifically read @xmath25 and @xmath26 , with @xmath27 . the energy density and the pressure from this model
are provided by the relations @xmath28 \ , , \\ & & p=\frac{b}{6 } \theta \left[\tau - 2 + \eta \left ( \ln |\tau + \delta| - \frac{2}{\tau + \delta } \right ) \right ] \,.\end{aligned}\ ] ] at high energies , when @xmath29 , the equation of state of the ymc evolves towards the equation of state of radiation @xmath30 approaching the infrared regime , at energies such that @xmath31 , the weak energy condition ( see _ e.g. _ @xcite ) is violated , and the equation of state parameter crosses the value @xmath32 .
notice however that , moving from the expression of @xmath22 determined in the asymptotically - free regime , while reaching the range of values for @xmath33 that ensure a dark energy behavior for the universe , a divergence is encountered in the effective coupling constant @xmath34 . although divergences do not occur in the observable quantities , which are the pressure @xmath35 , the energy density @xmath36 , and their ratio @xmath37 , the occurrence of a divergence in @xmath34 makes less trivial the act of relying on the extension of this procedure to higher - order loop expansions .
nevertheless , for the purposes of an analysis of the dark energy behavior , it has been shown that all the physical quantities behave like smooth functions in the range of the infrared values @xmath38 .
furthermore , the stability of the system accounting for interactions with matter and electromagnetic radiation , into which the ymc may decay , has been successfully checked . to achieve this goal , in @xcite
the authors have considered : the dynamical system provided by the first friedmann equation in presence of matter with energy density @xmath39 and ( electro - magnetic ) radiation with energy density @xmath40 @xmath41 the equations of motion for the energy densities component of ymc , matter and radiation that arise from conservation of the total energy - momentum tensor and from the decay of the ymc into matter . in comoving coordinates , in which @xmath42 , these latter equations read @xmath43 the decay rate @xmath44 is a free parameter of the model that enters the definition of the dimensionless dynamical system to be solved , equation ( [ h ] ) and equations ( [ emt ] ) .
the specific value of the rescaled parameter @xmath45 then affects the attractor coefficients in the analysis of the stability of the model .
consistency with big bang nucleosynthesis ( bbn ) @xcite , with the fractional density observed for dark energy ( @xmath46 ) and with the observational constraint on the equivalence between the energy densities of radiation and matter at the redshift of recombination , then set the initial conditions of the dynamical variables .
finally , the model has been shown to posses a dark energy tracking solution , and the dynamical system to have a fixed point , which is stable . for @xmath47 ,
the parameter of the equation of state takes the asymptotic value @xmath48 , provided that in @xmath22 one sets up @xmath49 .
a different value of this latter parameter , for instance @xmath50 , would rather entail @xmath51 .
a further analysis , developed along the same lines as the one reported above , has been deepened by the same authors in @xcite , and the investigation extended to the case of the three - loops effective action . within this latter work ,
the effective three - loops coupling constant reads @xmath52 \,,\end{aligned}\ ] ] and the main predictions have been substantially unaffected by the improvement in the perturbative analysis . for different choices of @xmath53 , which is the decay rate parameter rescaled by @xmath54 evaluated at present time , the parameter of the equation of state at present time takes the values @xmath55 ( given that @xmath56 ) or @xmath57 ( given that @xmath58 ) .
we emphasize that in both cases summarized above , the two - loops expansion within @xcite and the three - loops expansion within @xcite , the dark energy behavior arises from the perturbative computation of the effective lagrangian in the asymptomatically - free limit .
the validity of this procedure is then extended to the infra - red regime of the yang - mills theory , in order to recover a minimum in @xmath1 for the effective lagrangian , and derive the equation of state for the universe , which entails accelerated expansion .
but the occurrence of divergences in the effective action may shed some doubts on the validity of the conclusions for the dark energy behavior of the theory .
finally , we would like to point out that the gauge interactions taken into account here might not be necessarily considered to be the ones constituting the standard model of particle physics .
it is interesting to notice that an extra `` dark sector '' might be advocated to explain the gauge group here involved .
furthermore , it might be tempting to identify the gauge group copies with suitable candidates for dark matter , postulating the existence of `` dark copies '' of fermions colored under the extra `` dark gauge group '' .
the dark matter sector that is then introduced may eventually be connected to mirror standard model theories @xcite discussed in the literature .
within previous studies @xcite , the existence and the stability of a dark energy mechanism based on ymc has been investigated .
the analyses have been developed moving from the ultraviolet results recovered in the literature , up to three - loops in the effective action for yang - mills su(n ) fields , and results have been then extended in @xcite to the infrared regime , in order to derive a physical characterization of an accelerated expansion of the universe .
an important technical issue is that one that concerns the stability of the result at higher loops expansion , since the appearance of additional term in the effective action might indeed spoil stability , which totally relies on ultraviolet perturbative expansion .
furthermore , we know that the infrared regime of yang - mills su(n ) theories has a completely different behavior than the ultraviolet regime , the latter encoding asymptotic freedom , while the former showing a confinement behavior that depends on the number of gauge colors involved , and even on the number of colored fermions involved in the theory .
therefore , the main motivation of this study has been to prove that under mild and general assumptions , which are basically the existence of a minimum in @xmath1 in the non - perturbative effective lagrangian , a dark energy behavior is recovered .
then by making use of the non - perturbative techniques mutated from the functional renormalization group procedure , which is more adequate to be used in the confining infrared limit of the theory , it is possible to prove that a such a minimum indeed exists . at this purpose , we provided the explicit example of the su(2 ) yang - mills action , deriving the effective lagrangian for such a particular model and deepening the cases in which interactions with different forms of matter is considered .
the procedure , which might not be completely reliable for the precise determination of the coefficients of the effective non - perturbative lagrangian , is nevertheless enough to ensure that a minimum of @xmath1 exists for @xmath22 , and thus that a ymc works as reliable candidate to explain the origin of dark energy . in the following sections , we unfold detailed arguments in support of this thesis . within the perturbative ymc model for dark energy that we have reviewed in the previous section ,
the effective yang - mills lagrangian is that one calculated at one - loop in the seminal work by savvidy @xcite , namely @xmath59 where @xmath60 for su(n ) .
higher - loops correction have been then deployed in order to check whether the substance of the physical results remains unchanged , and how the details may vary .
+ instead of keep considering higher - loops improvements of , the main purpose of this work is to start from a general non - perturbative form of the action , _
i.e. _ @xmath61 we may then proceed to determine the form of @xmath62 by stating some general requirements that must be fulfilled in order to obtain a ymc that can work to explain the origin of dark energy .
the function @xmath63 will be in general equipped also with an energy scale @xmath8 in analogy with for dimensional reasons .
we notice indeed that @xmath63 must be a ( not necessarily analytic ) function of @xmath1 satisfying the following properties : 1 .
@xmath62 has a non trivial minimum in @xmath1 ; 2 .
@xmath62 possesses a perturbative limit , which reproduces the one - loop result derived by savvidy @xcite ; 3 .
@xmath62 shows the asymptotic behavior of being at least linear in @xmath1 , which in turn is linear to the bare yang - mills action .
this latter requirement can be formalized as it follows : in the ultraviolet regime @xmath64 it must hold the limit @xmath65 in what follows we will consider for simplicity a pure su@xmath66 yang - mills theory , the gauge field of which is not coupled to any other fundamental matter fields .
we want to stress that su@xmath66 yang - mills fields introduced here , as well as su(n ) yang - mills fields dealt with in @xcite , despite being suitable to build a model for cosmic dark energy should not be necessarily identified as standard model gauge fields .
the introduction of additional copy of su(n ) yang - mills fields might anyway allow to make contact with some mirror standard model theories @xcite that have been advocated in the literature of dark matter .
again , such a link is not necessary for our purposes and will not be analyzed within this investigation of dark energy ymc .
nevertheless it might represent some interesting directions to be followed in forthcoming studies . in what follows
, we discuss the cosmological consequences of the requirements we demanded above for the non - perturbative effective action , and shed light on the behavior of the ymc for the fate of cosmological dark energy .
we will assume a flat friedmann - lematre - robertson - walker ( flrw ) universe , the line element of which can be cast respectively in terms of comoving or conformal coordinates as it follows @xmath67,\end{aligned}\ ] ] where the cosmological time @xmath68 and the conformal time @xmath33 are related through @xmath69 .
we will consider the simplest case of a universe filled only with the ymc , and assume it to be minimally coupled to the gravity .
then the effective action reads for the metric , the riemann tensor and einstein equation respectively . ]
@xmath70 ~d^{4}x .
\label{s } \end{aligned}\ ] ] from now on , we will only denote with @xmath71 the determinant of the metric @xmath72 .
@xmath73 is the scalar ricci curvature , and @xmath22 is the effective lagrangian of the ymc , described by eq .
( [ l_eff_np ] ) .
varying @xmath74 with respect to the metric @xmath75 , one obtains the einstein equation @xmath76 , where the energy - momentum tensor of the ymc is given by @xmath77 we may set up a gauge that preserves the isotropy and the homogeneity of the flrw background . we write gauge fields as functions of the cosmological time @xmath68 , and choose their components so to satisfy @xmath78 and @xmath79 .
this choice indeed ensures that the total energy - momentum tensor @xmath80 is homogeneous and isotropic .
we then introduce the usual definition of the yang - mills tensor , cast in terms of the group structure constants @xmath81 , namely @xmath82 for the su@xmath66 gauge - group to which we are specializing our analysis , the structure constants reduce to @xmath83 .
furthermore , thanks to the gauge fixing we have selected above , and looking at the case of constant electric field ( we will assume for simplicity in the following a vanishing magnetic field ) , we can rewrite the yang - mills tensor in the simplified form @xmath84 this allows to express @xmath1 in a very simple form , _
i.e. _ @xmath85 , and each component of the energy - momentum tensor then rewrites as @xmath86 these tensors are not yet isotropic , their values depending on the direction of the color @xmath87 .
nevertheless , the total energy - momentum tensor @xmath88 is isotropic , and the corresponding energy density and pressure are given by @xmath89 consequently the equation of state ( eos ) of the ymc is immediately recovered to be @xmath90 it is worth discussing the mathematical properties of the eos @xmath91 .
on one hand , if we require the yang - mills theory to condensate , then the function @xmath63 must has a non trivial minimum , as we demanded in sec .
but this is equivalent to require that @xmath92 vanish at some point @xmath93 . in this point
the ymc has an eos of the cosmological constant with @xmath94 . around this critical point
the ymc dark energy models can account either for an eos characterized by @xmath95 or for an eos characterized by @xmath96 , thus encoding phantom matter behavior .
on the other hand , in the high - energy - scale regime , in which with @xmath97 , one would like to recover that the ymc exhibits an eos of radiation , characterized by @xmath98 , in analogy with the perturbative analysis @xcite . within the framework of the non - perturbative action
, this corresponds to the third requirement listed in sec .
the effective action should then scale for @xmath97 at least like the bare yang - mills action , _
i.e. _ @xmath99 .
+ in the following sections , we will show in detail that the ymc evolves from the eos with @xmath98 to @xmath94 while the universe is expanding .
the cosmological model we are about to analyze in this section entails three different sources for the energy - momentum tensor : i ) the dark energy , the role of which we assume to be played by the ymc ; ii ) the matter , including both baryons and dark matter , which is dealt with as non - relativistic dust with negligible pressure ; iii ) the radiation , the component of which consists of photons and possibly other massless particles , such as neutrinos .
we will describe the three components in terms of their eos , without accounting for any microscopic treatment in terms of fundamental matter fields .
+ since from eq .
( [ flrw ] ) we assumed _ ab initio _ the universe to be flat , fractional densities will sum up to one , namely @xmath100 .
indeed , fractional energy densities are defined as @xmath101 , @xmath102 , @xmath103 , and the total energy density is given by @xmath104 .
the overall expansion of the universe is determined by the friedmann equations : @xmath105 where the _ dot _ denotes the @xmath106 .
as a first preliminary investigation , we will assume that there is no interaction between the three energy components .
the dynamical evolution of these latter is determined by their equations of motion , which in turn follow from imposing the conservation of each component of the energy - momentum tensor , @xmath107 from eqs .
and , we can immediately obtain the standard evolutions of the matter and radiation components , @xmath108 and @xmath109 .
a less obvious but still rather simple task is solving the evolution of the ymc component .
inserting into , we obtain the following relation , @xmath110 which is in a quite compact form .
this equation is integrable for any regular enough @xmath63 . in particular
, we can easily derive the result @xmath111 where @xmath112 is a coefficient of proportionality that depends on the initial conditions .
at very high redshift , eq . entails an increase of the order - parameter @xmath1 that involves the limit @xmath113 .
thus , at very - high redshift the system transits towards the ultraviolet regime . within this limit ,
eq . encodes the eos parameter @xmath114 .
the ymc then starts behaving as the radiation component , as one would have expected since the theory is evolving towards asymptotic freedom at high energy . at small redshift
, the expansion of the universe requires the lhs of eq . to vanish , which occurs for the extremal value of @xmath115 , and the eos parameter then converges towards @xmath94 .
this ensure that the ymc behaves as an effective cosmological constant .
+ finally , notice that we may proceed as in @xcite , and take advantage of the observational constraint on the ratio between the dark - energy - density and the critical energy density , in order to fix the energy - scale @xmath116 that appears in the effective lagrangian @xmath62 .
the effective lagrangian is no more dependent on the parameter @xmath8 , and we can rescale @xmath1 as in the previous literature . in this section ,
we generalize the ymc dark energy model and take into account some effective interaction with dust matter . nevertheless , for the sake of simplicity , at this first stage of the analysis we will disregard the interaction between radiation and the ymc .
we will then describe the ymc dark energy and background matter interaction through one additional parameter @xmath117 .
the equations of the conservation of energy in and should be modified into @xmath118 the interaction parameter @xmath117 , in natural units , has the dimension of @xmath119 ^ 4 $ ] and has been introduced for phenomenological reasons .
its possible form will be addressed later .
if the ymc transfers energy to the matter , for instance if the ymc decays into pairs of matter particles , we should require the parameter @xmath117 to be positive .
vice - versa we should require the parameter @xmath117 to be negative .
+ we can then proceed to the study of the evolution of the system - .
it is convenient to introduce the so - called e - folding time @xmath120 , the derivative with respect to which we will denote with a prime .
we will denote as @xmath121 the dimensionless matter density @xmath122 .
the system - then reads @xmath123 using the above definitions , we can immediately recast the fractional energy densities of the ymc and the dust matter @xmath124 + it s useful to discuss the general properties of this dynamical system before specifying the form of the interaction term @xmath117 .
we can seek the fixed points of the system by imposing @xmath125 in eqs.- , and then look for the solutions @xmath126 and @xmath127 of of the simplified system derived from eqs . - .
this latter reads @xmath128 the stability of the solutions of these differential equations , and the possible existence of attractor solutions , will be analyzed in the forthcoming subsections , in which we will specialize the form of the coupling @xmath117 by assuming different linear combinations of the energy - density components considered so far . in this section ,
we parametrize the interaction as proportional to the ymc energy - density , namely @xmath130 . the trivial case @xmath131 reduces to the free ymc dark energy model studied above .
we will just consider the simplest case with @xmath112 being a non - zero dimensionless constant .
the evolution equations - then recast @xmath132 when the fraction density of the ymc is sub - dominant in the universe , we expect the effect on the dust to be small . only in the latest stage of expansion of the universe , where the ymc dark energy dominates its evolution , the effect of interaction on the dust component
can become important .
+ the critical point - now rewrites @xmath133 it s easy to see that the fractional energy densities of the ymc and the eos at this critical point do not depend on the details of the potential @xmath63 , but might rather have a dependence on the parameter @xmath112 .
it is straightforward to verify this from the very definition of eos calculated at the fixed point solution of eq . :
@xmath134 the constraint @xmath135 implies that @xmath136
. internal consistency of the model requires the solution of the system to be stable under perturbations .
the next step is then to require the critical point to be an attractor solution . in order to achieve
that , we first need to compute the eigenvalues of the linearized system - at the critical point : @xmath137 the solution is an attractor if @xmath138 in general we will need a specific form of @xmath63 to further discuss the nature of the critical point .
the next case to be considered hinges on an interaction of the form @xmath140 .
the evolution equations - now rewrites @xmath141 if we assume @xmath112 to be a non vanishing constant , we easily recover that @xmath142 .
this result might then lead to observational inconsistencies : the evolution of the dust component conflicts with the evolution of dust in the standard big - bang model .
we should then avoid considering such a form of interaction term in the early stage of evolution of the universe , at very high redshift .
nevertheless , even if we insisted in phenomenologically describing at small redshift the interaction between ymc dark matter and matter energy - density with a term of the form @xmath143 , we would find only a trivial critical point @xmath144 and @xmath145 .
thus , we must conclude that it is impossible to obtain an attractor solution for this kind of system . as a final case
, we can discuss a model where @xmath147 .
we limit again ourselves to the consideration of @xmath112 being a dimensionless constant . in the later stage ,
when the dark energy dominates the evolution , this model reduces to the case we studied in sec .
[ case1 ] , while in the dust dominated stage reduces to the case we studied in sec .
[ case2 ] . + in analogy to the discussion developed in the previous section ,
if we insisted in applying this model to the description of the early universe , the evolution of dust would turn to be incompatible with the prediction of the standard hot big - bang models .
nevertheless , if we want to develop a phenomenological model for small redshift , we may elaborate on this case .
the dynamical equations in - become @xmath148 and the system admits a critical point in @xmath149 the fractional energy density and the eos of the ymc at this critical point now read @xmath150 notice that now the observational constraint of @xmath135 sets a different range of allowed values for @xmath112 , namely @xmath151 .
we focus now to the analysis of the ymc model we have described in the previous sections , with a specific choice for the ir effective lagrangian .
we keep analyzing a ym theory that enjoys a su(2 ) gauge group , and show at non - perturbative level that a ymc forms that drives accelerated expansion of the universe at small redshift .
the functional renormalization group ( frg ) approach to non - abelian gauge theories will be particularly fruitful to our purposes , as it allows to introduce non - perturbative methods that can be treated as much as it is possible analytically .
the frg approach is a tool developed to study interacting qft and statistical systems in the non - perturbative regime , where no small coupling exists and perturbative techniques are not applicable .
the method is based upon a wilsonian momentum - shell wise integration of the path - integral : a mass - like regulator function @xmath152 suppresses quantum fluctuations with momenta lower than an ir momentum cutoff scale @xmath153 .
this allows us to define a scale - dependent effective action , the flowing action @xmath154 , which only contains the effect of quantum fluctuations with momenta greater than @xmath153 . by changing @xmath153
we can interpolate smoothly between the microscopic action @xmath155 and the full quantum effective action @xmath156 .
the scale - dependence of the flowing action is then given by the functional renormalization group equation ( frge ) @xcite and @xcite : @xmath157 herein , @xmath158 denotes the second functional derivative of the flowing action with respect to the fields , and constitutes a matrix in field space .
the super - trace @xmath159 includes a summation over all discrete indices and the fields , including a negative sign for grassmann valued fields , _ i.e. _ fermions and faddeev - popov ghosts .
the super - trace also includes a summation over the eigenvalues of the laplacian in the kinetic term .
the main technical advantage of the frge lies in its one - loop form , which nevertheless takes into account higher - loop effects , as it depends on the full , field - dependent nonperturbative regularized propagator @xmath160 ( see @xcite ) .
the frge has been extensively applied to @xmath161 yang - mills theories , for further references see @xcite and for the application of the frg to the study of ymc @xcite . solving
the equation exactly is a titanic task .
since we are mainly interested in qualitative and as much as possible analytic results , we will deploy some approximations .
( for the reader interested in the state of art of ymc in the frg framework , we refer to the work by eichhorn , gies and pawlowski @xcite , in which a numerical extrapolation between low and high momenta of full propagators were used to compute the gluon condensate . )
first of all we will replace @xmath162 in the rhs of equation with the bare action @xmath163 , doing so we are allowed to integrate both sides of the equation @xmath164 we select the bare action to be @xmath165 , that corresponds to the uv limit of our effective theory .
we will fix the integration constant requiring the effective action to vanish for vanishing field strength . to correctly invert the regularized propagator we need to include in the action a ( harmonic ) gauge - fixing and the associated faddeev - popov ghosts .
@xmath166 in the landau gauge @xmath167 , the trace over the gauge field space is restricted to the transverse sector @xmath168 with operators @xmath169 and @xmath170 , in which @xmath171 is the @xmath172 coupling and the barred quantities are made of background fields ; for more details on the actions and its variations we refer to the appendix [ actionym ] .
we will employ the simplest possible regulator functions ( mass - like cutoff ) @xmath173 in both the transversal gauge and ghost sector @xmath174 .
+ we can employ an integral representation ] nevertheless we will use the naive representation and implicitly regularize the final expression . ] of the logarithm , in order to find an exactly summed expression @xmath175 a wise choice of the background allow us to perform the traces as sums over the eigenvalues of the chosen kinetic operators .
in general the eigenvalues of the operator @xmath176 are not known , and the only known stable covariantly constant background is the self - dual one already employed in the frg context in @xcite ( the key properties needed for this work are summarized in the appendix [ selfdual ] ) . the effective lagrangian is finally recovered to be @xmath177 the `` magnetic field '' @xmath178 is the only variable of the self - dual background , which is related to the order parameter through @xmath179 . the next step is to remove the constant part from the integral that gets contribution from the lowest order expansion of the @xmath180 .
we then perform a change of variable , and reshuffle as @xmath181 the asymptotic behavior for small coupling constant @xmath171 of the integral is reproduced exactly at the lowest order the one - loop effective action .
furthermore , the ( unique ) non trivial minimum for is found to be at @xmath182 , as it can be read out from fig.1 .
[ picleff ] .,title="fig : " ] in order to check whether the effective lagrangian calculated in and derived for a su(2 ) ym theory can actually explain dark energy , we need to review if our example satisfies the properties we discussed in section [ snp ] . 1 . from fig.1
it s evident that the function has a non zero global minimum .
the exact position of this minimum can be computed numerically , and in terms of dimensionless quantities is found to be @xmath182 .
it is possible to compute the asymptotic expansion of @xmath183 for small values of the ym coupling constant @xmath171 . in this limit
we are able to reproduce the one - loop result derived by savvidy @xcite : if we perform the taylor expansion of @xmath184 , we find indeed @xmath185 3 .
@xmath183 shows an asymptotic behavior at least linear in @xmath1 .
this means that for @xmath186 the exponential in the integral tends to @xmath187 and the only @xmath1 dependence is the overall one , namely @xmath188 the condition @xmath189 then follows immediately . in the following sections , we show in detail that the ymc described by our toy model evolves from a radiation like component to a dark energy one .
we have already shown that in the case of a non - interacting ymc model the condensate evolution equation is implicitly solvable , and that the solution @xmath190 can be obtained by inverting the equation @xmath191 where @xmath112 is a coefficient of proportionality that depends on the initial conditions .
we can then fix the renormalization scale @xmath153 by comparing the `` predicted '' ymc fractional energy density with the measured dark energy fractional energy density ( @xmath192 ) finding for a big range of initial conditions @xmath193 .
this energy scale , as already noted in @xcite , is low compared to typical energy scales in particle physics , such as the qcd and the weak - electromagnetic unification , and this prevents the identification of this ymc as a condensate of some sm gauge fields
. then we can study the evolution of the ymc energy density and its eos for different values of @xmath194 , still obtaining the same asymptotic values .
results are summarized in fig .
[ pic2 ] on the left and @xmath195 on the right).,title="fig : " ] on the left and @xmath195 on the right).,title="fig : " ] for ymc models enjoying an interaction proportional to the ymc energy density , we have already discussed in section [ case1 ] the equation defining the existence of a fixed point derived from the differential equation evolution system - and the condition to be imposed on the coupling parameter @xmath112 in order to characterize an attractor solution .
here we report numerical computations on the position of the fixed point and on the value of the critical exponents at the fixed point .
the fixed point exists for every positive value of the coupling parameter and is always attractive , as shown in fig .
[ pic3 ] of the critical order parameter and matter density @xmath196 ( top panels ) and of the critical exponents @xmath197 at the fixed point ( bottom panels ) . in the bottom panel , on the left is reported the case in which @xmath129 , and on the right the case in which @xmath198.,title="fig : " ] of the critical order parameter and matter density @xmath196 ( top panels ) and of the critical exponents @xmath197 at the fixed point ( bottom panels ) .
in the bottom panel , on the left is reported the case in which @xmath129 , and on the right the case in which @xmath198.,title="fig : " ] of the critical order parameter and matter density @xmath196 ( top panels ) and of the critical exponents @xmath197 at the fixed point ( bottom panels ) .
in the bottom panel , on the left is reported the case in which @xmath129 , and on the right the case in which @xmath198.,title="fig : " ] of the critical order parameter and matter density @xmath196 ( top panels ) and of the critical exponents @xmath197 at the fixed point ( bottom panels ) . in the bottom panel , on the left is reported the case in which @xmath129 , and on the right the case in which @xmath198.,title="fig : " ]
the query whether the ymc may actually provide a consistent and physically reliable model for dark energy , along the lines of the analysis first developed in refs @xcite , has been addressed in this paper within the attempt of finding support to this theoretical hypothesis in the non - perturbative approach to the calculation of the yang - mills effective action .
in particular , we have first discussed the properties that the effective lagrangian @xmath62 must posses in order to drive the universe towards a cosmological dark energy phase .
if a condensate exists , _ i.e. _ if the non - perturbative effective action has a minimum in the ymc order - parameter @xmath1 , then the model can actually reproduce the dark energy behavior of the expanding universe at small redshift .
if the effective action scales at least like the bare yang - mills action for high - energy scale , at high redshifts it entails the eos of radiation .
moreover , internal consistency also requires that perturbative one - loop results must be still recovered in the appropriate asymptotic limit .
we have then focused on the particular example provided of the su@xmath66 yang - mills bare action .
we have deployed non - perturbative techniques mutated from the frg method , in order to show that for the su@xmath66 yang - mills a non - trivial minimum indeed exists , and that the high - energy scale regime approaches known perturbative results , and yields radiation dominated eos . despite our conclusions can be only based so far on this particular example , this successful check of the requirements necessary to have a viable ymc dark energy model seems to us extremely encouraging in strengthening the possibility that ymc can work as a model for dark energy .
the improvement of the non - perturbative techniques may allow in the future to extend the present analysis to the cases of su@xmath199 gauge - groups , or more in general to other classes of lie - groups . at this purpose
, we may either ask ourselves whether condensation can work for any su@xmath199 group , or whether consistency of the model necessarily predicts a maximal values of @xmath11 in order the mechanism to work .
not disconnected to these questions , it comes the query on the relation between the yang - mills fields involved , which are necessary in order the condensate to form , and the elementary - particle fields advocated to explain dark matter .
indeed , it would be tempting to try to link ymc dark energy models to other models for dark matter , such as the ones referred to in the literature @xcite as mirror standard model for dark matter .
we are grateful to yifu cai and roberto percacci for extremely useful discussions .
we consider the bare action of the yang - mills theory @xmath200 we split the field @xmath201 into a background plus a fluctuation @xmath202 .
the quadratic part of this action in the fluctuation field is @xmath203 where the bar quantities are made out of the background fields . to compute the inverse propagator we need to add also a gauge fixing action and the corresponding faddeev - popov ghost action : @xmath204 the frge splits into the trace over the transverse part and the longitudinal part of the connection field and the ghost sector : @xmath205 calculations are simplified by the fact that in the landau gauge @xmath167 the longitudinal trace
can be dropped out completely .
we may choose a background field configuration that allows to project onto the effective potential @xmath62 .
hence a covariantly constant field strength with @xmath206 suffices . since the spectrum of the laplace - type operators , like @xmath207 , or at least the heat - kernel trace for these operators has to be known
, we have a limited choice in the possible background field configurations . to avoid problems with tachyonic modes , which indicate the instability of a background , we project onto the only known stable covariantly constant background , which is self - dual , namely @xmath208
. then we set @xmath209 .
apart from @xmath210 all other non - zero components follow from the antisymmetry of the field strength tensor . due to the enhanced symmetry properties connected to the self - duality , zero modes exist .
these carry important information and have to be regularized carefully , since the standard choice for @xmath211 is zero on the zero mode subspace . with a degeneracy factor @xmath213 .
herein @xmath214 and @xmath215 is given by @xmath216 with the generators of the adjoint representation @xmath217 and therefore for a general gauge group depends on the direction of the unit vector @xmath218 .
a. g. riess _ et al .
_ [ supernova search team collaboration ] , astron . j. * 116 * , 1009 ( 1998 ) [ astro - ph/9805201 ] .
s. perlmutter _ et al .
_ [ supernova cosmology project collaboration ] , astrophys . j. * 517 * , 565 ( 1999 ) [ astro - ph/9812133 ] .
m. kowalski _ et al .
_ [ supernova cosmology project collaboration ] , astrophys .
j. * 686 * , 749 ( 2008 ) [ arxiv:0804.4142 [ astro - ph ] ] .
d. n. spergel _ et al . _
[ wmap collaboration ] , astrophys .
j. suppl .
* 148 * , 175 ( 2003 ) [ astro - ph/0302209 ] .
d. n. spergel _ et al . _
[ wmap collaboration ] , astrophys .
j. suppl .
* 170 * , 377 ( 2007 ) [ astro - ph/0603449 ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl . * 180 * , 330 ( 2009 ) [ arxiv:0803.0547 [ astro - ph ] ] . s. cole _ et al . _
[ 2dfgrs collaboration ] , mon . not .
soc . * 362 * , 505 ( 2005 ) [ astro - ph/0501174 ] .
m. tegmark _ et al . _
[ sdss collaboration ] , phys .
d * 74 * , 123507 ( 2006 ) [ astro - ph/0608632 ] .
s. capozziello , s. carloni and a. troisi , recent res .
astrophys .
* 1 * , 625 ( 2003 ) [ astro - ph/0303041 ]
. s. m. carroll , v. duvvuri , m. trodden and m. s. turner , phys .
d * 70 * , 043528 ( 2004 ) [ astro - ph/0306438 ] .
l. amendola and s. tsujikawa , _ dark energy : theory and observations _
( cambridge university press , cambridge , uk , 2010 ) .
u. ellwanger , m. hirsch and a. weber , z. phys .
c * 69 * ( 1996 ) 687 ; eur .
j. c * 1 * ( 1998 ) 563 ; b. bergerhoff and c. wetterich , phys .
d * 57 * ( 1998 ) 1591 ; [ hep - ph/9708425 ] .
j. kato , hep - th/0401068 .
j. m. pawlowski , d. f. litim , s. nedelko and l. von smekal , phys .
* 93 * ( 2004 ) 152002 [ hep - th/0312324 ] ; j. m. pawlowski , d. f. litim , s. nedelko and l. von smekal , aip conf . proc .
* 756 * , 278 ( 2005 ) [ arxiv : hep - th/0412326 ] . | models based on yang - mills condensate ( ymc ) have been advocated in the literature and claimed to be successful candidates to explain dark energy .
several instantiations of this simple idea have been considered , the most promising of which are reviewed here .
nevertheless , results previously attained heavily relied on the perturbative approach to the analysis of the effective yang - mills action , which is only adequate in the asymptotically - free limit , and were extended into a regime , the infrared limit , in which confinement is expected .
we show that if a minimum of the effective lagrangian in @xmath0 exists , a ymc forms that drives the universe toward an accelerated de sitter phase .
the details of the models depend weakly on the specific form of the effective yang - mills lagrangian . using non - perturbative techniques mutated from the functional renormalization group procedure
, we finally show that the minimum in @xmath1 of the effective lagrangian exists , thus ymc can actually take place .
the non - perturbative model has properties similar to the ones of the perturbative model . in the early stage of the universe ,
the ymc equation of state has an evolution that resembles the radiation component , i.e. @xmath2 .
however , in the late stage , @xmath3 naturally runs to the critical state with @xmath4 , and the universe transits from matter - dominated into dark energy dominated stage only recently , at a redshift the value of which depends on the initial conditions that are chosen while solving the dynamical system . |
recently , a diphoton signal excess at 750 gev was reported in the data of large hadron collider ( lhc ) run 2 with @xmath1 collisions at energy @xmath2 tev @xcite .
it can be explained by a sm singlet @xmath0 , with production cross section @xcite , @xmath3 the local significance of this excess is about 3.9 @xmath4 and 2.6 @xmath4 for atlas and cms , respectively , and it was not seen in the data of lhc run 1 with @xmath5 tev @xcite .
although no excesses in @xmath6 , @xmath7 , @xmath8 , dilepton and dijet channels @xcite were observed yet both in the old data of run 1 and the first data of lhc run 2 , this bump has stimulated great interests . in this paper , we consider the interest of building connections between the sm singlet scalar @xmath0 responsible for the diphoton excess and weakly - interacting massive dm models . inspired by the construction of sm higgs - portal dm models @xcite , in which the dm communicates with the sm particles via the higgs mediator , we propose @xmath0-portal dm model by replacing the higgs scalar with @xmath0 . for discussions about dm roles in diphoton excess in eq.([excess ] ) ,
@xcite .
our construction of @xmath0-portal dm obviously differs from the higgs - portal dm models due to obviously different mediator scalar mass and yukawa coupling .
but they indeed share a common feature , i.e. , they are both effective theories at the electroweak ( ew ) scale .
this paper is organized as follows . in sec .
[ sec2 ] , we define our model introduce model parameters . in sec
. [ sec3 ] , we consider the constraints on @xmath9-sector from diphoton excess in eq.([excess ] ) and 8-tev limits at lhc .
in particular , parameter space should be consistent with the 8-tev @xmath10 limit .
dm of type either a sm singlet scalar or dirac fermion are both addressed . in sec .
[ sec4 ] , we consider the constraints on dm - sector from dm relic abundance @xcite and direct detection limits @xcite . in sec . [ sec4 ]
we add a few comments on dm direct detection at lhc run 2 . finally , we conclude in sec .
we assume that the effective theory of new physics model includes a scalar @xmath0 responsible for the diphoton excess , a fermion @xmath9 charged under sm gauge group @xmath11 with electric charge @xmath12 in unit of @xmath13 , and a sm singlet scalar dm ( @xmath14 ) or a sm singlet fermion dm ( @xmath15 ) . the tev - scale effective lagrangian @xmath16 for this model is given by , @xmath17 where @xmath18 we identify @xmath19 gev and ew scale @xmath20 gev .
the yukawa coupling @xmath21 in the case for scalar dm has been normalized to a dimensionless parameter . for either a scalar or fermion
dm there are four following model parameters in eq.([lag ] ) , @xmath22 similar to higgs - portal singlet scalar dm @xcite , a @xmath23 parity , under which @xmath14 is odd and the others are even , is employed to keep dm stable .
instead of writing the interactions between @xmath0 and gluons and photons via operators with mass dimension five , in ( [ lag ] ) we consider an explicit realization via fermion @xmath9 . into vector - like quark models .
we temporarily assume that those fermions ignored here are not relevant for our study . ]
the advantage is obvious , as the number of model parameters are reduced .
see , e.g. , @xcite . .
limits in the data of lhc run 1 are shown in curves , above which regions are excluded.,width=384 ]
the decay channels for @xmath0 include @xmath24 and an extra one @xmath25 in the low mass region @xmath26 .
some of these magnitudes satisfy @xmath27 while the others such as @xmath28 are all of the same order .
obviously , larger branching ratio @xmath29 in eq.([width ] ) can be obtained by choosing larger @xmath12 . in this paper
we simply choose the electric charge 5/3 for @xmath9 for the following facts .
at first , the dark matter phenomenology discussed in sec .
iv does not significantly affected by the choice on electric charge slightly bigger than @xmath30 for @xmath9 as favored by the diphoton excess at 750 gev .
secondly , those results arising from electric charge @xmath30 in sec.iv are useful reference for the case where @xmath9 is also charged under @xmath31 singlet , the electric charge for @xmath9 can be alternatively chosen bigger than @xmath30 .
see our previous work @xcite for relevant discussion .
in contrast , if @xmath9 is further charged under @xmath32 , the mass region @xmath33 gev has been excluded by the 8-tev lhc data @xcite for the assumption @xmath34 .
similarly , see ref .
@xcite for lhc lower mass bounds for @xmath35 . ] . in the setting on small other than large decay width for @xmath0 ,
@xmath36 is restricted to be above @xmath37 .
the production cross section @xmath38 is mainly through gluon fusion , the magnitude of which depends on yukawa coupling @xmath39 and fermion mass @xmath36 .
given the fact that the decay width ratio @xmath40 is nearly fixed for the explicit choice on electric charge @xmath12 , the diphoton signal strength @xmath41 is sensitive to @xmath39 and fermion mass @xmath36 . in fig.[signal ]
the yellow band corresponds to the contours of observed diphoton signal strength , with limits in the data of lhc run 1 are shown in curves .
it indicates that the narrow region in the yellow band below the limit given by @xmath42tev can explain the diphoton excess and is consistent with the following limits @xcite at @xmath43-tev lhc simultaneously , @xmath44 it also implies that the value of @xmath39 as required is in the perturbative region for @xmath36 below 2 tev . :
feynman diagrams for dm annihilation .
@xmath45 : feynman diagram for dm - nucleon scattering cross section.,width=268 ] : feynman diagrams for dm annihilation .
@xmath45 : feynman diagram for dm - nucleon scattering cross section.,width=268 ]
in this section we take the dm view on our model parameters .
the yukawa interaction between dm and @xmath0 responsible for diphoton excess provides the @xmath46-channel annihilation @xmath47 as shown in the left panel of fig.[feynman ] .
this annihilation should account for the dm relic abundance as measured by the plank and wmap 9-year data @xcite , @xmath48 by employing micromegas @xcite we show in fig.[relic ] the contours of @xmath49 in the parameter space of @xmath21 and @xmath50 with dm either a scalar or dirac fermion for three representative choices on @xmath51 gev .
@xmath52 in fig.[relic ] shows that once @xmath53 is opened in the mass region @xmath54 , annihilation cross section significantly increases , which leads to smaller @xmath21 as required by the relic abundance in eq.([relic ] ) .
similar phenomenon occurs in the @xmath45 for dirac fermion dm , although it changes more mildly .
this difference may be due to the particular normalization we have chosen for scalar dm in eq.([lag ] ) .
the requirement of dark matter relic density as indicated by the right panel of fig.3 implies that perturbative analysis is not valid for fermion dm with mass roughly below 400 gev .
conversely , this requirement as indicated by the left panel of fig.3 implies that perturbative analysis is not valid for scalar dm with mass roughly above 2.5 tev instead . unlike the higgs - portal singlet dm model where dm - nucleon scattering proceeds via tree - level process , , in our model dm - nucleon scattering proceeds via a loop process instead
, as shown in the @xmath55 of fig.[feynman ] , where the intermediate vector boson in the @xmath55 is either sm gluon or photon . as a result , the @xmath0-nucleon scattering cross section @xmath56 is relatively suppressed in compared with a tree - level process .
fig.[si ] shows the contours of @xmath56 as function of dm mass for either scalar ( @xmath57)or dirac fermion ( @xmath55 ) dm , which clearly shows that it is consistent with all present direct detection limits .
it also implies that there is no prospect for discovery even in the further experiment at xenon1 t .
projected to the parameter space of @xmath21 and @xmath50 . @xmath52 and @xmath45 corresponds to a scalar and dirac fermion dm , respectively . in each case , we take three representative choices on @xmath51 gev .
the uncertainty for @xmath21 is due to the uncertainty of @xmath39 as shown in fig.[signal].,width=307 ] projected to the parameter space of @xmath21 and @xmath50 . @xmath52 and @xmath45 corresponds to a scalar and dirac fermion dm , respectively . in each case , we take three representative choices on @xmath51 gev .
the uncertainty for @xmath21 is due to the uncertainty of @xmath39 as shown in fig.[signal].,width=307 ] as function of dm mass for either scalar ( @xmath57)or dirac fermion ( @xmath55 ) dm for three representative choices on @xmath51 gev , which clearly show that our dm model is consistent with all present direct detection limits .
they also indicate that there is no prospect for discovery at further xenon1 t experiment .
similar to fig.[relic ] , the uncertainty for @xmath21 is due to the uncertainty of @xmath39 in fig.[signal].,width=307 ] as function of dm mass for either scalar ( @xmath57)or dirac fermion ( @xmath55 ) dm for three representative choices on @xmath51 gev , which clearly show that our dm model is consistent with all present direct detection limits .
they also indicate that there is no prospect for discovery at further xenon1 t experiment .
similar to fig.[relic ] , the uncertainty for @xmath21 is due to the uncertainty of @xmath39 in fig.[signal].,width=307 ]
the dm can be directly detected at lhc via dm pair production .
it may contribute to excess in sm multi - jets plus missing energy .
the signal strength of multi - jet process @xmath59 through gluon fusion is suppressed by many - body final states .
consider that the dm is typically in the mass range between 400 gev and 3 tev and the sm background is very large , a large luminosity is required for discovery of dm at the lhc run 2 .
this issue will be addressed elsewhere in detail @xcite .
in this paper we propose a type of new dm models , in which the scalar responsible for the diphoton excess at 13-tev lhc mediates the interactions between dm and sm . in the so - called @xmath0-portal dm models , after taking into account the lhc constraints and dm direct detection limits , we show that in the perturbative framework dm either a sm singlet scalar or dirac fermion can be allowed in a wide mass range between 400 gev and 3 tev . with high integrated luminosity ,
the dm can be directly detected in sm multi - jets and missing energy . | a new type of dark matter ( dm ) theories are proposed in the light of the standard model ( sm ) singlet scalar @xmath0 which is responsible for the diphoton excess at the lhc run 2 . in the so - called @xmath0-portal dm models , after taking into account the lhc constraints and dm direct detection limits , we show that in the perturbative framework dm as either a sm singlet scalar or dirac fermion can be allowed in a wide mass range between 400 gev and 3 tev .
the dm can be directly detected in sm multi - jets and missing energy . |
let us consider magnetotactic bacteria which move in the direction of their magnetic moment @xmath2 or opposite to it due to random switching of the direction of rotation of a rotary motor .
since in the synchronous regime the magnetic moment rotates synchronously with the field , so does the velocity of the bacterium .
the master equation accounting for a random switching rate @xmath3 of the direction of a rotary motor rotation reads @xmath4 taking into account that @xmath5 , where @xmath6 , eq . ( [ eq:1 ] ) may be rewritten for the two - dimensional case in a more simple way as follows @xmath7 there are several possible ways to analyse the properties of random trajectories of these circle swimmers .
a simple approach is based on the derivation from ( [ eq:1a ] ) of a closed set of equations for several moments characterizing their distribution @xmath8 .
it reads @xmath9 in order to derive a set of equations ( 3 ) the condition of normalization @xmath10 and the relation @xmath11 are used . since the set ( 3 )
is closed it is possible to obtain an analytical solution for @xmath12 , which characterizes the random process of the particle diffusion . the stationary solution for the moments @xmath13 reads @xmath14 @xmath15 relations ( [ eq:4],[eq:5 ] ) , as a result , give @xmath16 taking into account 2d character of particle motion , absence of cross - correlation @xmath17 in the stationary case and @xmath18 due to the symmetry , for the diffusion coefficient we have @xmath19 in the limit @xmath20 we have @xmath21 introducing as a characteristic time of tumbling @xmath22 relation ( [ eq:8 ] ) reads @xmath23 it coincides with the expression for the diffusion coefficient of tumbling bacterium @xcite with @xmath24 , where @xmath25 is angle by which the bacterium changes direction of motion after a tumbling event . since in our case direction of the bacterium motion changes to the opposite we have @xmath24 . in the limit of high frequency for the diffusion coefficient we have @xmath26 , where @xmath27 is the radius of a circular trajectory , which coincides with the expression of the diffusion coefficient in the elementary theory of brownian motion @xcite .
we should remark that besides the diffusion due to random switching of the rotary motor of the bacterium there is also thermal noise due to fluctuating forces and torques .
this problem is considered in @xcite . in @xcite calculation
is carried out by using time - correlation functions .
we will illustrate further that the diffusion coefficients of torqued swimmers may be calculated in a similar fashion as above on the basis of the fokker - planck equation for the joint distribution function of the position and orientation of the swimmer @xmath28 : @xmath29 where @xmath30 is the operator of infinitesimal rotations and @xmath31 and @xmath32 are translational and rotational diffusion coefficients respectively . in 3d case
the following closed set of equations for the moments may be obtained @xmath33 in the stationary case we have @xmath34 this gives , in agreement with @xcite , @xmath35 and @xmath36 similar to ( 14 ) relation in 2d case , up to our knowledge , was for the first time obtained in @xcite .
we see that the frequency dependence of the diffusion coefficient is similar to the one for the magnetotactic bacterium under the action of the poisson like noise .
the role of the switching rate @xmath3 is determined in this case by the rotary diffusion coefficient of the swimmer @xmath32 .
taking for the rotational drag coefficient of the bacterium the value @xmath37 @xcite we may estimate the characteristic switching rate due to the rotational diffusion as @xmath38 , which is much less than expected due to the switching of the rotary motors of a bacterium .
this shows that random trajectories observed in @xcite of bacteria in a rotating field are due to the random switching of rotary motors and not due to the random thermal noise .
another evidence for this comes from direct comparison of the trajectory observed in the experiment ( see fig.11 in @xcite ) and trajectory obtained numerically as described in @xcite and shown in fig.1 .
parameters used in generating the trajectory are similar to those of the experiment .
trajectory length is @xmath39 , @xmath40 . from the experiment estimated radius of rotation is @xmath41 and diffusion coefficient @xmath42 which give @xmath43 and , according to ( [ eq:7 ] ) , @xmath44 . concerning the relation ( [ eq:7 ] )
we may point out that for given frequency of rotating field @xmath45 the diffusion coefficient has a maximum at definite switching rate @xmath46 .
thus a bacterium , which is in unfavorable conditions and is confined to a definite area by a rotating field has a possibility to increase its diffusion coefficient for seeking more favorable conditions by adopting the switching rate of the rotary motor
. it may be worth to notice that optimal switching rate given by the relation @xmath46 in the case of a rotating field with frequency @xmath47 equals to @xmath48 - a value close to the characteristic switching and tumbling events of bacteria @xcite .
it is interesting to remark that similar mechanism of the enhancement of the diffusion coefficient of marine bacteria was considered in @xcite where it was illustrated that time modulation of the swimming velocity may increase the diffusion coefficient by an order of magnitude .
maximum of the diffusion coefficient for the centers of spiral waves in dependence on the correlation time of random noise was found in @xcite .
the set of equations ( 3 ) allows us to consider the transitory regime of establishment of the brownian diffusion .
seeking the solution exponential in time @xmath49 of linear homogeneous set of equations @xmath50 we obtain @xmath51 . as a result
the solution for these moments satisfying zero initial conditions reads @xmath52 and @xmath53 by integration we find the mean square displacement @xmath54 at @xmath55 for the mean square displacement we have @xmath56 at small times the motion of the bacterium is ballistic @xmath57 .
in the synchronous regime tangent vector @xmath58 to the trajectory , which we choose along the magnetic moment of a bacterium , rotates with the angular velocity of the field : @xmath59 where @xmath60 . since , according to the frenet equation , @xmath61 and @xmath62 ( the bacterium moves along its magnetic moment or opposite to it ) we have @xmath63 and thus @xmath64 .
it means that at constant rotary motor rotation direction the trajectory is an arc of circle with radius @xmath65 .
the radius vector of the curvature center of the trajectory is @xmath66 .
it means that the position of the curvature center jumps by @xmath67 at the switching of the rotary motor . for numerical implementation of the algorithm it is useful to introduce complex notations @xmath68 and @xmath69 . in dimensionless form
the equations for a random trajectory of the bacterium read ( @xmath70 ) @xmath71 for numerical simulation of the trajectory a set of switching times @xmath72 distributed according to the poisson law @xmath73 are introduced . at time moments @xmath74 the random variable @xmath75 equal to @xmath76 changes the sign .
the generated trajectory is used to obtain diffusion coefficient @xmath31 as a function of the switching rate @xmath3 .
two examples of computed mean square displacement as a function of time difference is shown in fig.2 .
it is evident that the relationship is not linear at small times .
slope of the best - fit line at large times is an estimate of @xmath77 . the estimated diffusion coefficient for a range of @xmath3 values is shown in fig.3 . all trajectories used consist of @xmath78 switches of swimming direction .
the results are in very good agreement with ( [ eq:7 ] ) . in @xcite , when considering sperm trajectory curvature fluctuations @xmath79 , it is found that they introduce effective diffusion coefficient given by ( @xmath80 ) @xmath81 where @xmath82 is a power spectrum of the curvature fluctuations , found at the circulation frequency @xmath83 . in our case of a synchronous regime of motion of bacteria in a rotating field @xmath84 is equal to the frequency of the rotating field @xmath45 .
the relation ( [ eq:20 ] ) is in agreement with the numerical simulation results .
taking into account that for the poisson process the probability of even number of switching events in time @xmath85 is @xmath86 and the probability of odd number of switching events is @xmath87 for the curvature time correlation function we have @xmath88 and the power spectrum of curvature fluctuations is @xmath89 relations ( [ eq:20 ] ) and ( [ eq:21 ] ) give the relation ( [ eq:7 ] ) .
it should be noticed that according to @xcite the relation ( [ eq:21 ] ) has a more broad context than a particular example considered here and includes the noise in chemotactic signalling system and other .
analysis of the random trajectories of the circle swimmers - the magnetotactic bacteria shows that experimentally observed random trajectories of bacteria are caused by the internal noise of rotary motors and are not due to the thermal fluctuations .
derived on the basis of the master equation relation for the diffusion coefficient is in an excellent agreement with the results of numerical simulations and allows one to determine the characteristic rate of the random switching events of rotary motors from the available experimental data .
authors acknowledge funding from the european union though the fp7 project bio2man4mri ( grant no.245542 ) .
ac is thankful to organizers of the 2nd french - brazilian meeting on nanoscience , nanotechnology and nanobiotechnology ( december 2012 , brazil ) .
15 s.ebbens , r.a.l.jones , a.j.ryan , r.golestanian , and j.r.howse , _ phys.rev.e _
_ 82 _ 015304(r ) ( 2010 ) m.sandoval , _
_ 87 _ 032708 ( 2013 ) f.kummel , b.ten hagen , r.witkowski , i.buttinoni , g.volpe , h.lowen , and c.bechniger , arxiv:1302.5787v ( 23 february 2013 ) b.m.friedrich , f.julicher , _ new journal of physics _ _ 10 _ 123025 ( 2008 ) d.takagi , a.b.braunschweig , j.zhang , and m.j.shelley , _ phys.rev.lett .
_ _ 110 _ 038301 ( 2013 ) k.erglis , qi wen , v.ose , a.zeltins , a.sharipo , p.a.janmey , and a.cebers , _ biophysical journal _
_ 93 _ 1402 ( 2007 ) a.cebers , _ jmmm _ _ 323 _ 279 ( 2011 ) k.martens , l.angelani , r.di leonardo , and l.bocquet , _ eur.phys.j.e _ _ 35 _ 84 ( 2012 ) h.berg , _ random walks in biology _ , _ princeton university press _ , ( 1983 ) ph.nelson , _ biological physics _ , _ w.h.freeman and company , new york _ ( 2003 ) c.douarche , a.buguin , h.safman , and a.libchaber , _ phys.rev.lett . _
_ 102 _ 198101 ( 2009 ) e.lauga , _ phys.rev.lett . _
_ 106 _ 178101 ( 2011 ) i. sendifia - nadal , s.alonso , v.perez-munuzuri , m.gomez-gesteira , v.perez-villar , l.ramirez-piscina , j.casademunt , j.m.sancho , and f.sagues , _ phys.rev.lett . _
_ 84 _ 2734 ( 2000 ) | properties of active colloids of circle swimmers are reviewed . as an particular example of active magnetic colloids
the magnetotactic bacteria under the action of a rotating magnetic field is considered .
the relation for a diffusion coefficient due to the random switching of the direction of rotation of their rotary motors is derived on the basis of the master equation .
the obtained relation is confirmed by the direct numerical simulation of random trajectory of a magnetotactic bacterium under the action of the poisson type internal noise due to the random switching of rotary motors .
the results obtained are in qualitative and quantitative agreement with the available experimental results and allows one to determine the characteristic time between the switching events of a rotary motor of the bacterium .
active systems magnetotactic bacteria circle swimmer diffusion properties of random trajectories of so - called circle swimmers have been studied by various groups of researchers @xcite .
various kinds of these swimmers and their properties have been investigated .
effective diffusion coefficients of circle swimmers subjected to thermal noise are calculated in @xcite .
circle swimmers with an asymmetric l - shape self - propelling due to the self - phoresis at illumination by light were synthesized and studied in @xcite .
circular dynamics of slightly bent self - propelling rods switching their direction of circulation due to thermal fluctuations are studied in @xcite .
a specific example of circle swimmers is magnetotactic bacteria , which in a rotating magnetic field in the absence of noise move along circular trajectories , if the frequency of the rotating field is less than a critical value @xcite .
a random walk of the centers of circular trajectory is observed due to random switching of the direction of rotation of their flagella @xcite .
here the properties of this random walk are studied on the basis of the master equation @xcite for the probability density distribution function @xmath0 of position of the bacterium @xmath1 and direction of its magnetic moment @xmath2 .
bacteria moves in @xmath2 direction or opposite to it .
it is illustrated that this method may be applied also to circle swimmers of other types . |
the existence of vesicles at thermal equilibrium is a long - standing , and still controversial , problem @xcite .
unilamellar and multi - lamellar vesicles ( mlvs ) were first induced in lyotropic lamellar phases with excess water by adding energy ( _ e.g. _ shear flow , ultrasound , electric field ) @xcite . in a very elegant work
, using membranes generated by a chemical reaction , hoffman and co - workers demonstrated that any of lamellae , unilamellar vesicles , or mlvs can be prepared in the same system , depending on the mechanical path chosen @xcite .
however in some special cases @xcite _ equilibrium _ unilamellar vesicles have been demonstrated .
these systems are all in the dilute regime ( large excess of water ) .
some experiments have suggested that mlvs ( or onions ) can be stabilized in the semi - dilute regime ( excess water ) @xcite .
theories to explain the stability of dilute unilamellar vesicles either describe a competition between the entropy of mixing and the curvature energy of the vesicles @xcite or a symmetry - breaking instability leading to a spontaneous curvature @xcite .
onions are also predicted to be stabilized in the dilute and semi - dilute regimes due to an unstable curvature energy ; in these cases , a transition towards unilamellar vesicles is avoided by imposing either a core energy @xcite or a cutoff in the entropy of the helfrich interactions @xcite .
indeed , simons and cates @xcite have discussed the stability of unilamellar vesicles and the transition to onions as the concentration increases even when the curvature energy is unfavorable , due to entropic reasons . in the concentrated regime ( homogeneous lamellar phase with no excess water ) , very monodisperse onions can be prepared under shear @xcite .
the vesicles fill space and remain in the one phase region without expelling solvent , by distorting into space - filling polyhedra @xcite .
the resulting texture is a lattice of disclinations which can be either disordered @xcite or ordered @xcite .
moreover many experimental systems `` spontaneously '' exhibit onions in regions of the phase diagrams that seem to be continuously linked to the lamellar phase @xcite .
this evolution from a texture of polydisperse vesicles to a texture of focal conics upon tuning a parameter , such as a co - surfactant concentration or the volume fraction of solvent , seems to be generic .
a striking signature of this evolution is the change in the rheology , from a viscous gel in the `` onion '' phase to a more fluid phase in the `` lamellar '' focal conic phase .
thus , whether an onion texture is at equilibrium or not seems yet unclear . in this work
we propose a model based on a non - quadratic elastic energy , which shows that a pure lamellar phase , or one of space - filling deformed monodisperse onions , is unstable with respect to a polydisperse space - filling packing ( we study an apollonian packing @xcite ) .
we compare these predictions to experiments on copolymer - doped lamellar phases , originally introduced by ligoure _
the bending free energy per area @xmath0 for membranes is usually taken , following helfrich , as the simplest quadratic function of the mean and gaussian curvatures , respectively @xmath1 and @xmath2 , where @xmath3 and @xmath4 are the two principal radii of curvature @xcite .
this suffices for large curvature radii @xmath5 , where @xmath6 is the layer thickness .
however , if onions exist , because of the high concentration of membrane , high curvatures ( small @xmath7 ) are expected near the core of the onions .
we will thus use the free energy , relative to a flat state , of a symmetric bilayer expanded to quartic order in the curvature radii @xcite : @xmath8 where @xmath9 and @xmath10 are the conventional bending and gaussian curvature moduli , and the non - linear moduli @xmath11 can be expected to scale as @xmath12 .
stability of a flat membrane requires @xmath13 . for @xmath14 an instability towards a phase with @xmath15 ( _ e.g. _ a bicontinuous phase ) is possible .
however , for @xmath16 spherical shells are stabilized , with free energy ( for radius @xmath7 ) @xmath17 given by @xmath18 where @xmath19 .
positive @xmath20 stabilizes spherical shells ( vesicles ) of finite radius .
unilamellar vesicles can not accommodate volume fractions close to space filling , so we consider the free energy of an onion of radius @xmath21 , constructed from @xmath22 shells of discrete radii @xmath23 set by the layer spacing @xmath24 , which we assume remains fixed : @xmath25 \label{eq:4}\end{aligned}\ ] ] where @xmath26 is the balance between the helfrich and non - linear elastic coefficients and @xmath27 . in the limit @xmath28 the second term above , due to the fourth order coefficients , is equivalent to the core energy introduced by fournier and durand @xcite . for large dilute monodisperse onions the free energy per unit volume scales as @xmath29 , leading for @xmath30 to stable onions with size @xmath31 .
however , in the concentrated regime monodisperse onions can not fill space without non - trivial deformations @xcite .
typically , onions formed under shear pack as polyhedra , with the deformation of layer @xmath32 concentrated along the edges of length @xmath33 with curvatures of order @xmath34 , where the smectic bending penetration length @xmath35 depends on the bending @xmath36 and compression @xmath37 moduli of the lamellar phase @xcite . summing over all shells in a deformed polyhedron leads to an energy per unit volume that scales as @xmath38/(r \xi d^3)$ ] . for @xmath39 this leads to an unphysical unilamellar vesicle phase with @xmath40 ( the lower cutoff ) .
however , a polydisperse phase can relax the deformation , leading , for a broad enough distribution to a space - filling distribution of spheres .
a candidate space - filling distribution is the apollonian packing , for which the number distribution of spheres @xmath41 , with a fractal dimension estimated from simulations to be @xmath42 @xcite ( see fig . [ fig : app1]c ) .
such a packing has been previously suggested to describe focal conic defects in smectic droplets @xcite . in principal we should calculate this distribution from a free energy that incorporates entropy .
although small on the typical scale of the elastic terms , this entropy should widen the parameter range in which onions are stable @xcite .
existing equations of state for , _
e.g. _ polydisperse hard spheres , fail for widely disperse distributions near space filling @xcite , so we have been unable to estimate this contribution to any useful accuracy .
entropic effects will most likely influence the upper and lower cutoffs of the distribution , and would eventually lead ( for highly fluctuating systems ) to an exponential , rather than power law , size distribution away from space filling @xcite . , as a function of @xmath26 .
apollonian onions are stable for @xmath43 .
( b ) schematic phase diagram in the @xmath44 plane , @xmath45 .
( c ) schematic packing of the apollonian distribution ( reprinted with permission from @xcite).,title="fig : " ] the normalized number distribution of onion sizes @xmath46 obeying @xmath47 , where @xmath48 is the volume of lamellar material , is @xmath49 the total energy of the distribution , @xmath50 , is @xmath51 this energy can then be minimized over the maximum onion size @xmath52 as a function of the single parameter @xmath53 to find the lowest free energy distribution ( fig . [ fig : app1]a ) .
although spherical shells are stable for @xmath30 , the stabilizing elastic constants @xmath11 assure that a finite and negative @xmath54 is required for onion formation .
onions are first stable with ` infinite radius ' , implying a smooth transition from lamellae to onions .
however , the scaling @xmath55 implies an unbounded total volume .
hence , the size distribution must have an upper cutoff , which in practice should be the smallest dimension of the sample container .
( in fact , the nature of the size distribution for apollonian packings of spheres is still under debate @xcite , and appears to depend crucially on the manner in which the packing is constructed . )
the maximum onion size then decreases , and for @xmath56 the stable phase is of unilamellar vesicles . at this point the calculation has presumably broken down , and either another phase such as micelles intervenes , or even higher order curvature terms become important . because @xmath57 , and the volume fraction @xmath58 is of order @xmath59 , the condition @xmath60 implies a phase boundary @xmath45 , as shown in fig .
[ fig : app1]b , in the @xmath61 plane . and copolymer / surfactant weight ratios @xmath62 of ( a ) @xmath63 , @xmath64 ; ( b ) @xmath63 , @xmath65 ; ( c ) @xmath63 , @xmath66 ; ( d ) @xmath67 , @xmath66 .
bottom : gel phase in contact with water ; ( e , crossed polarizers ) @xmath63 , @xmath68 ; ( f , differential interference constrast microscopy ) @xmath63 , @xmath69 .
scale bar @xmath70 . ]
as shown theoretically in @xcite , adsorption of amphiphilic copolymers to surfactant bilayers is expected to decrease @xmath71 , which eventually becomes negative .
we therefore prepare lamellar phases decorated with copolymers @xcite .
bilayers of thickness @xmath72 comprising cetylpyridinium chloride ( cpcl ) and octanol ( oct ) ( @xmath73 w / w ) are diluted in brine ( @xmath74}= 0.2 \rm{m}$ ] ) and decorated with symperonics f68 ( serva ) [ @xmath75 , where eo is ethylene oxide and po is propylene oxide ] ) as amphiphilic copolymer .
the bilayer volume fraction @xmath58 and the copolymer / surfactant weight ratio @xmath62 range from @xmath76 to @xmath77 and @xmath78 to @xmath79 , respectively .
the bilayer volume fraction @xmath58 and the copolymer / surfactant weight ratio @xmath62 range from @xmath76 to @xmath77 and @xmath78 to @xmath79 , respectively .
the lamellar structure is preserved upon adding copolymer , with a continuous hardening into a `` lamellar gel '' .
the location of this gel in the ( @xmath62,@xmath58 ) phase diagram was studied by ligoure _
they observed gel - like behavior over a range of critical copolymer / surfactant weight ratios that depends on the membrane fraction . ) , where @xmath62 and @xmath58 are respectively the copolymer / surfactant weight ratio and bilayer volume fraction .
experimental phase boundaries are inferred at ( @xmath80 ) .
the dotted line is a guide for the eye .
( b ) small- and large - angle static light scattering intensity for @xmath63 and @xmath66 . a crossover from power law to constant intensity
is noted at a length scale of order @xmath81 . ]
the marked variation of the mechanical properties of the bulk lamellar phase is accompanied by modifications of the optical textures .
the pictures shown in fig .
[ fig : pictures ] demonstrate that the lamellar gels are onion phases .
onions are clearly obtained for both neat lamellar phases and samples put in contact with solvent . in the latter case
individual onions detach from the solvent / gel interface ( fig .
[ fig : pictures]e , f ) , while in the former case defect textures characteristic of the onion phase are observed between crossed polarizers ( fig .
[ fig : pictures]a - c ) .
the onions are highly polydisperse , with an apparent maximum size which seems to decrease with increasing @xmath62
. different textures and the presence of oily streaks are discerned at higher @xmath58 ( fig .
[ fig : pictures]d ) .
( top ) and maximum radius @xmath82 ( bottom ) determined from @xmath83 , for samples with ( a ) @xmath63 and various @xmath62 ; ( b ) @xmath66 and various @xmath58 .
solid lines are fits to @xmath84 and @xmath85 , with @xmath86 .
( inset , a ) same data as in the main figure in a semi - logarithmic plot.,title="fig : " ] while microscopy indicates that the size distribution is very broad and varies with @xmath62 and @xmath58 , a quantitative determination of the size distribution using this technique is delicate .
however , assuming an apollonian distribution we can , for example , estimate the macroscopic elastic modulus as a mass ( or volume ) average of the elastic moduli of onions of different sizes @xmath87 .
the elastic energy stored in a linear deformation of strain @xmath88 of an onion of radius @xmath21 is @xmath89 , where the effective surface tension @xmath90 @xcite , leading to a modulus @xmath91 .
thus , an estimate of the volume - averaged elastic modulus is @xmath92 @xcite .
averaging for @xmath93 , we find @xmath94 for @xmath95 . the linear elastic plateau is measured in a couette rheometer along two lines in the phase diagram at constant @xmath62 and @xmath58 ( fig .
[ fig : scatter]a ) . for fixed @xmath58 ,
the modulus @xmath83 increases monotonically with increasing @xmath62 , while for fixed @xmath62 , the modulus varies non - monotonically with @xmath58 ( fig .
[ fig : moduli ] ) . to extract the maximum onion radius @xmath82 from @xmath83 we must estimate the surface tension @xmath96 . for interlamellar forces dominated by helfrich entropic undulations , @xmath97 , with @xmath6 the bilayer thickness @xcite .
interestingly , it has been shown @xcite that for a decorated lamellar phase the inter - bilayer interactions are still of an effective helfrich form , with an effective bilayer thickness @xmath98 where @xmath99 is the apparent thickness of the polymer layer . in the experimental range of @xmath62
we take @xmath100 with @xmath101 , although this expression is strictly valid only in the brush regime ( @xmath102 ) .
one can thus calculate the surface tension by taking @xmath103 and hence extract , from the experimental values of @xmath104 , the variations of @xmath82 with the two experimental parameters @xmath58 and @xmath62 . as can be seen in fig .
[ fig : moduli ] , for @xmath63 , @xmath82 varies between @xmath105 and @xmath106 , with an apparent divergence near @xmath107 ; while for @xmath66 , @xmath82 varies between @xmath108 and @xmath109 , with an apparent divergence near @xmath110 .
, @xmath82 is a decreasing function of @xmath58 , which may be be due to the nearby lamellar - to - vesicle phase boundary @xcite . ]
the critical - like behavior of the onion maximum size inferred experimentally is remarkably similar , qualitatively , to the prediction of fig .
[ fig : app1 ] . from the
observed critical - like variations in @xmath82 we have identified two points on the boundary between lamellar and onion phases , in the @xmath111 plane ( fig .
[ fig : scatter]a ) . using the relation between @xmath71 and the parameter @xmath53 we can estimate experimental values for @xmath71 .
the onion / lamellae phase boundary is given by @xmath112 with @xmath113 .
for the two copolymer / surfactant weight ratios @xmath62 for which the boundary has been determined ( fig .
[ fig : app1]c ) , we obtain @xmath114 for @xmath115 and @xmath116 for @xmath66 . in agreement with theoretical expectations @xcite , @xmath71 is found to be negative , of order @xmath117 , and to decrease with increasing weight ratio .
consistent with the theoretical phase diagram ( fig .
[ fig : app1]b ) and with previous experimental observations @xcite , we also find that the critical value of @xmath71 , below which the onion phase is stable , decreases when @xmath58 increases .
the radii @xmath82 estimated from the elastic moduli are similar to those observed by microscopy ( fig .
[ fig : pictures]e , f ) . for @xmath68
the largest observable onion is @xmath118 and the calculated @xmath119 ( fig .
[ fig : pictures]e ) , and for @xmath69 , the largest observable onion is @xmath120 ( in fig .
[ fig : pictures]f ) and the calculated @xmath121 .
we can also compare @xmath82 to static light scattering experiments ( fig .
[ fig : scatter ] ) .
the scattering intensity @xmath122 scales as a power law for large @xmath123 , @xmath124 , and has a plateau for smaller @xmath123 , indicating that there is no structure at lengths larger than the cross - over length @xmath125 .
this provides an estimate for the maximum onion size @xmath82 that is very close to the value obtained from @xmath104 ( @xmath126 ) .
while similar power law behavior for high @xmath123 has previously been observed for monodisperse onions obtained by shear , the behavior at large @xmath123 @xcite was very different , with a peak characteristic of the onions size observed in that case .
the flat intensity obtained here clearly indicates the intrinsic polydispersity of the onions .
figures [ fig : pictures]e , f also show wide size distributions . in summary , we have presented experimental evidence for an equilibrium phase of polydisperse multi - lamellar vesicles ( onions ) , and rationalized this by combining a non - linear helfrich elastic theory with an assumed polydisperse distribution of onions .
future work should incorporate the entropy and calculate the nature of the space - filling distribution .
we thank the kavli institute for theoretical physics at the university of santa barbara , where part of this work was initiated .
lr thanks c. ligoure for fruitful discussions and l. cipelletti for help in the light scattering experiments , and pdo thanks crpp for hospitality . | we demonstrate the possibility of a stable equilibrium multi - lamellar ( `` onion '' ) phase in pure lamellar systems ( no excess solvent ) due to a sufficiently negative gaussian curvature modulus .
the onion phase is stabilized by non - linear elastic moduli coupled to a polydisperse size distribution ( apollonian packing ) to allow space - filling without appreciable elastic distortion .
this model is compared to experiments on copolymer - decorated lamellar surfactant systems , with reasonable qualitative agreement . |
fission induced by nucleons at intermediate energies , _ i. e. _ from pion production threshold ( @xmath2 150 mev ) to a few gev , is important from both basic and applied viewpoints .
even if fission is explained as a decay process of residual nuclei formed at the end of the fast nucleon cascade in the spallation reaction , many details are not yet clarified and deserve further experimental and theoretical work .
important applications of intermediate energy fission are energy production with accelerator driven systems @xcite , radioactive waste transmutation@xcite and radiation shield design for accelerators : these applications require proton and neutron fission cross sections to be determined with high accuracy in a wide energy range . many experimental data have accumulated in the last sixty years : ref.@xcite gives a detailed review of the @xmath0 and @xmath1 measurements up to the beginning of the present century and ref.@xcite proposes a parametrization of @xmath0 cross sections based on the same experimental systematics . among the recent @xmath0 experiments a prominent role is played by kotov et al.@xcite , who give the cross sections tabulated in steps of 100 mev in the range from 200 mev to 1 gev for several actinides very important for applications , @xmath3 , @xmath4 , @xmath5 and @xmath6 , and for two pre - actinides , @xmath7 and @xmath8 , forming the eutectic system acting as a spallation target and as a coolant in an accelerator driven system . a similar role for @xmath1 experiments
is currently being played by the n_tof facility@xcite at cern , which can measure fission cross sections from thermal energies up to about 1 gev , with a pulsed neutron beam produced by 20 gev / c protons from the ps accelerator impinging on a lead spallation target . since measuring an absolute cross section
requires simultaneous determination of fission events and neutron flux , which is a very difficult task , the measurements performed up to the present time are relative to @xmath9 and absolute cross sections have been obtained by normalizing the experimental ratios to an evaluated @xmath9 cross section , commonly taken from the endf / b - vii.1 library@xcite up to @xmath10 = 30 mev and from the jendl / he-2007 library@xcite from 30 mev to 1 gev .
@xmath1 cross sections up to 1 gev have already been published for @xmath11 and @xmath5@xcite , as well as for @xmath7 and @xmath8@xcite .
preliminary data have been obtained for @xmath3 and @xmath12 .
it is experimentally known@xcite that at the lower extremum of the intermediate energy range ( @xmath2 100 - 150 mev ) the @xmath1 cross section is systematically lower than the @xmath0 cross section for a given target nucleus , the effect being larger for pre - actinides than for actinides , but the difference tends to decrease with increasing incident energy , so that at 1 gev the behaviour of protons and neutrons is expected to be quite similar and the corresponding fission cross sections of the same order .
main purpose of the present work is to check whether available @xmath0 and @xmath1 data for a given target nucleus can be reproduced with satisfactory accuracy using the same set of model parameters , or , at least , with very close values ; in the affirmative , where only @xmath0 data exist in the energy range of interest , it is reasonable to use them to predict @xmath1 data for the same target nucleus , or viceversa .
our work is similar in spirit to ref.@xcite , where use was made of the los alamos codes cem2k+gem2 ( cascade - exciton model plus generalized evaporation model ) and laqgsm+gem2 ( quark - gluon string model plus generalized evaporation model ) in order to reproduce @xmath0 cross sections for pre - actinides and actinides , from @xmath13 to @xmath6 , taken mainly from prokofiev s systematics@xcite , extending the calculations to a large energy range , from tens of mev to 5 gev . with cem2k+gem2 , good fits were obtained by adjusting only two parameters , the ratio , @xmath14 , of level density parameters in the fission and evaporation channels for fissioning nuclei with atomic number @xmath15 and a constant , @xmath16 , in a semi - empirical parametrization of the ratio , @xmath17 , of neutron and fission width for @xmath18 .
using the same parameters , @xmath1 , @xmath19 and @xmath20 cross sections were calculated for the same nuclei and reasonable agreement with experimental data was obtained in several cases . with laqgsm+gem2 , the set of parameters fitted from@xmath0 reactions made it possible to reproduce fission fragment production and spallation , fragmentation and evaporation products in heavy - ion induced reactions measured at gsi - darmstadt in inverse kinematics@xcite . finally , the latest versions of the above mentioned codes , introduced as event generators in los alamos transport codes , such as mcnp6 , where carefully tested not only in the calculations of fission - fragment yields and of particles emitted from them , as is usually done with transport codes , but also in the calculations of fission cross sections@xcite .
with the models used in the present work , fission induced by nucleons in the energy range of interest can be seen as a two - stage process : a fast cascade stage , initiated by the high energy projectile , and representable as a succession of two - body collisions , with emission of fast nucleons , light clusters , pions , etc , leaving an excited remnant , and the slow decay stage , where the remnant decays by evaporation , fission , or other mechanisms . in our system of codes , the intranuclear cascade is described by a recent c++ version of the lige intranuclear cascade model , incl++@xcite , the evaporation - fission model by a c++ version of gemini , gemini++@xcite@xcite , or a fortran version of abla07@xcite .
it is to be pointed out , however , that the description of intermediate energy fission as a two- stage process is not a general rule in commonly used models and codes : many of them treat fission as a three - stage process , with an intermediate pre - equilibrium stage between fast cascade and evaporation - fission .
such an intermediate stage was originally suggested at jinr dubna and resulted in the original version of the cascade - exciton model@xcite , already mentioned in the introduction , which finally evolved in the version@xcite currently used in the los alamos transport codes mcnpx , mcnp6 and mars15 .
the three - stage description is present in the event generators of many other well - known transport codes of general use .
coming back to the codes used in the present work , incl++@xcite is a time - like intranuclear cascade model . at the beginning of the cascade stage ,
the incident nucleon is located with its own impact parameter on the surface of a working sphere , centered on the target nucleus with a radius @xmath21 , where @xmath22 and @xmath23 are the radius and the diffuseness of the target nucleus , respectively .
particles move along straight - line trajectories between collisions in the working sphere and are divided into participants and spectators in the usual sense .
participants that leave the working sphere are considered as ejectiles . inside the working sphere , nucleons feel a potential that depends on energy and isospin .
the depth of the potential well decreases linearly with increasing energy , from ordinary values at the fermi level to zero at about 200 mev .
the isospin dependence is such that neutron and proton fermi levels have the same energy .
collisions are , of course , governed by pauli blocking , treated in a different way in the first and in the subsequent collisions .
the nucleons involved in the first collision are subject to a strict blocking : after the collision , both of them should lie outside the fermi sphere . in subsequent collisions ,
the blocking is applied stochastically , with a probability given by the product of final - state blocking factors .
a careful definition of the latter allows one to account for surface effects and for the depletion of the fermi sphere during the evolution of the cascade .
an important novelty of recent versions of the code is the introduction of a coalescence model based on phase space , which permits the emission of light clusters , with mass @xmath24 , during the cascade stage , in keeping with experimental evidence .
pions are produced in inelastic nucleon - nucleon collisions through the excitation and subsequent decay of @xmath25 resonances , which sets an upper limit of the order of 3 gev to the incident nucleon energy for the mechanism of pion production to be valid .
the lower energy limit is given by the requirement that the de broglie wave length of relative motion be much smaller than the range of nuclear forces , which in turn is smaller than the average distance of neighbouring nucleons and is commonly set to 200 mev , although the model performs reasonably well even at lower energies , as shown in fission calculations of the following section . an important characteristic of the model is the self - consistent determination of the stopping time of the cascade , which can be simply parametrized as @xmath26 fm / c , with @xmath27 the mass of the target nucleus . at @xmath28
many physical quantities , such as the excitation energy of the target nucleus and the average kinetic energy of the ejectiles , switch from a fast time evolution , dominated by intranuclear cascade , to a much slower evolution , which is taken as a signature of equilibration . thanks to this choice of the stopping time , it is not necessary to introduce a pre - equilibrium model describing the intermediate stage between the fast cascade and the evaporation - fission decay .
the effect of an explicit pre - equilibrium stage in the incl model on nucleon - induced reactions above 200 mev was tested in ref.@xcite and found not really necessary .
the effect might be more significant in the energy range from 100 to 200 mev , with particular reference to energy spectra and angular distributions of emitted particles , rather than the fission cross sections considered in the present work .
however , no clear evidence in this sense emerged from the latest iaea benchmark of spallation models@xcite , whose conclusions are presented in ref.@xcite .
gemini++@xcite@xcite is a statistical - model code which follows the decay of a compound nucleus by a series of sequential binary decays until such decays are forbidden by energy conservation or become improbable because of gamma - ray competition . differently from most statistical - model codes , light - particle evaporation is described by the hauser - feshbach formalism , which strictly conserves angular momentum , at the price of higher computational time with respect to the more common weisskopf - ewing formalism .
an important ingredient of the decay width is the nuclear level density as a function of excitation energy @xmath29 and angular momentum @xmath30 , described by a bethe - type formula @xmath31 \quad . \label{bethe}\ ] ] here , @xmath32 is the excitation energy and the thermal excitation energy @xmath29 is related to @xmath32 by the formula @xmath33 where @xmath34 is the yrast energy for angular momentum @xmath30 , @xmath35 and @xmath36 are the pairing and shell correction , respectively , which depend on excitation energy .
the level density parameter @xmath37 includes the damping of shell correction @xmath36 inspired by ignatyuk@xcite @xmath38 \quad , \label{a_ig}\ ] ] with @xmath39 , @xmath40 = 19 mev and @xmath41 = 50 .
the effective level density parameter @xmath42 is taken of the form@xcite @xmath43 with @xmath44 = 7.3 mev , @xmath45 = 12 mev and @xmath46 , with @xmath47 the mass number . the decay width for symmetric fission , dominant at high excitation energy ,
is given by the well - known bohr - wheeler formula , while the decay width for asymmetric fission is derived from moretto s formalism@xcite @xcite , based on the concept of conditional fission barrier , _ i. e. _ a saddle point configuration with fixed asymmetry of mass and charge of the prefragments .
liquid drop barriers are calculated by means of sierk s finite - range model@xcite , with shell and pairing corrections taken from ref.@xcite .
an important adjustable parameter in the code is the ratio of the effective level density parameter at the saddle point , @xmath48 , to the same quantity at ground - state deformation , @xmath49 , with a default value of 1.036@xcite .
fission transients are not considered , although it has been shown@xcite that the model can accommodate a short fission delay .
abla07@xcite is a statistical code describing the de - excitation of a nucleus in thermal equilibrium by particle evaporation , fission , or , above a prescribed excitation energy per nucleon , multifragmentation .
particle evaporation is treated in an extended weisskopf - ewing formalism , where a distribution of orbital angular momenta in the emission of nucleons and clusters is evaluated in semiclassical approximation , based on phase space arguments .
the model also allows generalized evaporation of excited clusters , which plays a role analogous to asymmetric fission in gemini++ .
an essential ingredient of the decay formalism is the nuclear level density as a function of excitation energy and angular momentum , described by a constant temperature formula at low energy and by a bethe - type formula ( [ bethe ] ) at high energy .
the asymptotic level density parameter , @xmath50 ( see formula ( [ a_ig ] ) ) , is energy - independent and given , in mev@xmath51 , by the original prescription of ref.@xcite @xmath52 where @xmath53 is the ratio of the surface of the deformed nucleus to that of a spherical nucleus with the same mass number @xmath47 , as in the finite - range liquid drop model@xcite .
the level density contains a collective enhancement factor , due to nuclear rotations and vibrations , depending on excitation energy@xcite .
the approach to fission contains elements of dynamics : the time evolution of the fission degree of freedom is treated as a diffusion process , determined by the interaction of collective degrees of freedom with a heat bath formed by the individual nucleons and described by a fokker - planck equation whose solution leads to a time - dependent fission width , @xmath54 .
an analytical approximation to such a solution and , consequently , to the time dependence of the fission width makes the problem computationally tractable . at low excitation energy ,
the code uses the standard model of a two - humped fission barrier , whose penetrability is computed in the approximation of full damping of the vibrational resonances in the intermediate well .
like in the gemini++ code , liquid drop barriers are computed in the frame of the finite - range model@xcite and shell and pairing corrections are taken from ref . @xcite .
as already stressed in the introduction , our main purpose is a simultaneous reproduction of @xmath0 and @xmath1 cross sections for the same target nuclei , possibly using the same , or , at least , rather close values of model parameters .
in fact , it is expected that the isospin dependence of the fission cross section is already contained in the models we use . in order to better understand how fission keeps memory of projectile isospin ,
it is worth recalling that the fission cross section increases with increasing fissility parameter , @xmath55 , where @xmath56 and @xmath47 are charge and mass of the fissioning nucleus . in the case of a proton induced reaction ,
the highest fissility parameter is obtained when the incident proton is captured by the target nucleus and is proportional to @xmath57 , with @xmath58 and @xmath27 the charge and mass of the target nucleus , while in the case of capture of an incident neutron the fissility parameter is smaller , since it is proportional to @xmath59 .
an analysis of charge and mass distributions of remnants undergoing fission at the end of the fast cascade stage , carried out in ref.@xcite with the cascade - exciton model code cem95@xcite for @xmath0 and @xmath1 reactions up to a projectile energy @xmath60 = 200 mev showed the relevance of remnants with charge @xmath61 in enhancing the @xmath0 cross section with respect to the @xmath1 cross section , where remnants can not have charges larger than @xmath58 . by considering the fissility parameter dependence of the liquid drop barriers , the authors of ref.@xcite explained also why the difference of @xmath0 and @xmath1 cross sections is much larger in pre - actinides than in actinides .
it is worth recalling that a cem95 analysis and interpretation of the differences between @xmath0 and @xmath1 cross sections for @xmath62 and @xmath8 in the energy range from 45 to 500 mev had already been published in ref.@xcite .
the analysis of differences of @xmath0 and @xmath1 reactions was further enriched with new experimental data in the pre - actinide region by the authors of ref.@xcite , who derived also an approximate analytical dependence of the @xmath63 ratio on @xmath64 at given projectile energy , @xmath65 .
for instance , at @xmath66 180 mev , they obtained @xmath67\quad , \label{sigma_ratio}\ ] ] valid for both actinides and pre - actinides . with increasing projectile energy up to the gev region
, it is expected that the differences in the behaviour of neutrons and protons become smaller , so that @xmath0 and @xmath1 cross sections tend to a common value . in keeping with the philosophy of the authors of the incl++ code@xcite@xcite ,
no parameters have been modified in the intranuclear cascade model , already optimized by reproducing a large amount of data related to the cascade stage , such as total reaction cross sections , double - differential spectra of emitted nucleons , pions and light clusters , while two basic fission model parameters have been taken as adjustable either in gemini++ or abla07 : the height of the liquid - drop fission barrier , @xmath68 , and the asymptotic level density parameter at the saddle point , @xmath48 , in abla07 , or the ratio of the level density parameter at the saddle point to that at the ground state deformation , @xmath69 , in gemini++ .
both parameters affect significantly the calculation of fission cross sections in the whole energy range of interest to the present work , from 100 mev to 1 gev , although the change of the barrier height is more important at low incident energy and the change of the level density parameter at high energy . in principle
, many sets of experimental data could be reproduced by modifying only one parameter , _ e.g. _
@xmath48 , at the the cost of using rather different values of it in the neutron and proton channels , particularly for actinides , as we did in our preliminary work@xcite . in the following sub - section results for absolute @xmath0 and @xmath1
cross sections are presented and discussed in comparison with experimental data , while a separate sub - section will be dedicated to @xmath1 cross sections relative to @xmath9 , with particular reference to the comparison with n_tof data .
the rationale behind our work is to adjust fission model parameters on the @xmath0 cross sections of ref.@xcite for @xmath7 , @xmath8 , @xmath3 , @xmath4 and @xmath6 and use similar parameter values in computing @xmath1 cross sections for the same targets . as alredy pointed out in our preliminary work@xcite
, the gemini++ and abla07 models can produce fits of comparable accuracy for actinides , while in the lead - bismuth region abla07 appears to work better than gemini++ .
therefore , we have decided to show the results of gemini++ calculations for actinides and those of abla07 calculations for lead and bismuth .
the experiment of ref.@xcite , labelled as kotov 2006 in the figures , yields 9 cross section values in the energy range from 207 mev to 1 gev .
we have evaluated the quality of our fits by means of a @xmath70 test , where the @xmath70 variable is defined as @xmath71 where the symbols are self - explanatory .
since we try to reproduce 9 experimental values , @xmath72 , by adjusting 2 parameters , the relevant @xmath70 distribution , @xmath73 , should have @xmath74 = 9 - 2 degrees of freedom and the quality of fit is evaluated by means of the cumulative probability @xmath75 table [ tabi ] shows the values of the parameters adjusted on the experimental @xmath0 data of ref.@xcite for actinides in our incl++/gemini++ calculations , _
i. e. _ the ratio of the level density parameter at the saddle point to the one at ground - state deformation , @xmath69 , and the global correction of the liquid - drop fission barriers , @xmath76 , in mev .
.model parameters and @xmath70 tests for proton - induced fission of actinides .
[ cols="^,^,^,^,^,^",options="header " , ] the energy dependence of fission cross sections of lead and bismuth is quite different from that in the actinide region : at low energy , let us say 100 mev , the @xmath1 cross section is 2 to 3 times smaller than the @xmath0 cross section and neither of them reaches a plateau at high energy , 1 gev .
therefore , knowledge of the @xmath0 cross section in this energy range yields less useful information on the @xmath1 cross section with respect to actinides .
an experiment performed at the institute of ref.@xcite produced the @xmath77 cross section at the same energies and the results , published in ref.@xcite ( vaishnene 2010 ) , are shown in the left panel of fig.[label7 ] , together with our calculations , which reproduce them satisfactorily with the default parameters of the abla07 model , _ i. e. _
@xmath78 = 1 and @xmath76 = 0 . fission cross sections .
experimental data are discussed in the text.,width=377 ] it is to be pointed out that @xmath0 data below 200 mev are consistent with ref.@xcite , with the exception of ref.@xcite ( flerov 1972 ) , whose data are somewhat lower than our calculation . at high energy , on the contrary , the recent result of ref.@xcite ( schmidt 2013 ) at 500 mev as well as the value of ref.@xcite ( enqvist 2001 ) at 1 gev are higher .
both experiments@xcite@xcite were performed in inverse kinematics .
new data in this energy interval are needed in order to clarify the situation .
@xmath1 data are available up to 200 mev from refs.@xcite ( smirnov 2004 ) and @xcite ( rhyzov 2006 ) , which are in mutual agreement .
they are reproduced in the right panel of fig.[label7 ] with the default value of @xmath48 , at the cost of increasing the fission barriers by a quantity @xmath76 = 0.3 mev .
fission of natural lead is very important for applications to accelerator - driven systems , where the eutectic system acting as a spallation target and as a coolant consists of @xmath7 and @xmath8 . as usual
, kotov@xcite provides @xmath0 data from 200 mev to 1 gev . an older measurement ( flerov 1972)@xcite below 200 mev appears to be compatible with kotov s data@xcite , while other old experiments@xcite,@xcite,@xcite ( koshin 1965 , brandt 1972 and remy 1971 , respectively ) at higher energies seem to suggest a systematically higher cross section .
as shown by the corresponding @xmath70 test in table [ tabii ] , our fit to the @xmath0 data of kotov@xcite , obtained by reducing the default @xmath48 parameter by a factor @xmath78 = 0.995 and by reducing the fission barrier by a quantity @xmath76 = -0.16 mev is not very satisfactory ; in particular , our calculations overestimate the low energy points , as shown in the left panel of fig.[label8 ] .
fission cross sections .
experimental data are discussed in the text.,width=415 ] the same difficulty appears also in @xmath1 calculations , carried out with @xmath78 = 1.005 and @xmath76 = 0 , which overestimate the experimental data below 200 mev , where there is good agreement of experiments ( prokofiev 1996)@xcite , ( shcherbakov 2001)@xcite , ( smirnov 2004)@xcite , ( nolte 2007)@xcite , as shown in the right panel of fig .
[ label8 ] .
the relative @xmath1 data measured at n_tof up to 1 gev@xcite will be discussed in the next sub - section . as far as the @xmath0 cross section is concerned , several measurements exist for @xmath8 and the majority of them ( steiner 1956)@xcite , ( de carvalho 1962)@xcite , ( konshin 1965)@xcite , ( brandt 1972)@xcite , ( bychenkov 1973)@xcite , ( debeauvais 1976)@xcite are in reasonable agreement with our reference experiment kotov 2006@xcite , as shown in the left panel of fig.[label9 ] ; the main exceptions are the point at 1 gev from ref.@xcite ( bochagov 1978 ) , which is much lower than the corresponding point from ref.@xcite , as well as ref.@xcite ( hudis 1976 ) , whose data are systematically higher than those of ref.@xcite . fission cross sections .
experimental data are discussed in the text.,width=415 ] the latter are well reproduced in our calculations by multiplying the default @xmath48 parameter by a factor @xmath78 = 1.01 and increasing the fission barriers by an amount @xmath76 = + 0.18 mev . among the @xmath1 experiments , a key role
is played by refs.@xcite,@xcite ( prokofiev 1996 , nolte 2007 ) , which are relative to @xmath79 scattering and can be considered as measurements of the absolute @xmath1 cross section , while refs@xcite ( shcherbakov 2001 ) , @xcite ( fomichev 2004 ) , @xcite ( laptev 2007 ) , @xcite ( tarro 2011 ) are relative to @xmath9 and refs.@xcite,@xcite ( smirnov 2004 and rhyzov 2006 , respectively ) , are relative to @xmath80
. only refs.@xcite,@xcite extend above 200 mev . using in @xmath1 calculations the same parameters adopted in fitting @xmath0 data
would underestimate the data of ref.@xcite from 200 to 400 mev , while reproducing the data of ref.@xcite relative to @xmath81 reasonably well . as a compromise between the two sets of data , we have adopted in @xmath1 calculations
@xmath78 = 1.01 and @xmath76 = 0 .
more details will be given in the next sub - section .
this sub - section is dedicated to @xmath1 measurements relative to the @xmath9 reaction , taken as a standard , even if no reliable data of its absolute cross section exist above 200 mev , with particular reference to n_tof@xcite measurements up to 1 gev .
for those cases@xcite@xcite where the experimental ratios measured at n_tof have been normalized in the original publications to the @xmath9 cross section from the jendl / he-2007 library a comparison of absolute cross sections has been added to that of relative cross sections .
the parameters adopted in the calculations are those given for the @xmath1 reactions in the sub - section on absolute cross sections .
the cross section ratio to be compared with measurements is the ratio of theoretical cross sections obtained with our monte carlo simulations , which make it possible to associate to any @xmath82 an uncertainty @xmath83 attributable to counting statistics .
if @xmath84 is the ratio in question , the relative error @xmath85 is obtained by the quadratic law of propagation of relative errors of two uncorrelated variables @xmath86 and @xmath87 @xmath88^{1/2 } \quad . \label{ratio}\ ] ] figure [ label10 ] compares calculated and experimental ratios for the @xmath89 reaction .
to @xmath9 fission cross section .
experimental data are discussed in the text.,width=415 ] our results agree with ref.@xcite , which still represents one of the most reliable measurements of @xmath1 cross sections in the intermediate energy region , and are somewhat higher than ref.@xcite .
preliminary n_tof results up to 1 gev , not reported in fig .
[ label10 ] , compare reasonably well with our calculations .
no @xmath0 data are available for @xmath11 , whose @xmath1 cross section relative to @xmath81 has been measured up to 400 mev in ref.@xcite ( lisowski 1991 b ) and at the n_tof facility up to 1 gev using parallel plate avalanche counters in ref.@xcite ( paradela 2010 ) and up to 200 mev using fission ionization chambers in ref.@xcite ; the two n_tof experiments are in good mutual agreement , therefore we take into account only the more extended measurement@xcite for comparison with our calculations , done with the same fission parameters as @xmath81 , namely @xmath69 = 1.050 and @xmath76 - 0.2 mev .
the left panel of fig.[label11 ] compares theoretical and experimental ratios , in overall good agreement , while the right panel compares our absolute cross section with the one obtained in ref.@xcite by normalizing the experimental ratio to the jendl / he-2007 evaluation of the @xmath9 cross section .
cross section of @xmath11 relative to @xmath81 .
right panel : absolute @xmath1 cross section of @xmath11 .
experimental data are discussed in the text.,width=377 ] as one can see from the right panel of fig .
[ label1 ] the disagreement increases with increasing neutron energy , because our calculated @xmath9 cross section shows a large plateau at high energy , while the jendl evaluation steadily decreases in the same energy range . a somewhat larger value of @xmath69 is necessary for reproduction of the fission measurements of @xmath90 .
fig.[label12 ] shows a comparison of available data relative to @xmath81 from refs.@xcite,@xcite with the theoretical cross section ratio obtained with @xmath69 = 1.10 and @xmath76 = 0 .
a positive value of @xmath76 would further bring the relative cross section to better agreement with experimental data .
new data up to 1 gev are expected in the near future from the n_tof experiment .
cross section of @xmath90 relative to @xmath81 .
experimental data are discussed in the text.,width=453 ] as already discussed in the preceding sub - section , the n_tof results for the @xmath1 cross section of @xmath3 are still preliminary and a more reliable @xmath9 cross section in the intermediate energy range is recommended for their normalization .
the difficulty with the present data@xcite clearly appears in the comparison of our calculated cross - section ratio , which reproduces the @xmath1 values of fig [ label4 ] , with the presently available experimental ratios@xcite@xcite , which are in reasonable mutual agreement and significantly lower than our results , as shown in fig.[label13 ] .
cross section of @xmath3 relative to @xmath81 .
experimental data are discussed in the text.,width=453 ] the left panel of fig.[label14 ] compares the ratios of the @xmath91 cross section to the @xmath9 cross section measured at n_tof@xcite with our calculated results , which turn out to be in overall agreement with the experiment . on the contrary , the right panel shows a significant disagreement of the corresponding absolute cross sections in the high energy region , owing to the normalization to the jendl / he-2007 evaluation of @xmath9 adopted in ref.@xcite .
cross section of @xmath5 relative to @xmath81 .
right panel : absolute @xmath1 cross section of @xmath5 .
experimental data are discussed in the text.,width=340 ] the relative @xmath1 cross section of @xmath6 , shown in fig .
[ label15 ] , compares reasonably well with the data of ref.@xcite below 150 mev and with those of ref.@xcite between 150 and 200 mev .
cross section of @xmath6 relative to @xmath81 .
experimental data are discussed in the text.,width=340 ] the two series of data partially overlap , while those of ref.@xcite are systematically higher than them and than our calculations as well .
the @xmath1 cross section of @xmath7 relative to @xmath81 has been measured at n_tof up to 1 gev@xcite .
the left panel of fig.[label16 ] shows a comparison of experimental and theoretical cross section ratios , computed by increasing the default @xmath92 parameter by a factor of 1.005 , without any change of fission barriers . as already noticed for the absolute @xmath1 cross section discussed in the preceding sub - section , the calculations overestimate the experimental data at low energies . for the sake of completeness , the right panel of fig .
[ label16 ] compares the calculated absolute cross section with the one obtained by normalizing the relative cross section measured at n_tof to the evaluated @xmath9 cross section of the jendl / he-2007 library .
the disagreement at high energies is mainly due to the difference of the two computed @xmath81 cross sections .
cross section of @xmath7 relative to @xmath81 .
right panel : absolute @xmath1 cross section of @xmath7 .
experimental data are discussed in the text.,width=377 ] finally , fig .
[ label17 ] shows the results of our @xmath1 calculations for @xmath8 , which slightly overestimate the data of ref.@xcite relative to @xmath81 below 200 mev , as shown in the left panel of the figure .
cross section of @xmath8 relative to @xmath81 .
right panel : absolute @xmath1 cross section of @xmath8 .
experimental data are discussed in the text.,width=453 ] this is the result of a compromise that takes into account the absolute data of ref.@xcite up to 400 mev , shown in the right panel of the figure together with the relative data of ref.@xcite normalized to the jendl / he-2007 cross section of the @xmath9 reaction .
also the measurement of ref.@xcite is relative to @xmath81 , but we do not know the values adopted for normalization , so that we could consider only the absolute cross section .
the disagreement with ref.@xcite above 500 mev is mainly due to the jendl / he-2007 cross section adopted in that reference for normalization . summing up ,
@xmath0 and @xmath1 cross sections have been reproduced with the same model parameters in the cases of @xmath93 , @xmath5 , @xmath6 and @xmath62 ; the same level density parameters , but different corrections to the heights of fission barriers have been adopted for @xmath90 , @xmath3 and @xmath8 ; finally , slightly different level density parameters and barrier height corrections have been used for @xmath7 : this is a peculiar case , since the natural element consists of sizable abundances of @xmath62 , @xmath94 and @xmath95 . in principle , each isotope should have its own model parameters , but , for the sake of simplicity , we have chosen to treat @xmath7 as an effective nucleus .
in the present work we have reproduced @xmath0 cross sections in the intermediate energy range for a number of actinides and pre - actinides relevant to nuclear energy applications by a two - parameter fit : an increase of the level density parameter at the saddle point with increasing fissility parameter is observed in specific isotopic chains , such as uranium nuclei in table i and lead nuclei in table ii , but the present data are not sufficient to determine a global trend in either region . therefore , the results of the present work can not be used for prediction of fission cross sections of isotopes for which neither @xmath0 or @xmath1 data exist in the intermediate energy range .
the model parameters that reproduce @xmath0 cross sections have been used as a first guess in the calculation of @xmath1 cross sections , where the comparison with experiments is complicated by the fact that the large majority of data are relative to fission standards , such as @xmath93 in the actinide region and/or @xmath8 in the pre - actinide region , but even for the standards absolute cross section data are not available above 200 mev .
@xmath1 cross sections are systematically lower than @xmath0 cross sections around 100 mev , but the differences are expected to reduce with increasing projectile energy , particularly for actinides , whose @xmath0 data exhibit a large plateau from 500 mev to 1 gev .
this effect is clearly seen in our calculations .
as far as @xmath1 cross sections are concerned , the only relative measurements extending up to 1 gev , due to the n_tof collaboration , are satisactorily reproduced by our calculations .
the disagreement between corresponding absolute cross sections is almost entirely due to the normalization to the jendl / he-2007 evaluation of the @xmath9 reaction adopted in n_tof papers@xcite,@xcite . a deeper comparison with evaluated data libraries such as jendl / he-2007 would require an extension of the energy range of our calculations : as already shown in the case of @xmath96 , whose calculations up to 3 gev look promising , the codes used in this work allow us to extend cross section calculations up to 3 gev , but moving the lower extremum down to 20 mev
could only be done by resorting to fully quantum - mechanical models , since the very concept of intranuclear cascade model loses its applicability at low energies .
a recent example of consistent calculations of @xmath0 and @xmath1 calculations in the energy range from 20 mev to 1 gev for a number of actinides is provided by ref.@xcite , where use is made of the mcfx system of codes , permitting a good reproduction of many experimental data at the cost of adjusting the heights of the fission barriers of remnants obtained from microscopic calculations .
this has been done by us , too , at the very phenomenologic level permitted by the @xmath76 parameter of tables i - ii .
an extension of our fission calculations to the energy range from 20 mev to 3 gev is planned for the nuclei already suggested as fission standards , @xmath93 and @xmath8 , with the specific scope of support to the analysis of @xmath1 cross sections measured at the n_tof facility@xcite at cern .
we acknowledge valuable discussions with drs .
sylvie leray , aleksandra keli - heil , nicola colonna , carlos paradela and laurent tassan - got .
s. g. mashnik , k. k. gudima , r. e. prael and a. j. sierk , proceedings of the workshop on nuclear data for the transmutation of nuclear waste , lanl report la - ur-04 - 1873 , los alamos , 2004 ; arxiv : nucl - th/0404018 .
r. j. charity , proceedings of the joint ictp - iaea advanced workshop on model codes for spallation reactions , ictp trieste , italy , 4 - 8 february 2008 , d. filges et al .
, eds . , iaea report indc(nds)-530 , vienna , p. 139
( 2008 ) .
a. keli , m. v. ricciardi and k. h. schmidt , proceedings of the joint ictp - iaea advanced workshop on model codes for spallation reactions , ictp trieste , italy , 4 - 8 february 2008 , d. filges et al . , eds . , iaea report indc(nds)-530 , vienna , p. 181
( 2008 ) .
s. lo meo , d. mancusi , c. massimi , g. vannini and a. ventura , proceedings of the 20@xmath97 international school on nuclear physics , neutron physics and applications , varna , bulgaria , september 17 - 21 , 2013 ; j. phys .
: conf . ser .
* 533 * ( 2014 ) 012024 .
a. d. carlson , s. chiba , f. j. hambsch , n. olsson and a. n. smirnov , proceedings of the international conference on nuclear data for science and technology , trieste , italy , may 19 - 24 , 1997 , part ii , p. 1223 | monte carlo calculations of fission of actinides and pre - actinides induced by protons and neutrons in the energy range from 100 mev to 1 gev are carried out by means of a recent version of the lige intranuclear cascade model , incl++ , coupled with two different evaporation - fission codes , gemini++ and abla07 . in order to reproduce experimental fission cross sections ,
model parameters are usually adjusted on available @xmath0 cross sections and used to predict @xmath1 cross sections for the same isotopes . |
the understanding of plasma equilibria is one of the most important problems in magnetohydrodynamics and arises in several fields including solar physics and thermonuclear fusion .
such an equilibria is often governed by the well known steady hydromagnetic equations @xmath0 which describe the balance of the lorentz force by pressure . here
@xmath1 and @xmath2 are respectively the magnetic field and the pressure .
notice that system ( [ eqn : ms])+([eqn : div1 ] ) is quite similar to steady inviscid fluid equations @xmath3 this analogy is due to the vectorial identity @xmath4 system of equations ( [ eqn : ms])+([eqn : div1 ] ) must be completed with some boundary conditions on @xmath1 and @xmath2 .
physical considerations suggest to prescribe the boundary normal field component : @xmath5 where @xmath6 satisfies the compatibility condition @xmath7 due to the equation @xmath8 . defining the inflow boundary as @xmath9 one can also prescribe the normal component @xmath10 of the current density and the pressure @xmath2 on @xmath11 @xmath12
@xmath13 one can notice that if the pressure is neglected , equations ( [ eqn : ms])+([eqn : div1 ] ) become @xmath14 equation ( [ eqn : fff ] ) means that the magnetic field and its curl , which represents the current density , are everywhere aligned .
the magnetic field is said _ beltrami _ or _ force - free _ ( ff ) . a usual way to tackle the problem ( [ eqn : fff ] ) + ( [ eqn : div2 ] ) consists to rewrite equation ( [ eqn : fff ] ) into the form @xmath15 where @xmath16 is a scalar function which can be a constant function or can depend on @xmath17 . in the former
, the @xmath1 field is said linear ff . in the latter ,
it is said non linear . + some partial results concerning existence of 3d solutions of equations ( [ eqn : ms])+([eqn : div1 ] ) in bounded domains are given in @xcite and @xcite .
linear force - free - fields were studied in @xcite . for the existence of non - linear ones the reader can refer to @xcite , @xcite .
the numerical solving of equations ( [ eqn : ms])+([eqn : div1 ] ) and equations ( [ eqn : fff])+([eqn : div2 ] ) is of importance in magnetohydrodynamics studies and in solar physics . as it is known , the reconstruction of the coronal magnetic field has is of a great utility in observational and theoretical studies of the magnetic structures in the solar atmosphere . in this paper
, we propose an iterative process for solving these equations ( section [ sec:2 ] ) .
a finite element method is proposed for solving each one of the arising problems .
our objective here is to expose an iterative method for solving the non - linear equations ( [ eqn : ms])+([eqn : div1 ] ) in a bounded and simply - connected domain .
the starting idea of the method consists to split the current density @xmath18 into the sum @xmath19 where the vector field @xmath20 is collinear to @xmath1 , while @xmath21 is perpendicular to @xmath1 .
the problem is decomposed formally into a curl - div system on @xmath22 and two first order hyperbolic equations on @xmath23 and @xmath24 .
+ more precisely , writing @xmath25 where @xmath26 is a scalar function and taking the divergence of ( [ omega123 ] ) , gives @xmath27 notice that the pressure satisfies a similar equation since @xmath28 equation ( [ eqn : ms ] ) becomes @xmath29 which means that @xmath30 if @xmath31 . in consideration of these remarks
, we are going now to propose an iterative process to solve non - linear systems ( [ eqn : ms])+([eqn : div1 ] ) . in this process the transport equation ( [ eqn : mu ] ) is perturbed by adding an artificial reaction term .
namely , we construct a sequence @xmath32 as follows : * the starting guess @xmath33 is chosen as the irrotational field associated to @xmath6 defined by @xmath34 this is a usual problem which can be reduced to a scalar neumann problem since the domain is simply - connected .
* for all @xmath35 , @xmath36 is solution of the system @xmath37 where @xmath38 is a small parameter and @xmath39 . * for all @xmath35 , @xmath40 and @xmath41 , where @xmath42 satisfies @xmath43 here @xmath44 and @xmath45 is a small parameter .
* for all @xmath35 , @xmath46 , with @xmath47 solution of @xmath48 where @xmath49 while @xmath50 is solution of the laplace problem @xmath51 notice that the appearance of the correction term @xmath52 is due to the fact that @xmath53 is not zero in general .
+ the convergence of this iterative process is not an easy matter .
we conjecture that it converges if @xmath54 is sufficiently small and @xmath55 in @xmath56 for some constant @xmath57 .
nevertheless , in the case of linear force - free fields ( in that case the algorithm is simplified since at each iteration @xmath58 , @xmath59 and @xmath42 is a fixed real ) boulmezaoud and amari @xcite proved that this process is super - convergent .
the proof of convergence in the general case is not given and remains an open question .
+ notice that the same algorithm can be used for computing linear or non - linear force - free fields which are solutions of ( [ eqn : div2])+([eqn : fff2 ] ) , provided that the computation of the pressure @xmath36 and the vector field @xmath60 are dropped .
here we give a short description of the finite elements methods we use for solving problems arising in the iterative process exposed above .
observe first that at each iteration of the algorithm one should solve two problems : + ( a ) a reaction - convection problem of the form : _ find @xmath61 solution of _
@xmath62 ( b ) a vector potential problem : _ find the pair @xmath63 satisfying _ @xmath64 we begin with the approximation of ( [ eqn : hyperbolic ] ) . it is well known that the direct application of a galerkin finite elements method to the singularly perturbed problem ( [ eqn : hyperbolic ] ) may lead to the appearance of spurious oscillations and instabilities . in the two last decades
, several methods were proposed to remove this drawback ( especially in the two dimensional case ) . among these methods
, one can recall the _ streamline diffusion _ method ( see brookes and hughes @xcite , see also , e. g. , johnson et _ al . _
@xcite ) , the _ discontinuous _ galerkin method ( see lesaint @xcite ) and _ bubble functions _ methods ( see , e. g. , brezzi et _ al . _ @xcite ) . here
we shall use the method of streamline diffusion .
thus , let us consider a family of regular triangulations @xmath65 of @xmath56 .
the discrete problem we consider is @xmath66 where @xmath67 here @xmath68 stands for the finite elements space @xmath69 where for each @xmath70 , @xmath71 denotes the space of polynomials of degree less or equal @xmath72 .
+ one can prove that the problem @xmath73 has a unique solution @xmath74 when @xmath75 .
moreover , if @xmath76 for some constant @xmath77 and if @xmath78 and @xmath79 for some @xmath80 , then @xmath81 where @xmath82 + now , we deal with the approximation of the curl - div system ( [ eqn : systemb ] ) , which can be dispatched into two problems : a variational problem ( @xmath83 ) in terms of @xmath84 and the fictitious unknown @xmath85 , and laplace equation ( [ eqn : lap_cor ] ) in terms of @xmath86 .
we only deal with the approximation of @xmath84 , since we shall see that the computation of the @xmath86 is useless .
denote by @xmath87 the space @xmath88 equipped with its usual norm .
the statement of problem ( [ eqn : systemb ] ) suggests the use of an @xmath87 approximation .
define @xmath89 the space @xmath90 let @xmath91 , @xmath92 two finite - dimensional subspaces and set @xmath93 we make the following assumptions @xmath94 @xmath95 the discrete version of problem ( [ eqn : systemb ] ) writes @xmath96 according to amrouche and _
@xcite , the problem @xmath97 has one and only one solution @xmath98 with @xmath99 , and @xmath100 a simple manner for constructing the spaces @xmath101 and @xmath102 is to use the @xmath103 conforming elements of ndelec @xcite ( see amrouche and _ al . _ ) . in that case , the following estimate holds @xmath104 .
+ an important feature of the discrete system @xmath97 is that only the discrete vector field @xmath105 is really unknown . actually , we know that @xmath106
. this property can be exploited from a practical viewpoint to reduce the discrete system to a smaller one by eliminating @xmath107 . in term of matrices ,
the system writes @xmath108 where @xmath109 is a symmetric and positive square matrix ( @xmath109 is not definite neither invertible ) .
we can state the following [ matrix_lemma ] let @xmath110 be a square positive , definite and symmetric matrix having the same size as @xmath111 .
then , the pair @xmath112 is solution of ( [ eqn : syst ] ) if and only if @xmath113 and @xmath114 is solution of @xmath115 in lemma [ matrix_lemma ] , the matrix @xmath109 and the rhs @xmath116 are not arbitrary . indeed , if @xmath117 denotes the matrix of the operator @xmath118 , then necessarily @xmath119 and @xmath120 .
these identities are the discrete counterpart of the continuous relations @xmath121 and @xmath122 .
a serious advantage of the new system ( [ eqn : dec_syst ] ) comparing with ( [ eqn : syst ] ) is that number of unknowns is reduced .
in this last section , we expose some computational results we obtain with a 3d code .
this code use the iterative method and the finite elements discretization exposed above to solve problem ( [ eqn : ms])+([eqn : div1 ] ) and problem ( [ eqn : fff2])+([eqn : div2 ] ) .
we compare the exact solution and the numerical solution and we show the behavior of the errors in terms of @xmath54 .
two exact solutions are used for the tests . * * test 1 ( a non - linear force - free - field ) . * + let @xmath123 the cylindrical coordinates with respect to a point @xmath124 ( @xmath125 and @xmath126 ) .
the pair @xmath127 is given @xmath128 this is a non - linear force - free field with @xmath129 table [ tab:1 ] shows the behavior of the residue @xmath130 and the product @xmath131 versus the iteration number .
this example illustrates the superconvergence of the algorithm .
+ lll @xmath132 & @xmath133 & @xmath134 + 0 & 0.09912 & 6.740e-15 + 1 & 0.00566 & 0.06781 + 2 & 0.00036 & 0.01939 + 3 & 2.644e-05 & 0.01910 + * * test 2 : ( bennet pinch ) * .
the pair @xmath127 is given by @xmath135 in table [ tab:2 ] , the relative @xmath136 errors on @xmath1 and @xmath2 after convergence of the algorithm are shown .
these error decreases as @xmath137 , which confirms the high accuracy of the method . + lll @xmath54 & @xmath138 & @xmath139 + 0.69282 & 0.03837 & 0.08648 + 0.23094 & 0.00492 & 0.01102 + 0.13856 & 0.00191 & 0.00396 + 0.09897 & 0.00108 & 0.00201 + + alber , h.d . :
existence of three - dimensional , steady , inviscid , incompressible flows with non - vanishing vorticity .
ann . * 292 * , 493528 ( 1992 ) amrouche , c. , bernardi , c . , dauge , m. , girault , v. : vector potentials in three - dimensional non - smooth domains , math .
methods appl .
* 9 * , 823864 ( 1998 ) .
boulmezaoud , t.z . ,
amari , t. : on the existence of non - linear force - free fields in three - dimensional multiply - connected domains .
paratre dans zeitschrift fr angewandte mathematik und physik ( zamp ) .
brooks , a. , hughes , t .
r. j. : streamline upwind / petrov - galerkin formulations for convection dominated flows with particular emphasis on the incompressible navier - stokes equations .
fenomech 81 , part i ( stuttgart , 1981 ) .
methods appl .
engrg . * 32 * , 199259 ( 1982 ) . | we propose an iterative finite element method for solving non - linear hydromagnetic and steady euler s equations .
some three - dimensional computational tests are given to confirm the convergence and the high efficiency of the method . |
low - dimensional layered organic conductors are a promising class of materials that exhibit high temperature superconductivity . among them , an antiferromagnetic dimer - mott ( dm ) insulator @xmath2-(bedt - ttf)@xmath1icl@xmath1 , which shows superconductivity at 14.2 k under high pressure ( @xmath38 gpa ) @xcite , has aroused great interests because its transition temperature is highest among organic superconductors known to date . at ambient pressure , @xmath2-(bedt - ttf)@xmath1icl@xmath1 is a dm insulator with an antiferromagnetic transition temperature @xmath4 of 22 k @xcite .
as typically seen in bandwidth - controlled mott transition systems such as @xmath5-(bedt - ttf)@xmath1@xmath6 , where @xmath6 = cu@xmath1(cn)@xmath7 , cu[n(cn)@xmath1](br , cl ) @xcite , @xmath2-(bedt - ttf)@xmath1icl@xmath1 also exhibits superconductivity when the antiferromagnetic phase is suppressed by applying pressure .
it has been , therefore , believed that antiferromagnetic spin fluctuation plays an essential role in the high-@xmath9 superconductivity @xcite .
the crystal structure of @xmath2-(bedt - ttf)@xmath1icl@xmath1 consists of an alternating stack of the conductive bedt - ttf layers and insulating icl@xmath1 layers along the @xmath10 axis as shown in fig.1(a ) @xcite .
although the charge transfer between these layers leads to a 3/4-filled band system , due to strong dimerization of two bedt - ttf molecules , the 3/4-filled band splits into bonding and antibonding bands , resulting in an effective half - filled band system [ see figs.1(b ) and 1(c ) ] .
the effective model to describe the electronic state is therefore considered to be the half - filled hubbard model , in which each dimer is regarded as a unit , and charge degrees of freedom inside of dimers are disregarded . according to band structure calculations @xcite
, @xmath2-(bedt - ttf)@xmath1icl@xmath1 has a quasi
one - dimensional ( q1d ) electronic structure with a pair of slightly warped open fermi surfaces owing to the cancelation of @xmath11 and @xmath12 [ see fig.1(b ) ] . this material , however , becomes a mott insulator due to the strong on - dimer coulomb repulsion @xmath13 .
thus , @xmath2-(bedt - ttf)@xmath1icl@xmath1 has been considered to be a typical q1d dm insulator with one hole per each dimerized molecules at ambient pressure . however , unlike @xmath5-(bedt - ttf)@xmath1@xmath6 , the phase diagram of @xmath2-(bedt - ttf)@xmath1icl@xmath1 under pressure simply can not be accounted for within the framework of the bandwidth - controlled mott transition system .
recent @xmath14c nmr measurements under pressure up to 3 gpa @xcite have revealed that the antiferromagnetic moment of 1@xmath15/dimer at ambient pressure , which is a hallmark of a dm insulator , decreases to 0.5@xmath15/dimer at 0.6 gpa .
this result implies that the dm insulator picture of @xmath2-(bedt - ttf)@xmath1icl@xmath1 may be violated under high pressure . indeed , first - principles band structure calculations @xcite have pointed out that the degree of dimerization is significantly reduced with increasing pressure , suggesting that the electronic structure of @xmath2-(bedt - ttf)@xmath1icl@xmath1 changes from the effective half - filled to a 3/4-filled band system under pressure .
in this situation , the effective model should be the 3/4-filled extended hubbard model that takes into account the charge degrees of freedom inside of dimers and interdimer coulomb interactions .
recent theoretical calculations based on the extended hubbard model @xcite have revealed that in weakly dimerized systems both charge and spin fluctuations play an important role in unconventional superconductivity , pointing out the need to revisit the pairing mechanism that drive the high-@xmath9 superconductivity of @xmath2-(bedt - ttf)@xmath1icl@xmath1 . since a detailed knowledge of the electronic structure is an essential starting point for understanding the superconducting pairing mechanism , experimental studies on the electronic structure of @xmath2-(bedt - ttf)@xmath1icl@xmath1 under high pressure
are highly desired .
however , the high - pressure effect on the electronic structure of @xmath2-(bedt - ttf)@xmath1icl@xmath1 have not been reported so far , except for the first - principles band structure calculations @xcite .
-(bedt - ttf)@xmath1icl@xmath1 .
( b ) arrangement of bedt - ttf molecules within the @xmath16 plane viewed along the long axis of the bedt - ttf molecule .
@xmath17 is the on - molecule coulomb interaction , @xmath18 is the intradimer coulomb interaction , and @xmath13 is the effective on - dimer coulomb interaction .
the transfer integrals , @xmath19 , are also shown .
the ovals represent dimers . in the strong dimerization limit , @xmath20 .
( c ) schematic picture of intra- and interdimer charge transfer excitations in a dm insulator .
@xmath21 denotes the dimerization gap .
( d ) optical reflectivity and ( e ) conductivity spectra of @xmath0-(bedt - ttf)@xmath1icl@xmath1 for @xmath22 and @xmath23 at various temperatures at ambient pressure .
solid lines in ( e ) indicate the hubbard and dimer bands .
the spectra in ( e ) are offset for clarity .
] the infrared ( ir ) optical conductivity is a powerful probe to elucidate low - energy excitations reflecting the electronic structure of matter .
especially , in strongly correlated organic compounds , characteristic electronic structures associated with the on - site and inter - site coulomb interactions emerge in the mid - ir region @xcite . in this study , in order to investigate the pressure evolution of the electronic structure of @xmath2-(bedt - ttf)@xmath1icl@xmath1 , we performed polarized reflectivity measurements under high pressure by using a synchrotron radiation light source of spring-8 and a diamond anvil cell ( dac ) @xcite .
we show that the electronic structure of @xmath2-(bedt - ttf)@xmath1icl@xmath1 evolves from effective half - filled to 3/4-filled with increasing pressure .
moreover , we find that the dimensionality of the electronic structure switches from q1d to 2d , in which the high-@xmath9 superconductivity emerges .
the rest of the paper is as follows . in sec.ii
we briefly present the experimental details of the optical conductivity experiments at ambient pressure and under high pressure .
the results of the optical conductivity at ambient pressure and under high pressure are presented in secs.iii a and b , and discussed in the following secs.iii c - e .
finally , our main results are summarized in sec.iv .
analytical details of the optical conductivity spectra under pressure are given in appendix .
single crystals of @xmath0-(bedt - ttf)@xmath1icl@xmath1 were grown by an electrochemical oxidation method @xcite . for high - precision reflectivity measurements , single crystals with very flat surfaces are required . in this study , deuterated @xmath0-(bedt - ttf)@xmath1icl@xmath1 single crystals , which have much clearer surfaces than the hydrogenated ones , were used .
one should note that in @xmath0-(bedt - ttf)@xmath1icl@xmath1 substitution effects by deuterated bedt - ttf molecules on the fundamental magnetic and optical properties are negligibly small .
polarized reflectivity measurements in the mid - ir region ( 6008000 @xmath24 ) at ambient pressure were performed in vacuum with a fourier transform microscope spectrometer along the @xmath25 ( which is perpendicular to the @xmath26 axis within the @xmath16 plane and corresponding to the stacking direction of the bedt - ttf molecules ) and @xmath26 axes .
the reflectivity in the range of 800045 000 @xmath24 was measured at room temperature and ambient pressure .
the optical conductivity was calculated by a kramers - kronig ( kk ) transformation . above 45 000
@xmath24 , the standard @xmath27 extrapolation was used . at low frequencies ,
a constant extrapolation was used since the material is insulating with a large optical gap of @xmath32000 @xmath24 ( which is estimated from the dc resistivity @xcite ) .
polarized reflectivity measurements under high pressure were performed by using a diamond anvil cell ( dac ) in the range of 5008000 @xmath24 . for simultaneous measurements of @xmath22 and @xmath23 ,
two small samples with flat surfaces were placed on the culet face of the diamond anvil with a small amount of apiezon m grease . to perform accurate measurements of the reflectivity for tiny samples , synchrotron radiation light at bl43ir in spring-8 was used . in this study , to obtain good hydrostaticity , we used glycerin as the pressure transmitting medium , which produces more hydrostatic pressure than solid media such as kbr and nacl @xcite .
the pressure was monitored with the ruby fluorescence method .
the reflectivity was determined by comparison with a gold thin film on the gasket . for data under pressure
, lorentz fitting analysis was used to obtain the optical conductivity spectra .
the polarized reflectivity measurements in the mid - ir region at ambient pressure were performed in vacuum along the @xmath25 and @xmath26 axes .
figure1(d ) shows the optical reflectivity spectra @xmath28 at various temperatures for @xmath22 and @xmath23 , which are consistent with previous results @xcite .
a large anisotropy of the reflectivity is clearly observed , reflecting the strong 1d electronic structure of @xmath0-(bedt - ttf)@xmath1icl@xmath1 .
several vibrational features can be seen below @xmath31500 @xmath24 ; these are attributed to the totally symmetric @xmath29 vibrational modes of the bedt - ttf molecule . in the mid - ir region ,
@xmath28 for @xmath22 shows broad features , as typically seen in a dm insulator .
the corresponding optical conductivity spectra @xmath30 obtained from the kk transformation are plotted in fig.1(e ) . at room temperature ,
@xmath30 for @xmath22 has a broad and asymmetric shape in the range of 20006000 @xmath24 . as lowering the temperature
, however , a peak structure centered at 3000 @xmath24 becomes sharper , which makes it clearer that the spectra for @xmath22 consists of two bands located at 3000 @xmath24 and 4000 @xmath24 .
in sharp contrast , @xmath30 for @xmath23 exhibits very low conductivity , reflecting the 1d nature of the electronic structure as was expected in the reflectivity spectra . in a dm insulator
, there are two absorption bands in the mid - ir region : the so - called hubbard and dimer bands corresponding to the inter- and intradimer charge transfers , respectively [ see fig.1(c ) ] .
the excitation energy of the interdimer charge transfer , @xmath31 , is given by the effective on - dimer coulomb interaction , @xmath13 , which causes splitting of the antibonding band .
in the extended hubbard model , the effective on - dimer coulomb interaction , @xmath13 , is represented by @xmath32 , where @xmath17 is the on - molecule coulomb interaction , @xmath18 is the intradimer coulomb interaction , and @xmath33 is the intradimer transfer integral @xcite . on the other hand ,
the excitation energy of the intradimer charge transfer , @xmath34 , is given by @xmath35 in the strong dimerization limit ( @xmath17 , @xmath18 @xmath36 @xmath37 ) .
recent band structure calculations based on the maximally localized wannier orbitals ( mlwo ) for a series of dimerized hall - filled organic salts @xcite have a remarkable development in estimates of model parameters .
the calculations based on the mlwo for @xmath0-(bedt - ttf)@xmath1icl@xmath1 have estimated the intradimer charge transfer @xmath35 @xmath38 @xmath39 to be 500 mev ( which corresponds to approximately 4000 @xmath24 ) , showing an excellent agreement with the experimental value of 4000 @xmath24 .
therefore the band at 4000 @xmath24 is assigned to the dimer band . on the other hand ,
the band at 3000 @xmath24 is characterized by the sharpness ; it s much sharper than that of the 2d @xmath5-type system @xcite , which reflects the narrow density of states in the 1d mott insulator @xcite .
this fact strongly supports that the band at 4000 @xmath24 is assigned to the hubbard band . in the strong dimerization limit , the central frequency of the hubbard band becomes smaller than that of the dimer band [ see fig.4(e ) for details ] .
therefore , the present result that the excitation energy of the hubbard band transition is lower than that of the dimer band transition indicates that @xmath0-(bedt - ttf)@xmath1icl@xmath1 is located in the strong dimerization regime at ambient pressure . in order to elucidate the pressure evolution of @xmath30 ,
the polarized reflectivity measurements under high pressure were performed by using a dac . in a dac , the reflectivity is measured at a sample / diamond interface , not at a sample / vacuum interface [ see fig.2(b ) ] .
therefore the large refractive index of diamond ( @xmath32.4 ) makes a difference between the spectra measured in a dac and vacuum , @xmath40 and @xmath28 ( see appendix for details ) .
in addition to this , multiple reflections in a thin layer of apiezon m grease between the sample and diamond also affects the @xmath40 spectra .
our analysis considering the above corrections have shown that the thickness of the thin layer of apiezon m grease is about 300 - 400 nm .
figure2(a ) shows the @xmath40 spectra for @xmath22 expected from @xmath28 measured in vacuum at ambient pressure , which well reproduces @xmath40 actually measured at a low pressure of 0.54 gpa [ see fig.2(c ) ] . .
solid lines indicate the fitting results .
( b ) optical conductivity spectra obtained from the fitting analysis for @xmath23 .
the spectra are offset for clarity . ] to extract @xmath30 from the measured @xmath40 spectra , we fit @xmath40 to a lorentz oscillator model ( see appendix for details ) .
we show examples of the fitting for @xmath22 and @xmath41 in figs.2(c)-(f ) and fig.3(a ) , respectively .
the measured @xmath40 is well reproduced by two lorentz oscillators in the mid - ir region , as well as several sharp phonons below @xmath31500 @xmath24 in the case of @xmath22 .
the corresponding optical conductivity spectra for @xmath22 and @xmath41 obtained from the fitting analysis are shown in figs.2(g)-(j ) and fig.3(b ) , respectively . at 0.54 gpa ,
@xmath30 for @xmath22 exhibits very similar behavior to that of ambient pressure ; the hubbard and dimer bands have been observed in the mid - ir region .
with further increase in pressure , the hubbard band becomes broader as shown in figs.2(h)-(j ) , but the central frequency does nt change so much with pressure . in sharp contrast , the dimer band shifts to much lower frequencies between 0.54 gpa and 2.9 gpa , and with increasing pressure the spectral weight ( sw ) becomes much larger . to clarify the pressure evolutions of the hubbard and dimer bands , we plot the pressure dependence of the central frequencies and sws of each band in figs.4(a ) and 4(b ) . between 0.54 gpa and 2.9 gpa ,
the central frequency of the dimer band rapidly decreases .
in contrast , the pressure dependence of the hubbard band is gentle compared to that of the dimer band . as a result , above 2.9 gpa , the central frequency of the dimer band becomes lower than that of the hubbard band .
as mentioned above , in the strong dimerization limit , the central frequency of the hubbard band should be smaller than that of the dimer band .
the present result therefore indicates that the simple dm insulator picture of @xmath0-(bedt - ttf)@xmath1icl@xmath1 has been violated above 2.9 gpa .
recent theoretical calculations based on the extended hubbard model @xcite have suggested that by weakening the degree of dimerization or by increasing the interdimer coulomb interaction , the dm insulating state can be suppressed and evolve into a co insulating state , where the electronic charge distributions are polarized inside of dimers [ see the schematic phase diagram of fig.4(d ) ] . according to the calculations ,
the intradimer charge transfer corresponding to the dimer band transition significantly softens compared to @xmath42 approaching the phase boundary between the dm and co insulating phases , resulting in a collective excitation @xcite . on the other hand , the hubbard band transition smoothly evolves into a charge transfer excitation induced by the inter - site coulomb interaction @xmath43 , as indeed observed in quarter - filled organic conductors in a co phase @xcite . very importantly , in this situation , the excitation energy of the dimer band transition can be smaller than that of the hubbard band transition .
in the light of this fact , the present result for @xmath0-(bedt - ttf)@xmath1icl@xmath1 that the excitation energy of the dimer band transition becomes smaller than that of the hubbard band under pressure strongly suggests that the electronic structure of @xmath0-(bedt - ttf)@xmath1icl@xmath1 changes from effective half - filled to 3/4-filled as a result of weakening of dimerization with increasing pressure .
such a decrease of dimerization under pressure has been discussed also in ( tmttf)@xmath1@xmath6 salts @xcite . under pressure , in addition to weakening of dimerization , an increase of the bandwidth has been observed .
figure4(c ) shows the pressure dependence of the total sws for @xmath22 and @xmath41 at the lowest temperature . the total sw continuously increases with increasing pressure for both polarization directions , reflecting the increase in the bandwidth due to the increase in the transfer integrals . therefore in @xmath0-(bedt - ttf)@xmath1icl@xmath1 ,
when the pressure is applied , the system would follow the track as sketched in fig.4(d ) .
what is remarkable here is that the system crosses the vicinity of the co phase , which can be associated with the superconducting pairing mechanism of @xmath0-(bedt - ttf)@xmath1icl@xmath1 as discussed later . in the case of @xmath5-(bedt - ttf)@xmath1@xmath6
, a slight application of pressure ( @xmath44 0.1 gpa ) induces a mott transition .
thus , the degree of dimerization is almost unchanged in the low - pressure region , so that the system remains in the strong dimerization limit .
indeed , this has been confirmed by the chemical pressure studies for @xmath5-(bedt - ttf)@xmath1cu[n(cn)@xmath1]br@xmath45cl@xmath46 @xcite , in which the central frequencies of neither the hubbard nor the dimer bands change with @xmath47 , indicating that the system is located far away from the co phase .
another important finding in this study is that the sw for @xmath41 significantly increases with increasing pressure .
the total sw for @xmath41 at 8.5 gpa becomes twice compared to that at ambient pressure , whereas for @xmath22 it increases only by 25% .
consequently , the anisotropy of the total sw becomes smaller .
this result indicates that the dimensionality of the electronic structure of @xmath0-(bedt - ttf)@xmath1icl@xmath1 evolves from q1d to 2d with increasing pressure .
according to the previous theoretical studies based on the fluctuation - exchange ( flex ) approximation including only the on - site coulomb interaction @xcite , with increasing pressure the nesting condition of the fermi surface becomes degraded owing to the dimensional crossover of the fermi surface . as a result , the magnetic structure described by the localized spins in the strongly correlated mott insulating state changes into an itinerant sdw state .
with further increase in pressure , @xmath4 decreases and the antiferromagnetic phase can be suppressed .
finally , unconventional superconductivity with @xmath48-wave symmetry emerges .
thus , the origin of the superconductivity of @xmath0-(bedt - ttf)@xmath1icl@xmath1 is considered to be associated with antiferromagnetic spin fluctuation . however , in the above calculations , intra- and interdimer coulomb interactions are not included .
our present study highlights the 3/4-filled nature of @xmath0-(bedt - ttf)@xmath1icl@xmath1 under pressure , in which the intradimer charge degrees of freedom and interdimer coulomb interactions become more important .
recent theoretical calculations analyzed by the random - phase approximation ( rpa ) and flex based on the extended hubbard model @xcite have pointed out that the superconducting phase appears near the charge - density - wave ( cdw ) phase as well as the sdw phase .
while the extended @xmath49-wave pairing symmetry is favored near the cdw phase , @xmath48-wave superconductivity is realized due to spin fluctuation near the sdw phase . since both superconducting gap
symmetries are classified by @xmath50 irreducible representation , the two superconducting states are cooperative with each other
. then charge fluctuation inside of dimers can enhance the spin - fluctuation induced superconductivity . in this context , the present experimental results for @xmath2-(bedt - ttf)@xmath1icl@xmath1 suggest that in addition to antiferromagnetic spin fluctuation , charge fluctuation plays an important role in the high-@xmath9 superconductivity of @xmath2-(bedt - ttf)@xmath1icl@xmath1 .
finally , we compare the @xmath51-@xmath52 phase diagrams obtained in the present study and the previous resistivity measurements using a cubic anvil cell @xcite .
according to the resistivity measurements , @xmath0-(bedt - ttf)@xmath1icl@xmath1 shows superconductivity above @xmath37 gpa , and @xmath9 reaches 14.2 k at 8.2 gpa [ see fig.4(e ) ] .
in the present study , we have focused on the mid - ir region , so we can not discuss details of the electronic structure near the fermi level such as a drude peak .
however , the optical response in the low - energy region below @xmath31000 @xmath24 seems to capture the features associated with the electronic structure near the fermi level . below 2.9 gpa , @xmath40 in the low - energy region
does nt change so much with temperature , reflecting the large optical gap independent of temperature . at 5.9 gpa
, however , @xmath40 in the low - energy region slightly increases with decreasing temperature from room temperature , and below 200 k it decreases with decreasing temperature and becomes temperature - independent below 100 k. this result indicates that the metal - insulator ( mi ) transition at 5.9 gpa occurs at about 200 k , which is consistent with the @xmath51-@xmath52 phase diagram obtained from the resistivity measurements . at 8.5 gpa , @xmath40 in the low - energy region increases with decreasing temperature and exhibits a maximum at 80 k , which suggests that the mi transition occurs at @xmath380 k. this implies that the highest pressure we reached corresponds to about 7 gpa in the @xmath51-@xmath52 phase diagram obtained from the resistivity measurements .
the discrepancy between the present study and the resistivity measurements may come from a difference in the hydrostatic stress conditions .
the magnetic properties of the low - temperature magnetically ordered phase have been investigated up to 3 gpa by the @xmath14c nmr measurements @xcite .
although we measured the optical conductivity down to low temperatures at each pressure , a clear difference in the optical spectra above and below the magnetic transition has not been observed .
this is because the electronic excitations associated with the hubbard and dimer bands are not largely affected by the magnetic order .
the experimental result that there is no large temperature variation at low temperatures in the optical spectra at 2.9 gpa and 5.9 gpa suggests that the system does nt enter an insulating state such as a co phase that drastically changes the electronic structure in the mid - ir region . therefore the antiferromagnetic mott insulating phase or the sdw phase should be realized in the low - temperature region above 3 gpa , as discussed above .
further experimental studies on the magnetic properties under higher pressures are desired .
in summary , we have performed optical conductivity measurements for @xmath0-(bedt - ttf)@xmath1icl@xmath1 under high pressure . at ambient pressure , two characteristic bands due to intra- and interdimer charge transfers have been observed in the optical spectra only for the polarization along the stacks of bedt - ttf molecules , strongly supporting that this salt is a q1d dm insulator at ambient pressure . with increasing pressure ,
however , the intradimer charge excitation shifts to much lower frequencies as a result of weakening of dimerization , implying that the electronic structure evolves from effective half - filled into 3/4-filled under pressure .
moreover , we have shown that the dimensionality of the electronic structure switches from q1d to 2d , in which the high temperature superconductivity emerges .
these results indicate that the phase diagram of @xmath0-(bedt - ttf)@xmath1icl@xmath1 is different from that of @xmath5-(bedt - ttf)@xmath1@xmath6 where the superconducting phase appears in the effective half - filled band system .
the present results suggest that the system approaches the charge - ordered state under pressure , in which the possible charge - fluctuation - mediated high-@xmath9 superconductivity emerges .
we thank c. hotta , s. iwai , and a. kawamoto for fruitful discussions .
synchrotron radiation measurements were performed at spring-8 with the approvals of jasri ( 2013a0089 , 2013a1118 , 2013b0089 , and 2013b1144 ) .
this work is supported by a grant - in - aid for scientific research ( grants no .
23540409 , no . 24540357 , no .
25287080 , no . 26287070 , no .
26800173 , and no .
15h00984 ) from mext and jsps , japan .
in a dac , the reflectivity is measured at a sample / diamond interface . according to the fresnel s formula ,
the reflectivity spectra @xmath28 at an interference between a sample and a transparent medium is given by @xmath53 ^ 2+k(\omega)^2}{[n(\omega)+n_0]^2+k(\omega)^2 } , \tag{a1}\end{aligned}\ ] ] where @xmath54 and @xmath55 are the real and imaginary parts of the complex refractive index of the sample , respectively , and the refractive indices @xmath56 for vacuum and diamond are 1 and 2.419 , respectively .
therefore @xmath40 measured in a dac generally becomes smaller than @xmath28 measured in vacuum .
in addition to this , multiple reflections in a thin layer of apiezon m grease between the sample and diamond [ see fig.2(b ) in the main text ] also affects @xmath40 , especially in the high - energy region @xcite ; this is because shorter wavelength lights are more likely to be influenced by such multiple reflections .
the reflectivity considering the multiple reflections can be given by @xcite @xmath57 where @xmath58 , @xmath59 is the complex refractive index of the sample , @xmath60 and @xmath61 are the refractive indices of diamond and apiezon m grease , and @xmath62 is the thickness of apiezon m grease .
our analysis considering the above equation have shown that the thickness of the thin layer is about 300 nm and 430 nm for the @xmath22 and @xmath23 samples , respectively .
figure2(a ) in the main text shows the @xmath40 spectra for @xmath22 expected from @xmath28 measured in vacuum at ambient pressure , which well reproduces @xmath40 actually measured at a low pressure of 0.54 gpa [ see fig.2(c ) ] . here , to calculate @xmath40 expected in a dac at ambient pressure , @xmath54 and @xmath55 were derived from the kk analysis of @xmath28 measured in vacuum at ambient pressure .
then , the obtained parameters were substituted into eq.(a2 ) with @xmath63 nm . to extract @xmath30 from the measured @xmath40 spectra , we fit @xmath40 to a lorentz oscillator model . in this model ,
the real and imaginary parts of the complex dielectric function , @xmath64 and @xmath65 , are given by @xmath66 where @xmath67 is the high - frequency dielectric constant , @xmath68 , @xmath69 , and @xmath70 are the plasma frequency , the central frequency , and the linewidth of the @xmath71-th lorentz oscillator , respectively .
the real and imaginary parts of the complex refractive index of the sample , @xmath54 and @xmath55 , are described by the above lorentzian parameters through the relations @xmath72 and @xmath73 .
then , by fitting the measured @xmath40 based on eq.(a2 ) , we can obtain the lorentzian parameters .
thus , we can extract @xmath30 through the relation @xmath74 .
we show the fitting results for @xmath22 and @xmath41 in figs.2(c)-(f ) and fig.3(a ) , respectively .
the corresponding optical conductivity spectra for @xmath22 and @xmath75 are shown in figs.2(g)-(j ) and fig.3(b ) , respectively . in @xmath0-(bedt - ttf)@xmath1icl@xmath1 , there exists only one equivalent dimer in the unit cell .
therefore only the acoustic mode can exist , in which polarizations in two adjacent dimers align in parallel with each other .
since the direction of polarization is expected to be evolved along the direction from the center of one bedt - ttf molecule to the other one in the dimer ( which is close to the @xmath25 axis ) , the acoustic mode appear nearly along the @xmath25 axis .
however , the direction of polarization and the @xmath26 axis are not perfectly perpendicular to each other , so the same acoustic mode should be observable in the polarization of @xmath41 in @xmath0-(bedt - ttf)@xmath1icl@xmath1 . | to elucidate the pressure evolution of the electronic structure in an antiferromagnetic dimer - mott ( dm ) insulator @xmath0-(bedt - ttf)@xmath1icl@xmath1 , which exhibits superconductivity at 14.2 k under 8 gpa , we measured the polarized infrared ( ir ) optical spectra under high pressure . at ambient pressure
, two characteristic bands due to intra- and interdimer charge transfers have been observed in the ir spectra , supporting that this salt is a typical half - filled dm insulator at ambient pressure . with increasing pressure , however , the intradimer charge transfer excitation shifts to much lower energies , indicating that the effective electronic state changes from half - filled to 3/4-filled as a result of weakening of dimerization .
this implies that the system approaches a charge - ordered state under high pressure , in which charge degrees of freedom emerge as an important factor .
the present results suggest that charge fluctuation inside of dimers plays an important role in the high - temperature superconductivity .
= 10000 |
topological properties of low - dimensional quantum many - body systems have been attracting considerable interest in both quantum information sciences and condensed matter physics .
remarkably , it is understood that important information of the topological phase is encoded in the von neumann entanglement entropy of its ground state.kitaev-preskill,levin-wen,zhang-vishwanath in a seminal paper , li and haldane@xcite proposed that the entanglement spectrum ( es ) labeled by the quantum numbers of symmetries preserved by the schmidt singular - value decomposition contains _ more _ information than the entanglement entropy . in particular , they found the largest eigenvalues of the ground state reduced density matrix mimics the physics edge spectrum of a gapped quantum system , thus revealing the bulk topological order .
a large number of works have followed , suggesting that the low - energystructure of es reveals the bulk topological order.fidkowski,cirac-2011,bernevig-2011,qi-2012 recently , a critique on this body of work was given by chandran , khemani , and sondhi@xcite , who posed a question,``how universal is the entanglement spectrum ? '' in this paper we will address this important question by studying the es of a family of one - dimensional symmetry protected topological ( spt ) phases.chen-gu-wen-2011,schuch,chen-gu-liu-wen the spt phases have robust gapless edge excitations and can not be continuously connected to trivial product state without either breaking the protecting symmetry or closing the energy gap . moreover
, the one - dimensional spt phases have been generally classified by group cohomologies completely.@xcite es of these phases are sufficiently simple and can be calculated analytically for model wave functions@xcite , thus providing an ideal setting for addressing this issue .
our results demonstrate that es of topological phases include both topological / universal and non - topological parts . in certain cases
the universal part can be isolated , and is extremely useful in revealing the bulk topological order .
the simplest example of spt phases is the haldane gapped phase of the antiferromagnetic spin-1 chain,@xcite which is protected by any one of the following discrete symmetries : time reversal symmetry , link inversion symmetry , or the @xmath6 symmetry comprising @xmath7 rotations about two orthogonal axes.gu-wen-2009,pollmann-oshikawa-2010,pollmann-oshikawa-2012 the fixed point wave functions for these phases are the valence bond solid ( vbs ) states.aklt we show their es contains a topological part , which is universal , and a non - topological or non - universal part .
the topological part can be isolated by a topological disentangler , a non - local unitary transformation .
for the familiar @xmath0 symmetric vbs states with odd - integer spins , their es is given by a single energy level with @xmath8-fold degeneracy .
however , the bulk topological order only gives rise to ( or protects ) a two - fold degeneracy , which is associated with the @xmath9 symmetry .
the remaining @xmath10-fold degeneracy is _ not _ protected by the bulk topological order .
furthermore , for vbs phases with higher symmetry , we can find corresponding high topological degeneracy and associated protecting symmetry via a generalized non - local unitary transformation .
specifically , we find in the @xmath11 symmetric vbs state with spin-@xmath12,@xcite the topological degeneracy are four - fold and the protecting symmetry can be identified as @xmath13 . in general , for the @xmath2 symmetric vbs states with integer spin-@xmath3,tu-2008,tzx , tu-2009,else - bartlett - doherty , quella-2013 the corresponding protecting symmetry is given by @xmath14 , and the topological degeneracy is @xmath5-fold .
the @xmath0 symmetric vbs states with arbitrary integer spin for a periodic spin chain is given by@xcite @xmath15where the schwinger spin - boson representation with a local constraint @xmath16 has been used as@xmath17one can construct an @xmath18 spin invariant model hamiltonians , whose ground states are exactly given by these vbs states.@xcite in the thermodynamic limit , the spin correlation function decays exponentially with a correlation length @xmath19 , implying that the vbs state is a gapped disordered state . for a finite length chain with open boundary condition ,
the ground state wave function is expressed as @xmath20where @xmath21 and @xmath22 can be chosen from @xmath23,@xmath24 , representing the eigenvalues of the edge spins of @xmath25 and @xmath26 .
there are total @xmath27 degenerate edge states . as model wave functions of a family of one - dimensional spt phases , the vbs states exhibit the simplest es . to calculate the es for the wave functions @xmath28where @xmath29
after the normalization of all wave functions are carefully considered , the resulting wave function is written as @xmath30where a ratio of the length dependent factors has been used : @xmath31 , and the entanglement energy level is just given by a single level @xmath32 with a degeneracy @xmath33 , which exactly corresponds to the degeneracy of the vbs wave function with one edge spin .
so the degeneracy of each edge spin and the degeneracy of the entanglement level has one - to - one correspondence.@xcite however , a natural question to ask is whether all the degeneracies of the entanglement energy level are topologically protected or not .
eigenvalue ; ( b ) the right edge spin is specified by its @xmath34 eigenvalue.,width=283 ]
in order to characterize the topological features of these vbs states , non - local string order parameters ( sop ) were first proposed for the spin-1 case,@xcite , and then generalized to arbitrary integer spins@xcite : @xmath35where @xmath36 , @xmath37 or @xmath38 . in the thermodynamic limit
, the sop is found to be @xmath39which reveals the fundamental difference between vbs states with odd- and even - integer spins .
only vbs states with odd - integer spins represent topologically nontrivial phases .
an alternative and more useful characterization was revealed through a non - local unitary transformationoshikawa-1992,kennedy - tasaki-1992 @xmath40from which the local spin operators are transformed to @xmath41where a string operator in @xmath42 involves all the sites from left edge site up to @xmath43 , while the string operator in @xmath44 involves all the sites from the site @xmath45 up to the right edge site .
these properties will become crucial in studying the _ transformed _ vbs wave function .
meanwhile , the non - local sop becomes to the local spin correlation via this transformation : @xmath46
for the convenience of performing the non - local unitary transformation , the vbs wave functions are expressed in terms of the eigenvalue basis of the left edge spin @xmath25 and the right edge spin in the @xmath47 : @xmath48where @xmath49 and @xmath50 .
the transformed vbs states are ferromagnetic long - range ordered states with two finite magnetizations @xmath51 and @xmath52 as local order parameters@xcite:@xmath53therefore the non - local transformation with the discrete symmetry @xmath1 plays the role of a topological disentangler , because it turns topological order of the original vbs state into conventional ferromagnetic order .
the eigenvalues @xmath21 and @xmath22 of the edge spins are used to classify the transformed degenerate ground states @xmath54 .
there exist four different classes of long - range ordered ferromagnetic states with the degeneracy @xmath55 each .
these degeneracies of each ferromagnetic ordered state only appear for @xmath56 odd - integer spins and results from the partial polarization of the spins .
the detailed ferromagnetic configurations are summarized in table i. .the transformed so(3 ) symmetric vbs states with odd - integer spins .
they are ferromagnetic states classified into four groups by the eigenvalues of the edge spins @xmath57 and @xmath58 , each with @xmath10-fold degeneracy . [ cols="^,^,^,^",options="header " , ] to consider the es of the transformed vbs state , we cut the spin chain into two parts a and b ( see fig.1b ) . when the edge spin @xmath59 is transformed into the @xmath60 basis ,
the vbs wave function is written as @xmath61where @xmath62and we have expressed the edge spins in the wave function for part a in terms of @xmath25 and @xmath63 basis , and the edge spins in the wave function of part b in terms of @xmath60 and @xmath47 basis .
when the non - local unitary transformation is applied to the vbs wave function , we separate the transformation into three parts @xmath64 , where @xmath65@xmath66 plays the essential role in determining the new es .
@xmath67 and @xmath68 take the eigenvalues of the corresponding edge spins
. then the transformed wave function is written as @xmath69and the total density matrix of @xmath70 is thus obtained . after the normalization factors are properly taken into account , we can trace out the degrees of freedom of the wave function of the part a , and the reduced density matrix is derived as @xmath71 _ { m , m^{\prime } } = \sum_{n=0}^{s}\sum_{m=0}^{s}\sum_{m^{\prime } = 0}^{s}\frac{1% } { s+1}\frac{c_{s}^{n}}{\sqrt{c_{s}^{m}c_{s}^{m^{\prime } } } } \nonumber \\ & & \text { \ \ \ \ } \times f\left ( s , n , m\right ) f\left ( s , n , m^{\prime } \right ) e^{i\pi \left ( \alpha + n-\frac{s}{2}\right ) \left ( m - m^{\prime } \right ) } .\end{aligned}\ ] ] to complete the summations , some further algebra is performed to rewrite the function @xmath72 in terms of the schwinger bosons .
when @xmath73 is even , the phase factor disappears @xmath74 , and the diagonal elements of the reduced density matrix is obtained @xmath75 _ { m , m^{\prime } } = \frac{\delta _ { m , m^{\prime } } } { s+1}.\ ] ] however , when @xmath73 is odd , there is a phase factor @xmath76 in the summations , and non - diagonal elements of the reduced density matrix is given by @xmath75 _ { m , m^{\prime } } = \frac{(-1)^{\alpha
-s/2}}{\left ( s+1\right ) } \delta _ { m+m^{\prime } , s}.\]]for the even - integer spins @xmath3 , all the non - diagonal matrix elements are zero because @xmath77 can not be even , and then @xmath78 is a diagonal matrix with the same element @xmath79 .
this implies that the es of the vbs states with even - integer spins remains the _ same _ after the non - local unitary transformation .
thus the degeneracy in es of vbs states with even - integer spins is _ not _ of topological nature , consistent with the absence of the non - local sop found previously and triviality of this phase.@xcite however , for the odd - integer spins @xmath3 , the above results show that @xmath80 has diagonal elements @xmath79 and skew diagonal elements @xmath81 , so it is a x - form matrix . for any edge configuration , @xmath78 has one eigenvalue @xmath82 , and the entanglement energy level is dramatically changed and given by @xmath83 with @xmath84 fold degeneracy . in the spin-1 case ,
the transformed vbs wave function is a ferromagnetic product state , and the entanglement energy level is non - degenerate__.__@xcite for the @xmath56 odd - integer spins , however , the transformed vbs wave functions are still entangled states , which are spin partially polarized ferromagnetic states with a degeneracy @xmath85 for each class . the es degeneracy of the transformed vbs states depends on the precise form of the resultant ferromagnetic states , and is thus non - topological . the crucial point , however , is that the non - local unitary transformation with the discrete symmetry @xmath9 has lifted the two - fold topological degeneracy , which is shared by all of these odd - integer spin vbs states belonging to the _ same _ topological phase@xcite .
meanwhile , the topological degeneracy and the topologically protecting symmetry can be read off from the non - local unitary transformation .
in order to put the above results in a large context , we consider vbs states with larger symmetry groups which represent spt phases with higher symmetry . we start with a topologically _ nontrivial _ vbs state formed in a spin-@xmath87 chain . viewing each spin-@xmath12 as formed by two spin-@xmath88 s , we can construct an @xmath86 symmetric vbs state@xcite .
as pointed out in ref.@xcite , the @xmath89 states of a spin-@xmath88 can be regarded as the four states of the spinor representation of @xmath86 .
similarly one can also represent the @xmath90 states of spin-@xmath12 as the five - dimensional vector irreducible representation ( ir ) of @xmath86 .
then we can view the vector ir as the symmetric component of the tensor product of two virtual spinor ir s , i.e. , @xmath91the numerals are the dimensions of the @xmath86 ir s .
the tensor product of two @xmath92 s on adjacent sites decomposes into @xmath93comparing eqs .
( [ decom1 ] ) and ( [ decom2 ] ) , we can regard eq .
( decom1 ) as the tensor product of two neighboring virtual spins after their respective partners have form @xmath86 singlet with other virtual spins
. then one can find that @xmath86 singlet @xmath94 and the antisymmetric @xmath95 appear in the decomposition but the symmetric @xmath96 is absent .
therefore , if @xmath97 , the @xmath86 symmetric vbs state where neighboring virtual @xmath98 s pair into @xmath86 singlet will be the ground state .
the operator @xmath99 can be expressed in terms of the @xmath86 generators @xmath100because the physical spin @xmath18 is a subgroup of @xmath86 , each ir of @xmath11 must decompose into an integral number of @xmath18 multiplets .
thus the @xmath101 discussed above must be expressible as the direct sum of @xmath18 ir obtained by decomposing the direct product of two @xmath102 multiplets . since the 14-dimensional ir is symmetric upon the exchange of site indices , it must only contain even - spin @xmath18 multiplets , i.e. @xmath103 . for an open chain , there exist a free spin-@xmath88 at each end of the chain , leading to 4-fold degeneracy which is topologically protected by a larger symmetry .
the corresponding topological es degeneracy should also be 4-fold . for comparison , in the @xmath0 symmetric vbs states
, topology only protects a half edge spin and 2-fold es degeneracy for odd - integer spins , while there is no protection at all for even - integer spins .
it is straightforward to promote the symmetry of the vbs states and demand that the spin-@xmath3 states on each site transform under the @xmath104-dimensional vector representation of @xmath105 , which can be formed by tensor decomposition of two virtual @xmath5-dimensional spinors.@xcite the main issue here is to identify the topological degeneracy .
since @xmath2 is a rank-@xmath3 algebra , there are @xmath3 mutually commuting cartan generators : @xmath106 .
at each site , the quantum states are classified by the eigenvalues of these cartan generators as @xmath107thus the single - site states are associated with @xmath3 quantum numbers @xmath108 , and they are subjected to the constraint @xmath109the topological feature of these vbs states can be characterized by the following generalized non - local sop @xmath110since the ground state is @xmath2 rotational invariant , the above non - local order parameters should all be equal to each other . to determine the value of these parameters , only @xmath111 needs to be evaluated . in the @xmath112 channel ,
the role of the phase factor in eq .
( eq : sop ) is to correlate the finite spin polarized states in the @xmath113 channel at the two ends of the string .
if nonzero @xmath113 takes the same value at the two ends , then the phase factor is equal to @xmath114 . on the other hand ,
if nonzero @xmath113 takes two different values at the two ends , then the phase factor is equal to @xmath115 .
thus the value of @xmath111 is determined purely by the probability of @xmath116 appearing at the two ends of the string .
it is straightforward to show that the probability of the states @xmath116 appearing at one lattice site is @xmath117 and thus @xmath118 . in the @xmath2
lie algebra , @xmath119 @xmath120 span an @xmath0 sub - algebra in which @xmath121 plays the role of flipping the quantum number @xmath122 .
this exponential operator can flip the quantum numbers of @xmath122 without disturbing the quantum states in all other channels .
thus the following non - local unitary transformation with @xmath123 discrete symmetry @xmath124can change the spin configuration in the @xmath125 channel into a ferromagnetically ordered one .
furthermore , by performing this non - local transformation successively in all the channels @xmath126 , all the configurations of the ground state will become ferromagnetically ordered .
applying @xmath127 to the _ cartan _ generators , it can be shown that @xmath128substituting this formula to eq .
( [ eq : sop ] ) , we find that @xmath129thus the non - local sop @xmath130 of the cartan generators become the ordinary two - point correlation functions of local operators . under this general non - local unitary transformation with the @xmath131 symmetry
, the topological order of the original vbs states is transformed into conventional ferromagnetic order , just like in the so(3 ) case .
when an open chain system is considered , there appear @xmath5 degenerate edge states at each end of the chain , which are also topologically protected by the discrete symmetry @xmath14 . accordingly the topological degeneracy in the es can be read off as @xmath5-fold .
we thus find that a higher symmetry in the spt phase protects a higher es degeneracy .
we have shown through explicit examples that entanglement spectra of topological ground states contain both universal and non - universal structures .
the topological degeneracy of the lowest entanglement energy level may be lifted or isolated by a generalized non - local unitary transformation or topological disentangler , which is determined by the minimal symmetry protecting the topological phases .
our results shed significant light on the issue of using entanglement spectrum to identify topological order .
one of the authors ( gmz ) would like to thank h. h. tu and t. xiang for their earlier collaborations and acknowledge the support of nsf - china through the grant no.20121302227 .
ky is supported by national science foundation through grant no . | one - dimensional valence bond solid ( vbs ) states represent the simplest symmetry protected topological phases .
we show that their ground state entanglement spectrum contains both topological and non - topological structures . for the @xmath0 symmetric vbs states with odd - integer spins ,
the two - fold topological degeneracy is associated with an underlying @xmath1 symmetry that protects the corresponding topological phase . in general , for the @xmath2 symmetric vbs states with integer spins @xmath3 , the corresponding protecting symmetry is identified as the @xmath4 symmetry , yielding the @xmath5-fold topological degeneracy .
the topological degeneracy and associated protecting symmetry can be identified by a non - local unitary transformation , which changes the topological order of the vbs states into conventional ferromagnetic order . |
superconductivity is one of most attractive topics in condensed matter physics area not only because of the great application potential of those superconductors ( scs ) with high transition temperature and critical field but also due to the importance of understanding the mechanism of cooper pairing , especially for the unconventional superconductivity beyond bardeen - cooper - schrieffer ( bcs ) mechanism . according to the ginzburg - landau ( gl ) theory , the value of gl parameter @xmath4 which is the ratio of penetration depth to coherence length classifies scs into two categories : type - i scs when @xmath5 and type - ii scs when @xmath6.@xcite for most of superconducting compounds , they belong to the type - ii scs and have been studied extensively .
in contrast , type - i scs are thought empirically to occur mainly in elementary metals and metalloids and type - i superconducting compounds are very rare.@xcite recently , however , serval binary and ternary compounds are found to be type - i scs , for instance , ybsb@xmath0,@xcite tasi@xmath0,@xcite lapd@xmath0ge@xmath0,@xcite larh@xmath0si@xmath0,@xcite ( lu , y , la)pd@xmath0si@xmath0,@xcite larhsi@xmath7,@xcite ag@xmath8pb@xmath0o@xmath9,@xcite scga@xmath7 and luga@xmath7.@xcite these studies break the empirical relation between type - i superconductivity and elemental metals and enlarge the family of type - i scs to binary and ternary compounds . for the binary bismuth compounds , some of them show superconductivity , such as kbi@xmath0(3.6 k),@xcite srbi@xmath7 ( 5.62 k),@xcite babi@xmath7 ( 5.69 k),@xcite rh@xmath7bi@xmath10 ( 2.94 k),@xcite and in@xmath0bi ( 5.9 k).@xcite for kbi@xmath0 , except superconducting transition temperature @xmath1 , the studies on its physical properties are scarce and its classification of superconductivity has not been identified yet.@xcite in this work , we performed the detailed characterization and analysis of physical properties for kbi@xmath0 single crystals in superconducting and normal states .
experimental and theoretical calculation results undoubtedly indicate that kbi@xmath0 is a type - i sc in the dirty limit . as far as we know ,
this is the first type - i sc in the bismuth compounds .
single crystals of kbi@xmath0 were grown by the flux method with k : bi = 1 : 9 molar ratio .
k pieces ( 99.9 @xmath11 ) and bi shot ( 99.9 @xmath11 ) were mixed and put into an alumina crucible , covered with quartz wool and then sealed into the quartz tube with partial pressure of argon .
the quartz tube was heated to 580 @xmath12c for 12 h and then slowly cooled to 280 @xmath12c where crystals were decanted with a centrifuge .
single crystals with typical size 1.8@xmath13@xmath141 mm@xmath15 were obtained and exhibit metallic luster .
x - ray diffraction ( xrd ) of powdered small crystals and a single crystal were performed using a bruker d8 x - ray machine with cu @xmath16 radiation ( @xmath17 0.15418 nm ) at room temperature .
rietveld refinement of the xrd patterns was performed using the code topas4.@xcite electrical transport , magnetization and specific heat measurements were performed in a quantum design ppms-14 .
first - principles electronic structure calculations were carried out with the projector augmented wave method @xcite as implemented in the vasp package.@xcite the generalized gradient approximation of perdew - burke - ernzerh @xcite was adopted for the exchange - correlation potential .
the kinetic energy cutoff of the plane - wave basis was set to be 350 ev . a supercell containing 2 k atoms and 4 bi atoms and a @xmath18 k - point mesh for the brillouin zone sampling were employed .
the gaussian smearing with a width of 0.05 ev was used around the fermi surface . in structure optimization ,
both cell parameters and internal atomic positions were allowed to relax until the forces were smaller than 0.01 ev / .
the spin - orbital coupling ( soc ) effect was included for the density of states ( dos ) calculations .
the large red and small blue balls represent k and bi atoms , respectively .
the green rectangle marks the ( 111 ) plane .
( b ) the kagom@xmath19 net of bi atoms in ( 111 ) plane .
( c ) powder xrd pattern of kbi@xmath0 .
( d ) xrd pattern of a kbi@xmath0 single crystal .
the extra peaks in xrd patterns ( labeled as the # symbol ) originate from residual bi flux .
inset : photo of a typical kbi@xmath0 single crystal .
the length of one grid in the photo is 1 mm . ] as shown in fig .
1(a ) , kbi@xmath0 is isostructural to mgcu@xmath0 ( laves phase).@xcite each unit cell of kbi@xmath0 has 8 k atoms and 16 bi atoms .
every four bi atoms form a tetrahedron and these tetrahedra connect each other by vertex - sharing to form a three - dimensional network .
k atoms arrange in a diamond lattice which is intertwined with the network of bi tetrahedra . on the other hand , it can be seen that there are two - dimensional ( 2d ) kagom nets of bi atoms in the ( 111 ) plane ( fig .
1(b ) ) , which connect together along [ 111 ] direction by other bi layers , and k atoms locate right above and below the center of hexagons of bi atoms in the 2d kagom nets .
the powder xrd pattern of kbi@xmath0 can be well indexed using the @xmath20 space group ( mgcu@xmath0-type structure ) ( fig .
the refined lattice parameters are @xmath21 0.95233(2 ) nm with @xmath22 5.88 , @xmath23 8.69 and @xmath24 1.28 , which are close to the values reported in literature ( @xmath21 0.95223(2)nm).@xmath25 there are some extra peaks that can be ascribed to the diffraction of residual bi flux .
the xrd pattern of a kbi@xmath0 single crystal indicates that the surface of crystal is parallel to the ( 111 ) plane ( fig .
1(d ) ) . the kbi@xmath0 crystals prefer to form triangle surface as shown in the inset of fig.1(d ) .
this characteristic is consistent with the single crystal xrd pattern as well as its crystallographic symmetry . of kbi@xmath0 single crystal at zero field .
inset : expanded view of the low - temperature data showing the superconducting transition .
the @xmath26 at zero field is marked by arrow .
( b ) low - temperature dependence of @xmath27 of kbi@xmath0 single crystal at various magnetic fields from 0 to 200oe . ] fig .
2(a ) shows the temperature dependence of electrical resistivity @xmath27 for kbi@xmath0 single crystal at zero field .
it can be seen that kbi@xmath0 exhibits metallic behavior and the curve @xmath27 is convex above 25 k , with a tendency to saturate at high temperature .
this is a typical shape of resistivity for the sample in which the dominant scattering mechanism is electron - phonon scattering .
the residual resistivity ratio ( rrr ) , defined as @xmath28(300 k)/@xmath28(4 k ) , is about 72.5 , indicating the high quality of single crystal . on the other hand , at zero field
, there is a sharp superconducting transition in @xmath27 curve with @xmath29 3.549 k and transition width @xmath30 0.1 k ( inset of fig .
the transition temperature is consistent with previous magnetic susceptibility measurement ( @xmath31 3.5 k).@xcite moreover , with increasing magnetic field , the superconducting transition shifts to lower temperatures gradually and the transition width also becomes wider for @xmath32 ( 111 ) ( fig .
surprisingly , at very low field ( @xmath33 200 oe ) , the @xmath1 has been suppressed below 2 k. = @xmath34/(1-@xmath35@xmath36 ) scaled by 4@xmath37 at @xmath33 10 oe with zfc and fc modes .
the @xmath26 is marked by arrow .
( b ) temperature dependence of @xmath38 at various magnetic fields with zfc mode .
( c ) low - field parts of @xmath39 at various temperatures with demagnetization correction.(d ) magnetization loop at @xmath40 2 k. for all of measurements , the field direction is parallel to the ( 111 ) plane of kbi@xmath0 single crystal . ] as shown in fig .
3(a ) , low - temperature magnetic susceptibility of kbi@xmath0 single crystal for @xmath41 ( 111 ) with @xmath33 10 oe exhibits diamagnetic signals with sharp transitions appearing at @xmath29 3.517 k ( @xmath42 0.10 k ) for both the zero field cooling ( zfc ) and field cooling ( fc ) modes .
this superconducting transition temperature is almost same as that obtained from transport measurement , confirming the superconducting ground state of kbi@xmath0 . after considering the demagnetization effect of sample by using the formula @xmath43 where @xmath35 is demagnetization factor and the calculated value is about 0.222,@xcite
the superconducting volume fraction estimated from the zfc data is very close to 100 % , indicating that the bulk superconductivity in kbi@xmath0 .
meanwhile , the large superconducting volume fraction for fc data suggests the flux pinning effect is rather weak or even absent .
further increasing field suppresses the @xmath1 gradually and when @xmath33 175 oe , the normal states behavior persists down to 2 k , similar to the trend in @xmath27 curves ( fig .
2(b ) ) . fig .
3(c ) shows the dc magnetization isotherms @xmath44 of kbi@xmath0 single crystal at various temperatures ( from 2 k to 3.6 k ) for @xmath45 ( 111 ) at the low - field range .
it can be seen that the magnetic fields corresponding to the maximum absolute values of magnetic moment are very close to those where @xmath44 become zero .
based on this , two different assumptions can be made : either ( i ) kbi@xmath0 is a type - ii sc that the upper critical field @xmath46 is very small and close to the lower critical field @xmath47 ; or ( ii ) kbi@xmath0 is a type - i sc and the departure from the ideal step - like transition at critical field may be attributed to a strong pinning of domain walls in the intermediate state or to unclear demagnetization.@xmath48 for the latter case , the critical field @xmath49 is defined as the field where the sample enters to the normal state ( @xmath50 0 ) . on the other hand , when temperature decreases , the critical field shifts to higher field . even at the lowest measuring temperature ( 2 k ) , however , the critical field is still remarkably small ( @xmath51 160 oe ) .
the full magnetization loop of kbi@xmath0 measured at 2 k for @xmath41 ( 111 ) is shown in fig .
it can be seen that the hysteresis is very small .
the small critical field accompanying with the weak magnetization hysteresis and the shape of @xmath39 loop similar to other type - i superconducting compounds , such as ybsb@xmath0,@xcite larhsi@xmath7,@xcite scga@xmath7 and luga@xmath7,@xcite suggest that the kbi@xmath0 is more like a type - i sc rather than a type - ii sc . vs. @xmath52 of kbi@xmath0 single crystal in a series of magnetic fields from 0 to 307 oe .
( b ) temperature dependence of specific heat from 2.2 k to 300 k at zero field .
the solid line shows the classic value of specific heat at high temperature limit .
inset : temperature dependence of the electronic specific heat plotted as @xmath53 vs. @xmath54 at zero field . ]
specific heats of kbi@xmath0 single crystal at various fields are shown in fig .
at zero field , there is a jump at 3.573 k , indicating the bulk superconductivity of kbi@xmath0 .
the transition temperature is also consistent with the @xmath1 obtained from resistivity and magnetization measurements .
more importantly , the specific - heat jump at superconducting transition at @xmath33 7 oe is much sharper and higher than that at zero field , which is a unique feature of type - i scs and indicates that there is a crossover from second- to first - order phase transition when field is applied .
this behavior has been observed in other type - i intermetallic scs , such as ybsb@xmath0,@xcite larhsi@xmath7,@xcite scga@xmath7 and luga@xmath7,@xcite therefore , specific heat measurement further supports that the kbi@xmath0 is a type - i sc . on the other hand , the superconducting transition shifts to lower temperature with increasing field and the superconductivity
is suppressed below 2.2 k when @xmath55 157 oe . in the normal state ( @xmath56 307 oe ) , the electronic specific heat coefficient @xmath57 and phonon specific heat coefficient @xmath58 are obtained using the linear fit @xmath59 .
the fitted @xmath57 and @xmath58 is 1.3(4 ) mj mol@xmath60 k@xmath61 and 3.10(4 ) mj mol@xmath60 k@xmath62 , respectively .
the latter one gives the debye temperature @xmath63 123.4(5 ) k using the formula @xmath64 .
the electron - phonon coupling @xmath65 can be estimated with the values of @xmath66 and @xmath1 using mcmillan s
theory,@xcite @xmath67 where @xmath68 is the repulsive screened coulomb potential and is usually between 0.1 and 0.15 .
setting @xmath69 0.13 , the calculated @xmath65 is 0.774 , implying that kbi@xmath0 is an intermediately or strongly coupled bcs sc .
4(b ) shows the specific heat of kbi@xmath0 measured from 2.2 k to 300 k at zero field .
the specific heat at high temperature approaches the value of @xmath70 at 300 k , where @xmath71 is the atomic number in the chemical formula ( @xmath72 3 ) and @xmath73 is the gas constant ( @xmath74 8.314 j mol@xmath60 k@xmath60 ) , consistent with the dulong - petit law .
the electronic specific heat @xmath75 at zero field ( inset of fig .
4(b ) ) is obtained by subtracting the lattice contribution from the total specific heat .
the extracted electronic specific heat jump at @xmath1 ( @xmath76 6.06 ) is much larger than the weakly coupled bcs value 1.43 , indicating the strongly coupled superconductivity in kbi@xmath0.@xcite the thermodynamic critical field @xmath77 can be obtained by integrating the differences of specific heats and the specific heats divided by temperature between the zero - field superconducting ( s ) and normal ( n ) states ( @xmath33 207 oe ) ( free energy analysis),@xcite
@xmath78 @xmath79 dt'\ ] ] @xmath80 where @xmath81 , @xmath82 , and @xmath83 are the differences of free energy , internal energy , and entropy between zero - field superconducting and normal states .
the relationship between the @xmath1 and critical fields determined from the magnetization , resistivity , specific heat measurements and free energy analysis is summarized in the @xmath84 phase diagram ( fig .
it can be seen that all of data are almost on one line , confirming the consistency of critical fields from different measuring methods and undoubtedly indicating that all of these critical fields are thermodynamic critical field @xmath77 .
the dot line gives the free energy fitting that using the bcs temperature dependence @xmath85 $ ] .
this gives the zero - temperature @xmath86 234.3(3 ) oe and @xmath1 = 3.561(1 ) k , close to those values obtained from different experimental methods .
phase diagram for kbi@xmath0 determined from @xmath87 ( squares ) , @xmath88 ( triangles ) , and @xmath89 ( rhombuses ) at various fields , @xmath39 at various temperatures ( circles ) as well as that calculated from free energy analysis from zero - field specific heat data ( stars ) .
the dot line is fitted by using @xmath85 $ ] . ] .
6 shows the calculated density of state ( dos ) of kbi@xmath0 .
the shape of dos is somewhat different from previous result,@xcite because the soc which is significant for bismuth is considered in present calculation .
the finite dos at fermi energy level ( @xmath90 ) indicates the metallic ground state , consistent with the experimental results .
the dos near @xmath90 is mainly contributed by bi-@xmath91 state and the contribution of k is negligible , indicating that k atoms transfer almost all of valence electron to the bi atoms , i.e. the valence state of k is + 1 and the three - dimensional network of bi can be thought as polyanion.@xcite .
black , red and blue lines correspond to the total dos , partial dos of bi-@xmath91 state and partial dos of k-@xmath92 state , respectively .
the dotted line represents the fermi energy level , @xmath90 . ] .physical parameters of kbi@xmath0 in superconducting and normal states .
[ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] [ tablekey copy(1 ) ] the calculated total dos at @xmath90 is 1.77 states / ev - f.u . and it gives the carrier density @xmath93 = 6.33@xmath1310@xmath94 @xmath95 using the free - electron model .
the calculated fermi wave vector @xmath96 equals 12.33 nm@xmath60 using the formula @xmath97 .
then , the effective electron mass can be determined as @xmath98 0.312 @xmath99 , where
@xmath100 is boltzmann constant and @xmath99 is the free electron mass . with
derived @xmath101 and @xmath93 , the london penetration depth is calculated as @xmath102 11.8 nm.@xcite meanwhile the coherence length is determined by using the bcs relation @xmath103,@xcite which gives @xmath104 1.76 @xmath105 m .
assuming a simple model of a spherical fermi surface ( @xmath106 1 ) , the mean free path @xmath107 is estimated as @xmath108 13.9 nm , where @xmath109 is the low - temperature normal state resistivity ( 5.76 @xmath110 cm at 4 k).@xcite it clearly indicates that the electronic mean free path is considerably smaller than the bcs coherence length ( @xmath111 0.008 ) , suggesting that kbi@xmath0 can be classified as a sc in the dirty limit . in the dirty limit , the gl parameter @xmath112 0.611,@xcite which is smaller than @xmath113 , further confirming that kbi@xmath0 is a type - i sc .
moreover , the zero - temperature ginzburg - landau ( gl ) coherence length in the dirty limit @xmath114 can be obtained from the relations @xmath115 .@xcite it gives @xmath116 133.7 nm .
according to the definition of @xmath117 , the derived zero - temperature gl penetration depth @xmath118 is 81.7 nm .
the superconducting and thermodynamic parameters of kbi@xmath0 are summarized in table i.
in summary , the single crystal of kbi@xmath0 has been grown from bi flux successfully .
resistivity , magnetization and specific heat measurements indicate that kbi@xmath0 shows a bulk superconductivity with @xmath1 = 3.573 k. further analysis of experimental results indicate that kbi@xmath0 is type - i bcs sc in the dirty limit with an intermediate or strong coupling strength .
the thermodynamic critical field is 234.3(3 ) oe and the calculated gl parameter @xmath2 0.611 .
this study not only deepens our understanding on type - i superconductivity but will also stimulate further work on discovering other type - i scs in binary or ternary compounds .
this work was supported by the ministry of science and technology of china ( 973 project : 2012cb921701 ) , the fundamental research funds for the central universities , and the research funds of renmin university of china ( ruc ) ( 15xnlf06 and 14xnlq03 ) , and the national natural science foundation of china ( grant no .
11574394 ) .
computational resources have been provided by the plhpc at ruc .
a. n. kuznetsov and t. f. fssler , z. anorg .
chem . * 628 * , 2537 ( 2002 ) .
a. aharoni , j. appli . phys . * 83 * , 3432 ( 1998 ) . v. h. tran , z. bukowski , p. wisniewski , l. m. tran , and a. j. zaleski , j. phys . : condens .
matter * 25 * , 155701 ( 2013 ) . | we report the detailed transport , magnetic , thermodynamic properties and theoretical calculation of kbi@xmath0 single crystals in superconducting and normal states .
kbi@xmath0 shows metallic behavior at normal state and enters superconducting state below @xmath1 = 3.573 k. moreover , kbi@xmath0 exhibits low critical fields in all of measurements , field - induced crossover from second to first - order phase transition in specific heat measurement , typical magnetization isotherms of type - i superconductors , and small ginzburg - landau parameter @xmath2 0.611 .
these results clearly indicate that kbi@xmath0 is a type - i superconductor with thermodynamic critical field @xmath3 234.3(3 ) oe . |
entanglement for two - qubit systems has been investigated extensively in many directions . quantifying and manipulating entanglement for systems of small dimension
have been covered a in wide area of research @xcite . however , there are a lot of applications that have been implemented on quantum information and computations by using these systems theoretically and experientially .
quantum teleportation , coding @xcite and cryptography represent the more important applications of entanglement . to implement these application with high efficiency , maximum entangled states ( mes )
are needed . despite it
is possible to generate maximum entangled states , they turn into partially entangled states ( pes ) due to their interaction with their own environments @xcite .
therefore , another process is needed to recover entanglement , called quantum purification , which requires infinite number of pairs to get one mes @xcite .
so , it is important to investigate the behavior of entanglement in the presences of different types of noise .
there are some classes of noise states whose teleportation fidelity can be enhanced if one of the two qubits subject to dissipative interaction with the environment via amplitude damping channel @xcite .
recently , montealegre et .
@xcite , have investigated the effect of different types of noise channels on the one - norm geometric discord and showed that the effect of the noise channel can be frozen .
metwally showed that the phenomena of sudden single and double changes and the sudden death of entanglement are reported for identical and non - identical noise @xcite .
also , the behavior of a multi - qubit system in the presence of the amplitude damping channel is investigated in @xcite , where the possibility of performing quantum teleportation with decohered states was discussed .
this motivates us to investigate two classes of multi - qubit systems , the ghz and the w - states , where the effect of the correlated or non - correlated noisy depolarized channel is considered .
the paper is organized as follows : in sec .
@xmath4 , we describe the initial state and their evolution in the presence of depolarized damping channel .
the behavior of entanglement for the ghz state is discussed in sec .
@xmath5 , while sec .
@xmath6 , is devoted for the w - state .
finally we conclude our results in sec . @xmath7 .
assume that a source generates a three - qubit state in the form of ghz or w - state defined as : @xmath8 during the transmission from the source to the users , alice , bob , and charlie , it is assumed that alice s and bob s qubits passes through the depolarized channel @xcite .
this channel transforms the single qubit into a completely mixed state with probability @xmath9 and leaves it untouched with probability @xmath10 . for a single qubit noise ,
the operators are given by @xmath11 where @xmath12 and @xmath13 are the identity and pauli operators for the three qubits respectively .
in computational basis , these operators can be written as : @xmath14 we consider that the suggested noise model is of one , two and three sides noisy channel type . for two and three sides noise , we consider that the noise could be correlated or non - correlated @xcite . for correlated noise , we mean that the qubit is affected by the same noise at the same time , while for non - correlated noise , the qubits may be affected by different noises at the same time .
the three - qubit system under the depolarizing noise acting on the first qubit of the quantum state @xmath15 is given by @xmath16 where @xmath17 and @xmath18 is one of the initial states ( 1 ) .
however , if the first two qubits are affected by the depolarized channel , then the initial state @xmath15 evolves as @xmath19 for correlated noise , while for non - correlated noise , the final state can be written as @xmath20 @xmath21 finally , if the three qubits are affect by the depolarizing channel ( 2 ) , then for correlated noise , the final state of the initial three qubits ( 1 ) evolves as , @xmath22 while for the non - correlated noise it is given by @xmath23 where @xmath24 and @xmath25 when @xmath26 stands for the final states for correlated and non - correlated noise if @xmath4 or @xmath5 qubits are affected , respectively . in the following
we shall investigate the dynamics of entanglement which is contained in the initial states ( 1 ) , where we investigate all the possible noises as described above .
for this aim , we use the tripartite negativity as a measure of entanglement .
this measure states that , if @xmath27 represents a tripartite state , then the negativity is defined as @xcite , @xmath28 where @xmath29 , @xmath30 are the negative eigenvalues of the partial transpose of the state @xmath31 with respect to the qubit @xmath32
correlated noise : + the initial three - qubit state @xmath33 , @xmath34 is given from ( 1 ) under the depolarizing noise on alice s qubit is given by @xmath35 however , if we assume that the first two qubits subject to correlated noise then the final state ( 5 ) is given by + @xmath36 + finally , if it is assumed that all the three qubits are subject to a correlated noise , then @xmath37 can be written as + of the ghz state under depolarize channel .
the solid and the dot curves for the correlated and non - correlated noise respectively noise.,title="fig : " ] ( -310,90)@xmath38 ( -145,-1)@xmath9 + @xmath39 + in fig.(1 ) , we investigate the behavior of entanglement for the three final states @xmath40 .
the general behavior displays that the entanglement decays as @xmath9 increases .
however , these decays depends on the number of the affected qubits with the depolarized channel . for the state @xmath41 , where only one qubit is depolarized , the entanglement decays smoothly , while for @xmath42 , when the strength of noise is maximum , the entanglement does nt vanish . for the state @xmath43 , where two qubits are depolarized
, the decay rate is larger than that depicted for the state @xmath41 .
although the entanglement vanishes for some specific value of @xmath44 , it re - increases again for farther values of @xmath9 . finally , the decay rate of entanglement is the largest one if the three qubits are depolarized as shown for the state @xmath45 . for this state
the entanglement decays gradually until vanishes completely at @xmath46 .
non - correlated noise : + now , we assume that two or three qubits are depolarized in non - correlated way .
if only two particle are depolarized then the final sate ( 6 ) can be written as + @xmath47 + where @xmath48 + , where it is assumed only two qubits are depolarized , while the dot curve when the 3-qubits are depolarized .
, title="fig : " ] ( -310,90)@xmath38 ( -145,-1)@xmath9 + however , if we assume that the three qubits pass through the depolarized channel ( 2 ) , then the final state ( 8) ( for non - correlated noise ) is given by + @xmath49 + where
@xmath50 , \nonumber\\ \mathcal{b}_1&=&\frac{1}{2}\left[(1-p)^3-p(1-p)+\frac{5}{9}p^2(1-p)-\frac{p^3}{27}\right ] , \nonumber\\ \mathcal{b}_3&=&\frac{1}{3}\left[p(1-p)^2+\frac{7p^2}{3}(1-p)+\frac{p^3}{3}\right ] , \nonumber\\ \mathcal{b}_4&=&\frac{p^2}{9}(1-p),\quad \mathcal{b}_5=\frac{p}{2}(1-p)^2+\frac{p^2}{9}(1-p ) , \nonumber\\ \mathcal{b}_6&=&\frac{1}{3}\left[p(1-p)^2+\frac{5p^2}{3}(1-p)+\frac{p^3}{3}\right].\end{aligned}\ ] ] + fig.(2 ) , shows the behavior of entanglement which is contained in the final sates @xmath51 and @xmath51 , where it is assumed that the noise is non - correlated . the general behavior shows that the entanglement decays hastily as the channel parameter @xmath9 increases .
the decay rate of entanglement depends on the numbers of depolarized qubits . however , the decay rate of entanglement when the three qubit subject to the noise is larger than that depicted when only two qubits are depolarized .
the phenomena of the sudden - death of entanglement appear clearly when the particles are affected by non - correlated noise . + from fig.(1 ) and fig.(2 ) , we can conclude that the decay of entanglement depends on the type of the noise ( correlated or non - correlated ) and the number of qubits which are subject to the noise . for correlated noise
, the entanglement decays smoothly to reach its minimum bounds .
these bounds depend on the number of polarized qubits , where they are non - zero for fewer number of polarized particles and completely vanish as the number of depolarized qubit increases
. however , the phenomena of the sudden death appears in the presences of non - correlated noise and as the number of polarized qubit increases , the entanglement dies for smaller values of the channel parameter @xmath9 .
now we assume that the user share a three - qubits - state of w - type @xmath52 , where @xmath53 is defined in ( 1 ) .
if we allow only one qubit to pass through the depolarized channel ( 3 ) , then the final state is given by @xmath54 however , if we assume that only two qubits are forced to pass through the noise channel ( 3 ) and that the noise is correlated , then the final state is given by , @xmath55 where @xmath56 when one , two and three qubits pass through the depolarized channel(3)in a correlated way , respectively . , title="fig : " ] ( -310,90)@xmath38 ( -145,-1)@xmath9 however , if the three qubits pass through the non - correlated noise then the final state is given by @xmath57 where @xmath58 fig.(3 ) , describes the behavior of entanglement for the final states ( 16),(17 ) and ( 19 ) , where we assume that , one , two and three qubits are subjected to the noise channel ( 3 ) , respectively .
the general behavior shows that the entanglement decays as the channel parameter @xmath9 increases .
the decay rate depends on the number of depolarized qubits .
it is clear that , if only one qubit is depolarized , then the entanglement decays smoothly to reach its minimum non - zero value .
these minimum values decrease as the number of depolarized qubits increases . and
@xmath59 , respectively . , title="fig : " ] ( -330,90)@xmath60 ( -145,-1)@xmath9 now , we assume that the effect of the is non - correlated .
if only two qubits are passing through the depolarized channel , then the noise output state is given by , @xmath61 where @xmath62 however , we assume that the three qubits of the w - states passing through the noise channel . in this case
the output state is given by @xmath63 where @xmath64 the behavior of entanglement for the output states ( 21 ) and ( 23 ) is described in fig.(4 ) .
in general , the entanglement decreases as the channel s strength increases .
however the decreasing rate is smaller than that depicted for correlated noise ( see fig.3 ) .
moreover , when all three qubits pass through the depolarized channel , the entanglement decays faster than that depicted for two depolarized qubits .
it is clear that for larger values of the strength parameter @xmath9 , the entanglement increases .
however , the upper bound of entanglement is smaller than that in the case of two - depolarized qubit . on the other hand , comparing the behavior of this entanglement with that for the ghz state
, we can see that the w - state is more robust than the ghz state , where for the ghz state the entanglement completely vanishes in a sudden way for smaller values of the channel strength .
in this contribution , we investigate the behavior of entanglement for two classes of multi - qubit states ; the ghz and w - states passing trough the depolarized channel . in our treatment , we assume that the effect of the noise channel includes two possibilities . in the first possibility
we assume that the qubits are subject to the same noise at the same time ( correlated noise ) .
the second possibility , which is called non - correlated noise , the particles are subjected to different noises at the same time .
analytical expressions are obtained for the final output states . for the correlated noise
, we calculate the final output states when one , two or three qubits are forced to pass through the depolarized channel .
however , for the non - correlated noise , we consider the case of two or three qubits passing through the depolarized channel .
the dynamics of the survival amount of entanglement is investigated against the channel strength .
the general behavior shows that the entanglement decreases as the channel parameter increases .
however , the decreasing rate depends on the number of depolarized qubits , the type of the noise and the type of the initial state . for _
correlated _ noise , the decay rate of entanglement decreases as the number of polarized qubits increases .
although the decay rate for the w - state is larger than that depicted for the ghz state , the entanglement completely vanishes when the three qubits are depolarized for the ghz state . on the other hand , for the _
non - correlated _ noise , the phenomena of the sudden death of entanglement appears when the ghz s qubits are depolarized .
if only two qubit qubits are depolarized , the entanglement vanishes at larger values of the channel parameter .
however , for the w - state , the phenomena of the entanglement sudden death does nt appear and the lower bound of entanglement is large .
_ in conclusion _ , the robustness or fragility of the initial states which pass through a depolarized channel depends on the type of the noise : correlated or non - correlated . in general ghz
is more fragile than w - state .
the phenomena of entanglement sudden death appears only for non - correlated noise . c. h. bennett , g. brassard , s. popescu , b. schumacher , j. a. smolin , and w. k. wootters , phys .
76 , 722 ( 1996 ) ; d. deutsch , a. ekert , r. jozsa , c. macchiavello , s. popescu , and a. sanpera , phys .
77 , 2818 ( 1996 ) ; m. zukowski , a. zeilinger , m. a. horne , and a. k. ekert , phys . rev .
71 , 4287 ( 1993 ) ; c. h. bennett , g. brassard , c. creepeau , r. jozsa , a. peres , and w. k. wootters , phys . rev . lett
70 , 1895 ( 1993 ) ; n. metwally , phys . rev . a 66 054302 ( 2002 ) ; n. metwally and a .- s .
f. obada , phys . lett .
a , 352 , 45 ( 2006 ) .
n. metwally,``dynamics of encrypted information in superconducting qubits with the presence of imperfect operations '' , j. opt .
* 389 - 396 ( 2012 ) .
m. a. nielsen and i. l. chuang , quantum computation and quantum information ( cambridge university press , cambridge , 2000 ) .
obada , n. araf , and n. metwally``quantum communication in the presence of noisy local environmants '' journal of the association of arab universities for basic and applied sciences ( jaabas ) * 12 * , 5560 ( 2012 ) . | some properties of multi - qubit systems interacting with noisy environment is discussed .
the amount of the survival entanglement is quantified for the ghz and w - states .
it is shown that the entanglement decay depends on the noise type ( correlated or non - correlated ) , number of interacted qubits with the environment and the initial state which passes through this noisy environment . in general
, the ghz is more fragile than the w - state .
the phenomena of entanglement sudden death appears only for non - correlated noise .
dynamics of multi - qubit systems in noisy environment + n. metwally@xmath0 and a. almannaei @xmath1 + @xmath2math . dept . , college of science , university of bahrain , kingdom of bahrain + @xmath3 math .
dept .
, faculty of science , aswan university , sahari 81528 , aswan , egypt |
recently , renewal of interest has been evoked to the fractional angular momentum ( fam ) in two - dimensional ( 2d ) space in relation with the correspondence between quantum mechanical wave - functions and classical periodic - orbits@xcite .
the fam is only possible in 2d multiply - connected - space , since in three- or higher - dimensional space , the angular momentum being integer or half - integer is completely determined by the commutation relation of angular - momentum operators . in 2d space ,
the angular momentum operator has only one component , which does not give rise to a unique determination of the angular - momentum eigenvalue , and the common belief of integer - quantization is based on the @xmath1-periodic boundary - condition which , however , is not justified .
wilczek in his pioneering work proposed for the first time a quantum mechanical model consisting of a charged particle and magnetic flux in 2d space to demonstrated the fractional eigenvalues of angular momentum known as anyon @xcite . in the existence of gauge field , however , we have both the kinetic angular momentum ( kam ) and the canonical angular momentum ( cam ) , which are different , because of the velocity - dependent forces @xcite .
it is true that `` the generator of rotation should be the cam and is prescribed by noether s theorem as a conserved quantity''@xcite , which is gauge dependent and integer - quantized @xcite . while the kam is gauge invariant dynamic quantity , generally fractional because of the gauge field @xcite .
although the consistence of fractional cam with the aharonov - bohm ( ab ) phase - interference was shown long ago @xcite it remains a long standing open question whether or not the fractional cam plays a role in quantum physics .
it is worthwhile to remark that the gauge potential of an ab flux can only shift the angular momentum eigenvalues by a common fractional number but can not change the integer - quantization rule , namely the eigenvalue - space of angular momentum is still @xmath0 @xcite .
the existence of fam in a wide class of 2d central potentials without gauge field has been discussed more recently @xcite by the localization of coherent wave functions on classical orbits , which imposes a special boundary condition leading to the unusual angular phase of wave functions . following the interesting studies of ref .
@xcite , which provide an exactly solvable model both quantum mechanically and classically , we consider a charged particle in the gauge field of magnetic flux - string perpendicular to the 2d plane with the central scalar - potentials in addition and study the non - integer quantization of angular momentum ( niqam ) in terms of quantum - classical correspondence .
since the coherent state is constructed by the superposition of angular momentum eigenstates and thus the probability clouds have to possess the same rotational symmetry as that of classical orbits , which is not necessarily @xmath1-periodic . as a consequence the niqam with the level - space being greater or less than @xmath0 appears naturally along with the correspondence principle .
by the explicit calculation it is shown that only the shift of cam eigenvalues by the gauge field , which gives rise to a common topological phase for all wave functions , is in agreement with the correspondence .
the validity of wave functions with a topological phase is further confirmed by the expectation values of kam operator , which coincide with the classical values of kam @xmath2 . the long standing problem whether or not the fam is related to the cam is resolved and the quantum mechanical model of anyon proposed by wilczek @xcite emerges as a special case of @xmath3 with @xmath1-rotational - symmetry of classical orbits .
more specifically , we refer here to the particular zero - energy states , which can be obtained analytically for both classical- and quantum - solutions @xcite . on the other hand the zero - energy states are of importance in various fields such as * * the cold - atom collisions@xcite , the construction of vortex lattices @xcite , and quantum cosmology @xcite .
we consider a charged particle of charge @xmath4 and mass @xmath5 in a gauge field of infinitely long magnetic flux - line of total flux @xmath6 located at the origin of 2d space and central scalar - potential of form @xcite @xmath7 where @xmath8 and @xmath9 . outside the flux - line ( @xmath10 ) ,
the lorentz force on the charged particle is always zero because of the vanishing magnetic field @xmath11 , while the vector potential in the polar coordinate is seen to be @xmath12 with @xmath13 being the unit - vector of angular direction . in order to establish the quantum - classical correspondence , we ought to evoke the canonical variables . in the polar coordinates @xmath14 , the lagrangian of the system is seen to be @xmath15-ea_{0}(r)+l_{wz}%\ ] ] where @xmath16 is so - called the wess - zumino topological interaction - term with the parameter @xmath17 being the dimensionless magnetic flux in the quantum - unit @xmath18 .
the wess - zumino term does not affect the equation of motion but the initial value of angular momentum .
canonical momentums corresponding to the coordinate variables @xmath19 , @xmath20 are defined by@xmath21 here @xmath22 is cam , while @xmath23 is the kam .
then the hamiltonian is @xmath24 from canonical equations we find that both the cam and kam are conserved quantity @xmath25 for the case of zero - energy and nonvanishing initial kam @xmath26 we have @xmath27-e\frac{\gamma_{\upsilon}}{r^{\upsilon}}=0.\ ] ] we assume an initial - value that @xmath28 where @xmath29 is an arbitrary dimensionless - quantity .
thus in the considered case only the cam is shifted by the flux . introducing a dimensionless variable@xmath30^{1/\mu}}%\ ] ]
we obtain the equation of particle trajectories such as @xcite @xmath31 the general solution of eq.(8 ) is given in ref.@xcite @xmath32 \label{9}%\ ] ] using atomic unit @xmath33 with @xmath34 , which is the dimension of length , we set @xmath35 in the numerical evaluation .
the classical orbits with the initial * * * * angle setting to zero @xmath36 are shown in figs . 1 - 5 ( solid green curves ) , which depend on the initial angular momentum @xmath29 only for given scalar potential . * * in general , we have closed orbits ( figs . 1 - 4 ) for @xmath37 @xmath38 , while open trajectories for @xmath39 seen from fig.5 . from eq.(9 ) it is obvious that the rotation symmetry of classical orbits depends on the power - index @xmath40 such that the orbits are invariant under a rotation - angle @xmath41 , and @xmath1-symmetry holds only for @xmath42 .
we demonstrate in this paper that the angular momentum quantization can be determined only by the rotation symmetry of classical orbits based on the requirement of quantum - classical correspondence .
the rotation angles of symmetry for closed orbits in figs.1 - 4 are 2@xmath43 , @xmath44 , @xmath45 , and @xmath46 respectively .
in the polar coordinates @xmath47 , the zero - energy schr@xmath48dinger equation @xmath49\psi+ea_{0}\psi=0\ ] ] becomes @xmath50@xmath51 by the separation of variables @xmath52 .
the eigenvalue solution of angular part eq .
( 10 ) is @xmath53 with @xmath54 being the normalization constant and the cam eigenvalue @xmath55 to be determined . for the usual requirement of 2@xmath43-periodic boundary - condition , @xmath56
, one can obtain the integer angular momentum quantization and the normalization constant @xmath57 .
however the 2@xmath43-periodic boundary - condition is not justified in the @xmath58-space , although the potential @xmath59 is symmetric under rotation .
it has been demonstrated that a macroscopic quantum state , here su(2 ) coherent state @xcite can be constructed with probability density of wave functions well localized on the classical orbits for the quantum - classical correspondence , which results in a special boundary condition of the angular momentum eigenstates such that rotational period of wave function is not necessarily 2@xmath43 but should be the same as that of classical orbits @xmath60@xcite .
thus the cam eigenvalue is no longer integer but should be set as @xmath61 where @xmath62 is a integer .
the angular momentum now is quantized with an eigenvalue - space@xmath63 the integer - quantization is only a special case of @xmath42 .
the normalization constant becomes@xmath64 the radial equation is@xmath65 in the choice of eq .
( 12 ) for @xmath55 , the @xmath62-th kam eigenvalue is shifted by the flux number that @xmath66 indicating the dynamic effect of the gauge potential in contradiction with the classical solution in which gauge field does not apply a torque on the particle .
we do have the other choice of @xmath55 such that@xmath67 while the kam eigenvalue @xmath68 does not depend on the flux in consistence with classical solution .
the common - phase @xmath69 factor called the topological phase @xcite does not change the angular momentum quantization eq.(13 ) nor the normalization constant eq.(14 ) since whole eigenfunctions have the same additional angular phase .
it is the main goal of the present paper that only the choice of cam eq.(17 ) gives rise to the exact quantum - classical correspondence .
* * we demonstrate in this section that only the cam eigenvalues eq.(17 ) consist with the quantum - classical correspondence , such that the probability densities of coherent wave functions are well localized on classical orbits . introducing the dimensionless radius @xmath70 and @xmath71 , where @xmath72 is a quantity with dimension of length , we obtain @xmath73 where the parameter @xmath74 is defined by@xmath75 and the corresponding kam eigenvalues @xmath76 given by eq.(18 ) do not depend on the flux number .
square - integrable solutions of eq .
( 19 ) are found in terms of bessel functions of the first kind @xcite@xmath77 with the normalization constant given by @xmath78}{\gamma(1/2 + 1/\mu)\gamma\lbrack l_{n}^{k}/\left\vert \mu\right\vert -1/\mu ] } } % \ ] ] where we have set the parameter @xmath79 and thus @xmath80 according to the definition of @xmath74 .
if the following conditions are satisfied:@xmath81 * * we have bound states corresponding to the classical closed - orbits as demonstrated in ref @xcite .
for the zero - energy @xmath82 there exist classical solutions of closed - orbits with any non - zero angular momentum when @xmath83 , while the condition of bound states in quantum mechanics is @xmath84**. * * it has been shown that * * @xcite * * the normalizable quantum solutions can be classified to two classes : ( 1 ) bound states ( @xmath84 ) for @xmath37 @xmath38 corresponding to the classical closed - orbits ( figs .
1 - 4 , solid green curves ) ( 2 ) scattering states ( @xmath85 for @xmath39 @xmath86which are normalizable , corresponding to classical open - orbits ( fig . 5 , solid green curves ) .
at the region @xmath87 the solutions of wave functions are not square - integrable @xcite , which we do not discuss in the present paper .
the complete eigenfunctions of the schr@xmath48dinger equation can be written in the explicit form as @xcite@xmath88 following refs.@xcite we construct * * the stationary su(2 ) coherent - state for the central - scalar - potentials in the standard way as @xcite@xmath89 with @xmath90 and @xmath76 given in eqs.(17 ) , ( 18 ) respectively .
an additional angular momentum number @xmath91 with @xmath92 being a integer - parameter is introduced to adjust the probability density , which does not change the rotational symmetry of wave - function density but the spatial expansion of it . in the following section
we will see that the parameter @xmath92 is not arbitrary but is related to the classical kam @xmath2 such that the expectation value of kam operator coincides with the classical value .
* * the wave functions are normalized to unity in the angular range of @xmath20 being from @xmath93 to @xmath43**. * * from eq .
( 24 ) we can see that the common topological phase - factor resulted by the gauge potential does not affect the probability density of wave functions in agreement with the classical assumption of no torque * * applied on the charged particle . in figs . 1 - 5
the practical values of parameters are given by @xmath94 , @xmath95 @xmath96 , @xmath97 @xmath98 , @xmath99 , and @xmath100 @xmath101 respectively .
in all figures the total number of eigenfunctions is @xmath102 .
the density @xmath103 of coherent - state wave function possesses the same rotation symmetry of the classical orbits . figs . 1 - 5 ( a )
show that the probability densities of coherent wave functions are well localized on the classical orbits in each case indicating the exact quantum - classical correspondence .
we thus obtain the angular - momentum quantizations with the level - spaces @xmath104 and @xmath105 ( of @xmath0 ) respectively .
the cam spectrum is shifted by the corresponding flux number @xmath106 . * * and the closed classical - orbit ( solid green curve ) for @xmath107 .
( a ) shift of cam ; ( b ) shift of kam.,width=249,height=364 ] .,width=268,height=384 ] .,width=249,height=364 ] .,width=249,height=364 ] and the open classical - orbit ( solid green curve ) for @xmath108(a ) shift of cam ( b ) shift of kam.,width=249,height=364 ] besides the ab interference @xcite the fractional spins have many important physical applications , for example , two - vortex system in a superfluid film @xcite , necklace - ring @xcite , chiral - wave superconductor@xcite or quantum billiard@xcite inside a boundary defined by the wedge - shaped section of a circle .
_ shift of the kam eigenvalues by the gauge potential and break down of the correspondence : _ we , of course , can have the other choice of cam eigenvalues i.e. eq.(12 ) , which does not depend on the flux and as a consequence the spectrum of kam is shifted by the gauge potential as shown in eq.(16 ) . in this case
we do not have the topological phase of the wave functions but have the flux - dependent kam eigenvalues instead .
the quantization of angular momentum , namely the level - space of eigenvalues , is not changed by the gauge potential . replacing the kam eigenvalues @xmath76 in eqs.(23,24 ) by eq.(16 ) the corresponding probability densities of coherent states
are depicted in figs.1 - 5 ( b ) , from which we see that probability densities are no longer localized no the classical orbits indicating additional torques applied on the particle in contradiction with the classical assumption .
_ expectation values of angular momentum _
: we now calculate the expectation values of kam operator@xmath109\ ] ] in the su(2 ) coherent - state to find the explicit relation between the adjusting - parameter @xmath110 and the initial classical - kam @xmath111 in order to confirm further the quantum - classical correspondence .
the average of kam operator * * in the su(2 ) coherent state eq.(24 ) with the topological phase is evaluated as @xmath112\ ] ] substituting the corresponding parameter values of @xmath40 , @xmath92 , and @xmath113 into the above formula one can see the exact agreement with the initial - values of classical kam such that@xmath114 for @xmath29= @xmath115respectively .
however , for the cam eigenvalue - choice of eq.(12 ) , namely the shift of kam eigenvalue by the gauge potential * * , * * the average of kam operator becomes@xmath116\ ] ] which , of course , disagrees with classical values .
in summary , the niqam in two - dimensional space is demonstrated in terms of quantum - classical correspondence with an exactly solvable model , such that the probability clouds of macroscopic quantum states , here the su(2 ) coherent wave functions , are well localized on the classical orbits .
the eigenfunctions of angular momentum have to possess the same rotational symmetry of classical orbits , in which the rotation - period can be greater or less than @xmath1 depending on the power - index @xmath40 of the central potential only . as a consequence
the level - space of angular momentum spectrum is less or greater than @xmath0 and the integer - quantization ( level - space @xmath0 ) is possible only if @xmath3 .
the gauge potential of ab - flux does not affect the angular momentum quantization but can shift the spectrum of angular momentum by a common value . by explicit calculations
it is shown that the quantum - classical correspondence results in the unambiguous determination of cam eigenvalues with a common topological - phase in the wave functions , the probability densities of which coincide with the classical orbits for any power - index of potentials .
the quantum mechanical model of anyon proposed by wilczek @xcite and latter clarified as the fractional cam in ref.@xcite becomes a special case of the present model with @xmath3 .
this work was supported by national nature science foundation of china ( grant no.10775091 ) . | we study the quantum - classical correspondence in terms of coherent wave functions of a charged particle in two - dimensional central - scalar - potentials as well as the gauge field of a magnetic flux in the sense that the probability clouds of wave functions are well localized on classical orbits . for both closed and open classical orbits , the non - integer angular - momentum quantization with the level - space of angular momentum being greater or less than @xmath0
is determined uniquely by the same rotational symmetry of classical orbits and probability clouds of coherent wave functions , which is not necessarily @xmath1-periodic .
the gauge potential of a magnetic flux impenetrable to the particle can not change the quantization rule but is able to shift the spectrum of canonical angular momentum by a flux - dependent value , which results in a common topological phase for all wave functions in the given model .
the quantum mechanical model of anyon proposed by wilczek ( phys .
rev .
lette .
48 , 1144 ) becomes a special case of the arbitrary - quantization . |
supersymmetric models where the gravitino ( @xmath0 ) is the lightest supersymmetric particle have been considered in the literature for some time @xcite .
more recently they have enjoyed a resurgence motivated by new models of dynamical supersymmetry breaking @xcite and by possible explanations @xcite of the puzzling cdf @xmath14 event @xcite .
most of the recent effort has been devoted to studying the new signals for supersymmetry via electroweak - interaction processes , that accompany such scenarios by virtue of the newly allowed @xmath15 decay ( _ i.e. _ , diphoton signals from @xmath16 and analogous signals in @xmath17 production of a variety of supersymmetric channels ) @xcite , or via direct gravitino production ( _ i.e. _ , single photon signals from @xmath18 @xcite or @xmath19 @xcite ) .
much less emphasis has been placed on strong - interaction processes at hadron colliders @xcite .
an important part of this phenomenological effort should be directed at obtaining direct experimental information on the gravitino mass . however , this information will not come from the kinematical effects of such a very light particle , but instead from the dynamical effect that each interaction vertex involving gravitinos is inversely proportional to the gravitino mass .
lower bounds on the gravitino mass from single - photon searches at lep already exist , but are limited by kinematics at lep 1 @xcite and by dynamics at lep 2 @xcite .
limits on the gravitino mass based on monojet and multijet final states at the tevatron require that the scalar partners of the goldstino ( @xmath20 and @xmath21 ) in spontaneously broken susy models are very light@xcite .
limits also exist from astrophysical and cosmological considerations@xcite , although , as noted in ref.@xcite , these limits are competitive with collider limits only if one again assumes that the scalar partners of the goldstino are very light .
more stringent constraints have been obtained in ref.@xcite .
the purpose of this paper is to examine another manifestation of light gravitinos , in the context of hadron colliders . )
collider , although the discussion applies with little modification to the case of the lhc ( @xmath22 ) collider too . ] two gravitino - mass dependent processes involving the strong interactions come to mind : @xmath23 the first process ( @xmath24 ) proceeds via the usual supersymmetric qcd diagrams , but receives in addition contributions from new gravitation - induced diagrams involving exchange of gravitinos in the @xmath25- and @xmath26-channels @xcite . for sufficiently light gravitino masses ( @xmath27 ) ,
the gravitation - induced contributions dominate the supersymmetric - qcd ones , otherwise the reverse is true .
this process is also kinematically constrained at the tevatron to @xmath28 .
the second process ( @xmath29 ) only receives contributions from gravitation - induced diagrams and is therefore only relevant for sufficiently light gravitino masses .
however , its kinematical reach at the tevatron is much greater ( @xmath30 ) .
we find that for sufficiently light gravitinos ( @xmath31 ) ev , the second process has a larger cross section even within the kinematical reach of the first process ( @xmath32 ) .
it is also important to consider the gluino decay modes . for sufficiently light gravitinos ( @xmath27 @xcite )
these are dominated by @xmath33 ; otherwise they proceed in complicated ways of which @xmath34 is an example .
we then obtain a partitioning of the @xmath35 ) space into four regions of low / high " values of the parameters , roughly delimited by @xmath36 and @xmath37 .
note that in each of these regions we expect a signal dominated by a different number of jets .
these facts are summarized in the following sketch .
@xmath38 the sparse literature on this subject contains explicit expressions only for the @xmath39 initiated sub - processes ( @xmath40 ) @xcite and phenomenological analyses @xcite only for the left half ( _ i.e. _ , low @xmath41 " ) of the above table .
the above comments and processes generally apply to squarks as well , with gluinos replaced by squarks ( @xmath42 ) and gluons replaced by quarks ( @xmath43 ) .
a sample squark decay for heavier gravitinos is @xmath44 . in section 2
we present the cross sections for different channels at a hadron collider giving a monojet signal and an outgoing gravitino . in section 3
we present a discussion of our numerical results . in the appendix we provide a derivation of the interaction lagrangian involving gravitinos , stating clearly the conventions we use .
we include the feynman rules relevant to our calculation and also provide comparisons with others using different conventions .
processes involving gravitinos suffer unitarity violation at high energies due to the non - renormalisability of the supergravity lagrangian@xcite .
to preserve unitarity till energies of the order of a few tev typically requires @xmath45 ev .
hence , we shall take @xmath46 ev as our lower bound on the gravitino mass in what follows .
in @xmath17 collision there are three parton sub - processes giving rise to a single jet along with a gravitino . as we mention above , the jet is initiated by either an outgoing gluino or squark and the undetected gravitino is associated with missing transverse energy .
the different parton sub - processes are @xmath47 , @xmath48 , and @xmath49 .
below we present the differential cross section @xmath50 for each sub - process .
the feynman diagrams for each sub - process are provided in figs .
[ fig : qqbar ] , [ fig : qg ] and [ fig : gg ] respectively . the hadronic cross section @xmath51 is obtained by convoluting the sub - process cross section @xmath52 of partons @xmath53 with parton distribution functions @xmath54 .
( because of the relation between mandelstam variables @xmath55 and @xmath26 , the partonic cross sections below may be expressed differently . ) the differential cross section for the sub - process @xmath47 is given by @xmath56 ^ 2 } + \sum_{j=1,2 } \frac{2u^3(u-\mg^2)}{3[u - m_{\tilde{q_j}}^2]^2}\nonumber\\ & & -\sum_{j=1,2}\frac { 4st^3}{3s[t - m_{\tilde{q_j}}^2 ] } -\sum_{j=1,2}\frac { 4su^3}{3s[u - m_{\tilde{q_j}}^2 ] } \biggr].\label{eq : qqbar}\end{aligned}\ ] ] above @xmath57 , where @xmath58 is newton s gravitational constant . the sums over @xmath59 run over the two squark mass - eigenstates .
left - right squark mixing is not relevant for this channel ( nor are they for the other two channels ) , as explained in appendix c. the total partonic cross section is obtained by integrating the above from @xmath60 to 0 .
we have verified that our above result agrees with the differential cross section obtained in the second reference in ref.@xcite for @xmath61 ( modulo colour factors ) , where @xmath62 represents a photino .
the differential cross section for the sub - process @xmath63 is given by ( @xmath64 represents the two squark mass eigenstates ) @xmath65 ^ 2}\nonumber\\ & & - \frac{2u^2\biggl ( ( u-\msqj^2)(s-\msqj^2 ) + t(\mg^2-\msqj^2)\biggr)}{3[u - m_{\tilde{g}}^2]^2 } -{4st^2u\over 3s^2}\nonumber\\ & & -\frac { 4t^2u(s-\msqj^2)}{3s[t - m_{\tilde{q_j}}^2 ] } + \frac { 4u^2s(s-\msqj^2)}{3s[u - m_{\tilde{g}}^2]}\nonumber\\ & & + \frac { 4stu(s-\msqj^2)}{3s^2 } + \frac { 4t^2u^2}{3[t - m_{\tilde{q_j}}^2][u - m_{\tilde{g}}^2]}\nonumber\\ & & + \frac { 8t^3u}{3s[t - m_{\tilde{q_j}}^2 ] } -\frac { 4stu^2}{3s[u - m_{\tilde{g}}^2 ] } \biggr].\label{qg}\end{aligned}\ ] ]
the above cross section is new and has not been obtained elsewhere in the literature .
the total partonic cross section is obtained by integrating the above from @xmath66 to 0 .
when obtaining the total hadronic cross section for this channel we sum over both squark masses and then multiply the result by a factor of 2 to include the sub - process @xmath67 .
the differential cross section for the sub - process @xmath68 is given by @xmath69 ^ 2 } + ( \mg^2t^2/s ) { t(t-\mg^2)+u(u-\mg^2)\over[t - m_{\tilde{g}}^2]^2}\biggr)\nonumber\\ & & + { 4\over3}\biggl ( { u^3(t+\mgl^2)\over[u - m_{\tilde{g}}^2]^2 } + ( \mg^2u^2/s ) { t(t-\mg^2)+u(u-\mg^2)\over[u - m_{\tilde{g}}^2]^2}\biggr)\nonumber\\ & & + { 8 ( s-\mgl^2 ) st u\over 3s^2 } -{8 ( s-\mgl^2 ) t^2
u\over 3s[t - m_{\tilde{g}}^2]}\nonumber\\ & & -{8 ( s-\mgl^2 ) t u^2\over 3s[u - m_{\tilde{g}}^2 ] } + { 8 ( s-\mgl^2 ) t^2
u^2/s\over 3[t - m_{\tilde{g}}^2][u - m_{\tilde{g}}^2 ] } \nonumber\\ & & -{8 ( s-\mgl^2 ) t^2 u\over 3s[t - m_{\tilde{g}}^2 ] } -{8 ( s-\mgl^2 ) t u^2\over 3s[u - m_{\tilde{g}}^2 ] } \biggr]\label{gg}\end{aligned}\ ] ] the total partonic cross section is obtained by integrating the above from @xmath60 to 0 . the above differential cross section disagrees with the result in ref.@xcite by a factor of @xmath70 . however if one replaces the definition of @xmath71 in ref.@xcite by the ` standard ' definition @xmath72 then our results agree . while obtaining the above differential cross sections one must be careful while summing over polarization states of the incoming gluon(s ) .
we have used @xmath73 where @xmath74 are the momenta of the incoming particles@xcite .
( for example , for @xmath75 , we pick the incoming quark momentum as @xmath76 . ) for @xmath68 one could have alternatively used @xmath77 and included the contribution of ghosts as in ref.@xcite .
the hadronic cross section corresponding to any parton sub - process is given by @xmath78 above @xmath53 run over all partons , valence and sea , that participate in the sub - process .
@xmath79 is the ratio of the parton momentum to the hadron momentum .
@xmath80 , the centre of mass energy of the tevatron , is 1.8 tev .
@xmath81 , @xmath82 for the @xmath83 and @xmath39 channels and @xmath84 for the @xmath85 channel . for the @xmath86 and @xmath39 channels we set the factorisation scale @xmath87 in the parton distribution functions @xmath88 to be the gluino mass , while for the @xmath85 channel we set @xmath87 to be the squark mass .
we set the renormalisation scale @xmath89 equal to @xmath90 and use the world average @xmath91 in the modified minimal subtraction scheme .
the three parton - level cross sections given in the previous section have been integrated over the parton distribution functions as indicated above .
we assume both squark masses are equal for simplicity .
we start with the @xmath39 initiated process : @xmath5 , shown by the dashed lines in fig .
[ fig : crossgg ] , as a function of @xmath41 , for a few choices of the gravitino mass ( the cross section scales with @xmath92 ) . for reference
we also show the gluino pair - production cross section obtained in ref.@xcite ( solid lines ) , which gives the order of magnitude of the traditional supersymmetric signals at hadron colliders . for low gluino masses the latter process dominates while for larger gluino masses the single gluino cross section dominates .
here we disagree with ref.@xcite that the @xmath93 cross section is greater than the cross section for @xmath94 via gluon fusion for @xmath95 ev .
the gluino - gravitino channel may also proceed from a quark anti - quark initial state ( @xmath96 ) .
the hadronic cross section in this case depends on the squark mass in addition to the gluino mass .
these are shown in fig .
[ fig : crossqqbar ] for @xmath97 ev and for various choices of the squark mass ( solid lines ) , and also for the special case of @xmath98 ( dotted line ) . in this figure
we also show ( dashed line ) the additional contribution to this channel discussed above ( _ i.e. _ , from @xmath5 ) .
it is evident that the @xmath3 channel generally dominates over the @xmath5 channel for @xmath99 gev .
this is at variance with the observation in ref.@xcite that gluon fusion is the dominant sub - process .
as noted in ref.@xcite , the @xmath100 cross section via gluon fusion is fairly independent of the gravitino mass for @xmath101 ev ( see fig .
[ fig : crossgg ] ) .
therefore , as the @xmath102 cross section scales with @xmath92 , we may conclude from fig .
[ fig : crossqqbar ] that the latter process dominates over the former for gravitino masses as high as @xmath103 , depending on the gluino ( and squark ) mass .
( note that the results of ref.@xcite that we use for @xmath104 include only the gluon fusion sub - process . )
the last channel to consider is that which originates from the parton - level processes @xmath105 .
the corresponding cross section is shown in figs .
[ fig : crossqg ] and [ fig : crossqgg ] . for squark masses greater than 500 gev and gluino masses
less than 500 gev the @xmath83 channel dominates over the @xmath85 channel .
however for low squark masses the @xmath85 channel dominates . putting all light - gravitino signals together , one sees that these cross sections are higher than the traditional gluino pair - production one for gravitino masses as high as @xmath106 , depending on the gluino ( and squark ) mass . for @xmath107 ,
these cross sections have a kinematical reach in gluino / squark mass about twice as deep as the traditional gluino pair - production one .
this reach decreases quickly with increasing gravitino mass , being surpassed by the traditional process for @xmath108 and higher .
the study of standard model backgrounds to the above processes is beyond the scope of this paper , and in fact has been considered previously in the literature @xcite .
these authors have shown that on imposing suitable cuts , a sizeable signal may be observable over background .
they then use limits on the multijet cross section at the tevatron to exclude certain regions of the @xmath109 parameter space . here
we simply comment that our signal calculations can exceed those in ref .
@xcite , and therefore one could expect an even more detectable signal than previously anticipated .
the task of determining the actual observable signal is best left to the experimentalists .
we hope that our calculations of the total cross sections will help in deciding whether these signals are worth pursuing in earnest .
the work of j.k . and r.r .
has been supported by the world laboratory .
the work of j.l . has been supported in part by doe grant no . de - fg05 - 93-er-40717 and that of d.v.n . by doe grant no .
de - fg05 - 91-er-40633 .
j.k . and r.r .
would like to thank takeo moroi for many useful conversations .
j.k . would like to thank j. bagger for useful clarifications with respect to ref .
we find it necessary to include the exact recipe for obtaining the feynman rules from the relevant terms in supergravity lagrangian , since there are many conventions available in standard references .
problems arise when one uses feynman rules from different sources without properly adapting them to a single scheme .
for example , the conventions of ref .
@xcite , @xcite and @xcite all differ from each other . for the future convenience of the reader , as well as
ours , we provide a self - consistent set of feynman rules and present the detailed steps involved in obtaining them , along with necessary comparisons with other references .
our goal is to provide a set of feynman rules for interactions involving gravitinos that is consistent with ref .
we define our covariant derivative as @xmath110 with commutation relation of @xmath111 generators defined as @xmath112 = if^{abc } t^c,\ ] ] which lead to the tensor field @xmath113 these definitions follow those in ref .
@xcite , and differ from those in other texts . for example , in ref .
@xcite the authors elect to use anti - hermitian @xmath114 matrices , while refs .
@xcite and @xcite use hermitian @xmath114 s which , however , are negative of what we use
. these different conventions result in different feynman rules . for example , the feynman rules for the quark - quark - gluon vertex are @xmath115 in refs .
@xcite , @xcite , and @xcite , respectively .
similarly , the feynman rule for the 3-gluon vertex depends on the conventions used .
throughout the article , we use the flat space metric of @xmath116 we may form a numerically invariant tensor @xmath117 @xmath118 that transforms as a vector in o(1,3 ) using pauli matrices @xmath119 , @xmath120 these are the clebsch - gordan coefficients which relate the ( @xmath121,@xmath121 ) of @xmath122 to the vector of @xmath123 . here
dotted indices transform under the ( 0,@xmath121 ) of the lorentz group , while those with undotted indices transform under the ( @xmath121,0 ) conjugate representation .
spinors with upper and lower indices are related through the @xmath124-tensor : @xmath125 where the antisymmetric tensor @xmath124 s are normalized as @xmath126 which holds true for @xmath124-tensors with dotted indices .
the advantage of this scheme is that the mixed tensors are symmetric : @xmath127 these tensors are used to raise and lower the indices of the @xmath128-matrices : @xmath129 it is then straighforward to relate two - component spinors to four - component spinors through the realization of the dirac @xmath130-matrices : @xmath131 where @xmath132 the generators of the lorentz group in the spinor reprsentation are given by @xmath133 with this choice of @xmath130-matrices , dirac spinors contain two weyl spinors , @xmath134 while majorana spinors contain only one : @xmath135 we define @xmath136 as @xmath137 projection operators are defined accordingly : @xmath138
we start with the general supergravity lagrangian given in chapter xxv of ref .
@xcite and adapt it to the conventions listed in appendix b. ( note that ref .
@xcite uses the flat space metric diag(-1,1,1,1 ) , amongst other differences . ) below we explicitly write down the terms of the lagrangian relevant to our calculation.@xmath139 \nonumber\\ & & -\frac{i}{2m}[\psi_\mu\sigma^{ab}\sigma^\mu\bar\lambda^a + \bar\psi_\mu\bar\sigma^{ab}\bar\sigma^\mu\lambda^a]f^a_{ab } , \label{eq : lagrangian } \end{aligned}\ ] ] where @xmath140 is the khler metric .
@xmath141 s are scalar superpartners of chiral fermion @xmath142 s , @xmath143 are the usual field strength tensors of the gauge fields @xmath144 whose superpartners are gauginos @xmath145 , and @xmath146 is the gravitino field .
covariant derivative @xmath147 s are : @xmath148 where the killing vectors @xmath149 are defined in terms of the killing potential @xmath150 . for
the minimal khler potential @xmath151 , the killing vectors and the killing potential take the form of @xmath152 four - spinors are constructed using two - spinors as @xmath153 @xmath154 @xmath155 where superscript @xmath156 is for dirac spinors , and @xmath157 for majorana spinors .
the extra @xmath158 s in gauginos are introduced following ref .
@xcite . after adding appropriate mass terms to ( [ eq : lagrangian ] )
, we obtain the lagrangian in four - spinor notation : @xmath159 where @xmath160 includes kinetic terms : @xmath161 and others include @xmath162- , and @xmath163-particle interaction terms : @xmath164\gamma^\mu \lambda^a\right]f^a_{\alpha\beta}\label{eq : grg},\end{aligned}\ ] ] from which our feynman rules will be derived .
note that an extra factor of @xmath165 is necessary in the feynman rules to account for the majorana nature of gauginos and gravitinos .
mixing between left- and right - squarks is proportional to the mass of their quark counterpart . in our calculation of cross sections at the fermilab @xmath166-@xmath167 collider ,
the effect of mixing is practically nonexistent .
therefore we replace @xmath168 and @xmath169 with the mass eigenstates @xmath170 and @xmath171 .
relative signs between terms within ( [ eq : qsqgr ] ) , ( [ eq : qsqgrg ] ) and ( [ eq : grg ] ) are related to the definition of the fermions in ( [ eq : lambdam ] ) and ( [ eq : psim ] ) , and to the fact that @xmath169 s transform as anti - triplets of @xmath172 , i.e. , their generators are @xmath173 instead of @xmath174 ( see pages 208 and 223 of ref .
terms ( [ eq : qsqgr],[eq : qsqgrg],[eq : grg ] ) include interactions with gravitinos .
if the gravitino is very light , which is the scenario we are pursuing , spin-@xmath175 components of the gravitino decouple from the spin-@xmath176 components , and it interacts with matter as a massless goldstino with derivative couplings . thus we may use an effective lagrangian with a massless goldstino @xmath177 by making the substitution , @xmath178 after simple rearrangements , terms ( [ eq : qsqgr],[eq : qsqgrg],[eq : grg ] ) become @xmath179\slashp_g\gamma^5\lambda^a ) -\frac{g_sf^{abc}}{4\sqrt{6 } m m_{3/2 } } ( \bar\psi_g[\slashv^b,\slashv^c]\slashp_g\gamma^5\lambda^a)\end{aligned}\ ] ] where subscripts @xmath180 , @xmath165 , @xmath181 , and @xmath182 of the momenta are for @xmath183 , @xmath184 , gravitino , and gluino , respectively . resulting feynman rules
are listed in fig .
[ fig : feynrules ] . for the sake of completeness
, we include the effects of left - right squark mixing in the feynman rules , where @xmath185 implies no mixing . in section 4.5 of ref .
@xcite , the author uses the on - shell condition of external particles to replace derivatives in the lagrangian by masses of external particles .
this replacement is valid for internal particles as well as off - shell contributions cancel , which is characteristic of the effective lagrangian . in table 1
we provide the feynman rules for gravitino couplings obtained by applying this prescription .
our rules appear different from those in ref .
@xcite because of the different conventions that we use ( note the differences in the definitions of @xmath186 and @xmath136 , and gluinos ) .
note that the 4-point interaction quark - squark - gluon - gravitino was overlooked in ref .
@xcite .
.alternate feynman rules for gravitino couplings of ref .
@xcite using our conventions .
the index in parentheses refers to the diagram in fig .
[ fig : feynrules ] . the rules for ( e),(f ) , @xmath187 and @xmath188 , which were not included in ref .
@xcite , are unchanged from fig . [ fig : feynrules ] . [ cols="<,<",options="header " , ] 99 p. fayet , phys .
b 69 ( 1977 ) 489 ; b 70 ( 1977 ) 461 ; b 84 ( 1979 ) 421 ; b 86 ( 1979 ) 272 ; b 175 ( 1986 ) 471 .
j. ellis , k. enqvist , and d. v. nanopoulos , phys .
b * 147 * ( 1984 ) 99 .
d. dicus , s. nandi , and j. woodside , phys .
d * 41 * ( 1990 ) 2347 and phys .
d * 43 * ( 1991 ) 2951 .
d. stump , m. wiest , and c .- p .
yuan , phys .
d * 54 * ( 1996 ) 1936 ; s. dimopoulos , m. dine , s. raby , and s. thomas , phys .
* 76 * ( 1996 ) 3494 ; s. dimopoulos , s. thomas , and j. wells , phys . rev .
d * 54 * ( 1996 ) 3283 ; k. babu , c. kolda , and f. wilczek , phys .
* 77 * ( 1996 ) 3070 .
s. ambrosanio , g. kane , g. kribs , s. martin , and s. mrenna , phys .
* 76 * ( 1996 ) 3498 and phys .
d * 54 * ( 1996 ) 5395 .
j. l. lopez and d. v. nanopoulos , mod .
lett . a * 11 * ( 1996 ) 2473 and phys .
d * 55 * ( 1997 ) 4450 .
s. park , in proceedings of the 10th topical workshop on proton - antiproton collider physics , fermilab , 1995 , edited by r. raja and j. yoh ( aip , new york , 1995 ) , p. 62 .
d. dicus , s. nandi , and j. woodside , phys .
b * 258 * ( 1991 ) 231 .
j. l. lopez , d. v. nanopoulos , and a. zichichi , phys .
* 77 * ( 1996 ) 5168 and phys .
d * 55 * ( 1997 ) 5813 .
d. dicus and s. nandi , hep - ph/9611312 .
j. ellis , k. enqvist , and d. v. nanopoulos , phys .
b * 151 * ( 1985 ) 357 ; m. nowakowski and s. d. rindani , phys .
b * 348 * ( 1995 ) 115 ; t. gherghetta , nucl .
b * 485 * ( 1997 ) 25 ; j. grifolds , r. mohapatra , and a. riotto , hep - ph/9610458 and hep - ph/9612253 .
a. brignole , f. feruglio and f. zwirner , hep - ph/9703286 .
m. a. luty and e. ponton , hep - ph/9706268 . t. bhattacharya and p. roy , phys .
b * 206 * ( 1988 ) 655 ; nucl .
b * 328 * ( 1989 ) 469 ; nucl .
b * 328 * ( 1989 ) 481 ; r. casalbuoni , s. de curtis , d. dominici , f. feruglio and r. gatto , phys .
b * 216 * ( 1989 ) 325 .
j. babcock , d. sivers , and s. wolfram , phys .
d * 18 * ( 1978 ) 162 ( see appendix b ) .
edition ( revised and expanded ) by j. wess and j. bagger ( princeton university press , princeton , 1992 ) h. haber and g. kane , phys .
( 1985 ) 75 .
t. moroi , ph.d .
thesis , effects of the gravitino on the inflationary universe " , hep - ph/9503210 . ,
2nd edition by i. j. r. aitchison and a. j. g. hey ( adam hilger , philadelphia , 1989 ) by c. itzykson and j. zuber ( mcgraw hill , new york , 1985 ) by s. pokorski ( cambridge university press , cambridge , 1987 ) by t. cheng and l. li ( oxford university press , oxford , 1984 ) | we consider the production of gravitinos ( @xmath0 ) in association with gluinos ( @xmath1 ) or squarks ( @xmath2 ) at hadron colliders , including the three main sub - processes : @xmath3 , @xmath4 , and @xmath5 .
these channels become enhanced to the point of being observable for sufficiently light gravitino masses ( @xmath6 ) , as motivated by some supersymmetric explanations of the cdf @xmath7 event . the characteristic signal of such events
would be monojets , as opposed to dijets obtained in the more traditional supersymmetric process @xmath8 .
searches for such events at the tevatron can impose lower limits on the gravitino mass . in the appendix , we provide a complete set of feynman rules for gravitino interactions used in our calculation . act-11/97 + cern - th/97 - 120 + ctp - tamu-30/97 + doe / er/40717 - 44 + july 1997 + hep - ph/9707331 0.5 cm * light - gravitino production at hadron colliders * 1.0 cm jaewan kim@xmath9 , jorge l. lopez@xmath10 , d.v . nanopoulos@xmath11 ,
+ raghavan rangarajan@xmath9 , and a. zichichi@xmath12 + 0.75 cm @xmath9astroparticle physics group , houston advanced research center ( harc ) + the mitchell campus , the woodlands , tx 77381 , usa + @xmath10 bonner nuclear lab , department of physics , rice university + 6100 main street , houston , tx 77005 , usa + @xmath13center for theoretical physics , department of physics , texas a&m university + college station , tx 778434242 , usa , and + academy of athens , chair of theoretical physics , division of natural sciences + 28 panepistimiou avenue , 10679 athens , greece + @xmath12university and infn - bologna , italy and cern , 1211 geneva 23 , switzerland + 0.50 cm 0.5
cm [email protected] + [email protected] + [email protected] + [email protected] + @xmath10present address : shell e&p technology company , bellaire technology center , p. o. box 481 , houston , tx 77001 - 0481 |
one of the major discoveries of eit on soho ( delaboudinire et al .
1995 ) was the existence of conspicuous large - scale euv propagating intensity disturbances .
these intensity fronts sometimes have the appearance of an almost circular wave front and are frequently called eit or euv waves ( e.g. , moses et al . 1997 ; thompson et al . 1998 ; thompson et al . 1999 ) .
euv waves are normally first seen in close proximity to a flaring and erupting active region and they subsequently propagate over a significant fraction of the visible surface before they become too faint to be detected .
eit observations have found that they travel at speeds 50 - 400 km / s .
statistical studies showed that they are mostly associated with coronal mass ejections ( cmes ) and not with flares ( e.g. , biesecker et al .
there is a tendency for the fastest euv waves to be associated with similar phenomena in h@xmath2 , and soft x - rays and to produce type - ii radio bursts ( e.g. , moreton 1960 ; athay and moreton 1961 ; klassen et al .
2000 ; kahler and hudson 2001 ; cliver at al . 2001 ; narukage et al .
2002 ; hudson et al . 2003 ; gilbert et al . 2004
; vrnak et al .
there is an extensive literature on this subject ( e.g. , wills - davey and thompson 1999 ; delanne 2000 ; klassen et al . 2000 ; kahler and hudson 2001 ; wu et al . 2001 ; narukage et al .
2002 ; ofman and thompson 2002 ; gilbert et al .
2004 ; zhukov and auchre 2004 ; podladchikova and berghmans 2005 ; warmuth and mann 2005 ; veronig et al .
2006 ; vrnak et al .
reviews on the topic can be found in chen and fan ( 2005 ) , gopalswamy et al .
( 2006 ) , pick et al .
( 2006 ) , schwenn et al .
( 2006 ) , warmuth ( 2007 ) . despite the extensive research on these phenomena ,
significant controversy remains over the physical origin(s ) of the euv waves .
one interpretation states that euv waves are global fast - mode waves triggered by the associated flare or cme ( e.g. , thompson et al .
1999 ; wang 2000 ; wu et al . 2001
; ofman and thompson 2002 ) .
the expectation of finding fast - mode waves in the corona was originally considered by uchida ( 1968 ) .
fast - mode waves can travel over large distances over the solar surface since they can propagate at right angles with respect to the ambient , radial to a first approximation , magnetic field of the quiet sun regions which occupies most of the solar surface , particularly during solar minimum conditions .
moreover , they have a compressive character which can lead to the intensity enhancement associated with the euv waves .
in what follows we will refer to this interpretation as _ fast - mode wave_. a sufficiently impulsive flare and/or cme can generate a wave or even a shock - wave which will then give rise to signatures over an extended range of wavelengths and physical regimes in the solar atmosphere such as , euv waves , moreton waves ( h@xmath2 ) , soft x - rays and type ii radio bursts .
another school of thought states that the euv waves are not waves at all , but rather the signature of the large scale propagation of the associated cme .
we refer to such interpretations as pseudo - waves .
one suggestion for a pseudo - wave is the disk projection of the large - scale current shells enveloping the erupting flux rope of the associated cme ( delanne et al .
2008 ) . using a 3-d mhd flux rope model
these authors proposed that these currents could be associated with enhanced ohmic dissipation leading to coronal heating .
this would then result into the enhanced euv intensities of the wave fronts .
it is not known whether this enhanced heating is steady or impulsive in nature . in what follows
we will refer to this interpretation as _ current - shell _ model for brevity .
a hybrid model , with features from both wave and pseudo - wave models is the model of chen et al .
( chen et al . 2002 ; chen , fang and shibata 2005 ; chen and fang 2005 for a review ) .
this model is based on the eruption of a flux rope and predicts two types of waves : a very fast , super - alfvnic piston - driven shock straddling the erupting flux rope which may be identified as the coronal counterpart of a moreton wave and a slower pseudo - wave associated with successive opening the overlying magnetic field induced by the erupting flux rope .
the latter type of wave is a pseudo - wave , which chen et al .
identify with the euv waves and is the result of plasma compression , similar to the conclusion of delanne and aulanier ( 1999 ) , who showed that fast expansion of the magnetic field should compress plasma at the boundaries between expanding stable flux domains , leading to enhanced emission . we will therefore treat the chen et al . model in tandem with the delanne et al
model since both models assert that the euv waves are formed around the errupting flux rope by either enhanced current shells and/or plasma compression .
note here that delanne et al .
( 2008 ) found that the disk projection of the current shells produces more contrast than that from the plasma compression .
another possibility for a pseudo - wave is successive reconnections between the expanding large - scale cme with small - scale ( cool loops ) quiet sun structures ( e.g , attrill et al .
2007a , b ; van driel - gesztelyi et al .
these reconnection events will give rise to enhanced heating which will in turn give rise to the bright front intensity . in what follows
we will refer to this interpretation as _
reconnection fronts_. finally , another interesting suggestion regarding the nature of euv waves is that they represent _ mhd solitons _ ( wills - davey , deforest @xmath3 stenflo 2007 ) . for solitary waves the wave speed is dependent on the pulse amplitude ( a function of density enhancement ; wills - davey , deforest @xmath3 stenflo , 2007 ) . as a general trend , the most well - defined coronal waves have higher density enhancement and travel faster .
more observational signatures from this mechanism need to be worked out in order to be able to distinguish , for instance , between solitary and non - solitary waves . from the above it
is clear that understanding the physical origin of euv waves will supply important constraints on coronal conditions ( magnetic field , temperature and density in the corona ) , for a fast - mode wave interpretation , or on cme initiation and early expansion , for a pseudo - wave interpretation .
part of the problem for conclusively determining the nature of euv waves stems from the limitations of the eit observations used almost exclusively to study these waves : relatively low cadence ( 12 min ) , one temperature ( normally the 195 channel ) and a single vantage point .
these limitations are largely overcome with the stereo mission ( kaiser et al 2008 ) .
the extreme ultraviolet imaging telescope ( euvi ; wuelser et al .
2004 ) of the secchi instrument suite ( howard et al .
2008 ) excels over eit because of its faster cadence ( 2.5 min or less during campaign periods ) , simultaneous imaging at different temperature channels , and distinct vantage points allowing a truly 3d view of the sun .
indeed , previous analyses of euv waves with high cadence euvi data showed that they propagate at higher speeds than previously determined , closer to the magnetoacoustic speed , ( long et al .
2008 ; veronig et al .
additionally , gopalswamy et al .
( 2009 ) detected wave reflection from a coronal hole . finally , patsourakos @xmath3 vourlidas ( 2007 ) found evidence of kink - like oscillations of quiet sun loops in the wake of an euv wave .
mhd simulations showed that euv waves can induce similar oscillations , at least in the source ar loops ( ofman 2007 ) .
all the above findings lend strong support to a wave interpretation .
however , these studies relied on a single viewpoint for their analyses . here , we present the first stereoscopic study of an euv wave which occurred on december 07 2007 . also for the first time , we perform a comprehensive comparison of predictions and/or expectations from each candidate wave interpretation and test them against the stereo observations .
we demonstrate how all the unique advantages of the stereo observations place new and strong constraints on the nature of euv waves .
we analyzed the observations of an euv wave seen by the two stereo spacecraft ( sc ; hereafter a and b ) from @xmath1 04:35 - 05:15 ut during 7 december 2007 .
the stereo spacecraft were significantly separated by @xmath4 .
the wave was initiated from the small active region 10977 , which was close to disk center as viewed by a and towards the east limb as viewed by b ( figure [ fig : context ] ) .
the overall configuration of the corona was rather typical of solar minimum conditions : simple active regions , a low - latitude coronal hole , with the rest of the disk dominated by quiet sun .
close to the time of the wave launch , a small b1.4 class flare started at around 04:35 in the same ar and peaked at around 04:41 ut .
an associated cme was seen later in coronagraph data .
this cme was rather slow , reaching speeds of @xmath1 300 km / s in the coronagraphic fields of view , as determined by _ cactus _
( robbrecht @xmath3 berghmans 2005 ) and was not particularly bright .
we focus here on euv data and total brightness white - light images collected by the euvi and the cor1 coronagraph ( thompson et al .
2003 ) on secchi , respectively .
euvi takes full disk images in euv channels centered around 171 , 195 , 284 and 304 ( hereafter referred as 171 , 195 , 284 and 304 ) .
euvi has @xmath1 1.6 arcsec pixels and our observations have a cadence of 2.5 , 10 and 20 minutes for 171 , 195 - 304 and 284 , respectively .
cor1 is an internally occulted coronagraph observing from 1.3 - 4.0 @xmath5 .
the cor1 pixel size is 15 arcsec and nominal cadence is 15 minutes . the euvi and cor1 data were processed with the standard _
secchi_prep _ routine .
note that the euvi and cor1 a & b image pairs are synchronized ; namely , the observations were taken at the _ same _ time on the sun .
therefore , we do not have to worry about the different light travel times from the sun to the two stereo spacecraft
. a running ratio movie of the euv wave in the 195 channel for both spacecraft can be found online ( video1.mov ) and representative snapshots from this movie are shown in figure [ fig : wave_195 ] .
each frame corresponds to the ratio of a given 195 image with the previous one .
note here that the cadence of the 195 euvi data ( @xmath1 10 minutes ) is similar to that of eit ( @xmath1 12 minutes ) : the basic difference is in the multi - view point aspect of the euvi observations together with the higher sensitivity of euvi compared to eit .
the latter helps to better observe faint transients like euv waves . running ratio images
emphasize changes of the brightness , location , and structure of emitting features between two subsequent images and highlight faint propagating disturbances such as eit waves .
these images correspond to the temporal derivative of the observed intensities .
the dark regions do not necessarily represent true dimmings ( e.g. chertok @xmath3 grechnev 2005 ) .
we provide a movie of plain ( non - differenced ) 195 images ( video2.mov ) , in which each frame has been first processed by a wavelet - based technique ( stenborg , vourlidas @xmath3 howard 2007 ) to increase the image contrast .
the propagation of a rather sharp wave - front is evident in this movie .
the wave first appears at @xmath1 04:35 ut at the periphery of the active region , following expanding loop motions which started around 04:15 .
after 04:35 , the wave starts to expand over the solar surface , becoming progressively more diffuse and faint ; after 04:45 it attains a quasi - circular shape and looks similar to data from both spacecraft .
the wave seems to avoid the equatorial coronal hole eastwards of the source ar . by 05:15
the wave covers a significant part of the visible solar disk , but by this time is rather diffuse and faint .
all these characteristics are pertinent to typical solar minimum euv waves ( e.g. , moses et al .
1997 ; thompson et al .
1998 ) .
however , we noticed a peculiarity in the wave propagation early on : around 04:35 the wave seems to propagate in opposite directions as viewed by the two instruments !
it propagates eastward in euvi - a and westward in euvi - b ( see video1.mov , video2.mov and figure [ fig : wave_195 ] ) . at this time
the wave is incomplete ; it is not forming a full circle .
higher cadence data are required to resolve this puzzling observation as well as to establish firmer conclusions about the relative timings between different aspects of the event ( e.g. , flare , loop openings , wave ... ) .
for this reason , we turn our attention to the 171 images which were taken at a relatively high cadence of 2.5 min .
we process them with the a wavelet - based technique to increase the image contrast ( stenborg , vourlidas @xmath3 howard 2007 ) .
video3.mov contains an 171 a & b movie of the area around the source active region .
several snapshots ( in running ratio format to emphasize the motions ) from the movie are given in figure 5 .
video4.mov contains a 171 a & b movie over a more extended area , covering a significant fraction of the solar disk , to capture the large - scale evolution of the wave .
figure [ fig : wvl_171 ] has several snapshots from this movie in running ratio format .
several remarks can now be made .
first , loops start to slowly rise around 04:11 in a and almost 8 minutes later in b ( figure [ fig : wvl_171 ] and video3.mov ) .
they are directed towards west ( east ) in b ( a ) , similarly to what is seen in the first column of figure [ fig : wave_195 ] .
if we now consider the expanding loops in a indicated by an arrow in the right panel of figure [ fig : wvl_171 ] we note that they are seen more or less face - on .
given the relatively large separation ( @xmath4 ) between a and b these loops should be then seen edge - on in b , as indicated by the arrow in the left panel of figure [ fig : wvl_171 ] .
therefore , the oppositely - directed early signature of the wave is associated with the expanding motions of the _ same _ loops observed from the two well - separated viewpoints . between @xmath1 04:31:00 - 04:33:30
the expanding loops undergo a sudden jump and can not be traced anymore ; a wave front starts to develop ( figure [ fig : wvl_171 ] and video3.mov ) .
this jump occurs slightly before the start of the associated flare ( @xmath1 04:35 ) .
given that flare onset and impulsive cme acceleration seem to simultaneously occur ( e.g. , zhang et al .
2001 ) , it is very possible that the observed loops jump also marked the the start of the impulsive acceleration of the associated cme .
the first almost quasi - circular wave front forms around 04:41 ; the wave also seems to interact with a prominence to the east of the active region at around 04:43 ( figure [ fig : wvl_171_wave ] and video2.mov ) .
the most important findings of this section , namely that the wave appeared different , i.e. it exhibited different shapes and direction of propagation , during its early phases from the two distinct vantage points , when loop expansion was clearly observed , ( i.e. figures [ fig : wave_195 ] , [ fig : wvl_171 ] , [ fig : wvl_171_zoom ] ) while later on basically the same quasi - circular pattern was observed have significant implications .
basically , both pseudo - wave theories expect a quasi - circular wave shape _ at all times _ ( see delanne et al .
2008 ; attrill et al .
2007a ) which is clearly in disagreement with our observations .
furthermore , the fact that during later times the wave seems similar on both euvis could be a problem for the _ current shell _ model of delanne et al .
( 2008 ) which expects the wave to be the disk projection of a very high altitude current shell ( 280 - 407 mm ) , which should appear significantly different from well - separated viewpoints .
this is probably not the case for the _ reconnection fronts _ model of attrill et al .
( 2007a ) which assumes that the wave is formed at lower altitudes .
on the other hand , a quasi - circular shape is consistent with a wave theory , for which the wave is simply the projection of the 3d wave dome onto the solar surface ( e.g. uchida 1968 , wang 2000 ) .
to summarize , we find that the _ transition _ of the wave appearance , from an asymmetric loop expansion early on to a more symmetric and diffuse front , seems to be against the expectations of pseudo - wave theories .
these theories invoke symmetric loop expansion at all times , either via ` passive ' flux rope eruption or successive reconnections . on the other hand ,
the observed appearance is consistent with a wave driven by an impulsive loop expansion .
we now study the wave behavior as a function of temperature .
figure [ fig : multi_temp ] contains almost simultaneous ( within a minute ) running ratio images of the wave seen in all euvi - a , in which the wave was closer to disk center compared to euvi - b .
note that we used the same 20-minute running ratio images in all channels , which is the slowest cadence among the euvi channels .
all ratio images have been scaled to the same ratio range and a spatial 3-pixel wide median filter has been applied in order to reduce the noise .
azimuthally - averaged radial intensity - ratio profiles of the images of figure [ fig : multi_temp ] are given in figure [ fig : plot_lena ] .
they were produced using a procedure similar to that described by podlachikova @xmath3 berghmans ( 2005 ) : the image data were first projected onto a spherical polar coordinates system with its center on the flare cite , and we then averaged those maps over polar angle to obtain the azimuthally - averaged radial intensity - ratio profiles of figure [ fig : plot_lena ] . in this plot
the wave front is located ahead of a region exhibiting a large - scale minimum of the intensity ratio ( radial distances @xmath6 250 pixels ) . as can be seen from figures [ fig : multi_temp ] and [ fig : plot_lena ]
the wave is visible in the coronal channels of euvi 171,195 and 284 .
it is better defined in 195 while its visibility decreases when going to 284 and 171 . on the other hand ,
the wave is barely visible in the mainly chromospheric / transition region 304 channel .
the wave in this channel was diffuse , and looked similar to what seen in the 284 channel .
the intensity increase associated with the wave takes values in the range of @xmath1 5 - 30 @xmath7 .
the wave propagates over quiet sun regions , where the bulk of the coronal plasma lies in the temperature range 1 - 2 mk , with a peak in the quiet sun differential emission measure distribution around 1.5 mk ( e.g. brosius et al . 1995 ) , a temperature close to the peak of the temperature response function of the 195 channel .
the fact that the wave is best seen in this temperature range implies that it represents mostly a density increase without significant plasma heating taking place .
considering also the small intensity increase we conclude that the observed wave represents a small ( i.e. , linear ) density perturbation of the ambient quiet sun , without any significant plasma heating taking place .
this behavior is consistent with coronal propagation of a fast - mode wave .
the wave has a compressive character and when propagating over the qs corona will compress the ambient plasma , thereby producing intensity increases which could be the observed wave .
given the much higher densities in the chromosphere / transition region it would be more difficult for a small amplitude wave to produce significant compression there , which explains the lack of a 304 signature .
when the eruption is very fast ( @xmath6 several hundred @xmath8 ) a shock wave can be produced .
this will give rise to significant density and temperature enhancements over a larger temperature range and to sharp wave fronts observed not only in euv but also in chromospheric emissions @xmath9 and soft - xrays ( i.e. the ` brow ' and s wave events ) as well type - ii radio bursts ( e.g. see the review of warmuth 2007 ) .
the observed multi - temperature behavior of the wave poses problems to pseudo - wave interpretations .
consider first the _ current shell _ model of delanne et al .
according to this model the wave signature is due to ohmic heating in large - scale current layers at high altitudes occurring around the erupting flux rope .
note here that our best current theoretical understanding of coronal heating points to rather small - scale currents as a possible heating agent ( e.g. klimchuk 2006 ) .
moreover , observations show little correlation between large - scale currents and sxr intensities ( fisher et al .
1998 ) .
note now that the density drops - off very rapidly with height in the solar atmosphere ( it can drop by a factor of @xmath10 5 - 10 between the base of the corona and 250 mm ; see for example figure 1.20 in the compilation of density measurements of aschwanden 2005 ) and the euv intensities are proportional to the line of sight integral of the density squared .
one would then expect that a disk observation in the euv , such as those of the euv waves , would be largely dominated by the contribution of the lower ( and more dense ) layers and not from the higher layers .
however , the _ current shell _
model finds that most of the contribution to the current density should originate from high - altitudes ( @xmath6 250 mm ) , delanne et al .
the above discussion implies that the _ current shell _ model may not lead to any sizeable coronal signal at all .
we now turn our attention to the _ reconnection front _ model of attrill et al .
the observations showed a minimal wave signature in the 304 channel .
note here that the wave did not look substantially better in ( plain ) 304 images processed by wavelets .
this operation has a similar effect on the data as differencing ( i.e. enhance subtle features ) , but applies to plain images which takes out some of the ambiguities in the interpretation resulting from taking differences .
moreover , the 304 data of figure [ fig : multi_temp ] were largely smoothed by a box - car of 15 pixels to bring up the wave .
the _ reconnection front _ model of attrill et al .
relies upon small - scale reconnections between the laterally expanding cme and small loops in the surrounding qs .
the energy content in these reconnections would be similar to that associated with the large array of transient phenomena known to occur over the qs which are widely believed to be the result of small - scale reconnections ( see the review of parnell 2002 for example ) .
therefore , one would expect brightenings of similar magnitude in the frame of the attrill et al .
model which should be easily observable .
note here that the magnitude of these brightenings would be smaller that those occurring in active regions because of the substantially weaker qs magnetic fields .
the typical timescales in the chromosphere / transition region as well as the lifetimes of these transients are in the range 100s - few minutes .
this means that the relatively low cadence of the 304 data ( 10 min ) , could cause us to miss the moment of maximum emission of some of the brightenings but not all of them .
this would give rise to a rather inhomogeneous and more importantly incomplete wave - fronts which is probably not the case .
higher - cadence 304 observations are required to show if the above is really the case .
the footpoints of coronal structures , where the bulk of their tr / chromospheric emissions comes from , represent a very sensitive monitor of coronal conditions . in the case of steady or quasi - steady heating , for example ,
the tr emission is proportional to the coronal pressure ( e.g. klimchuk , patsourakos and cargill 2008 ; equation a12 ) .
impulsive heating would give rise to sizeable footpoint emissions , turning them visible in chromospheric / tr emissions like 304 ( antiochos et al . 2003 ; de pontieu et al . 1999 ; hansteen 1993 ; klimchuk 2006 ; klimchuk et al .
2008 ; martens et al . 2000 ; patsourakos and klimchuk 2008 ; spadaro et al .
2006 ; winebarger , warren and flaconer 2008 ) . and
this occurs irrespectively of the spatial location of the heating along individual field lines ; thermal conduction and mass flows are very efficient at redistributing any excess heat along the field .
finally , it is well - known observationally that the quiet sun transition region and chromosphere exhibit stronger variability than the overlying corona ( e.g. berghmans , clette and moses 1998 ; patsourakos and vial 2002 ; teriaca , madjarska , and doyle 2002 ) .
therefore , if an euv wave is associated with some sort of impulsive heating , one may expect a significant chromospheric / transition region signature maybe stronger than the coronal one .
our observations show the contrary . the small impact of the wave in the transition region and the chromosphere may not be a problem for the _ current shell _ model .
this is because the heating in that model occurs on much higher field lines than for the _ reconnection front _ model . and
as we saw before footpoint emissions are proportional to the coronal pressure , which in turn is also proportional to the field line length and therefore height for the same temperature ( e.g. the scaling - law of rosner , tucker and vaiana 1978 ) .
another potential problem for pseudo - waves concerns their width . assuming a rather low limit for the coronal cooling time of 1000 s and for a typical euv wave propagation speed of @xmath1 300 @xmath11
one finds that a pseudo - wave should result in a very extended @xmath1 3@xmath12 km ( or @xmath1 0.5 @xmath5 ) band of emission in its wake .
this is clearly in disagreement with the rather sharply defined wave - fronts observed during the early stages of euv waves ( see for example video1.mov and particularly video2.mov with the plain wavelet images ) .
for instance we estimated a full wave width of @xmath1 160 arcsec ( i.e. @xmath1 110 mm ) for a snapshot of the wave taken at 04:45:30 ; see figure [ fig : wave_195 ] .
this is only the @xmath13 of @xmath5 .
we finally comment on the 304 channel emission .
although the passband is dominated by the he ii line with a peak temperature of 0.08 mk ( upper chromosphere and lower transition region ) , this passband also contains a contribution from si xi at 303.32 with a peak formation temperature of @xmath14 mk which is more diffuse than the chromospheric he ii emission ( e.g. stenborg , vourlidas , & howard 2008 ) . however , spectrally resolved full disk cds observations by thomson @xmath3 brekke ( 2000 ) showed that the si ix contribution in qs areas is very small ( 4 @xmath7 of the he ii ) and diffuse . in our event ,
the wave signatures in the 304 images are diffuse and faint ( similar to the 284 observations ) which suggests a likely coronal origin for them .
this is problematic for the _ reconnection front _ interpretation where , according to the preceding discussion , we should expect low corona / chromospheric signatures _ in addition _ to the coronal signal .
thanks to the availability of the two secchi viewpoints , we are able to estimate the 3d location of the wave .
we manually selected several points along a segment of the wave front in an euvi running ratio images pair ( figure [ fig : triagul ] ) .
the segment covered a significant fraction of the wave front so that our measurement can be considered as representative of the structure .
we then fitted these points with a parabolic function for each telescope and determined the image x , y coordinates of the corresponding vertexes . approximating the wave front with a parabola
is a reasonable but by no means unique choice : cme fronts almost universally exhibit concave - outwards fronts and wave fronts can be spherically symmetric .
triangulation of the vertex locations using the _ scc_triangulate _ routine ; ( see inhester ( 2006 ) for the basics of triangulation and stereoscopy ) supplied the 3d coordinates ( x , y , z ) of the vertex of the parabola .
the height @xmath15 ) of the wave above the solar surface is about 0.13 @xmath5 or @xmath16 mm ( the error is the standard deviation of the heights determined from 10 repetitions of the above procedure ) and is comparable to the coronal scale - height for quiet sun temperatures ( @xmath1 70 mm for a temperature of 1.5 mk ) .
this is consistent with fast - mode wave propagation over quiet sun areas , since the wave perturbs the ambient coronal plasma with its bulk confined within a coronal scale - height .
a height of @xmath1 90 mm is probably too small for large - scale current layers around an erupting flux rope of the _ current shell _ model .
our stereoscopic height observations of the wave of figure [ fig : triagul ] were taken around 30 minutes after loop expansion was first observed and not very far from when the cme first emerged in the cor1 field of view ( 1.5 @xmath17 ) which means that the height of the errupting structure would have been substantially large . moreover , delanne et al .
( 2008 ) expect that the current shells most contributing to the wave should be high - lying ( @xmath6 250 mm ) . finally , our height measurement is probably too large to be consistent with the _ reconnection front _ model .
the fraction of quiet sun loops which extend to substantially large heights ( @xmath6 10 mm ) represents a small fraction of the qs magnetic flux ( 10 @xmath7 maximum ) ; with the majority of qs loops not reaching heights larger than 10 mm ( close et al . 2004 ) . in the frame of the attrill et al model ,
we expect more reconnections to occur between the shorter qs loops and the expanding flux - rope .
also note that since the magnetic field strength decays with height , reconnections employing low - lying qs loops would be more energetic than those employing larger ones .
therefore , the radiative signal of the wave in the frame of the _ reconnection front _ model will be dominated by the contribution of the low - lying loops .
larger loops could prevent some of the interactions between the lower loops and the expanding cme , but possibly not all of them since the small loops are more numerous . from the above , and given the lack of any predictions on the height of reconnections for the _ reconnection front _ model , we speculate that the wave height in this scenario would be that of qs cool loops ; namely 5 - 10 mm tall ( e.g. , aiouaz and rast 2006 ; dowdy , rabin , and moore 1986 ; feldman , widing , and warren 1999 ; close et al . 2004 ; patsourakos , gouttebroze , and vourlidas 2007 ; peter 2001 ; snchez almeida et al .
2007 ; tu et al . 2005 ) . to further test the suggestion that euv waves could be pseudo - waves directly linked to the cme expansion , we performed forward modeling of the associated cme as seen by cor1 .
we used the flux rope cme model ( graduated cylindrical shell ; gcs ) of thernisien , howard and vourlidas ( 2006 ) .
gcs is a purely geometrical 3d model aiming to reproduce the large - scale appearance of cmes in coronagraph data and in particular the cme envelopes : it does not aim to reproduce the fine internal structure of cmes .
it makes no assumption about the physical mechanisms producing the observed cme envelopes .
the model consists of a tubular section attached to two cones , which approximate the main body of the flux - rope and its legs , respectively .
the model is formulated in terms of few adjustable parameters , i.e. location , longitude and latitude of the source region , height of the cme front , etc ( see thernisien et al 2006 for details ) .
the gcs model makes it easy to calculate projections onto the plane of the sky from any viewpoint ( e.g. stereo , soho ) and compare them with coronagraphic cme observations ( cor1 in our case ) ; the free parameters of the model are modified until the best match is found simultaneously in both viewpoints .
this requirement makes the model sensitive to changes of its free parameters and therefore results in more accurate cme fits than possible with a single viewpoints ( thernisien et al 2009 ) .
we applied the gcs model to an image pair of total brightness cor1 images taken around 05:15 ( upper panel of figure [ fig : fit_cme ] ) . at this time
, the wave - associated cme has just emerged above the cor1b occulter while it was never seen in cor1a data .
moreover , the euv wave was still visible on the disk but was rather diffuse ( lower panel of figure [ fig : fit_cme ] ) . to construct fig .
[ fig : fit_cme ] we did the following : ( 1 ) we found the gcs model best reproducing the associated cme observed by cor1 ( upper panel of figure [ fig : fit_cme ] ) and , ( 2 ) projected the upper ( i.e. spherical ) section of the model onto the co - temporal euvi running ratio images containing the wave ( lower panel of figure [ fig : fit_cme ] ) .
note that step ( 1 ) reproduced the large - scale cme seen by cor1-b ( top left panel of figure [ fig : fit_cme ] ) , and indicated that the cme in cor1-a should lie behind the occulting disk ( top right panel of figure [ fig : fit_cme ] ) , as it was observed . in step
( 2 ) self - similar expansion of the flux - rope like cme was assumed as suggested by both observations ( schwenn et al .
2003 ; cremades @xmath3 bothmer 2004 ; thernisien et al . 2006 ; thernisien et al .
2009 ) and models ( trk and kliem 2005 ; attrill et al
. 2007a , delanne et al . 2008 ) .
according to pseudo - wave models , the euv waves correspond to either the disk projection of the cme envelope for the _ current shell model _ or to the low coronal lateral extent of the cme for the _ reconnection fronts model_. therefore , the procedure of figure [ fig : fit_cme ] allows to validate the realism of pseudo - wave models .
note here that our procedure calculates the projection of the upper section of the cme only , and not of its footpoints , which are line - tied to their low corona and presumably correspond to areas of strong dimmings . in principle
, the cme envelope will laterally expand either by being ` pushed ' by the erupting flux rope or by successive reconnections with qs loops . in any case
, the projection of the cme on the low corona the pseudowave will always follow the cme evolution . since the gcs model of figure [ fig : fit_cme ] is constrained by the position and morphology of the cme in the coronagraph data , it accounts for any possible deflections of the cme from its original propagation direction ( e.g. , thernisien et al .
2006 ) . however , the cme projection disagrees with the actual observed wave in two important aspects : ( i ) it is displaced with respect to the wave and , ( ii ) its projected area is smaller than the wave .
the displacement between the projected cme and the wave is a strong argument towards a wave interpretation ( wave and cme propagate independently ) and against pseudo - wave models ( there is a direct link between the cme and the wave ) .
the latter would need to invoke special conditions to account for such displacements .
besides , the often reported cme deviations ( due to nearby coronal hole boundaries ) are based on single viewpoint observations ( lasco ) and therefore may suffer from projection effects .
note here that delanne et al .
( 2008 ) show a projection of the current shell from their model on two snapshots of an observed wave , and a good match was found ( figure 11 of delanne et al .
( 2008 ) ) . however , this lacked both the co - temporal snapshot of the associated cme and the multiple lines of sights used in our modeling , thereby not fully constraining the problem .
this underlines the importance of multiple viewpoints and of coronagraphs going as close as possible to the solar surface .
the size disparity between cme projection and wave poses another problem for pseudo - wave models which by definition assume they are equal .
indeed , if we consider the very rapid decrease of density with height and the strong dependence of euv intensity on it ( @xmath18 , with @xmath19 the path length along the line of sight ) the signal from a disk observation of a cme in the euv would be dominated by the contribution of its lower sections rather of its upper sections .
actually , since cmes expand with height , the disk projection of the cme of figure [ fig : fit_cme ] largely overestimates its actual size which makes the comparison between the observed wave and pseudo - wave models even more problematic .
as explained above , the disk projection of the gcs model corresponds to the projection of the upper part of cme onto the solar disk , and therefore does not correspond to the flux - rope footpoints . in figure
[ fig : fit_wave ] we do the inverse .
we first find a flux - rope model with a projection on the disk that matches the observed euv wave simultaneously in a and b. then we compare its projection on the cor1 images .
we believe that this is a good method to test the pseudo - wave model prediction ; namely that the outer cme envelope is the wave . again , however , we find that the pseudo - wave interpretation can not account for the observations .
the forward modeling of pseudo - waves of this section is by no means unique , in the sense that different geometries and/or eruption scenario may be employed .
however , as explained above , the employed geometry ( i.e. , flux - rope like and self - similar expansion ) find substantial backing in both observations and modeling and therefore represents a reasonable choice .
finally note here that if cme and euv waves had been different facets of the very same phenomenon ( i.e. a pseudo - wave ) , then our fittings of figures 9 and 10 would have been mutually consistent ( e.g. the cmes of figures 9 and 10 should have been the same ) , which is not the case . for our pseudo - wave modeling we had to use co - temporal images of the wave and the cme .
because of its slow speed , the cme took about 40 minutes after the first appearance of the wave to enter the cor1 field of view . by that time
the wave had already traveled over a significant fraction of the solar area and was very diffuse .
we have to wait for observations of more impulsive cmes to perform cme - wave comparisons , with a better defined wave signature .
having tested the pseudo - wave interpretation against the observations in the above , we turn to the wave interpretation .
we perform a fast - mode wave simulation for the observed euv wave following wang ( 2000 ) .
first , a potential full source surface extrapolation of the global coronal magnetic field was carried out using mdi synoptic maps centered at the day of our observations .
for each traced field line , a hydrostatic isothermal atmosphere at 1.5 mk was then attached following a scaling - law relating density and base magnetic field .
the above steps supplied the 3d distribution of the fast - mode speed in the corona .
finally , a number of rays have been initiated at the periphery of the active region , and tracked as a function of time .
essentially the fast - mode wave is diffracted away from regions with high gradients in the fast - mode speed .
this happens in active regions ( upward wave diffraction ) and in coronal holes ( diffraction away from the coronal hole ) .
less diffraction occurs in quiet sun , where the fast - mode speed gradients are smoother .
note that the model makes no explicit assumption about the physical mechanism triggering the wave . for
more on the model refer to wang ( 2000 ) .
two snapshots of our simulations are given in figure 10 where we plot the projection of the wave front ( white dotted line ) on the solar surface . a viewing point from soho has been assumed .
the simulated snapshots are separated by 15 minutes , which is also approximately the time separation ( 12 minutes ) between the 2 running - ratio eit images ( delaboudinire et al . 1995 ) which complement figure 11 .
we see that the fast - mode simulation essentially manages to track the large scale appearance of the wave both in terms of size and shape .
it follows rather closely the location and shape of the wave between the two successive snapshots and even reproduces some details of the wave , like the avoidance of the coronal hole .
it is interesting to note that if the wave is launched at the center of the active region , it will expand very rapidly in the radial direction which is at odds with the observations .
the radial expansion is due to the very rapid decrease of the fast - mode speed with height , which leads to very strong wave diffraction in the upward direction . instead
, the observations in section 2.1 show that the wave is low - lying and appears in the periphery of the active region .
this is also a strong argument against the flare as a trigger of the wave since the wave would form at the heart of the active region in that case .
we have performed the first detailed analysis of an euv wave from two vantage points as observed by the secchi disk imagers and coronagraphs .
the high cadence and stereoscopic observations allowed us for the first time to precisely determine _ where _ ( i.e. in the ar periphery ) and _ when _
( i.e. closely associated with the impulsive acceleration phase of the cme and the start of the associated flare ) the first signature of the euv wave is seen .
our findings place new and tighter constraints on the physics of euv waves and are summarized as follows : * high - cadence ( 2.5 min ) images showed a strong association between expanding loops and the euv wave onset and significant differences in the wave appearance during early stages as viewed by the two euv imagers ; these differences largely disappear later .
* the wave first appears at the ar periphery when an abrupt jump of the expanding loops occurs within an interval of 2.5 minutes and before the first flare signature . *
the wave is seen more prominently in emissions formed in the @xmath1 1 - 2 mk range ; little , if anything , is seen in the chromosphere / transition region .
the wave represents a rather modest intensity increase of @xmath20 over the background .
* triangulations showed that the wave was low - lying with a height of @xmath1 70 mm . *
the on - disk projection of a geometrical flux rope model of the associated cme did not show good agreement with the observed wave ; a disparity of the size and the location of the projection and of the observed wave was found .
* simulations of fast - mode wave were found to be in a good agreement with the observed wave .
the comparison of our findings with the expectations of the different proposed mechanisms for euv waves were discussed in the previous sections and are summarized in table-1 .
we consider that a fast - mode wave is the most probable interpretation .
pseudo - wave theories run into significant trouble when trying to reconcile several important elements of the observations : an asymmetric - to - symmetric transition in the wave appearance as seen from the two sc , a rather poor chromospheric / transition region signature ( inconsistent at least with the _ reconnection front _ model ) , probably too small wave height for the _ current shell _ model and too large for the _ reconnection front _ model and a missmatching cme on - disk projection with respect to the observed wave . on the other hand ,
a fast - mode wave interpretation is consistent with all observations .
we wish to iterate it here , that by euv waves we are exclusively referring to large scale propagating intensity enhancements , which reach significant distances from the source region , especially during periods of solar minimum when the coronal landscape is rather simple .
we do not exclude the possibility that other propagating intensity fronts which follow closely the expanding dimmings could be indeed pseudo - waves .
however , these dimmings would normally never reach the rather global scales of the euv waves we are considering here . statistical studies of euv dimming showed an average width of only @xmath21 ( reinard and biesecker 2008 ) .
we think that models such as the hybrid model of chen et al
. may be able to explain both the dimming and the ` true ' wave .
further mhd simulations mapping the parameter space of different cme - wave - dimming scenarios are required .
.comparison of the euvi wave observations to the predictions of various proposed mechanisms to explain euv waves .
the first column gives the physical mechanism , the second column the observables and the third column outlines the observations first and the corresponding model predictions next .
tr / chromo = transition region / chromosphere . [ cols="<,<,<",options="header " , ] we can now draw the following picture about the observed euv wave .
coronal loops start to slowly expand ( e.g. figure [ fig : wvl_171 ] and video3.mov ) .
this happens clearly before the onset of the flare .
the angular separation of our observations was large enough ( @xmath22 ) to allow us to observe the _
same _ loops from strikingly different perspectives early on in the event ( edge - on and face on ; first column of figure [ fig : wave_195 ] , figure [ fig : wvl_171_zoom ] ) . just before the start of the associated flare ,
the loops abruptly disappear and can not be traced anymore ; a well - defined wave front forms at the periphery of the active region .
this front could be generated by a large scale disturbance induced by either the impulsive cme acceleration or the associated flare .
the latter possibility can be rather safely excluded for the following reasons : ( 1 ) the associated flare was weak , ( 2 ) we found in section 2.5 that launching the wave in the ar core , as expected from a flare - associated blast - wave , would not lead to wave fronts consistent with the observations .
it is possible that the catastrophic energy release that apparently destroys the ascending loops within 2.5 minutes is also responsible for the launch of the wave which starts to propagate freely .
a freely - propagating wave is consistent with the fact that the wave becomes more diffuse as it propagates farther from the active region ( e.g. parker 1961 ; hundhausen 1985 ; wartmuth 2007 ) . before closing
we note that clearly more work from both observational and theoretical approaches is required .
first , although the analyzed event could be considered as a ` typical ' solar minimum euv wave , our study needs to be extended to larger sets of euv wave events seen by stereo . more emphasis
should be given to higher cadence wave observations in the 195 channel , in which waves are best visible , in order to obtain a larger number of usable measurements per event .
moreover , higher cadence is needed also in the 304 channel in an attempt to search for evidence of transient brightenings that could be associated with reconnection fronts .
combination of stereoscopic wave observations by stereo with the ultra - high cadence euv observations from sdo will certainly help into that direction .
finally , we note that the theories of euv waves have reached various degrees of sophistication .
however none of these theories are yet at a level to construct synthetic euv images out of the particular physical mechanism to allow direct comparison with the observations .
for this reason , we relied in this study upon morphological concepts like wave fronts , cme projections , geometrical shapes etc , which have led us to physical insights but are not sufficiently quantitative , at least to our satisfaction .
we welcome the recent development of full sun mhd models of linker et al .
( 2008 ) which treat in a self - consistent manner both the cme eruption ( i.e. the wave driver ) and plasma thermodynamics ( i.e. what it ultimately required for calculating realistic euv intensities ) .
the simulated euv images showed the appearance of propagating intensity fronts similar to typical euv waves .
the analysis of these fronts showed they were associated with fast - mode waves .
the secchi data used here were produced by an international consortium of the naval research laboratory ( usa ) , lockheed martin solar and astrophysics lab ( usa ) , nasa goddard space flight center ( usa ) , rutherford appleton laboratory ( uk ) , university of birmingham ( uk ) , max@xmath23planck@xmath23institut for solar system research ( germany ) , centre spatiale de lige ( belgium ) , institut d optique thorique et applique ( france ) , and institut dastrophysique spatiale ( france ) .
we sincerely thank the referee for many usefull comments that led to a significant improvement of the manuscript .
we thank g. attrill , l. van driel - gesztelyi , e. robbrecht , h. hudson , s. plunnket , j. linker , s. krucker and m. pick for useful discussions .
delaboudinire , j .-
p . , artzner , g.e . ,
brunaud , j. , gabriel , a.h . ,
hochedez , j.f .
, millier , f. , song , x.y .
, au , b. , dere , k.p . , howard , r.a .
, kreplin , r. , michels , d.j . ,
moses , j.d . , defise , j.m . ,
jamar , c. , rochus , p. , chauvineau , j.p . ,
marioge , j.p . ,
catura , r.c . ,
lemen , j.r . , shing , l. , stern , r.a . , gurman , j.b . ,
neupert , w.m . ,
maucherat , a. , clette , f. , cugnon , p. , and van dessel , e.l . : 1995 , _ solar phys . _ * 162 * , 291 .
hundhausen , a. j. , some macroscopic properties of shock waves in the heliosphere , in collisionless shocks in the heliosphere : a tutorial review , geophys .
monogr . ser . ,
34 , edited by r. g. stone and b. t. tsurutani , p. 37 , agu , washington , d. c. , 1985 .
pick , m. , forbes , t.g . , mann , g. , cane , h.v . ,
chen , j. , ciaravella , a. , cremades , h. , howard , r.a . ,
hudson , h.s . , klassen , a. , klein , k.l . ,
lee , m.a . ,
linker , j.a . ,
maia , d. , mikic , z. , raymond , j.c .
, reiner , m.j . ,
simnett , g.m . ,
srivastava , n. , tripathi , d. , vainio , r. , vourlidas , a. , zhang , j. , zurbuchen , t.h . ,
sheeley , n.r . , and marqu , c. : 2006 , _ space science reviews _ * 123 * , 341 .
schwenn , r. , raymond , j.c . , alexander , d. , ciaravella , a. , gopalswamy , n. , howard , r. , hudson , h. , kaufmann , p. , klassen , a. , maia , d. , munoz - martinez , g. , pick , m. , reiner , m. , srivastava , n. , tripathi , d. , vourlidas , a. , wang , y .- m . , and
zhang , j. : 2006 , _ space science reviews _ * 123 * , 127 . | one of the major discoveries of the extreme ultraviolet imaging telescope ( eit ) on soho were intensity enhancements propagating over a large fraction of the solar surface .
the physical origin(s ) of the so - called ` eit ' waves is still strongly debated .
they are considered to be either wave ( primarily fast - mode mhd waves ) or non - wave ( pseudo - wave ) interpretations .
the difficulty in understanding the nature of euv waves lies with the limitations of the eit observations which have been used almost exclusively for their study .
they suffer from low cadence , and single temperature and viewpoint coverage .
these limitations are largely overcome by the secchi / euvi observations on - board the stereo mission .
the euvi telescopes provide high cadence , simultaneous multi - temperature coverage , and two well - separated viewpoints .
we present here the first detailed analysis of an euv wave observed by the euvi disk imagers on december 07 , 2007 when the stereo spacecraft separation was @xmath0 .
both a small flare and a cme were associated with the wave .
we also offer the first comprehensive comparison of the various wave interpretations against the observations .
our major findings are : ( 1 ) high - cadence ( 2.5 min ) 171 images showed a strong association between expanding loops and the wave onset and significant differences in the wave appearance between the two stereo viewpoints during its early stages ; these differences largely disappeared later , ( 2 ) the wave appears at the active region periphery when an abrupt disappearance of the expanding loops occurs within an interval of 2.5 minutes , ( 3 ) almost simultaneous images at different temperatures showed that the wave was most visible in the 1 - 2 mk range and almost invisible in chromospheric / transition region temperatures , ( 4 ) triangulations of the wave indicate it was rather low - lying ( @xmath190 mm above the surface ) , ( 5 ) forward - fitting of the corresponding cme as seen by the cor1 coronagraphs showed that the projection of the best - fit model on the solar surface was not consistent with the location and size of the co - temporal euv wave and ( 6 ) simulations of a fast - mode wave were found in a good agreement with the overall shape and location of the observed wave .
our findings give significant support for a fast - mode interpretation of euv waves and indicate that they are probably triggered by the rapid expansion of the loops associated with the cme . |
the search for continuous wave ( cw ) sources rapidly rotating neutron stars ( ns ) that emit quasi - monochromatic gravitational waves ( gw s ) is one of the most computationally intensive tasks for the data analysis of gw detectors .
surveys of wide areas of the sky and/or large frequency and spin - down ranges are computationally bound ( brady et al . , 1998 ) ; the only viable strategy is to set up hierarchical algorithms , where `` coherent '' and `` incoherent '' stages are alternated in order to maximize the signal - to - noise ratio ( snr ) , based on the cpu power available ( brady and creighton , 1999 ; schutz and papa , 1999 ) .
the algorithms investigated so far deal only with _
isolated _ sources . to search for a ns orbiting a binary companion has always been considered computationally intractable , as one would need to correct also for the doppler phase shift caused by the orbital motion of the gw source around a companion object ( star , back hole , or planet ) : the maximum time of coherent integration , before the signal power is spread over more than one frequency bin , is between @xmath0 sec ( for a source like the hulse - taylor binary pulsar ) and @xmath1 sec ( for a ns with a jupiter - size planet in a 4-months period orbit ) .
this would add five more search parameters .
nonetheless , there are several important reasons for start addressing this problem at this time : ( i ) we would like to quantify what `` intractable '' means , and estimate the computational costs as a function of the search parameters ; ( ii ) the only known gw source in the high - frequency band is sco x-1 , a ns orbiting a low - mass companion ( bildsten , 1998 ) ; if our current astrophysical understanding is correct , sco x-1 would be detectable by geo600 ( in narrow - band configuration ) and ligo , at a snr@xmath2 in two years of full coherent integration ; ( iii ) the continuous monitoring of all known ns s is planned , and around 50 radio pulsars are in a binary system ; ( iv ) we are now starting the design of software codes to search for cw sources during the first science runs carried out by the detectors , but their general structure is likely to be used for several years . the aim of this contribution is to estimate the additional processing power needed to correct for the orbital motion of a cw source orbiting a binary companion , with emphasis on targeting known ns s .
in order to disentangle the extra computational costs involved in dealing with the ns orbital parameters , we will make the following assumptions : the source location in the sky is exactly known
so that one can perfectly remove the phase doppler shift due to the detector motion and the signal is monochromatic , at frequency @xmath3 . for a general blind search of ns s
possibly in binary orbits , the total computational burden would be ( roughly ) the product of the one quoted for isolated sources , times the estimate that we present here .
the gravitational waveform is given by @xmath4}\ } , \label{h}\ ] ] where @xmath5 and @xmath6 are assumed ( as usual ) constant , and @xmath7 is the doppler phase modulation induced by the orbital motion of the source around the companion ; @xmath8 it the signal parameter vector .
we assume that the orbit is keplerian , and elliptical in shape .
the doppler correction to the phase of the signal due to the orbital motion is therefore : @xmath9\ , , \label{phid}\ ] ] where @xmath10 and @xmath11 are the polar angles describing the direction to the detector , with respect to an appropriate reference frame attached to the binary system , @xmath12 is the speed of light and @xmath13 is the eccentric anomaly .
it is related to the mean angular velocity @xmath14 ( where @xmath15 is the orbital period ) and the mean anomaly @xmath16 by the kepler equation : @xmath17 , where @xmath18 is an initial phase , @xmath19 . the additional parameters on which one must launch a search are therefore five in the elliptical case , say @xmath20 , where @xmath21 , and three in the circular orbit case , say @xmath22 . as usual
, @xmath3 is not a search parameter which requires a filter mesh . a rigorous way of estimating the search costs can be worked out by approaching the data analysis through a geometrical picture ( sathyaprakash and dhurandhar , 1991 ; dhurandhar and sathyaprakash , 1994 ; balasubramanian et al . , 1996 ; owen 1996 ; brady et al . , 1998 ) : the signal is a vector in the vector space of data trains and the @xmath23-parameter family of signals traces out an @xmath23-dimensional manifold which is termed as the signal manifold . on this manifold
one introduces a proper distance and therefore a metric @xmath24 defined as the fractional loss of snr the _ mismatch _ @xmath25 caused by the wrong choice of the filter parameters . the spacing of the grid of filters is decided by the fractional loss due to the imperfect match that can be tolerated . fixing the mismatch @xmath25 , fixes the grid spacing of the filters in the parameter space @xmath26 .
the number of filters @xmath27 is then just : @xmath28^{n}\ , v_{\cal p}\,,\quad\quad\quad v_{\cal p } = \int_{{\cal p } } \sqrt{\det ||\gamma_{ij}|| } d { \vec \lambda}\ , , \label{nf}\ ] ] where @xmath29 is the proper volume , and @xmath30 for eccentric ( circular ) orbits .
the general expressions of @xmath29 and @xmath27 , with signal model given by eqs .
( [ h ] ) and ( [ phid ] ) , are very complex . nonetheless ,
in the limit of long ( @xmath31 ) and short ( @xmath32 ) observation times , with respect to the orbital period , one can obtain analytical closed form expressions which are actually quite simple .
the expansions in these two regimes agree remarkably well with the exact expression over most of the p - range , see figure 1 .
we give here some details for the circular orbit case , and just sketch the key result for eccentric orbits .
the circular orbit case is important because it provides us several insights into the problem via a comparatively easier computation ; moreover , several binary radio pulsars have effectively @xmath33 and sco x-1 is essentially in a circular orbit ; in addition , when `` blind '' searches will be implemented , it is likely that they will be restricted to ns s orbiting a companion in circular orbit , in order to keep the computational burden affordable .
the total number of filters ( for a @xmath34 mismatch ) is given by : @xmath35 where the parameter volume one needs to cover is : @xmath36 here we have ambitiously chosen parameter values that are appropriate for a ns / ns binary in a few hours orbit . notice the radically different dependence on @xmath37 in the two regimes , and the @xmath38 dependence of the computational costs . if the parameter values are known in advance with an error @xmath39 , then the number of filters reduces by a factor @xmath40 & \quad\quad ( 2\pi t / p \gg 1 ) \nonumber\\ 2.2\times
10^{-4 } \prod_{j=1}^3\,\left [ \left(\delta \lambda^j/\lambda^j\right)/10^{-2}\right ] & \quad\quad ( 2\pi
t / p \ll 1 ) \nonumber\\ \end{array } \right .
\label{g0}\ ] ] clearly , any prior information on the value of the source parameters greatly decreases the number of templates , and it is easy to check that a single - filter search can be performed if : @xmath41 in the case of eccentric orbits , the computational burden is of course much higher , but the results show a behaviour which is very similar to the one for @xmath33 . as in the previous case ,
the costs increase dramatically as the observation time covers more than @xmath42 rad of the source orbital phase . for @xmath43
one finds : @xmath44 @xmath45 contains the dependency on the eccentricity , and @xmath46 is normalized to the parameter values given in eq .
( [ vmax0 ] ) ; notice that now , with a 5-dimensional parameter space , the number of filters is proportional to @xmath47 .
we can now apply the results of the previous section to some of the known ns in binary systems which will be targeted by laser interferometers , and estimate the processing power involved _ only _ in the coherent correction of the orbital motion .
sco x-1 orbits a low - mass companion with a period @xmath48 days , in an orbit which is essentially circular ( here we will assume @xmath33 ) ; the position of the ns on the orbit is known with an error @xmath49 rad , and the projected semi - major axis is @xmath50 cm , with @xmath51 .
the uncertainties surrounding the frequency at which gw s are emitted suggest to cover a frequency band up to @xmath52 hz .
it is easy to verify that for integration times longer than @xmath53 hours , one needs to correct for the source orbital motion .
it is also evident , from the results of the previous section , that the number of filters changes dramatically when the integration time goes from @xmath54 hours to a day ; in fact : @xmath55 notice also that if one allows for possible ( small ) departures from a perfectly circular orbit , the number of templates further increases .
we analyze now the case of radio pulsars , and assume @xmath56 sec and @xmath57 .
we have considered the 44 ns s with a binary companion ( the total number of radio pulsars is 706 ) included into the catalogue by taylor et al .
( 1993,1995 ) .
seven binary radio pulsars emit at frequencies below 10 hz , and are therefore outside the observational band ; for 21 sources the parameter measurements coming from radio observations are so precise that one can simply use the quoted values of the parameters and fully correct for the orbital motion ; the remaining 16 radio pulsars require a search over a limited parameter range . for these sources , all the ns s whose spin - down values yield upper - limits on the gw amplitude which are above the sensitivity curve of all proposed detectors ( including ligo iii ) would require at most 10 templates for @xmath56 sec .
four ns s require a number of templates in the range @xmath58 ( but are well below the ligo iii sensitivity ) and eight sources , for which we do not have as yet measurements of the spin - down , would require a very substantial number of orbital filters , more than @xmath59 .
we have presented some preliminary results regarding the additional computational costs involved in the correction of the doppler phase shift induced in the cw signal of a ns orbiting a binary companion ; we have given general expressions to compute the number of filters which are required to carry out the search as a function of the mismatch , time of integration and parameter space .
a more thorough description of the issues presented here is currently in preparation , and we plan to start soon the investigation of hierarchical algorithms , that would conceivably speed up the search in a considerable way .
balasubramanian r. , sathyaprakash b. s. , dhurandhar s. v. 1996 , phys . rev .
d 53 , 3033 | we analyze the computational costs of searches for continuous monochromatic gravitational waves emitted by rotating neutron stars orbiting a companion object . as a function of the relevant orbital parameters , we address the computational load involved in targeted searches , where the position of the source is known ;
the results are applied to known binary radio pulsars and sco - x1 . |
bundles of filamentous polymers like f - actin form primary structural components of a broad range of cytoskeletal structures including stereocilia , filopodia , microvilli , cytoskeletal stress fibers , or the sperm acrosome .
actin - binding proteins allow the cell to tailor the dimensions and mechanical properties of the bundles to suit specific biological functions .
in particular , the mechanical properties of these bundles play key roles in cellular functions ranging from locomotion @xcite to mechanotransduction @xcite and fertilization @xcite . in view of this ubiquity ,
a detailed understanding of bundle mechanics is fundamental to gaining a mechanistic understanding of cellular function @xcite .
quantifying the governing mechanical principles of these fundamental cytoskeletal constituents could also prove valuable in the design of biomimetic nanomaterials .
in - vitro experiments recently investigated the role of actin - binding proteins like fascin , @xmath0-actinin and @xmath1-plastin in mediating bundle mechanical properties @xcite .
already an inspection of bundle conformations from fluorescence microscope images makes it evident that the properties of the various crosslinking proteins must be quite distinct . while bundles formed by fascin show a compact form and remain straight over several microns , bundles formed by @xmath0-actinin or filamin are wiggly and very lose @xcite .
the mechanical properties of actin bundles formed by different crosslinking proteins was quantified by a fluctuation analysis @xcite , which measures the magnitude of their thermal fluctuations .
it was found that the apparent bundle bending stiffness can be varied over a substantial range by changing the type and relative concentration of the crosslinker .
these intriguing mechanical properties can be understood in terms of the wormlike bundle model ( wlb ) , which describes bundles as an assembly of semiflexible filaments interconnected by crosslinking proteins @xcite ; for an illustration of the bundle architecture see fig .
[ fig : bundle_illustration ] .
unlike the standard wormlike chain model ( wlc ) @xcite , the wormlike bundle model ( wlb ) exhibits a state - dependent bending stiffness @xcite that derives from a generic competition between the bending and twist stiffness of individual filaments and their relative motion mediated by the stiffness of the crosslinkers .
an important aspect of the wlb model is that crosslinks may be very efficient in constraining the lateral excursions of filaments within the bundle but much less so in inhibiting their axial motion .
this is possible as the relative axial sliding of two crosslinked filaments probes not only the elastic properties of the crosslinking protein but also those of the binding domain at which the protein is attached to the filament
. the latter may be of quite different rigidity than the protein itself .
unfortunately there are no single - molecule experiments yet which would quantify the mechanical and binding properties of actin crosslinking proteins attached to a pair of f - actin filaments . in the wlb model the mechanical properties of the crosslinks are described by a single shear - stiffness @xmath2 for the relative sliding of the constituent filaments .
an important finding within the wormlike bundle model is that the mechanical properties of bundles can be classified into three distinct bending regimes that are mediated by both crosslink type and equally importantly by bundle dimensions , namely diameter and length @xcite .
taking into account the mechanical properties of filaments and crosslinker at a microscopic level is a virtue of the model making it quite verstile .
it reduces to a well defined continuum limit but is equally applicable to bundles with as few as two filaments .
this microscopic perspective may prove a valuable starting point to address more complex questions related to bundle mechanical properties .
this may include problems such as disorder , lattice defects @xcite , or filament fracture @xcite .
while previously we have provided a formulation only for plane - bending in two dimensions @xcite , here we give a full derivation of the wlb hamiltonian in three spatial dimensions .
this includes bending as well as twist deformations .
we explore the predictions of the wlb model for experimental observables such as the tangent - tangent correlation function and dynamic response functions .
moreover , we discuss the effects of different bundle geometries on their mechanical properties . in this respect
, we will view microtubules as a bundle of ( proto-)filaments arranged on the surface of a cylinder ; compare fig .
[ fig : quantities_explained ] .
this shell - like bundle architecture is contrasted with a uniform distribution of filaments as found for f - action bundles @xcite .
finally , we discuss how helicity influences bundle mechanics .
th filament in the cross - sectional plane ( @xmath3-plane ) is given by @xmath4 .
_ f - actin bundle architecture _ ( left ) : filaments are arranged on a square lattice .
_ microtubule architecture _
( right ) : filaments are arranged on the surface of a cylinder.,title="fig : " ] th filament in the cross - sectional plane ( @xmath3-plane ) is given by @xmath4 .
_ f - actin bundle architecture _ ( left ) : filaments are arranged on a square lattice .
_ microtubule architecture _
( right ) : filaments are arranged on the surface of a cylinder.,title="fig : " ] th filament in the cross - sectional plane ( @xmath3-plane ) is given by @xmath4 .
_ f - actin bundle architecture _ ( left ) : filaments are arranged on a square lattice .
_ microtubule architecture _
( right ) : filaments are arranged on the surface of a cylinder.,title="fig : " ] th filament in the cross - sectional plane ( @xmath3-plane ) is given by @xmath4 .
_ f - actin bundle architecture _ ( left ) : filaments are arranged on a square lattice . _ microtubule architecture _
( right ) : filaments are arranged on the surface of a cylinder.,title="fig : " ]
we consider bundles of length @xmath5 that consist of @xmath6 parallel filaments .
while the filaments may form a disordered lattice structure , we will focus our attention here to the cases of regular arrangements .
in particular we will treat in detail the square - lattice and the cylindrical tube ( see fig . [
fig : quantities_explained ] ) .
each filament is modeled mechanically as an extensible worm - like polymer with stretching stiffness @xmath7 , bending rigidity @xmath8 and twist rigidity @xmath9 .
filaments are irreversibly crosslinked to their nearest - neighbors by crosslinks with mean axial spacing @xmath10 .
the crosslinks are modeled to be compliant in shear along the bundle axis with finite shear stiffness @xmath2 , and to be inextensible transverse to the bundle axis , thus constraining the interfilament distance , @xmath11 , to be constant ( see fig .
[ fig : quantities_explained ] ) .
this assumption , which neglects crosslink stretching deformations , is based on the recognition that the shearing mode involves deformation of the crosslink _ and _ its binding domain to the filament .
the resulting stiffness may indeed be much lower than that of a crosslink in isolation .
filament stretching is characterized by the axial displacement @xmath12 of filament @xmath13 at axial position @xmath14 along the backbone . to describe bundle bending and twisting we define @xmath15 to be a material frame fixed to the bundle central line at each arclength position @xmath14 .
the vector @xmath16 is the tangent to the space curve traced out by the central line , while the two vectors @xmath17 and @xmath18 lie within the cross - section of the bundle .
the position of each filament in the cross - section is parametrized in terms of a vector @xmath19 , where @xmath20 and @xmath21 are the material - frame coordinates of filament @xmath13 ; they are constants independent of arclength @xmath14 and deformation of the central line ( see fig . [
fig : quantities_explained ] ) .
as one moves along the bundle backbone the material frame rotates according to frenet - seret equations , @xmath22 .
the rate of rotation is given by the generalized curvatures @xmath23 , which , in addition to the axial displacement @xmath24 , represent the basic kinematic degrees of freedom of the bundle .
neglecting all nonlinear effects we are now going to develop a simple expression for the bundle energy which is harmonic in its degrees of freedom , axial displacement @xmath24 and generalized curvatures @xmath25 .
this wlb hamiltonian consists of three contributions , @xmath26 .
the first term corresponds to the standard wormlike - chain hamiltonian @xmath27\,.\ ] ] writing this we assume that each filament follows effectively the same space - curve as the center - line .
while , in general , one should account for the curvatures , @xmath28 , of each individual filament separately , this would only lead to correction factors that can be neglected for our purposes .
consider , for example , planar bending of the central line , @xmath29 , where @xmath30 is the radius of curvature .
the filaments that lie at a distance @xmath31 away from the central line naturally have a different radius of curvature , @xmath32 , and thus a different bending energy .
the magnitude of the correction term relative to @xmath33 is small since it scales as @xmath34 , i.e. we assume the typical curvatures to be smaller than the bundle radius ( for more details see appendix [ sec : relat - betw - omeg ] ) . as to the twist degree of freedom , in eq .
( [ eq : h0_omega ] ) we do not allow for the possibility of relative twisting of the individual filaments ( see section [ sec : pretwisted - bundles ] and appendix [ sec : relat - betw - omeg ] for a discussion of this effect ) .
we assume that twist is only due to the `` bundle - twist '' @xmath35 of the central line .
this assumption is reasonable in tightly bound bundles , where the filaments are connected by many crosslinks . in this state
the filaments and their relative orientation can be assumed to be locked - in by the crosslinker .
the second term in the hamiltonian , @xmath36 , accounts for filament stretching .
it depends on the difference in axial displacement , @xmath24 , between two crosslinks at arclength positions @xmath37 and @xmath38 , respectively .
@xmath39 ^ 2\nonumber\\ & \to&\frac{k_s\delta}{2}\int_0^lds \left({\frac{\partial u_k}{\partial s}}\right)^2\,,\end{aligned}\ ] ] where we have performed the continuum limit @xmath40 to arrive at the second line .
the spring constant @xmath41 is the single filament stretching stiffness on the scale of the crosslink spacing @xmath10 .
the particular form for @xmath7 depends on the system under consideration . for high crosslink concentrations ( small @xmath10 )
, the segment behaves as a homogeneous elastic beam , characterized by a youngs modulus @xmath42 and @xmath43 .
the combination @xmath44 that enters the hamiltonian is independent of @xmath10 , as it should : the mechanical stretching stiffness of a beam can not depend on the properties of the crosslinks . for small concentrations of crosslinks ( large @xmath10 ) entropic effects become relevant and
the stretching stiffness is that of a thermally fluctuating wormlike chain with persistence length @xmath45 . in this case one
has @xmath46 , which implies that the combination @xmath47 _ does _ depend on the crosslink spacing @xmath10 .
this is related to the fact that the formation of a crosslink suppresses thermal undulations ( reduces entropy ) and thus increases the entropic stretching stiffness . equating both stretching stiffnesses
, @xmath48 one finds the critical crosslink concentration , @xmath49 , at which the cross - over from enthalpic to entropic elasticity takes place . in the case of microtubules , which will be treated in section [ sec : micr - nanot ] , the crosslink spacing @xmath10 is given by the tubulin - size and the stretching stiffness is modeled as for an elastic beam . the third energy contribution , @xmath50 , results from the crosslink - induced coupling of neighboring filaments .
the relative axial motion of a filament pair @xmath51 at a given point of the backbone is described by the crosslink shear displacement , which is the sum of a geometric contribution , @xmath52 , and the relative stretching of neighboring filaments , @xmath53 .
the geometric part results from the arclength mismatch between the two filaments , induced by a bending and twisting of the bundle central line . as in eq .
( [ eq : hstretch ] ) we first write the shear energy as a sum over all crosslink positions @xmath37 and then perform the continuum limit , to get @xmath54 ^ 2\nonumber\\ & \to&\frac{k_\times}{2\delta}\int_0^lds \left ( \delta_{lk } + \delta u_{lk } \right)^2\,,\end{aligned}\ ] ] where @xmath2 is the shear stiffness of the individual crosslink . for any bundle deformation
, the associated value of @xmath52 can be compensated for by stretching the filaments , making the shear energy vanish when @xmath55 . at the same time , however , this would increase the stretching energy , which may be unfavorable if the stretching stiffness @xmath7 is rather large . for deformations on the scale of the bundle length ( @xmath56 )
the ratio of both energies gives the important parameter @xmath57 , which quantifies the relative strength of both deformation modes @xcite . as a final ingredient to the model
we need to calculate the dependence of @xmath52 on the bundle curvatures , @xmath25 . without going into the details of an explicit derivation
, we here just give the resulting expression . for more details
we refer the reader to appendix [ sec : deriv - shear - deform ] .
the special case relevant for the microtubule geometry is also dealt with in some detail in @xcite . to linear order in @xmath25 we find @xmath58 where we defined @xmath59 as the distance between the filament pair and @xmath60 as the angle of @xmath61 with respect to the @xmath62-axis .
furthermore , @xmath63 are defined as the cross - sectional coordinates of the _ midpoint _ between the filament pair .
in contrast to @xmath64 , which is an expansion in the generalized curvatures @xmath25 , the shear energy @xmath65 is a function of the _ integrated _ curvatures , since @xmath66 . for terms beyond the harmonic contribution to be negligible
, one has to assume that the shear displacement is sufficiently small , @xmath67 , where @xmath68 is some microscopic length - scale related to the size of the crosslink . in terms of the bundle curvatures this implies @xmath69 , which is much more restrictive than the range @xmath70 , over which @xmath64 can be approximated by a harmonic form . as a consequence
the bundle is only allowed to make small excursions from its initial state , an assumption which is usually well satisfied in bundles of stiff polymers like actin or microtubules , but certainly breaks down under extreme loading conditions ( e.g. to describe post - buckling ) or in more flexible objects like dna . with this `` weakly - bending '' assumption
we reformulate the generalized curvatures in terms of the lab - frame euler angles @xcite @xmath71 as reference state we take @xmath72 , @xmath73 and @xmath74 , which corresponds to a straight , but pre - twisted bundle that points along the @xmath75-axis . for small excursions around this reference state we can linearize the equations such that @xmath76 and the angles are now measured relative to the reference state .
as expected the pretwist @xmath77 leads to a coupling of the angles @xmath78 and @xmath79 . in the following
we are primarily concerned with the case of vanishing pretwist .
then , the coupling terms vanish and we can simply set @xmath80 we will come back to the case of pretwist in section [ sec : pretwisted - bundles ] . for the purpose of illustration
we provide some examples of how the shear displacement @xmath81 depends on the geometry of the bundle cross - section and the deformation of its central line .
if the bundle consists of only two filaments @xcite ( geometry of a ribbon ) we have @xmath82 as the central line and the mid - line between the two filaments are identical . in this case
@xmath81 simplifies to @xmath83 which is illustrated in fig . [ fig : delta_two_filaments ] .
note , that twist ( @xmath35 ) does not contribute , as both filaments twist around the central line symmetrically .
leads to the arclength mismatch , @xmath84 .
the filaments have to stretch the relative amount @xmath85 , in order to keep the crosslink ( dashed line ) undeformed with zero shear energy . ] ) taking an angle @xmath86 with the cylinder axis @xmath87 .
the resulting arclength mismatch is given by @xmath88 .
] a particular case of this two - filament bundle has been considered in a set of articles @xcite , where the crosslinks are assumed to be rigid with respect to shear , i.e. @xmath89 . to satisfy
a vanishing arclength mismatch @xmath81 one thus requires @xmath90 .
this means that the unit vector @xmath18 , which points from one filament to the other , must not rotate in the direction of the tangent @xmath91 . in other words
, @xmath18 must equal the the bi - normal .
thus the ribbon orientation is completely specified by the space curve traced by the central line .
a second example is the axoneme in eukaryotic flagellae @xcite .
there , filaments ( microtubule doublets ) are arranged on the surface of a cylinder , just as the protofilaments in a single microtubule .
switching to polar coordinates in the cross - sectional plane ( see fig .
[ fig : quantities_explained ] ) , @xmath92 , @xmath93 , we find @xmath94 a similar expression , disregarding the possibility of twisting , has been given by mohrbach _
et al . _
the structure of the second and third term is the same as in eq .
( [ eq : delta_two ] ) , additionally taking into account the modified orientation in the cross - sectional plane , as described by the angle @xmath95 . the origin of the first term is illustrated in fig .
[ fig : delta_mt ] and elaborated on in appendix [ sec : deriv - shear - deform ] .
consider now the case of planar bending .
this will make the connection to continuum elasticity particularly clear . under planar
deformation the bundle is described by the one variable @xmath96 such that the shear deformation is simply @xmath97 . here
, @xmath98 is the displacement of the bundle transverse to the bundle axis ( in the y - direction ) .
this latter form makes clear that the shear deformation represents one part of the strain tensor component @xmath99 . the second part , @xmath100 is the continuum version of relative filament stretching ( see eq .
( [ eq : hshear ] ) ) .
the elastic symmetries relevant to the wlb model depend on the arrangement of the filaments in the cross - section .
if there is rotational symmetry with respect to the bundle axis , one speaks of transversly isotropic elastic bodies @xcite . while , in general , this has five first order elastic constants , our model has only two , augmented by the ( second order ) bending / twisting elasticity of the individual filaments which is not accounted for in continuum elasticity .
the simplification arises from assuming transverse inextensibility as well neglecting cross - sectional shape changes .
the latter is , for example , important in the failure of hollow tubes under bending .
one relevant effect is the brazier effect @xcite , which describes the increasing ovalization of the cross - section under the action of a bending moment . in the present formulation of the model these nonlinearities
are not accounted for . for a discussion of potential modifications to include cross - sectional deformations ,
we refer the reader to the outlook section at the end of this article
. it should be clear from the way the hamiltonian was derived that the model is applicable to bundles with arbitrary ( ordered / disordered ) arrangements of filaments in the cross - section . in the remainder of this article
we will focus our attention to bundles with highly symmetric cross - sections , where the filaments either form a rectangular array or a hollow tube . in view of recent experiments probing the mechanical or statistical properties of individual bundles in vitro @xcite , we head at a description of the bundle in terms of effective bending and twist rigidities .
these are defined with respect to the standard worm - like chain model .
to arrive at the proper expressions we have to integrate out the internal stretching variable @xmath101 , which in general is not observable in experiment . to show how this works , we symbolically write the partition function as @xmath102 where @xmath103 signifies the set of euler angles @xmath104 .
the constrained partition function then reads @xmath105 . the integration over the @xmath101-variables
can easily be performed .
as the hamiltonian is harmonic we are left with only gaussian integrals , which are evaluated in fourier space .
the resulting potential of mean force @xmath106 can be written in the form of a wormlike chain hamiltonian @xmath107\ ] ] with effective bending and twist rigidities @xmath108 and @xmath109 , respectively .
we note that in the symmetric situations considered here there is no coupling between the different deformation modes bending and twisting ( see appendix [ sec : exact - solution-2d ] ) .
in contrast to the usual wlc , the effective bend and twist rigidities are in general dependent on the mode - number @xmath110 and thus on the wavelength of the deformation . this effect and
the discussion of its consequences is the central topic of the remaining sections .
in the following sections we will focus our attention to bundles with @xmath111filaments that form a rectangular array ( see fig . [ fig : quantities_explained ] ) . the angle @xmath0 that specifies the orientation of the filament - pair in the cross - section is then @xmath112 as filaments are either arranged along the @xmath113- or the @xmath62-axis . as mentioned above the different deformation modes decouple in harmonic order .
we can thus investigate bending independently from twisting .
also the two space - directions decouple and we can reduce the model to an effective two - dimensional description @xcite .
the shear hamiltonian reduces to @xmath114 where we used eq .
( [ eq : delta_mismatch ] ) with @xmath115 . by following the recipe outlined above
we eliminate the axial strain variable @xmath24 . by approximating @xmath24 by a linearly increasing function of @xmath13 ( see discussion below ) we arrive at the following result for the effective bending rigidity should be substituted by @xmath116 .
also the length - scale @xmath117 acquires a more complex ( but essentially equivalent ) dependence on bundle - size @xmath6 .
] as defined in eq .
( [ def_eff_rigidities ] ) @xmath118\,.\ ] ] here , we have defined a characteristic wavelength @xmath119 and a dimensionless bending rigidity @xmath120 . in terms of the quantities @xmath117 and @xmath121
the previously defined @xmath57 can be rewritten as @xmath122 .
if the filaments behave as homogeneous elastic beams , @xmath121 is just a number independent of bundle geometry or crosslink spacing . for any numeric computation we will , for specifity , assume that @xmath123 , which corrresponds to beams with square cross - sections @xcite . the characteristic feature of eq .
( [ eq : kappalinear ] ) , is the wavelength dependence ( see fig . [
fig : kappabend_q ] ) . for wavelengths
@xmath124 in the interval @xmath125 the bending stiffness decreases as @xmath126 . in ref .
@xcite we have termed this the intermediate or shear dominated regime as the bending rigidity is proportional to the shear stiffness of the crosslinks .
it is in this parameter regime that the bundle behaves qualitatively different than either a homogeneous beam ( obtained in the `` fully coupled '' limit of @xmath127 ) or an assembly of `` decoupled '' filaments ( @xmath128 ) . another important feature of eq .
( [ eq : kappalinear ] ) , which is independent of the specific @xmath129-dependence , is the ratio of maximal to minimal bending rigidity , @xmath130 .
this only depends on the number of filaments and the dimensionless bending rigidity @xmath121 . on @xmath13 , the distance from the bundle central line . decreasing the dimensionless shear - stiffness @xmath131 the strain is reduced but not in a linear fashion .
the outer layers of the bundle remain stretched stronger than the inner ones . ]
while integrating out the stretching variables @xmath24 can be performed exactly , eq .
( [ eq : kappalinear ] ) is based on the additional assumption that axial strains are linearly increasing through the cross - section , @xmath132 . the exact profile for @xmath24
is calculated in the appendix and displayed in fig .
( [ fig : axialstrain ] ) ; compared to the linear profile it shows an enhancement of strain towards the bundle periphery .
however , the ensuing value for the bending stiffness is largely insensitive to the linear approximation for a discussion of the error in the continuum limit . in general , the error is smaller for smaller bundles and vanishes for a bundle consisting of only two filaments . ] .
we speculate that the nonlinearities in the axial strain may eventually be important for nonlinear material properties , as for example strain induced rupture .
the increased strain in the outermost filaments brings them closer to their threshold for rupture and thus makes them more susceptible to this mode of failure .
in order to make contact with continuum models for beam bending we perform a continuum limit , by letting @xmath133 but keeping the bundle aspect ratio @xmath134 constant . then , bundle length @xmath5 has to grow with @xmath135 as @xmath136 .
in particular , this implies that fewer and fewer modes @xmath110 belong to the decoupled regime ( where @xmath137 ) .
eventually , this regime , where filaments bend independently ( @xmath138 ) is no longer accessible . in effect
this means that the bending stiffness @xmath8 of the individual filaments can be neglected , just as in `` normal '' continuum elasticity , where higher order gradients ( @xmath139 ) are not accounted for from the start . in this continuum limit
the result from the linearization assumption , eq .
( [ eq : kappalinear ] ) , reduces to the timoshenko model for beam bending @xcite , which was recently used to interpret bending stiffness measurements on microtubules @xcite and carbon nanotube bundles @xcite , @xmath140 to allow for comparison with continuum elasticity we have used the expressions @xmath141 and @xmath142 applicable for homogeneous beams of square cross - section and defined the shear - modulus @xmath143 . on the other hand
, one can equally derive a continuum limit from the exact expression for @xmath108 ( as presented in appendix [ sec : bending ] ) .
this gives @xmath144 both expressions , eq .
( [ eq : kappatimo ] ) and eq .
( [ eq : kappacontinuum ] ) are compared in fig .
( [ fig : timo.wlb ] ) .
one infers that the approximation ( timoshenko theory ) overestimates the exact solution by no more than @xmath145 .
the difference can partly be compensated for by introducing a shear - correction factor ( @xmath146 ) in the denominator of eq .
( [ eq : kappatimo ] ) . in this section
the implications of a mode - number dependent bending rigidity is further elaborated by discussing the concept of the persistence length .
the persistence length of a single wlc may be defined in terms of the competition of bending and thermal energies , @xmath147 . with this definition , bending rigidity and persistence length are basically identical . in the framework of the wlb
this would lead to a mode - number dependent persistence length @xmath148 .
for the wlc the persistence length is also the length - scale over which the tangent - tangent correlation function decays @xmath149\,.\ ] ] this simple exponential form is no longer valid for the wlb as can be illustrated by considering planar bending with @xmath150 .
then the tangent - tangent correlation function is easily inferred from the angular fluctuations @xcite as @xmath151 \,,\ ] ] with @xmath152 \,,\ ] ] and @xmath153 .
forcing such a complex expression into the form given by eq .
( [ eq : tagengt - tangent ] ) implies an arclength - dependent persistence length , @xmath154 . at short distances ,
one recovers the decoupled regime and @xmath155 , while at long distances , @xmath156 as found in the fully coupled regime .
note , that there is no immediate relation between this @xmath154 and the @xmath157 defined above .
the fourier - transform of @xmath157 is , instead , given by the following expression @xmath158 this quantity appears in the elastic energy expressed in real - space as @xmath159 which is a non - local function of arclength .
the length - dependence obtained here is markedly different from that found in the correlation function , eq .
( [ eq : lp_effective ] ) .
while @xmath154 is constant at large distances , @xmath160 decays exponentially and vanishes over the length - scale @xmath161 , which corresponds to the onset of the fully - coupled regime .
a similar nonlocal energy function is obtained when considering the fluctuation properties of elastic membranes .
in - plane shear- and compression - modes lead to a renormalized bending rigidity for the out - of - plane fluctuations @xcite .
in contrast to bundles , however , there the kernel is long - ranged @xmath162 , which asymptotically leads to a flat membrane phase . concluding this section
, we find that it is impossible to speak of a single persistence length without specifying the precise experimental conditions as well as the observable under consideration ( here tangent - tangent correlation function ) .
the wlb model presents a framework within which a length - dependent bending stiffness can be understood .
however , we would like to emphasize that the fundamental quantity is the @xmath163-dependent @xmath164 as presented in eq .
( [ eq : kappalinear ] ) or below in eq .
( [ eq : kappamt ] ) .
it depends on wavenumber in a universal way independent of the type of measurement .
the dependence on bundle length , in contrast , arises through a specific transformation to real space and may produce different results depending on the observable under consideration . in our previous publications
@xcite we have calculated several thermodynamic observables and showed that the mode - dependent bending stiffness of a wlb may lead to drastic modifications of their scaling behavior . here , we widen the scope of this analyis by discussing dynamic observables . in analogy to the usual overdamped dynamics of a wlc
we can discuss the dynamics of a wlb by substituting the mode - dependent bending stiffness in the standard langevin equation of motion for the transverse bending modes @xmath165 @xmath166 with this one obtains for the reduced correlation function @xmath167 with the relaxation times @xmath168 .
for wlcs ( constant @xmath169 ) one finds a scaling regime at times @xmath170 , where @xmath171 @xcite . for wlbs , on the other hand
, one has to use the @xmath129-dependent effective bending stiffness , eq .
( [ eq : kappalinear ] ) , instead . as the term @xmath172 multiplies the @xmath173 contribution in eq .
( [ eq : corrfct ] ) one finds the correlations to grow in time as @xmath174 .
of course , this result is only valid as long as the @xmath175-term dominates the effective bending stiffness , i.e. as long as the bundle is in the intermediate regime .
in fact , there will be a complex cross - over scenario @xmath176 with the cross - over times @xmath177 and @xmath178 governing the cross - over from the decoupled to the intermediate and the fully - coupled regimes . at times larger than @xmath179 the correlation function saturates . from the correlation function
one can furthermore calculate the response function @xmath180 measuring the linear response to transverse forces . using the fluctuation - dissipation theorem and the kramers - kronig relations one finds @xmath181
this contrasts with the response function @xmath182 for stretching forces @xmath183 which is sometimes taken as a starting point to determine the high - frequency response of entangled solutions of stiff polymers @xcite , the transverse response , on the other hand , has been argued to relate to a `` microrheological modulus '' @xcite that is measured locally by imbedding probe beads into the network .
unfortunately , for a wlc both functions are hardly indistinguishable , and only differ by the constant factor @xmath184 .
the high - frequency behavior in both cases is @xmath185 . in the case of a wlb , things are somewhat different , as the additional factor @xmath186 in the longitudinal response function not only changes the prefactor but also modifies the functional form with respect to frequency ( see fig .
[ fig : chiperppar ] ) .
( top ) with longitudinal response function @xmath187 ( bottom ) for a bundle of @xmath188 filaments ( in units of @xmath189 and assuming @xmath190 ) .
frequencies are plotted in units of @xmath191 .
the different curves correspond to different values of @xmath192 .
the two asymptotic scaling regimes with @xmath193 correspond to the decoupled ( top ) and fully - coupled regime ( bottom ) , respectively . the intermediate @xmath194 is sharper in @xmath180 than in @xmath182 even though the latter function shows overall a stronger variation with frequency.,title="fig : " ] ( top ) with longitudinal response function @xmath187 ( bottom ) for a bundle of @xmath188 filaments ( in units of @xmath189 and assuming @xmath190 ) .
frequencies are plotted in units of @xmath191 .
the different curves correspond to different values of @xmath192 .
the two asymptotic scaling regimes with @xmath193 correspond to the decoupled ( top ) and fully - coupled regime ( bottom ) , respectively .
the intermediate @xmath194 is sharper in @xmath180 than in @xmath182 even though the latter function shows overall a stronger variation with frequency.,title="fig : " ] for similar reasons as in the discussion of the correlation function one expects an intermediate regime with @xmath194 , at least in the transverse response .
this should lead to measurable signatures in microrheological experiments on bundled f - actin systems .
due to the additional @xmath129-dependence in the denominator , the longitudinal response , @xmath182 shows a smooth cross - over between the two asymptotic regimes of fully - coupled and decoupled bending , as explained in the figure .
we now turn to the discussion of the twist mode . in this case
the euler angles @xmath79 and @xmath78 in eq .
( [ eq : delta_mismatch ] ) are zero such that the shear hamiltonian reduces to @xmath195 where we defined the geometric factor @xmath196 . as may already be apparent from comparing eq .
( [ eq : hbendingonly ] ) with eq .
( [ eq : htwistonly ] ) the stretching deformation @xmath101 couples differently to twist ( @xmath197 ) as to bending ( @xmath198 ) . the difference being the additional derivative occuring in eq .
( [ eq : htwistonly ] ) .
effectively this means that the resulting twist rigidity will not depend on mode - number @xmath163 but receive a constant correction to the single filament value @xmath199 .
let us first assume that the filaments are inextensible .
then the @xmath101-terms in the hamiltonian identically vanish and the effective twist rigidity can simply be read off from the terms multiplying @xmath200 , @xmath201\,.\end{aligned}\ ] ] here we have defined @xmath202 , which is similar to @xmath117 defined above , with @xmath8 substituted by @xmath9 .
unlike @xmath117 , however , it multiplies the constant length @xmath11 , the lateral distance between neighboring filaments .
as anticipated , there is no mode - number dependence . by introducing bundle diameter @xmath203 and effective shear modulus @xmath204 ( as in the discussion leading to eq .
( [ eq : kappatimo ] ) ) the second term takes a form well known from continuum theory , @xmath205 @xcite .
thus , we find an effective twist rigidity , eq .
( [ eq : kappatwist ] ) , that just describes the simple additive superposition of two contributions , shear - induced rigidity ( second term ) and twist rigidity of the individual filaments ( first term ) . in the continuum limit @xmath133 the latter contribution can naturally be neglected as it only grows with @xmath6 as compared to the @xmath206 increase of the shear - induced rigidity .
now we allow for finite axial displacements @xmath24 .
this is commonly referred to as cross - sectional warping ; under twist deformations the bundle cross - sections do not stay plane but deform and acquire a curvature .
just as in the case of the bending rigidity , the exact solution for the twist rigidity only differs little from the foregoing simplified analysis .
some details about the derivation are presented in appendix [ sec : twisting ] .
the resulting axial displacements @xmath24 can be found in fig .
( [ fig : cont_limit_twist ] ) the classic solution of saint - venant @xcite is approached closer and closer for increasing the shear - stiffness @xmath2 .
( @xmath207 , corresponding to one row of the rectangular array ) for different non - dimensional crosslink shear stiffness @xmath208 . the classic solution for a beam ( dashed line )
is approached as @xmath89 . in the opposite limite , @xmath209 , there is no warping and @xmath210 , as filaments remain uncoupled in this limit . ]
in this section we want to turn our attention to the case of microtubules , which we model as bundles with filaments arranged on the surface of a cylinder . for the time being we assume that in the groundstate the microtubule is untwisted such that the @xmath6 protofilaments are oriented along the cylinder axis .
this assumption is in fact only valid for microtubules with @xmath211 protofilaments @xcite .
this class , nevertheless , seems to be the most frequent in in - vitro polymerization experiments .
there is an ongoing debate in the literature about the dependence of microtubule bending rigidity on length @xcite .
early buckling experiments indicated a length - dependence @xcite , while an improved version of the same experiments later gave a negative result @xcite .
@xcite performed a mode analysis of microtubule contours .
their data is compatible with a length - dependent bending rigidity but the authors vote for a cautious interpretation of the results in view of the close proximity to the noise level .
recently , experiments by pampaloni _
@xcite measured the transverse fluctuations of grafted microtubules to establish an increasing persistence - length for microtubule lengths up to @xmath212 .
using a high - precision tracer technique , taute _ et al . _
@xcite analyzed shorter microtubules and found the persistence length to level off at @xmath213 for lengths shorter than @xmath214 .
similar values have been obtained in ref .
@xcite ( @xmath215 m ) and in ref .
@xcite ( @xmath216 m ) , and explained with the help of the wlb model . on short length - scales ( in the decoupled regime )
the effective bending rigidity is constant because it reflects the stiffness of the individual proto - filaments . in contrast , kis _
et al . _
@xcite have found a decreasing bending stiffness even for microtubule sections as short as several hundred nanometers . while the discussion about these partially conflicting measurements is ongoing , we would like to point out that different techniques do not necessarily have to come to the same conclusion , once the idea of the bending rigidity as a fundamental material parameter is given up . in the context of the wlb model
the fundamental quantity is a mode - dependent bending rigidity . as discussed in section [ sec : tang - tang - corr ] , any dependence on bundle length is a `` secondary '' effect that will depend on the observable under consideration . by using the same procedure as in the case of the rectangular bundle we find for the microtubular bending rigidity @xmath217\,,\ ] ] where the relevant length - scale @xmath117 is now defined as @xmath218 .
note , that this expression results in the bending stiffness in the fully - coupled regime ( where the microtubule behaves as a simple beam ) to scale as @xmath219 , in contrast to bundles with a homogeneous cross - section ( as the one discussed before ) , where @xmath220 . with eq .
( [ eq : kappamt ] ) direct comparison with experimental data can be made .
in particular , the ratio of maximal to minimal bending rigidity can most easily be determined , @xmath221 . for the latter equality we assumed the protofilament to have a circular cross - section ( @xmath222 ) . for microtubules with @xmath211 protofilaments
this results in a universal ratio @xmath223 . in this way , using the the range of values @xmath224 mm ( which we assume to represent the protofilament stiffness ) one can estimate the maximal persistence length of long microtubules to be in the range @xmath225 mm .
compared with experiments @xcite these values seem to be too large . given the large error - bars in any of the mentioned experiments this calculation may , nevertheless , be acceptable .
furthermore , as we will discuss below , microtubule helicity can provide a mechanism to reduce the apparent persistence length by reducing the effect of shear - induced coupling . for given microtubule length the persistence length
is then predicted to decrease with increasing helicity thus improving the comparison with experiment .
finally , let s turn to the case of pure twist deformations .
due to the symmetry of the circular cross - section no warping is possible .
we find for the microtubule twist rigidity , similar to eq .
( [ eq : kappatwist ] ) , @xmath226\ , , \end{aligned}\ ] ] where @xmath227 .
in the previous sections we have restricted our attention to non - helical bundles and assumed that in the ground - state the filaments point along the bundle axis . as a final application of our model
, we will here discuss the question of helicity , or pretwist , and its influence on bundle mechanics .
this aspect is important not only for some types of microtubules but also for f - actin bundles and is reflected in the role that helicity plays in providing an explanation for the existence of a ( thermodynamically ) preferred bundle size @xcite .
the discussion of pretwisted bundles proceeds in two steps .
we first assume the bundle to form with all filaments straight . starting from this reference state crosslink binding
may add a driving force for twisting the bundle if the straight state does not allow for optimal accessibility of the crosslink binding sites .
another effect may be the helicity of the filaments themselves that favor a twisted bundle over an untwisted one . without elaborating on the precise mechanism that leads to pretwisted bundles , we incorporate bundle pretwist by substituting the generalized twist - curvature @xmath35 by @xmath228 . in doing
so the new energetic ground - state is at @xmath229 and @xmath230 . to obtain the effective bending rigidity we linearize the curvatures around this ground - state , as performed in eq .
( [ eq : omega ] ) . inserting this result into the shear deformation , eq .
( [ eq : delta_mismatch ] ) , one finds terms like @xmath231 which depend nonlinearly on arclength @xmath14 . thus a transformation to fourier - space is no longer helpful as different modes would remain coupled .
we therefore resort to an alternative approach , and determine the effective bending rigidity by numerical integration of the mechanical equilibrium equations in real space , @xmath232 .
specifically , we determine the end - deflection @xmath233 of a bundle of 4 filaments under a point force @xmath234 at the distal end ( @xmath235 ) , given clamped boundary conditions at the proximal end ( @xmath236 ) . the effective bending rigidity
is then obtained from the expression , @xmath237 , and displayed in fig .
( [ fig : pretwist ] ) as a function of the crosslink shear stiffness , @xmath2 , and a series of values for the pretwist , @xmath238 . for simplicity
we have assumed the filaments to be inextensible , @xmath239 , and thus restricted ourselves to the decoupled and the intermediate regimes . for increasing pretwist the apparent stiffness decreases and asymptotically approaches the value without shear - stiffness .
thus , in pretwisted , helical bundles the crosslinks only have a limited ability to mechanically couple the different filaments together .
the higher the twist the more the filaments act as if they were independent @xcite .
the reason for this behavior is that in pretwisted bundles the filaments exchange their place and those that start on the top of the bundle soon are on the bottom .
thus , crosslink sites that would stay behind and lead to large shear displacements in untwisted bundles , can now catch up to make the effective shear deformation smaller . ) as a function of shear stiffness @xmath2 ( in units of @xmath240 ) of a bundle of four filaments under a tip force @xmath234 .
the bending stiffness is calculated from the determined end - deflection @xmath233 by @xmath241 .
the full curves are for finite pretwist @xmath242 , while the dashed curves represent the limiting cases of zero and infinite pretwist , repsectively . ]
we have presented a detailed study of the elastic and dynamic properties of bundles of semiflexible filaments ( wormlike bundle model , wlb ) .
it is found that a competition between the elastic properties of the filaments and those of the crosslinks leads to renormalized effective bend and twist rigidities that can become mode - number dependent .
the strength and character of this dependence varies with bundle architecture , such as the arrangement of filaments in the cross section and pretwist .
two paradigmatic cases of bundle architecture have been discussed ( see fig . [
fig : quantities_explained ] ) : the first assumes filaments to be arranged homogeneously throughout the cross - section , for example on a square or triangular lattice .
this geometry is particularly relevant for f - actin bundles .
the second architecture has the filaments arranged on the surface of a cylinder as is the case for microtubules . for all bundle architectures ,
the bending rigidity depends on mode - number @xmath163 as @xmath126 .
this is the shear dominated regime , as the bending rigidity is proportional to the shear stiffness , @xmath2 , of the crosslinks @xcite .
it is in this parameter regime that the bundle behaves qualitatively different than either a homogeneous beam ( obtained in the `` fully coupled '' limit ) or an assembly of `` decoupled '' filaments .
each architecture has its own universal ratio of maximal to minimal bending rigidity , independent of the specific type of crosslink induced filament coupling . for microtubules ( without pretwist )
we find the ratio @xmath243 , which is in reasonable agreement with the available experimental data .
an important factor in determining the strength of crosslink - induced filament coupling is the pretwist ( helicity ) of the bundle .
numerical computation shows that the effective bending rigidity decreases with increasing the pretwist .
this has interesting consequences for microtubules , where the amount of pretwist depends on the number of protofilaments , @xmath6 .
different microtubule types are therefore predicted to have different variations in their effective bending rigidity .
these predictions could be tested in experiments that are able to select the microtubule type and measure their bending rigidity independently .
we have discussed several further observables , static and dynamic , that could be relevant to experiments .
we have shown that the concept of the persistence length becomes ambiguous and depends on the observable used .
the usual definition via the tangent - tangent correlation function is shown to lead to a persistence - length that depends on the scale of observation .
further observables that are effected by the mode - dependent bending rigidity are the force - extension relation or the buckling force @xcite .
interestingly , in the intermediate regime the latter is constant and independent of bundle length .
the dynamic properties of bundles are characterized by a complex cross - over scenario which is in one - to - one correspondence with the three regimes of decoupled , intermediate and fully - coupled bending . while decoupled and fully - coupled bending display the usual @xmath244 in the correlation function , it is shown that the shear - coupling leads to an intermediate asymptotic regime , where the correlations only grow as @xmath245 .
the response functions for longitudinal and transverse forces also reflect these different regimes .
in contrast to the wlc , they are not just proportional to each other but show distinct frequency dependences
. these findings may be relevant for microrheological experiments , with imbedded bead particles directly coupling to the transverse bundle fluctuations .
possibilities for future studies may be to look into the effects of filament fracture or lattice defects .
the elastic energy represents a harmonic approximation which should be extended to include nonlinear effects . especially for microtubules
one may expected nonlinear effects to play an important role in bundle mechanics .
for example , it may be important to consider that protofilaments in their unstressed state are not straight but bend radially outwards .
aditional complications could also arise from the fact that some microtubules are not transversly isotropic as we have assumed here , but have a `` seam '' , where protofilaments are offset relative to their neighbors .
failure modes under axial compressive forces have been discussed in a model for microtubules that starts from a transversly isotropic shell theory @xcite .
it would be interesting to compare the results of a generalized wlb model to include cross - sectional deformations with their results for the critical buckling forces .
one mode of failure , the ovalization of the microtubule cross - section ( brazier effect ) may for example be taken into account by adjusting the cross - sectional coordinates with the help of an `` ovalization parameter '' .
this would then have to be determined together with the other degrees of freedom from the equilibrium equations .
the authors acknowledge fruitful discussions with andreas bausch , mark bathe , mireille claessens and karen winkler .
ch acknowledges the von - humboldt feodor - lynen , the marie - curie eurosim and the anr syscom program for financial support .
ef is grateful to the deutsche forschungsgemeinschaft for support through grant fr 850/8 - 1 , and to the german excellence initiative via the nim program .
we also acknowledge the hospitality of the aspen center for physics where part of this work was completed .
in this appendix the effective bending and twist rigidities are calculated exactly from the wlb hamiltonian , eqs .
( [ eq : h0_omega ] ) - ( [ eq : delta_mismatch ] ) . to this end
the @xmath101-variables have to be integrated over to define an effective wlc hamiltonian .
first , we have to show that there are no terms in the shear hamiltonian , eq . ( [ eq : hshear ] ) , that would couple the different euler angles @xmath246 . to this end
we use eq .
( [ eq : delta_mismatch ] ) in eq .
( [ eq : hshear ] ) and specialize to a square cross - section .
the resulting shear hamiltonian can then be written as @xmath247\,,\end{aligned}\ ] ] where @xmath248 and @xmath249 . here
, each filament is labeled by the pair of indices @xmath250 , which denotes its location in the @xmath251th row and the @xmath252th column of the square cross - section .
the only terms that couple the different euler angles are @xmath253 and @xmath254 .
these identically vanish because of the symmetric arrangement of the filaments , @xmath255 and @xmath256 .
the remaining calculations are performed in fourier - space . for the transformation we use @xmath257-modes , which are appropriate for bundles with pinned boundary conditions .
writing the @xmath101-dependent part of the hamiltonian in matrix form @xmath258 one needs the following formula valid for a gaussian integral @xmath259 where @xmath260 is obtained from @xmath261 having found the solution @xmath260 , we can finally bring eq .
( [ eq : hsymbolic ] ) in the form of eq .
( [ def_eff_rigidities ] ) to read off the effective bend and twist rigidities . here
we solve eq .
( [ eq : eq_for_u ] ) for the case of bending of the square bundle .
it proves useful to introduce the dimensionless crosslink shear stiffness @xmath262 .
we can then write eq .
( [ eq : eq_for_u ] ) as @xmath263 we also defined the discrete second derivative @xmath264 .
note , that we are working in fourier - space so all quantities should carry an additional subscript relating to the mode - number @xmath110 . as different modes
do nt mix , there is no harm in dropping it for the moment . at the outer edges of the bundle @xmath265 we find @xmath266 and @xmath267 the eq .
( [ eq : varukfourier ] ) is solved by @xmath268 , where @xmath269 and we have defined @xmath270 . the constants @xmath271 are adjusted to fulfil the boundary conditions eqs .
( [ eq : bcouter ] ) and ( [ eq : bcinner ] ) .
this solution for the axial stretching variable @xmath24 is plotted in fig .
[ fig : axialstrain ] . to obtain the approximated bending rigidity of eq .
( [ eq : kappalinear ] ) one first has to insert the assumption @xmath272 into the hamiltonian and then minimize with respect to the single variable @xmath273 . this way one finds @xmath274 which has to be reinserted into the hamiltonian to yield eq .
( [ eq : kappalinear ] ) . in the case of pure twisting the same analysis can be done to calculate the effective twist rigidity .
the difference to the bending case is that now the full 2d boundary value problem has to be solved .
the equation determining the axial stretching @xmath275 of the filament indexed by @xmath250 is @xmath276 where the operator @xmath277 is the two - dimensional version of the discrete laplacian encountered above .
the finite difference operator @xmath278 is defined as @xmath279 . as for the case of pure bending , here , the twist variable @xmath280 enters only via the boundary terms .
the classic theory to calculate the twist rigidity of beams has been given by saint - venant @xcite . in this approach
the axial displacements are found by solving laplace s equation @xmath281 on the appropriate domain of the cross - section .
we see that eq .
( [ eq : equilibrium_twist ] ) reduces to the laplace equation in the limit @xmath282 , which is the reason why in fig . [
fig : cont_limit_twist ] the continuum limit is approached with increasing shear stiffness at fixed bundle size .
the remaining difference stemming from the discretization into @xmath6 filaments represents only a small effect . in saint - venant theory
it is well known that for rectangular domains two types of solutions appear , depending on the aspect - ratio of the cross - section .
we illustrate the different symmetry properties of these solutions in fig .
[ fig : torsionrechteck_20zu20 ] . of filament @xmath250 . left : square cross - section of @xmath283 filaments .
right : @xmath284 filaments , title="fig : " ] of filament @xmath250 . left : square cross - section of @xmath283 filaments .
right : @xmath284 filaments , title="fig : " ]
this appendix gives some details on the description of the bundle kinematic degrees of freedom .
the goal is to relate the generalized curvatures @xmath285 of filament @xmath13 , to the @xmath25 of the bundle central line . that these need not necessarily coincide is best illustrated by discussing an example .
if the central line is twisted but not bent , @xmath35 remains as the only non - vanishing component of the curvature vector .
the filaments themselves , however , twist _ and _ bend as they trace out a helical path with radius @xmath31 .
their bending energy is @xmath286 , of fourth order in @xmath35 . as a second example consider the bending of the central line , @xmath287 , where @xmath31 is the radius of curvature .
the filaments that lie at a distance @xmath11 away from the central line naturally have a different radius of curvature , @xmath288 , and thus a different bending energy .
the correction is again of higher order and scales as @xmath289 .
it turns out , that all effects like the two just mentioned only contribute to higher order .
to lowest order we will find that @xmath290 .
we define the vector @xmath291 to point to filament @xmath13 at arclength position @xmath14 .
the central line of the bundle is thereby given by @xmath292 .
the position of each filament in the cross - section is parametrized by @xmath293 , where @xmath20 and @xmath21 are constants independent of arclength @xmath14 and deformation of the bundle .
this , in particular , implies that in the reference state the filaments are always straight and untwisted .
we also need the derivative of @xmath294 with respect to @xmath14 , @xmath295 where we have used the frenet - seret equations and defined @xmath296 . with this
the tangent to filament @xmath13 is given by @xmath297 with an appropriate normalization factor , @xmath6 .
one finds , that the filament tangent is parallel to the central - line only if the twist vanishes , @xmath298 . for finite bundle twist
, the filament tangent is rotated around the vector @xmath294 relative to the central tangent @xmath87 .
the magnitude of the rotation is @xmath299 and depends on the distance @xmath300 of the filament to the central line . in order to derive expressions for the remaining two unit vectors @xmath301 and @xmath302 let us assume , for the time being , that no other rotation is allowed that may re - orient the local frame of filament @xmath13 relative to the central frame . with this assumption filaments
are not allowed to twist relative to their neighbors as this would correspond to a rotation around @xmath303 . in the bundles we consider , this internal twist motion
can safely be neglected as crosslinks provide for permanent rigid inter - filament connections . as explained below
, these internal twist modes may nevertheless be important at the time of bundle formation . the local frame
can then be related to the central frame by @xmath304 one can now express the generalized curvatures of the filaments in terms of the @xmath25 of the central line . for simplicity we here consider the case of pure twisting where @xmath229 .
one then finds @xmath305 note , that the difference in length between the central line and the individual filaments should be taken into account in the parametrization in terms of arclength , however , only contributes to higher orders .
more formally , @xmath306 .
in addition to twist , filaments also bend @xmath307 and @xmath308 which reduces to the well known expression for the curvature of a helix if @xmath309 .
finally , let us comment on what would happen if we did allow the filaments to twist individually , and relative to their neighbors .
this filament twist can be described by an additional rotation , @xmath310 of the local unit vectors , @xmath311 around the tangent @xmath303 .
the new vectors are then given by @xmath312 and the bundle twist , @xmath35 , can be calculated as before @xmath313 the twist energy should thus be written as @xmath314 where we defined the moments of the filament twist distribution @xmath315 . as explained above , in crosslinked bundles this internal twist
can be assumed to be quenched at the time of bundle formation . in this case
we need not treat the @xmath316 as dynamical variables for the discussion of bundle deformation .
a finite @xmath317 nevertheless gives the bundle a certain helicity and imposes pretwist , as discussed in section [ sec : pretwisted - bundles ] .
to calculate the geometric part @xmath81 of the shear deformation , an expression for the arclength mismatch between the two points on the filament pair is needed .
the general expression is @xmath318 where the first contribution , @xmath319 , derives from the possibility that the two points do not correspond to the same point , @xmath14 , on the bundle central line . by using eqs .
( [ rkprime ] ) , ( [ tangent_k ] ) and expanding one finds @xmath320\,,\end{aligned}\ ] ] where , we defined @xmath321 pointing from filament @xmath13 to filament @xmath322 . as in the formulation of @xmath64
, only the leading order terms have been accounted for .
the @xmath323-term only contains bending deformations .
this may be seen by setting @xmath324 appropriate for a pure twisting of the central line .
then @xmath325 as the vector @xmath61 lies within the cross - section , that is perpendicular to the tangent .
the last term corresponds to the arclength - difference acquired between two filaments at different distance to the central - line ( @xmath326 ) .
it is clear that the filament farther out has to go a longer distance , so its crosslinking sites will stay back in comparison to that of its neighbor closer in the center of the bundle .
however , twist also produces shear deformation between filaments which lie at equal distance to the central line ( @xmath327 ) .
this is embodied in the extra term @xmath319 , which we treat now ( also see fig . [
fig : delta_mt ] ) . to derive an expression for the shear displacement in this case ,
assume that the two filaments lie , separated by a distance @xmath59 , on the surface of a cylinder of radius @xmath31 .
twisting the cylinder makes the filaments wind around it , each taking an angle @xmath328 to the cylinder axis .
the shear deformation then simply is @xmath329 . for arbitrary orientation of the filament pair
we have to write @xmath330 , where @xmath331 is the tangent to the midline between the filament pair . in agreement
to what has been said above , this mechanism does not contribute to the shear displacement when the filament pair @xmath51 is connected by crosslinks in radial direction . in this case ,
@xmath332 , and the shear deformation @xmath333 . | bundles of filamentous polymers are primary structural components of a broad range of cytoskeletal structures , and their mechanical properties play key roles in cellular functions ranging from locomotion to mechanotransduction and fertilization . we give a detailed derivation of a wormlike bundle model as a generic description for the statics and dynamics of polymer bundles consisting of semiflexible polymers interconnected by crosslinking agents .
the elastic degrees of freedom include bending as well as twist deformations of the filaments and shear deformation of the crosslinks .
we show that a competition between the elastic properties of the filaments and those of the crosslinks leads to renormalized effective bend and twist rigidities that become mode - number dependent .
the strength and character of this dependence is found to vary with bundle architecture , such as the arrangement of filaments in the cross section and pretwist .
we discuss two paradigmatic cases of bundle architecture , a uniform arrangement of filaments as found in f - actin bundles and a shell - like architecture as characteristic for microtubules .
each architecture is found to have its own universal ratio of maximal to minimal bending rigidity , independent of the specific type of crosslink induced filament coupling ; our predictions are in reasonable agreement with available experimental data for microtubules .
moreover , we analyze the predictions of the wormlike bundle model for experimental observables such as the tangent - tangent correlation function and dynamic response and correlation functions . finally , we analyze the effect of pretwist ( helicity ) on the mechanical properties of bundles .
we predict that microtubules with different number of protofilaments should have distinct variations in their effective bending rigidity . |
experimental creation of bose - einstein condensate ( bec ) has led to emergence of a new field of nonlinear wave dynamics owing to a remarkable richness of nonlinear wave patterns supported by this medium .
first , vortices and bright and dark solitons were observed and their dynamics was studied theoretically in framework of the mean - field approach ( see , e.g. @xcite and references therein ) .
then , dispersive shocks generated by a large and sharp disturbance of bec were found in experiment @xcite and explained theoretically @xcite in framework of the whitham theory of modulations of nonlinear waves ( see also numerical experiment in @xcite ) . at last ,
the stationary waves generated by supersonic flow of bec past obstacles have been recently observed @xcite .
they were studied in @xcite where the main focus was on the nonlinear component representing a train of solitons ( in the simplest case of a single soliton @xcite ) or , more precisely , having a form of a modulated nonlinear periodic wave .
the theory developed in @xcite shows that there exist stationary spatial solutions of the gross - pitaevskii ( gp ) equation which describe nonlinear waves supported by a supersonic bec flow with mach number @xmath0 where @xmath1 is velocity of the oncoming flow at @xmath2 and @xmath3 is the sound speed of the long - wavelength linear waves .
the density @xmath4 of the condensate , as well as the components of the velocity field , depend on the variable @xmath5 alone , where @xmath6 is the slope of the phase lines with respect to the @xmath7 axis and it is supposed that the velocity of the oncoming flow is directed along the @xmath8 axis . then , the mach cone for sound waves with infinitely large wavelength corresponds to the slope @xmath9 and it was shown in @xcite that the spatial ( oblique ) solitons have @xmath10 , that is they are located inside the mach cone . in particular
, it was shown that the shallow solitons are formed close to the mach cone , @xmath11 , and they are asymptotically described by the korteweg - de vries ( kdv ) equation ; and deep solitons have @xmath12 ( i.e. they are formed at small angles with respect to direction of the oncoming flow ) and are asymptotically described by the nonlinear schrdinger ( nls ) equation . on the contrary ,
the linear waves are generated outside the mach cone and they have @xmath13 .
in fact , these linear waves had been observed in numerical simulations some years ago @xcite but the theory of their generation has not been developed so far .
the aim of this paper is to develop such a theory .
our analysis is based on the use of the mean - filed description of bec dynamics in the framework of the gp equation @xmath14 where @xmath15 is the order parameter ( `` condensate wave function '' ) , @xmath16 is the potential which confines atoms of a bose gas in a trap and/or describes interaction of the bec with the obstacle , and @xmath17 is an effective coupling constant arising due to inter - atomic collisions with the @xmath18-wave scattering length @xmath19 , @xmath20 @xmath21 being the atomic mass .
here we consider the bose - einstein condensate with repulsive interaction between particles for which @xmath22 .
as suggested by the experimental set - up @xcite , we consider a two - dimensional flow of a condensate , so that the condensate wave function @xmath23 depends on only two spatial coordinates , @xmath24 .
to simplify the theory , we assume that the characteristic size of the obstacle is much less than its distance from the center of the trap , so that the oncoming flow can be considered as uniform with constant density @xmath25 of atoms and constant velocity @xmath26 directed parallel to @xmath8 axis ( see also estimates in @xcite ) .
it is convenient to transform eq .
( [ 2 - 2 ] ) to a `` hydrodynamic '' form by means of the substitution @xmath27 where @xmath28 is density of atoms in bec and @xmath29 denotes its velocity field , and to introduce dimensionless variables @xmath30 where @xmath31 is the bec healing length and numerical constants are introduced for future convenience . as a result of this transformation
we obtain the system ( we omit tildes for convenience of the notation ) @xmath32=0 , \end{split}\ ] ] where @xmath33 .
since we shall consider waves far enough from the obstacle , the potential is omitted in ( [ 2 - 5 ] ) .
we are interested in linear waves propagating on the background flow with @xmath34 , @xmath35 , @xmath36 .
hence , we introduce @xmath37 and linearize the system ( [ 2 - 5 ] ) with respect to small deviations @xmath38 . as a result
we obtain the system @xmath39 which describes propagation of linear waves in bec with a uniform flow .
we obtain the applicability condition of these equations , if we notice that in the linear wave @xmath40 and the nonlinear terms of the order of magnitude @xmath41 can be neglected as long as they are much less than the linear ones @xmath42 .
thus , we get the criterion @xmath43 hence , if @xmath44 is large enough , the linear theory is applicable to description of waves outside the mach cone far enough from the obstacle , where the density amplitude @xmath45 of the wave satisfies the condition ( [ 10a ] ) . for harmonic waves
@xmath46 $ ] the system ( [ 3 - 1 ] ) yields at once the dispersion relation @xmath47 where @xmath48 .
now we consider the stationary wave patterns far enough from the obstacle where the condition ( [ 10a ] ) is supposed to be fulfilled .
in fact , this problem is analogous to the kelvin theory of ship waves generated by a ship moving in a deep water , but with a different dispersion law ( [ 3 - 2 ] ) .
hence , we shall follow kelvin s method in its form presented in @xcite .
first , we notice that in a stationary wave pattern @xmath49 and , hence , the components of the wavevector @xmath50 are the functions of the space coordinates @xmath51 connected with each other by the relationship @xmath52 where we have taken into account that for chosen geometry of the bec flow the wave must propagate upwind i.e. @xmath53 .
next , the `` ship wave '' pattern corresponds to a modulated two - dimensional wave where the wavevector @xmath54 is a gradient of the phase @xcite , @xmath55 hence , the components @xmath56 satisfy the condition @xmath57 which , with an account of ( [ 3 - 3 ] ) , yields the equation for @xmath58 @xmath59 where @xmath60 is defined by the derivative of an implicit function ( [ 3 - 3 ] ) : @xmath61 it follows from eq .
( [ 3 - 6 ] ) that @xmath62 and @xmath58 are constant along the characteristics defined as solutions of the equation @xmath63 at last , since at large distances from the obstacle , the latter can be considered as a point - like source of waves , the resulting stationary wave is centered , that is the characteristics intersect at the point @xmath64 of the location of the obstacle .
hence , we obtain the solution @xmath65 where @xmath66 is the angle between the @xmath8 axis and the radius - vector @xmath67 of the point @xmath68 with wavevector @xmath56 ; see fig . 1 where other convenient parameters are also defined , namely , the angle @xmath69 between the wavevector @xmath54 and a negative direction of the @xmath8 axis , and the angle @xmath70 between vectors @xmath54 and @xmath67 .
normal to it at the point @xmath68 and coordinates @xmath66 and @xmath69 defining the parameters of the wave pattern at this point . , width=226,height=226 ]
thus , we have @xmath71 @xmath72 then elementary calculation gives for ( [ 4 - 2 ] ) the expression @xmath73 and eq .
( [ 4 - 4 ] ) yields @xmath74 thus , we have found that for a fixed value of @xmath69 the components @xmath56 are constant along the line @xmath75 with @xmath66 defined by eq .
( [ 4 - 5 ] ) and the length of the wavevector given by ( [ 4 - 6 ] ) .
therefore the phase ( [ 3 - 4 ] ) can be conveniently calculated by integration along the line @xmath75 with constant vector @xmath54 , so that @xmath76 it means that the lines of constant phase ( e.g. the wave crests ) @xmath77 are determined in parametrical form by eq .
( [ 4 - 5 ] ) and @xmath78 where @xmath79 is given by eq .
( [ 4 - 6 ] ) and @xmath80 can be calculated from @xmath81 which gives , after elementary algebra , the expression @xmath82 this expression permits one to express eq .
( [ 5 - 2 ] ) as @xmath83 and eq .
( [ 4 - 5 ] ) as @xmath84 at last , the curves with constant phase @xmath77 are given in cartesian coordinates by the formulas @xmath85 thus , we have found the expressions describing the linear wave pattern in a parametric form where the parameter @xmath69 changes in the interval @xmath86 for @xmath87 we have @xmath88 and @xmath89 , that is small values of @xmath69 correspond to the wave before the obstacle .
the boundary values @xmath90 correspond to the lines @xmath91 that is the curves of constant phase become asymptotically the straight lines parallel to the mach cone . the general pattern is shown in fig . 2
we have compared the above approximate analytical theory with the exact numerical solution of the gp equation , which in non - dimensional units ( [ 2 - 4 ] ) takes the standard form @xmath92 which corresponds for @xmath93 to the system ( [ 2 - 5 ] ) with @xmath94 in our simulations the obstacle was modeled by an impenetrable disk with radius @xmath95 .
such an obstacle introduces large enough perturbation into the flow to generate an oblique solitons pair behind it ( see @xcite ) .
we assume that at the initial moment @xmath96 there is no disturbance in the condensate , so that it is described by the plane wave function @xmath97 corresponding to a uniform condensate flow . for large enough evolution time
, the wave pattern around the obstacle tends to a stationary structure .
an example of such a structure for @xmath98 is shown in fig . 2 . and
the radius of an impenetrable obstacle to @xmath95 . dashed line corresponds to the linear analytical theory for the line of constant phase .
it corresponds to eqs .
( [ 5 - 6 ] ) but the curve is shifted to 1 unit of length to the left from the center of the obstacle for better fitting to numerics . it should be noted that this distance is negligibly small in our theory corresponding to a point - like obstacle .
a pair of oblique dark lines behind the obstacle correspond to spatial solitons studied in @xcite.,width=302,height=226 ] a dashed line corresponds to the analytic theory developed in the preceding section ; as we see it agrees very well with the numerical results for @xmath98 .
the condition ( [ 10a ] ) indicates that the nonlinear effects grow up with increase of @xmath44 . to demonstrate this explicitly ,
we have compared the wavelength @xmath99 at @xmath89 calculated using the developed linear analytic theory with the same parameter obtained from our full numerical simulations . according to linear theory ,
the wavelength at @xmath89 ( i.e. @xmath87 ) is constant and equal to @xmath100 where we have used eq .
( [ 4 - 6 ] ) . in fig . 4 we compare this dependence of the wavelength @xmath99 on the mach number @xmath44 with the results of numerical simulations at the point with @xmath101 .
as we see , eq . ( [ 6 - 3 ] ) is very accurate for values of @xmath44 satisfying the condition ( [ 10a ] ) and discrepancy between analytical and numerical results slightly increases with increase of @xmath44 . in general , this plot confirms validity of a linear theory in the region of its applicability .
we have developed here the theory of linear waves generated by the flow of bec past an obstacle .
the linear approximation is correct for small enough amplitudes of the perturbation .
this condition is satisfied in the case of small disturbance introduced by the obstacle and not too high values of the mach number .
our numerical simulations confirm the analytical theory in the region of its applicability . | using stationary solutions of the linearized two - dimensional gross - pitaevskii equation , we describe the `` ship wave '' pattern occurring in the supersonic flow of a bose - einstein condensate past an obstacle .
it is shown that these `` ship waves '' are generated outside the mach cone .
the developed analytical theory is confirmed by numerical simulations of the flow past body problem in the frame of the full non - stationary gross - pitaevskii equation . |
in the local universe , luminous and ultra - luminous infrared galaxies ( lirgs & ulirgs ) have long been proposed as ideal candidates for forming massive elliptical galaxies @xcite .
they contain vast quantities of molecular gas and most show evidence of recent or ongoing merging activity , along with relatively high star - formation rates .
observations of these and other advanced mergers have long suggested that their luminosities are equivalent to or greater than massive elliptical galaxies @xcite , and @xcite , hereafter paper i. + the arguments _ against _ lirgs / ulirgs , and mergers in general forming _
l@xmath0 _ or _
m@xmath0 _ elliptical galaxies are based on kinematic arguements . until recently
, kinematic studies of lirgs / ulirgs have used the the infrared co stellar absorption lines at 1.6@xmath3 and 2.29@xmath3 to measure central velocity dispersions ( @xmath1@xmath2 ) .
these stellar band - heads are prominent in late - type stars and lie within observable infrared atmospheric windows .
+ @xmath1@xmath6 s have been previously measured for @xmath7 60 lirgs / ulirgs and two non - lirg / ulirg merger remnants @xcite , yielding a median @xmath1@xmath2 @xmath7 150 km s@xmath8 , far less than an _ m@xmath0 _ elliptical galaxy .
+ @xcite used the mg_ib _
( @xmath9 @xmath7 5200 ) and triplet ( @xmath9 @xmath7 8500 , hereafter cat ) to measure @xmath1@xmath2 in a sample of 13 visually selected mergers , producing a median of @xmath1@xmath2 @xmath10 200 km s@xmath8 .
@xcite , hereafter paper ii , measured @xmath1@xmath2 for 38 optically selected single - nuclei merger remnants , including 10 lirgs and 2 ulirgs , using the cat absorption lines and found a median @xmath1@xmath2 @xmath10 211 km s@xmath8 for the entire sample and a median of 196 km s@xmath8 for the lirgs / ulirgs .
+ in paper ii , _
k_-band photometry from paper i was combined with @xmath1@xmath11 to test whether and where merger remnants lie on the _ k_-band fundamental plane @xcite .
most of the merger remnants did lie on or within the scatter of the fundamental plane .
a small number of predominantly lirg / ulirg remnants sat offset from the fundamental plane in a tail - like feature , ( see figure 1 in paper ii ) .
the offset was due to high surface brightness ( @xmath12@xmath13@xmath14@xmath15@xmath16 ) , not small @xmath1@xmath2 , suggesting the presence of a younger ir - bright population .
+ paper ii also demonstrated that the co band - head produced consistently _ smaller _
@xmath1@xmath2 for lirgs / ulirgs than cat .
this suggested previous infrared studies may have _ underestimated _ their masses .
unfortunately , to date , _ all _ published @xmath1@xmath11 and @xmath1@xmath6 of remnants lie within the above - mentioned tail - like offset from the _ k_-band fundamental plane , making it difficult to separate whether the discrepancy is related to galaxy properties or problems with one of the stellar lines .
@xcite ( hereafter sg03 ) first noted that in their sample of 25 ( mostly s0 ) early - type galaxies , a signficant fraction also showed @xmath1@xmath6 @xmath12 @xmath1@xmath17 .
the largest discrepancies were found in s0s .
they proposed that dust enshrouded stellar disks , visible only in the ir produced smaller @xmath1@xmath18 .
an e0 and 6 non - lirg / ulirg merger remnants found to lie on the fundamental plane were observed in queue mode with the gemini near - infrared spectrograph ( gnirs ) on gemini - south ( program gs-2007a - q-17 , p.i .
rothberg ) using the short camera with the 111 l / mm grating and [email protected] @xmath20 99@xmath19 slit ( r @xmath7 6200 ) .
observations were centered on the @xmath21co(2,0 ) 2.29@xmath3 band - head .
previously published optical and infrared spectroscopic data were also used ( see papers i and ii for details ) , bringing the total to 8 non - lirg and 6 lirg merger remnants .
all @xmath1@xmath2 were either extracted or corrected ( for data from the literature ) to a 1.53 _ h_@xmath8 kpc central aperture . + low - resolution ( r @xmath7 1200 ) spectra were obtained with spex on the nasa infrared telescope facility for 11 merger remnants .
observations were made using the short cross - dispersed mode ( 0.8@xmath3 @xmath12 @xmath9 @xmath12 2.4@xmath3 , _ r _ @xmath7 1200 ) with the [email protected] @xmath20 15@xmath19 slit .
these observations were used to measure brackett @xmath22 ( br@xmath22 ) equivalent widths ( ew ) and supplement measurement of co indices for published @xmath1@xmath6 . + a comparison sample of 23 pure elliptical galaxies was assembled from the literature to look for the same discrepancy between @xmath1@xmath23 and @xmath1@xmath6 observed in merger remnants and sg03 .
all @xmath1@xmath2 obtained from the literature were corrected to a 1.53 _ h_@xmath8 kpc aperture .
figure 1 shows a three - panel comparison among the lirg merger remnants ( _ left _ ) , non - lirg merger remnants ( _ center _ ) and elliptical galaxies ( _ right _ ) .
the data were fit with a double weighted least - squares fit to compare with a slope of unity to search for discrepancies in @xmath1@xmath2 .
both the non - lirg merger remnants and the elliptical galaxies have a slope within 1@xmath1 of unity , while the the lirg merger remnants yield a slope of [email protected] .
in paper ii it was postulated that the cat and co stellar features were probing two different stellar populations ; the former probing older , late - type giants contributed from the progenitor spirals , and the latter sensitive to young red supergiants ( rsgs ) or asymptotic giant branch ( agb ) stars formed during the merger . a direct way to test this is to measure ews of the cat and co features and compare those with stellar populations models in order to age - date each population .
the premise is that the younger stellar populations are created within a disk formed when gas from the progenitor spirals dissipates towards the barycenter of the merger ( see @xcite ) . at _
k_-band , the light from this population dominates during certain epochs @xcite ( hereafter m05 ) .
+ figure 2 shows stellar population models from m05 ( _ solid lines _ ) for a simple burst population with a salpeter imf and solar metallicity for the cat@xmath0 and co indices .
cat@xmath0 ews and co@xmath25 indices were measured for each merger remnant and used in conjunction with the models to age - date each remnant as plotted . in some cases , the measured indices intersected multiple points in the models .
the degeneracy was broken by measuring the ew of the paschen triplet ( pat ) absorption lines ( near 8500 ) and comparing with predicted values from m05 , or measuring the ews of br@xmath22 emission ( 2.165@xmath3 ) and comparing with predicted values from starburst99 @xcite . in some cases
the degeneracy for co ages could not be broken due to the absence of br@xmath22 .
double values are plotted in figure 2 with a dotted line connecting them . + overplotted in figure 2 ( _ right _ ) next to each merger remnant are the fractional @xmath1@xmath2 differences ( @xmath1@xmath26 - @xmath1@xmath27)/@xmath1@xmath26 .
the lirg merger remnants show larger fractional differences than the non - lirg remnants ( except ngc 7252 ) .
the results in figure 2 appear to support the premise that the presence of young populations are responsible for the smaller @xmath1@xmath2 observed in lirgs
. the final test will be to confirm whether these young populations reside in rotating disks located within the central few kpc of the lirg / ulirg merger remnant using spatially resolved spectroscopic observations of the co band - head . | recently , controversy has erupted over whether gas - rich spiral - spiral mergers are capable of forming _
m@xmath0 _ ellipticals .
measurements of @xmath1@xmath2 from the 2.29@xmath3 co band - head for local lirg / ulirgs , suggest they are not .
ir - bright mergers are often cited as the best candidates for forming massive ellipticals , so the recent observations have raised doubts about both the toomre merger hypothesis and the fundamental assumptions of @xmath4-cdm galaxy formation models .
however , kinematics obtained with the calcium ii triplet at 8500 suggest mergers are forming _ m _ @xmath5 _ m@xmath0 _ ellipticals . in this work , we show that kinematics derived from the co stellar absorption band - head leads to a significant underestimation of the masses of lirgs / ulirgs .
this is primarily due to the presence of a young population affecting co band - head measurements . |
@xmath0 meson weak decays are very important for testing the standard model and probing new physics .
the two @xmath0 factories have accumulated large data sets while tevatron and lhcb will have even more data .
the main task for theorists is to compute all experimental observables in a reliable way .
this amounts to compute hadronic matrix elements reliably .
there are several methods in the literature .
the earliest is the so called navie factorization @xcite . in this method
the matrix element has no renormalization scheme and scale dependence , so can not cancel the corresponding dependence of the wilson coefficients .
futher more , it can not account for `` nonfactorizable '' contributions .
the generalized factorization @xcite takes into account the one - loop radiative corrections to recover the renormalization scheme and scale dependence for the hadronic matrix elements .
but at quark level , to avoid the infrared divergence , it has to assume the external quarks to be off - shell by @xmath1 .
this results in gauge dependence . to account for `` nonfactorizable '' contributions ,
a phenomenological parameter @xmath2 is introduced and it is assumed that @xmath2 is universal .
but actually @xmath2 is process - dependent . in order to overcome the shortcomings of the above methods , in 1999 , beneke ,
_ et al . _ ( bbns ) @xcite proposed a new scheme based on qcd . in this
talk , i will concentrate on our work @xcite about the chirally enhanced power corrections and infrared divergence problem and the application of the qcd factorization approach .
my talk is organized as follows : in section 2 , i shall give an simple introduction to qcd factorization .
section 3 is devoted to the chirally enhanced power corrections and the cancellation of the infrared divergence in vertex corrections . in section 4@xmath36
, the application of qcd factorization including chirally enhanced power corrections to the two - body charmless @xmath0-decays is reviewed .
section 7 is for conclusions .
beneke , _ et al . _ ( bbns ) proposed a new factorization scheme based on qcd @xcite .
they pointed out that , in the heavy quark limit ( @xmath4 @xmath5 @xmath6 ) @xmath7 where the short distance hard scattering kernels @xmath8 is calculable order by order in perturbative theory ; the long distance quantities , e.g. decay constants , form factors , light - cone distribution amplitudes are inputs form either experimental measurements or other theory ; @xmath9 is a light meson or a charmonium state , @xmath10 ( contains the spectator in @xmath0 ) is any light or heavy meson . if @xmath10 is heavy ( for example @xmath11 ) , the second line in eq.([eq : bbns ] ) is @xmath12 suppressed .
( 300,100)(0,0 ) ( -50,-405 ) at the zeroth order of @xmath13 , eq.([eq : bbns ] ) reduces to naive factorization . at the higer order ,
the corrections can be computed systematically . the renormalization scheme and scale dependence of @xmath14 is restored . in the heavy quark limit ,
the `` nonfactorization '' contributions is calculable perturbatively .
it does not need to introduce @xmath15 .
the strong phase is suppressed by either @xmath13 or @xmath12 .
@xmath16-exchange and @xmath16-annihilation diagrams are @xmath12 suppressed .
there is no long distance interactions between @xmath9 and ( @xmath17 ) .
@xmath18 includes tree diagram , non - factorizable gluon exchange , e.g. vertex corrections [ fig.[fig : fig1](a)-(d ) ] , penguin corrections [ fig.[fig : fig1](e)-(f ) ] , @xmath19 .
@xmath20 includes hard spectator scattering [ fig.[fig : fig1](g)-(h ) ] , @xmath19 later on we shall apply eq.([eq : bbns ] ) to @xmath0 meson rare decays .
superficially , @xmath21 @xmath3 @xmath22 is a small number . but in some cases , such power suppression fails numerically .
for example @xmath23 is always multiplied by a formally power suppressed but chirally enhanced factor @xmath24 @xmath25 @xmath26 ( where @xmath27 @xmath25 @xmath28 , @xmath29 are current quark masses , and @xmath30 denote pseudoscalar mesons ) . in the heavy quark limit , @xmath4 @xmath31 @xmath32 , @xmath24 @xmath31 @xmath33 .
but for @xmath4 @xmath3 5gev , @xmath24 @xmath3 @xmath34 , no @xmath12 suppression .
so we should consider the chirally enhanced power corrections .
for example , in @xmath0 @xmath31 @xmath35 , dominant contribution to the amplitude @xmath3 @xmath36 , and @xmath37 @xmath3 @xmath38 , so one half comes from the chirally enhanced power corrections .
possible sources of power corrections are : high twist wave functions , quark transverse momentum @xmath39 , and annihilation topology diagrams . in this talk ,
i only discuss the chirally enhanced power corrections from high twist wave functions @xcite .
all the chirally enhanced power corrections involve twist-3 light cone distribution amplitudes @xmath40 and @xmath41 .
if we want to include chirally enhanced power corrections consistently , we must prove that the hard scattering kernels are infrared finite , at least for vertex corrections .
the decay amplitude of @xmath0 @xmath31 @xmath42 reads @xmath43 for the hard kernel @xmath18 in eq.([eq : bbns ] ) , there is no contributions from twist-3 wave function insertion for @xmath44 and @xmath45 operators .
but for @xmath46 , it is subtle ! for twist-3 light cone distribution amplitudes ,
if we use dimensional regularization , the infrared divergences of fig.1(a)-(d ) can not cancel !
this is because the twist-3 wave functions are only defined in @xmath47 @xmath25 @xmath48 @xmath49 @xmath50 dimensions .
we need to give gluon a small mass and regularize the infrared integrals in 4-dimensions . after a lengthy calculation , for the operator @xmath51 ,
the vertex correction terms are @xmath52 \label{eq : q5a } \\ & & \!\!\!\!\!\!\!\!\!\!\!\ ! { \rm fig.1(b)}\ { \sim}\ + \frac{{\alpha}_{s}}{4{\pi } } \frac{c_{f}}{n_{c}}\frac{{\phi}_{\sigma}(v)}{\bar{v } } \big [ \frac{{\ln}^{2}{\lambda}}{2}+2{\ln}(-\bar{v}){\ln}{\lambda } -4{\ln}\bar{v}{\ln}{\lambda}+{\ln}{\lambda}+{\rm finite~terms}\big ] \label{eq : q5b } \\ & & \!\!\!\!\!\!\!\!\!\!\!\ ! { \rm fig.1(c)}\ { \sim}\ + \frac{{\alpha}_{s}}{4{\pi } } \frac{c_{f}}{n_{c}}\frac{{\phi}_{\sigma}(v)}{v } \big [ { \ln}^{2}{\lambda}-2{\ln}(-v){\ln}{\lambda}+3{\ln}{\lambda } + { \rm finite~terms } \big ]
\label{eq : q5c } \\ & & \!\!\!\!\!\!\!\!\!\!\!\ ! { \rm fig.1(d)}\ { \sim}\ -\frac{{\alpha}_{s}}{4{\pi } } \frac{c_{f}}{n_{c}}\frac{{\phi}_{\sigma}(v)}{\bar{v } } \big [ { \ln}^{2}{\lambda}-2{\ln}(-\bar{v}){\ln}{\lambda}+3{\ln}{\lambda } + { \rm finite~terms } \big ] \label{eq : q5d } \end{aligned}\ ] ] where @xmath53 @xmath25 @xmath54 . in the above
, we can see that after summing over fig.[fig : fig1](a)-(d ) , the infrared divergences cancel only when @xmath55 @xmath25 @xmath56 , where @xmath57 @xmath25 @xmath58 .
this is not a surprise because we neglected the contribution of three - body twist-3 light cone distribution amplitudes ( lcdas ) . for the asymptotic form , @xmath55 @xmath25 @xmath59 which is the same for both @xmath60 and @xmath61 meson .
after our work @xcite , bbns @xcite also discussed the infrared divergence problem .
they implemented the equation of motion @xmath62 to show the infrared safety without assuming @xmath55 to be symmetric .
however eq.([eq : motion ] ) is justified only when 2-particle twist-3 lcdas are asymptotic , i.e. @xmath63 @xmath25 @xmath50 and @xmath55 @xmath25 @xmath59 by neglecting the contribution from the 3-particle twist-3 lcdas .
our result is based on a more general situation than that bbns considered .
therefore , bbns result is consistent with ours .
we can prove at the order of @xmath64 that the decay amplitude is independent of the renormalization scale .
we can also prove the gauge invariance when chirally enhanced power corrections are included .
the detailed proof can be found in @xcite .
we do not discuss it here .
now we come to the application below .
now we give the prediction of the branching ratios and @xmath67 asymmetries .
we included the contributions of the hard spectator scattering and annihilation topology .
we also include the chirally enhanced power corrections .
we compute the branching ratios ( @xmath68 ) and @xmath67 asymmetries of @xmath0 @xmath31 @xmath65 : @xmath69 , @xmath35 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath19 @xmath19 ; @xmath0 @xmath31 @xmath66 : @xmath74 , @xmath75 , @xmath76 @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath19 @xmath19 .
all our numerical results can be found in the tables in refernce @xcite . from our calculated resulte
we see that : owing to including the chirally enhanced power corrections , the scale @xmath81 dependence of @xmath68 is smaller ; @xmath68 of @xmath82 @xmath31 @xmath83 @xmath84 ( pure tree ) and @xmath82 @xmath31 @xmath85 ( pure penguin ) are in good agreement with data ; @xmath68 of @xmath86 @xmath31 @xmath87 is larger than data ; @xmath68 of @xmath0 @xmath31 @xmath35 seem smaller than data ; for @xmath0 @xmath31 @xmath71 , if consider di - gloun fussion , we can fit the data ; for @xmath86 @xmath31 @xmath88 , our prediction is much smaller than data ; for @xmath86 @xmath31 @xmath89 , the result is very small , only weak annihilation diagram contributes ; there are large uncertainties from ckm elements , form - factors , annihilation parameters .
but a global analysis can fit the data very well . for @xmath67 asymmetries ,
the numberial results are not reliable because vertex , penguin , hard spectator scattering and annihilation diagrams can all give imaginary part to decay amplitudes , so strongly affect @xmath90 . because of space limitation , for more detail , see @xcite . for @xmath0 @xmath31 @xmath66 ,
the differences from @xmath0 @xmath31 @xmath65 are : if the emitted meson is vector meson ( vertex diagrams and penguins ) , then the twist-3 wave functions of @xmath91 do not contribute ( power suppressed , so can be neglected ) .
but for hard spectator scattering , even the emitted meson is vector one , there are still twist-3 contributions .
note , there are large uncertainties on @xmath67 asymmetries .
so we can not make good predictions on @xmath90 in qcdf scheme
. the calculated branching ratios of @xmath0 @xmath31 @xmath66 for @xmath92 @xmath31 @xmath93 and @xmath92 @xmath31 @xmath94 transitions and @xmath67 asymmetries can be found in reference @xcite .
beneke , _ et al .
_ @xcite have done a global analysis on @xmath0 @xmath31 @xmath65 .
but their fitted @xmath95 @xmath3 @xmath96 , a bit large compare with the standard ckm fit .
now there are many new data on both @xmath0 @xmath31 @xmath65 and @xmath66 .
it is necessary to do the global fit of @xmath0 @xmath31 @xmath65 , @xmath66 at the same time .
we have done it ! for @xmath0 @xmath31 @xmath69 , @xmath35 , the fit is sensitive to @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 , and @xmath103 . to include seven @xmath0 @xmath31 @xmath66 decay channels , only @xmath104 , @xmath105
are newly involved sensitive parameters .
so the inclusion of @xmath0 @xmath31 @xmath66 will lead more stringent test to qcdf .
the experimental constrains of @xmath67 asymmetries are not implemented , because the qcdf s predictions on @xmath67 asymmetries in @xmath0 decays are rough !
so we do not use @xmath67 asymmetries for the global fit .
the best fit values of the branching fractions is presented in table.[tab : tab8 ] .
+ the main results are : + the best fit of @xmath95 @xmath3 @xmath106 which is consistent with recent fit results
@xmath107 @xmath108 @xmath95 @xmath108 @xmath109@xcite . + for @xmath0 @xmath31 @xmath88 , the best fit is around @xmath110 , + @xmath111 exp : the babar and belle average is @xmath112 + for @xmath113 @xmath31 @xmath114 , the best fit is @xmath115 , + @xmath111 exp : belle @xmath116 cleo @xmath117 . + for @xmath113 @xmath31 @xmath118 , the best fit is @xmath119 , + @xmath111 exp : babar @xmath120 belle @xmath121 + for @xmath113 @xmath31 @xmath122 , the best fit @xmath3 @xmath123 , + @xmath111 exp : lp03 @xmath3 @xmath124 ichep04 @xmath3 @xmath125 . + for more details , see reference @xcite .
we include : i ) chirally enhanced power corrections , ii ) weak annihilation , iii ) hard spectator scattering .
we list the computed branching ratios and @xmath67 asymmetries of @xmath126 @xmath31 @xmath65 , @xmath66 in reference @xcite .
only @xmath126 @xmath31 @xmath127 @xmath128 , @xmath129 @xmath130 , @xmath131 @xmath132 , @xmath73 @xmath128 have large branching ratios .
we have shown that * chirally enhanced power corrections must be included in qcdf * twist-3 light cone distribution amplitude @xmath133 , @xmath41 must be considered simultaneously * the infrared divergences in vertex corrections cancel only twist-3 wave function of light pseudoscalar is symmetric , so chirally enhanced power enhanced power corrections can be included consistently * the calculated branching ratios for @xmath0 @xmath31 @xmath65 , @xmath66 charmless decays are , principally , in agreement with data * gloabl analysis of @xmath0 @xmath31 @xmath65 , @xmath66 can fit the data very well including @xmath86 @xmath31 @xmath88 and @xmath95 @xmath3 @xmath106 * @xmath126 @xmath31 @xmath65 , @xmath66 charmless decays are waiting to be tested in tevatron and lhcb 0 m. wirbel , b. stech , and m. bauer , _
z. phys . _ * c29 * , 637 , ( 1985 ) ; + m. bauer , b. stech , and m. wirbel , _ z. phys . _ * c34 * , 103 , ( 1987 ) .
r. fleischer , _ phys .
lett . _ * b321 * , 259 ( 1994 ) ; _ ibid . _
_ b332 _ , 419 ( 1994 ) ; _ z. phys . _ * c62 * , 81 ( 1994 ) ; a. ali , g. kramer , and c. d. l , _ phys .
* d59 * , 014005 ( 1999 ) ; _ ibid . _ * d58 * , 094009 ( 1998 ) ; y. h. chen , h. y. cheng , b. tseng , and k. c. yang , _ phys .
rev . _ * d60 * , 094014 , ( 1999 ) .
m. beneke , g. buchalla , m. neubert and c. t. sachrajda , _ phys .
lett . _ * 83 * , 1914 , ( 1999 ) ; _ nucl .
phys . _ * b591 * , 313 , ( 2000 ) .
d. s. du , d. s. yang , g. h. zhu , _ phys .
lett . _ * b509 * , 263 ( 2001 ) ; _ phys .
* d64 * , 014036 ( 2001 ) .
d. s. du , h. j. gong , j. f. sun , d. s. yang , g. h. zhu , _ phys .
rev . _ * d65 * , 074001 ( 2002 ) .
d. s. du , h. j. gong , j. f. sun , d. s. yang , g. h. zhu , _ phys .
_ * d65 * , 094025 ( 2002 ) ; erratum - ibid . *
d66 * , 079904 ( 2002 ) .
d. s. du , j. f. sun , d. s. yang , g. h. zhu , _ phys .
* d67 * , 014023 ( 2003 ) .
j. f. sun , g. h. zhu , d. s. du , _ phys .
* d68 * , 054003 ( 2003 ) .
m. beneke , g. buchalla , m. neubert and c. t. sachrajda , _ nucl .
* b606 * , 245 , ( 2001 ) .
a. hcker , h. lacker , s. laplace , f. le diberder , _ eur .
_ * c21 * 225 ( 2001 ) . | some recent progress in china in the study of charmless @xmath0 decays with qcd factorization is reviewed .
chirally enhanced power corrections and infrared divergence problem are stressed . |
mixed ruthenates with perovskite based crystal structures have been receiving considerable attention of late @xcite , both because of their interesting magnetic properties and because of the recent discovery of superconductivity in the layered ruthenate , sr@xmath1ruo@xmath3 @xcite . despite the rarity of 4d based magnetic materials ,
srruo@xmath0 is a robustly ( curie temperature , @xmath4165 k ; magnetization , @xmath51.6 @xmath6/ru ) ferromagnetic metal occurring in a distorted cubic perovskite structure @xcite , and @xmath7 can be even further increased by doping with pb @xcite .
however , magnetism is easily suppressed by doping with ca , although ca / sr states are far removed from the fermi level and accordingly may not be expected to influence the electronic properties of srruo@xmath0 very drastically .
furthermore , sr@xmath1yruo@xmath8 which has essentially the same crystal structure as srruo@xmath9 but with every second ru substituted by y , is antiferromagnetic , with estimates of the saturation magnetization even higher than the parent compound ( @xmath10 although the critical temperature , @xmath11 is reduced to 26 k. the variety of magnetic and electronic properties observed in these superficially similar compounds already poses an interesting theoretical challenge ( cf .
, for instance , non superconducting cuprates , which despite their large variety , always show strong antiferromagnetism in cu - o planes ) .
besides , there are a number of interesting observations that deserve attention .
these include the fact that srruo@xmath0 is the only known ferromagnetic metal among the 4@xmath12 oxides .
as such interesting differences are expected from the much more abundant 3@xmath12 oxide magnets .
for example , much stronger spin - orbit effects compared to the 3d systems may be anticipated , and these may manifest themselves in the magneto - crystalline and magneto - optical properties .
in fact , srruo@xmath0 does show an abnormally high magneto - crystalline anisotropy for a pseudo cubic material @xcite to the extent that it is difficult to measure its saturation magnetization using standard measurements of hysteresis loops , and resulting in some confusion in the older experimental literature .
more recently , klein and co - workers @xcite have measured strong magneto - optic properties in srruo@xmath0 epitaxial films .
4@xmath12 ions generally have more extended @xmath12 orbitals than the corresponding 3@xmath12 ions , and as a result @xmath13 oxides tend to have greater overlap and hybridization between the transition metal and o 2@xmath14 orbitals . besides a tendency towards greater itinerancy , this can lead to more interplay between structural degrees of freedom and the magnetic and electronic properties .
as mentioned , additional interest in these ruthenates comes from their apparent proximity to superconductivity , and possible new insights into the problem of high - temperature superconductivity that may emerge from their study .
although the layered perovskite sr@xmath1ruo@xmath3 has a modest @xmath15 of 1k ( there have been very recent , unconfirmed reports of signatures of superconductivity at up to 60k in the double perovskite sr@xmath1yruo@xmath2 with cu doping @xcite ) , it was suggested that this material may be an unconventional superconductor .
this is based largely on several similarities with the cuprates : sr@xmath1ruo@xmath3 is iso - structural with the first discovered high-@xmath15 superconductor , shows highly two dimensional electronic properties , and of course is close to magnetic phases , particularly srruo@xmath0 , and sr@xmath1yruo@xmath2 . however , the evidence for strong electron correlations in ruthenates is by far not yet as compelling as the body that has been accumulated for the cuprates and many other 3@xmath12 oxides , and the question of whether these ruthenates can be treated within the framework of conventional band theory , or require a strong - correlation based theory , remains open .
several photoelectron spectroscopy experiments have been reported for sr@xmath1ruo@xmath3 , which because of its layered crystal structure is more amenable to such studies than nearly cubic srruo@xmath0 .
yokoya et al .
@xcite and lu et al .
@xcite , using angle resolved photoemission ( arpes ) , both report observation of fermi surface sections and extended van hove features somewhat like those in the local density electronic structure calculations , although the positioning of the van hove singularity relative to the fermi energy differs and the dispersion is generally somewhat weaker than in the calculations , probably due to correlations , but possibly because of strong electron - phonon and -magnon interactions .
similarly , schmidt et al .
@xcite observed the valence bands of sr@xmath1ruo@xmath3 using arpes , and found uppermost occupied bands with a width reduced by a factor of 2 compared with band structure calculations .
unfortunately , arpes is highly sensitive to the quality of samples and particularly sample surfaces .
interestingly , polycrystalline but otherwise apparently high quality sr@xmath1ruo@xmath3 samples can be non - metallic @xcite .
angle integrated photoemission is a more robust technique ; using it , yokoya et al .
@xcite found good agreement between the experiment and density functional calculations , but observe a correlation satellite to the d - band ( using resonant photoemission ) .
based on these measurements they estimated an effective hubbard @xmath16 of 1.5 ev , which is at least three times smaller than similar estimates for the cuprate superconductors , casting some doubt on suggestion that sr@xmath1ruo@xmath3 and related ruthenates are very strongly correlated .
one of the most decisive arguments in favor of the importance of strong correlations in high-@xmath15 cuprates is the failure of conventional local - density - approximation band structure calculations to describe even qualitatively the antiferromagnetism in the undoped parent compounds .
similarly , the key question for these ruthenates may be which approximation , strongly correlated or band structure based , is best suited to explaining the variety of magnetic properties .
one of the main purposes of this work is to determine whether a ( similar to the cuprates ) failure of the conventional , mean - field type , band calculations , is present in these ruthenates . within a strong - correlation scenario , the ferromagnetism in metallic srruo@xmath17 results from the double exchange mechanism , while antiferromagnetism in insulating sr@xmath1yruo@xmath2 is due to superexchange via two ( unlike the 3d oxides and cu perovskites ) oxygen ions .
this is appealing because in the mott - hubbard picture the main factor controlling the magnetic properties is carrier concentration , which is indeed different in those two materials : in ( sr , ca)ruo@xmath0 ruthenium is four - valent , that is , its @xmath12-band is populated by 4 electrons , while in sr@xmath1yruo@xmath2 the nominal valency of ru is 5 , and the number of @xmath12 electrons is 3 . on the other hand , integer occupancy does not favor the double exchange scenario , and , besides , it is unclear how the mott - hubbard model provides a mechanism for suppressing magnetism in caruo@xmath18 finally , as we discuss in detail below , conventional band theory in all the cases we test does yield the correct magnetic ground state , in contrast to the cuprates and similar correlated 3@xmath19oxides .
thus , contrary to some recently suggested superconductivity scenarios based on strong correlations@xcite , it seems likely that if strong correlations play some role , it is more of a quantitative than of a qualitative nature .
on the other hand , we note that even if sr@xmath1ruo@xmath3 and other ruthenates are not strongly correlated , the superconductivity could still be unconventional , for instance arising from a magnetic mechanism . in this regard ,
a number of measurements indicate anomalously large scattering of electrons by spin fluctuations@xcite , complicated by a strong magnetoelastic coupling@xcite .
cyclotron masses@xcite , the specific heat , and the paramagnetic susceptibility @xcite are all strongly renormalized .
while there is always the possibility of ascribing this renormalization to strong correlations , the simplest explanation may be strong electron - phonon - magnon interactions .
further , an abnormally large transport coupling constant , @xmath20 is required to rationalize the temperature dependence of the resistivity with the calculated drude plasma energies@xcite , although this value is consistent with the specific heat enhancement .
unusual temperature dependencies of the hall effect @xcite were found in caruo@xmath0 and in srruo@xmath0 .
we shall return to the transport properties later in the paper ; it is plausible that they can be reconciled with the conventional one - electron mechanism , despite the unusual @xmath21dependences .
the main purpose of the present paper is to study the magnetic phases and the relative importance of correlation and band structure effects for obtaining the magnetic properties .
we focus on the double perovskite sr@xmath1yruo@xmath2 and the ferromagnetic - paramagnetic transition in ( sr , ca)ruo@xmath0 with increasing ca content , and we shall show that conventional band theory is fully able to describe the variegated magnetic properties in this family of materials .
as mentioned , srruo@xmath0 occurs in an orthorhombic , pbnm , gdfeo@xmath0 structure , which has four formula units per cell .
it is interesting to note that this is the same generic structure as lamno@xmath0 and related manganites that have received considerable recent attention because of the discovery of colossal magnetoresistance effects in some of these .
further srruo@xmath0 has the same nominal @xmath12 electron count as lamno@xmath0 , although unlike lamno@xmath0 it is a ferromagnetic metal even without doping . in lamno@xmath0
the distortion from the ideal cubic perovskite crystal structure consists of both rotations of the o octahedra and jahn - teller distortions of them to yield mn - o bond length variations of more than 10% .
this is understood in ionic terms as a result of the fact that the high spin mn ion with this electron count has a half full majority spin e@xmath22 orbital favoring a jahn - teller distortion .
in contrast , srruo@xmath0 occurs with a reduced magnetic moment and its distortion consists of almost rigid rotations of the o octahedra with practically no accompanying variations in the ru - o bond lengths .
caruo@xmath0 occurs in the same crystal structure and symmetry as srruo@xmath0 , also with no evident jahn - teller distortion of the o octahedra , but with approximately twice larger rotations .
such rotations are common in perovskite based materials and are usually understandable in terms of ionic size mismatches between the a and b site cations .
such an explanation is consistent with the trend observed in ( sr , ca)ruo@xmath0 since the ca@xmath23 ionic radius is approximately 0.15 smaller than sr@xmath23 .
although caruo@xmath0 is paramagnetic , it is believed to be rather close to magnetic instability .
sr@xmath1yruo@xmath2 is an antiferromagnetic insulator that occurs in a distorted but well ordered double perovskite structure .
this is derived from the perovskite , srruo@xmath0 by replacing every second ru by y , such that the remaining ru ions form an fcc lattice .
the structural units are thus ru - o and y - o octahedra , with the sr ions in the a site positions providing charge balance .
each ru - o octahedra shares a single o atom with each neighboring y - o octahedra , and vice versa , but there are no common o ions shared between different ru - o octahedra .
the primary distortions from the ideal perovskite derived structure , consist of ( 1 ) a substantial breathing of the octahedra to increase the y - o distance to 2.2 at the expense of the ru - o distances which become 1.95 and ( 2 ) rotations of the octahedra to reduce the closest sr - o distances , consistent with the ionic sizes .
these distortions reduce the symmetry to monoclinic ( p21/n )
. a related view of the crystal structure is based on the fact that y , like sr is fully ionized in such oxides , and accordingly is a spectator ion providing space filling and charge to the active ru - o system but playing no direct role in the electronic or magnetic properties . from this point of view , sr@xmath1yruo@xmath2 consists of independent rigid , but tilted , ( ruo@xmath2)@xmath24octahedral clusters , arranged on a slightly distorted fcc lattice . hopping then proceeds between ru ions in neighboring ruo@xmath2 clusters via two intervening o ions since y is tri - valent , the ru is formally 5-valent ( @xmath25 ) in this compound instead of formally tetra - valent as in perovskite srruo@xmath0 . in the octahedral crystal field ,
the ru @xmath26 orbitals lie below the @xmath27 orbitals , so that in the high spin state the majority spin ru @xmath26 manifold would be fully occupied , and all other ru 4d orbitals unoccupied .
this jahn - teller stable configuration is consistent with the experimental observation that the bond angles and bond lengths within the ru - o octahedra are almost perfectly equal , but the ru moment of 1.85 @xmath6/ru measured using neutron diffraction is considerably smaller than the 3 @xmath6/ru that would be expected in the high - spin configuration .
first principles studies of srruo@xmath0 have shown that its electronic structure involves rather strong ru - o covalency , and that o @xmath14-derived states participate substantially in the magnetism and the electronic structure near the fermi energy , which is important for understanding the transport properties . as will be discussed below
, a similar covalency is present in caruo@xmath0 and the differences in the magnetic ground states of caruo@xmath0 and srruo@xmath0 are due to band structure effects related to the modulation of the ru - o hybridization by the structural distortion . in this regard
, it should be noted that ru@xmath28o hybridization may be even stronger in sr@xmath1yruo@xmath29 , based on the expectation that the o @xmath30 manifold would be even higher in energy with respect to the ru @xmath12 states .
the similar ru - o distances in srruo@xmath0 and sr@xmath1yruo@xmath2 ( less than 0.03 longer than in sr@xmath1yruo@xmath2 ) and the fact that 5 is not a common oxidation state for ru also suggest strong covalency in the double perovskite . here
we report density functional calculations of the electronic and magnetic properties of sr@xmath1yruo@xmath2 .
these confirm the strongly hybridized view of these materials and provide an explanation for the electronic and magnetic properties .
the electronic structure of srruo@xmath0 has been described elsewhere@xcite . here
we repeat , for completeness , the main results , and also discuss some quantitative differences between the published calculations .
there have been two recent band structure calculations for srruo@xmath0@xcite . in both
works the calculations were performed for both an idealized cubic perovskite structure and the experimental crystal structure .
et al_@xcite interpreted their experimental measurements in terms of the band structure calculated within the local spin density approximation ( lsda ) using the linear muffin - tin orbitals ( lmto ) method .
singh @xcite used the general potential linearized augmented plane - waves ( lapw ) method to calculate electronic and magnetic properties .
the two studies yielded reasonably similar results for the electronic structures near the fermi energy although some noticeable differences are present .
important for interpreting experimental results are the differences in the density of states and in the fermi velocities .
the latter were found in ref . to be almost isotropic , while in ref .
strong anisotropy of the fermi velocity ( about 30% in each channel ) was reported .
the ratio @xmath31 found in ref .
is 50% larger than that in ref . .
most important , the overall shape of the density of states within a @xmath32 ry window at the fermi level is rather different .
it is known that the accuracy of the atomic sphere approximation calculations can be difficult to control for materials with open crystal structures and low site symmetries due to sensitivity to the computational parameters ( e.g. , basis set , inclusion of empty spheres in lattice voids , linearization parameters etc . ) .
since we wanted to use lmto - asa technique in analyzing the calculated band structure , we have repeated the lapw calculations reported in ref . using a standard lmto - asa package _
stuttgart-4.7_. we found it necessary to include 10 empty spheres per formula unit to achieve adequate space filling in the distorted structure ( in the cubic perovskite structure this was not needed ) .
the result appeared to be much closer to the lapw results of ref .
than to the lmto ones of ref . ; ref .
does not mention use of any empty spheres , in which case insufficient space filling could have influenced the calculation .
the results given here are from lapw calculations , except where specifically noted otherwise .
calculations for srruo@xmath0 in the ideal perovskite structure yielded a spin moment of 1.17 @xmath6 per formula unit , while calculations including the experimentally observed rotations yielded a larger moment of 1.59 @xmath6 in accord with recent experimental results .
only a portion of the total moment resides on the ru sites ( 64% in the lapw mt sphere , and 67% in the lmto atomic sphere ) .
the electronic density of states has a gap in the spin majority channel which is only 20 mry above the fermi level .
the fact that srruo@xmath0 is so close to a half - metal is important for understanding its transport properties , and the fact that they are so sensitive to magnetic ordering ( and , correspondingly , to temperature ) .
experimentally , caruo@xmath0 is a paramagnetic metal .
this fact suggests that the rotation of ruo@xmath2 the octahedra is antagonistic to magnetism ( since larger rotations constitute the main structural difference between caruo@xmath0 and srruo@xmath0 ) .
however , this conjecture is apparently at odds with the calculated result that the equilibrium magnetization in srruo@xmath0 is smaller in an ideal cubic perovskite structure than in the actual distorted one . as a first step to understanding this ,
we have extended our calculations to caruo@xmath0 in its experimental structure .
details of the method are as in ref .. the resulting density of states is shown in fig.[cr3-dos ] .
we find that indeed the magnetism is suppressed in this case , though in a very borderline fashion . fixed spin moment calculations of the total energy as a function of spin magnetization for caruo@xmath0 show a very extended flat region , extending to near 1.5 @xmath6 per formula unit .
this is reminiscent of fcc pd which also shows such a flat region .
this borderline state implies a high spin susceptibility and explains the fact that low doping can induce a ferromagnetic state .
further , para - magnon like spin excitations should be very soft in this material and magnetic impurities may be expected to induce giant induced local moments .
there are already some reports that this is the case in caruo@xmath0 @xcite . having shown that the ferromagnetism in srruo@xmath0 and its suppression in caruo@xmath0 can be described using band structure methods
, we turn to the question of why these two perovskites have different magnetic properties .
to determine whether the key difference between the materials is structural we have performed calculations for caruo@xmath0 using the crystal structure of srruo@xmath0 .
these calculations yield a spin magnetization of 1.68 @xmath6 per formula unit and a magnetic energy of 0.06 ev / ru , very similar to srruo@xmath33 .
calculations for the intermediate structure formed by a linear average of the experimental caruo@xmath0 and srruo@xmath0 structures , yield a similar spin moment of 1.53 @xmath6 per formula unit and a magnetic energy of only 0.029 ev / ru ( note the similarity of the magnetizations and large variation of magnetic energy ) . to within the accuracy of our calculations
this paramagnetic - ferromagnetic energy difference becomes zero just at the experimental caruo@xmath0 structure .
since ferromagnetism in the ( sr , ca)ruo@xmath0 is apparently strongly coupled to the rotation of the octahedra , alloying the a site cation is expected to be an effective means for tuning the magnetic properties .
alloying caruo@xmath0 with larger divalent cations should generally induce ferromagnetism while alloying srruo@xmath0 with smaller cations should suppress ferromagnetism .
baruo@xmath0 , while a known compound , occurs in a different crystal structure and is not magnetic .
however , pb can be partially substituted on the sr site , and it is known that introduction of this slightly larger divalent cation does increase @xmath7 in srruo@xmath0 . later in the paper
we shall analyze the transformation of the band structure of ( sr , ca)ruo@xmath0 upon increase of the tilting in more detail and will show that the nonmonotonic dependence of the equilibrium magnetization on tilting is a straightforward consequence of a natural evolution of the band structure near @xmath34 with the structural distortion .
the electronic and magnetic structure of sr@xmath1yruo@xmath2 was calculated using the full experimental crystal structure of battle and macklin @xcite except that the very small ( 0.23% ) lattice strain was neglected .
additional calculations were performed for idealized structures neglecting the tilting of the octahedra to help understand the role of this distortion , which changes the angles and distances along the ru - o - o - ru hopping paths .
these local density approximation calculations were performed using the general potential lapw method @xcite including local orbital extensions @xcite to accurately treat the o @xmath35 states and upper core states of sr and y as well as to relax any residual linearization errors associated with the ru @xmath12 states .
a well converged basis consisting of approximately 2700 lapw basis functions in addition to the local orbitals was used with o sphere radii of 1.58 a.u . and
cation radii of 2.10 a.u .
this self - consistent approach has a flexible representation of the wavefunctions in both the interstitial and sphere regions and makes no shape approximations to either the potential or charge density .
as such it is well suited to materials with open structures and low site symmetries like sr@xmath1yruo@xmath2 .
in addition , we used the lmto method in the atomic sphere approximation and tight - binding representation@xcite ( stuttgart code , version 4.7 ) to get better insight in the calculated electronic structure .
the lmto - asa method is less accurate than the full - potential lapw but it provides more flexibility in the way how the results are represented and how they can be analyzed in tight - binding language .
calculations were performed at the experimental structure for ferromagnetic ( f ) and the observed antiferromagnetic ( af ) orderings .
the af ordering is 0.095 ev / ru lower in energy than the f ordering , and has an insulating gap in the band structure , consistent with the experimental ground state .
the insulating gap of 0.08 ev is between majority and minority spin states and may yield only a weak optical signature .
the ru moment as measured by the magnetization within the ru lapw sphere is 1.70 @xmath6 for the af state and 1.80 @xmath6 with the f ordering , in reasonable accord with the neutron scattering results the similar moments with different spin configurations suggests that a local moment picture of the magnetism is appropriate for sr@xmath1yruo@xmath2 .
this is in contrast to perovskite srruo@xmath0 .
similar to srruo@xmath0 there are substantial moments within the o lapw spheres as well as the ru spheres , amounting to approximately 0.10 @xmath6/o ( af ordered ) and 0.12 @xmath36/o ( f ordered ) .
these can not be understood as tails of ru 4d orbitals extending beyond the lapw sphere radii , since such an explanation is inconsistent with the radial dependence of these orbitals , but rather they arise from polarization of the o ions due to hybridization , which is evidently strong both from this point of view and from the calculated electronic structure , discussed below .
the total local moment per formula unit is of mixed ru and o character and amounts to 3 @xmath6/f.u .
, which is approximately 60% ru derived and 40% o derived ( the interstitial polarization of 0.5 - 0.7 @xmath6/cluster derives from both ru and o , but is assigned as mostly o in character based on the extended 2p orbitals of negative o ions and the small o sphere radius , and results of lmto - asa calculations which do not have any interstitial volume ) .
the calculated exchange splittings of the o 1s core levels are 80 to 95 mev depending on the particular o site .
the o polarizations may be observable in neutron experiments if o form factors are included with ru in the refinement . such an experiment
is strongly suggested by the present results .
projections of the electronic density of states ( dos ) of antiferromagnetic sr@xmath1yruo@xmath2 onto the lapw spheres are shown in fig .
[ dos - syr - af ] , where majority and minority spin projections onto a ru ion and the six o ions in its cluster are shown .
the dos in two spin channels are similar in shape apart from an exchange splitting throughout the valence energy region and shows evidence of a strongly hybridized electronic structure .
the details of this structure are deferred to the tight binding analysis below , except to mention the exchange splitting of the essentially pure o 2p states between -4 and -6 ev relative to the fermi energy ( @xmath34 ) and the fact that there are substantial ru 4d contributions to the minority spin channel between -4 and -6 ev as well as o contributions above @xmath34 implying that the average ru 4d occupancy is considerably higher than d@xmath37 .
although assigning charge in a crystal to various atoms is an ambiguous procedure , integration of the d - like dos implies an average near d@xmath38 similar to perovskite srruo@xmath0 .
the magnetic moments derive from polarization of three bands near @xmath34 by an exchange splitting of 1 ev .
the f ordered dos ( fig .
[ dos - syr - f ] ) is very similar to that in the af state , but the exchange splitting is somewhat smaller and the bandwidth somewhat larger , resulting in a slight semimetallic overlap of majority and minority spin bands at @xmath34 which reduces the spin moment from 3.0 to 2.97 @xmath6/f.u .. parallel calculations were performed using a structure in which the tilting of the ruo@xmath2 clusters is suppressed . as with the actual experimental structure , the af ordering is lower in energy than the f ordering .
however , in this case the band structures are metallic for both orderings , showing that the tilting is crucial for the insulating state . as will be discussed below
there is a substantial coupling between the magnetic order and this structural degree of freedom .
somewhat unexpectedly , the easiest compound to understand is the sr@xmath1yruo@xmath2 double perovskite .
sr and y , as is common in perovskites , are fully ionic , so that the states around the fermi level barely have any sr or y character .
thus , as mentioned , this compound can be viewed as consisting of rigid ruo@xmath39 octahedra , arranged on an fcc lattice , and loosely connected to each other .
we will show below that this intuitive picture provides very good qualitative and quantitative interpretation of the full - scale band structure calculation .
in contrast with sr@xmath1ruo@xmath3 or sr@xmath40ca@xmath41ruo@xmath9 no octahedra share oxygens . the octahedra are slightly tilted , which we shall neglect for the moment ( the effect of tilting is in a certain sense important and will be discussed later ) .
accordingly , we begin by discussing a single cluster .
the electronic structure of a single ruo@xmath2 cluster is governed by the relative position of ru @xmath12 and o @xmath14 levels , and the corresponding hopping amplitudes .
the ru @xmath12 states are split by the crystal field into two manifolds consisting of 3 @xmath26 and 2 @xmath27 levels , respectively , and these are separated by @xmath42 ev .
the o @xmath14 levels are subject to a crystal field splitting at least three times smaller , and yield 9 @xmath43 states , which form @xmath44 bonds with ru , plus three @xmath45 states , which participate in the @xmath46 bonding . after including @xmath47 hopping ,
the system of levels becomes , for each spin channel : 13 nonbonding : 4@xmath48 5 bonding : 2@xmath49 and 5 antibonding : 2@xmath50 where @xmath51 are pure ionic levels , and @xmath52 ^ 2 + 16t_\sigma ^2}\},$ ] @xmath53 ^ 2 + 16t_\pi ^2}.$ ] the actual ordering of levels in ruo@xmath8 as shown on fig.[levels ] is @xmath54 .
the last inequality leads to a substantial gap ( @xmath55 2 ev ) between the antibonding @xmath56 and the antibonding @xmath57 bands in the solid .
this large gap is only partially due to the crystal field , and arises largely from the stronger ( relative to @xmath58 @xmath59 bonding .
the exchange splitting is , naturally , weaker than this enhanced crystal field splitting , and the hund s rule does not apply to the high - lying antibonding @xmath60 states , which remain empty in both spin channels . neglecting those states , there are 21 levels to be occupied by 39 valence electrons . here
hund s rule does apply and tells us to populate all 21 spin - majority states , and all but the three antibonding @xmath56 levels in the spin - minority channel .
thus for the electronic properties of the crystal these six spin - up and spin - down @xmath56 molecular orbitals are of primary relevance ; the symmetry of these orbitals is the same as for @xmath61 states in a transition metal ion .
we now use this information to analyze the electronic structure of crystalline sr@xmath1yruo@xmath2 .
when a solid is built out of the clusters , the molecular levels broaden into bands , which however remain quite narrow in this material .
although the main intercluster hopping occurs via @xmath62 matrix elements ( here and below we mean for @xmath12 the ru - o molecular orbitals with the effective @xmath19symmetry ) , the intercluster distance is large and the effective hopping amplitude is small .
thus one may conjecture that the valence band formed out of the majority spin molecular @xmath56 orbitals , and the corresponding minority spin band do not overlap , and that the crystal , in either the ferro- or antiferromagnetic states , remains insulating .
a more detailed analysis , as discussed later in the paper , reveals a difference between the ferro- and antiferromagnetic ordering , namely that the bandwidth is slightly larger , and the exchange splitting slightly smaller in the former case .
in fact , our lda calculations , described above , yield an insulating antiferromagnetic ground state , with a small gap of about 0.07 ev ; they also give a metastable semimetallic ferromagnetic state , with a band overlap of a few mev .
let us now analyze this band structure in the tight - binding terms .
a nearest neighbor model should be a good starting approximation .
let us begin with the ferromagnetic case , and consider the undistorted crystal structure ( no tilting of oxygen octahedra ) .
the main parameter is now the @xmath63 hopping amplitude , @xmath64 in the nearest neighbor approximation , the three @xmath56 bands do not hybridize with each other .
each of them , however , disperses according to @xmath65 @xmath66 and the corresponding permutations of @xmath67 including @xmath68 hopping , the bands hybridize among themselves , resulting in a further increase in the bandwidth .
the calculated lda bands have widths of approximately 1.1 ev , corresponding to @xmath69 ev .
@xmath68 hopping effects are responsible for the deviations from the dispersion above by about 0.1 ev .
importantly , there is no repulsion between the valence bands and the conduction bands , because they are fully spin polarized with the opposite spins . this situation changes , however , in the antiferromagnetic case .
the observed magnetic ordering corresponds to ferromagnetic 001 planes stacked antiferromagnetically .
each ruo@xmath2 cluster has thus 4 neighbors with the same and 8 neighbors with the opposite spin .
correspondingly , of three @xmath56 derived bands one ( @xmath70 ) remains essentially the same as in the ferromagnet , and two other loose their dispersion to the first order in @xmath71 since the relevant neighboring clusters have only states of the opposite spin at this energy .
instead , for those bands there is a hybridization between the valence and the conduction bands , because now the orbitals with the same spin on the neighboring clusters belong to these different bands .
the hybridization matrix element is @xmath72 and produces an additional bonding energy @xmath73 per cluster ( @xmath74 ev is the exchange splitting ) .
this yields about 0.08 ev which is very close to the calculated lda energy difference ( 0.12 ev ) between the afm and fm configurations in the ideal undistorted structure . in the actual crystal structure
the oxygen octahedra are tilted by about 12@xmath75 , so that @xmath76 is reduced by about 15% ( neglecting @xmath77 etc . ) , yielding @xmath78 ev .
our first - principles lapw calculations give for the bandwidth approximately 0.9 ev , that is @xmath79 ev , and @xmath80 ev . the calculated lapw energy difference is 0.095 ev , the same reduction from the undistorted case as given by the simple tight binding estimate above .
it is worth noting that while this mechanism gives an effective antiferromagnetic exchange interaction @xmath81 the underlying physics is very similar to , but not identical with , the usual superexchange interaction in 3@xmath12 oxides , @xmath82 .
the differences are that instead of metal - oxygen - metal hopping here the relevant hopping is direct cluster - cluster hopping and the energy denominator is the band gap due mainly to intracluster exchange , rather than coulomb correlations described by a hubbard @xmath16 .
one can also estimate the neel temperature , using the above value for @xmath83 to do that , let us begin with noting that this system represents a very good approximation to the antiferromagnetic nearest neighbor fcc model .
the strong magnetoelastic coupling discussed below does not favor non - collinear spin configurations , so the direction of the cluster magnetic moments is fixed .
the magnetic coupling @xmath84 with the next nearest neighbors can be safely neglected .
indeed , it is governed by the @xmath62 hopping .
although @xmath85 is larger than @xmath86 usually by a factor of the order of 2 , the larger distance , in the canonical scaling @xmath87 gives a factor of @xmath88 and the energy denominator in the equation for @xmath89 is about 10 times larger . taken together , one expects @xmath84 to be at least two order of magnitudes smaller than @xmath83 the antiferromagnetic fcc ising model is well studied@xcite . despite magnetic frustration
, it has a neel temperature of approximately 1.76@xmath90 for the spin 1/2 and approximately 1.33 @xmath90 for the spin 1 , which in our case corresponds to 700900 k. the measured @xmath11 is 26 k , in apparent severe disagreement with our estimate . it is tempting to ascribe this to intracluster hubbard - like correlation effects , which can increase the gap and reduce @xmath83 moreover , since the @xmath26 band width is only 1 ev , even a moderate hubbard repulsion could affect @xmath91 .
one can get a very rough upper estimate this effect as follows : the energy of the coulomb repulsion of two electrons placed in two @xmath26 orbital on the same cluster is ( assuming about equal population on ru and o ) @xmath92 where @xmath93 is the coulomb repulsion of two electrons localized on two neighboring oxygens etc .. @xmath94 ev ; @xmath95 is believed to be about 1.5 ev@xcite , and for @xmath96 we use 3 ev , keeping in mind that the charge - transfer metal - oxygen energy for the 3d oxides is about 4.5 ev and the metal - oxygen distance is 50% smaller there . then , we arrive at @xmath97 ev .
it is unclear to what extent this @xmath16 will be reduced by screening by surrounding cluster and by intracluster charge redistribution , but this effect would definitely be substantial .
anyway , using 3 ev as a very safe upper bound , we get for the lower bound on 2@xmath91 approximately 0.03 ev , which corresponds to @xmath15 of at least 300 k. thus , strong correlations alone can not explain anomalously low neel temperature of this compound .
another possibility to reduce the transition temperature is magnetoelastic coupling , which is subject of the next section .
the fact that magnetic excitations and phonons are coupled in ruthenates is known@xcite , but not well understood from a microscopic point of view . in the case of sr@xmath1yruo@xmath2 it is , however , reasonably clear : with increasing tilting angle the @xmath98 hopping must decrease and with it the antiferromagnetic stabilization energy and effective exchange constant @xmath83 this is confirmed by our first - principle results . in other words , magnetic excitations flipping the spin of a ruo@xmath2 cluster are coupled with this phonon mode , changing the tilting angle ( which is the soft mode for the transition from the cubic structure into the tilted one ) .
a dimensionless coupling constant may be defined as @xmath99 where @xmath100 is the phonon coordinate .
here @xmath101 is the displacement of oxygens from their equilibrium positions , and @xmath102 is the frequency of the phonon . very roughly , @xmath103 where @xmath104 is the energy difference between the cubic and the distorted structure , taken per one oxygen , and @xmath12 is the equilibrium oxygen displacement . from our calculations , @xmath105 mev .
experimentally , @xmath106 .
thus , @xmath107 @xmath108 .
now , using @xmath109 where @xmath110 is the tilting angle , we can estimate @xmath111 @xmath112 in fact , linear interpolation of @xmath91 between the cubic and equilibrium structure gives the same number for @xmath113 thus @xmath114 is about 0.17 for this phonon mode , which means that the characteristic ( e.g. , zero - point motion ) amplitude of the librations of the octahedra around their equilibrium position will produce sizable changes in the effective exchange constant .
the thermodynamics of such a system is interesting and unusual , but its discussion goes beyond the scope of this paper .
it is important to note , however , that the long - range order in the nearest - neighbors antiferromagnetic fcc ising model appears exclusively because of the finite temperature entropy contribution to the free energy@xcite . while at @xmath115 there is an infinite number of degenerate states , ordered in two dimensions and disordered in the third one , at @xmath116 this degeneracy
is lifted because of different spectra of low - energy spin - flip excitations in the different ground states .
as long as such spin - flip excitations are coupled with the phonons , the standard consideration of the afm fcc ising model does not apply , and the transition is not necessarily at @xmath117 .
however , the long - range two - dimensional afm correlations should be present up to @xmath118 , and could in principle be seen in some experiments .
the above discussion of sr@xmath1yruo@xmath2 magnetic properties was based on the molecular ( cluster ) picture , and we observed that oxygen plays a crucial role in formation of the magnetic state .
the same conclusion can be obtained , starting from the extended band picture .
the standard approach to magnetism in the band theory goes back to stoner and slater@xcite .
they considered non - interacting electrons in the paramagnetic state , and added their exchange interaction in an average form , @xmath119 where @xmath120 is total magnetization and @xmath121 is independent of @xmath120 .
the magnetic susceptibility of such a system can be written as @xmath122 where @xmath123 is pauli susceptibility .
if magnetization is measured in bohr magnetons , then @xmath124 the density of states per spin at the fermi level .
the instability occurs when @xmath125 diverges , that is , when @xmath126 becomes larger than 1 .
. [ stoner ] can of course be viewed as an approximation in the framework of the general linear response theory .
however , such an approximation is highly uncontrollable , and even the splitting of the right - hand part of eq .
[ stoner ] into two terms can not be derived in a systematic way .
more instructive is application of the stoner method to the density functional theory . in dft
, total energy change is exactly written as sum of the change in the one - electron energy , which for small @xmath120 is @xmath127 and the change in the interaction energy , which is ( @xmath128 . here
@xmath129 is the effective kohn - sham magnetic field averaged over the sample ( because stoner theory assumes a uniform internal ferromagnetic field ) , and , @xmath130 the utility of the stoner approach in dft is due to the fact that usually there are very few orbitals whose occupancy substantially influences @xmath131 and therefore @xmath121 is easy to calculate in a quasiatomic manner , using , for instance , the quasiatomic loop in standard lmto codes . in practice , in quasiatomic calculations
one changes the occupation of a given orbital , transferring some charge from the spin - up to spin - down quasiatomic level , recalculates the lsda potential and determines how large the induced splitting of quasiatomic levels is .
if different kinds of atoms in a solid contribute to the density of states at the fermi level , one has to take into account the magnetization energy for each of them .
this means that the total stoner @xmath121 for such a solid is the average of the individual ( quasiatomic ) @xmath121 s with the squared partial density of states . indeed , suppose the states at the fermi level are a superposition of orbitals from several atoms , so that @xmath132 ( where @xmath133 labels the atoms ) . applying a uniform magnetic field creates a magnetization @xmath134 where @xmath135 is magnetization of the @xmath133-th atom . by definition ,
the intraatomic energy change is @xmath136 thus , the total @xmath137 so formulated , the stoner theory applies to infinitesimally small changes in magnetization and essentially determines whether or not the paramagnetic state is stable against ferromagnetism .
it is , however , a reasonable assumption that this theory holds , approximately , for finite magnetizations as well .
one has , however , to modify the one - electron energy term @xmath138 to account for the energy dependence of the density of states , within the rigid - band approximation .
then , the spin splitting producing a given magnetization @xmath120 can be defined as @xmath139 where @xmath140 is the density of states averaged between the fermi level of the spin - up and spin - down subbands .
for the one - electron energy one obtains @xmath141 because one has to move @xmath142 electrons up by @xmath143 integrating this expression , one arrives at the so - called extended stoner theory@xcite , which uses the following expression for the total magnetization energy : @xmath144 minimization of this energy leads to the extended stoner criterion , which states that stable ( or metastable ) values of the magnetic moment are those for which @xmath145 and @xmath146 the paramagnetic state is ( meta)stable when @xmath147 stoner theory is , in principle , formulated for a ferromagnetic instability . however , unless the fermi surface topology specifically favors ( or disfavors ) the antiferromagnetic instability with a given vector * q * , one can assume that @xmath148 indeed , in many cases if a material comes out magnetic from the calculations , the energy difference between ferro- and antiferromagnetic ordering is small compared with the magnetic stabilization energy .
as we shall see , this is the case in sr@xmath1yruo@xmath8 but not in srruo@xmath9 and the reason is that in the latter the stoner factor @xmath121 is very different for ferro- and antiferromagnetic arrangements .
now let us consider how one can describe the magnetism in sr@xmath1yruo@xmath2 from the stoner point of view .
calculation of stoner parameters @xmath121 s is straightforward in the lmto method@xcite , which divides space into atomic spheres . in the popular stuttgart lmto - tb package it is possible to change occupancy of any atomic orbitals and to calculate the resulting change in atomic parameters , in particular the shift of the corresponding band center @xmath149 with the spin - up and spin - down occupancies split by @xmath150 the stoner parameter is ( @xmath151 we obtain @xmath152 of about 0.7 ev , and , importantly , find that the o@xmath153 states in ruthenates also have substantial stoner parameter , @xmath154 ev . the density of the ru @xmath12 states is approximately twice larger than that of the three o @xmath14 states
thus , the total stoner parameter for sr@xmath1yruo@xmath2 is @xmath155 ev . correspondingly , the paramagnetic state is unstable unless @xmath156 states / spin / ev / formula . the paramagnetic lmto density of states of cubic sr@xmath1yruo@xmath2 ( that is , with breathing , but with no tilting distortion ) near the fermi level is shown in fig .
[ syr - dos - nm ] .
it has the narrow @xmath56 band half filled , and @xmath157 is close to 4.5 states / spin / ev / formula .
this is much larger than @xmath158 , so the paramagnetic states is very unstable . on the other hand ,
the average density of states in the @xmath56 band is not that large , @xmath159 ev=2 states / spin@xmath160ev , thus in the cubic structure this band will not be fully polarized . integrating the density of states shown in fig .
[ syr - dos - nm ] , we obtain the extended stoner plot for sr@xmath1yruo@xmath2 ( fig . [ sryc - st ] ) , and observe that the equilibrium magnetization is slightly smaller than 2 @xmath6 , and the ground state is semimetallic , in agreement with the self - consistent spin - polarized lmto calculations , as well as with the more accurate lapw calculations .
another fact that one can observe from fig .
[ sryc - st ] is that if oxygen would not contribute in the total stoner factor , that is , if total @xmath121 would be only @xmath161 ev , equilibrium magnetization would be very small , approximately 0.4 @xmath6 , and of course with a much smaller gain in energy . as we shall see below ,
this is the case in srruo@xmath0 , where in the antiferromagnetic structure oxygen ions can not polarize by symmetry . in sr@xmath1yruo@xmath2
, however , oxygen fully contributes into the magnetic stabilization energy both in ferro- and in antiferromagnetic structure , so the next order mechanisms decides which magnetic order realizes .
such an additional mechanism is discussed in the previous section hybridization repulsion between the filled and the empty @xmath56 bands , which stabilizes the antiferromagnetic structure . the main structural difference from the double perovskite , sr@xmath1yruo@xmath2 is that now all oxygen ions are shared between two rutheniums , so one can not make use of a single ruo@xmath2 cluster concept . as with sr@xmath1yruo@xmath8
we shall start by analyzing the band structure with non - tilted octahedra , that is , with the cubic perovskite structure .
per cubic cell we have , in each spin channel , two ru @xmath27 states , strongly hybridized with 3 o @xmath45 orbitals , and three ru @xmath26 states , hybridized with 6 o @xmath43 orbitals . in the nearest neighbor approximation , these @xmath46 bands do not mix with the @xmath44 bands , and the @xmath44 bands , in turn , consist of three sets of mutually non - interacting @xmath162 @xmath163 , and @xmath164-like bands . the nearest neighbor
tb hamiltonians have the form ' '' '' @xmath165 and @xmath166 ' '' '' 2 where @xmath167 etc . for each @xmath26 manifold three bands
appear : one non - bonding at @xmath168 and one bonding - antibonding pair at @xmath169 ^ 2 + 16t_\pi ^2(s_x^2+s_y^2)}.$]analysis of the calculated band structure shows that @xmath170 so , neglecting oxygen - oxygen hopping @xmath171 the dispersion is approximately @xmath172 where @xmath173 ev .
ru @xmath174 orbitals are split off from the @xmath26 orbitals by about 3 ev . as in sr@xmath1yruo@xmath8
the crystal field effect on oxygen states is weaker : the o @xmath175 states are less than 2 ev below o p@xmath176 states .
the energy distance between ru @xmath27 and o @xmath45 levels is nearly 5 ev , so a good approximation is @xmath177 applying lwdin perturbation theory to fold down the oxygen states , we get for ( antibonding ) @xmath60 bands the effective hamiltonian ' '' '' @xmath178 ' '' '' 2 which yields two bands with dispersion @xmath179 the formal valency of ru in sr@xmath40ca@xmath41ruo@xmath0 is 4
. the total number of electrons , populating the ru - o valence bands , is 22 .
this means that the bonding ( mostly oxygen ) @xmath60 bands are filled , as well as the bonding and nonbonding @xmath56 bands .
the conduction band is the antibonding @xmath56 band , with its 6 states filled by 4 electrons .
this band has a strong ( logarithmic ) van hove singularity at half filling .
however , direct oxygen - oxygen hopping @xmath180 ev , which we have initially neglected , moves this singularity upwards to the position which corresponds to approximately 63% filling ( 3.8 electrons ) and makes the singularity sharper .
this is the pronounced peak at @xmath34 in our first principles paramagnetic dos@xcite .
such a situation , where the fermi level nearly exactly hits a logarithmic peak in the density of states , is energetically unfavorable , and leads to an instability , which can be either magnetic , or a sufficiently strong lattice distortion , or both .
the calculated partial densities of ru ( d ) and of the three o ( p ) states at the fermi level in sr@xmath40ca@xmath41ruo@xmath0 are approximately 70% and 30 % , respectively .
correspondingly , @xmath181 ev . without the oxygen stoner parameter , @xmath182 as mentioned above , our lapw calculations yield for srruo@xmath0 in the cubic structure relatively small magnetization of 1.17 @xmath183 .
the reason for that is that the density of state is piled near the fermi level , and drops quickly when one goes away from it .
[ sr3+cr3-st ] shows how this is reflected in the effective density of states @xmath184 : it decreases rapidly with magnetization , and becomes equal to @xmath158 at @xmath185 . for a moderate
tilting , corresponding to actual srruo@xmath0 structure , @xmath186 is smaller than in the cubic structure , but it decreases rather slowly with @xmath120 and remains larger than @xmath158 much longer .
two questions arise in this connection : why the ground state is ferromagnetic , and not antiferromagnetic , and why in the actual crystal structure is caruo@xmath0 not magnetic at all ? the first question is particularly easy to answer . in an antiferromagnetic structure ,
oxygen ions occur between opposite spin ru ions , and thus by symmetry have zero net polarization .
correspondingly , the total stoner parameter @xmath121 is smaller and so is magnetic stabilization energy and the equilibrium magnetization on ru . as we shall see below , tilting has a substantial effect on the effective density of states , and for large tiltings the ground state becomes paramagnetic .
it follows from the above discussion , however , that the ground state is always either ferro- or paramagnetic .
this situation is in sharp contrast with classical localized magnetic materials , like nio or feo , where even when ferromagnetism is imposed , oxygen is polarized only very weakly , and the magnetization of the metal ion is even smaller than for the antiferromagnetic state .
it is also in contrast with the ferromagnetic colossal magnetoresistance mn oxides , where the antiferromagnetic ground state is destroyed by the double exchange interaction , competing with superexchange .
these ruthenates have integer occupancy of the valence band , and thus double exchange is not operative
. however , covalency effects , which are strong because of the large @xmath47 hopping and the near - degeneracy of the ru @xmath187 and o @xmath188 states , are operative .
this strong covalency is what requires part of the magnetic moment to reside on the oxygen , since exchange splitting the ru @xmath12 states without the o would require disrupting the covalent bonding . in the crystal structures where it is possible to maintain o moments without ferromagnetic ordering , an antiferromagnetic state is likely to form ( as in sr@xmath1yruo@xmath189 but where it is not possible , like srruo@xmath9
a ferromagnetic ground state occurs instead .
it is also worth noting that besides the double perovskite sr@xmath1yruo@xmath8 where oxygen ions can polarize both in the ferro- and antiferromagnetic structure , and single perovskites sr@xmath40ca@xmath41ruo@xmath0 , there exist intermediate layered structures , which consist of perovskite ( sr , ca)o@xmath1 layers . using the same arguments
, we conjecture that if such compounds are magnetic , the effect of oxygen will cause ferromagnetic ordering inside layers , while interlayer coupling is strong ferromagnetic if the layers are sharing apical oxygens , but may be antiferromagnetic if they are connected by intermediate rocksalt layers ( like in sr@xmath1ruo@xmath3 ) .
the observed crystal structure of both srruo@xmath0 and caruo@xmath0 is characterized by a substantial tilting of the ruo@xmath2 octahedra . in srruo@xmath0
the octahedra are rotated by 8@xmath190 and in caruo@xmath0 the distortion is about twice larger . in fig.[sr3+cr3 ] we show the density of states in the @xmath56 band for these three different structures .
there are two interesting effects on the electronic structure , associated with tilting .
one is that hybridization between the @xmath191 and @xmath60 bands becomes possible .
this broadens the logarithmic singularity in the density of states . at the same time
the bands become more narrow and the gap between the antibonding @xmath56 and @xmath60 bands grows . on the other hand ,
the unit cell is quadrupled so new bragg reflections appear .
these yield pseudogaps at the new brillouin zone boundaries , occurring at energies close to half filling ( e.g. , along @xmath192 and @xmath193 directions ) as well at two - third filling ( e.g. , along @xmath194 direction ) .
this second pseudogap thus appears to be near the fermi level .
one factor , band narrowing , tends to increase the equilibrium magnetization , but another one , the second pseudogap at the fermi level , works against it .
the actual trend looks like this : at small distortions the equilibrium magnetization grows . at some critical distortion magnitude , which is not far from the observed equilibrium distortion for srruo@xmath0 ,
the magnetization reaches a maximum and starts to decline .
the first principles calculations show little difference between srruo@xmath0 and caruo@xmath17 provided the same crystal structure is used , so the main difference in the observed behavior is indeed due to the different distortion magnitudes . to understand the changes caused by the tilting distortion it is instructive to look at the extended stoner plots for different distortions .
[ sr3+cr3-st ] shows such plots for srruo@xmath0 in the experimental structure , in the cubic ( ideal perovskite ) structure , and for caruo@xmath0 , as well as for caruo@xmath0 in the srruo@xmath0 structure .
one may immediately note the extreme instability of the cubic structure , due to discussed peak at the fermi level .
however , because the density of states is piled up near the fermi level , the resulting exchange splitting is small compared with the band width . for the moderate
tilting , like that in the experimentally observed srruo@xmath0 structure , the peak broadens and it takes larger exchange fields to fully split this peak into occupied and unoccupied peaks .
finally , at even larger tiltings , corresponding to caruo@xmath0 , the peak is suppressed . in the effective density of states , as shown on fig.[sr3+cr3-st ] , this results in a nearly flat plateau , extending from @xmath195 to @xmath196 @xmath197 accidentally , this plateau matches nearly exactly @xmath198 calculated as described in the previous section . in other words , the total energy of caruo@xmath0 is nearly independent of magnetization up to @xmath199 @xmath200 the total energy as a function of magnetization is shown on fig.[fixedm ] , where the results of the fixed - spin - moment lapw calculations are compared with the same energy differences in the stoner theory@xcite .
we conclude that although caruo@xmath0 is nonmagnetic in its ground state , long - wave paramagnons should be extremely soft in this compound .
this should effect the transport , magnetic and electronic properties .
unusual transport properties of srruo@xmath0 are due to the following three peculiarities : ( 1 ) while the dos in both spin subsystems are nearly the same , the partial plasma frequency in the majority spin channel is 3 times larger than in the minority spin channel , a manifestation of the proximity to the half - metallic regime , which would occur if the magnetization were 2 @xmath36 instead of 1.59 @xmath197 ( 2 ) there is strong coupling between electrons , phonons , and magnons , which probably produces substantial spin - flip scattering of electrons , and ( 3 ) in both spin channels the fermi surfaces consist of several sheets of complicated topology : hole - like , electron - like , and open , so that hole - like and electron - like parts compensate each other .
let us start with the electric resistivity , and assume for simplicity that two bands are present , spin - up and spin - down .
let us further assume that the sources of the resistivity are scattering of electrons by phonons , with the coupling constant @xmath201 and by magnons , with the coupling constant @xmath202 since the dos are approximately equal , @xmath203 st / ry , also @xmath204 the specific heat renormalization in each band is now @xmath205which would need to be @xmath206 to agree with experiment@xcite . in the lowest - order variational solution of the boltzmann equation , given by allen@xcite
( see also ref . ) , the resistivity of such a system at sufficiently high temperature is @xmath207 where the so - called `` scattering - in '' term which is usually small in cubic crystals is neglected , and @xmath208 is the partial plasma frequency squared ( one can find in the literature @xcite a so - called `` two - current formula '' which gives the same result when the `` scattering - in '' term is neglected .
there are some differences between the formulas of refs . and , which we discuss in the appendix .
) . from our first principles calculations , @xmath209 ev and @xmath210 ev . in the nonmagnetic phase @xmath211 ev , the same as the total plasma frequency in the ferromagnetic phase . at @xmath212 k and up to the curie temperature
the resistivity is reported to be linear@xcite . the linear coefficient ( @xmath213 @xmath214cm / k ) corresponds to @xmath215 very close to the number extracted from the electronic specific heat . above @xmath7 the resistivity changes slope , remaining linear up to at least several hundred kelvin .
the slope is , however , smaller than below @xmath216 and corresponds to @xmath217 where @xmath218 is the electron-_para_magnon coupling constant .
this value differs from that quoted in ref . , because of considerable differences in the calculated band structure .
thus , we conclude that the high - temperature resistivity of srruo@xmath0 indicates rather strong electron - paramagnon , and even stronger electron - magnon coupling , with the reservation that probably in this system one can not really separate electron - phonon and electron - magnon scattering completely because the corresponding degrees of freedom are coupled .
the problem noted in ref .
, namely that at high temperatures the mean free path is comparable to the lattice parameter , yet no saturation is seen in the resistivity , remains .
the resistivity of caruo@xmath0 has also been studied . in the studies reported in the literature
@xcitethe high - temperature resistivity shows the same slope as in srruo@xmath9 in full agreement with our observation that the electronic structure of both compounds is very similar . at low temperatures ,
however , the resistivity behaves very differently , namely it increases nearly linearly at small @xmath219 with a large slope .
the slope decreases eventually and at the room temperature the behavior becomes similar to that of srruo@xmath220 this low - temperature linearity indicates that the excitations responsible for resistivity ( apparently , paramagnons ) soften at @xmath221 indicating a magnetic instability at @xmath115 or at slightly negative @xmath219 ( in curie - weiss sense ) .
this is in agreement with our result that caruo@xmath0 is on the borderline of a ferromagnetic state .
again , paramagnons are strongly coupled with phonons , and this leads to the large coupling strength .
the low - temperature resistivity has also attracted attention .
experimentally , the resistivity initially increases rather quickly .
allen et al @xcite observed a low - temperature power law @xmath222 klein et al @xcite found that below 10 k the resistivity can be reasonably well fit with a quadratic law , but an even better fit ( linear ) , was found for up to 30 k for the dependence of resistivity on magnetization .
we interpret this observation as follows : a stronger than @xmath223 increase indicates that the excitations , responsible for the low - temperature scattering , have a sublinear dispersion .
conventional magnons , with @xmath224 dispersion , produce @xmath225 in good agreement with the experiment .
in fact , the experimental exponent is even below 2 , which is easily accountable for by the fermi surface effects : part of the temperature dependence comes from the term @xmath226 if it is proportional to @xmath227 this not the case in srruo@xmath9 where one of the two fermi surfaces ( majority spin ) is a small sheet with heavy electrons ( similarly , magnetic alloys where momentum conservation does not hold show @xmath228 see ref . ) .
a possible problem with this interpretation of the low - temperature resistivity is that as was already noted@xcite , in elemental ferromagnets , the magnon - limited resistivity is almost three orders of magnitude smaller than what would be needed to explain the low - temperature resistivity of srruo@xmath0 ( where @xmath229 @xmath230 @xmath214cm / k@xmath231 this can be resolved if we invoke the anomalously large magnon - phonon coupling , which , as discussed above , originates from the crucial role played by oxygen in magnetic properties of ruthenates .
the strong electron - magnon coupling at low temperature in srruo@xmath0 is closely related to the large electron - paramagnon coupling at high temperatures and in caruo@xmath220 one can make a rough estimate of the characteristic frequency of magnons responsible for the resistivity : the schindler - rice formula@xcite , derived for the @xmath232 paramagnon scattering should be qualitatively applicable here , because we also have light electrons which carry current and are scattered by magnons into a heavy , transport - inert band .
this formula reads @xmath233 \\ j_n(x ) & = & \int_0^x\frac{4z^ndz}{\sinh ^2(x/2)},\end{aligned}\ ] ] and has asymptotic behavior at @xmath234 as @xmath235 and at @xmath236 as @xmath237 where @xmath238 is characteristic energy of the magnons . using the experimental number ( @xmath239 @xmath214cm / k@xmath240 and assuming that the magnon - limited part of the high - temperature resistivity is @xmath241 @xmath242cm / k , we arrive at @xmath243 k , which is a low , but not impossible number .
the hall coefficient in srruo@xmath0 and caruo@xmath0@xcite has attracted considerable attention . in both compounds
the hall constant @xmath244 shows an unusual temperature dependence , changing sign at @xmath245 k. at this point , however , the similarity ends . for each given temperature
the hall resistivity @xmath246 in caruo@xmath0 is nearly perfectly proportional to the field , as it should be for ordinary hall processes . in srruo@xmath9 to the contrary , @xmath247 decreases substantially with temperature , and only well above @xmath15 does @xmath246 become a linear function of @xmath248 this closely resembles the so - called extraordinary hall effect in ferromagnets .
the physics of the extraordinary hall effect is as follows : below @xmath249 the internal magnetic field is much larger than that applied in a typical hall experiment .
however , the hall currents induced in different magnetic domains mutually cancel . the applied field acts by aligning domains and lifting this cancellation .
this process defines the large slope of @xmath247 in low fields . at a field close to the saturation magnetization
@xmath250 all domains are aligned and further change of the hall current is due to the applied field itself ( the ordinary hall effect ) .
it is tempting to associate the nonlinear field dependence of the hall resistivity in srruo@xmath0 with this effect .
however , this hypothesis has been discounted by the authors of ref . for the following reason : in standard extraordinary hall effect theory the intersection of the linear low - field and high - field asymptotes occurs at @xmath251 in srruo@xmath0 the position of the intersection is roughly the same for all temperatures below @xmath249 and falls between 3 and 4 t. magnetometer data
show that @xmath250 is about 0.12 t at @xmath252 k , and , naturally , drops to zero at @xmath253 furthermore , a closer look at the data reveals that the slope of the hall coefficient changes in a smooth manner , unlike conventional ferromagnets , where it changes rather sharply near @xmath254 @xmath255 studies of the bulk magnetization in polycrystalline@xcite and single crystal@xcite samples of srruo@xmath0 show that the magnetization is not saturated even at applied fields of several t. this has been ascribed to the strong magnetocrystalline anisotropy , as measured by kanbayashi@xcite , and expected for a 4d magnet .
although hysteresis measurements for the thin film samples on which hall measurements were taken apparently showed saturation near 1.5 t , we speculate that the domains may not yet be fully aligned at this field , yielding a continuing non - linear field dependence in the hall resistivity . the sign reversal of the hall conductivity has received even more attention . in the literature
two explanations can be found : one@xcite assumes different temperature dependences for the electron and hole scattering rates , because of the different scattering mechanism ( phonon vs. magnon ) , which may then yield a strong temperature dependence of the hall resistivity , and a sign change .
it has been argued@xcite that this hypothesis should not work , since caruo@xmath0 is nonmagnetic , but still shows a sign - changing hall effect . instead the authors of ref . suggested that the sign may change because the number of electrons and holes in the energy window @xmath256 around the fermi energy may change with @xmath257 however , the sign reversal in caruo@xmath0 and srruo@xmath0 could be due to different physical reasons .
this possibility is also suggested by the very different field dependence of the hall resistivity in the two cases . on the other hand , it follows from our calculations , and is also indicated by various experiments , that caruo@xmath0 is on the verge of a magnetic instability , and the interplay between the phonons and paramagnons may play much the same role as the interplay between the phonons and magnons in srruo@xmath18 furthermore , besides the temperature dependence of the relaxation rates and the temperature broadening of the fermi level , there is yet another effect which may cause the sign change in srruo@xmath0 .
the exchange splitting must be very temperature dependent in srruo@xmath220 unlike common ferromagnets like fe , where the curie temperature corresponds to disordering of local moments , here the magnetization _ disappears at t_@xmath258 including the _ local _ magnetization .
thus , the spin splitting changes with the temperature , essentially disappearing around @xmath259 this is in contrast with most ferromagnets where an effective local spin splitting exists well above @xmath249 without any macroscopic magnetization .
thus , the band structure itself is strongly temperature dependent .
this effect can be operative in srruo@xmath0 in addition to the two other possible mechanisms .
the prerequisite for any of these mechanisms to be relevant is that there is strong compensation between the hole - like and the electron - like contributions from the different bands . to check this ,
we have calculated the hall conductivity @xmath260 and the hall coefficient @xmath261 for all the individual bands in srruo@xmath0 and caruo@xmath9 following the procedure described in ref .. the results are shown in fig .
[ hallsr3 ] .
it was observed by schultz _ et al_@xcite that quantitative calculations of the hall coefficient are extremely sensitive to sampling of the brillouin zone ; it is impractical for these 20 atom per unit cell structures to calculate the first principles band structure at a * k-*mesh comparable with the ultradense meshes used in ref . for elemental metals and instead we have relied on interpolation between first principles band energies calculated at 100 points in the irreducible wedge of the zone .
thus , our calculations shown in fig . [ hallsr3 ] can not be taken quantitatively , but rather illustrate the qualitative fact , that the hall conductivity has different signs in different bands and spin channels .
the net hall conductivity is defined by strong cancellation of hole- and electron - like contributions from different bands sc which in turn is very sensitive to relative position of different bands .
evidently , this balance can be easily violated by such temperature - dependent factors as lattice distortion , magnetization , and relaxation times . the mechanism suggested by klein et
al @xcite is also possible , since the net hall conductivity does change sign within a few hundred k around @xmath34 .
finally , very recent measurements@xcite of the hall coefficient in mixed sr@xmath40ca@xmath41ruo@xmath0 samples showed that for intermediate concentrations it does not change sign with temperature , suggesting that the sign reversals in pure compounds are accidental and unrelated .
at this time there is already a fairly substantial body of experimental literature on these ruthenates , including magnetic measurements , spectroscopic studies , specific heat data and determinations of electronic transport and superconducting properties .
these measurements demonstrate unusual and perhaps unexpected properties , and many of these have been ascribed to correlation effects . for example , the specific heats in the metallic compounds show substantial enhancements over the bare band structure values , superconductivity occurs in a layered material in apparent proximity to magnetic phases , quasiparticle bands measured by arpes show weaker dispersion than band structure calculations , satellites are observed in angle integrated photoemission spectra , and the transport properties of the metallic phases are unusual , showing e.g. sign reversals in the hall coefficient . since this evidence clearly suggests something unusual about the perovskite - derived ruthenates , it is tempting to ascribe it to strong correlation effects , particularly since these effects are all either qualitatively in the direction expected for a correlated system or can conceivably arise from the additional complexity introduced by correlation effects . on the other hand ,
chemical trends lead to the expectation that , all things being equal , 4d ru oxides should be less prone to strongly correlated behavior than the corresponding 3d oxides , and much less prone to such effects than cuprates .
this is because of the much more extended 4d orbitals in ru ions which should lead to stronger hybridization , better screening and lower effective hubbard @xmath16 .
furthermore , although much of the data are at first sight qualitatively in accord with general expectations for a correlated system , they have not been quantitatively explained in these terms , and there are some data that are rather difficult to understand purely in terms of a correlated scenario , most notably the disappearance of magnetism upon doping ca in the srruo@xmath0 system , and the ferromagnetic ground state in the integer occupancy ( and thus not a double - exchange system ) compound , srruo@xmath0 .
we have performed first principles , band structure based calculations within the lsda for srruo@xmath0 , caruo@xmath0 and sr@xmath1yruo@xmath262 although this approach fails miserably in systems that are truly strongly correlated , it does yield the correct magnetic and electronic states in these materials , including quantitative agreement with known magnetic properties in all cases in these ruthenates . moreover , the different magnetic behaviors can be fully understood in terms of simple and straightforward one - electron tight binding models and stoner theory .
although interpretation of the transport properties in terms of a conventional one - electron picture and bloch - boltzmann theory is not as straightforward , we show that such an approach is not inconsistent with the existing body of experimental evidence . a key notion for understanding
the transport in these systems is strong electron - phonon-(para)magnon coupling , which in turn can be understood in the framework of the band theory . in agreement with expectations based on chemical trends
, rather strong hybridization is found between the ru 4d and o 2p states in these materials . while antagonistic to a strong correlation scenario this is in large part responsible for the unusual properties in our band picture , including the very fact that magnetism occurs at all in a 4d metallic oxide .
this strong hybridization leads to a ferromagnetic direct exchange interaction between ru and o , and the cooperation between ru and o contributions to the stoner parameter leads to the magnetic ground states . as a result ,
the o ions in these ruthenates make substantial contributions to the magnetization density , which may be observable in neutron scattering experiments with o form factors included in the refinements .
the importance of @xmath263 hybridization also leads to a strong coupling of magnetic and structural degrees of freedom , resulting in for example the destabilization of the ferromagnetic state due to octahedral tilting in caruo@xmath0 .
one consequence of our scenario is that when ru ions are bonded to the same o , as neighboring ru ions are in the perovskite structure , the interaction between them will be strongly ferromagnetic .
this means that magnetic fluctuations in layered ruthenates like sr@xmath1ruo@xmath3 and the associated ruddlesden - popper ( rp ) series of compounds , are predicted to have predominantly ferromagnetic character in plane , although alternating layers , or perovskite blocks in the rp series may be coupled antiferromagnetically to each other via superexchange through rocksalt blocks .
such ferromagnetic fluctuations would be pair - breaking for singlet ( @xmath264 or @xmath19wave ) superconductivity , but not for triplet superconductivity , as suggested for instance by rice and sigrist@xcite for sr@xmath1ruo@xmath3 .
in fact , for triplet pairing both magnetic fluctuations and phonons provide cooper attraction .
finally , when ru ions are not connected at all via common o ions , like in sr@xmath1yruo@xmath2 the ru - ru coupling is via two intervening o ions , both of which are strongly hybridized with and ferromagnetically coupled to the nearest ru , but couple to each other via an antiferromagnetic superexchange interaction .
this results in an antiferromagnetic state .
the strength and importance of covalent transition metal - oxygen interactions , combined with magnetism and metallicity is perhaps unique to these ruthenates .
already a number of interesting physical properties have been found among these compounds , and no doubt more interesting physics remains to be found in this family .
we acknowledge enlightening discussions with j.s .
dodge , r.p .
guertin , lior klein , mark lee , and w.e .
work at the naval research laboratory is supported by the office of naval research .
computations were performed using the dod hpcmo computing centers at navo and asc
. introduced by fert and campbell @xcite and often used since then , the two - current conduction model of transport in ferromagnetic metals assumes that the spin - up and spin - down distribution function changes are independent .
this formula was derived for a very specific case of @xmath265 scattering in transition metals and under a number of simplifying assumptions . in the original papers @xcite
no clear distinction was made between the spin - flip scattering ( which always influences the total conductivity ) and the spin - flip conductivity ( which vanishes when scattering - in is neglected ) . in a later publication @xcite this distinction has been made , but this publication is less well known .
we find it instructive to present here a systematic derivation for the two - current model of electric transport in ferromagnets , and to show how it is related to the general theory of multiband conductivity .
this derivation is based on allen s variational approach to multiband boltzmann equation@xcite and clearly shows some limits to the applicability of the two - current model .
the boltzmann equation for electric transport in metals is @xmath266 where the subscripts @xmath267 include both the band index and the spin and @xmath268 is the transition probability matrix element , which can also be written conveniently in terms of the scattering matrix elements @xmath269 as @xmath270 for static ( impurity ) scattering , or the corresponding expression for the scattering by phonons , magnons etc .
the change of the distribution function for given * k * is @xmath271 so that @xmath272 and for @xmath234 @xmath273 allen introduced ( and called `` disjoint representation '' ) the following approximation : for each sheet of the fermi surface @xmath274 is proportional to * v*@xmath275 @xmath276 where the coefficients @xmath277 depend on the band ( and , in a magnet , on the spin ) . in this approximation ,
the last equation may be solved ( we assume * e*@xmath278 and omit subscript @xmath279 at @xmath280 @xmath281 or @xmath282 where the shorthand notation on the left - hand side is obvious and @xmath283 2 the first term here accounts for the scattering - out , and the second for the scattering - in processes .
solving now for @xmath277 s , we find @xmath284 since the total current density is @xmath285 and the conductivity is @xmath286 in notations of fert and campbell@xcite this is @xmath287 analogous to eq .
57 of ref .. campbell - fert formula is different when the nondiagonal elements are not neglected .
it gives instead @xmath288 for the purpose of this paper we shall neglect scattering - in completely , because it is defined by an average over the sign - changing quantity @xmath289 , so that the matrix @xmath290 is diagonal .
we shall also define the partial resistivities differently from ref .
, namely so that @xmath291 so defined , the @xmath292 is proportional to the spin - flip scattering rate
. then , for a simple ferromagnet with the two fermi surface sheets , one for spin - up and another for spin - down electrons , we have @xmath293 or @xmath294 this coincides with the fert - campbell formula , if the scattering - in term is neglected ; otherwise , it is different .
allen , h. berger , o. chauvet , l. forro , t. jarlborg , a. junod , b. revaz , and g. santi , phys . rev . *
b 5 * , 4393 ( 1996 ) .
t. kiyama , k. yoshimura , k. kosuge , y. ikeda , and y. bando , phys . rev . *
b 54 * , r756 ( 1996 ) .
note that we have not optimized the structure either for the paramagnetic or for the magnetic structures in our fixed - moment calculations .
due to strong magnetoelastic interactions the equilibrium tilting angle in a magnetic state should be slightly different from that without magnetization .
however , the qualitative statement that caruo@xmath0 is on the verge of magnetic instability should be true even if magnetoelastic effects are taken into account . | the magnetic properties of ruthenates with perovskite derived structures , particularly ( ca , sr)ruo@xmath0 and sr@xmath1yruo@xmath2 are studied within the context of band structure based stoner theory .
first principles calculations are used to demonstrate that in all cases the correct magnetic behavior and order can be obtained without recourse to strong correlation effects , and that the insulating character of sr@xmath1yruo@xmath2 is reproduced .
the different magnetic states of srruo@xmath0 and caruo@xmath0 are shown to be due to the different structural distortions in these materials , most significantly the larger rotation of the octahedra in the ca compound .
caruo@xmath0 is found to be on the verge of a ferromagnetic instability , leading to the expectation of giant local moments around magnetic impurities and other anomalous effects in analogy with fcc pd metal .
oxygen 2p derived states hybridize strongly with ru d states in all three compounds , and o , through this hybridization plays an unusually large role in the magnetic properties .
this involvement of o is responsible for the strong magneto - structural coupling that is found in the calculations .
transport properties of caruo@xmath0 and srruo@xmath0 are analyzed using the calculated fermiology .
unusually large magnon and paramagnon couplings are found , which are consistent with reported measurements of the low temperature specific heat and the resistivity coefficient . 2 |
in the last decade it has been reached a general consensus that ultra - relativistic heavy - ion collisions ( urhics ) at the relativistic heavy - ion collider ( rhic ) and the large hadron collider ( lhc ) create a hot and dense strongly interacting quark and gluon plasma ( qgp ) @xcite .
a main discovery has been that the qgp has a very small shear viscosity to density entropy , @xmath0 , which is more than one order of magnitude smaller than the one of water @xcite , and close to the lower bound of @xmath9 conjectured for systems at infinite strong coupling @xcite .
a key observable to reach such a conclusion is the elliptic flow @xcite , @xmath10 with @xmath11 being the azimuthal angle in the transverse plane and the average meant over the particle distribution . in fact , the expansion of the created matter generates a large anisotropy of the emitted particles that can be primarily measured by @xmath12 .
an hadronic system would generate a @xmath12 about a factor 2 - 3 smaller than the measured one @xcite and this is one of the hints that the system created at rhic and lhc is of non - hadronic nature .
the origin of @xmath12 is the initial spatial eccentricity , @xmath13 of the overlap region in non - central collisions , which is responsible for different pressure gradients in the transverse plane thus favoring flow preferably along the @xmath14 direction rather than @xmath15 direction .
the observed large @xmath12 is considered a signal of a very small @xmath0 because it means that the system is very efficient in converting @xmath16 into an anisotropy in the momentum space @xmath12 , a mechanism that would be strongly damped in a system highly viscous that dissipates and smooths anisotropies @xcite .
quantitatively both viscous hydrodynamics @xcite , and transport boltzmann - like approaches @xcite agree in indicating an average @xmath0 of the qgp lying in the range @xmath17 .
along with the existence of a deconfined qgp matter and the understanding of its properties , the urhic program offers the opportunity to verify the picture in which at vey high energy the two colliding nuclei are described as two sheets of color glass condensate ( cgc ) @xcite , see @xcite for reviews .
the cgc in the nuclei would be primarily generated by the very high density of the gluon distribution function at low @xmath14 ( parton momentum fraction ) , which triggers a saturation of the distribution for @xmath3 below a saturation scale , @xmath18 .
the determination of the shear viscosity @xmath0 of the qgp and the search for the cgc are related .
in fact , the main source of uncertainty for @xmath0 comes from the unknown initial conditions of the created matter @xcite that imply quite different eccentricities @xmath16 .
a simple geometrical description through the glauber model @xcite predicts a @xmath16 smaller at least 25 - 30@xmath19 than the eccentricity of the cgc , for most of the centralities of the collisions , see for example results within the kharzeev - levin - nardi ( kln ) model @xcite , factorized kln ( fkln ) model @xcite , monte carlo kln ( mc - kln ) model @xcite and dipole model @xcite .
in fact , the saturation of gluon distribution function is stronger in the central overlapping region , making the space distribution sharper than the one coming from a simple geometrical overlap .
other cgc models like the ones based on the solution of the classic yang - mills ( cym ) equations predict a somewhat smaller initial eccentricity @xcite .
however the effect we discuss in the present paper is mainly related to the saturation effect in momentum space and not in the @xmath20space that determines the eccentricity , even if certainly to consider cym initial conditions is an important task considering the success it has in predicting the collective flow anisotropies @xmath21 with @xmath22 @xcite . in this article , we report our results about the computation of the elliptic flow of the quark - gluon - plasma produced in relativistic heavy ion collisions , simulating the expansion of the fireball by solving the relativistic boltzmann equation for the parton distribution function tuned at a fixed shear viscosity to entropy density ratio @xmath0 @xcite . in this context
the advantage of using kinetic theory is that starting from a one - body phase space distribution function @xmath23 , and not from the energy - momentum tensor @xmath24 , it is straightforward to initialize simulations from a non - equilibrium distribution function , like the one characterizing the kln model , while the hydrodynamics relies on the gradient expansion of the stress tensor and therefore is applicable only if initial deviations from local equilibrium are small .
it is worth noticing that kinetic theory has also attracted interest as a tool to derive equations for the hydrodynamical evolution of a fluid starting from a microscopic theory @xcite . in this study
we assume that the initial condition in configuration and momentum space arising from the melting of the cgc is given by the kln model .
in fact the kln initial condition that has been largely employed to study the dynamics of hic and the viscosity of the qgp @xcite .
the uncertainty in the initial condition translates into an uncertainty on @xmath0 of at a least a factor of two as estimated by mean of several viscous hydrodynamical approaches @xcite .
more explicitly , the experimental data of @xmath25 at the highest rhic energy are in agreement with a fluid with @xmath26 according to viscous hydrodynamics simulation , assuming a standard glauber initial condition . assuming an initial fkln or mc - kln space distribution the comparison favors a fluid at @xmath27 .
the reason is the larger initial @xmath16 of the fkln , which leads to larger @xmath12 unless a large @xmath0 is considered .
we point out that the implementation of the shattered cgc initialization in hydrodynamics takes into account only the different space distribution respect to a geometric glauber model , discarding the key and more peculiar feature of the damping of the distribution for @xmath3 below the @xmath18 saturation scale .
we have found by mean of kinetic theory that this has a significant impact on the build - up of @xmath12 . in order to frame our work correctly , we refer to the standard picture of the evolution of the fireball produced in the heavy ion collision , so we can identify the moment in which our approach is valid .
the commonly accepted picture goes as follows : at @xmath28 the two high energy colliding nuclei are described as two thin sheets of color - glass condensate , which after the collision turns at @xmath29 into an out - of - equilibrium glasma state , which consists of a background of chromomagnetic and chromoelectric flux tubes , on the top of which quantum fluctuations are produced and cause the decay of the glasma itself to a partong liquid as the system expands .
this parton system then equilibrates and the evolution is governed by viscous hydro . during the very early evolution of the glasma , the initial negative longitudinal pressure becomes positive because of the breaking of the color strings , see for example @xcite .
the decay causes the fields become very weak within a fraction of fm / c : we denote this time by @xmath30 . from this time a description of the system as a pure interacting partons liquid becomes reliable , and can be described by kinetic theory .
we therefore start our simulations from @xmath30 .
we assume @xmath31 fm / c in this work ; however we have checked that our main results on the elliptic flow are not affected significantly by shifting the initial time to reasonably larger times . within our framework the longitudinal pressure @xmath32 at @xmath33 is positive , in agreement with recent calculations within classic - yang mills statistical approach @xcite or within kinetic theory at fixed @xmath0 with a gauge field in abelian dominance @xcite .
our main purpose in this study is to take into account strong deviations from local equilibrium in the initial condition , which naturally arise when the kln model is used in the calculations : these kind of deviations are not affordably treated within viscous hydrodynamics even if in principle non vanishing bulk and shear energy tensor at @xmath30 may account for it . anyway , we will show that our description reproduces the hydrodynamics results when a proper thermalized initial condition is prepared in our simulations .
the plan of the article is as follows . in section
ii , we review the initial conditions we use in the simulations , namely the factorized glasma the glauber initial conditions . in section iii , we review our approach to kinetic theory at fixed @xmath0 .
in section iv , we present our results on thermalization and isotropization of the fireball , as well as on the time evolution of the parton distribution function in the expanding system . in section
v , we present and discuss our results on the elliptic flow . finally in section
vi , we draw our conclusions .
in order to prepare the initial condition in coordinate and momentum space of a melted glasma state we adopt the model which was firstly introduced by kharzeev , levin and nardi @xcite ( kln model ) , with particular reference to the factorized - kln ( fkln ) approach as introduced in @xcite , in which the coordinate space distribution function of gluons arising from the melted glasma is assumed to be @xmath34 where @xmath35 corresponds to the momentum space distribution in the @xmath36 factorization hypothesis @xcite , @xmath37 and @xmath38 . here
@xmath39 correspond to the longitudinal momentum fraction carried by the gluons belonging to the two colliding nuclei which produce a gluon with transverse momentum @xmath3 and rapidity @xmath15 ; the ultraviolet cutoff in the @xmath36 integral in eq .
is @xmath40 gev@xmath41 for the case of au - au collisions at @xmath42 gev , while we take @xmath43 gev@xmath41 for the case of pb - pb collisions at @xmath44 tev ; @xmath45 denotes the strong coupling constant , which is computed at the scale @xmath46 according to the one - loop @xmath47 function but frozen at @xmath48 in the infrared region as in @xcite . in eq .
@xmath49 denote the probability to find one nucleon at a given transverse coordinate , namely @xmath50 where @xmath51 is the inelastic cross section and @xmath52 corresponds to the usual thickness function of the glauber model .
equation is based on the factorization hypothesis , which is known to work fairly in the case of proton - proton as well as proton - nucleus collisions but it is violated in the case of a nucleus - nucleus collision @xcite . however , we do not aim at considering the exact spectrum of a glasma , that may be computed for example within the cym approach @xcite ; rather we wish to study another issue , that is how the initial nonequilibrium distribution in momentum space , in particular one with a saturation scale built in , affects the building up of the elliptic flow .
furthermore , the kln initial condition is still largely employed in hydrodynamics simulations to test the impact of initial conditions on different observables @xcite . for simplicity , and also to uniform our language to that commonly adopted in hydro simulations , we still refer to the initial condition specified by eq . as to the melted - glasma initial condition .
the main ingredient to specify in eq . is the unintegrated gluon distribution function ( ugdf ) for partons coming from nucleus @xmath53 , which is assumed to be : @xmath54~ , \label{eq : phia}\end{aligned}\ ] ] which embeds saturation of the distribution for @xmath55 ; a similar equation holds for partons belonging to nucleus @xmath56 .
following @xcite we take the saturation scale for the nucleus @xmath53 as @xmath57 with @xmath58 , and similarly nucleus @xmath56 .
the scale @xmath59 regulates the average value of the saturation scale on the transverse plane .
equation implies that for a fixed value of @xmath14 , the larger the nucleon density the larger the saturation scale , since @xmath60 for large @xmath52 ; however in the limit @xmath61 , which is realized in the peripheral region of each of the two colliding nuclei , equation implies that @xmath18 does not vanish as @xmath52 .
in fact if @xmath62 then from eq .
we get @xmath63 , which implies that @xmath18 is nonzero in this limit , and coincides with the saturation scale of a nucleon @xcite . for our simulations of au - au collisions at rhic energy
we take @xmath64 gev@xmath65 , which gives @xmath66 gev in the case @xmath67 , at @xmath68 , @xmath69 gev and @xmath70 , where the average is understood in the transverse plane .
this numerical value is smaller than the one commonly used in hydrodynamic simulations , where @xmath71 gev , see for example @xcite , and that we used in our previous work @xcite .
we prefer to use a smaller value of the average saturation scale in our simulations because in this way we can shift from the fkln initial condition to the other ones obtaining the same final spectra , without the need to tweak the kinetic freezout parameters when changing among the several initial conditions . in this way ,
the differences between the various systems we study here are confined in the initial configuration , while the dynamics of the freezout is not affected by the initial conditions . however , to show the impact of the saturation scale we will discuss both the cases @xmath72 gev and @xmath73 gev in agreement with the estimates of @xcite , for the pb - pb collisions simulations at the lhc energy .
( color online ) saturation scale @xmath18 averaged over the transverse plane for the cases of a au - au collision at @xmath42 gev ( green dashed line ) and a pb - pb collision at @xmath44 tev ( solid indigo line ) .
all the calculations correspond to gluons produced at @xmath68 and @xmath74 gev , and @xmath75 gev.,width=321 ] in fig .
[ fig : qsav ] which we plot the averaged value of @xmath18 on the transverse plane , @xmath76 with @xmath18 given by eq . and the weight function by eq . , for the cases of a au - au collision at @xmath42 gev and a pb - pb collision at @xmath44 tev , for gluons at @xmath68 and @xmath74 gev , corresponding to @xmath70 in the case of the au - au collision and to @xmath77 in the case of the pb - pb collision .
in our study we employ transport theory as a base of a simulation code of the fireball expansion created in relativistic heavy - ion collision @xcite , therefore the gluon distribution function @xmath78 evolves according to the relativistic boltzmann transport ( rbt ) equation : @xmath79~ , \label{eq : be}\ ] ] where @xmath80 $ ] is the collision integral , @xmath81&=&\int d\gamma_2 d\gamma_{1^\prime } d\gamma_{2^\prime } ( f_{1^\prime}f_{2^\prime } - f_1 f_2)\nonumber\\ & & \times|{\cal m}|^2\delta^4(p_1 + p_2 - p_{1^\prime } - p_{2^\prime})~,\end{aligned}\ ] ] with @xmath82 , and @xmath83 corresponds to the transition amplitude . at variance with the standard use of transport theory , in which one fixes a set of microscopic processes
whose scattering matrix is fed into the collision integral , we have developed an approach that , instead of focusing on specific microscopic calculations or modelings for the scattering matrix , fixes the total cross section in order to have the wanted @xmath0 . by means of this procedure
we are able to use the boltzmann equation to simulate the dynamical evolution of a fluid with specified shear viscosity , in analogy to what is done within hydrodynamical simulations .
the advantage of the kinetic theory approach at fixed @xmath0 , compared to hydro simulations , is twofold : firstly starting from @xmath23 , and not from @xmath24 , it is direct to incorporate non - equilibrium initial conditions .
secondly we do not need to specify an ansatz for the deviations @xmath84 from equilibrium due to viscous corrections .
this approach to kinetic theory has been also considered in @xcite where it has been shown that transport theory at fixed @xmath0 reproduces the results of viscous hydrodynamics for one - body observable like @xmath24 or entropy density also in the limit in which the system is not in the dilute regime .
this is not surprising because looking at the boltzmann collision integral in terms of viscosity allows the analytical derivation of second order viscous hydrodynamics @xcite . in this article
we consider only the @xmath85 processes to compute the collision integral .
one may think that the introduction of the higher order processes would result naturally in a change of the available phase space of partons , but once the system is close to an hydro regime where many collisions happens in a very short time respect to the expansion time scale and system size , the detail of the single scattering is lost and what matters is the viscosity of the fluid .
of course one can expect that the microscopic details of the scattering arises at large @xmath3 where a hydro - like description in terms of a gradient expansion of the stress tensor breaks down ; in the regime of small and moderate @xmath3 , which is the one we are interested to in this study , the single collision phase space change is not relevant .
we have indeed checked that we are in a regime where the phase space of the single scatterings are not relevant by changing the microscopic two - body scattering matrix from anisotropic to isotropic , but renormalizing the total cross section in such a way to keep the same @xmath0 .
we have found that leaving unchanged all the other parameters , the elliptic flow is not affected by this change for @xmath86 gev , finding some deviations at larger @xmath3 @xcite .
a similar result is seen if @xmath87 collisions are switched on and off once again the cross sections are such to keep @xmath0 fixed @xcite .
once @xmath0 is fixed , we compute the total cross section in each cell of the coordinate space of our grid . to this end
we need an analytical relation between @xmath88 , temperature , cross section and density ; as shown in @xcite , the chapmann - enskog approximation supplies such a relation with quite good approximation , in agreement with the results obtained using the green kubo formula .
therefore , we fix @xmath0 and compute the pertinent total cross section by mean of the relation @xmath89 which is valid for a generic differential cross section @xmath90 as proved in @xcite . in the above equation @xmath91 , with @xmath92 the screening mass regulating the angular dependence of the cross section , while @xmath93\ !
h\left(\frac{a^2}{y^2}\right ) \label{g_ce}\end{aligned}\ ] ] with @xmath94 the bessel function and @xmath95 corresponding to the ratio of the transport and the total cross section @xmath96 and @xmath97 $ ] .
the @xmath98 is the proper function accounting for the pertinent relaxation time @xmath99 associated to the shear transport coefficient .
the maximum value of @xmath100 , namely @xmath101 , is reached for isotropic cross section and eq.([eq : sigma ] ) reduces to the relaxation time approximation with @xmath102 ; a smaller value of @xmath98 means that a higher @xmath103 is needed to reproduce the same value of @xmath0 .
however , we notice that in the regime were viscous hydrodynamic applies ( not too large @xmath0 and @xmath3 ) the specific microscopic detail of the cross section is irrelevant and our approach is an effective way to employ transport theory to simulate a fluid at a given @xmath0 .
therefore from one hand we are in hydrodynamical regime from the other that our approach can not be and is not meant as a tool to infer a microscopic description of the qgp .
( color online ) ratio of shear viscosity over entropy density ratio implemented in the present study .
green solid line corresponds to the case in which we take @xmath104 in the quark - gluon - plasma phase , which we then connect smoothly to the value in the hadron phase by means of a kinetic freezout . for completeness
we also show the orange line corresponding to the case @xmath105 in the quark - gluon - plasma phase , which is once again connected smoothly to the value in the hadron phase.,width=321 ] the shear viscosity over entropy density ratio which we use in our study is plot in fig . [
fig : etasus ] . in the plasma phase @xmath0
is taken to be a constant , whose numerical value is fixed case by case : in the figure the cases @xmath104 and @xmath105 are represented , but we also perform some simulations with larger values of the ratio . we implement a kinetic freezout by assuming that @xmath0 increases smoothly in a temperature range from the plasma phase to a hadron phase values which is fixed by referring to the estimates in @xcite . in this way we take into account scatterings in the hadron phase as well , which however give a very tiny contribution to the collective flow because of the damping due to the larger viscosity and the reduce eccentricity at this later stage . in the following
, we will consider three different types of initial distribution function in the phase - space , which have been considered also in our previous study @xcite : two of them are commonly employed in hydro simulations , while the third one represents the novelty of the present study and relies on the ability of the transport approach to include the saturation scale in the initial distribution function , as shown in fig .
[ fig : spettri_cgc ] . for simulations at the rhic energy we refer to au - au collision at @xmath106gev
; we present here results for @xmath107 fm , @xmath108 fm and @xmath109 fm . for simulations at the lhc energy we will refer to pb - pb collisions at @xmath110tev , focusing on the case @xmath108 fm .
( color online ) initial @xmath111spectra at midrapidity for the case of a au - au collision at @xmath112 gev , with an impact parameter @xmath108 fm.,width=321 ] the standard initial condition for simulations of the plasma fireball created at rhic is based on the glauber model , with an @xmath113-space distribution given by the a standard mixture @xmath114 and a @xmath115-space thermalized spectrum in the transverse plane at a time @xmath116 with a maximum initial temperature @xmath117 . following the nomenclature introduced in
@xcite we will refer to this case as th - glauber corresponding to full circles in fig.[fig : spettri_cgc ] . within previous studies based on hydro simulations
the impact of an initial factorized glasma state has been performed considering an @xmath113-space distribution given by the fkln ( or mc - kln ) , while in the momentum space the spectrum has been considered thermalized at @xmath118 fm / c @xcite ; we refer to this case as th - fkln ; among other things this initial condition , when implemented in hydro simulations , leads to the conclusion that the liquid created by the melting of the glasma is characterized by @xmath119 @xcite at rhic , at least for non - central collisions .
the third initial conditions is the full fkln initial conditions where , beyond the @xmath113- space , the saturated distribution in @xmath115-space proper of the model is implemented as well , see fig .
[ fig : spettri_cgc ] solid thick line with diamonds .
we have verified that our main results on thermalization times and elliptic flows are not affected by the choice of @xmath30 . in this case as initial time we take @xmath120 fm / c with no assumption about thermalization in the transverse plane , as it should the case of any implementation of the kln model since its proper momentum distribution is out of equilibrium . this initial condition is not usually considered in hydrodynamics because there it is implicitly assumed the system is locally thermalized in the transverse plane , and in fact the initial transverse energy density profile is connected to the initial @xmath113-space fkln distribution assuming equilibrium thermodynamic relations . in the context of viscous hydrodynamics it is possible to include initial non - equilibrium conditions by introducing an initial non - vanishing value for the viscous tensor @xmath121 .
this has been mostly studied for the shear stress tensor and it has been seen to have a quite small impact on the @xmath12 at least on the bulk of the system @xcite
. however the non - equilibrium implied by fkln implies a change in the trace of the energy - momentum tensor that could be related in hydrodynamical language to a finite bulk stress tensor at @xmath30 . to our knowledge
it has never been investigated the relation between an initial non vanishing bulk @xmath121 and the initial non - equilibrium implied by the saturation scale in the distribution function @xmath122 .
similarly we introduce th - glauber , th - fkln and fkln initial conditions for the lhc runs . in the case of th - glauber and th - fkln the maximum temperature is taken to be @xmath123 mev and the system is assumed to be thermalized in the transverse plane at @xmath124 fm / c ; for the case of the fkln initial condition , we keep @xmath120 fm / c .
this time should correspond to the time interval in which the color strings decay forming the strongly interacting parton liquid .
( color online ) multiplicity as a function of @xmath125 at rhic and lhc energies as indicated in the figure .
experimental data are taken from refs .
@xcite.,width=321 ] the particle multiplicity is fixed to correctly reproduce the experimental one for all the three cases .
this is shown in fig .
[ fig : multe ] where we plot the particle multiplicity , measured at central rapidity in units of the number of pairs , as a function of the number of participants , for the cases of the fkln and th - glauber initializations .
experimental data are taken from refs .
@xcite . in the case of the th - glauber and th - fkln
we have assumed boost invariance in the longitudinal direction for the initial configuration ; on the other hand , for the case of the fkln initial condition a small @xmath126dependence comes from the distribution in eq . . nevertheless in both the calculations
the total number of particles is the same , as it is shown in fig .
[ fig : multe ] .
( color online ) _ upper panel : _ time evolution of spectra . _ middle panel : _ comparison of the final spectra obtained by simulations with experimental data @xcite . _ lower panel : _ time evolution of temperature @xmath127 . in the inset
we plot @xmath128 for the early stage of the collision . in all panels @xmath108
the th - fkln case is not shown because for both spectra and temperature evolution we do not find visible deviations from the th - glauber initialization.,title="fig:",width=321 ] + ( color online ) _ upper panel : _ time evolution of spectra .
middle panel : _ comparison of the final spectra obtained by simulations with experimental data @xcite .
_ lower panel : _ time evolution of temperature @xmath127 . in the inset we plot @xmath128 for the early stage of the collision . in all panels @xmath108 fm .
the th - fkln case is not shown because for both spectra and temperature evolution we do not find visible deviations from the th - glauber initialization.,title="fig:",width=321 ] + ( color online ) _ upper panel : _ time evolution of spectra .
_ middle panel : _ comparison of the final spectra obtained by simulations with experimental data @xcite .
_ lower panel : _ time evolution of temperature @xmath127 . in the inset we plot @xmath128 for the early stage of the collision . in all panels @xmath108 fm .
the th - fkln case is not shown because for both spectra and temperature evolution
we do not find visible deviations from the th - glauber initialization.,title="fig:",width=317 ]
in fig . [ fig : spettri_cgc2 ] we collect the initial spectra , @xmath129 , integrated over the momentum rapidity window @xmath130 , for the case of a au - au collision @xmath42 gev for the fkln and th - glauber initial conditions at their respective initial times @xmath30 and at the final time @xmath131 fm / c . for the case of the th - fkln , we find that the spectra are the same as in the case of the th - glauber , therefore we do not plot them in the figure .
we have also shown by the dashed line the spectrum at @xmath132 fm / c for the case of the fkln initial condition .
our main aim is to focus on the impact of the initial conditions implied by the fkln on the elliptic flow ; however we discuss before the time evolution of spectra and the related isotropization .
the results in fig .
[ fig : spettri_cgc2 ] are obtained with @xmath104 .
we notice that initially the fkln spectrum is quite far from a thermalized spectrum ; in fact , it embeds the saturation effects which should be proper of the melted glasma .
nevertheless the spectrum thermalizes in the transverse plane within @xmath133 fm / c , since its @xmath3 dependence becomes exponential with a slope very similar to the th - glauber . in the middle panel
we compare our results with the experimental data of the star collaboration for the charged hadrons @xcite . in order to make this comparison
we have assumed the quark - hadron duality which implies that for each parton a single hadron is produced ; moreover we have multiplied our results by the factor @xmath134 which takes into account the facts that experimental data correspond to charged hadrons only , and that the rapidity range of experimental data is different from ours .
we want only to show that the final spectra between th - glauber and fkln have a very simlar shape which make more meaningful the comparison of the elliptic flow discussed in the next section .
the data are mainly shown to have a reference . in the lower panel of fig .
[ fig : spettri_cgc2 ] ( c ) , we have plot the ratio @xmath135 , evaluated in the local rest frame of the fluid and representing the temperature in the case of a thermalized system , as a function of time .
the change of slopes of the curves suggest that initially the cooling of the fireball is dominated by the longitudinal expansion ; as soon as the transverse expansion begins ( @xmath136 ) , the cooling becomes faster and indeed the fireball reaches the freezout energy density within few fm / c . in the inset of fig .
[ fig : spettri_cgc2 ] , we plot the quantity @xmath137 . in the case of 1d expansion without dissipation , implies that a thermalized system @xmath138 .
we find that in the case of the fkln ( solid green line ) the product @xmath137 is strongly dependent on time because the system initially is quite far from equilibrium , and the initial non - equilibrium generates some entropy to thermalize ; however at @xmath139 we find @xmath137 saturates to a constant value , signaling the system has thermalized in the whole momentum space .
we notice a tiny evolution also for these cases that we have indicated as thermal .
the reason is that the initial spectra are thermal only in the transverse plane , while they are boost - invariant along the longitudinal direction .
furthermore as one can expect my solving the boltzmann under boost invariance at mid - rapidity @xmath140 should decrease as @xmath141 with @xmath142 that is smaller than @xmath143 @xcite .
this causes a tiny evolution to the constant value , that would disappear when the distribution is thermal also in the longitudinal direction . at larger times
@xmath140 decreases faster because the 3d expansion sets in and at @xmath144 one has @xmath145 as shown in the main panel of fig .
[ fig : spettri_cgc2 ] with the thin solid line indicating the @xmath146 behavior .
we remark that even if we have chosen to start our simulation at @xmath147 fm / c , our final result on thermalization time required by the kln initial condition is unaffected if we shift the initial time to a reasonably larger value .
a posteriori our results on short thermalization times are quite natural : in fact , we have assumed the fluid expands with a very small shear viscosity , which naturally implies that the coupling among the partons are nonperturbative , hence leading efficiently the distribution to the fixed point of the boltzmann equation , namely a thermalized distribution .
these results on fast thermalization are in agreement with earlier studies showing that two - body collisions are insufficient to achieve a fast thermal equilibrium @xcite , and three - body processes are necessary .
in fact in that case the perturbative qcd two - body cross section is used , which corresponds to @xmath0 about one order of magnitude larger than in our case ; hence , to achieve fast thermalization three - body processes have to be fed into the collision integral .
the difference with @xcite is that we do not focus on the microscopic details which lead to thermalization , as discussed in section iii : we are interested to simulate the dynamics of a fluid with few macroscopic properties specified , namely the equation of state and the ratio @xmath0 .
thus we normalize the cross section locally to get the wanted @xmath0 , and our scattering rates even if obtained effectively by two - body scattering are very large , leading eventually to a fast thermalization . _
( color online ) _ time evolution of spectra at midrapidity for pb - pb collision at @xmath148 tev , with an impact parameter @xmath108 fm .
, title="fig:",width=313 ] + in fig .
[ fig : spettri_cgc3 ] , we plot our results for the spectra for a pb - pb collision at @xmath44 tev , corresponding to a typical lhc collision . as expected , the qualitative behaviour of spectra is unchanged shifting from the rhic to the lhc energy . in this case however the effect of the saturation scale is smaller , because the initial temperature for the th - glauber initial condition is larger , as it can be inferred from the spectra .
hence the @xmath111spectra for the thermalized case is indeed quite close to the fkln one , as shown in fig.[fig : spettri_cgc3 ] ( upper panel ) .
also in the case of the pb - pb collision , starting from @xmath120 fm / c for the fkln melted glasma initial condition , time evolution of spectra shows that the transverse thermalization occurs within @xmath149 fm / c . at lhc
the time scale at which @xmath150 is constant appears to be quite similar to the one at rhic in fig.[fig : spettri_cgc2 ] ( c ) .
, for several initial conditions .
upper panel results refer to au - au collisions at @xmath151 gev . , title="fig:",width=313 ] +
we have studied also the isotropization of pressures for the expanding system . the energy - momentum tensor is defined locally as @xmath152 where @xmath7 corresponds to the invariant distribution function . in our simulations
the energy - momentum tensor is defined in each cell and it is computed in the local rest frame of the fluid , subtracting the radial flow contribution ; we then define locally transverse and longitudinal pressures , @xmath153 and @xmath32 respectively , as @xmath154 finally we average @xmath155 over several cells in order to lower the effect of statistical fluctuations . in fig .
[ fig : ptot2 ] we plot our results about @xmath155 as a function of time , for several initial conditions and for the case of strong coupling @xmath104 .
in particular in the figure we have averaged over a square in the central region of the transverse plane whose side is of @xmath156 fm , while in the longitudinal direction we have averaged over @xmath157 . at early times
the longitudinal pressure is zero due to the bjorken initial conditions .
our findings suggest that independently on the initial condition implemented , the system becomes nearly isotropic ( @xmath158 ) within a @xmath159 for @xmath104 .
we have selected the region @xmath157 checking that the particle in this space region are nearly correspondent to all the particles with momentum rapidity @xmath160 to which we are mainly interested . we have also checked that the maximum value of @xmath155 marginally depends on the space region selected while its maximum values increase up to @xmath161 if one selects the central region ( see also fig.[fig : ptot3 ] ) , while it goes down to @xmath162 if one considers the entire fireball in both transverse and longitudinal direction .
a similar behavior is observed at lhc as can be seen for the th - glauber case also in the left panel of fig.[fig : ptot3 ] .
, for the case of the th - glauber initial condition , for several values of the ratio @xmath0 . left panel corresponds to pb - pb collision at @xmath44 tev ; right panel corresponds to au - au collision at @xmath42 gev .
selection in transverse plane is @xmath163 and space - time rapidity @xmath157 .
all the calculations refer to collisions with an impact parameter @xmath108 fm.,width=313 ] it is interesting to compute how the isotropization is affected by the @xmath0 of the fluid . to this end in fig .
[ fig : ptot3 ] we show our results for pressures at the rhic ( right panel ) and lhc ( left panel ) energies , for the glauber initial condition ( for the fkln initial condition we obtain similar results ) , for different values of the ratio @xmath164 . in particular , the value @xmath165 corresponds approximately to the perturbative qcd estimate @xcite .
as expected , the larger the @xmath0 , the smaller the ability of the system to remove the initial pressure anisotropy and a perturbative fluid would not be able to sufficiently isotropize the pressure .
the ratio @xmath155 reaches its maximum value slowly as @xmath0 increases : turning from @xmath104 to @xmath166 the capability of the system to reach isotropization is lost while it is barely reached for @xmath167 but @xmath168 . at lhc the trend is similar but the larger density to reach slightly larger values of @xmath155 .
similar results on incomplete isotropization have been found in @xcite , where schwinger mechanism causes the color strings decaying into a parton liquid and shear viscosity is taken into account via the relaxation time ansatz .
respect to @xcite we consider a full 3d+1 simulation and we do not employ the relaxation time approximation , solving the kinetic equation with the full boltzmann integral . on the other hand , we miss the field dynamics that causes oscillations of @xmath155 which in the early phase generates a peak in the @xmath155 at @xmath169 and in 3d+1 could also lead to instabilities in three dimensions in the very early stage of the expansion . the results in fig .
[ fig : ptot2 ] and [ fig : ptot3 ] are comparable with previous calculations @xcite in which a dynamics for the color glasma fields is introduced ; this comparison is quite meaningful because it is well known that in the very initial stage of its dynamics , the glasma is characterized by @xmath170 due to strong fields in the longitudinal direction .
this negative pressure can not be reproduced within our simulation code at the moment because we have not yet implemented a dynamics for the classical chromoelectric and chromomagnetic fields .
nevertheless as shown in @xcite the ratio @xmath155 becomes positive within @xmath171 fm / c depending on the coupling strength , and unless the system is in weak coupling the strength of the fields becomes negligible for @xmath172 @xcite , which justifies our assumption that starting the simulations at @xmath173 fm / c the evolving fireball can be described by a parton liquid with positive pressure .
the main goal of our study is the computation of the differential elliptic flow for different initial conditions , both at rhic and lhc energies .
the main motivation has been discussed in the first part of section iii .
, as a function of time for three different initial conditions .
upper panel refers to au - au collisions at @xmath151 gev ; lower panel to pb - pb collisions at @xmath148 tev . in all calculations
the impact parameter @xmath108 fm , @xmath104 .
components of the energy - momentum tensor have been integrated over @xmath157 .
for the fkln initial condition we have chosen @xmath120 fm / c , while for the thermal distributions @xmath174 fm / c.,title="fig:",width=313 ] + , as a function of time for three different initial conditions .
upper panel refers to au - au collisions at @xmath151 gev ; lower panel to pb - pb collisions at @xmath148 tev . in all calculations
the impact parameter @xmath108 fm , @xmath104 .
components of the energy - momentum tensor have been integrated over @xmath157 .
for the fkln initial condition we have chosen @xmath120 fm / c , while for the thermal distributions @xmath174 fm / c.,title="fig:",width=313 ] gev , and of a pb - pb collisions at @xmath175 tev . in all calculations the impact parameter @xmath108 fm and @xmath104 .
for the fkln initial condition we have chosen @xmath120 fm / c , while for the thermal distributions @xmath174 fm / c.,title="fig:",width=313 ] + gev , and of a pb - pb collisions at @xmath175 tev . in all calculations the impact parameter @xmath108 fm and @xmath104 . for the fkln initial condition we have chosen @xmath120 fm / c , while for the thermal distributions @xmath174 fm / c.,title="fig:",width=313 ] before discussing the elliptic flow we show the results obtained for the momentum eccentricity , @xmath176 which is usually computed also in hydrodynamics @xcite and whose evolution is not related to the freezout by the cooper - frye hypersurface as insted happens for @xmath12 .
the average in the above equation is understood in the transverse plane at midrapidity .
we compute this quantity to show that especially for the bulk of the system the kinetic boltzmann approach determines a hydro - like evolution . in fig .
[ fig : ep ] , we plot the momentum eccentricity as a function of time , for the three different initial conditions we study in this article , for both rhic and lhc energies . for completeness in fig .
[ fig : ecce ] we plot the space eccentricity @xmath177 as a function of time for the cases of collisions at rhic ( upper panel ) and lhc ( lower panel ) energies .
results in fig .
[ fig : ep ] show that momentum anisotropy is built - up within the first few fm / c , which means that the elliptic flow is being developed in the same time range . however , in this time range @xmath177 in the case of the fkln initial condition is comparable to that of the th - fkln .
eventually the system enters the freezout and almost free streaming regime , and the @xmath177 for the two cases evolve in a different way .
this suggests that when the major part of the momentum anisotropy is formed , the fireball expands in the transverse plane almost in the same way both in the th - fkln and in the fkln cases ; therefore the difference in elliptic flow in the two cases has to be generated mainly by the different momentum distribution in the initial stage .
our results on @xmath178 are in fair agreement with those obtained within hydro simulations @xcite .
a strict quantitative comparison with the aforementioned results is not feasible because the results in @xcite are obtained using simulations in 2 + 1 dimensions with some differences in the initial temperature and density profile and a different temperature dependence of the hadron gas shear viscosity and also a different freezout temperature . even with these differences
we can compare our results with those of the eos i case ( perfect gas eos ) in @xcite , see their fig .
the time evolution of @xmath178 has some difference in the two cases , in particular in our case it grows faster with time , which is likely due to the differences mentioned above , nevertheless the asymptotic value are in good agreement with the th - glauber case . at midrapidity @xmath130 for different initial conditions and @xmath0 as in the legend .
all the calculations refer to au - au collisions at @xmath112 gev . from the upper to the lower panel the impact parameter @xmath179 fm , @xmath108 fm and @xmath109 fm respectively.,title="fig:",width=313 ] + at midrapidity @xmath130 for different initial conditions and @xmath0 as in the legend .
all the calculations refer to au - au collisions at @xmath112 gev . from the upper to the lower panel the impact parameter @xmath179 fm , @xmath108 fm and @xmath109 fm respectively.,title="fig:",width=313 ] + at midrapidity @xmath130 for different initial conditions and @xmath0 as in the legend .
all the calculations refer to au - au collisions at @xmath112 gev . from the upper to the lower panel the impact parameter @xmath179 fm , @xmath108 fm and @xmath109 fm respectively.,title="fig:",width=313 ] in fig .
[ fig : v2pt ] , we collect our results for the differential elliptic flow for the case of a au - au collision at @xmath42 gev , with three different impact parameters : from the upper to the lower panel the impact parameter is @xmath179 fm , @xmath108 fm and @xmath109 fm respectively .
the case @xmath108 fm corresponding to the @xmath180 centrality class at rhic has been already considered in @xcite , but with a different value of @xmath181 : here @xmath182 gev in the transverse plane at @xmath70 , while @xmath183 gev in @xcite .
we have used a different value of @xmath181 in the present calculations mainly because in this way we do not need to manipulate other parameters , among them the initial temperature as well as the t dependence of @xmath0 in the cross - over region that regulates the kinetic freezeout temperature , in order to obtain final spectra which are in agreement among the different initial conditions ; moreover we want to check that within a reasonable uncertainty on @xmath184 the effect we measure on the elliptic flow persists .
spectra corresponding to these data are plot in fig .
[ fig : spettri_cgc2 ] for the case @xmath108 fm ( for the other values of @xmath185 we obtain similar results ) . in fig .
[ fig : v2pt ] we have split the results for each value of @xmath185 into two panels : the left one corresponds to our hydro - like calculations , in which the initial distribution is assumed to be thermalized in the transverse plane . on the right panel
we plot the results obtained assuming the nonequilibrium distribution in the initial condition . to guide the eye , in the figure we also plot experimental data for @xmath12 in the relevant centrality class from the star collaboration @xcite .
the @xmath25 from the kinetic simulations are quite close to the data for the th - glauber and fkln initializations with @xmath104 .
however we remind that no hadronization process is yet included in our approach , so the agreement with the charged hadrons has to be taken with caution even if it indicates that the azimuthal asymmetries generated are in the right ball park .
the glauber initial condition is in agreement with experimental data for @xmath104 ; instead , in the case of th - fkln the elliptic flow for the same value of @xmath0 overshoots data except for very central collisions : in order to reproduce better experimental data one , @xmath0 has to be increased by about a factor of two .
these results are in agreement with the ones obtained from viscous hydrodynamics @xcite , showing the solidity and consistency of our transport approach at fixed @xmath0 .
the necessity of a larger @xmath0 to reproduce data on elliptic flow is usually understood in terms of the larger initial eccentricity of the th - fkln fireball compared to the glauber one , which would result in fact in a larger anisotropy in momentum space unless viscosity is large enough to damp the flow @xcite . in the right panel of fig .
[ fig : v2pt ] , we present the result for the fkln model , when the proper distribution function is implemented in both the @xmath113 and @xmath115 spaces .
we find that fkln with a @xmath104 gives a @xmath25 quite similar to the th - glauber , even if the initial eccentricity in this case is the same as the one of the th - fkln case .
for fkln with @xmath105 the differential elliptic flow would be too small .
in other words the initial out - of - equilibrium fkln distribution reduces the efficiency in converting @xmath16 into @xmath12 .
our interpretation is that the initial large eccentricity of the fkln configuration is compensated by the key feature of an almost saturated initial distribution in @xmath115-space below the saturation scale @xmath18 and probably by the softer tail at @xmath186 .
we notice that changing the centrality class we have found that the effect of the initial distribution on the elliptic flow becomes smaller as we turn from peripheral to central collisions , as it can be seen in fig .
[ fig : v2pt ] .
we obtain similar results at the lhc energy , fig .
[ fig : v2ptl ] , where we collect the results of our computation of @xmath25 for @xmath108 fm and for @xmath75 gev ( upper panel ) and @xmath187 gev ( lower panel ) , in order that @xmath184 is in agreement with the estimates of @xcite .
we can see that the th - fkln overestimate the @xmath25 at both @xmath104 and @xmath156 and probably even a @xmath188 is needed to account for the data .
when the full fkln is implemented we can see that the damping effect already discussed at rhic is such that with an @xmath105 one predicted a @xmath25 quite close to experiments .
this result we measure on the elliptic flow can be understood naively connecting @xmath12 to the slope of the spectrum , which corresponds to the inverse temperature if the spectrum is a thermal one .
in fact , the elliptic flow can be understood as a larger slope of the momentum spectrum in the out of plane @xmath189 direction respect to the @xmath190 one caused by a larger pressure in the @xmath189 direction due the elliptical shape .
the net effect in terms of the difference of the particle yields between the two directions is larger if the spectra are decreasing exponentially respect to the case in which they are nearly flat as a function of @xmath3 . for the different initial conditions and @xmath0 as in the legend , for the case @xmath75 gev where @xmath181 is defined in eq .
. _ lower panel .
_ same quantity for @xmath187 gev .
all the calculations refer to pb - pb collisions at @xmath175 tev , with an impact parameter @xmath108 fm.,title="fig:",width=313 ] + for the different initial conditions and @xmath0 as in the legend , for the case @xmath75 gev where @xmath181 is defined in eq . .
_ lower panel .
_ same quantity for @xmath187 gev .
all the calculations refer to pb - pb collisions at @xmath175 tev , with an impact parameter @xmath108 fm.,title="fig:",width=313 ] at midrapidity @xmath130 and at @xmath74 gev for different initial conditions , computed at @xmath104 .
all the calculations refer to au - au collisions at @xmath112 gev .
, width=313 ] in fig .
[ fig : v2riass ] we plot the differential elliptic flow at @xmath74 gev for different initializations , as a function of the number of participants .
we have shown results for @xmath191 fm corresponding to @xmath192 , @xmath179 fm corresponding to @xmath193 , @xmath108 fm corresponding to @xmath194 and @xmath109 fm corresponding to @xmath195 .
this comparison is meaningful since it permits to visualize and summarize the dependence of @xmath25 on the centrality class , comparing the impact of the initial distribution on the final @xmath12 .
in particular the discrepancy between th - fkln and fkln initializations becomes less relevant for more central collisions , implying that the effect of the initial momentum distribution is not negligible if one considers non - central collisions .
we also note that for central collisions at rhic th - glauber and th - fkln for @xmath104 predict the same @xmath12 and the effect is of kln generating larger @xmath12 disappears .
this is seen also in viscous hydro simulation and it is a further confirmation that our approach converge to viscous hydro when the same thermal initial conditions are employed . at @xmath196 as a function of the evolution time for all the different initial conditions and @xmath0 values .
upper panel refers to au - au collisions at @xmath112 gev , with an impact parameter @xmath108 fm .
lowere panel corresponds to pb - pb collisions at @xmath175 tev , with an impact parameter @xmath108 fm.,title="fig:",width=8 ] + at @xmath196 as a function of the evolution time for all the different initial conditions and @xmath0 values .
upper panel refers to au - au collisions at @xmath112 gev , with an impact parameter @xmath108 fm .
lowere panel corresponds to pb - pb collisions at @xmath175 tev , with an impact parameter @xmath108 fm.,title="fig:",width=8 ] in fig .
[ fig : v2_time ] , we plot the elliptic flow as a function of time for particles at @xmath197 gev . both the cases of the rhic and lhc runs are shown in the figure , for noncentral collisions with @xmath108 fm . in the figure we plot quantities as a function of @xmath198 rather than @xmath199 , where @xmath30 is the time at which dynamical evolution begins , because anisotropy is formed thanks to particle interactions which are effective at @xmath200 . in the case of
rhic the behavior clearly splits into two classes : one is represented by the thermal distributions ; the other one is given by the case in which the initial spectrum is the fkln one . in the case of lhc
this distinction is quantitatively less relevant but still present , probably because as already said the difference between the thermal distribution and the kln one is less pronounced .
the main point which we emphasize is that @xmath201 for the thermal distributions is large from the very beginning of the dynamical evolution , while it is quite reduced for fkln .
this is a further clue that the initial distribution is very important for the build - up of @xmath12 , a statement that we have already supported discussing the evolution of eccentricities and momentum anisotropy .
we further note that for @xmath202 , which is the time range roughly corresponding to the thermalization time for all the cases considered here , @xmath201 of fkln becomes very similar to th - fkln and th - glauber .
this observation further confirms that it is the initial out - of - equilibrium and nearly saturated distribution that dampens the efficiency in converting the space eccentricity into the @xmath12 . therefore , even if thermalization sets in quickly , as assumed in hydrodynamics @xmath203 , the actual evolution from the initial state to the thermalized one can not be neglected in studying the elliptic flow , especially at rhic energy were most of the anisotropy in momentum space develops in the first 3 fm / c at least for semi peripheral collisions .
gev , with an impact parameter @xmath108 fm and @xmath104.,title="fig:",width=313 ] + gev , with an impact parameter @xmath108 fm and @xmath104.,title="fig:",width=313 ] tev , with an impact parameter @xmath108 fm and @xmath104.,title="fig:",width=313 ] + tev , with an impact parameter @xmath108 fm and @xmath104.,title="fig:",width=313 ]
before going to the conclusions we want to discuss the range of values covered by the phase space distribution functions .
this is an issue that recently has attracted a particular interest because it could be possible that with the high density reached in the initial stage of the collision the bosonic quantum nature of gluons triggers a bose - einstein condensation @xcite .
the invariant distribution function , @xmath7 , is defined as @xmath204 where @xmath205 counts the number of particles in the phase space volume given by @xmath206 and @xmath207 corresponds to the number of gluonic degrees of freedom . on our grid
we have @xmath208 and @xmath209 ; moreover , we take @xmath210 and we sample the momentum space with @xmath211 . we integrate over momentum rapidity : this is necessary since at initial time @xmath33 we assume @xmath212 , which implies the initial distribution to be singular ,
namely @xmath213 the subscript @xmath214 reminds that the above equation is valid only at @xmath33 . because of the delta - function the measurable quantity at @xmath33 is @xmath215 , which can be obtained by integrating @xmath7 over momentum rapidity ; when we compute the distribution function at later times , for consistency we also integrate over momentum rapidity . for later convenience
we also integrate over the azimuthal angle @xmath216 .
we then define @xmath217 which on each cell of the configuration space reads @xmath218 with @xmath219 and the sum is understood over all the particles that at time @xmath199 are in the specified cell . here
we present the evolution of @xmath220 in the central region of a qgp fireball to have a first estimate of the value explored with time and transverse momentum @xmath3 . in figs .
[ fig : fk ] and [ fig : fkl ] we plot the invariant distribution functions for two different initial conditions .
the upper panels of the figures correspond to the fkln , while the lower panels correspond to the glauber initial condition ; fig .
[ fig : fk ] is the result for the au - au collisions at the rhic energy , while fig .
[ fig : fkl ] corresponds to the case of pb - pb collisions at the lhc energy . in order to obtain the results shown in the figure , we have integrated eq . over the region @xmath221 fm , @xmath222 fm and @xmath223 , which embed the hottest and densest parts of the initial fireball .
however , the results are qualitatively similar if we chose to compute @xmath224 by means of eq . only in the central cell .
the natural consequence of the longitudinal ( as well as the transverse ) expansion is that the distribution function becomes smaller and smaller as time increases .
in fact , from eq .
we expect @xmath225 due to the longitudinal expansion close to luminal velocity . in fig.[fig :
fkl ] ( upper panel ) we can observe that at @xmath226 the initial @xmath227 but the rapid longitudinal expansion along with the rapid cooling down is such the @xmath7 decreases very quickly with and at @xmath228 it is already smaller than @xmath229 . on the hand at @xmath230 not only at @xmath30 the @xmath231 but due to the high scattering rates that drive a rapid thermalization the decrease of @xmath232 with @xmath146 is fully damped at even at @xmath233 we have @xmath234 .
a similar trend with even larger value of @xmath235 is observed at lhc and not only for the fkln but also for the th - glauber initial conditions .
we remind that in our formulation we have not introduced the bosonic enhancement factors in the collision integral , which however are expected to give a relevant effect if @xmath236 .
still our result shows that the longitudinal expansion coupled to the thermalization process keeps the occupation number quite large for a relatively long time interval and therefore it is reasonable to expect that the formation of a transient bose - einstein condensate also in the expand glasma in hic and not only for the static case of a system in a box @xcite .
the results of @xcite are of course quite interesting and we plan to improve our calculation in the near future to take into account the bose enhancement factors in the collision integral , as well as the inelastic processes which convert gluons to quarks . as for the effect discussed on the build - up of @xmath12 in section iv
, we do not expect significant impact of the bose - einstein enhancement factors @xmath237 in the collisions integral because , as discussed , only for small @xmath3 , @xmath7 is relatively large and furthermore only in the very central region of the fireball . while our effect take place for @xmath238 .
moreover , the effect we find is mainly due to the initial non - equilibrium dynamics and not to the fixed point of the kinetic equation which instead determines the final state .
further investigations are in progress .
in this article we have presented our results on thermalization , isotropization and building - up of the elliptic flow for fireballs produced in relativistic heavy ion collisions both at rhic and lhc energies .
we have put emphasis on the role of a nonequilibrium initial condition on the generation of the collective flow , using as a model the fkln - glasma initial condition which differs from the thermal distribution for the presence of a saturation scale in the low @xmath3 spectrum .
our study is based on kinetic theory at fixed @xmath0 . in our study
we have neglected the initial time evolution of the chromo - electric and chromo - magnetic fields ( the glasma ) produced immediately after the collision , whose intensity can be directly related to the saturation scale @xmath18 in the cgc before the collision .
therefore our approach could be justified as soon as the initial strong glasma longitudinal fields decay into particle quanta and its longitudinal pressure becomes positive .
the characteristic time for this decay is of the order of @xmath239 @xcite .
the advantage to use kinetic theory from @xmath30 rather than hydro is that the nonequilibrium initial distribution is not problematic as kinetic theory is built to study the evolution of a generic @xmath23 distribution function . as a non - equilibrium initial distribution in our simulations
we have used the spectrum obtained within the fkln model , which embeds a saturation scale in momentum space . in both rhic and lhc runs
we have found that thermalization times are @xmath240 0.8 - 1 fm / c .
moreover we have studied the time scales for the isotropization of the expanding fireball , defining transverse and longitudinal pressures , @xmath153 and @xmath32 respectively , and computing the time evolution of @xmath155 .
we have found that isotropization is not reached if @xmath241 , in agreement with @xcite ; this result is easily understood , since higher viscosity implies lower cross sections among partons and dynamics in the longitudinal direction and transverse plane are decoupled . on the other hand
it is significant that for a fluid with @xmath242 also isotropization ( @xmath243 ) occurs very quickly @xmath244 . we have then focused our attention on the elliptic flow production when the initial distribution has a saturation scale built in it .
we have found that the amount of elliptic flow produced in heavy ion collisions depends not only on the pressure gradients and the @xmath0 of the system , but also on the initial distribution in momentum space . in particular ,
an initial condition characterized by a momentum distribution with a saturation scale generates smaller @xmath12 respect to the thermal one .
this result is quite general , and we expect it should be valid , besides qgp in urhics , for systems like cold atoms in a magnetic trap which are characterized by a value of @xmath0 close to the qgp one @xcite . assuming the fkln distribution as the one arising from the melting of the glasma , the effect of the initial nonequilibrium distribution affects the estimate of @xmath0 of about a factor of two .
however we have seen also that this effect is maximal for semi - peripheral collisions , becoming quite small for very central collisions .
the relevance of our results is further enhanced by the fact that th - fkln with @xmath245 would generate a low @xmath246 respect to the available data .
this is the main conclusion of @xcite , namely that th - fkln is not able to describe for the experimental observations , might be revised . in order to make more precise comparison with experimental data we are currently implementing also fluctuating initial conditions which will allow to extend the present study to all the @xmath247 harmonics relevant in hic s
this will allow to see if fkln can account for the measured @xmath246 or the non - equilibrium damps such harmonics even more than what seen in hydrodynamics . in @xcite it is found that harmonics up to the fifth order can be reproduced by combining cym early - time with hydro late - evolution evolutions ; however in the calculations of @xcite a small deviation from equilibrium has to be assumed in order to use viscous hydrodynamical equations .
therefore it will be interesting to compute the higher order harmonics combining the cym initial spectrum with the dynamics embedded in the kinetic equations where the assumption of sudden thermalization can be relaxed .
finally , we have also computed the invariant gluon distribution functions , @xmath248 , for the fkln glasma and glauber initial conditions .
we have found that the initial longitudinal expansion strongly affects the evolution of these distributions at early times , causing their lowering of one order of magnitude within 1 fm / c for @xmath249 gev . on then other hand at @xmath250
the longitudinal expansion is compensated by the fast thermalization and even without including @xmath251 enhancement factors @xmath252 stays larger than unity for a time interval @xmath253 .
this suggested that even in the fast expanding qgp in hic a transient bose - einstein condensation can take place as suggested in @xcite .
however the effect of bose statistics should be relevant at @xmath254 , while our effect is related to the initial non - equilibrium and not to the final exact distribution . in order to verify quantitatively
how this attractor works in presence of the longitudinal expansion we have to perform simulations with bose enhancement factors in the collision integral kernel , which is beyond the scope of the present study but that we will perform in the near future .
_ acknowledgements . _
the authors acknowledge enlightening discussions with m. chernodub , j. y. ollitrault and n. su .
m. r. acknowledges mei huang for her kind hospitality at institute of high energy physics in beijing , where part of this work was completed . v. g. and f. s. acknowledge the erc - stg funding under the qgpdyn grant .
h. niemi , g. s. denicol , p. huovinen , e. molnar and d. h. rischke , phys . rev .
* 106 * ( 2011 ) 212302 z. xu and c. greiner , phys
. rev .
c * 79 * ( 2009 ) 014904 z. xu , c. greiner and h. stocker , phys .
* 101 * ( 2008 ) 082302 f. gelis , e. iancu , j. jalilian - marian and r. venugopalan , ann .
nucl . part .
* 60 * , 463 ( 2010 ) .
e. iancu and r. venugopalan , in * hwa , r.c .
et al . : quark gluon plasma * 249 - 3363 .
l. mclerran , arxiv:0812.4989 [ hep - ph ] ; hep - ph/0402137 .
f. gelis , int . j. mod
. phys .
a * 28 * , 1330001 ( 2013 ) .
d. kharzeev , e. levin and m. nardi , nucl .
a * 747 * , 609 ( 2005 ) ; d. kharzeev , e. levin and l. mclerran , phys .
b * 561 * , 93 ( 2003 ) ; d. kharzeev and m. nardi , phys .
b * 507 * , 121 ( 2001 ) ; d. kharzeev and e. levin , phys .
b * 523 * , 79 ( 2001 ) ; t. hirano and y. nara , nucl .
a * 743 * , 305 ( 2004 ) .
s. plumari , v. baran , m. di toro , g. ferini and v. greco , phys .
b * 689 * , 18 ( 2010 ) .
s. plumari , a. puglisi , m. colonna , f. scardina and v. greco , j. phys .
ser . * 420 * , 012029 ( 2013 ) .
s. plumari , a. puglisi , f. scardina and v. greco , phys .
c * 86 * ( 2012 ) 054902 .
j. -p .
blaizot _ et al .
_ , arxiv:1305.2119 [ hep - ph ] ; j. -p .
blaizot _ et al .
_ , nucl .
a904 - 905 * 2013 * , 829c ( 2013 ) . j. -p .
blaizot , f. gelis , j. liao , l. mclerran and r. venugopalan , nucl .
phys . a904 - 905 * 2013 * ( 2013 ) 829c c. gale , s. jeon , b. schenke , p. tribedy and r. venugopalan , phys .
* 110 * , 012302 ( 2013 ) .
r. ryblewski and w. florkowski , phys .
d * 88 * , 034028 ( 2013 ) .
f. gelis and t. epelbaum , arxiv:1307.2214 [ hep - ph ] .
k. fukushima , arxiv:1307.1046 [ hep - ph ] .
j. -p .
blaizot , t. lappi and y. mehtar - tani , nucl .
a * 846 * , 63 ( 2010 ) .
t. lappi , phys .
b * 703 * , 325 ( 2011 ) .
t. lappi , eur .
j. c * 71 * , 1699 ( 2011 ) . c. shen , u. w. heinz , j. -f .
paquet and c. gale , arxiv:1308.2440 [ nucl - th ] . c. shen , u. w. heinz , j. -f .
paquet , i. kozlov and c. gale , arxiv:1308.2111 [ nucl - th ] .
h. song , s. bass and u. w. heinz , arxiv:1311.0157 [ nucl - th ] .
u. heinz and r. snellings , ann .
nucl . part .
sci . * 63 * , 123 ( 2013 ) .
j. s. moreland , z. qiu and u. w. heinz , nucl .
a904 - 905 * 2013 * , 815c ( 2013 ) . j.
l. albacete , a. dumitru and y. nara , j. phys .
ser . * 316 * , 012011 ( 2011 ) .
z. qiu and u. w. heinz , phys .
c * 84 * , 024911 ( 2011 ) .
h. song , s. a. bass and u. heinz , phys .
c * 83 * , 054912 ( 2011 ) [ erratum - ibid .
c * 87 * , 019902 ( 2013 ) ] .
a. monnai and t. hirano , phys .
b * 703 * , 583 ( 2011 )
. t. hirano , p. huovinen and y. nara , phys .
c * 84 * , 011901 ( 2011 ) .
h. song and u. w. heinz , phys .
b * 658 * , 279 ( 2008 ) .
p. romatschke , int .
j. mod . phys .
e * 19 * , 1 ( 2010 ) .
d. a. teaney , arxiv:0905.2433 [ nucl - th ] .
p. huovinen and d. molnar , phys . rev .
c * 79 * , 014906 ( 2009 ) .
a. el , z. xu and c. greiner , phys . rev .
c * 81 * ( 2010 ) 041901 p.
b. arnold , g. dmoore and l. g. yaffe , jhep * 0305 * , 051 ( 2003 ) .
k. tsumura and t. kunihiro , arxiv:1311.7059 [ physics.flu-dyn ] .
d. bazow , u. w. heinz and m. strickland , arxiv:1311.6720 [ nucl - th ] .
s. plumari , a. puglisi , f. scardina and v. greco , j. phys .
ser . * 446 * , 012025 ( 2013 ) .
b. alver _ et al .
_ , phys .
c * 83 * , 024913 ( 2011 ) . | in this article we report on our results about the computation of the elliptic flow of the quark - gluon - plasma produced in relativistic heavy ion collisions , simulating the expansion of the fireball by solving the relativistic boltzmann equation for the parton distribution function tuned at a fixed shear viscosity to entropy density ratio @xmath0 .
our main goal is to put emphasis on the role of a saturation scale in the initial gluon spectrum , which makes the initial distribution far from a thermalized one .
we find that the presence of the saturation scale reduces the efficiency in building - up the elliptic flow , even if the thermalization process is quite fast @xmath1 and the pressure isotropization even faster @xmath2 .
the impact of the non - equilibrium implied by the saturation scale manifests for non - central collisions and can modify the estimate of the viscosity respect to the assumption of full thermalization in @xmath3-space .
we find that the estimate of @xmath0 is modified from @xmath4 to @xmath5 at rhic and from @xmath6 to @xmath4 at lhc .
we complete our investigation by a study of the thermalization and isotropization times of the fireball for different initial conditions and values of @xmath0 showing how the latter affects both isotropization and thermalization .
lastly , we have seen that the range of values explored by the phase - space distribution function @xmath7 is such that at @xmath8 the inner part of the fireball stays with occupation number significantly larger than unity despite the fast longitudinal expansion , which might suggest the possibility of the formation of a transient bose - einstein condensate . |
the minimal supersymmetric extension of the standard model ( mssm ) @xcite features many potential sources for cp violation @xcite beyond the standard model ( sm ) of particle physics .
most of these additional phases are associated with flavor mixing . in the flavor conserving sector , the @xmath1 gaugino mass parameter @xmath2 is conventionally chosen to be real and positive . the cp - violating , complex parameters are then the higgsino mass parameter @xmath3 , the @xmath4 and @xmath5 gaugino mass parameters @xmath6 and @xmath7 , respectively , and the trilinear scalar coupling parameters @xmath8 , @xmath9 these phases in general contribute to electric dipole moments ( edms ) , in particular to those of the neutron , and the thallium and mercury atoms . in fact , scenarios where ( at least ) one of the phases appearing in eq .
( [ eq : phases ] ) is large are often in conflict with current experimental upper bounds on these edms @xcite .
however , large phases are not excluded by these constraints , even if first generation sleptons are rather light . for example
, a certain degree of fine - tuning allows for cancellations among the different terms contributing to the edms @xcite .
the edm bounds can also be fulfilled by including lepton flavor violating couplings in the slepton sector @xcite .
see also refs .
@xcite .
it is clear that cp observables outside the low energy edm sector have to be measured to independently determine or constrain possible susy phases at colliders . in that respect ,
t - odd and cp - odd asymmetries based on triple or epsilon products have been proposed @xcite . for the ilc @xcite
, triple product asymmetries have been intensively studied in the production and decay of neutralinos @xcite and charginos @xcite , also using transversely polarized beams @xcite . at the lhc
@xcite , triple product asymmetries have been studied for the decays of stops @xcite , sbottoms @xcite , staus @xcite , and neutralinos which originate from squark decays @xcite .
the triple - product asymmetries can be of the order of @xmath10 , since they already appear at tree level due to spin correlations .
in particular , it was shown that the cp - violating effects in stop decays @xcite can be measured at the @xmath11 confidence level at the lhc for a luminosity of @xmath12 , using the technique of momentum reconstruction for on - shell decay chains @xcite .
we are thus motivated to explore the discovery reach at the lhc for cp violation in the production of a neutralino - chargino pair @xmath13 the contributing tree - level diagrams involving @xmath14 , @xmath15 squark and @xmath16 exchange are shown in fig .
[ fig : process2 ] . the phases @xmath17 and @xmath18 in the neutralino and chargino sector will trigger cp violation .
cp - sensitive contributions to the differential cross section will originate at tree level from the spin and spin - spin correlations among the neutralinos and charginos .
these can be analyzed with the help of the visible decay products of the neutralino and chargino .
leptonic decays are especially useful @xcite . on the one hand , hadronic decays lead to final states with very large sm backgrounds , and therefore do not lead to viable signals at the lhc . on the other hand ,
the construction of t - odd observables requires the measurement of the charges of final state particles , which is difficult , if not impossible , for jets .
we therefore use the leptonic two - body decays of the chargino and neutralino as `` spin analyzers '' : @xmath19 see fig .
[ fig : process ] for a schematic picture of the production and decay process .
the lhc signature is three isolated leptons ( two of them with same flavor and opposite charge ) and missing energy , carried away by the two neutralinos @xmath20 and the neutrino @xmath21 .
this process , known as the trilepton signal @xcite , has low qcd and sm background , and thus has been studied in detail as a susy discovery channel at the tevatron @xcite , and also at the lhc @xcite . in this paper
we first calculate the amplitude squared for the entire process of neutralino - chargino pair production and decay in the spin density matrix formalism @xcite .
the explicit formulas allow us to identify the cp - sensitive parts in the spin and spin - spin correlations , see section [ cp asymmetries ] . from those we define t - odd asymmetries of triple- and epsilon products . in section [ numerical analysis ]
, we numerically analyze these asymmetries , the production cross sections , and the neutralino and chargino branching ratios in general mssm scenarios with complex @xmath3 and @xmath6 .
we discuss several production modes @xmath22 , and two - body decays via gauge bosons and three - body decays .
the asymmetries analyzed in sections [ cp asymmetries ] and [ numerical analysis ] depend on the 4-momenta of the quarks in the initial state , and of the chargino and/or neutralino in the final state .
these can in general not be determined exactly . in section
[ construct ] , we therefore discuss approximations for the initial quark and intermediate gaugino momenta , which work quite efficiently to enhance the asymmetries . in section
[ discovery reach ] , we comment on the minimal luminosity required at the lhc to observe cp - violating effects in the trilepton mode , before we summarize and conclude in section [ summary and conclusions ] .
the appendices contain a review of chargino and neutralino mixing ( appendix [ chargino and neutralino mixings ] ) , of the relevant parts of the mssm lagrangian with complex couplings ( [ lagrangians and couplings ] ) , definitions of momenta and spin vectors as well as details of phase space integration ( [ phase space ] ) , analytical expressions for the amplitude squared in the spin - density matrix formalism ( [ density matrix formalism ] ) , and a discussion of how our asymmetries depend on the lepton charges ( [ cpasymmetries ] ) .
( 100,100)(20,40 ) ( 50,125)[lb]@xmath23 ( 50,65)[lb]@xmath24 ( 105,90)[]@xmath25 ( 150,125)[lb]@xmath26 ( 150,65)[lb]@xmath27 ( 122,100)(166,123 ) ( 166,77)(122,100 ) ( 77,100)(33,122 ) ( 33,77)(77,100 ) ( 77,100)(122,100 ) 44.5 ( 77,100)1.5 ( 122,100)1.5 ( 100,100)(-30,40 ) ( 75,135)[lb]@xmath23 ( 75,55)[lb]@xmath24 ( 110,95)[lb]@xmath14 ( 125,135)[lb]@xmath26 ( 125,52)[lb]@xmath27 ( 100,122)(144,130 ) ( 100,77)(144,69 ) ( 100,122)(55,130 ) ( 55,69)(100,77 ) ( 100,77)(100,122)5 ( 100,77)1.5 ( 100,122)1.5 ( 100,100)(-60,40 ) ( 75,135)[lb]@xmath23 ( 75,60)[lb]@xmath28 ( 90,100)[]@xmath15 ( 125,60)[lb]@xmath27 ( 125,133)[lb]@xmath26 ( 144,69)(100,122 ) ( 100,77)(144,130 ) ( 100,122)(55,130 ) ( 55,69)(100,77 ) ( 100,77)(100,122)5 ( 100,77)1.5 ( 100,122)1.5
in this section , we first identify the cp - sensitive parts in the amplitude squared of the entire process of neutralino - chargino pair production and their subsequent leptonic two - body decays , see eqs .
( [ eq : prod])-([eq : decaychar ] ) and fig .
[ fig : process ] . in order to probe these parts ,
we then define several t - odd products of 4-momenta and the corresponding asymmetries .
explicit expressions for the squared amplitude , relevant parts of the mssm lagrangian with complex couplings , and details of phase - space integration , are summarized in appendices a to d. in this and the following section we consider all ( products of ) 4-momenta to be observables . since not all of these momenta can be measured in actual lhc experiments , we will address solutions to this complication later in section [ construct ] .
the amplitude squared @xmath29 for neutralino - chargino pair production and decay , see fig . [
fig : process ] , can be decomposed into contributions from the neutralino spin correlations , the chargino spin correlations , the neutralino - chargino spin - spin correlations , and a spin - independent part , see eq .
( [ eq : matrixelement2 ] ) . at tree level ,
cp - sensitive parts can only originate from the spin and the spin - spin correlations .
they receive contributions from the exchange of the @xmath16 and the squarks @xmath30 , @xmath15 .
the cp - sensitive contributions to the neutralino spin correlations are and @xmath15 contributions , given by @xmath31 and @xmath32 , do not contain sufficiently many different couplings to be sensitive to any cp - violating phase . note that we have assumed vanishing @xmath33 mixing for first generation squarks , making their couplings chiral , whereas the @xmath16-@xmath34-@xmath35 couplings have no fixed chirality . ]
\sigma_{pd_1}(w \tilde u_l ) & \propto & { \rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ }
\;{\mathcal e}_{\tilde\chi^0_i } , \label{eq : neutwu}\\ [ 2 mm ] \sigma_{pd_1}(w \tilde
d_l ) & \propto & { \rm im}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } \;{\mathcal e}_{\tilde\chi^0_i } , \label{eq : neutwd}\\ [ 2 mm ] \sigma_{pd_1}(\tilde u_l \tilde d_l ) & \propto & { \rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } \;{\mathcal e}_{\tilde\chi^0_i}. \label{eq : neutud}\end{aligned}\ ] ] these spin correlation terms explicitly depend on the imaginary parts of the products of the @xmath16-@xmath37-@xmath38 couplings @xmath39 , the @xmath40-@xmath41-@xmath37 couplings @xmath42 , and the @xmath43-@xmath41-@xmath44 couplings @xmath45 .
the imaginary parts are manifestly cp - sensitive , _
i.e. _ sensitive to the phases @xmath18 , @xmath46 in the neutralino - chargino sector , and are each multiplied by the t - odd epsilon product @xmath47 \,\equiv \,\varepsilon_{\mu\nu\alpha\beta}~ p_{u}^{\mu}~p_d^{\nu}~p_{\tilde\chi^0_i}^{\alpha}~p_{\ell_1}^{\beta } \ , .
\label{eq : epsilon_neut}\ ] ] we use the convention @xmath48 , and here and in the following we not put a bar on any of the @xmath28 or @xmath49 quark indices , as long as the statements made apply for both @xmath50 or @xmath51 production processes . similarly , the cp - sensitive contributions to the chargino spin correlations are @xmath52 \sigma_{pd_3}(w \tilde u_l ) & \propto & { \rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } \ ; { \mathcal e}_{\tilde\chi^\pm_j } , \label{eq : charwu}\\ [ 2 mm ] \sigma_{pd_3}(w \tilde d_l ) & \propto & { \rm im}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } \;{\mathcal e}_{\tilde\chi^\pm_j } , \label{eq : charwd}\\ [ 2 mm ] \sigma_{pd_3}(\tilde u_l \tilde d_l ) & \propto & { \rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } \;{\mathcal e}_{\tilde\chi^\pm_j } \ , , \label{eq : charud}\end{aligned}\ ] ] with the short - hand notation for the epsilon product @xmath53 \,\equiv\,\varepsilon_{\mu\nu\alpha\beta}~ p_{u}^{\mu}~p_d^{\nu}~p_{\tilde\chi^\pm_j}^{\alpha}~p_{\ell_3}^{\beta } \ , .
\label{eq : epsilon_char}\ ] ] note that the chargino spin correlations probe the same coupling combinations as the neutralino spin correlation terms . finally , the cp - sensitive contributions to the spin - spin correlations are @xmath54 \sigma_{p d_1 d_3}(w
\tilde u_l ) & \propto & { \rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \}\ ; f , \label{eq : neutcharwu}\\ [ 2 mm ] \sigma_{p d_1 d_3}(w \tilde d_l ) & \propto & { \rm im}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \}\;f , \label{eq : neutcharwd}\\ [ 2 mm ] \sigma_{p d_1 d_3}(\tilde u_l \tilde d_l ) & \propto & { \rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } \;f , \label{eq : neutcharud}\end{aligned}\ ] ] with the short - hand notation for the kinematical function @xmath55 + ( p_u \cdot p_{\tilde\chi_i^0})[p_d , p_{\tilde\chi_j^\pm},p_{\ell_1},p_{\ell_3 } ] \nonumber \\[1 mm ] & & + ( p_d \cdot p_{\ell_3})[p_u , p_{\tilde\chi_i^0},p_{\tilde\chi_j^\pm},p_{\ell_1 } ] + ( p_d \cdot p_{\tilde\chi_j^\pm})[p_u , p_{\tilde\chi_i^0},p_{\ell_1},p_{\ell_3 } ] , \label{eq : xf2 } \end{aligned}\ ] ] see eq . ( [ eq : f2 ] ) in appendix [ squared amplitude of production and decay ] . again
the spin - spin correlations contain the same dynamics , _
i.e. _ the same coupling combinations , as the spin correlations .
note that the neutralino and chargino decays will not yield additional cp - sensitive contributions , since these are two - body decays via scalar particles .
the t - odd spin - spin correlation term @xmath56 is analogous to the corresponding term in neutralino @xcite or chargino @xcite pair production at the ilc , since all these processes have the same kinematical structure . each of the epsilon products @xmath57 , eq . ( [ eq : epsilon_neut ] ) , and @xmath58 , eq .
( [ eq : epsilon_char ] ) , or the tensor product @xmath56 , eq .
( [ eq : xf2 ] ) , is t - odd .
that is since each of the spatial components of the four - momenta changes sign under a naive time reversal , @xmath59 . due to cpt invariance
, these t - odd products are multiplied with the cp - sensitive , imaginary parts of products of couplings , @xmath60 , @xmath61 , @xmath62 , or @xmath63 .
the task is to define observables that project out the cp - sensitive parts in the spin and the spin - spin correlation terms in the amplitude squared .
this can be achieved by defining for the several possible t - odd products @xmath64 , @xmath58 , or @xmath56 the corresponding t - odd asymmetries of the cross section @xmath65 for neutralino - chargino production and decay @xmath66 |t|^2 d{\rm lips}\ ; d{\rm pdf } } { \int |t|^2 d{\rm lips}\;d{\rm pdf } } , \label{eq : toddasym}\ ] ] with the amplitude squared @xmath29 , eq .
( [ eq : matrixelement2 ] ) , the lorentz invariant phase - space @xmath67 ( [ eq : phasespace ] ) , and the short - hand notation @xmath68 , see eq .
( [ eq : sigmahadronic ] ) .
we obtain explicit expressions for the asymmetries by inserting the amplitude squared @xmath29 ( [ eq : matrixelement2 ] ) into eq .
( [ eq : toddasym ] ) : @xmath69 { \mathcal a}({\mathcal e}_{\tilde\chi^\pm_j } ) & = & \frac{\int { \rm sign}({\mathcal e}_{\tilde\chi^\pm_j})\ , \sigma_{pd_3}d_1 \ ; d{\rm lips } \ ; d{\rm pdf } } { \int p~d_1~d_3~d{\rm lips } \ ; d{\rm pdf } } , \label{eq : adependencechar}\\ [ 2 mm ] { \mathcal a}(f ) & = & \frac{\int { \rm sign}(f)\ , \sigma_{p d_1 d_3}\ ; d{\rm lips } \ ; d{\rm pdf } } { \int p~d_1~d_3~d{\rm lips } \ ; d{\rm pdf } } , \label{eq : adependenceneutchar}\end{aligned}\ ] ] using eqs .
( [ eq : dsigmaneut ] ) and ( [ eq : dsigmachar ] ) , and the phase space element @xmath67 as given in eq .
( [ eq : phasespace ] ) , where we have already used the narrow width approximation of the propagators , see eq .
( [ eq : narrowwidth ] ) . in the numerators of the asymmetries @xmath70 , eqs .
( [ eq : adependenceneut])-([eq : adependenceneutchar ] ) , only those spin or spin - spin correlations remain , which contain the corresponding t - odd products @xmath71 , @xmath58 , or @xmath56 .
the other terms of the amplitude squared vanish due to the phase space integration over the sign of the t - odd product , @xmath72 . in the denominator ,
all spin and spin - spin correlation terms vanish , and only the spin - independent parts contribute .
this last statement remains true even after applying acceptance cuts on the final state momenta , as long as these cuts are cp symmetric .
note that in general the largest asymmetries are obtained by using t - odd products of 4-momenta that match exactly the kinematical dependence of the cp - sensitive terms in the amplitude squared . in the literature , these are therefore sometimes referred to as _ optimal observables _ @xcite .
other combinations of momenta lead in general to smaller asymmetries .
, as used _
e.g. _ in ref .
typically their statistical significance is increased by @xmath73 @xcite , compared to the asymmetries given in eq .
( [ eq : toddasym ] ) .
however , we show in section [ construct ] that these optimal observables are more difficult to measure experimentally than the asymmetries we analyze . ] by construction , the asymmetries @xmath70 , eqs .
( [ eq : adependenceneut])-([eq : adependenceneutchar ] ) , are manifestly lorentz invariant . due to the boost between the partonic center - of - mass frame and the lab frame , the use of triple products , which are not lorentz invariant , generally leads to smaller asymmetries .
one famous example is the triple product of the three outgoing leptons @xcite @xmath74 that triple product probes the cp - sensitive terms in the spin - spin correlations and in the neutralino spin correlations , but not in the chargino spin correlations . since it does not directly match the kinematical dependence of the cp - sensitive terms , and since it is not lorentz invariant , the corresponding asymmetry will be reduced by a factor of about @xmath75 to @xmath76 compared to the maximal asymmetries as given in eqs .
( [ eq : adependenceneut])-([eq : adependenceneutchar ] ) .
these however contain the initial quark and the intermediate neutralino / chargino momenta , which can not be measured directly at the lhc and thus need to be approximated , whereas the lepton momenta appearing in eq .
( [ tripleterm ] ) are measurable with high accuracy . in section
[ construct ] we will compare numerically the sizes of different experimentally measurable cp - asymmetries around a benchmark scenario .
another complication arises because the initial state at the lhc , which contains two protons but no antiprotons , is not cp self - conjugate .
thus lhc experiments will strictly speaking not be able to measure true cp - odd asymmetries .
however , it is sufficient for our purpose that our asymmetries are odd under naive time reversal .
this ensures that at tree level they are non - zero only in the presence of non - trivial cp phases .
finally , one needs to be careful when summing over events with different lepton charges , since some of the asymmetries change sign when the sign of a lepton charge is flipped .
this is discussed in appendix [ cpasymmetries ] . .
mssm benchmark scenario .
[ tab : scenario ] [ cols="^,^,^,^,^,^,^,^ " , ] we are now ready to present some numerical results .
we first study the production of the second lightest neutralino and chargino pair at the lhc , @xmath78 , since we find the largest asymmetries there ; we will comment on channels involving the lighter chargino and heavier neutralino states in the next subsection . for our reference scenario ,
the remaining @xmath79 branching ratios , not listed in table [ tab : masses ] , are @xmath80 , @xmath81 , @xmath82 , summed over @xmath83 , and @xmath84 , @xmath85 @xcite . in fig .
[ fig : mu_m2_asymm_neut ] ( b ) , we show the @xmath86@xmath2 dependence of the cross section @xmath87 for neutralino - chargino pair production , which reaches several @xmath88 fb for light neutralinos and charginos .
our benchmark scenario is indicated by a cross in the @xmath86@xmath2 plane .
the neutralino and chargino branching ratios @xmath89 and @xmath90 are shown in fig .
[ fig : mu_m2_asymm_neut ] ( c ) , ( d ) .
the ( lorentz invariant ) asymmetry for the t - odd product @xmath91 $ ] , which appears in the neutralino spin correlations , is shown in fig .
[ fig : mu_m2_asymm_neut ] ( a ) . for the asymmetry we use the short - hand notation
@xmath92 $ ] .
that asymmetry will probe the cp - sensitive parts of the neutralino spin correlations , as discussed in section [ cp asymmetries ] .
looking at fig .
[ fig : mu_m2_asymm_neut ] ( a ) , we see that our benchmark point lies on a line , approximately reaching from @xmath93 gev to @xmath94 gev , where the asymmetry obtains its maximum . in the vicinity of that line
there is a level - crossing of the neutralino states @xmath95 and @xmath96 for @xmath97 , and between @xmath98 and @xmath95 for @xmath99 .
the level crossing of the neutralinos is also reflected by the abrupt change in the cross section , see fig .
[ fig : mu_m2_asymm_neut ] ( b ) .
note that for non - vanishing phases there is no true level - crossing since the neutralino masses , _
i.e. _ the singular values of the neutralino matrix @xcite , are driven apart by the complex off - diagonal terms .
$ ] in @xmath100 , see eq .
( [ eq : toddasym ] ) , and ( right ) the production cross section @xmath87 in fb , for neutralino - chargino pair production @xmath78 and subsequent leptonic two - body decays @xmath101 ; @xmath102 and @xmath103 at the lhc at @xmath104 tev .
the susy parameters are given in table [ tab : scenario ] .
, title="fig:",scaledwidth=48.0% ] ( -5.5,8 ) @xmath105 $ ] in % $ ] in @xmath100 , see eq .
( [ eq : toddasym ] ) , and ( right ) the production cross section @xmath87 in fb , for neutralino - chargino pair production @xmath78 and subsequent leptonic two - body decays @xmath101 ; @xmath102 and @xmath103 at the lhc at @xmath104 tev .
the susy parameters are given in table [ tab : scenario ] .
, title="fig:",scaledwidth=48.0% ] ( -5.5,8)@xmath106 in fb in fig .
[ fig : phases_asymm_sig ] , we show the phase dependence of the asymmetry @xmath107 $ ] ( left ) , and the corresponding neutralino - chargino cross section @xmath87 .
the asymmetry receives contributions from the different production channels of @xmath16 , @xmath14 , and @xmath15 exchange .
the contribution from @xmath108 exchange is shown in fig .
[ fig : phases_asymm_part ] ( a ) , that of @xmath109 exchange in fig .
[ fig : phases_asymm_part ] ( b ) , the @xmath110 exchange in fig .
[ fig : phases_asymm_part ] ( c ) , and finally the @xmath111 exchange in fig .
[ fig : phases_asymm_part ] ( d ) .
we can see that the individual contributions can be as large as @xmath112 , however the different contributions enter with opposite sign , and thus cancel each other partly .
see the sum of all their contributions to the asymmetry @xmath113 $ ] in fig .
[ fig : phases_asymm_sig ] ( left ) .
$ ] in @xmath100 , see eq .
( [ eq : toddasym ] ) , on @xmath86-@xmath2 ( left ) and on the phases ( right ) for neutralino - chargino pair production @xmath78 and subsequent leptonic two - body decays @xmath101 ; @xmath102 and @xmath114 at the lhc at @xmath104 tev
. the susy parameters are given in table [ tab : scenario ] .
compare with figs .
[ fig : mu_m2_asymm_neut ] ( a ) and [ fig : phases_asymm_sig ] ( left ) , where the asymmetry @xmath92 $ ] of the neutralino spin correlations is shown . ,
title="fig:",scaledwidth=48.0% ] ( -5.5,8 ) @xmath115 $ ] in % $ ] in @xmath100 , see eq .
( [ eq : toddasym ] ) , on @xmath86-@xmath2 ( left ) and on the phases ( right ) for neutralino - chargino pair production @xmath78 and subsequent leptonic two - body decays @xmath101 ; @xmath102 and @xmath114 at the lhc at @xmath104 tev
. the susy parameters are given in table [ tab : scenario ] .
compare with figs .
[ fig : mu_m2_asymm_neut ] ( a ) and [ fig : phases_asymm_sig ] ( left ) , where the asymmetry @xmath92 $ ] of the neutralino spin correlations is shown . , title="fig:",scaledwidth=48.0% ] ( -5.5,8)@xmath115 $ ] in % in fig .
[ fig : asymm_char ] ( left ) , we show the @xmath86@xmath2 dependence of the asymmetry of the t - odd product @xmath116 $ ] , which appears in the chargino spin correlations . for the asymmetry we use the short - hand notation @xmath115 $ ] , to indicate the different momenta used . in fig .
[ fig : asymm_char ] ( left ) , our benchmark scenario is indicated by a cross in the @xmath86@xmath2 plane . for that scenario
we show the phase dependence of the asymmetry in the right panel . by comparing its size with the corresponding asymmetry which probes the neutralino spin correlations , see figs .
[ fig : mu_m2_asymm_neut ] and [ fig : phases_asymm_sig ] , for the @xmath86@xmath2 dependence and the phase dependence , respectively , we see that the asymmetry which probes the chargino spin - correlations is about half as large .
this is only due to kinematics , as discussed in section [ cp asymmetries ] , since the combinations of products of imaginary couplings are the same for the neutralino and the chargino spin correlations .
the asymmetry , @xmath117 $ ] , which probes the chargino spin correlations , receives contributions from the different production channels of @xmath16 , @xmath14 and @xmath15 exchange .
the individual contributions from @xmath108 , @xmath109 , @xmath110 and @xmath111 exchange can be as large as @xmath118 ( not shown ) , however they enter with opposite sign , such that they cancel each other partly in their sum for the asymmetry @xmath92 $ ] .
the individual contributions have the same dynamical dependence on the cp phases as those for the neutralino spin correlations , compare fig .
[ fig : asymm_char ] ( right ) with fig .
[ fig : phases_asymm_sig ] ( left ) , and are only about a factor of @xmath75 smaller .
$ ] in @xmath100 , and ( right ) the triple product asymmetry @xmath119 in @xmath100 , see eq .
( [ eq : toddasym ] ) , for neutralino - chargino pair production @xmath120 and subsequent leptonic two - body decays @xmath121 ; @xmath122 and @xmath123 at the lhc at @xmath104 tev
. the susy parameters are given in table [ tab : scenario ] .
compare their sizes with the other asymmetries in figs .
[ fig : phases_asymm_sig ] ( left ) and [ fig : asymm_char ] ( right ) .
, title="fig:",scaledwidth=48.0% ] ( -4.5,8)@xmath124 $ ] in @xmath100 $ ] in @xmath100 , and ( right ) the triple product asymmetry @xmath119 in @xmath100 , see eq .
( [ eq : toddasym ] ) , for neutralino - chargino pair production @xmath120 and subsequent leptonic two - body decays @xmath121 ; @xmath122 and @xmath123 at the lhc at @xmath104 tev .
the susy parameters are given in table [ tab : scenario ] .
compare their sizes with the other asymmetries in figs .
[ fig : phases_asymm_sig ] ( left ) and [ fig : asymm_char ] ( right ) .
, title="fig:",scaledwidth=48.0% ] ( -5.6,8)@xmath119 in @xmath100 in fig .
[ fig : phases_asymm_spinspin ] ( left ) , we show the phase dependence of the cp asymmetry @xmath124 $ ] , which probes the cp - sensitive parts @xmath56 , see eq .
( [ eq : xf2 ] ) , of the spin - spin correlations in neutralino - chargino production and decay .
that asymmetry is lorentz - invariant , in contrast to the triple product asymmetry @xmath125 , see eq .
( [ tripleterm ] ) , which we show in fig .
[ fig : phases_asymm_spinspin ] ( right ) . since the triple product asymmetry is not lorentz invariant , and since it is not an optimal observable in the sense that the triple product does not match the kinematical dependence of the cp - sensitive terms in the spin or spin - spin correlations , it is greatly reduced compared to the other asymmetries . in section
[ construct ] , we will discuss alternatives to the triple product asymmetry . in fig .
[ fig : asym_sigma_mq ] , we show the dependence of the cross section @xmath126 ( left ) , and of the four different asymmetries ( right ) on the squark masses , @xmath127 . since the different interference contribution depend sensitively on the squark masses , also the cross section and the asymmetries are very sensitive . for increasing squark masses
, the @xmath16 boson contribution dominates , which results in the asymptotic values @xmath128 fb , whereas the asymmetries almost vanish in this limit .
( 16,8 ) ( -4,-20 ) ( 5.0,-20 ) as we can see from table [ tab : masses ] , the cross sections for @xmath131 production are much larger than the @xmath132 cross sections .
channels can wash out the strong cp signal from @xmath133 production , if the different @xmath134 modes are not separated .
however , for the present benchmark scenario , the pollution from the @xmath135 three - body decays can probably be made quite small , in particular if final states are excluded where all three leptons have the same flavor .
since @xmath135 decays produce rather soft leptons , they can be suppressed by putting a cut on the energy or @xmath136 of the unpaired lepton @xmath137 .
similarly , one may be able to suppress contributions from @xmath138 contributions through cuts on the paired leptons @xmath139 .
in particular , the neutralino two - body decay chain has a well - defined upper edge of the @xmath140 invariant mass distributions , with most events not too far from this edge .
a cut on the @xmath141 invariant mass should therefore suppress unwanted contributions .
depending on the masses , this could also suppress @xmath142 , if open .
] however , in scenarios with @xmath143 , it is difficult to simultaneously satisfy the inequalities @xmath144 these are needed to enable the neutralino and chargino two - body decays , eqs .
( [ eq : decayneut ] ) , ( [ eq : decaychar ] ) , respectively , while suppressing undesirable @xmath145 decays .
these decays , if allowed , reduce the branching ratio @xmath146 we are interested in
. moreover , the cp asymmetry for the neutralino @xmath147 $ ] changes sign with an intermediate @xmath148 compared to an intermediate @xmath149 in the neutralino decay chain @xcite .
thus the two contributions , from intermediate @xmath149 and @xmath150 , will tend to cancel , if they can not be distinguished experimentally on an event - by - event basis .. ] note that in the mssm one generally has @xmath151 , while @xmath1 gauge invariance implies @xmath152 .
thus some fine - tuning of parameters is needed to achieve @xmath153 without getting also automatically @xmath154 . as an example of a somewhat fine - tuned scenario , we keep the hierarchy @xmath155 as in our benchmark scenario , table [ tab : scenario ] , but choose a slightly smaller @xmath156 gev ( and close to @xmath157 gev ) , which enables the two body decay @xmath158 .
we also adopt the phase @xmath159 , which gives maximal asymmetries for this scenario .
we then scan the parameter space around that modified benchmark point .
although the production cross section @xmath160 reaches up to @xmath75 pb for small @xmath161 $ ] does not exceed @xmath162 .
the neutralino branching ratio @xmath163 gets slightly reduced since now the @xmath164 are lighter than @xmath149 , and reaches not more than @xmath0 .
the chargino branching ratio can reach @xmath165 , since only the chargino decay channels into leptons are open . similarly , the @xmath166 production cross section reaches @xmath167 pb , with @xmath168 , but the asymmetries are again small , @xmath169\lsim 1 \%$ ] . in a second approach ,
we invert the hierarchy between the left and right slepton states , taking @xmath170 .
then we choose @xmath171 sufficiently heavy to close the channel @xmath172 , but still allow for @xmath173 and @xmath174 , such that there is only the contribution from the left slepton to the neutralino asymmetry , and no cancellations appear .
however we can not find asymmetries larger than @xmath175 > 0.5\%$ ] .
so far we have assumed rather large cp phases and small sfermion masses .
this will generate too large electric dipole moments of sm fermions , unless the first and second generation @xmath176-parameters are fine - tuned accurately in both size and phase .
this could be ameliorated by choosing sufficiently large slepton masses .
the phase sensitivity of spin and spin - spin correlations could then be studied via chargino and neutralino decays into real @xmath177 and @xmath178 gauge bosons ; background reduction and charge determination would again force one to focus on purely leptonic decays of the gauge bosons .
however , the asymmetries for the neutralino decay @xmath179 with the subsequent leptonic decay @xmath180 are reduced by a factor @xmath181 , where @xmath182 is the left ( right ) sm coupling of the @xmath183 to charged leptons .
thus one loses almost an order of magnitude in all neutralino spin asymmetries .
in addition , the overall statistical significance is reduced , since @xmath184 summed over @xmath185 .
this reduces the total cross section for the purely leptonic final state , see the discussion at the end of section @xmath186 in ref . @xcite . due to the majorana properties of the neutralinos , the neutralino spin asymmetries identically vanish for decays into a higgs bosons , @xmath187 , since the couplings obey @xmath188 .
a similar argument holds for the chargino decay into a @xmath16 boson , @xmath189 .
if the @xmath16 momentum is reconstructed , which is in principle possible for hadronic decays , the reduction factor @xmath190 to the chargino spin asymmetries due to the left @xmath191 and right @xmath192 @xmath193-@xmath98-@xmath16 couplings is typically of the order of @xmath194 to @xmath195 @xcite . for leptonic decays the overall statistical significance is reduced due to the branching ratio @xmath196 , for @xmath185 .
lastly , for the chargino decay via a @xmath16 , and the neutralino decay via an on - shell @xmath183 boson , one would have to fight the large sm background from @xmath197 production , of order @xmath198 pb @xcite .
thus maximally p - violating chargino and neutralino decays , which can be realized only if sleptons are light , are ideal for analyzing the cp - violating effects in chargino - neutralino production .
there are also regions of parameter space where some of the charginos and neutralinos only have three - body decays .
this happens if the sleptons ( and squarks ) are heavier than the charginos and neutralinos in question ; in addition , the mass difference between these states and the lightest neutralino has to be smaller than the masses of the @xmath16 and @xmath183 boson .
three - body decays , which proceed via the exchange of virtual sfermions , gauge and higgs bosons , could then provide additional cp - violating contributions to the asymmetries @xcite .
( this also happens for two - body decays into spin-@xmath167 @xmath183 @xcite and @xmath16 @xcite bosons . )
although that could be interesting , as those cp - violating contributions can be of the order of @xmath0 @xcite , it could be more difficult to disentangle the two different cp - violating contributions from production and decay .
since the three - body decay scenario would require quite different calculations and phenomenology , we defer its analysis to another work .
so far we have discussed those cp asymmetries , based on the optimal epsilon products , which exactly match the kinematical dependence of the cp - sensitive terms in the amplitude squared .
for example , in the neutralino spin correlations this is the epsilon product @xmath199 $ ] , see eq .
( [ eq : epsilon_neut ] ) .
this epsilon product contains the quark and neutralino momenta , which are not directly accessible at the lhc .
even if we knew all particle masses , we would not be able to completely reconstruct the momenta in the event .
there are @xmath200 unknowns : the @xmath186-momenta of the @xmath98 from the neutralino decay , and the invisible sneutrino ( @xmath201 ) from the chargino decay , as well as the @xmath202-components of the @xmath28 and @xmath23 momenta , but only @xmath203 kinematical constraints : @xmath76 from energy - momentum conservation , and @xmath186 invariant mass constraints for the decays of the chargino , the neutralino , and the slepton .
a complete reconstruction is thus impossible , since the decay chain on the chargino side is too short due to the invisible sneutrino . in this section
we thus discuss different methods to approximate the intermediate neutralino and the initial partonic momenta .
we have checked that all our results for the asymmetries and cross sections are in agreement with the public code ` madgraph ` @xcite .
since the estimates we present in the following can not be perfect , we quantify the dilution of the asymmetries @xmath70 due to the momenta replacements , and the dilution of the cross section @xmath65 due to selection cuts .
the figure of merit will be the statistical significance , which roughly scales like @xmath204 for a given luminosity @xmath205 , assuming @xmath206 acceptance . for our benchmark point , with @xmath207 production and decay , see table [ tab : scenario ] , we have @xmath208 , @xmath209 , @xmath210 .
the task is to obtain larger significances than that of the triple product asymmetry @xmath211 , for which we have @xmath212 for @xmath213 . as a preselection
, we apply cuts on the transverse momentum and rapidity of each lepton @xmath214 where @xmath215 is the difference of the azimuthal angles of a lepton pair in radiant , and @xmath216 their rapidity difference .
the preselection are standard cuts to isolate leptons , which are for example included as basic cuts in ` madgraph ` @xcite . about @xmath217 of our signal events
pass these cuts .
we replace the neutralino momentum @xmath218 by the lepton momenta @xmath219 from its decay , such that the epsilon product becomes @xmath220 \to [ p_u , p_d , p_{\ell_1}+ p_{\ell_2},p_{\ell_1}]= [ p_u , p_d , p_{\ell_2},p_{\ell_1}]$ ] .
the corresponding asymmetry is reduced from @xmath221 to @xmath222 .
we can improve the approximation of the neutralino momentum by focusing on events where the lepton pair @xmath141 from the neutralino decay has a large invariant mass , @xmath223 , which optimizes the significance for @xmath224 .
this cut ensures that the sum of the 3-momenta of the leptons is small in the @xmath95 rest - frame .
note that a cut on this quantity might be needed to reduce sm backgrounds where the lepton pair originates from a virtual photon .
in addition we only keep those lepton pairs with a large transverse momentum sum , @xmath225 , which optimizes the significance for @xmath226 .
about two thirds of the preselected events pass both these cuts , but the asymmetry @xmath227 $ ] is enhanced from @xmath222 to @xmath228 .
note that the sign of the asymmetry depends on the charge of the near @xmath229 and far lepton @xmath230 from the neutralino decay in the following way , see also the discussion in appendix [ cpasymmetries ] , @xmath231 & = & -\mathcal{a } [ p_u , p_{d},p_{\ell_1 ^ - } , p_{\ell_2^+ } ] \\&= & + \mathcal{a } [ p_u , p_{d},p_{\ell_2^+ } , p_{\ell_1 ^ - } ] \\&= & -\mathcal{a } [ p_u , p_{d},p_{\ell_2 ^ - } , p_{\ell_1^+ } ] .
\label{charge } \end{aligned}\ ] ] the sign change in the first step originates from eq .
( [ eq : sigmaad1 ] ) , and the sign change in the second and third steps is due to the interchange of the momenta of @xmath229 and @xmath230 in the antisymmetric epsilon product .
these relations are important , since we need not determine from which vertex , near or far , the leptons @xmath229 and @xmath230 originate .
instead in the epsilon product one just groups them according to their charge , and uses for example the asymmetry @xmath232 $ ] , see also the discussion in appendix g of ref .
one can ensure that the two leptons stem from the neutralino decay , if one requires that the lepton @xmath137 from the chargino decay has different flavor than the opposite - sign same - flavor lepton pair @xmath233 from the neutralino decay
. the sign of @xmath234 $ ] , eq .
( [ eq : epsilon_neut ] ) , depends on the direction of the incoming @xmath28-quark .
( the @xmath23 antiquark then obviously comes from the opposite direction . ) in most events @xmath235 the @xmath28-quark will be more energetic than the antiquark @xmath23 , due to the characteristic momentum distributions of valence @xmath28 and sea - quarks @xmath23 in the pdfs .
so the entire event will be mostly boosted in the direction of the incoming @xmath28-quark .
the z - component of the sum of all three lepton momenta , @xmath236 , coincides in @xmath237 of the cases with the direction of the incoming @xmath28-quark momentum .
however , guessing wrong will immediately reduce the asymmetry , since then the event is included with the wrong sign .
the efficiency of the guess can be considerably enhanced to over @xmath238 , if we instead require that only the lepton pair from the neutralino decay has a large component in the direction of the beam , @xmath239 .
about half of the preselected events pass the cut for @xmath240 , which optimizes the significance . with these approximated quark momenta , @xmath241 = \pm 1 , \label{eq : quarkapproxd}\end{aligned}\ ] ]
we obtain for the asymmetry @xmath242 = -20 \%$ ] .
if we now also replace the neutralino momentum and include the cuts on the invariant mass and the transverse momentum of the lepton pair , we obtain an asymmetry of @xmath243 = 17 \%$ ] , with a cut efficiency of @xmath244 after preselection cuts , such that the corresponding significance is @xmath245 , for @xmath246 .
this value has to be compared with the significance @xmath247 for the triple product asymmetry @xmath248 , with the lepton preselection cuts only .
although our cuts and momenta approximations can more than double the significance , it will be difficult to measure the asymmetries at the lhc due to the low production cross section .
we will comment on this issue in the next section .
the trilepton signal for cp - conserving neutralino - chargino pair production at the lhc has been studied by the atlas @xcite and cms @xcite collaborations .
they have focused on the benchmark point su2 , given by @xmath249 tev , @xmath250 gev , @xmath251 , @xmath252 , @xmath253 , which lies within the focus point region of the msugra parameter space .
thus it is characterized by heavy squarks and sleptons of order @xmath186 tev , but relatively light charginos and neutralinos of order @xmath254 gev to @xmath255 gev .
while the heavy squarks and sleptons only provide rather small production cross sections , the light gauginos provide good susy discovery potential in multi - lepton events . in general , the most important sm backgrounds to the trilepton signal
are from @xmath256 , @xmath257 , and @xmath258 production @xcite .
the fully leptonic @xmath259 events can be reduced by rejecting lepton pairs ( with opposite signs and same flavor ) which have an invariant mass @xmath260 gev around @xmath261 .
the leptonic decays of @xmath262 , @xmath263 , can generate a third , but rather soft , lepton from the leptonic @xmath264 decay .
stringent isolation cuts on the lepton tracks can reduce those backgrounds , and also soft leptons from bremsstrahlung , other hadron decays , and photon conversions , which otherwise have large contributions .
the dominant source of low momentum trileptons are heavy flavor sm processes , like @xmath265 , @xmath266 , for @xmath267 , as pointed out in ref @xcite , and a cut on @xmath268 around @xmath269 gev is proposed . for the cp - conserving su2 scenario
, the atlas collaboration @xcite has shown that with appropriate cuts , the susy trilepton signal gets reduced from about @xmath270 fb to @xmath186 fb , with a remaining sm background of @xmath271 fb @xcite .
thus atlas expects a @xmath272 discovery over background in the cp - conserving su2 scenario for a luminosity of @xmath273 @xmath274 .
if we assume a similar loss in the signal events , due to the cuts , of about an order of magnitude , we can estimate the discovery potential for cp violation in the trilepton signal .
for our scenario for @xmath275 production as given in table [ tab : scenario ] , we have shown that the asymmetry @xmath243 $ ] can reach up to @xmath276 .
the necessary cuts for the asymmetry will be in some sense equivalent to those which isolate the signal .
we would then need at least a luminosity of @xmath277 for @xmath278 standard deviations .
clearly the exact answer can only be given after performing a detailed experimental study .
however the trilepton signal is probably not best suited to study susy cp violation at the lhc , and thus we also defer a detailed monte carlo study taking into account the above mentioned backgrounds and cuts .
in the complex mssm , we have analyzed the potential to observe cp violation from the gaugino and higgsino phases @xmath17 and @xmath18 in neutralino - chargino pair production at the lhc , @xmath279 .
their subsequent leptonic two - body decays give rise to a trilepton signal , with low qcd and sm backgrounds .
the trilepton signal is well known as a clean susy discovery channel at the tevatron , and also at the lhc . in order to find the optimal cp asymmetries in the trilepton signal at the lhc ,
we have calculated the amplitude squared in the spin density matrix formalism . from the explicit formulas we have identified the cp - sensitive contributions , which already appear at tree level in the neutralino and chargino spin and spin - spin correlations .
we have then defined optimal cp asymmetries , which base on epsilon products that exactly match the kinematic dependence of the cp - violating spin correlations . after performing a systematic scan in the mssm parameter space , we have found that these asymmetries can reach up to @xmath280 for scenarios with light squarks , neutralinos and charginos , in particular for parameter points near level crossings where the neutralinos strongly mix . only the cancellations between the different exchange contributions from squark @xmath281 and @xmath16 boson exchange in the production ,
prevent the asymmetries from attaining larger values .
these optimal asymmetries would however require a reconstruction of the initial quark and intermediate neutralino / chargino momenta , which is only possible to a certain degree at the lhc .
we thus have discussed different replacement and approximation strategies , which are best realized by the analysis of triple products of the outgoing three lepton momenta , and by efficient methods to estimate the initial partonic systems .
for example , we have shown that , by using appropriate cuts on the final lepton momenta , the direction of the initial quark momenta can be guessed right in over @xmath238 of the events , such that the resulting asymmetries can still reach values up to @xmath282 .
we have checked that all our results are in agreement with the public code ` madgraph ` @xcite .
these washout effects , compared to the optimal asymmetries , are caused by the strong partonic boosts at the lhc , and by cancellations between the different contributions from the spin and spin - spin correlations . due to these effects
, we conclude that susy cp violation in the trilepton signal will be difficult to observe , in particular when squarks and gauginos are much heavier than @xmath283 gev .
we have estimated that one would need at least a luminosity of @xmath284 , for @xmath278 standard deviations to observe cp - violating effects at the lhc .
we thank uli nierste for inspiring discussions and motivations to analyze triple products in the trilepton signal .
we thank k. rolbiecki for helpful comments and discussions .
sb and ok thank the instituto de fsica , universidade de so paulo , for kind hospitality .
this work was supported in part by a grant funded jointly by the dfg ( germany ) and fapesp ( so paulo state , brazil ) , and in part by the bmbf , and by cnpq - brazil .
ok was supported by micinn project fpa.2006 - 05294 , and by a cpan ( spain ) fellowship .
jk was supported in part by the arc centre of excellence for particle physics at the terascale and in part by the initiative and networking fund of the helmholtz association , contract ha-101 ( `` physics at the terascale '' ) . * appendix *
the masses and mixing angles of the charginos follow from their mass matrix @xcite @xmath285 with the @xmath1 gaugino mass parameter @xmath2 , the higgsino mass parameter @xmath3 , the ratio @xmath286 of the vacuum expectation values of the two neutral higgs fields , and the mass @xmath287 of the @xmath16 boson .
the chargino mass matrix can be diagonalized by two complex unitary @xmath288 matrices @xcite , @xmath289 such that the chargino masses satisfy @xmath290 .
the complex symmetric mass matrix of the neutralinos in the photino , zino , higgsino basis ( @xmath291 ) , is given by @xcite @xmath292 here @xmath293 , @xmath294 , @xmath295 being the weak mixing angle , @xmath296 , @xmath297 , @xmath6 is the @xmath298 gaugino mass parameter , and @xmath261 is the mass of the @xmath183 boson .
we diagonalize the neutralino mass matrix by a complex , unitary @xmath299 matrix @xcite @xmath300 such that the neutralino masses satisfy @xmath301 .
the @xmath1 gaugino mass parameter @xmath2 has been chosen real and positive by absorbing its possible phase via field redefinitions .
gaugino fields .
we then also need to redefine the fermion and/or sfermion fields in order to keep the @xmath1 gaugino fermion sfermion couplings real and positive . keeping the fermion sfermion couplings to higgsinos as well as @xmath302 and @xmath5 gauginos real
then requires phase transformations of these fields as well .
this illustrates that only the _ relative _ phases between @xmath2 and the other gaugino masses , and between @xmath2 and @xmath3 , are physical . ]
thus we parametrize cp - violation in the neutralino and chargino sector by the phases of the complex parameters @xmath303 and @xmath304 .
the interaction lagrangians for @xmath16 boson exchange in the production are @xcite @xmath305\tilde\chi^+_j + \mbox{h.c.},\end{aligned}\ ] ] with the weak coupling constant @xmath306 , @xmath307 , @xmath308 . in the photino , zino , higgsino basis ( @xmath291 )
, the couplings are @xmath309 the interaction lagrangians for @xmath28-squark exchange are @xmath310 { \scr l}_{d \tilde u\tilde\chi^+ } & = & g\,\bar d\,l_{\tilde u j}^{l}\,p_r \,\tilde\chi_j^{+c}\,\tilde u_l + { \rm h.c . } , \end{aligned}\ ] ] and those for @xmath49-squark exchange are @xmath311 { \scr l}_{u \tilde d\tilde\chi^+ } & = & g\,\bar u\,l_{\tilde d j}^{l}\,p_r \,\tilde\chi_j^{+}\,\tilde d_l + { \rm h.c . } .\end{aligned}\ ] ] in the photino , zino , higgsino basis , the couplings are @xmath312 , \label{eq : flu}\\[2 mm ] l_{\tilde u j}^{l } & = & -v_{j1 } , \label{eq : luj}\end{aligned}\ ] ] and @xmath313 , \label{eq : fld}\\[2 mm ] l_{\tilde d j}^{l } & = & -u_{j1}. \label{eq : ldj}\end{aligned}\ ] ] we neglect mixings in the quark and squark generations .
the interaction lagrangians for neutralino decay @xmath314 , followed by @xmath315 with @xmath316 are @xcite @xmath317 with the couplings @xcite @xmath318 , \label{eq : fl}\\[2 mm ] f_{\ell i}^r & = & \sqrt{2 } \sin \theta_w \left(\tan\theta_w n_{i2}^*-n_{i1}^*\right ) .
\label{eq : fr}\end{aligned}\ ] ] the interaction lagrangians for chargino decay @xmath319 , followed by @xmath320 are @xcite @xmath321 { \cal l}_{\nu \tilde\nu\tilde\chi^0 } & = & g f_{\nu k}^{l } \bar{\nu } p_r \tilde{\chi}^0_k \tilde{\nu}_l+ \mbox{h.c.},\end{aligned}\ ] ] with @xmath322 for completeness , we also give the couplings in the bino , wino , higgsino @xmath323 basis @xcite . in that basis
the neutralino mass matrix is diagonalized by the complex , unitary matrix @xmath183 , similar to eq .
( [ eq : neutn ] ) .
the quark - squark gaugino couplings are @xmath324 with @xmath325 , and @xmath326 and @xmath327 as given in eqs .
( [ eq : luj ] ) , ( [ eq : ldj ] ) . with these definitions
the cp - sensitive imaginary parts of product of couplings become @xmath328 & & \phantom{a } + \frac{1}{\sqrt{2 } } z_{i2}z_{i3 } u_{j2}^\ast v_{j1}^\ast + z_{i2}^2 u_{j1}^\ast v_{j1}^\ast \big\ } , \\[4 mm ] { \rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } & = & \frac{1}{\sqrt{2 } } { \rm im}\big\ { v_{j1}^\ast \big ( \frac{t_w}{3 } z_{i1}z_{i3 } u_{j2}^\ast + \frac{t_w}{3 } z_{i1}z_{i2 } u_{j1}^\ast \nonumber\\[2 mm ] & & \phantom{lalalalal } + \frac{1}{\sqrt{2 } } z_{i2}z_{i3 } u_{j2}^\ast + z_{i2}^2 u_{j1}^\ast \big ) \big\ } , \\[4 mm ] { \rm im}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } & = & \frac{1}{\sqrt{2 } } { \rm im}\big\ { u_{j1}^\ast \big ( \frac{t_w}{3 } z_{i1}z_{i4 } v_{j2}^\ast -\frac{t_w}{3 } z_{i1}z_{i2 } v_{j1}^\ast \nonumber\\[2 mm ] & & \phantom{lalalalal } -\frac{1}{\sqrt{2 } } z_{i2}z_{i4 } v_{j2}^\ast + z_{i2}^2 v_{j1}^\ast \big ) \big\ } , \\[4 mm ] { \rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } & = & \frac{1}{2 } { \rm im}\big\ { u_{j1}^\ast v_{j1}^\ast \big ( z_{i2}^2 - \frac{t_w^2}{3 ^ 2 } z_{i1}^2 \big ) \big\}.\end{aligned}\ ] ]
in the center - of - mass ( cms ) system of the incoming quarks , we parametrize the momenta with the scattering angle @xmath329 , and the azimuth @xmath330 to be chosen zero , @xmath331 & & \hat p_{\tilde\chi_i^0}^{\mu } = ( \hat e_i,-\hat q\sin\hat\theta,0 , \hat q\cos\hat\theta),\quad \hat p_{\tilde\chi_j^\pm}^{\mu } = ( \hat e_j,\hat q\sin\hat\theta,0,-\hat q\cos\hat\theta),\qquad \end{aligned}\ ] ] with the cms energy of the partons @xmath332 , and @xmath333 with @xmath334 . we label variables in the cms system by a hat in our notation .
the energies and the @xmath202-components of the momenta of the outgoing neutralino @xmath27 and chargino @xmath38 in the laboratory ( lab ) frame @xmath335 are obtained by a lorentz boost with @xcite @xmath336 the partons have energy fractions @xmath337 , @xmath338 , of the proton energy @xmath339 in the laboratory ( lab ) frame , such that @xmath340 , and @xmath341 in the laboratory frame , the momenta for the subsequent decays of the neutralino @xmath342 ; @xmath343 and of the chargino @xmath344 ; @xmath345 , are given by @xmath346 p_{\ell_2}^{\mu } & = & e_{\ell_2}(1,\ , \sin \theta_2 \cos \phi_2 , \ , \sin \theta_2 \sin \phi_2 , \ ,
\cos \theta_2),\\[2 mm ] p_{\ell_3}^{\mu } & = & e_{\ell_3}(1,\ , \sin \theta_3 \cos \phi_3 , \ , \sin \theta_3 \sin \phi_3 , \ , \cos \theta_3 ) ,
\label{eq : momenta2}\end{aligned}\ ] ] with the energies @xmath347 & & e_{\ell_3 } = \frac{m^2_j - m^2_{\tilde\nu_\ell } } { 2(e_j-|{\mathbf p}_j|\cos\theta_{d_3 } ) } , \label{eq : energies}\end{aligned}\ ] ] and the decay angles , @xmath348 , @xmath349 , and @xmath350 . for the description of the polarization of the neutralino @xmath37 and chargino @xmath38 we choose three spin vectors in the laboratory frame @xmath351 s^{1,\,\mu}_j=\left(0,\frac{{\bf s}^2_j\times{\bf s}^3_j } { |{\bf s}^2_j\times{\bf s}^3_j|}\right ) , s^{2,\,\mu}_j=\left(0 , \frac{{\bf p}_u\times{\bf p}_j } { |{\bf p}_u\times{\bf p}_j|}\right ) , s^{3,\,\mu}_j=\frac{1}{m_j } \left(|{\bf p}_j| , \frac{e_j}{|{\bf p}_j|}{\bf p}_j \right ) . \qquad \label{eq : spinvecchar}\end{aligned}\ ] ] they form an orthonormal set @xmath352 with the unit momentum vectors @xmath353 .
the lorentz invariant phase - space element for neutralino - chargino production and their subsequent two - body decay chain , see eqs .
( [ eq : prod])-([eq : decaychar ] ) , can be decomposed into two - body phase - space elements @xcite @xmath354 the several parts of the phase space elements are given by @xmath355 with @xmath356 , @xmath357 , and @xmath358 , @xmath359 .
for the calculation of the amplitude squared of neutralino - chargino production , eq .
( [ eq : prod ] ) , and their two - body decay chains , eqs .
( [ eq : decayneut])-([eq : decaychar ] ) , we use the spin - density matrix formalism of ref .
the amplitudes squared in the helicity formalism are given , _
e.g. _ in ref .
@xcite . for the production of the neutralino - chargino pair , @xmath360 with momentum @xmath361 and helicity @xmath362 , the un - normalized spin - density matrix is @xmath363 the helicity amplitudes are @xmath364 \left [ \bar u(p_j,\lambda_j ) \gamma_\mu ( o^{l\ast}_{ij } p_l+o^{r\ast}_{ij } p_r ) v(p_i,\lambda_i ) \right ] , \nonumber \\ \label{eq :
tw}\\ t_p^{\lambda_i\lambda_j}(t,\tilde u ) & = & g^2 \delta_t(\tilde u_l ) \left [ \bar u(p_i , \lambda_i)f_{u i}^{l\ast } p_l u(p_u ) \right ] \left [ \bar v(p_d ) l_{\tilde u j}^{l } p_r v(p_j,\lambda_j ) \right ] , \label{eq : tu}\\ t_p^{\lambda_i\lambda_j}(u , \tilde d ) & = & -g^2 \delta_u(\tilde d_l ) \left [ \bar v(p_d)f_{d i}^l p_r v(p_i,\lambda_i ) \right ] \left [ \bar u(p_j , \lambda_j ) l_{\tilde d j}^{l\ast } p_l u(p_u ) \right ] , \label{eq : td}\end{aligned}\ ] ] with the propagators @xmath365 and @xmath366 , @xmath367 , @xmath368 .
the feynman diagrams are shown in fig .
[ fig : process2 ] . having introduced a set of spin four - vectors @xmath369 with @xmath370 for the neutralino @xmath27 , see eq .
( [ eq : spinvecneut ] ) , and @xmath371 with @xmath372 for the chargino @xmath26 , see eq .
( [ eq : spinvecchar ] ) , the production matrix ( [ eq : rhopdef ] ) can be expanded in terms of the pauli matrices @xmath373 using the bouchiat - michel formulas for massive spin @xmath374 particles @xcite @xmath375({\not\!p}+m),\\ v(p,\lambda^\prime)~\bar v(p,\lambda)&=&\frac{1}{2}~[\delta_{\lambda^\prime\lambda}+ \gamma_5{\not\!s}^a \sigma^a_{\lambda'\lambda}]({\not\!p}-m).\end{aligned}\ ] ] the expansion coefficient @xmath376 of the production density matrix ( [ eq : rhop ] ) is independent of the chargino and neutralino polarizations .
it can be composed into contributions from the different production channels @xmath377 with @xmath378 , \label{eq : pww } \\[2 mm ] p(\tilde u_l \tilde u_l ) & = & \frac{g^4}{4}|\delta_t(\tilde u_l)|^2 |f_{u i}^l|^2 |l_{\tilde u j}^{l}|^2 ( p_u \cdot p_i)(p_d\cdot p_j ) , \label{eq : puu } \\ [ 2 mm ] p(\tilde d_l \tilde d_l ) & = & \frac{g^4}{4}|\delta_u(\tilde d_l)|^2 |f_{d i}^l|^2 |l_{\tilde d j}^{l}|^2 ( p_u \cdot p_j)(p_d\cdot p_i ) , \label{eq : pdd } \\ [ 2 mm ]
p(w \tilde u_l ) & = & \frac{\sqrt 2}{4}g^4\delta_s(w ) \delta_t^\ast(\tilde u_l ) \big [ 2 { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{l\ast}_{ij } \ } ( p_u \cdot p_i)(p_d\cdot p_j ) \nonumber \\ & & + { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_i m_j(p_u \cdot p_d ) \big ] , \label{eq : pwu } \\[2 mm ] p(w \tilde d_l ) & = & -\frac{\sqrt 2}{4}g^4\delta_s(w ) \delta_u^\ast(\tilde d_l ) \big [ 2 { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{r\ast}_{ij } \ } ( p_u \cdot p_j)(p_d\cdot p_i ) \nonumber \\ & & + { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_i m_j(p_u \cdot p_d ) \big ] , \label{eq : pwd } \\[2mm]p(\tilde u_l \tilde d_l ) & = & -\frac{g^4}{4 } \delta_t(\tilde u_l)\delta_u^\ast(\tilde
d_l ) { \rm re}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } m_i m_j ( p_u \cdot p_d ) , \label{eq : pud } \end{aligned}\ ] ] with the couplings as defined in appendix [ lagrangians and couplings ] .
the terms for @xmath376 are the same for the charge conjugated process , @xmath379 .
the coefficients @xmath380 , which describe the polarization of the neutralino @xmath37 , decompose into @xmath381 & & \sigma_p^a(w \tilde u_l)+\sigma_p^a(w \tilde d_l)+\sigma_p^a(\tilde u_l \tilde d_l ) , \label{eq : sigmaneut}\end{aligned}\ ] ] with @xmath382 \nonumber \\ & & -{\rm im}\{o^l_{ij } o^{r\ast}_{ij } \ } m_j[p_u , p_d , p_i , s_i^a ] \big\ } , \label{eq : sigmapaww } \\ [ 2 mm ] \sigma_p^a(\tilde u_l \tilde u_l ) & = & \frac{g^4}{4}|\delta_t(\tilde u_l)|^2 |f_{u i}^l|^2 |l_{\tilde u j}^{l}|^2 m_i(p_d \cdot p_j)(p_u\cdot s_i^a ) , \label{eq : sigmapauu } \\ [ 2 mm ] \sigma_p^a(\tilde d_l \tilde d_l ) & = & -\frac{g^4}{4}|\delta_u(\tilde d_l)|^2 |f_{d i}^l|^2 |l_{\tilde d j}^{l}|^2 m_i(p_u \cdot p_j)(p_d\cdot s_i^a ) , \label{eq : sigmapadd } \\ [ 2 mm ] \sigma_p^a(w
\tilde u_l ) & = & \frac{\sqrt 2}{4}g^4 \delta_s(w ) \delta_t^\ast(\tilde u_l ) \big\ { 2 { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{l\ast}_{ij } \ } m_i(p_d \cdot p_j)(p_u\cdot s_i^a ) \nonumber \\ & & + { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_j \left[(p_d \cdot p_i)(p_u \cdot s_i^a)-(p_u \cdot p_i)(p_d \cdot s_i^a ) \right ] \nonumber \\ & & -{\rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_j[p_u , p_d , p_i , s_i^a ] \big\ } , \label{eq : sigmapawu } \\[2 mm ] \sigma_p^a(w \tilde d_l ) & = & \frac{\sqrt 2}{4}g^4 \delta_s(w ) \delta_u^\ast(\tilde d_l ) \big\ { 2 { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{r\ast}_{ij } \ } m_i(p_u \cdot p_j)(p_d\cdot s_i^a ) \nonumber \\ & & + { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_j \left[(p_u \cdot p_i)(p_d \cdot s_i^a)-(p_d \cdot p_i)(p_u \cdot s_i^a ) \right ] \nonumber \\ & & -{\rm im}\{f_{d i}^{l\ast } l_{\tilde
d j}^{l } o^{l\ast}_{ij } \
} m_j[p_u , p_d , p_i , s_i^a ] \big\ } , \label{eq : sigmapawd } \\[2 mm ] \sigma_p^a(\tilde u_l \tilde d_l ) & = & \frac{g^4}{4 } \delta_t(\tilde u_l)\delta_u^\ast(\tilde d_l ) m_j \big\ { -{\rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } [ p_u , p_d , p_i , s_i^a ] \nonumber \\ & & + { \rm re}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } \left[(p_u \cdot p_i)(p_d \cdot s_i^a)-(p_d \cdot p_i)(p_u \cdot s_i^a ) \right ] \big\ } , \label{eq : sigmapaud } \ ] ] with the short - hand notation @xmath383 = \varepsilon_{\mu\nu\rho\sigma } \ , p_a^\mu \ , p_b^\nu \ , p_c^\rho \ , p_d^\sigma , \quad { \rm and } \quad \varepsilon_{0123}=1 .
\label{eq : epsilon}\end{aligned}\ ] ] with our choice of the spin vectors @xmath384 ( [ eq : spinvecneut ] ) for neutralino @xmath27 , @xmath385 is the longitudinal polarization of neutralino , @xmath386 is the transverse polarization in the production plane and @xmath387 is the polarization normal to the production plane . only if there are non - vanishing cp phases @xmath46 and/or @xmath388 in the chargino and neutralino sector , the polarization @xmath387 normal to the production plane is non - zero . thus it is a probe for cp violation in the production of a neutralino - chargino pair . to obtain @xmath389 for the charge conjugated process , @xmath390 , one has to change the signs of eqs .
( [ eq : sigmapaww])-([eq : sigmapaud ] ) .
the coefficients @xmath391 , which describe the polarization of the chargino @xmath392 , decompose into @xmath393 & & \sigma_p^b(w \tilde u_l)+\sigma_p^b(w \tilde d_l)+\sigma_p^b(\tilde u_l \tilde d_l ) , \label{eq : sigmachar}\end{aligned}\ ] ] with @xmath394 \nonumber \\ & & + { \rm im}\{o^l_{ij } o^{r\ast}_{ij } \ } m_i[p_u , p_d , p_j , s_j^b ] \big\ } , \label{eq : sigmapbww}\\[2 mm ] \sigma_p^b(\tilde u_l \tilde u_l ) & = & -\frac{g^4}{4}|\delta_t(\tilde u_l)|^2 |f_{u i}^l|^2 |l_{\tilde u j}^{l}|^2 m_j(p_u \cdot p_i)(p_d\cdot s_j^b ) , \label{eq : sigmapbuu } \\ [ 2 mm ] \sigma_p^b(\tilde d_l \tilde d_l ) & = & \frac{g^4}{4}|\delta_u(\tilde d_l)|^2 |f_{d i}^l|^2 |l_{\tilde d j}^{l}|^2 m_j(p_d \cdot p_i)(p_u\cdot s_j^b ) , \label{eq : sigmapbdd } \\ [ 2 mm ]
\sigma_p^b(w \tilde u_l ) & = & -\frac{\sqrt 2}{4}g^4 \delta_s(w ) \delta_t^\ast(\tilde u_l ) \big\ { 2 { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{l\ast}_{ij } \ } m_j(p_u \cdot p_i)(p_d\cdot s_j^b ) \nonumber \\ & & + { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_i \left[(p_u \cdot p_j)(p_d \cdot s_j^b)-(p_d \cdot p_j)(p_u \cdot s_j^b ) \right ] \nonumber \\ & & -{\rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_i[p_u , p_d , p_j , s_j^b ] \big\ } , \label{eq : sigmapbwu } \\[2 mm ] \sigma_p^b(w \tilde d_l ) & = & -\frac{\sqrt 2}{4}g^4 \delta_s(w ) \delta_u^\ast(\tilde d_l ) \big\ { 2 { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{r\ast}_{ij } \ } m_j(p_d \cdot p_i)(p_u \cdot s_j^b ) \nonumber \\ & & + { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_i \left[(p_d \cdot p_j)(p_u \cdot s_j^b)-(p_u \cdot p_j)(p_d \cdot s_j^b ) \right ] \nonumber \\ & & -{\rm im}\{f_{d i}^{l\ast } l_{\tilde
d j}^{l } o^{l\ast}_{ij } \ } m_i[p_u , p_d , p_j , s_j^b ] \big\ } , \label{eq : sigmapbwd } \\[2 mm ] \sigma_p^b(\tilde u_l \tilde d_l ) & = & -\frac{g^4}{4 } \delta_t(\tilde u_l)\delta_u^\ast(\tilde d_l ) m_i \big\ { -{\rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } [ p_u , p_d , p_j , s_j^b ] \nonumber \\ & & + { \rm re}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } \left[(p_d \cdot p_j)(p_u \cdot s_j^b)-(p_u \cdot p_j)(p_d \cdot s_j^b ) \right ] \big\}. \label{eq : sigmapbud}\end{aligned}\ ] ] with our choice of the chargino spin vectors @xmath395 ( [ eq : spinvecchar ] ) , @xmath396 is the longitudinal polarization of chargino @xmath392 , @xmath397 is the transverse polarization in the production plane and @xmath398 is the polarization normal to the production plane . to obtain @xmath399 for the charge conjugated process , @xmath390 , one has to change the signs of eqs .
( [ eq : sigmapbww])-([eq : sigmapbud ] ) .
the coefficients @xmath400 , which contain the spin vectors @xmath369 of the neutralino @xmath37 and @xmath371 of the chargino @xmath392 , are the spin - spin correlation terms .
they decompose into @xmath401 & & \sigma_p^{ab}(w \tilde u_l)+\sigma_p^{ab}(w \tilde d_l)+\sigma_p^{ab}(\tilde u_l \tilde d_l ) , \label{eq : sigmapab}\end{aligned}\ ] ] with @xmath402 , \label{eq : sigmapabww}\\[2 mm ] \sigma_p^{ab}(\tilde u_l \tilde u_l ) & = & -\frac{g^4}{4}|\delta_t(\tilde u_l)|^2 |f_{u i}^l|^2 |l_{\tilde u j}^{l}|^2 m_i m_j ( p_u \cdot s_i^a)(p_d\cdot s_j^b ) , \label{eq : sigmapabuu } \\ [ 2 mm ] \sigma_p^{ab}(\tilde d_l \tilde d_l ) & = & -\frac{g^4}{4}|\delta_u(\tilde d_l)|^2 |f_{d i}^l|^2 |l_{\tilde d j}^{l}|^2 m_i m_j ( p_d \cdot s_i^a)(p_u\cdot s_j^b ) , \label{eq : sigmapabdd } \\ [ 2 mm ]
\sigma_p^{ab}(w \tilde u_l ) & = & -\frac{\sqrt 2}{4}g^4 \delta_s(w ) \delta_t^\ast(\tilde u_l ) \big [ 2 { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{l\ast}_{ij } \ } m_i m_j(p_u \cdot s_i^a)(p_d\cdot s_j^b ) \nonumber \\ & & + { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } g^{ab } + { \rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } f^{ab } \big ] , \label{eq : sigmapabwu } \\[2 mm ] \sigma_p^{ab}(w \tilde d_l ) & = & \frac{\sqrt 2}{4}g^4 \delta_s(w ) \delta_u^\ast(\tilde d_l ) \big [ 2 { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{r\ast}_{ij } \ } m_i m_j(p_d \cdot s_i^a)(p_u \cdot s_j^b ) \nonumber \\ & & + { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } g^{ab } -{\rm im}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } f^{ab } \big ] , \label{eq : sigmapabwd } \\[2 mm ] \sigma_p^{ab}(\tilde u_l \tilde d_l ) & = & \frac{g^4}{4 } \delta_t(\tilde u_l)\delta_u^\ast(\tilde d_l ) \big [ { \rm re}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } g^{ab } -{\rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } f^{ab } \big ] , \label{eq : sigmapabud } \ ] ] and the short - hand notations for the products @xmath403 \nonumber \\ & & + ( p_u \cdot s_i^a ) \left [ ( p_i \cdot p_j)(p_d \cdot s_j^b ) - ( p_d \cdot p_j ) ( p_i \cdot s_j^b)\right ] \nonumber \\ & & + ( p_u \cdot p_i ) \left [ ( p_d \cdot p_j)(s_i^a \cdot s_j^b ) - ( p_j \cdot s_i^a ) ( p_d \cdot s_j^b)\right ] \nonumber \\ & & + ( p_u \cdot s_j^b ) \left [ ( p_i \cdot p_j)(p_d \cdot s_i^a ) - ( p_d \cdot p_i ) ( p_j \cdot s_i^a)\right ] \nonumber \\ & & + ( p_u \cdot p_j ) \left [ ( p_d \cdot p_i)(s_i^a \cdot s_j^b ) -(p_d \cdot s_i^a)(p_i \cdot s_j^b)\right ] , \label{eq : f1ab } \\[2 mm ] f^{ab } & = & \phantom{+ } ( p_u \cdot s_i^a)[p_d , p_i , p_j , s_j^b ] + ( p_u \cdot p_i ) [ p_d , p_j , s_i^a , s_j^b ] \nonumber \\ & & + ( p_d \cdot s_j^b)[p_u , p_i , p_j , s_i^a ] + ( p_d \cdot p_j ) [ p_u , p_i , s_i^a , s_j^b ] \label{eq : fab } \\[2 mm ] & = & \phantom{+ } ( p_d \cdot s_i^a)[p_u , p_i , p_j , s_j^b ] + ( p_d \cdot p_i)[p_u , p_j , s_i^a , s_j^b ] \nonumber \\ & & + ( p_u \cdot s_j^b)[p_d , p_i , p_j , s_i^a ] + ( p_u \cdot p_j)[p_d , p_i , s_i^a , s_j^b ] .
\label{eq : f3ab } \end{aligned}\ ] ] the terms for @xmath404 are the same for the charge conjugated process , @xmath390 .
similar to the production matrix , the matrix for neutralino decay @xmath405 , eq .
( [ eq : decayneut ] ) , can also be expanded in terms of the pauli matrices @xmath406 for the chargino decay @xmath407 , eq .
( [ eq : decaychar ] ) , we write @xmath408 the expansion coefficients are given by @xcite @xmath409 \sigma^a_{d_1 } & = & \,^{\;\,+}_{(- ) } g^2 |f^{r}_{\ell i}|^2 m_i ( s^a_i \cdot p_{\ell_1 } ) , \label{eq : sigmaad1 } \\ [ 2 mm ] d_3 & = & \frac{g^2}{2 } |v_{j1}|^2 ( m_j^2 -m_{\tilde\nu_{\ell}}^2 ) , \label{eq : d3 } \\ [ 2 mm ] \sigma^b_{d_3 } & = & \ ; \,^{\;\,-}_{(+ ) } g^2 |v_{j1}|^2 m_j ( s^b_j \cdot p_{\ell_3 } ) , \label{eq : sigmabd3}\end{aligned}\ ] ] where the sign in parenthesis in eq .
( [ eq : sigmaad1 ] ) holds for the charge conjugated process @xmath410 , and in eq .
( [ eq : sigmabd3 ] ) for the conjugated process @xmath411 . the decay factor for the subsequent slepton decay @xmath412 , is @xmath413 with all couplings as defined in appendix [ lagrangians and couplings ] .
having defined all density matrices for production and decay , the amplitude squared of the combined process of neutralino - chargino production , eqs .
( [ eq : prod ] ) , and their two - body decay chains , eqs .
( [ eq : decayneut])-([eq : decaychar ] ) , is written in the spin - density matrix formalism as @xcite @xmath414 \sum_{\lambda_i,\lambda^\prime_i,\lambda_j,\lambda^\prime_j}~ \rho_{p } ( \tilde\chi^0_i\tilde\chi^\pm_j)^ { \lambda_i\lambda^\prime_i\lambda_j\lambda^\prime_j}~ \rho_{d_1}(\tilde\chi^0_i)_{\lambda^\prime_i\lambda_i}~ d_2(\tilde\ell_r)~ \rho_{d_3}(\tilde\chi^\pm_j)_{\lambda^\prime_j\lambda_j } .
\label{eq : matrixelement}\end{aligned}\ ] ] the amplitude squared is composed of the propagators @xmath415 with mass @xmath416 and width @xmath417 of the particles . inserting the density matrices @xmath418 , eq .
( [ eq : rhop ] ) , @xmath419 , eq .
( [ eq : rhod1 ] ) , and @xmath420 , eq .
( [ eq : rhod3 ] ) , into the formula for the amplitude squared , eq . ( [ eq : matrixelement ] ) , we obtain @xmath421 \left [ p~d_1~d_3 + \sigma^a_p~\sigma^a_{d_1}~d_3 + \sigma^b_p~\sigma^b_{d_3}~d_1 + \sigma^{ab}_p~~\sigma^a_{d_1}~\sigma^b_{d_3 } \right ] d_2(\tilde\ell_r ) , \label{eq : matrixelement2}\end{aligned}\ ] ] with an implicit sum over @xmath422 and @xmath423 . the amplitude squared @xmath29 is now decomposed into an unpolarized part ( first summand ) , the spin correlations of the neutralino ( second summand ) , those of the chargino ( third summand ) , and the spin - spin correlations of the neutralino and chargino ( fourth summand ) .
we give the amplitude squared @xmath29 , eq .
( [ eq : matrixelement2 ] ) , for the process @xmath424 , followed by @xmath405 , and @xmath425 . to obtain @xmath29 for the charge conjugated process , @xmath426 , followed by @xmath405 , and @xmath427 , one has to reverse the sign of the neutralino spin correlations ( second summand ) , and the spin - spin correlations ( fourth summand ) .
the unpolarized part ( first summand ) , and the spin correlations of the chargino ( third summand ) stay the same .
for the neutralino decay into a positively charged selectron , @xmath410 , there is an additional sign change for the neutralino spin correlations ( second summand ) , and the spin - spin correlations ( fourth summand ) , due to the sign change of @xmath428 , see eq .
( [ eq : sigmaad1 ] ) . for the propagators in eq .
( [ eq : matrixelement2 ] ) , we use the narrow width approximation @xmath429 which is justified for @xmath430 , which holds in our case with @xmath431 .
note , however , that the naive @xmath432-expectation of the error can easily receive large off - shell corrections of an order of magnitude and more , in particular at threshold , or due to interferences with other resonant or non - resonant processes @xcite . the hard partonic cross section in the laboratory frame
is then obtained by integrating the squared amplitude , eq .
( [ eq : matrixelement2 ] ) , over the lorentz invariant phase space element , eq . ( [ eq : phasespace ] ) , @xmath433 finally the hadronic cross section @xmath434 in the laboratory frame is obtained by integrating the partonic cross section @xmath435 , eq .
( [ eq : sigmapartonic ] ) , over the parton distribution functions @xmath436 which depend on the factorization scale @xmath3 , and the momentum fractions @xmath437 , @xmath438 of the quarks of the proton center - of mass energy @xmath439 in the laboratory frame , such that @xmath340
. there is a combinatorial color factor of @xmath440 , since the pdfs typically sum the quark color ( @xmath186x@xmath186 ) , but here only quarks with pairing color - anticolor ( @xmath186 pairs ) contribute .
the factor of @xmath75 takes into account the two possible assignments that the quark ( antiquark ) originates form the left ( right ) incoming proton beam , and vice versa .
the factor @xmath441 takes efficiently into account the dominant qcd radiative corrections for the production cross section , with typical value @xmath442 @xcite . for simplicity however , we set @xmath443 in our numerical calculations . by using the completeness relations for the neutralino spin vectors @xmath444 the terms with products of the neutralino spin vectors in eq .
( [ eq : matrixelement2 ] ) can be written @xmath445 & & \sigma_{pd_1}(w \tilde u_l ) + \sigma_{pd_1}(w \tilde d_l ) + \sigma_{pd_1}(\tilde u_l \tilde d_l ) , \label{eq : dsigmaneut}\end{aligned}\ ] ] with @xmath446\nonumber\\[1 mm ] & & -|o^r_{ij}|^2 ( p_u \cdot p_j)\left [ ( p_d \cdot p_i ) ( p_i \cdot p_{\ell_1 } ) - m_i^2 ( p_d \cdot p_{\ell_1 } ) \right ] \nonumber\\ [ 1 mm ] & & + { \rm re}\{o^l_{ij } o^{r\ast}_{ij } \ } m_i m_j \left [ ( p_u \cdot p_i)(p_d \cdot p_{\ell_1 } ) -(p_d \cdot p_i)(p_u \cdot p_{\ell_1 } ) \right ] \nonumber \\[1 mm ] & & + { \rm im}\{o^l_{ij } o^{r\ast}_{ij } \}m_i m_j [ p_u , p_d , p_i , p_{\ell_1 } ] \big\ } , \label{eq : sigmapdaww } \\ [ 2 mm ] \sigma_{pd_1}(\tilde u_l \tilde u_l ) & = & \frac{g^6}{4}|\delta_t(\tilde u_l)|^2 |f^{r}_{\ell i}|^2|f_{u i}^l|^2 |l_{\tilde u j}^{l}|^2 ( p_d \cdot p_j ) \left [ ( p_u \cdot p_i ) ( p_i \cdot p_{\ell_1 } ) -m_i^2(p_u \cdot p_{\ell_1})\right ] , \nonumber \\
\label{eq : sigmapdauu } \\
\sigma_{pd_1}(\tilde d_l \tilde d_l ) & = & -\frac{g^6}{4}|\delta_u(\tilde d_l)|^2 |f^{r}_{\ell i}|^2|f_{d i}^l|^2 |l_{\tilde d j}^{l}|^2 ( p_u \cdot p_j ) \left[(p_d \cdot p_i ) ( p_i \cdot p_{\ell_1 } ) -m_i^2(p_d \cdot p_{\ell_1})\right ] , \nonumber \\ \label{eq : sigmapdadd } \\
\sigma_{pd_1}(w \tilde u_l ) & = & \frac{\sqrt 2}{4}g^6 \delta_s(w ) \delta_t^\ast(\tilde
u_l)|f^{r}_{\ell i}|^2 \big\ { \nonumber \\ [ 1 mm ] & & 2 { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{l\ast}_{ij } \ } ( p_d \cdot p_j ) \left[(p_u \cdot p_i ) ( p_i \cdot p_{\ell_1 } ) - m_i^2 ( p_u \cdot p_{\ell_1})\right ] \nonumber \\[1 mm ] & & + { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_i m_j \left[(p_u \cdot p_i)(p_d \cdot p_{\ell_1})-(p_d \cdot p_i)(p_u \cdot p_{\ell_1})\right ] \nonumber \\[1 mm ] & & + { \rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_i m_j[p_u , p_d , p_i , p_{\ell_1 } ] \big\ } , \label{eq : sigmapdawu } \\[2 mm ] \sigma_{pd_1}(w \tilde d_l ) & = & \frac{\sqrt 2}{4}g^6 \delta_s(w ) \delta_u^\ast(\tilde d_l ) |f^{r}_{\ell i}|^2 \big\{\nonumber \\[1 mm ] & & 2 { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{r\ast}_{ij } \}(p_u \cdot p_j ) \left[(p_d \cdot p_i ) ( p_i \cdot p_{\ell_1 } ) -m_i^2(p_d \cdot p_{\ell_1})\right ] \nonumber \\[1 mm ] & & + { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_i m_j \left[(p_d \cdot p_i)(p_u \cdot p_{\ell_1})-(p_u \cdot p_i)(p_d \cdot p_{\ell_1})\right ] \nonumber \\[1 mm ] & & + { \rm im}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_i m_j[p_u , p_d , p_i , p_{\ell_1 } ] \big\ } , \label{eq : sigmapdawd } \\[2 mm ] \sigma_{pd_1}(\tilde u_l \tilde d_l ) & = & \frac{g^6}{4 } \delta_t(\tilde u_l)\delta_u^\ast(\tilde d_l ) |f^{r}_{\ell i}|^2 m_i m_j\big\ { { \rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } [ p_u , p_d , p_i , p_{\ell_1 } ] \nonumber \\ & & + { \rm re}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } \left[(p_d \cdot p_i)(p_u \cdot p_{\ell_1})-(p_u \cdot p_i)(p_d \cdot p_{\ell_1})\right ] \big\}. \label{eq : sigmapdaud } \end{aligned}\ ] ] to obtain @xmath447 for the charge conjugated process , @xmath426 , and @xmath411 , one has to change the signs of eqs .
( [ eq : sigmapdaww])-([eq : sigmapdaud ] ) , due to the sign change of @xmath380 , see eq .
( [ eq : sigmaneut ] ) . for the neutralino decay into a positively charged selectron , @xmath410
, there is an additional sign change in eqs .
( [ eq : sigmapdaww])-([eq : sigmapdaud ] ) , due to the sign change of @xmath428 , see eq .
( [ eq : sigmaad1 ] ) .
analogously , by using the completeness relations for the chargino spin vectors @xmath448 the terms with products of the chargino spin vectors in eq .
( [ eq : matrixelement2 ] ) can be written @xmath449 & & \sigma_{pd_3}(w \tilde u_l)+ \sigma_{pd_3}(w \tilde d_l)+ \sigma_{pd_3}(\tilde u_l \tilde d_l ) , \label{eq : dsigmachar}\end{aligned}\ ] ] with @xmath450 \nonumber \\[1 mm ] & & -|o^l_{ij}|^2 ( p_u \cdot p_i ) \left[(p_d \cdot p_j ) ( p_j \cdot p_{\ell_3})-m_j^2(p_d \cdot p_{\ell_3})\right ] \nonumber \\[1 mm ] & & + { \rm re}\{o^l_{ij } o^{r\ast}_{ij } \ } m_i m_j \left[(p_u \cdot p_j ) ( p_d \cdot p_{\ell_3})-(p_d \cdot p_j)(p_u \cdot p_{\ell_3})\right ] \nonumber \\[1 mm ] & & -{\rm im}\{o^l_{ij } o^{r\ast}_{ij } \ } m_i m_j[p_u , p_d , p_j , p_{\ell_3 } ] \big\ } , \label{eq : sigmapdbww}\\[2 mm ] \sigma_{pd_3}(\tilde
u_l \tilde u_l ) & = & \frac{g^6}{4}|\delta_t(\tilde u_l)|^2 |v_{j1}|^2|f_{u i}^l|^2 |l_{\tilde u j}^{l}|^2 ( p_u \cdot p_i ) \left[(p_d \cdot p_j ) ( p_j \cdot p_{\ell_3})-m_j^2(p_d \cdot p_{\ell_3})\right ] , \nonumber \\&&\label{eq : sigmapdbuu } \\ \sigma_{pd_3}(\tilde d_l \tilde d_l ) & = & - \frac{g^6}{4}|\delta_u(\tilde d_l)|^2 |v_{j1}|^2|f_{d i}^l|^2 |l_{\tilde d j}^{l}|^2 ( p_d \cdot p_i ) \left[(p_u \cdot p_j ) ( p_j \cdot p_{\ell_3})-m_j^2(p_u \cdot p_{\ell_3})\right ] , \nonumber \\&&\label{eq : sigmapdbdd } \\ \sigma_{pd_3}(w \tilde
u_l ) & = & \frac{\sqrt 2}{4}g^6 \delta_s(w ) \delta_t^\ast(\tilde u_l ) |v_{j1}|^2\big\{\nonumber \\ & & 2 { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{l\ast}_{ij } \}(p_u \cdot p_i ) \left[(p_d \cdot p_j ) ( p_j \cdot p_{\ell_3})-m_j^2(p_d \cdot p_{\ell_3})\right ] \nonumber \\[1 mm ] & & + { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \}m_i m_j \left[(p_d \cdot p_j)(p_u \cdot p_{\ell_3})-(p_u \cdot p_j)(p_d \cdot p_{\ell_3})\right ] \nonumber \\[1 mm ] & & + { \rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_i m_j[p_u , p_d , p_j , p_{\ell_3 } ] \big\ } , \label{eq : sigmapdbwu } \\[2 mm ] \sigma_{pd_3}(w \tilde d_l ) & = & \frac{\sqrt 2}{4}g^6 \delta_s(w ) \delta_u^\ast(\tilde d_l ) |v_{j1}|^2\big\{\nonumber \\ & & 2 { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{r\ast}_{ij } \}(p_d \cdot p_i ) \left[(p_u \cdot p_j ) ( p_j \cdot p_{\ell_3})-m_j^2(p_u \cdot p_{\ell_3})\right ] \nonumber \\[1 mm ] & & + { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_i m_j \left[(p_u \cdot p_j)(p_d \cdot p_{\ell_3})-(p_d \cdot p_j)(p_u \cdot p_{\ell_3})\right ] \nonumber \\[1 mm ] & & + { \rm im}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_im_j[p_u , p_d , p_j , p_{\ell_3 } ] \big\ } , \label{eq : sigmapdbwd } \\[2 mm ] \sigma_{pd_3}(\tilde u_l \tilde d_l ) & = & \frac{g^6}{4 } \delta_t(\tilde u_l)\delta_u^\ast(\tilde d_l)|v_{j1}|^2 m_i m_j \big\ { { \rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } [ p_u , p_d , p_j , p_{\ell_3 } ] \nonumber \\[1 mm ] & & + { \rm re}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } \left[(p_u \cdot p_j)(p_d \cdot p_{\ell_3})-(p_d \cdot p_j)(p_u \cdot p_{\ell_3})\right ] \big\}. \label{eq : sigmapdbud}\end{aligned}\ ] ] the terms @xmath451 are the same for the charge conjugated process , @xmath390 , and @xmath411 , due to the sign change of both @xmath391 , see eq . ( [ eq : sigmachar ] ) , and @xmath452 , see eq .
( [ eq : sigmabd3 ] ) . finally , by using the completeness relations for both the chargino and neutralino spin vectors , eq .
( [ eq : completenessneut ] ) and eq .
( [ eq : completenesschar ] ) , respectively , the terms with products of the chargino and neutralino spin vectors in eq .
( [ eq : matrixelement2 ] ) can be written @xmath453 & & \sigma_{pd_1d_3}(w \tilde u_l)+ \sigma_{pd_1d_3}(w \tilde d_l)+ \sigma_{pd_1d_3}(\tilde u_l \tilde d_l ) , \label{eq : sigmapd1d3ab}\end{aligned}\ ] ] with @xmath454 & & + { \rm re}\{o^l_{ij } o^{r\ast}_{ij } \ } m_i m_j g_3 + { \rm im}\{o^l_{ij } o^{r\ast}_{ij } \ } m_i m_j f \big ] , \label{eq : sigmapd1d3abww}\\[2 mm ] \sigma_{pd_1d_3}(\tilde u_l \tilde u_l ) & = & \frac{g^8}{4}|\delta_t(\tilde u_l)|^2 |f^{r}_{\ell i}|^2|v_{j1}|^2|f_{u i}^l|^2 |l_{\tilde u j}^{l}|^2 g_1 , \label{eq : sigmapd1d3abuu } \\ [ 2 mm ] \sigma_{pd_1d_3}(\tilde d_l \tilde d_l ) & = & \frac{g^8}{4}|\delta_u(\tilde d_l)|^2 |f^{r}_{\ell i}|^2|v_{j1}|^2|f_{d i}^l|^2 |l_{\tilde d j}^{l}|^2 g_2 , \label{eq : sigmapd1d3abdd } \\ [ 2 mm ] \sigma_{pd_1d_3}(w \tilde u_l ) & = & \frac{\sqrt 2}{4}g^8 \delta_s(w ) \delta_t^\ast(\tilde
u_l)|f^{r}_{\ell i}|^2|v_{j1}|^2 \big [ 2 { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{l\ast}_{ij } \ } g_1 \nonumber \\ & & + { \rm re}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ }
m_i m_j g_3 + { \rm im}\{f_{u i}^l l_{\tilde u j}^{l\ast } o^{r\ast}_{ij } \ } m_i m_j f \big ] , \qquad \label{eq : sigmapd1d3abwu } \\[2 mm ] \sigma_{pd_1d_3}(w \tilde d_l ) & = & -\frac{\sqrt 2}{4}g^8 \delta_s(w ) \delta_u^\ast(\tilde d_l)|f^{r}_{\ell i}|^2|v_{j1}|^2 \big [ 2 { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{r\ast}_{ij } \ } g_2 \nonumber \\ & & + { \rm re}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_i m_j g_3 -{\rm im}\{f_{d i}^{l\ast } l_{\tilde d j}^{l } o^{l\ast}_{ij } \ } m_i m_j f \big ] , \label{eq : sigmapd1d3abwd } \\[2 mm ] \sigma_{pd_1d_3}(\tilde u_l \tilde d_l ) & = & -\frac{g^8}{4 } \delta_t(\tilde u_l)\delta_u^\ast(\tilde d_l ) |f^{r}_{\ell i}|^2|v_{j1}|^2 m_i m_j \big [ \nonumber \\[1 mm ] & & { \rm re}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } g_3 -{\rm im}\{f_{u i}^{l\ast } f_{d i}^{l\ast } l_{\tilde u j}^{l } l_{\tilde d j}^{l } \ } f \big ] , \qquad \quad \label{eq : sigmapd1d3abud}\end{aligned}\ ] ] and the short - hand notations for the kinematic functions @xmath455
\big[(p_d \cdot p_j ) ( p_j \cdot p_{\ell_3})-m_j^2(p_d \cdot p_{\ell_3})\big],\qquad \quad \label{eq : g1}\\[2 mm ] g_2 & = & \big[(p_d \cdot p_i ) ( p_i \cdot p_{\ell_1})-m_i^2(p_d \cdot p_{\ell_1})\big ] \big[(p_u \cdot p_j ) ( p_j \cdot p_{\ell_3})-m_j^2(p_u \cdot p_{\ell_3})\big],\qquad \quad \label{eq : g2}\end{aligned}\ ] ] @xmath456 \nonumber \\ & & + ( p_u \cdot p_{\ell_1 } ) \left [ ( p_i \cdot p_j)(p_d \cdot p_{\ell_3 } ) - ( p_d \cdot p_j ) ( p_i \cdot p_{\ell_3})\right ] \nonumber \\ & & + ( p_u \cdot p_i ) \left [ ( p_d \cdot p_j)(p_{\ell_1 } \cdot p_{\ell_3 } ) - ( p_j \cdot p_{\ell_1 } ) ( p_d \cdot p_{\ell_3})\right ] \nonumber \\ & & + ( p_u \cdot p_{\ell_3 } ) \left [ ( p_i \cdot p_j)(p_d \cdot p_{\ell_1 } ) - ( p_d \cdot p_i ) ( p_j \cdot p_{\ell_1})\right ] \nonumber \\ & & + ( p_u \cdot p_j ) \left [ ( p_d \cdot p_i)(p_{\ell_1 } \cdot p_{\ell_3 } ) -(p_d \cdot p_{\ell_1})(p_i \cdot p_{\ell_3})\right ] , \label{eq : f1 } \\[2 mm ] f & = & \phantom{+ } ( p_u \cdot p_{\ell_1})[p_d , p_i , p_j , p_{\ell_3 } ] + ( p_u \cdot p_i ) [ p_d , p_j , p_{\ell_1},p_{\ell_3 } ] \nonumber \\ & & + ( p_d \cdot p_{\ell_3})[p_u , p_i , p_j , p_{\ell_1 } ] + ( p_d \cdot p_j ) [ p_u , p_i , p_{\ell_1},p_{\ell_3 } ] \label{eq : f2 } \\[2 mm ] & = & \phantom{+ } ( p_d \cdot p_{\ell_1})[p_u , p_i , p_j , p_{\ell_3 } ] + ( p_d \cdot p_i)[p_u , p_j , p_{\ell_1},p_{\ell_3 } ] \nonumber \\ & & + ( p_u \cdot p_{\ell_3})[p_d , p_i , p_j , p_{\ell_1 } ] + ( p_u \cdot p_j)[p_d , p_i , p_{\ell_1},p_{\ell_3 } ] .
\label{eq : f3 } \end{aligned}\ ] ] to obtain the terms @xmath457 for the charge conjugated process , @xmath390 , and @xmath427 , one has to change the signs of eqs .
( [ eq : sigmapd1d3abww])-([eq : sigmapd1d3abud ] ) , due to the sign change of @xmath452 , see eq .
( [ eq : sigmabd3 ] ) . for the neutralino decay into a positively charged selectron , @xmath410
, there is an additional sign change in eqs .
( [ eq : sigmapd1d3abww])-([eq : sigmapd1d3abud ] ) , due to the sign change of @xmath428 , see eq .
( [ eq : sigmaad1 ] ) .
in this appendix we discuss the dependence of our asymmetries on the charges of the near and far leptons , @xmath229 and @xmath230 respectively , from the neutralino decay , and on the charge of the lepton @xmath137 from the chargino decay ( which also determines the chargino charge ) .
note first that changing the chargino charge also requires a charge conjugation of the initial state , _
i.e. _ @xmath458 , becomes @xmath459 .
since the total charge of the three leptons is easy to measure , @xmath460 production can be distinguished from its charge conjugated production , @xmath461 . due to the majorana nature of the neutralino ,
the near lepton produced in its decay chain has equal probability to have positive or negative charge , independent of the charge of the initial state .
the chargino spin correlation asymmetry depends on these charges as follows : @xmath462 & = & \mathcal{a } [ p_{\bar u},p_d , p_{\tilde \chi_j^- } , p_{\ell_3 ^ - } ] , \end{aligned}\ ] ] see the remarks regarding the sign change in the relevant contribution to the amplitude squared for the charge conjugated processes in appendix [ squared amplitude of production and decay ] , after eq .
( [ eq : sigmapdbud ] ) .
the asymmetry of the neutralino - chargino spin - spin correlations satisfies @xmath463 & = & \mathcal{a}[f(p_{\bar u } , p_d , p_{\tilde \chi_i^0 } , p_{\tilde \chi_j^- } , p_{\ell_1^+ } , p_{\ell_3 ^ -})]\nonumber \\ & = & -\mathcal{a}[f(p_{\bar u } , p_d , p_{\tilde \chi_i^0 } , p_{\tilde \chi_j^- } , p_{\ell_1 ^ - } , p_{\ell_3 ^ - } ) ] , \ ] ] see the remarks after eq .
( [ eq : f3 ] ) , and the t - odd product @xmath56 as given in eq .
( [ eq : xf2 ] ) .
for the neutralino spin correlation asymmetry we have @xmath464 & = & -\mathcal{a } [ p_u , p_{\bar d},p_{\tilde \chi_i^0 } , p_{\ell_1 ^ - } ] \nonumber \\ & = & -\mathcal{a } [ p_{\bar u},p_d , p_{\tilde \chi_i^0 } , p_{\ell_1^+ } ] \nonumber \\ & = & + \mathcal{a } [ p_{\bar u},p_d , p_{\tilde \chi_i^0 } , p_{\ell_1 ^ - } ] , \label{eq : signnspin}\end{aligned}\ ] ] see the remarks after eq .
( [ eq : sigmapdaud ] ) .
although these relations concern the asymmetries which contain the inaccessible quark momenta and neutralino and/or chargino momenta , these relations are essential to understand the properties under charge conjugation of the accessible asymmetries , where these momenta have been replaced and approximated , as discussed in section [ construct ] . as discussed in section [ neutmom ] , the neutralino momentum can not be reconstructed and thus is replaced by the momentum of the far lepton @xmath230 . as shown in section [ neutmom ] , the corresponding asymmetry depends on the charge of the near ( @xmath229 ) and far ( @xmath230 ) leptons as follows @xmath465 & = & -\mathcal{a } [ p_u , p_{\bar d},p_{\ell_1 ^ - } , p_{\ell_2^+ } ] \nonumber\\&= & + \mathcal{a } [ p_u , p_{\bar d},p_{\ell_2^+ } , p_{\ell_1 ^ - } ] \nonumber\\&= & -\mathcal{a } [ p_u , p_{\bar d},p_{\ell_2 ^ - } , p_{\ell_1^+ } ] .
\label{eq : chargen } \end{aligned}\ ] ] the sign change in the first step originates from eq .
( [ eq : sigmaad1 ] ) , and the sign change in the second and third steps is simply due to the interchange of the momenta of @xmath229 and @xmath230 in the antisymmetric epsilon product .
thus we need not determine from which vertex , near or far , the leptons @xmath229 and @xmath230 originate . instead in the epsilon product one would just group the leptons according to their charge , and use the corresponding asymmetry @xmath466&= & -\mathcal{a } [ p_u , p_{\bar d } , p_{\ell^- } , p_{\ell^+}].\end{aligned}\ ] ] the triple product asymmetry fulfills similar relations to eq .
( [ eq : chargen ] ) , concerning the interchange of the near and far leptons , and of their charges , @xmath467 which each hold for the chargino charge fixed @xmath468 or @xmath469 . thus also for the triple product asymmetry , only the charges of the two leptons from the neutralino decay have to be tagged , and for measurements one would use the triple product asymmetry @xmath470 concerning now the charged conjugated production , the neutralino spin correlation asymmetry obeys @xmath471 & = & - \mathcal{a } [ p_{\bar u } , p_{d } , p_{\ell^+ } , p_{\ell^-}].\end{aligned}\ ] ] note however , that no such simple relations for charge conjugation of the initial state hold neither for the asymmetries where the quark momenta are approximated , see eqs .
( [ eq : quarkapproxu ] ) , ( [ eq : quarkapproxd ] ) , nor for the triple product asymmetry , @xmath472&\neq & -\mathcal{a}[p_{\bar u}^{\rm aprx } , p_{d}^{\rm aprx } , p_{\ell^+ } , p_{\ell^- } ] , \\
\mathcal{a}(\mathbf{p}_{\ell^+ } , \mathbf{p}_{\ell^- } , \mathbf{p}_{\ell_3^+})&\neq & -\mathcal{a}(\mathbf{p}_{\ell^+ } , \mathbf{p}_{\ell^- } , \mathbf{p}_{\ell_3 ^ -}).\end{aligned}\ ] ] the initial partonic systems @xmath473 and @xmath474 have different boost distributions due to the pdfs , thus the efficiencies to guess the initial quark directions are different .
recall that the triple product asymmetry is not lorentz invariant . thus also here
the different boost distributions of the initial partonic systems , @xmath473 as opposed to @xmath474 , will give different values for the triple product asymmetry for @xmath475 production , and its charge conjugated process @xmath476 , in general .
moreover , as pointed out in section [ construct ] , the asymmetries @xmath477 $ ] , as well as @xmath478 , receive contributions both from the neutralino spin correlations and from the spin - spin correlations . since these two contributions react differently under charge conjugation , also the corresponding asymmetries have a different magnitude . 99 h. e. haber and g. l. kane , phys .
* 117 * , 75 ( 1985 ) ; h. p.
nilles , phys .
* 110 * ( 1984 ) 1 ; m. drees , r. godbole and p. roy , _ theory and phenomenology of sparticles _ , world scientific , singapore ( 2004 ) ; h. baer and x. tata , _ weak scale supersymmetry : from superfields to scattering events _ , cambridge , uk : univ .
( 2006 ) .
see , for example : h. e. haber , nucl .
* 62 * , 469 ( 1998 ) [ arxiv : hep - ph/9709450 ] ; t. ibrahim and p. nath , arxiv : hep - ph/0210251 ; rev .
* 80 * , 577 ( 2008 ) [ arxiv:0705.2008 [ hep - ph ] ] . for recent reviews on edm searches and constraints to the mssm phases ,
see for example , y. li , s. profumo and m. ramsey - musolf , jhep * 1008 * , 062 ( 2010 ) [ arxiv:1006.1440 [ hep - ph ] ] ; j. r. ellis , j. s. lee and a. pilaftsis , jhep * 0810 * , 049 ( 2008 ) [ arxiv:0808.1819 [ hep - ph ] ] ; m. j. ramsey - musolf and s. su , phys . rept .
* 456 * , 1 ( 2008 ) [ arxiv : hep - ph/0612057 ] ; m. pospelov and a. ritz , annals phys . *
318 * , 119 ( 2005 ) [ arxiv : hep - ph/0504231 ] .
t. ibrahim and p. nath , phys .
d * 58 * , 111301 ( 1998 ) [ erratum - ibid .
d * 60 * , 099902 ( 1999 ) ] [ arxiv : hep - ph/9807501 ] ; m. brhlik , g. j. good and g. l. kane , phys .
d * 59 * , 115004 ( 1999 ) [ arxiv : hep - ph/9810457 ] ; s. yaser ayazi and y. farzan , phys .
d * 74 * ( 2006 ) 055008 [ arxiv : hep - ph/0605272 ] .
a. bartl , w. majerotto , w. porod and d. wyler , phys .
d * 68 * , 053005 ( 2003 ) [ arxiv : hep - ph/0306050 ] .
s. a. abel , a. dedes , h. k. dreiner , jhep * 0005 * ( 2000 ) 013 .
[ hep - ph/9912429 ] ; r. m. godbole , s. pakvasa , s. d. rindani , x. tata , phys .
* d61 * ( 2000 ) 113003 .
[ hep - ph/9912315 ] ; k. choi , e. j. chun , k. hwang , phys .
* d63 * ( 2001 ) 013002 .
[ hep - ph/0004101 ] ; m. frank , h. hamidian , j. phys .
g * g24 * ( 1998 ) 2203 - 2211 .
[ hep - ph/9706510 ] .
j. a. aguilar - saavedra _ et al .
_ [ ecfa / desy lc physics working group ] , arxiv : hep - ph/0106315 ; t. abe _ et al . _
[ american linear collider working group ] , arxiv : hep - ex/0106055 ; k. abe _ et al .
_ [ acfa linear collider working group ] , arxiv : hep - ph/0109166 ; j. a. aguilar - saavedra _ et al .
_ , eur .
j. c * 46 * ( 2006 ) 43 [ arxiv : hep - ph/0511344 ] ; t. behnke _ et al .
_ [ ilc collaboration ] , arxiv:0712.2356 [ physics.ins-det ] ; j. brau _ et al .
_ [ ilc collaboration ] , arxiv:0712.1950 [ physics.acc-ph ] ; g. aarons _ et al . _ [ ilc collaboration ] , arxiv:0709.1893 [ hep - ph ] .
a. bartl , h. fraas , o. kittel and w. majerotto , phys .
d * 69 * , 035007 ( 2004 ) [ arxiv : hep - ph/0308141 ] .
a. bartl , k. hohenwarter - sodek , t. kernreiter , o. kittel and m. terwort , jhep * 0907 * , 054 ( 2009 ) [ arxiv:0905.1782 [ hep - ph ] ] . for further studies with neutralino
two - body decays at the ilc , see a. bartl , t. kernreiter and o. kittel , phys .
b * 578 * , 341 ( 2004 ) [ arxiv : hep - ph/0309340 ] ; s. y. choi , m. drees , b. gaissmaier and j. song , phys . rev .
d * 69 * , 035008 ( 2004 ) [ arxiv : hep - ph/0310284 ] ; j. a. aguilar - saavedra , nucl .
b * 697 * ( 2004 ) 207 [ arxiv : hep - ph/0404104 ] ; s. y. choi and y. g. kim , phys .
d * 69 * , 015011 ( 2004 ) [ arxiv : hep - ph/0311037 ] .
for studies with neutralino three - body decays at the ilc , see y. kizukuri and n. oshimo , phys .
b * 249 * ( 1990 ) 449 ; s. y. choi , h. s. song and w. y. song , phys .
d * 61 * , 075004 ( 2000 ) [ arxiv : hep - ph/9907474 ] ; j. a. aguilar - saavedra , phys .
b * 596 * , 247 ( 2004 ) [ arxiv : hep - ph/0403243 ] ; a. bartl , h. fraas , s. hesselbach , k. hohenwarter - sodek and g. a. moortgat - pick , jhep * 0408 * , 038 ( 2004 ) [ arxiv : hep - ph/0406190 ] ; s. y. choi , b. c. chung , j. kalinowski , y. g. kim and k. rolbiecki , eur .
j. c * 46 * , 511 ( 2006 ) [ arxiv : hep - ph/0504122 ] .
o. kittel , a. bartl , h. fraas and w. majerotto , phys .
d * 70 * , 115005 ( 2004 ) [ arxiv : hep - ph/0410054 ] .
a. bartl , k. hohenwarter - sodek , t. kernreiter , o. kittel and m. terwort , nucl .
b * 802 * , 77 ( 2008 ) [ arxiv:0802.3592 [ hep - ph ] ] . for further studies with chargino
two - body decays at the ilc , see s. y. choi , a. djouadi , m. guchait , j. kalinowski , h. s. song and p. m. zerwas , eur .
j. c * 14 * , 535 ( 2000 ) [ arxiv : hep - ph/0002033 ] ; a. bartl , h. fraas , o. kittel and w. majerotto , phys .
b * 598 * , 76 ( 2004 ) [ arxiv : hep - ph/0406309 ] ; j. a. aguilar - saavedra , nucl .
b * 717 * , 119 ( 2005 ) [ arxiv : hep - ph/0410068 ] ; h. k. dreiner , o. kittel and a. marold , phys .
d * 82 * , 116005 ( 2010 ) [ arxiv:1001.4714 [ hep - ph ] ] . for studies with chargino
three - body decays at the ilc , see y. kizukuri and n. oshimo , arxiv : hep - ph/9310224 ; a. bartl , h. fraas , s. hesselbach , k. hohenwarter - sodek , t. kernreiter and g. moortgat - pick , eur .
j. c * 51 * , 149 ( 2007 ) [ arxiv : hep - ph/0608065 ] . s. y. choi , m. drees and j. song , jhep * 0609 * , 064 ( 2006 ) [ arxiv : hep - ph/0602131 ] .
g. a. moortgat - pick _
et al . _ ,
phys . rept .
* 460 * , 131 ( 2008 ) [ arxiv : hep - ph/0507011 ] ; a. bartl , k. hohenwarter - sodek , t. kernreiter and h. rud , eur .
j. c * 36 * ( 2004 ) 515 [ arxiv : hep - ph/0403265 ] ; a. bartl , h. fraas , s. hesselbach , k. hohenwarter - sodek , t. kernreiter and g. a. moortgat - pick , jhep * 0601 * , 170 ( 2006 ) [ arxiv : hep - ph/0510029 ] ; a. bartl , h. fraas , k. hohenwarter - sodek , t. kernreiter , g. moortgat - pick and a. wagner , phys .
b * 644 * , 165 ( 2007 ) [ arxiv : hep - ph/0610431 ] ; a. bartl , k. hohenwarter - sodek , t. kernreiter and o. kittel , jhep * 0709 * , 079 ( 2007 ) [ arxiv:0706.3822 [ hep - ph ] ] .
g. aad _ et al . _
[ the atlas collaboration ] , `` expected performance of the atlas experiment - detector , trigger and physics , '' [ arxiv:0901.0512 [ hep - ex ] ] , p. 1604 and 1643 .
a. bartl , e. christova , k. hohenwarter - sodek and t. kernreiter , phys .
d * 70 * , 095007 ( 2004 ) [ arxiv : hep - ph/0409060 ] .
j. ellis , f. moortgat , g. moortgat - pick , j. m. smillie and j. tattersall , eur .
j. c * 60 * , 633 ( 2009 ) [ arxiv:0809.1607 [ hep - ph ] ] .
g. moortgat - pick , k. rolbiecki and j. tattersall , arxiv:1008.2206 [ hep - ph ] .
for a recent study of cp - violating rate asymmetries in stop production and decay at the lhc , see o. kittel and a. pilaftsis , arxiv:1108.3314 [ hep - ph ] .
a. bartl , e. christova , k. hohenwarter - sodek and t. kernreiter , jhep * 0611 * , 076 ( 2006 ) [ arxiv : hep - ph/0610234 ] ; f. f. deppisch and o. kittel , jhep * 1006 * , 067 ( 2010 ) [ arxiv:1003.5186 [ hep - ph ] ] .
h. dreiner , o. kittel , s. kulkarni and a. marold , phys .
d * 83 * , 095012 ( 2011 ) [ arxiv:1011.2449 [ hep - ph ] ] .
a. bartl , h. fraas , t. kernreiter and o. kittel , eur .
j. c * 33 * , 433 ( 2004 ) [ arxiv : hep - ph/0306304 ] ; p. langacker , g. paz , l. t. wang and i. yavin , jhep * 0707 * , 055 ( 2007 ) [ arxiv : hep - ph/0702068 ] . g. moortgat - pick , k. rolbiecki , j. tattersall and p. wienemann , jhep * 1001 * , 004 ( 2010 ) [ arxiv:0908.2631 [ hep - ph ] ] .
m. r. buckley , h. murayama , w. klemm and v. rentala , phys .
d * 78 * , 014028 ( 2008 ) [ arxiv:0711.0364 [ hep - ph ] ] .
h. baer , c. h. chen , f. paige and x. tata , phys .
d * 50 * , 4508 ( 1994 ) [ arxiv : hep - ph/9404212 ] .
z. sullivan and e. l. berger , phys .
d * 78 * , 034030 ( 2008 ) [ arxiv:0805.3720 [ hep - ph ] ] ; [ arxiv:0909.2131 [ hep - ph ] ] .
h. baer , k. hagiwara and x. tata , phys .
* 57 * ( 1986 ) 294 , and phys . rev .
* d35 * ( 1987 ) 1598 ; p. nath and r. l. arnowitt , mod .
a2 * ( 1987 ) 331 ; k. t. matchev and d. m. pierce , phys . rev .
* d60 * ( 1999 ) 075004 [ hep - ph/9904282 ] ; h. baer , m. drees , f. paige , p. quintana and x. tata , phys . rev .
* d61 * ( 2000 ) 095007 [ hep - ph/9906233 ] ; a. dedes , h. k. dreiner , u. nierste , p. richardson , [ hep - ph/0207026 ] ; h. k. dreiner , s. grab , t. stefaniak , phys .
* d84 * ( 2011 ) 015005 .
[ arxiv:1103.1883 [ hep - ph ] ] .
v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
* 95 * , 151805 ( 2005 ) [ hep - ex/0504032 ] ; t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
* 101 * , 251801 ( 2008 ) .
[ arxiv:0808.2446 [ hep - ex ] ] .
d. atwood and a. soni , phys .
d * 45 * , 2405 ( 1992 ) ; m. diehl and o. nachtmann , z. phys .
c * 62 * , 397 ( 1994 ) ; b. grzadkowski and j. f. gunion , phys .
* 350 * , 218 ( 1995 ) [ arxiv : hep - ph/9501339 ] . k. nakamura _ et al .
_ [ particle data group ] , j. phys .
g * 37 * , 075021 ( 2010 ) .
h. k. dreiner , c. hanhart , u. langenfeld and d. r. phillips , phys .
d * 68 * ( 2003 ) 055004 [ arxiv : hep - ph/0304289 ] ; h. k. dreiner , o. kittel and u. langenfeld , phys . rev .
d * 74 * ( 2006 ) 115010 [ arxiv : hep - ph/0610020 ] ; eur .
j. c * 54 * ( 2008 ) 277 [ arxiv : hep - ph/0703009 ] ; h. k. dreiner , s. heinemeyer , o. kittel , u. langenfeld , a. m. weber and g. weiglein , eur .
j. c * 62 * ( 2009 ) 547 [ arxiv:0901.3485 [ hep - ph ] ] ; h. k. dreiner , s. grab , d. koschade , m. kramer , b. oleary and u. langenfeld , phys . rev .
d * 80 * ( 2009 ) 035018 [ arxiv:0905.2051 [ hep - ph ] ] .
g. aad _ et al . _ [ atlas collaboration ] , eur .
j. c * 71 * , 1682 ( 2011 ) [ arxiv:1103.6214 [ hep - ex ] ] .
o. kittel and f. von der pahlen , phys .
d * 72 * , 095004 ( 2005 ) [ arxiv : hep - ph/0508267 ] .
j. alwall , m. herquet , f. maltoni , o. mattelaer , t. stelzer , jhep * 1106 * , 128 ( 2011 ) .
[ arxiv:1106.0522 [ hep - ph ] ] .
k. hagiwara _ et al .
_ , phys .
d * 73 * ( 2006 ) 055005 [ arxiv : hep - ph/0512260 ] ; d. berdine , n. kauer and d. rainwater , phys .
lett . * 99 * , 111601 ( 2007 ) [ arxiv : hep - ph/0703058 ] ; n. kauer , phys .
b * 649 * , 413 ( 2007 ) [ arxiv : hep - ph/0703077 ] . | in the cp - violating minimal supersymmetric standard model , we study the production of a neutralino - chargino pair at the lhc . for their decays into three leptons ,
we analyze cp asymmetries which are sensitive to the cp phases of the neutralino and chargino sector .
we present analytical formulas for the entire production and decay process , and identify the cp - violating contributions in the spin correlation terms .
this allows us to define the optimal cp asymmetries .
we present a detailed numerical analysis of the cross sections , branching ratios , and the cp observables . for light neutralinos , charginos , and squarks , the asymmetries can reach several @xmath0 .
we estimate the discovery potential for the lhc to observe cp violation in the trilepton channel .
adp-11 - 30/t752 + do - th 11/24 + arxiv:1110.6131 [ hep - ph ] + * cp asymmetries in the supersymmetric trilepton signal at the lhc * s. bornhauser _ department of physics and astronomy , university of new mexico , albuquerque , nm 87131 , usa _ _ bethe center for theoretical physics and physikalisches
institut der universitt bonn , nussallee 12 , d-53115 bonn , germany _
o. j. p. boli _ instituto de fsica , universidade de so paulo , so paulo , sp , brazil . _ j. s. kim _ fakultt fr physik , tu dortmund , otto
hahn
str . 4 , d44221 dortmund , germany , and + arc centre of excellence for particle physics at the terascale , school of chemistry and physics , university of adelaide , south australia 5005 , australia _
o. kittel _ departamento de fsica terica y del cosmos and cafpe , + universidad de granada , e-18071 granada , spain _ |
the experiment was performed at the variable energy cyclotron centre , kolkata , using @xmath1o ion beams of energies 117 , 125 , 145 and 160 mev .
self - supporting @xmath2c of thickness @xmath7 514 @xmath8g/@xmath9 was used as a target .
the @xmath0 particles were detected using a si(sb ) telescope ( @xmath7 10@xmath8 m @xmath10e , @xmath7 5 mm e ) .
the typical solid angle covered by the telescope was @xmath7 0.3 msr .
the calibration of the telescope was done using elastically scattered @xmath1o ion from a au target , as well as a @xmath11th-@xmath12 source .
inclusive energy distributions for the lcps were measured in the laboratory angular range of 11@xmath13 - 29@xmath13 , which covered the angular range @xmath7 25@xmath13 - 70@xmath13 in the center - of - mass frame .
the center - of - mass ( c.m .
) energy spectra of the @xmath0 particles at different beam energies are shown in fig .
[ fig1:epsart ] .
because of the inclusive nature of the spectra , there may be some admixture of contributions of other direct reaction mechanism ( such as pre - equilibrium emission , etc . ) to the equilibrium emission spectra at forward angles in particular . in fig .
[ fig2:epsart ] , the energy spectra obtained at lab angles 11@xmath13(@xmath14= 26@xmath13 , open triangles ) and 29@xmath13(@xmath14= 66@xmath13 , solid inverted triangles ) at @xmath15=145 mev are compared ; it is seen that the slopes of the both spectra are completely matched with each other .
it is thus clear that the effects of other direct reaction mechanisms are not significant at the measured angles .
energy spectra ( c.m . ) of @xmath0 particles obtained at different angles for different beam energies .
the symbols represent experimental data .
the dash - dot - dashed and solid lines represent cascade calculations with rldm and optimized values of spin dependent ` deformability ' parameters , respectively ( see table [ tab2:table ] ) .
all experimental and calculated spectra , starting from the lowest angle , were multiplied by 10@xmath16 , 10@xmath17 , 10@xmath13 , 10@xmath18 , 10@xmath19 , 10@xmath20 , respectively . ]
angular distributions , ( d@xmath21/d@xmath22)@xmath23 , obtained by integration of the c.m .
energy distributions for the beam energies of 117 , 125 , 145 and 160 mev , are shown in fig .
[ fig3:epsart ] as a function of @xmath14 . in all cases , the values of ( d@xmath21/d@xmath22)@xmath23 are found to be constant over the whole range of observed c.m .
so , ( d@xmath21/d@xmath24)@xmath23 is @xmath25 1/sin@xmath14 , which is characteristic of the emission from an equilibrated composite nucleus .
comparison of the energy spectra ( c.m . ) of @xmath0-particles obtained at c.m .
angles 26@xmath13(open triangles ) and 66@xmath13 ( inverted triangles ) for beam energy 145 mev . ]
the average velocities , @xmath26 of the @xmath0-particles emitted at different angles have been extracted from the average energy , @xmath27 , which has been calculated using the expression @xmath28 where the index _
i _ covers the whole energy spectrum in the laboratory .
the parallel ( @xmath29 ) and perpendicular ( _ _ v__@xmath30 ) components ( with respect to the beam direction ) of @xmath31 at each angle are plotted in fig .
[ fig4:epsart ] .
it is seen that they fall on a circle with the center at cn velocity , @xmath32 , and radius of average velocity in c.m . , @xmath33 , which implies that the average velocities ( as well as energies ) of the @xmath0 particles are independent of the c.m .
emission angles .
it again indicates that the @xmath0 particles are emitted from a fully energy equilibrated source moving with the velocity @xmath32 .
angular distribution of @xmath0 particles as a function of c.m .
angles , @xmath14 .
symbols represents the experimental data and solid lines show fit to the data obtained using ( d@xmath21/d@xmath22)@xmath23 = constant . ]
it is confirmed from the experimental results that the @xmath0-particles are emitted from a fully energy equilibrated composite which may be a cn .
so the statistical model code cascade @xcite has been used to explain the present experimental data because this code is based on the assumption that the evaporation of particles takes place from an excited cn which is in full equilibrium with respect to all degrees of freedom .
average velocity curve .
solid circles correspond to experimental data and solid lines show fit to the data obtained using the equation _ _ v__@xmath34 = @xmath35 - ( @xmath29 - @xmath32)@xmath19 where @xmath33 is the average velocity of @xmath0-particles in the c.m . frame .
the arrow indicates the position of @xmath32 . ] the spin distribution in cascade is usually derived from the known fusion cross section using the strong - absorption model wherever data are available ; otherwise , it is calculated theoretically .
the decay of the cn by particle emission is calculated in the framework of hauser - feshbach formalism @xcite .
the probability that the parent nucleus ( excitation energy @xmath36 , spin @xmath37 , parity @xmath38 ) emits a particle @xmath39 with an orbital angular momentum @xmath40 , kinetic energy @xmath41 ( @xmath41=@xmath36-@xmath42 ) , and spin @xmath43 is given by @xmath44}}}t_{l}(\epsilon_{x})d\epsilon_{x } , % \end{align } \end{split}\ ] ] where @xmath42 , @xmath45 and @xmath46 are the excitation energy , spin , and parity of the daughter nucleus , respectively , @xmath47 and @xmath48 are the level - densities of the excited parent nucleus and daughter nucleus , respectively , and , _ s = @xmath45 _ + _ s _ is the channel spin .
the transmission coefficients @xmath49(@xmath41 ) for the scattering of particle @xmath50 on the daughter nucleus ( inverse process ) are obtained by using standard optical model ( om ) potential for elastic scattering . in the om transmission coefficient calculation , the parameters have been taken from the references given in the table [ tab1:table ] .
p8 cm + + + critical angular momentum @xmath51 as in table [ tab2:table ] + diffuseness parameter @xmath52 = 1@xmath53 + + + + ( 1 ) neutrons : wilmore and hodgson @xcite .
+ ( 2 ) protons : perey and perey @xcite . +
( 3 ) @xmath0-particles : huizenga and igo @xcite .
+ ( 4 ) multiplication factor of the om radius : rfact = 1 + + + + ( 1 ) fermi - gas level - density formula with empirical parameters from dilg _
et al_. @xcite .
+ + + + ( 1 ) fermi - gas level - density formula with parameters from + ldm ( myers and swiatecki @xcite ) + ( 2 ) level - density parameter _ a _ = _ a_/8 mev@xmath17 + + + + @xmath54 = @xmath55(1+@xmath56 @xmath57+@xmath58 @xmath59 ) ,
@xmath56 , @xmath58 are given in table [ tab2:table ] .
+ + + + ( 1 ) _ e1 _
= 0.001 + ( 2 ) _ m1 _
= 0.01 + ( 3 ) _ e2 _
= 5.0 + the standard form of cascade is quite successful in explaining the lcp evaporation in the light ion induced reaction in general , where the cn is assumed to be nearly spherical .
however , in case of heavy - ion - induced reaction , there is appreciable deviation between the experimental and the predicted lcp evaporation spectra .
this deviation is attributed to the deformation of the excited compound system which is angular momentum dependent .
therefore , to explain the lcp energy spectra , the effects of the deformation of the cn should be included in the statistical model calculations .
the deformation affects the particle spectra in two ways .
first , it lowers the effective emission barrier , and second , it increases the moment of inertia .
the first effect modifies the transmission coefficients for the evaporated particles which may be taken care of by increasing the radius parameter of om potential .
however , the change in moment of inertia affects the level density and the slope of the particle spectrum .
this can be taken care of by incorporating the spin - dependent deformability parameters @xcite .
for level density calculations , excited energies have been divided into three regions .
region i ( low excitation energy , _ e _ @xmath60 3 to 4 mev ) : here , the experimentally known discrete levels are used for all the nuclei produced in the cascade . in some cases ,
known high - spin states at higher excitation energy are included as yrast levels in region ii .
region ii ( medium excitation energy , 4 mev @xmath60 _ e _ @xmath60 7.5 mev ) : analytical level density formula is used in this region .
the parameters @xmath61 and @xmath10 are deduced empirically for each nucleus from the work of vonach _ et al .
_ @xcite and dilg _
et al_. @xcite .
the excitation energy is corrected for the parity effects .
region iii ( high excitation energy , _ e _ @xmath62 e@xmath63 ) : shell effects and parity corrections are neglected in this region . the same formula is then used but with ldm parameters taken from ref .
@xcite . between regions
ii and iii , the level density parameters are interpolated linearly .
the parameters are given in table [ tab1:table ] .
the level density used in regions ii and iii for a given angular momentum _
j _ and excitation energy _
e _ is given by well - known fermi - gas expression @xcite with equidistant single - particle levels and a constant level density parameter @xmath64 : @xmath65,\end{aligned}\ ] ] where _ t _ is the thermodynamic temperature and @xmath10 is the pairing correction .
the rotational energy , @xmath66 , is expressed as , @xmath67 the effective moment of inertia , @xmath54 , is written as , @xmath68 where _
i@xmath69 _ ( @xmath70 ) is the rigid - body moment of inertia , @xmath56 and @xmath58 are deformability parameters , _
r@xmath69 _ is radius parameter , and @xmath71 is the level density parameter .
so , from the above equations , it is clear that by changing _
r@xmath69 _ , _ a _ , @xmath56 , and @xmath58 , it may be possible to reproduce the experimental spectra . by increasing _
r@xmath69 _ , both transmission coefficient and level density will be affected .
it reduces the potential barrier , leading to the increase of transmission coefficient .
simultaneously , the increase in _
r@xmath69 _ leads to increase of the _ i@xmath72 _ and _ vis - - vis _ the available phase space .
we have chosen _
= 1.29 which reproduced the lower energy part of the spectra .
a similar value was used in ref .
the level density parameter @xmath71 was taken to be _
a_/8 as in previous works @xcite.to reproduce the present experimental spectra , we have only changed the deformability parameters @xmath56 and @xmath58 , as in the previous works @xcite .
the measured energy spectra have been compared with the respective cascade @xcite calculations which have been shown in fig.1 .
the critical angular momenta , _ j@xmath73 _ , used in the calculation , are 20 , 21 , 22 , and 23(@xmath53 ) for the bombarding energies of 117 , 125 , 145 , and 160 mev , respectively @xcite .
the other input parameters are given in table [ tab1:table ] .
the dash - dot - dashed lines represent the results of the cascade calculation with the radius parameter _
r@xmath69 _ = 1.29 and the default values of the deformability parameters , @xmath74 = 3.7@xmath75 10@xmath76 and @xmath77 = 1.1@xmath75 10@xmath78 , at all beam energies which have been predicted by rotating liquid drop model @xcite .
the low - energy parts of the experimental @xmath0 particle spectra are found to match with the theoretical spectra , but higher energy parts do not . to reproduce the whole spectra
, we have followed the procedure proposed by huizenga _
the deformability parameters @xmath56 and @xmath58 have been suitably optimized to reproduce the experimental spectra , which in effect modified the phase space for statistical decay by relocation of the yrast line .
the results with these modified deformability parameters , are shown by solid lines in fig .
1 and the optimized values of deformability parameters , @xmath79 and @xmath80 , are given in table [ tab2:table ] .
cccccccc @xmath15 & _ e _ & @xmath51 & @xmath81 & @xmath74 & @xmath77 & @xmath79 & @xmath80 + + 117 & 67 & 20 & 13 & 3.7 @xmath7510@xmath76 & 1.1 @xmath7510@xmath78 & 1.9 @xmath7510@xmath82 & 2.0 @xmath7510@xmath83 + 125 & 70 & 21 & 14 & 3.7 @xmath7510@xmath76 & 1.1 @xmath7510@xmath78 & 2.1 @xmath7510@xmath82 & 2.0 @xmath7510@xmath83 + 145 & 79 & 22 & 15 & 3.7 @xmath7510@xmath76 & 1.1 @xmath7510@xmath78 & 2.3 @xmath7510@xmath82 & 2.0 @xmath7510@xmath83 + 160 & 85 & 23 & 15 & 3.7 @xmath7510@xmath76 & 1.1 @xmath7510@xmath78 & 2.5 @xmath7510@xmath82 & 2.0 @xmath7510@xmath83 +
it is apparent from table [ tab2:table ] that the deformability parameters had to be modified from the corresponding rldm values considerably to fit the experimental spectra . following the empirical procedure given in refs .
@xcite , it can be shown that the above change in the deformability parameters is indicative of substantial enhancement of the deformation of the excited composite over the corresponding ground state ( rldm ) deformation .
before we look further into the origin of such deformation , it may be worthwhile to see if there is any other alternative explanation of the observed deviation of the experimental spectra from the respective standard statistical model prediction .
though the present formalism of angular momentum dependent level density is largely successful in explaining experimental lcp spectra , the magnitude of enhancement required is quite large and lacking proper explanation @xcite .
so an alternative approach , based on frozen degrees of freedom , has been proposed , which has been shown to reproduce the data quite well @xcite . in this approach , it is assumed that the deformation of the cn is frozen during the decay , i.e. , there is no change of shape of the nascent final nucleus ; so the phase space is calculated using rldm deformation of the parent nucleus , rather than that of the usual daughter nucleus .
this indicates that the dynamical effects such as shape relaxation should be taken into account to properly understand the phenomenon of particle evaporation from a deformed cn .
the formalism of `` frozen deformation '' has been applied in the present cascade calculation . in the case of @xmath0 particle emission
, the effect of frozen deformation on the energy spectrum may be taken into account in the following way .
as the @xmath0 particles are emitted predominantly at the initial stage of the decay cascade , the deformation may be `` frozen '' at its value corresponding to the highest angular momentum that the cn may have , which is @xmath84 .
so in the present cascade calculation , the deformation has been kept fixed throughout by freezing the value of @xmath85 , which has been calculated using a fixed value of @xmath86 using eq .
[ eq : i_eff ] with @xmath56 and @xmath58 obtained from rldm ( see table [ tab2:table ] ) .
the results of the cascade calculation with frozen deformation have been shown in fig .
[ fig5:frozen ] along with the experimental data .
it is observed that the cascade prediction with frozen deformation @xcite for @xmath87 ( solid curve ) is also in fair agreement with the experimental data . to investigate
further into the nature of deformation , we have computed the values of average effective radius parameter,@xmath88 , for all sets of deformability parameters ( table [ tab2:table ] ) . following @xcite , the effective radius parameter , _ r@xmath72 _ , is defined as @xmath89 , and the average effective radius parameter is calculated using the expression , @xmath88 = @xmath90 .
then the deviation of @xmath88 from _
r@xmath69 _ is the measure of deformation , and the deviation of the same from the corresponding rldm value may be considered as an enhancement .
the values of _
r@xmath72 _ estimated in the case of ` frozen ' deformation approach varied between 1.48 ( at 117 mev ) and 1.58 ( at 160 mev ) , which are , though higher than the corresponding rldm values ( @xmath91 @xmath7 1.38 - 1.42 ) , quite lower than the values obtained with the best fitted parameter set ( @xmath92 @xmath7 1.52 - 1.66 ) . from the above discussion , it may be said that , though the present study indicates deformation of the excited composite which is higher than the corresponding rldm value , uncertainty still remains about the actual magnitude of deformation .
this limitation of the present study notwithstanding , there is , at least qualitative , indication about some enhanced deformation , which may be linked with orbiting as one of the contributing factors .
comparison of the experimental energy spectra ( c.m . ) of @xmath0-particles ( triangles ) with the same obtained by theoretical calculations for beam energy 160 mev .
solid curve represents the cascade calculations done using ` frozen ' deformation ( @xmath85 ) obtained using rldm deformability parameters for @xmath93 . ]
the energy and angular distribution of @xmath0 particles , emitted in the reactions @xmath1o ( 117 , 125 , 145 and 160 mev ) + @xmath2c , have been measured .
it has been observed from the angular distribution and average velocity plots that the @xmath0-particles are emitted from fully energy equilibrated sources , at all beam energies .
the measured energy spectra have been compared with the same predicted by the standard statistical model calculations .
it has been found that the experimental @xmath0 particle energy spectra are properly explained by cascade by optimizing the deformability parameters , which are quite different from the respective rldm values .
this appears to be indicative of appreciable change ( enhancement ) of deformation .
however , the understanding does not seem to be so simple and straightforward when one compares the above with the results of cascade calculation with frozen deformation , which has also been shown to be almost equally effective in explaining the data . in this case , the effective ` frozen ' deformation turned out to be smaller than that obtained using the optimized parameters .
so , the uncertainties about the magnitudes of the actual cn deformations notwithstanding , it can , only qualitatively , be said that equilibrium orbiting , which is similar to particle evaporation in time scale , could also be one of the contributing factors for the observed deformation .
however , the present models are too simplistic to predict the actual deformation of the cn ; more realistic event - by - event monte carlo calculations , taking into account the initial deformation , spin distribution of the cn and their subsequent evolutions , should be performed to have a proper understanding of the cn deformation .
in addition , new experimental inputs ( such as measurement of deformation from gdr studies @xcite ) are needed for more comprehensive understanding of the process .
99 d. shapira , j. l. c. ford , jr . , and j. gomez del campo , phys . rev .
c * 26 * , 2470 ( 1982 ) . c. bhattacharya , a. dey , s. kundu , k. banerjee , s. bhattacharya , s. mukhopadhyay , d. gupta , t. bhattacharjee , s. r. banerjee , s. bhattacharyya , t. rana , s. k. basu , r. saha , s. bhattacharjee , k. krishan , a. mukherjee , d. bandopadhyay , and c. beck , phys .
c * 72 * , 021601(r ) ( 2005 ) .
aparajita dey , c. bhattacharya , s. bhattacharya , s. kundu , k. banerjee , s. mukhopadhyay , d. gupta , t. bhattacharjee , s. r. banerjee , s. bhattacharyya , t. k. rana , s. k. basu , r. saha , k. krishan , a. mukherjee , d. bandopadhyay , and c. beck , phys . rev .
c * 76 * , 034608 ( 2007 ) .
w. dnnweber , a. glaesner , w. hering , d. konnerth , r. ritzka , w. trombik , j. czakanski and w. zipper , phys .
61 * , 927 ( 1988 ) .
d. shapira , r. novotny , y. d. chan , k. a. erb , j. l. c. ford , jr .
, phys . lett . *
114b * , 111 ( 1982 ) .
d. shapira , d. schull , j. l. c. ford jr .
, b. shivakumar , r. l. parks , r. a. cecil and s. t. thornton , phys .
lett . * 53 * , 1634 ( 1984 ) .
b. fornal , f. gramegna , g. prete , r. burch , g. derasmo , e. m. fiore , l. fiore , a. pantaleo , v. paticchio , g. viesti , p. blasi , m. cinausero , f. lucarelli , m. anghinolfi , p. corvisiero , m. taiuti , a. zucchiatti , p. f. bortignon , d. fabris , g. nebbia , and j. a. ruiz , phys .
c * 44 * , 2588 ( 1991 ) .
aparajita dey , s. bhattacharya , c. bhattacharya , k. banerjee , t. k. rana , s. kundu , s. mukhopadhyay , d. gupta , and r. saha , phys .
c * 74 * , 044605 ( 2006 ) .
s. kundu , a. dey , k. banerjee , t. k. rana , s. muhkopadhayay , d. gupta , r. saha , s. bhattacharya , and c. bhattacharya , phys .
c * 78 * , 044601 ( 2008 ) .
i. m. govil , j. r. huizenga , w. u. schrder , and j. tke , phys .
b * 197 * , 515 ( 1987 ) .
r. k. choudhury , p. l. gonthier , k. hagel , m. n. namboodiri , j. b. natowitz , l. adler , s. simon , s. kniffen and g. berkowitz , phys .
b * 143 * , 74 ( 1984 ) .
giovanni larana , david j. moses , winifred e. parker , morton kaplan , douglas logan , roy lacey , john m. alexander , and robert j. welberry , phys .
c * 35 * , 373 ( 1987 ) .
g. larana , r. moro , a. brondi , p. cuzzocrea , a. donofrio , e. perillo , m. romano , f. terrasi , e. vardaci , and h. dumont , phys .
c * 37 * , 1920 ( 1988 ) .
b. fornal , g. prete , g. nebbia , f. trotti , g. viesti , d. fabris , k. hagel , and j. b. natowitz , phys .
c * 37 * , 2624 ( 1988 ) .
b. fornal , f. gramegna , g. prete , g. nebbia , r. smith , g. derasmo , l. fiore , a. pantaleo , g. viesti , p. blasi , f. lucarelli , i. iori , and a. moroni , phys .
c * 41 * , 127 ( 1990 ) . c. bhattacharya , m. rousseau , c. beck , v. rauch , r. nouicer , r.m .
freeman , o. stezowski , d. mahboub , s. belhabib , a. hachem , e. martin , a. dummer , s. j. sanders , and a. szanto de toledo , nucl .
a * 654 * , 841c-844c ( 1999 ) .
d. bandyopadhyay , s. k. basu , c. bhattacharya , s. bhattacharya , k. krishan , a. chatterjee , s. kailas , a. navin , and a. srivastava , phys .
c * 59 * , 1179 ( 1999 ) .
d. bandyopadhyay , c. bhattacharya , k. krishan , s. bhattacharya , s. k. basu , a. chatterjee , s. kailas , a. shrivastava , and , k. mahata , eur .
j. a * 14 * , 53(2009 ) . winifred e. parker , morton kaplan , david j. moses , giovanni larana , douglas logan , roy lacey , john m. alexander , dalva m. decastrorizzo , paul deyoung , robert j. welberry , and john t. boger , phys .
c * 44 * , 774 ( 1991 ) .
n. n. ajitanand , g. la rana , r. lacey , david j. moses , louis c. vaz , g. f. peaslee , d. m. de castro rizzo , morton kaplan , and john m. alexander , phys .
c * 34 * , 877 ( 1986 ) .
m. kildir , g. la rana , r. moro , a. brondi , e. vardaci , a. donofrio , d. fessas , e. perillo , v. roca , m. romano , and f. terrasi , g. nebbia , g. viesti , and g. prete , phys .
c * 51 * , 1873 ( 1995 ) .
n. n. ajitanand , r. lacey , g. f. peaslee , e. duek , and john m. alexander , nucl .
instr . and meth . a * 243 * , 111 ( 1986 ) .
g. viesti , b. fornal , d. fabris , k. hagel , j. b. natowitz , g. nebbia , g. prete , and f. trotti , phys .
c * 38 * , 2640 ( 1988 ) . w. hauser and h. feshbach , phys . rev . *
87 * , 366 ( 1952 ) .
j. r. huizenga , a. n. behkami , i. m. govil , w. u. schrder , and j. toke , phys .
c * 40 * , 668 ( 1989).d . wilmore and p. e. hogson , nucl .
phys . * 55 * , 673 ( 1964 ) . c. m. perey and f. g. perey , at .
data nucl .
data tables * 17 * , 1 ( 1976 ) .
j. r. huizenga and g. igo , nucl .
* 29 * , 462 ( 1962 ) .
s. cohen , f. plasil , and w. j. swiatecki , ann . phys . * 82 * , 557 ( 1974 ) .
w. e. ormand , p. f. bortignon , a. bracco , r. a. broglia , phys .
c * 40 * , 1510 ( 1989 ) .
b. lauritzen and g. bertsch , phys . rev .
2412 ( 1989 ) , and references therein . b. fornal , g. viesti , g. nebbia , g. prete and j. b. natowitz , phys .
c * 40 * , 664 ( 1989 ) .
b. fornal , f. gramegna , g. prete , r. burch , g. derasmo , e.m .
fiore , l. fiore , a. pantaleo , v. paticchio et al .
, phys . lett . *
255b * , 325 ( 1991 ) .
deepak pandit , s. mukhopadhyay , srijit bhattacharya , surajit pal , a. de , s. bhattacharya , c. bhattacharya , k. banerjee , s. kundu , t. k. rana , a. dey , g. mukherjee , t. ghosh , d. gupta , and s. r. banerjee , phys .
c * 81 * , 061302 ( 2010 ) . | the energy spectra of the @xmath0 particles emitted in the reactions @xmath1o ( 7 - 10 mev / nucleon ) + @xmath2c have been measured in the center of mass angular range of 25@xmath3 @xmath4 70@xmath3 .
the experimental energy spectra have been compared with those obtained from the statistical model calculation with `` deformability '' parameters predicted by rotating liquid drop model ( rldm ) and also fitted the same with optimized `` deformability '' parameters , which are quite different from the respective rldm values .
the data have also been found to be explained quite well using `` frozen '' deformation approximation , where the `` deformability '' parameters have been kept fixed at rldm values of the parent nucleus throughout the decay process .
the effective radius in the latter case is smaller compared to that obtained using the optimized parameters ; however , in both cases , the deformations ( effective radii ) are larger than the corresponding rldm values .
so , considering the uncertainties in the estimation of actual compound nucleus deformations , it can , only qualitatively , be said that equilibrium orbiting , which is similar to particle evaporation in time scale , could also be one of the contributing factors for the observed deformation .
@xcite , ganes @xcite , cacarizo @xcite .
the article has been arranged as follows .
the experimental setup has been described in sec .
[ sec2:level ] .
the experimental results and analysis have been presented in sec .
[ sec3:level ] .
the details of the statistical model calculation has been given in sec .
[ sec4:level ] .
the extracted values of the deformation of @xmath5si@xmath6 and their possible implications have been discussed in sec .
[ sec4_5:discussion ] . finally , the summary and conclusion have been given in sec .
[ sec5:level ] . |
when we look at the night sky , we see that galaxies seem to be arranged in a particular way .
one might expect that galaxies would be distributed randomly , much as grains of sand would if you threw a handful across the floor , but instead , they seem to trace elegant structures ; galaxy clusters are connected to each other by long filaments , interspersed with large voids , where few or no galaxies are seen .
the drivers behind the formation of these large - scale structures have been the subject of intense study and debate for over thirty years ; how did the universe go from being smooth and homogeneous just after the big bang to the clumpy , clustered universe we see today ? at the core of current theories for the formation of these structures is the premise that the evolution of the total mass distribution is described by the gravitational collapse of primordial density fluctuations , and that this evolution is traced by the evolution of galaxies .
overdense regions , or ` haloes ' , are predicted to undergo mergers to build haloes of increasing mass , with galaxies forming from the baryonic matter in these haloes .
this framework of ` biased ' hierarchical buildup @xcite has proven to be remarkably successful in explaining several important aspects of galaxy and large - scale structure formation .
this paradigm is however not without its problems .
a good example of these problems concerns the evolution of massive ( @xmath6 ) galaxies .
we might expect that massive galaxies form slowly , with many halo mergers needed to build up the required large baryon reservoirs , and indeed some galaxies do appear to form in this way @xcite .
there is however evidence that many massive galaxies may form at high redshift @xcite and on short timescales @xcite , directly counter to early , ` naive ' model predictions .
intriguingly , recent surveys ( e.g. * ? ? ?
* ; * ? ? ?
* ) have uncovered a huge population of distant , ir bright sources that are plausible candidates for being rapidly forming , massive galaxies , however they are so numerous that even recent models have difficulty in explaining their counts , and invoke a wide variety of solutions ( e.g. @xcite ) .
it seems likely therefore that there are strong , but subtle links between distant , ir / sub - mm bright galaxies , and the formation of large - scale structures .
therefore , we need observations that relate the properties of these galaxies with the underlying dark matter distribution . motivated by this
, we have used data from the spitzer wide area infrared extragalactic survey ( swire , @xcite ) to select large samples of distant ultraluminous infrared galaxies ( ulirgs , l@xmath7l@xmath8 ) and study their clustering evolution with redshift .
we assume @xmath9 km s@xmath10 mpc@xmath10 , @xmath11 , and @xmath12 .
the results presented here were originally published in @xcite .
to select high redshift ulirgs , we use the 1.6@xmath13 m emission feature , which arises due to photospheric emission from evolved stars .
when this feature is redshifted into one of the irac channels then that channel exhibits a ` bump ' @xcite . a complete discussion of the source selection and characterization methods is given in lonsdale et al 2006 ( in preparation ) , which we summarize here .
our sources are taken from the elais n1 and elais n2 fields , and the lockman hole , covering 20.9 square degrees in total .
we first selected those sources fainter than r=22 ( vega ) , and brighter than 400@xmath13jy at 24@xmath13 m . within this set , we selected two samples that displayed a ` bump ' in the 4.5@xmath13 m and 5.8@xmath13 m channels , i.e. where @xmath14 for one sample ( the ` b2 ' sample ) and where @xmath15 for the other sample ( the ` b3 ' sample ) .
this resulted in a total of 1689 b2 sources and 1223 b3 sources .
for both samples we used hyper - z @xcite to estimate redshifts , the results from which place most of the b2 sources within @xmath0 , and most of the b3 sources within @xmath16 . from the best fits we also derived ir luminosities and power sources ; the requirement that the sources have @xmath17jy demands an ir luminosity of @xmath18l@xmath19 for all the sources , with ( most objects having ) a starburst as the dominant power source , with star formation rates of @xmath20m@xmath19yr@xmath10 .
similarly , the presence of the 1.6@xmath13 m feature demands a minimum mass of evolved stars of @xmath21m@xmath19 .
both the b2 and b3 sources are thus good candidates for being moderately massive galaxies harboring an intense , obscured starburst , making them similar in nature to both local ulirgs , and high - redshift smgs .
we measured the angular clustering of both samples using the methods described in @xcite , which we summarize here .
we found that the levels of angular clustering seen in the three fields were consistent with each other to within 0.5@xmath22 , and so combined the angular clustering measures for each sample over the three fields . to quantify the strength of clustering
, we fit both datasets with a power law , @xmath23 , where @xmath24 and @xmath25 is the clustering amplitude ; @xmath26 for the b3s and @xmath27 for the b2s . to convert these angular clustering amplitudes to spatial clustering amplitudes
, we invert limbers equation : @xmath28^{2}}{\int_{a}^{b}\left(\frac{dn}{dz}\right)^{2}e(z ) d_{\theta}^{1-\gamma}(z)f(z)(1+z)dz}\right]^{\frac{1}{\gamma}}\ ] ] where @xmath29 parametrizes the redshift evolution of @xmath30 , and : @xmath31/2 ) } , e(z ) = [ \omega_{m}(1+z)^{3 } + \omega_{\lambda}]^{\frac{1}{2}}\ ] ] we derived @xmath32 from the photometric redshift distributions @xcite . for the b2 sources
this is a gaussian centered at @xmath33 with a fwhm of @xmath34 , and for the b3 sources this is a gaussian centered at @xmath35 with a fwhm of @xmath36 .
the resulting correlation lengths are @xmath37mpc for the b2 sources , and @xmath38mpc for the b3 sources .
to place these clustering results in context , we consider two models .
the first parametrizes the spatial correlation function , @xmath39 , as a single power law in comoving coordinates , where the comoving correlation length , @xmath40 , is : here the choice of @xmath42 determines the redshift evolution @xcite .
several cases are usually quoted .
first is ` comoving clustering ' , where haloes expand with the universe , and @xmath43 ; in this case clustering remains constant .
second is the family of models for which @xmath44 , for which clustering increases with time .
examples of this family include ( a ) ` stable ' clustering , for which @xmath45 ( in this case the haloes are frozen in _ proper _ coordinates , ( b ) the predicted evolution of clustering of the overall dark matter distribution , where @xmath46 @xcite , and @xmath47 at z=0 @xcite , ( c ) ` linear ' clustering , where @xmath48 . a cautionary note to this
is that detailed interpretations of clustering evolution from these models suffer from several theoretical flaws @xcite , and so should be thought of as qualitative indicators rather than quantitative predictions .
we therefore simply use the ` stable ' and ` linear ' models as indicators of the possible range of halo clustering amplitude with redshift .
the second class of model comprises those in which comoving correlation lengths increase with _ increasing _ redshift .
these models introduce ` bias ' , @xmath49 , between the galaxies and the underlying dark matter .
an example of such models are the ` fixed mass ' models @xcite , which predict the clustering strength of haloes of a specified mass at any given redshift . in figure [ r0plot ]
we plot the @xmath42 model for dark matter , stableand ` linear ' epsilon models normalized to the b2 and b3 clustering strengths , the fixed halo mass models for halo masses of @xmath50m@xmath19 , @xmath51m@xmath19 , and @xmath52m@xmath19 , the @xmath30 values for the b2 and b3 galaxies , and the spatial correlation lengths of other galaxy populations taken from the literature . with the the uncertainties described earlier
firmly in mind , we use figure [ r0plot ] to explore the relationships between our samples , the underlying dark matter , and other galaxies . both the b2s and the b3s
are strongly clustered , with correlation lengths much higher than that predicted for the overall dm distribution .
both b2s and b3s cluster significantly more strongly than optical qsos at their respective epochs , and b3s cluster more strongly than smgs . based on the @xcite models
, then we derive _ approximate _ 1@xmath22 halo mass ranges of @xmath53m@xmath54 for the b3s , and @xmath55m@xmath56 for the b2s .
interestingly , halo masses comparable to these were recently derived for an independent sample of high - redshift ulirgs by @xcite .
the most interesting comparison is however between the two samples themselves .
the clustering evolution of qsos with redshift @xcite may mean that there is a ` minimum ' host halo mass for qso activity , below which no qso is seen , of @xmath57m@xmath19 .
the correlation lengths for the b2 and b3 samples are consistent with the same conclusion but for a @xmath58 ` minimum ' dm halo mass . taken together ,
these results imply that a minimum halo mass is a ` threshold ' factor for _ all _ forms of luminous activity in galaxies , both starbursts and agn .
it is also interesting to speculate on what the host haloes of b2 and b3 sources contain at lower and higher redshifts .
we might expect that a halo hosting a b3 source could contain an optically bright lbg at @xmath59 , followed by a b3 at @xmath60 , possibly accompanied by other ( near - ir selected ) star forming systems @xcite , before evolving to host a rich galaxy cluster at low redshifts .
the occupants of a halo hosting a b2 galaxy would however probably be different .
we would expect that such a halo could contain an smg at @xmath60 , and optically fainter lbgs at @xmath61 ( though probably not lbgs at @xmath62 ) . at lower redshifts such
a halo might host a radio - bright agn and or ero at @xmath63 , and a ( poor to rich ) cluster at @xmath64 .
we conclude that ulirgs at @xmath65 as a class likely signpost stellar buildup in galaxies in clusters at @xmath64 , with higher redshift ulirgs signposting stellar buildup in galaxies that will reside in more massive clusters at lower redshifts . | we present measurements of the spatial clustering of ultraluminous infrared galaxies in two redshift intervals , @xmath0 and @xmath1 .
both samples cluster strongly , with @xmath2mpc for the @xmath1 sample , and @xmath3mpc for the @xmath0 sample , making them among the most biased galaxies at these epochs .
these clustering amplitudes are consistent with both populations residing in dark matter haloes with masses of @xmath4 .
we infer that a minimum dark matter halo mass is an important factor for all forms of luminous , obscured activity in galaxies at @xmath5 . adopting plausible models for
the growth of dm haloes with redshift , then the haloes hosting the @xmath1 sample will likely host the richest clusters of galaxies at z=0 , whereas the haloes hosting the @xmath0 sample will likely host poor to rich clusters at z=0 . |
* 2m8 cm when a finger points at the moon , one must not mistake the finger with the moon .
old zen buddhist advice . & with four parameters i can fit an elephant and with five i can make him wiggle his trunk . attributed to j. von neuman @xcite .
+ + some thoughts and some equations transcending their greatness , also transcend their purpose . during the last quarter of the xix century ludwig boltzmann ( * ? ? ?
9 ) derived an equation which predicts the proportion @xmath0 particles , in an ensemble of @xmath1 non interacting particles , that are in a state with particle energy @xmath2 @xmath3 where @xmath4 is called the _ particle partition function _ or when dealing with molecules , the _ molecular partition function_. when only two energy levels are dealt with , the ratio of @xmath5 particles in energy level @xmath6 with @xmath7 particles in energy level @xmath8 predicted by equation [ e : boltzmann ] is @xmath9 in an extension of eq , ( [ e : boltzmann ] ) for _ degenerate _ systems ( when more then one states @xmath10 have the same energy ) a statistical weight , @xmath11 equal to the number of superimposed levels , is included .
then @xmath12 which is the boltzmann distribution law in its most general form .
the average kinetic energy is @xmath13 the middle term in eq .
( [ e : meanenerg ] ) includes the _ statistical weights _ @xmath14
which account for , so called , _ degenerate _ levels .
the molecular partition function is useful only when the system of interest can be considered to be made up of noninteracting particles , molecules with no appreciable intermolecular forces .
only then , can we define and enumerate the states of the system in either terms of quantum mechanical energy states of individual molecules , or classical positions and moments of individual molecules .
when interactions between molecules occur , the description of the states of the system must include potential energy terms , such as @xmath15 , which are functions of intermolecular distances .
equation ( [ e : boltzmann ] ) may be rewritten for ensembles of interacting particles as @xmath16 where @xmath17 , @xmath18 is the particle mass , @xmath19 is velocity and @xmath20 is potential energy . equation ( [ e : maxwboltz ] ) is the _ maxwell - boltzmann _ distribution function .
if there are only two possible states in the system eq .
( [ e : maxwboltz ] ) becomes @xmath21 when dealing with cell membranes , a boltzmann equation is expressed as free energy in voltage units ( the electrical potential difference existing across cell membranes ) in general the form used looks like @xmath22 where @xmath23 is used to more clearly specify that we deal with electrical potential differences , @xmath24 is customarily referred to as slope factor(see for example @xcite ) .
since these situations deal with ensembles of particles the slope factoris usually parametrized @xmath25 , at room temperature if : the ionic valence , _ _
z__=1 , _ f _ is the faraday constant , _
r _ is the gas constant and _ t _ is the absolute temperature . in electrophysiology , an equation of the form ( [ e : boltzvolt ] ) was introduced for the first time by @xcite ( h&h for brevity ) , and was used to describe the distribution , inside or outside axons , of hypothetical particles associated with na^+^ and k^+^ currents crossing the nerve membrane . since the @xcite work was seminal for electrophysiology , a plethora of papers have used boltzmann functions in connection with electrical properties of cells and isolated ionic channels @xcite . in electrophysiology , however equation ( [ e : boltzvolt ] ) is modified @xcite , @xmath26 represent _ cell trans membrane potential _ (
_ membrane potential _ for short ) expressed plainly as @xmath23 and since eq .
( [ e : boltzvolt ] ) takes values between 0 and 1 , @xmath26 is taken as the membrane potential where @xmath27 and is usually termed @xmath28 , eq .
( [ e : boltzvolt ] ) thus becomes @xmath29 when eq .
( [ e : boltzelect ] ) is used in the original fashion of h&h , to represent trans membrane distribution of some charged particle , @xmath30 is expressed in respect to the potential at which 50% of the particles are in one side of the membrane , and 50% is at the other side .
( [ e : boltzelect ] ) is thus reduced to a situation where a dependent variable @xmath30 may be fitted by some nonlinear optimization procedure to an independent variable @xmath23 ( usually expressed in mv ) using eq .
( [ e : boltzelect ] ) .
the optimization procedure enables to estimate the parameters @xmath31 and @xmath24 . in h&h work (
* pg 501 , eq .
1 ) , @xmath32 which is presented here with post h&h membrane potential sign conventions .
@xmath33 was estimated to be close to the resting membrane potential .
h&h also used a boltzmann function to estimate properties of hypothetical particles gating or triggering the mechanism controlling na^+^ conductance in nerve ( * ? ? ?
* part i , pp .
503504 ) the form of the boltzmann function in this case was @xmath34 where @xmath35 is trans membrane potential , @xmath36 is work done while molecule from the inside to the outside of the membrane , @xmath37 is valence of the molecule or number of positive charges on the molecule and @xmath38 is the absolute charge of an electron . with large @xmath35 , eq .
( [ e : gating1 ] ) was used in the limit form @xmath39 to estimate @xmath37 of hypothetical particles ; in ec ( [ e : gatinglim ] ) , @xmath40 is a constant . quoting ( * ? ? ?
* part i , pp .
503504 ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ whose distribution changes must bear six negative electronic charges , or , if a similar theory is developed in terms of the orientation of a long molecule with a dipole moment , it must have at least three negative charges on one end and three positive charges on the other .
a different but related approach is to suppose that sodium movement depends on the presence of six singly charged molecules at a particular site near the inside of the membrane . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ but _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ details of the mechanism will probably not be settled for some time , but it seems difficult to escape the conclusion that the changes in ionic permeability depend on the movement of some component of the membrane which behaves as though it had a large charge or dipole moment . if such components exist it is necessary to suppose that their density is relatively low and that a number of sodium ions cross the membrane at a single active patch .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in modern terms , @xmath41 expresses _ availability _ of sodium channels to be activated when the membrane depolarizes . yet
, the term _ channel _ was never used by @xcite in their famous series of papers , which never mentioned the term _ probability _ in a statistical sense . when h&h work was carried out ,
the existence of cellular plasma membranes has not been proven , and nothing was known on the nature of macromolecules associated with nerve excitation .
the @xcite model is an outstanding example of model fitting to data based on bright intuition .
it took close to 20 years for evidence to appear proving the existence of gating currents due to charge movements preceding channel activation as predicted by h&h intuition @xcite .
equations of the form of eq .
( [ e : boltzelect ] ) have become extremely popular to describe diverse biological situations @xcite , where it is referred as a boltzmann function or as a boltzmann equation .
part of the popularity is the easy availability of software which fits boltzmann functions to data , without much knowledge of the fitting procedure or the statistical properties of the parameters derived from the procedure .
the purpose of this paper is to explore the plasticity of the boltzmann function to fit data , some aspects of the optimization procedure to fit the function to data and on how to use this plastic function to differentiate the effect of a treatment ( anything that may change the system ) on data , when effect is marred by uncertainty .
to test the goodness of fitting curves to data , random data with known statistical properties were generated using monte carlo simulation @xcite . for this purpose sets of pairs @xmath42
were generated at @xmath43 ( in the fashion of ( * ? ? ?
* 7 ) ) with @xmath44 defined using @xmath45 like in eq .
( [ e : expboltz ] ) , and @xmath46 thus the random normal variate @xmath47 contains @xmath48 as @xmath49 [ eq . ( [ e : ranvar ] ) ] , the experimental error .
the uniform pseudo - random variables in the closed interval @xmath50 $ ] [ @xmath51 $ ] ] required by the @xcite algorithm to generate @xmath52 , were produced using the mersenne twister of @xcite with improved initialization ( www.math.sci.hiroshima-u.ac.jp/ m - mat / mt / mt2002/emt19937ar.html ) coded in gnu c++ ( g++ v. 4.9.2 ) , the seed required to initialize the algorithm was obtained from the ubuntu linux 15.04 entropy gathering device * _ /dev
/ random_*. a copy of the program source is provided in a digital annex to this manuscript . for the purpose of the simulations ,
initially , @xmath24 was set to 6 mv and @xmath31 was set to -40 mv , the choices are arbitrary and irrelevant , they approximate with modern conventions the values for nerve determined by h&h and may be taken as a small tribute to their monumental work after its 60^th^ anniversary . the values chosen may easily be replaced by any other as long as the variance modeled is kept .
curves were adjusted to data using a simplex minimization @xcite .
the statistical significance analysis of the differences between boltzmann curves was done using @xcite statistics as described in sections [ s : compboltz ] .
results are presented as medians and their 95% confidence interval ( ci ) determined with the hodges and lehman method .
comparisons of two samples were done with the nonparametric mann - whitney ( wilcoxon ) test ; for all nonparametric statistical procedures please refer to @xcite .
statistical differences between samples were considered significant when the probability that they stem from chance was @xmath53 ( @xmath54 ) .
all statistical procedures were carried out using the program _
est _ which is freely available in c++ source form , and compiled for linux , windows and macintosh os x from .
+ n(0,0.1)$ ] plotted vs @xmath55 . symbols in * panel b * represent data produced as @xmath56 plotted vs @xmath57 . in each case 20 sets of 10 data pairs were produced by monte carlo simulation and were fitted to eq .
( [ e : boltzelect ] ) with a simplex algorithm @xcite .
curves fitted are presented as solid lines drawn through the data points .
parameters calculated with the simplex algorithm describing the curves were : * panel a * , @xmath58 and @xmath59 , calculated for @xmath60 pairs ; * panel b * , @xmath61 and @xmath62 , calculated for @xmath60 pairs .
other details in the text of the communication.,width=453 ] to show the plasticity of the boltzmann function to fit data , 2 groups of 20 sets of data were generated using monte carlo simulation as explained before ( section [ s : montecarlo ] ) .
the first group was simulated about a boltzmann function [ eq . ( [ e : boltzelect ] ) ] with uncertainty as follows @xmath63 for a set of @xmath64 , with @xmath65 and @xmath66 ; data generated are presented as diverse symbols in fig .
[ fig : figure1]a . the solid line in fig .
[ fig : figure1]a obeys eq .
( [ e : boltzelect ] ) , with @xmath67 and @xmath68 , both values determined with the simplex algorithm for @xmath60 pairs of simulated data .
the second group was simulated about the well known michaelis - menten rectangular hyperbola plotted semi logarithmically @xcite with uncertainty added , as follows @xmath69 for a set of @xmath70 ; the results are presented as diverse symbols in fig .
[ fig : figure1]b . the solid line in fig .
[ fig : figure1]b obeys eq .
( [ e : boltzelect ] ) , with @xmath71 and @xmath72 , both values determined with the simplex algorithm for @xmath60 pairs of simulated data in the decimal semilogarithmic plane @xmath73 , where michaelis - menten data [ eq .
( [ e : michmen ] ) ] become a sigmoid function as seen in fig [ fig : figure1]b . the right model
right data relationship in the case of fig .
[ fig : figure1]a is clear ; random variables from a boltzmann process are fitted with a boltzmann equation , and the fit is quite good .
the case of fig .
[ fig : figure1]b is different , a michaelis - menten process is not a equal to a boltzmann process , and yet the boltzmann model represented by the solid line seems to fit the data in fig .
[ fig : figure1]b well , and predicts correctly several properties of the random michaelis - mented process used to generate the data , @xmath74 and @xmath75 . as mentioned before @xmath24 is the slope of the boltzmann function when @xmath76 .
the michaelis - menten equation has a more general form which is used to describe drug receptor interactions ( * ? ? ?
( viii-9 ) ) , this form is sometimes also called _ general logistic function _
32 ) @xmath77 where @xmath78 is interpreted as the number of drug molecules interacting with a receptor ( the _ molecularity _ of the reaction ) , and @xmath79 is the maximum drug effect
. for which @xmath80 } \right ] _ { x = k } = \left [ x \frac{dy}{dx}\right ] _ { x = k}=\frac{\eta \ ; y_{m}}{4}\ ] ] at the point where @xmath81 . then , with @xmath82 and @xmath83 , the following is true for the form used in eq .
( [ e : michmen ] ) and fig .
[ fig : figure1]b @xmath84 } \right ] _ { x = k } = \left [ \frac{x}{\log_{10}(e ) } \cdot \frac{dy}{dx}\right ] _ { x = k}=\frac{\eta \ ; y_{m}}{4 \ , \log_{10}(e ) } \simeq 0.5756\ldots\ ] ] the value predicted with the data in fig .
[ fig : figure1]b ( @xmath85 ) , thus underestimates the real value of @xmath86 used to calculate the data [ eq .
( [ e : decslope ] ) ] by @xmath87% .
since @xmath88 , the estimate of @xmath86 was better .
it then follows that , in spite of the apparently good fit between the data and the model in fig .
[ fig : figure1]b , eq .
( [ e : boltzelect ] ) not only lacks any mechanistic meaning regarding the data in the figure , but could lead to estimate a slope value which is wrong and , if used for this purpose , to assume a wrong molecularity for the michaelis - menten kind of reaction . ) as expressed by eq .
( [ e : boltz_pdf ] ) .
* panel b : * eq .
( [ e : boltzelect ] ) . in both cases
the functions were plotted with @xmath89 and @xmath90.,width=453 ] [ tab : samplepar ] m4.8cm*3m4 cm * parameter * & * value * & * sem * & * 95 % ci * + * sample mean * & @xmath91 & @xmath92 & @xmath93 + * theoretical mean * & 0 & @xmath94 & @xmath95 + * sample variance * & 52.4558 & + * theoretical variance * & @xmath96 & & + * sample skewness * & @xmath97 & & + * theoretical skewness * & @xmath98 & & + * sample kurtosis * & @xmath99 & & + * theoretical kurtosis * & @xmath100 & & + sem = standard error of the mean ; ci = confidence interval .
theoretical parameter values are : mean and variance are moments @xmath101 and @xmath102 in eq .
( [ e : fourmom ] ) ; skewness is @xmath103 in eq .
( [ e : skewness ] ) : kurtosis is @xmath104 in eq .
( [ e : kurtosis ] ) . two hundred thousand variables distributed as @xmath105 were generated by rearranging eq . ( [ e : boltzelect ] ) as @xmath106 where @xmath107 was a random number uniformly distributed in @xmath50 $ ] calculated with the genrand_res53 ( ) function of @xcite .
some statistical properties of this set are shown in table [ tab : samplepar ] .
the discrepancies between estimated sample parameters and the theoretical values in the table are very small , and most of them are related to the numerical approximations needed in computer simulation and the precision used to store the test file , set to 15 significant digits here ; the discrepancies are reduced when * long double * functions are used instead of * double * functions in the c++ program , for example .
table [ tab : samplepar ] , does not stress enough that boltzmann processes _ per se _ are highly stochastic independently of the inherent uncertainty of any data collecting empirical procedure ; the variability of @xmath108 depends only on @xmath109 as indicated by the 2^nd^ moment ( @xmath110 ) in equation set ( [ e : fourmom ] ) .
figure [ fig : figure3 ] simulates an experiment where 30 values of @xmath111 obeying eq .
( [ e : boltzelect ] ) were recorded and an empirical cdf ( _ edf _ ) was constructed ( stepped line ) for the set @xmath112 , in both cases @xmath113 .
the edf is a step function that jumps up by @xmath114 at each of the @xmath115 data points [ eq .
( [ e : edf ] ) ]
. the edf estimates the cdf underlying the points in the sample and converges with probability 1 according to the glivenko - cantelli @xcite theorem .
as indicated in the figure , the edf for 30 points which has no measurement uncertainty , deviates significantly from the theoretical cdf , _ due only to the stochasticity inherent to the boltzmann pdf_. the parameters for the simulations in figure [ fig : figure3 ] were chosen to mimic @xmath116 values commonly observed in electrophysiology ( * ? ? ?
* pg 501 , eq .
the maximum difference between the 2 curves in figure [ fig : figure3 ] is 0.203 ( arrow in figure [ fig : figure3 ] ) , the smirnoff test @xcite in used to compare the two curves the probability that the differences between the two cdfs in the figure stems from chance will be estimated as @xmath117 , which leads to a statistical error of type i , the incorrect rejection of a true null hypothesis . under experimental conditions , usually 15 to 20 points are recorded in a single experiment which is replicated in some 3 to 5 subjects ( or preparations ) , under conditions where preparation deterioration , recording noise or instrumental limitations add uncertainty to the variable studied . .
the stepped line is the empirical cdf calculated via monte carlo simulation for a sample of 30 points generated as @xmath118 $ ] [ eq .
( [ e : generbw ] ) in te text of this work ] , where @xmath119 $ ] a uniform random variable in the closed interval @xmath120 $ ] . in this figure @xmath113 .
arrow indicates maximum distance between empirical and boltzmann cdfs .
see the text for other details.,width=453 ]
_ determinism _ is a metaphysical philosophical position stating that for everything that happens there are conditions such that , given those conditions , nothing else could happen .
many mathematical models of physical systems are deterministic .
mathematical models which are not deterministic because they involve randomness are called _
stochastic_. even systems that involve no randomness involve _ uncertainty _ , for reasons such as : 1
. limitations of the observation instrument make the measurements fuzzy .
optical instruments are a very intuitive example of fuzziness .
actually , measuring instrument introduces fuzziness , usually called _ uncertainty _ , which is dependent on the instruments _ limit of resolution_. 2 .
members of a population are not all equal .
individual variability and biological diversity are essential to life .
3 . observing reality with a scope ( aim or purpose ) modifies the object observed @xcite .
this is specially relevant to quantum physics , but applies to any measurement ( draining current , compressing with a caliper , heating , etc . ) to , hopefully , a minor extent .
the object measured changes more or less cyclically in time .
the height of the mont blanc peak is a well known case @xcite .
the dimension of the observed object depends on the scale of measurement ( fractals ) @xcite .
coastlines , river beds or lengths , national borders , are classical examples .
the system studied is deterministic but unpredictable ( chaotic ) , due to strong nonlinearity or sensitivity to initial conditions , such as weather , population growth , seismic activity and many more @xcite . 7 .
the variables measured are any of a variety of mathematical inequalities asserting a fundamental limit to the precision with which certain pairs of physical properties of a particle , can be known simultaneously @xcite .
equations of the form of eq .
( [ e : boltzelect ] ) have become extremely popular to describe diverse biological situations @xcite .
the purpose of this paper is to explore the plasticity of the boltzmann function to fit data , some aspects of the optimization procedure to fit the function to data and on how to use this plastic function to differentiate the effect of changing system conditions , also called : _
treatment_. due to the reasons discussed in connection with eq.([e : hessmin ] ) , neither linear transformation nor nonlinear regressions are useful to estimate the uncertainty of unknown , linearly dependent , parameters @xmath121 of boltzmann functions , fitted to data ; their variances due to intra - experimental uncertainty can not be separated and can not be measured independently .
this shortcoming is more serious given that , as shown in section [ s : plasticity ] and figure [ fig : figure1 ] , boltzmann functions may be fitted to data for which the function has no mechanistic model value .
it has been said that many physicists have little knowledge of statistics.@xmath122 this mostly arises because there is little need for statistics in physics @xcite .
the main difference between physics and biology is that physics methods are applied to measure phenomena that are not too uncertain ; most physical constants are known precisely to well beyond 5 decimal places .
but as high resolution modern physics reaches its resolution limits , such as the most significant finding in recent times , the existence of the higgs boson , it depends critically on separating a small signal from its surrounding uncertainty by statistical means ( * ? ? ?
* see figure 3 ) .
variability and diversity are fundamental for preservation of life ; consequently , biological phenomena may be known only with significant uncertainty , and biological parameters can not be usually known with more that 5% accuracy .
the problem is certainly worse in biology and medicine , where experimenters may have little mathematical background , or even may be skeptic about the value of mathematics , statistics and mathematical modeling . quoting ( * ? ? ?
* how to avoid making a fool of yourself .
the role of statistics ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it is widely held by non - statisticians , like the author , that if you do good experiments statistics are not necessary .
they are quite right .
at least they are right as long as one makes an exception of the important branch of statistics that deals with processes that are _ inherently _ statistical in nature , so called _ stochastic _ processes @xmath122 .
the snag of course is that doing good experiments is difficult .
most people need all the help they can get to prevent them making a fool of themselves by claiming that their favourite theory is sustained by observations that do nothing of the sort . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in biology , a set of @xmath18 experiments is performed , then a boltzmann or other function is fitted to each experiment , and the values of the unknown model parameters ( @xmath31 and @xmath24 , for example ) obtained under different experimental conditions , are statistically compared without taking into account the intra experimental uncertainty stemming from the dispersion of data about the function fitted .
usually the experimenters are unaware that such practice is prone to result in statistical type i error , which occur when the null hypothesis is true , but is rejected ; this is asserting something that is absent , a false hit .
a type i error may be compared with a so - called false positive ( a result that indicates that a given condition is present when it actually is not present ) @xcite .
consequently , it is wise to distrust small or weakly statistically significant differences between boltzmann function parameters fitted to data under different experimental conditions .
the handicap resulting from being unable to estimate intra experimental uncertainty of unknown model parameters under , say , two experimental conditions , is most annoying when the question to answer is : is @xmath123 different from @xmath124 ? .
this is so since it the uncertainty of @xmath125 can not be separated from any other @xmath126 .
yet under the condition of _ paired comparisons _
, when _ the same system _
is studied under two experimental conditions which may be described with equations like eq .
( [ e : boltzelect ] ) , say @xmath127 and @xmath128 , the question : is @xmath127 different from @xmath128 ?
, may be answered more accurately ( with less likelihood of statistical errors of type i or ii @xcite ) .
this is so since eq .
( [ e : boltzelect ] ) is a cdf , and data described by the equation under different , _ but paired _ , experimental conditions may be used to build edfs which may be compared ( see section [ s : pdfcdf ] ) , no matter to which pdf they correspond ( section [ s : compboltz ] of this communication ) , using kolmogorov - smirnov statistics ( * ? ? ? * see ch . 6 for practical aspects ) .
if kolmogorov - smirnov statistics indicates that treatment modifies the system studied , but the uncertainty about which parameter(s ) is(are ) changing will not be dispelled .
a nonparametric statistical comparison between individual model variables , combined with the kolmogorov - smirnov test , will reduce uncertainty to assess treatment efficacy more precisely than any of the two approaches _ per se _ @xcite .
the boltzmann function in eq .
( [ e : maxwboltz ] ) through ( [ e : gating1 ] ) form , tells how likely it is that @xmath129 particles are in a given state out of @xmath130 possible states , not the probability that a given particle is in such state .
it is definitely not a pdf .
a condition that any continuous pdf , say @xmath131 , must fulfill is @xmath132 which is not the case of the boltzmann distribution function . for any continuous pdf , such as @xmath131 ,
@xmath133 exists and is a random variable @xmath51 $ ] uniformly distributed in the interval [ 0,1 ] , the so called _ standard uniform distribution _
@xmath51 $ ] , sometimes also referred to as _ rectangular distribution _
@xmath134 ( * ? ? ?
* pg , 155 ) since it has an expectation @xmath135 and and any value in the closed interval @xmath50 $ ] occurs with a probability @xmath136 . .
this is demonstrated by the following theorem ( * ? ? ?
156 , theo .
7.1.1 ) .
[ t : cdf_unif ] if @xmath55 is a random variable having a cdf @xmath137 then the random variable @xmath138 has the rectangular distribution @xmath139 .
this follows at once from the fact that the cdf of @xmath140 is @xmath141 = \begin{cases } 1 , & y > 1 \\
y , & 0 < y \leqslant 1 \\ 0 , & y \leqslant 0 \end{cases}\ ] ] which is the cdf of the rectangular distribution @xmath139 . theorem ( [ t : cdf_unif ] ) is obviously true for the boltzmann function such as eqs .
( [ e : maxwboltz ] ) through ( [ e : gating1 ] ) .
thus the boltzmann function ( @xmath142 ) is a cdf , of the pdf @xmath143 ^ 2}=\frac{1}{2 \kappa \left[1 + \cosh\left ( \frac{v - v_{\text{\textonehalf}}}{\kappa } \right ) \right]}\ ; .\ ] ] subject to necessary condition that @xmath144 , where @xmath145 is the set of real numbers , then @xmath146 and eq .
( [ e : boltz_pdf ] ) hold . the function expressed by eq .
( [ e : boltz_pdf ] ) is the boltzmann pfd .
two pdfs are equal if they have the same moment generating function ( _ mgf _ ) @xcite .
the mgf of the boltzmann pdf may be obtained as follows : @xmath147 and since the mgf is generally defined as @xmath148 then @xmath149 @xmath150 is the _
k^it^ _ _ central moment _ of @xmath36 . then , if @xmath151 the first four central moments , @xmath152 , of @xmath153 are @xmath154 as may be appreciated in fig .
[ fig : figure2]a the boltzmann pdf described by eq .
( [ e : boltz_pdf ] ) is symmetric about @xmath155 .
the skewness ( * ? ? ?
265 for details on skewness ) of @xmath156 is @xmath157 although the plot in fig .
[ fig : figure2]a evokes the gaussian bell , the distributions are different , @xmath158 has a kurtosis ( * ? ? ?
265 for details on kurtosis ) exactly equal to @xmath159 higher than the gauss pdf kurtosis which is is exactly 3 .
the _ hessian _ matrix of a function @xmath160 is a matrix of second partial derivatives of the form @xmath161\ ] ] if @xmath162 are all linearly independent , then @xmath163 is the diagonal matrix : @xmath164\ ] ] where @xmath165 are the matrix _ eigenvalues_. the hessian matrix , of eq .
( [ e : boltzelect ] ) is the non diagonal matrix : @xmath166 \\ \kappa\left[\kappa
-w\tanh \left(\xi\right)\right]&w\left[2 \kappa w \tanh \left(\xi\right)\right ] \end{matrix } \right ] \ ] ] where @xmath167 ; @xmath168 is symmetric and non singular , which indicates that @xmath121 are not linearly independent .
also @xmath169 = \dfrac{1}{4}\left [ \begin{matrix } -1&0 \\ 0&1 \end{matrix}\right ] \ ] ] which evaluating the matrix as a determinant ( called the _ discriminant _ ) , has a value of - with eigenvalues \{-1,1}. it is a negative definite matrix ; meaning that no mater what is the value of @xmath24 , the surface containing all the solutions of @xmath170 has a maximum at @xmath155 which is a _
critical point _ of @xmath171 . if @xmath172 is a set of @xmath115 random variables ordered so that @xmath173
. then @xmath174 , called by @xcite an edf , has the following properties : @xmath175 then to measure how close @xmath137 , a distribution function , is from @xmath176 @xcite defined the supremum @xmath177 where the bars indicates absolute values , _ x _ is the point where the supremum occurs . by the glivenko
cantelli theorem @xcite , if the sample comes from distribution @xmath137 , then @xmath178 converges to 0 almost surely .
@xcite strengthened this result , by effectively providing the rate of this convergence .
the following two theorems are described inspired on theo .
( 1 ) and ( 2 ) in @xcite ) .
[ t : kolmogorov ] for every fixed @xmath179 if @xmath180 denotes the probability that @xmath181 then @xmath182 where @xmath183 is the cumulative distribution function , which for @xmath184 , is given by either of the following equivalent relations @xmath185 for @xmath186 , @xmath187 .
the @xcite proof of theo . ( [ t : kolmogorov ] ) demands that @xmath188 is a uniform random variable such as @xmath51 $ ] .
equation ( [ e : cumulprob ] ) characterizes the cdf of any continuous pdf .
another theorem , due to @xcite states : [ t : smirnov ] let @xmath189 be a sets of @xmath190 mutually independent random variables ordered so that @xmath191 for which @xmath192 may be defined . then if @xmath193 define @xmath194 and suppose that @xmath195 where @xmath196 is a constant . then for every fixed @xmath197 @xmath198 theorem ( [ t : smirnov ] ) is a generalization of theo .
( [ t : kolmogorov ] ) to compare two empirical distribution functions based on @xcite statistics and is the base of the so called kolmogorov - smirnov test .
if @xmath190 points are drawn to produce a sample @xmath199 from a process described by @xmath200 are ordered as @xmath201 where all @xmath202 comply with the conditions in equation set ( [ e : edf ] ) and all have the form @xmath203 where @xmath49 is a random variable distributed as some pdf @xmath204 with expectation @xmath205 and variance @xmath206 .
the subindex @xmath207 introduced in equation ( [ e : expboltz ] ) allows for different experimental boltzmann systems , or for the same boltzmann system under different experimental condition often called _
treatments_. then , for two experimental samples we may define the set @xmath208 which has a _ supremum _
, @xmath209 the supremum of a subset @xmath210 of a totally or partially ordered set @xmath211 is the least element of @xmath211 that is greater than or equal to all elements of @xmath210 .
if the supremum exists , it is unique meaning that there will be only one supremum .
the supremum expressed by eq .
( [ e : d ] ) , may easily be evaluated numerically . in practice eqs .
( [ e : deltaminf ] ) and ( [ e : d ] ) are defined only for @xmath212 , and thus , un practice , the supremum is @xmath213 according to kolmogorov - smirnov statistics if @xmath214 , with @xmath215 , the null hypothesis may be rejected with an @xmath216 confidence level .
@xmath217 may be calculated with the algorithm of @xcite , is tabulated by @xcite and may also be expressed as @xmath218 where @xmath219 is tabulated by ( * ? ? ?
122 and table 55 ) .
this manuscript was written in latex using _ texstudio _ for linux ( also available for apple os x and ms windows , http://www.texstudio.org ) , an open source free tex editor .
51 [ 1]#1 [ 1]`#1 ` urlstyle [ 1]doi : # 1 s. peigneur , c. sevcik , j. tytgat , c. castillo , and g. dsuze .
subtype specificity interaction of bactridines with mammalian , insect and bacterial sodium channels under voltage clamp conditions .
_ febs j. _ , 2012 . in press ,
doi : 10.1111/j.1742 - 4658.2012.08808.x .
h. koegel , s. kaesler , r. burgstahler , s. werner , and c. alzheimer . unexpected down - regulation of the hik1 ca^2+^-activated k^+^ channel by its opener 1-ethyl-2- benzimidazolinone in hacat keratinocytes .
inverse effects on cell growth and proliferation .
_ j. biol .
_ , 278:0 33233330 , 2003 .
p. schuster .
boltzmann and evolution : some basic questions of biology seen with atomistic glasses . in g.
gavallotti , j. yngvasson , and w. l.reiter , editors , _ boltzmanns legacy _ , lectures in mathematics and physics , pages 217241 .
european mathematical society publishing house , zrich , switzerland , 2008 .
b. cambien , r. rezzonico , s. vitale , b. rouzaire - dubois , j .-
dubois , r. barthel , b. k. soilihi , b. mograbi , a. schmid - alliana , and h. schmid - antomarchi .
silencing of hslo potassium channels in human osteosarcoma cells promotes tumorigenesis .
j. cancer _ , 123:0 365371 , 2008 .
observation of a new boson at a mass of 125 gev with the cms experiment at the lhc .
arxiv:1207.7235.v2 , january 2013 .
european organization for nuclear research ( cern ) cern - ph - ep/2012 - 220 2013/01/29 cms - hig-12 - 028 .
e. j. arins , a. m. simonis , and j. m. van rossum .
_ molecular pharmacology .
the mode of action of biologically active compounds _ , volume 1 of _ medicinal chemistry _ , chapter drug - receptor interaction : interaction of one or more drugs with one reeptor system , pages 119286 . academic press , new york , 1964 . | sigmoid semilogarithmic functions with shape of boltzmann equations , have become extremely popular to describe diverse biological situations .
part of the popularity is due to the easy availability of software which fits boltzmann functions to data , without much knowledge of the fitting procedure or the statistical properties of the parameters derived from the procedure .
the purpose of this paper is to explore the plasticity of the boltzmann function to fit data , some aspects of the optimization procedure to fit the function to data and how to use this plastic function to differentiate the effect of treatment on data and to attest the statistical significance of treatment effect on the data .
boltzmann , energy states , probability , ion channels |
current classification schemes for the two latest spectral classes , l dwarfs @xcite and t dwarfs @xcite , are defined in different spectral regions .
l dwarf classification is tied to the red optical region ( 630010000 ) , enabling straightforward comparison with m dwarf classification schemes @xcite , and hence a well - defined demarcation of the m and l spectral classes .
t dwarf classification is tied to the near - infrared region ( 12.5 @xmath5 ) , taking advantage of both greater relative brightness ( @xmath4 75% of the emergent flux ) and the presence of strong h@xmath0o and defining ch@xmath1 bands .
classifications in different spectral regions can not be expected to be identical , however , due to the competing influences of temperature , gravity , and composition at different wavelengths .
this is particularly the case for late - type m , l , and t dwarfs , whose molecular - rich atmospheres also contain condensate clouds , predominantly affecting near - infrared spectral flux @xcite .
the influence of these chemical and physical variables muddle the characterization of the l / t transition , a spectromorphological boundary that encompasses a dramatic shift in near - infrared colors ( j k @xmath6 2 @xmath7 0 ) , condensate opacity , and overall spectral energy distribution .
clearly , the evolution of both near - infrared and optical spectra must be examined to fully characterize this critical transition .
the red optical spectra of t dwarfs are of interest in their own right , as they encompass unique features and diagnostics of physical parameters .
early investigation of the prototype of this class , gliese 229b @xcite , revealed strong h@xmath0o and weak ch@xmath1 molecular absorption ( counterparts to the strong bands that dominate the near - infrared spectrum ) , cs i alkali lines , and a steep red spectral slope shortward of 1 @xmath5 @xcite .
the latter feature , attributable to the pressure - broadened wings of the 7665/7699 k i resonant doublet lines @xcite , is responsible for the extremely red optical / near - infrared colors of this object ( r j @xmath4 9.1 ; matthews et al . 1996 , golimowski et al .
1998 ) , a property that has been exploited in the search for t dwarfs in wide - field surveys such as the two micron all sky survey ( * ? ? ?
* hereafter 2mass ) and the sloan digital sky survey ( * ? ? ?
* hereafter sdss ) .
@xcite have shown that these heavily - broadened lines and their 5890/5896 na i counterparts are sensitive to both effective temperature ( t@xmath8 ) and specific gravity .
the much weaker cs i lines are key tracers of atmospheric rainout @xcite .
additional molecular bands of feh and crh reported in the spectra of mid - type t dwarfs @xcite are possible tracers of condensate cloud opacity across the l / t transition @xcite .
finally , persistent h@xmath2 emission detected in the t dwarf 2mass 1237 + 6526 @xcite , whose atmosphere is predicted to be too cool to support a substantial chromosphere @xcite , is not yet understood and has been the subject of some debate @xcite . in this article
, we continue the spectroscopic investigation of t dwarfs initiated in ( * ? ? ?
* paper i ) by examining their red optical spectra .
this article complements that of @xcite , which discusses the optical l / t transition in detail . in @xmath92
we describe the acquisition and reduction of the data , obtained with the low resolution imaging spectrograph ( * ? ? ? * hereafter lris ) mounted on the keck i 10 m telescope . in @xmath93
we present the reduced spectra , analyzing in detail the spectral features observed . in @xmath94
we present a classification scheme for mid- and late - type t dwarfs using red optical spectra that is parallel to , but independent of , the near - infrared scheme of @xcite .
in @xmath95 we discuss our results , including the physical properties of the peculiar t dwarf 2mass 0937 + 2931 , a possible means of disentangling gravity and temperature in red optical spectra , and a comparison between t dwarf optical and near - infrared classifications .
we summarize our results in @xmath96 .
we acquired lris red optical spectra for a sample of 13 t dwarfs identified in 2mass @xcite and sdss @xcite on three nights , 2000 march 5 , 2001 february 20 , and 2003 january 2 ( ut ) .
an observing log is given in table 1 .
conditions during our march 2000 and january 2003 observations were clear with seeing @xmath10 1@xmath11 ; our february 2001 observations were made through light clouds with 1@xmath11 seeing .
lris is nominally a dual spectrograph with a 5500 dichroic mirror separating incoming light into two ( red and blue ) channels . for all observations
, we used only the red channel with a mirror bypassing the dichroic .
the og570 order - blocking filter was employed to suppress higher - order light , and a 1@xmath11 ( 4.7 pixels ) slit was used with the 400 lines mm@xmath12 grating blazed at 8500 , yielding 630010100 spectra with @xmath47 resolution ( r @xmath4 1200 ) .
dispersion on the chip was 1.9 pixel@xmath12 .
targets were nominally acquired using the facility guide camera ; objects that were too optically faint to be seen in the guider image were placed into the slit by blind offset from a nearby visible source , using offset values determined from 2mass coordinates .
individual exposures ranged from 1200 to 1800 sec , with up to four observations obtained on a particular night .
observations of the b1 v flux standard hiltner 600 @xcite were obtained for flux calibration on all runs using the same instrumental configuration . to correct for telluric absorption , we observed either da / dc / dz white dwarfs selected from @xcite or late - f / early - g stars before or after the target observations and at similar airmass .
these stars typically have few or no features in the red optical telluric absorption bands .
quartz lamp flat - field exposures ( reflected off of the interior dome ) were observed at the start of each run to calibrate detector response , and near arc lamp exposures were obtained immediately after the target observations for wavelength calibration .
science data were reduced using standard iraf routines .
first , we removed the image bias . for the march 2000 data ,
this was done by median - combining a series of one - second bias frames , and subtracting this image from the science and calibration data . for the february 2001 and january 2003 data , the bias level for each row was derived from the mean counts in the overscan region and subtracted .
images were trimmed and divided by the median - combined and normalized flat - field exposures to correct for detector response .
spectra were then extracted using the iraf apextract package .
curvature of the dispersion line was determined from either the telluric or flux standard star exposures and used as a template for the target extractions .
all spectral data were optimally extracted with a cleaning algorithm to eliminate cosmic rays and spurious pixels .
wavelength calibration was done using the arc lamp exposures and line identifications from the nist atomic line database .
we then computed a telluric correction from the telluric calibrator spectra by interpolating over atmospheric o@xmath0 ( @xmath468506900 b - band , @xmath476007700 a - band ) and h@xmath0o ( @xmath471507300 , @xmath481508350 , and @xmath489509650 ; stevenson 1994 ) bands , and then dividing this modified spectrum by the original calibrator spectrum .
this correction spectrum was applied to both target and flux standard data to eliminate telluric absorption from the spectrum .
we note that we acquired the wrong star as the telluric calibrator for 2mass 0937 + 2931 .
wd 0924 + 199 ( a.k.a pg 0924 + 199 , ton 1061 ) is classified dc5 @xcite , with a possible indication of h@xmath2 absorption and an 11,000 k blackbody continuum .
our observations show a roughly 5000 k blackbody continuum , weak and narrow h@xmath2 absorption , a strong ca ii resonance triplet ( 8498 , 8542 , & 8662 ) and a ba ii line at 6497 .
this object is likely a background f / g - type dwarf , and fortuitously works fine as a telluric calibrator . finally , flux calibration was done by correcting our hiltner 600 observations to the spectrophotometric data given by @xcite , being careful to interpolate over telluric absorption regions and the balmer h@xmath2 line .
this correction was applied to the t dwarf data to produce the final calibrated spectra .
if multiple spectra were obtained on a particular night , these were coadded with a sigma - clipping algorithm to improve signal - to - noise .
reduced spectra for the t dwarfs are shown in figure 1 in order of their near - infrared spectral classifications .
features present in the spectra are noted in figure 1 and listed in table 2 ; we discuss these features in detail below .
the strongly pressure - broadened na i and k i resonance doublets suppress most of the optical flux of t dwarfs @xcite . the broad wings of both alkali doublets can be seen clearly in the highest signal - to - noise spectrum of our sample , that of 2mass 0559@xmath31404 , shown in figure 2 with absolute flux density ( calibrated using i@xmath13 photometry and parallax measurements from dahn et al .
2002 ) plotted on a logarithmic scale .
the peak - up in flux between these features is clearly evident , and their shapes are consistent with heavily pressure - broadened atomic lines @xcite . other alkali line features present in these spectra include the 8521/8943 cs i and 7800/7948 rb i resonant doublets . examination of figure 1 shows that the cs i lines are clearly present in all of the t dwarfs observed , although they become rather weak in the latest - type objects .
rb i lines lie very close to the core of the strong k i doublet , and are generally only seen in our higher signal - to - noise spectra ( e.g. , figure 2 ) .
we have measured pseudo - equivalent widths ( pews ) for each of the cs i and rb i lines as follows : first , we subtracted a linear fit to the local pseudo - continuum , @xmath14 , to yield the line profile , @xmath15 .
next , we fit each line profile to a lorentzian function of the form @xmath16 where @xmath17 , @xmath18 ( the transition wavelength in ) , and @xmath19 ( the effective line width in ) were left as free parameters .
fits were made iteratively and confirmed by eye to eliminate spurious matches to noise spikes .
finally , pseudo - equivalent widths were derived from @xmath20 uncertainties , or upper limits in the cases where no line could be visually identified , were derived from the uncertainty in the continuum fit and residuals in a 40 region of the spectrum around each alkali line after the line profile fit . in cases where the continuum was essentially undetected
, no pew upper limits could be derived .
values are listed in table 3 .
early- and mid - type t dwarfs have cs i pews @xmath6 79 , with the exception of 2mass 0755 + 2212 , whose short wavelength spectrum is partly filled in by flux from a nearly - aligned background galaxy @xcite .
both cs i lines weaken in the later - type objects , particularly the 8521 line , which is weakest and only marginally detected in the t8 2mass 0415@xmath30935 ( [email protected] ) .
the residual strength of the 8943 line may be the result of contamination by ch@xmath1 absorption ( @xmath93.2 ; figure 4 ) .
these trends are consistent with chemical equilibrium calculations by @xcite , which predict a gradual chemical depletion of cs to cscl at t @xmath10 1300 k , accelerated at t @xmath10 1000 k as nacl ( and atomic na ) converts to na@xmath0s and liberates elemental cl .
note that k is depleted to kcl around the same temperature @xcite , weakening the k i wing and reducing the suppression of the local pseudo - continuum around the cs i lines . that the cs i lines nevertheless wane in the later t dwarfs
provides supporting evidence that elemental cs is in fact depleted in their photospheres .
the 7948 rb i line could be measured in about two - thirds of the objects in our sample .
line strengths for the early- and mid - type t dwarfs vary substantially , likely due to the decreased signal - to - noise in this spectral region .
surprisingly , many of the latest - type t dwarfs shown detectable , albeit marginal , 7948 rb i absorption .
figure 3 shows a close - up around this line for five t dwarfs with positive detections ; for 2mass 0727 + 1710 , gliese 570d , and 2mass 0415@xmath30935 , these detections are clearly marginal . nonetheless , the resurgence of the rb i line could be a real effect . as with k , atomic rb
is depleted to rbcl for t @xmath10 1000 k @xcite .
the lower depletion temperature for rb implies a much higher relative abundance than cs in the later t dwarfs , particularly as rb is roughly 20 times more abundant in a solar metallicity mixture @xcite .
furthermore , the weakening of the k i red wing is likely to enhance rb i pews .
the 7800 rb i line is only detected in our two brightest t dwarfs , 2mass 0559@xmath31404 and 2mass 1503 + 2525 , with pews of 6@xmath215 and 12@xmath215 , respectively .
one glaring absence among the alkali features is the 6708 li i resonance doublet .
2mass 0559@xmath31404 shows a possible feature close to this line ( figure 2 ) , but it is offset at 6690 and is likely a residual noise spike .
all of the objects in our sample are presumably substellar based on their low t@xmath8s @xcite , and a good fraction likely have masses below the 0.06 m@xmath22 limit for li core fusion depletion @xcite . however ,
none exhibit li i absorption .
atomic li is chemically depleted to licl , lih , and lioh for t @xmath10 1500 @xcite , at hotter temperatures than na , k , rb , and cs .
a general weakening of the li i line is observed in substellar late - type l dwarfs @xcite , and the absence of this line in the t dwarfs is consistent with a temperature effect .
however , we note that these observations are not consistent with current spectral models that use the @xcite chemical abundances , which continue to exhibit detectable li i absorption down to t@xmath8 @xmath6 600800 k @xcite .
confirmation of chemical depletion for all of the alkali species will require examination of their respective chloride bands at far - infrared wavelengths .
the most prominent molecular features in the red optical spectra of t dwarfs are the h@xmath0o bands at 9250 and 9450 .
these bands strengthen considerably toward the latest - type objects in our sample , but are generally weaker than their near - infrared counterparts .
as is the case in the near - infrared , the steam bands overlap telluric h@xmath0o absorption at 89509700 ( strongest from 93009650 ; stevenson 1994 ) ; however , the hotter brown dwarf bands clearly extend blueward to 9250 .
the 9896 wing - ford feh band is seen quite clearly in many of the spectra of figure 1 ; figure 2 of @xcite shows a close - up of this feature and the 9969 crh band .
both bands are strongest in mid - type t dwarfs like 2mass 1534@xmath32952ab , but are weak or absent in the latest - type t dwarfs .
the identification of the crh band may be in error , as @xcite find that absorption in this region for m and l dwarfs is entirely attributable to feh absorption . while opacity data from @xcite clearly show a crh bandhead at this wavelength , its visibility depends on the relative abundance of crh to feh , as their opacities are roughly equal in strength at these wavelengths @xcite .
@xcite predicts an equilibrium abundance crh / feh @xmath6 10@xmath23 for 1800 @xmath10 t @xmath10 2500 k , but a higher relative abundance of @xmath40.3 for t @xmath10 1500 k. hence , it is entirely possible that the 9969 crh band is by feh absorption in m and l dwarf spectra , but detectable in the cooler t dwarf spectra .
higher - resolution observations are required to test this possibility .
the shorter - wavelength counterparts of the feh and crh bands at 8692 and 8611 , respectively , are present in the spectrum of sdss 1254@xmath30122 , but absent in the remaining spectra .
a weak signature of cah at 67507150 , seen in m and l dwarf spectra , also appears to be present but weak in the spectra of 2mass 0559@xmath31404 ( figure 2 ) , sdss 1254@xmath30122 , and 2mass 1503 + 2525 , and is not seen in any of the other spectra .
in chemical equilibrium , cah and elemental ca are expected to be depleted to the solid minerals perovskite ( catio@xmath24 ) , grossite ( caal@xmath0o@xmath1 ) , hibonite ( caal@xmath25o@xmath26 ) , and gehlenite ( ca@xmath0al@xmath0sio@xmath27 ) at t @xmath10 1900 k @xcite .
the detection of this weak band in mid - type t dwarfs is therefore unexpected and may be indicative of non - equilibrium processes .
however , improved opacities for cah are needed before this scenario can be properly explored . while ch@xmath1 is ubiquitous in the near - infrared spectra of t dwarfs , only the relatively weak 8950 ch@xmath1 band , noted by @xcite in gliese 229b , is likely to be present in the red optical spectra of the latest - type t dwarfs .
figure 4 shows a close - up of this spectral region for six t dwarfs including gliese 229b .
absorption at the base of the 8943 cs i line clearly becomes very broad in the later - type objects , and is strongest in 2mass 0415@xmath30935 .
as shown in this figure , the breadth of this absorption is much greater than that expected for a pressure - broadened line , which exhibits a lorentzian line profile at detunings of a few 10s of angstroms @xcite .
furthermore , cs i pews are weakening in the latest - type t dwarfs , while the broad absorption in this region is clearly strengthening .
below the spectra we plot a ch@xmath1 opacity spectrum generated from the hitran database ( * ? ? ?
* see references in burrows et al .
1997 ) . while the broad trough around the cs i line , and a weak feature at 8835 in the spectrum of 2mass 0415@xmath30935 , are generally consistent with the ch@xmath1 opacity data , none of the t dwarf spectra , including that of gliese 229b , exhibit the strongest feature in the opacity data at 8875 .
however , it is important to consider that the ch@xmath1 opacity data is based on laboratory and planetary measurements made at temperatures significantly below those of a typical t dwarf photosphere ; hence , the overall band shape may be somewhat different in the latter environment . pending improved opacity data , we consider the detection of ch@xmath1 at 8950 to be tentative .
three other ch@xmath1 features at 7300 , 8000 , and 10000 seen in planetary and laboratory spectra @xcite are intrinsically weaker than the 8950 band and are not seen in the t dwarf data .
h@xmath2 emission at 6563 is detected in three objects in our sample : 2mass 1047 + 2124 , 2mass 1237 + 6526 ( previously reported in burgasser et al .
2000a ) , and sdss 1254@xmath30122 .
figure 5 shows a close - up of the h@xmath2 spectral region for these three objects .
note that the emission lines in 2mass 1047 + 2124 and sdss 1254@xmath30122 are very weak , detected at the 2.2 and 3@xmath28 levels , respectively , while the emission in 2mass 1237 + 6526 is substantial .
none of the other t dwarfs exhibit discernible emission at 6563 .
emission fluxes and 3@xmath28 upper limits are listed in table 4 , along with @xmath29 estimates , calculated from 2mass j - band magnitudes and using a linear interpolation of the j - band bolometric correction as a function of near - infrared spectral type , @xmath30 ( spt(t5 ) = 5 , etc . ) .
this relation is based on luminosity estimates for 2mass 0559@xmath31404 ( * ? ? ?
* @xmath31 ) , gliese 229b ( * ? ? ?
* @xmath32 ) , and gliese 570d ( * ? ? ?
* @xmath33 ) .
for sdss 1254@xmath30122 , we used an average @xmath34 between 2mass 0559@xmath31404 and the latest - type l dwarfs ( * ? ? ?
* @xmath35 ) .
note that pew measurements of the h@xmath2 line is prohibited in most of these objects by the absence of a detected continuum .
the detection of h@xmath2 emission in three t dwarfs appears to stand contrary to the reduced ionization fractions predicted in their very cool atmospheres @xcite .
the lack of ionized material prevents the coupling of magnetic field lines to the upper atmosphere , thereby discouraging the formation of a substantial chromosphere via collisional heating @xcite .
however , the relative emission luminosity of 2mass 1047 + 2124 and sdss 1254@xmath30122 is roughly 2 orders of magnitude lower than that of active mid - type m dwarfs ( * ? ? ?
* @xmath36 ) , and is substantially lower than that of active l dwarfs @xcite , as shown in figure 3 of @xcite .
hence , the emission in these objects is consistent with reduced , but existent , chromospheric activity . the strong , persistent emission in 2mass 1237 + 6526
is discussed in further detail in @xcite .
one other emission feature at 7740 is detected in the spectrum of 2mass 0755 + 2212 , noted in figure 1 .
as described in @xcite , this object is aligned with a background galaxy , and the observed emission line is consistent with h@xmath2 redshifted to @xmath37 .
the background galaxy also contributes continuum flux in the blue portion of the spectrum of 2mass 0755 + 2212 , filling in some of the alkali absorption features .
the spectral features described above show clear trends with near - infrared spectral type : an increasing spectral slope and strengthening h@xmath0o absorption ; a peak in the cs i line strengths in the early-/mid - type t dwarfs ; loss of feh and crh absorptions at 8692 and 8611 in the early - type t dwarfs , and at 9896 and 9969 in the late - type t dwarfs ; and the possible emergence of ch@xmath1 at 8950 in the latest - type t dwarfs . hence , the makings of a spectral sequence in the red optical that parallels the near - infrared sequence are apparent .
the low signal - to - noise in most of these spectra implies that a detailed scheme is generally not possible .
we have therefore aimed at deriving a rough classification that parallels the near - infrared scheme , enabling a general comparison between near - infrared and optical spectral morphologies .
we stress that our optical scheme is based entirely on red optical features , and that we have no _ a priori _ expectation that near - infrared and optical spectral morphologies must necessarily coincide .
we have augmented our observed spectral data with t dwarf lris data from @xcite and @xcite .
all of the spectra in our full sample have therefore been acquired and reduced in an identical manner ( with the exception of telluric absorption correction ) and can therefore be reliably compared .
our procedures for classification followed the basic tenets of the mk method @xcite .
we chose four t dwarfs
sdss 1254@xmath30122 ( t2 ) , 2mass 0559@xmath31404 ( t5 ) , sdss 1624 + 0029 ( t6 ) , and 2mass 0415@xmath30935 ( t8 ) and the l8 2mass 1632 + 1904 @xcite as our spectral standards .
three of the t dwarfs are near - infrared standards in the @xcite classification scheme , while sdss 1624 + 0029 replaces the t6 near - infrared standard 2mass 1225@xmath32739ab , which has been identified as an unequal - magnitude binary @xcite .
note that the spectra of 2mass 1632 + 1904 and sdss 1624 + 0029 have not been corrected for telluric absorption . to improve the signal - to - noise of the spectra of sdss 1254@xmath30122 and 2mass 0415@xmath30935 , we combined our data with those of @xcite .
we did not use the t1 and t3 near - infrared standards sdss 0837@xmath30000 and sdss 1021@xmath30304 because of their low signal - to - noise data .
this choice leaves sdss 1254@xmath30122 as the only early - type t dwarf standard , so classifications between l8 and t5 are not intended to be robust ; see @xcite for a more detailed discussion on the optical spectra of early - type t dwarfs .
we have nevertheless left placeholders for the unassigned intermediate subtypes ( t0 , t1 , t3 , t4 , and t7 ) to allow a consistent comparison with near - infrared classifications . to avoid confusion ,
we hereafter designate near - infrared and optical classifications as t@xmath38 and t@xmath39 , respectively .
all of the spectral standards are plotted together in figure 6 in both linear and logarithmic scales .
the trends discussed above are readily apparent over this short sequence .
more importantly , the various standards are clearly distinguishable , and hence represent distinct morphological classes .
in other words , the selected standards would have been chosen to represent separate optical classes regardless of their near - infrared types .
classification of the remaining spectra in our full sample was first done by visual comparison to the spectral standards .
we have found that the optimal means of doing this is by comparison on a logarithmic scale ( e.g. , right panel in figure 6 ) .
this method permits simultaneous examination of spectral slopes ranging over two orders of magnitude in flux density with weaker molecular and atomic features .
we have made visual classifications for all objects with lris spectra previously typed t@xmath380 or later ; values are listed in table 7 . in general , visual types for the t dwarfs correspond well with the near - infrared classifications , agreeing to within @xmath211 subtype for nearly every source .
one exception is sdss 0423@xmath30414 , which has an earlier - type spectrum than our l8 standard ; @xcite , who classify this object as l7.5 in the optical , discusses this mismatch and its possible physical basis in detail .
2mass 0937 + 2931 is assigned a peculiar visual classification because of its extreme red spectral slope , far in excess of any of the standards .
we discuss this object in further detail in @xmath95.2 . a more quantitative approach to classification is through the use of spectral indices , ratios of flux or flux density that measure the strengths of particular absorption or pseudo - continuum features .
we examined the behavior of 20 indices sampling the cs i , feh , crh , h@xmath0o , and spectral slope features on our t dwarf spectra .
some of these were taken from the literature and possibly modified @xcite , and some are of our own construction .
the most useful indices are those that show the clearest trends and greatest contrast with spectral type in the standards .
all of the indices were compared , and those that best satisfied these criteria were chosen .
eight promising indices were identified , defined in table 5 .
the cs i indices are essentially identical to those defined by @xcite , measuring the strengths of the 8521 and 8943 cs i lines , respectively .
these indices appear to peak around t@xmath392 , then weaken toward the later subclasses , consistent with the behavior of the corresponding pews .
the h@xmath0o index samples the 9250 band and may be weakly affected by telluric absorption , although it has been defined to avoid the strongest absorption longward of 9300 @xcite .
this index increases dramatically beyond roughly t@xmath392 .
the feh(a ) and crh(a ) indices measure the shorter wavelength bands of these molecules , which disappear around type t@xmath392 , after which the indices essentially measure a limited range of spectral slope .
the feh(b ) and crh(b ) indices measure the longer wavelength bands , and show rather complex behavior , weakening in the late l and early t dwarfs , strengthening again in the mid - type t dwarfs , and finally weakening in the late t dwarfs .
@xcite have proposed that this trend is attributable to the disruption of clouds across the l / t transition .
their complex behavior limits the use of the feh(b ) and crh(b ) indices to spectral types t@xmath395 and later . finally , the color - e index measures the red spectral slope due to k i absorption , but is defined at longer wavelengths than the color - d @xcite or pc3 @xcite indices because of the absence of detectable flux shortward of 8000 in many of the spectra . this index is useful earlier than roughly t@xmath392
( i.e. , around the l / t transition ) , but appears to saturate ( with considerable scatter ) in the mid- and late - type t dwarfs , possibly due to a turnover in the strength of the k i red wing ( however , see @xmath95.2 ) . taking into account similar behavior between pairs of indices and differing sensitivity for early and late t types ,
we chose to use the csi(a ) , feh(b ) , color - e , and hybrid crh(a)/h@xmath0o ratios for our classification .
these ratio values are plotted versus optical type in figure 7 .
ratio values for the five spectral standards are listed in table 6 , and values for the remaining t dwarf spectra in our sample , as well as the associated standard types for each ratio , are listed in table 7 .
final classifications were derived by averaging those ratio types ( feh(b ) was not used for types earlier than t@xmath395 , and color - e was not used for types later than t@xmath392 ) and rounding off to the nearest whole subclass .
individual ratio types for each object agree within roughly one standard class .
note that sdss 0837@xmath30000 and sdss 1021@xmath30304 are assigned uncertain classifications because of the substantial gaps between the nearest standard subclasses . with the exception of these two objects , all of our spectral ratio - derived classifications
agree with our visually - derived classifications within one subclass , implying that our suite of indices accurately represent the overall spectral morphology .
furthermore , there is excellent agreement between the optical ratio classifications and the near - infrared classifications , again with the singular exception of sdss 0423@xmath30414 .
based on these results , we conclude that classification of mid- and late - type t dwarfs in the optical is feasible , either by visual comparison to the selected standards or through the use of spectral indices defined in table 5 . however , our scheme is only accurate to within one subclass , and is therefore not as precise as current near - infrared schemes .
the behavior of the color - e and crh(a)/h@xmath0o indices also suggest that early - type t dwarfs may be delineated in the optical , suggesting that a clear segregation of l and t dwarfs at these wavelengths is possible .
optical spectral observations of additional objects in this spectral type range will have to be made , however .
we note that disagreement between the optical and near - infrared spectral morphologies of the key l / t transition object , sdss 0423@xmath30414 , does point to possible difficulties in this regime @xcite .
the only object in our sample whose red optical spectrum is truly unique in comparison to the spectral standards is the t@xmath386p / t@xmath397p 2mass 0937 + 2931 .
this object stands out in the near - infrared because of its enhanced cia h@xmath0 absorption ( resulting in a highly suppressed k - band peak ) and absence of the 1.25 @xmath5 k i lines , both of which point to a metal - poor and/or high gravity atmosphere @xcite . as noted in @xmath9 4.2 ,
the optical spectrum is peculiar because of its strong spectral slope , also a consequence of low atmospheric metallicity and/or high surface gravity @xcite . to examine the relative contribution of these secondary effects ,
we have compared the spectrum of 2mass 0937 + 2931 to theoretical models from @xcite .
these models do not incorporate the most recent line broadening theory of @xcite , nor do they include feh absorption , a relatively strong feature in the spectrum of 2mass 0937 + 2931 . nevertheless , as the current state of the art , they serve to illustrate general trends . as discussed in @xcite , increased surface gravity and decreased metallicity both result in a higher pressure photosphere , and hence enhancement of the pressure - sensitive k i and na i line wings .
this effect is strong enough to overcome the reduced column depth and decreased chemical abundance resulting from a higher gravity or lower metallicity atmosphere , respectively , which tend to weaken other chemical features , such as the 9250 h@xmath0o band .
decreased temperature also strengthens the na i and k i lines , but results in stronger h@xmath0o bands as well ( see @xmath9 5.2 ) .
when we compare these trends to the optical spectrum of 2mass 0937 + 2931 ( figure 8) , it becomes clear that the relatively weak h@xmath0o band seen in these data requires a moderately high t@xmath8 , while the strong na i and k i require both a high gravity and low metallicity atmosphere .
the best fitting models from @xcite , t@xmath8 = 800 and 1000 k , g = @xmath40 cm s@xmath41 , and @xmath42 are shown in figure 8 along with the observed spectrum .
both models show somewhat stronger inner k i wings than the data , consistent with the analysis of @xcite that the line - broadening theory used in @xcite overestimates the line opacity by as much as a factor of 10 shortward of 9000 .
nevertheless , the reasonable fit of the models to the data indicate a low metallicity , high gravity atmosphere for this object .
note that the constraint on t@xmath8 is fairly weak ; a robust estimate could be obtained by measurement of this object s parallax .
the substantial proper motion of 2mass 0937 + 2931 ( @xmath43 ; burgasser et al .
2003a ) suggests that it is either quite close to the sun or has a high space velocity .
@xcite have estimated a j - band spectrophotometric distance of roughly 9 pc , implying @xmath44 @xmath6 60 km s@xmath12 , which suggests ( but does not guarantee ) membership in the galactic thick disk or halo populations . such an object would have both a high surface gravity ( as it must be both old and therefore more massive for its temperature ) and a metal - deficient atmosphere @xcite .
@xcite have recently identified an l - type metal - deficient dwarf which , like 2mass 0937 + 2931 , exhibits strong k i and metal hydride absorption , relatively blue near - infrared colors , and substantial space motion . determining whether 2mass 0937 + 2931 is a cool analog to this substellar subdwarf requires additional astrometric data and refined spectral models .
nonetheless , the observed peculiarities in its optical and near - infrared spectra clearly indicate that 2mass 0937 + 2931 is a unique t dwarf requiring further investigation . if we ignore metallicity variations ( i.e. , excluding 2mass 0937 + 2931 ) , then differences amongst the spectra in our sample are determined largely by temperature and gravity effects .
the na i and k i resonant doublet lines , which generally dominate this spectral region , are highly sensitive to both parameters .
so too is the 9250 h@xmath0o band .
however , while the alkali lines strengthen with increasing gravity , h@xmath0o absorption decreases because of the reduced column depth of the atmosphere @xcite .
the contrasting behavior of these features suggests that they may be used to distinguish temperature and gravity effects in t dwarf spectra . as our sample likely includes objects with ages of roughly 15 gyr ( assuming they are drawn from the galactic disk population ; reid & hawley 2000 ) and 700 @xmath10 t@xmath8 @xmath10 1200 k @xcite , and hence surface gravities ( @xmath45 ) spanning the range @xmath46 cm s@xmath41 @xcite , demonstrative gravity effects may be present . as noted in @xmath9 4.3 , the color - e index ( which measure the k i red wing ) shows substantial scatter with optical spectral type for classes later than t@xmath392 ( figure 7d ) , while the combined crh(a)/h@xmath0o index is reasonably monotonic over this region .
figure 9 compares the color - e and crh(a)/h@xmath0o indices for all of our telluric - calibrated spectra .
the general trends with t@xmath8 and gravity are noted at the bottom , derived by measuring the indices on the spectral models of @xcite . for crh(a)/h@xmath0o
@xmath10 0.7 ( spt @xmath47 t@xmath392 ) , there is a clear spread in the color - e values , with 2mass 0937 + 2931 being the most significant outlier .
the simplest explanation for this divergence is intrinsic scatter in the indices , due to perhaps poor signal - to - noise in some of the spectra .
however , we note two suggestive trends .
first , the four objects with h@xmath2 emission ( including sdss 0423@xmath30414 ; kirkpatrick et al . 2003 ) generally have steeper spectral slopes ( larger color - e ) than other objects with similar crh(a)/h@xmath0o values . @xcite
have noted that for late - type m and l field dwarfs , h@xmath2 emission is generally found amongst the older , and hence more massive , objects .
if this relation follows through into the t dwarf regime , then we expect the active objects to have higher surface gravities , and therefore systematically larger color - e ratios .
second , one object in our sample has an independent age , and hence gravity , determination , the companion t@xmath388/t@xmath397 gliese 570d @xcite . at 25 gyr
, this brown dwarf has a surface gravity of ( 12)@xmath48 cm s@xmath41 @xcite .
below this object is the [email protected]/t@xmath397 2mass 1217@xmath30311 @xcite , which @xcite have suggested may be a low - gravity t dwarf based on its somewhat brighter k - band peak ( implying weaker cia h@xmath0 absorption ) .
again , the relative positions of these sources in figure 9 are consistent with the expected trends .
these examples suggest that temperature- and gravity - based features could be disentangled , enabling a means of resolving the brown dwarf age / mass / temperature degeneracy . independent age and/or gravity information ( e.g. , kinematics , companionship to well - studied stars , or cluster membership ) or improved spectral models are needed to verify and calibrate these potentially useful diagnostics .
one of our motivations for deriving a classification scheme for t dwarfs in the optical is to compare the evolution of spectral morphology between optical and near - infrared wavelengths . for this discussion ,
we focus on the mid- and late - type t dwarfs , t@xmath385/t@xmath395 and later . as discussed above , classifications for these objects in the two wavelength regimes considered here are remarkably consistent , implying a coherent evolution of features over 0.62.5 @xmath5 with spectral type
. this may be somewhat surprising , given the competing temperature and gravity dependencies of defining h@xmath0o and ch@xmath1 bands @xcite and k i and na i resonance line wings @xcite , the relatively weak gravity dependence of the optical cs i lines @xcite , and the strong gravity dependence of cia h@xmath0 in the near - infrared @xcite .
it may be the case that , despite the wide range of possible surface gravities in our sample , these particular t dwarfs have fairly similar masses and ages and hence little variation in @xmath45 .
such a case would imply that temperature is the @xmath49 discriminant in current t dwarf classification schemes .
however , the suggestive trends noted in the previous section do point to separable gravity / temperature diagnostics being present within our sample , and hence a non - negligible range of surface gravities .
this possibility leads us to consider that competing gravity and temperature effects may compensate for each other amongst all of the optical and near - infrared diagnostic spectral features , and that t subtypes in either regime may not in fact map directly onto a temperature scale .
the latter point is an important problem , as classifications are commonly used as proxies for t@xmath8 and luminosity in population studies .
our results indicate that comparing optical and near - infrared morphologies can not , at the precision of our near - infrared and optical classifications , resolve dependencies between gravity and temperature in the spectra of mid- and late - type t dwarfs .
however , developing the spectral diagnostics described in previous section may lead to a two - dimensional system that could be tied directly to temperature and gravity scales .
the agreement between optical and near - infrared classifications for mid- and late - type t dwarfs also suggests that two possible causes for disagreements amongst late - type l and early - type t dwarf classifications condensate cloud opacity and/or duplicity @xcite may not be important for later - type t dwarfs .
in general , t dwarf spectra and photometry are well - matched to models without condensate cloud opacity ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* however , see marley et al .
2002 ; burrows et al .
2002a ) , although low - level near - infrared photometric @xcite and spectroscopic @xcite variability , attributed to condensate clouds , has been observed in some t dwarfs .
these time - dependent variations are typically quite small , however , and do not significantly change the overall spectral morphology @xcite . as for duplicity ,
later - type t dwarf doubles are not affected by the substantial redistribution of spectral flux that occurs across the l / t transition @xcite , which allows a late - type l dwarf to dominate optical emission while its early - type t dwarf companion dominates near - infrared emission .
this scenario may explain the presence of ch@xmath1 at 2.2 @xmath5 in the l6.5 unequal - magnitude binary 2mass 0920 + 3517ab @xcite , as well as the near - infrared / optical classification discrepancy for sdss 0423@xmath30414 @xcite .
in contrast , both of the known binaries in our sample , 2mass 1225@xmath32739ab and 2mass 1534@xmath32952ab @xcite , have essentially identical optical and near - infrared types .
it is important to consider , however , that a larger , and perhaps time - resolved , optical sample is required to rigorously examine the influence of variability and duplicity on differences between optical and near - infrared t dwarf spectra .
we have examined lris red optical spectra for a sample of t dwarfs .
these spectra are largely dominated by pressure - broadened na i and k i features , which are highly sensitive to temperature , gravity , and metallicity .
weaker lines of cs i and rb i are also present , and appear to follow the trends of current chemical equilibrium calculations .
molecular bands of h@xmath0o , cah , crh , feh , and possibly ch@xmath1 are also present .
h@xmath2 emission is seen in three objects , weakly in sdss 1254@xmath30122 and 2mass 1047 + 2124 , but strong in the active t dwarf 2mass 1237 + 6526 .
none of the spectra exhibit the 6708 li i line , consistent with chemical depletion of this atomic species for t @xmath10 1500 k. trends in the observed strengths of these features have allowed us to define a classification scheme in the red optical , accurate to within one subclass . tying our scheme to selected near - infrared spectral standards , we are able to make a consistent comparison between optical and near - infrared types , and hence a critical examination of spectral morphologies in these two spectral regions .
we find excellent agreement between these classifications amongst the mid- and late - type t dwarfs , suggesting that either current classification schemes are largely temperature - based , or that temperature and gravity effects compensate for each other throughout the 0.62.5 @xmath5 region .
disentangling these physical parameters may be possible by comparing the strength of the k i red wing and the 9250 h@xmath0o band , potentially leading to a two - dimensional classification system that can be mapped onto temperature and gravity scales .
both gravity and metallicity effects help explain the peculiar spectrum of 2mass 0937 + 2931 , which may be an old , massive , and slightly metal - poor thick disk or halo brown dwarf . further astrometric data and improved spectral models will enable a more rigorous characterization of this unique brown dwarf . clearly ,
while very little of the total luminosity of t dwarfs is emitted shortward of 1 @xmath5 , there are an abundance of diagnostics present in this spectral region that can be used to examine the physics of these cool brown dwarfs .
we are grateful to our keck observing assistants joel aycock , julie rivera , and terry stickel , and instrument specialists / support astronomers paola amico , tom bida , and bob goodrich for their assistance during the data acquisition ; and to the nasa tac for its allocation of time for this project .
we also thank our anonymous referee for her / his insightful comments and criticisms .
a. j. b. acknowledges support provided by nasa through hubble fellowship grant hst - hf-01137.01 awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
j. d. k. acknowledges the support of the jet propulsion laboratory , california institute of technology , which is operated under contract with the national aeronautics and space administration .
a. b. acknowledges funding through nasa grants nag5 - 10760 and nag5 - 10629 .
this publication makes use of data from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center , funded by the national aeronautics and space administration and the national science foundation .
2mass data were obtain through the nasa / ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
portions of the data presented herein were obtained at the w. m. keck observatory which is operated as a scientific partnership among the california institute of technology , the university of california , and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w. m. keck foundation .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
leggett , s. k. , golimowski , d. a. , fan , x. , geballe , t. r. , & knapp , g. r. 2003 , in proceedings of the 12th cambridge workshop on cool stars , stellar systems , and the sun , ed .
a. brown , t. r. ayres , & g. m. harper ( boulder : univ .
colorado press ) , p. 120
lllccllc 2mass j04151954@xmath30935066 & t@xmath388/t@xmath398 & 2001 feb 20 & 6000 & 1.201.38 & wd 0413 - 074 & da4 & 1 + 2mass j05591914@xmath31404488 & t@xmath385/t@xmath395 & 2000 mar 5 & 3600 & 1.201.22 & wd 0552 - 041 & dc9 & 2 + 2mass j07271824 + 1710012 & t@xmath387/t@xmath398 & 2000 mar 5 & 1800 & 1.00 & wd 0747 + 073.1 & dc9 & 1 + 2mass j07554795 + 2212169 & t@xmath385:/t@xmath396 & 2001 feb 20 & 2400 & 1.131.19 & sao 79820 & g0 v & 1 + 2mass j09373487 + 2931409 & t@xmath386p / t@xmath397 & 2000 mar 5 & 3600 & 1.01 & ( wd 0924 + 199 ) & ( dc5 ) & 1 + 2mass j10475385 + 2124234 & [email protected]/t@xmath397 & 2001 feb 20 & 4800 & 1.061.22 & hd 93583 & g0 v & 3 + 2mass j12171110@xmath30311131 & [email protected]/t@xmath397 & 2000 mar 5 & 1800 & 1.111.16 & wd 1225 - 079 & dza5 & 3 + 2mass j12255432@xmath32739466ab & t@xmath386/t@xmath396 & 2000 mar 5 & 3600 & 1.481.50 & wd 1225 - 079 & dza5 & 3 + 2mass j12373919 + 6526148 & [email protected]/t@xmath397 & 2000 mar 5 & 1800 & 1.46 & & & 3 + & & 2001 feb 20 & 2400 & 1.491.52 & sao 15828 & f8 v & 3 + sdssp [email protected] & t@xmath382/t@xmath392 & 2001 feb 20 & 3600 & 1.081.13 & hd 111942 & g0 v & 4 + gliese 570d & t@xmath388/t@xmath397 & 2000 mar 5 & 3600 & 1.331.36 & wd 1444 - 174 & dc9 & 5 + & & 2001 feb 20 & 4800 & 1.081.13 & hd 131878 & g0 v & 5 + 2mass j15031961 + 2525196 & [email protected]/t@xmath396 & 2003 jan 2 & 1200 & 1.351.40 & bd+26 2652 & g0 & 6 + 2mass j15344984@xmath32952274ab & [email protected]/t@xmath396 & 2001 feb 20 & 2400 & 1.581.63 & hd 138874 & f7 v & 1 + lccl na i & 5890 & 3s @xmath50s@xmath51 @xmath3 3p @xmath50p@xmath52 & 1,2 + na i & 5896 & 3s @xmath50s@xmath51 @xmath3 3p @xmath50p@xmath51 & 1,2 + h@xmath2 & 6563 & 3d @xmath50d@xmath53 @xmath3 2p @xmath50p@xmath52 & 1 + cah & 67507050 & 0 - 0 band of a@xmath50@xmath54-x@xmath50@xmath55 & 3 + k i & 7665 & 4s @xmath50s@xmath51 @xmath3 4p @xmath50p@xmath52 & 1,2 + k i & 7699 & 4s @xmath50s@xmath51 @xmath3 4p @xmath50p@xmath51 & 1,2 + rb i & 7800 & 5s @xmath50s@xmath51 @xmath3 5p @xmath50p@xmath56 & 1 + rb i & 7948 & 5s @xmath50s@xmath51 @xmath3 5p @xmath50p@xmath56 & 1 + ch@xmath1 & 88009200 & 4(@xmath57@xmath58,@xmath57@xmath24 ) & 4 + cs i & 8521 & 6s @xmath50s@xmath51 @xmath3 6p @xmath50p@xmath52 & 1 + crh & 8611 bandhead & 0 - 0 band of a@xmath59@xmath55@xmath60-x@xmath59@xmath55@xmath60 & 5 + feh & 8692 bandhead & 1 - 0 band of a@xmath61@xmath62-x@xmath61@xmath62 & 6 + cs i & 8943 & 6s @xmath50s@xmath51 @xmath3
6p @xmath50p@xmath51 & 1 + h@xmath0o & 92509400 & 3(@xmath57@xmath58,@xmath57@xmath24 ) & 7 + h@xmath0o & 94509800 & 2(@xmath57@xmath58,@xmath57@xmath24 ) + 2@xmath57@xmath0 & 7 + feh & 9896 bandhead & 0 - 0 band of a@xmath61@xmath62-x@xmath61@xmath62 & 6 + crh & 9969 bandhead & 0 - 1 band of a@xmath59@xmath55@xmath60-x@xmath59@xmath55@xmath60 & 5 + llccccccccccc sdss 1254@xmath30122 & t@xmath382/t@xmath392 & 8520@xmath212 & [email protected] & & 8942@xmath212 & [email protected] & & ... & @xmath63 7 & & 7947@xmath218 & [email protected] + 2mass 0559@xmath31404 & t@xmath385/t@xmath395 & 8522@xmath212 & [email protected] & & 8945@xmath212 & [email protected] & & 7800@xmath2130 & 6@xmath215 & & 7950@xmath2114 & [email protected] + 2mass 0755 + 2212 & t@xmath385:/t@xmath396 & 8521@xmath213 & [email protected] & & 8944@xmath212 & [email protected] & & ... & @xmath63 2 & & 7949@xmath2111 & [email protected] + 2mass 1503 + 2525 & [email protected]/t@xmath396 & 8519@xmath212 & [email protected] & & 8941@xmath212 & [email protected] & & 7800@xmath2130 & 12@xmath215 & & 7944@xmath2115 & [email protected] + 2mass 1534@xmath32952ab & [email protected]/t@xmath396 & 8517@xmath212 & [email protected] & & 8940@xmath212 & [email protected] & & ... & @xmath63 130 & & 7943@xmath218 & 15@xmath2112 + 2mass 1225@xmath32739ab & t@xmath386/t@xmath396 & 8522@xmath212 & [email protected] & & 8944@xmath212 & [email protected] & & ... & @xmath63 24 & & ... & @xmath63 10 + 2mass 0937 + 2931 & t@xmath386p / t@xmath397 p & 8520@xmath212 & [email protected] & & 8942@xmath212 & [email protected] & & ... & ... & & 7949@xmath2117 & 10@xmath216 + 2mass 1047 + 2124 & [email protected]/t@xmath397 & 8522@xmath213 & [email protected] & & 8946@xmath214 & [email protected] & & ... & @xmath63 25 & & ... & @xmath63 7 + 2mass 1237 + 6526 & [email protected]/t@xmath397 & ... & @xmath63 9 & & 8941@xmath213 & 8@xmath214 & & ... & ... & & ... & @xmath63 60 + & & 8522@xmath213 & [email protected] & & 8943@xmath212 & [email protected] & & ... & @xmath63 40 & & ... & @xmath63 14 + 2mass 0727 + 1710 & t@xmath387/t@xmath398 & 8519@xmath213 & [email protected] & & 8943@xmath212 & [email protected] & & ... & @xmath63 50 & & 7949@xmath2110 & 16@xmath217 + 2mass 1217@xmath30311 & [email protected]/t@xmath397 & 8521@xmath212 & [email protected] & & 8944@xmath213 & [email protected] & & ... & @xmath63
18 & & ... & @xmath63 8 + gliese 570d & t@xmath388/t@xmath397 & 8519@xmath212 & [email protected] & & 8942@xmath212 & [email protected] & & ... & @xmath63 18 & & ... & ... + & & 8523@xmath213 & [email protected] & & 8944@xmath212 & [email protected] & & ... & @xmath63 18 & & 7951@xmath217 & 11@xmath217 + 2mass 0415@xmath30935 & t@xmath388/t@xmath398 & 8520@xmath219 & [email protected] & & 8943@xmath212 & [email protected] & & ... & @xmath63 12 & & 7948@xmath217 & 11@xmath216 + llcc sdss 1254@xmath30122 & t@xmath382/t@xmath392 & [email protected] & @xmath64 + 2mass 0559@xmath31404 & t@xmath385/t@xmath395 & @xmath65 & @xmath66 + 2mass 0755 + 2212 & t@xmath385:/t@xmath396 & @xmath67 & @xmath68 + 2mass 1503 + 2525 & [email protected]/t@xmath396 & @xmath69 & @xmath70 + 2mass 1534@xmath32952ab & [email protected]/t@xmath396 & @xmath71 & @xmath72 + 2mass 1225@xmath32739ab & t@xmath386/t@xmath396 & @xmath73 & @xmath74 + 2mass 0937 + 2931 & t@xmath386p / t@xmath397 p & @xmath75 & @xmath76 + 2mass 1047 + 2124 & [email protected]/t@xmath397 & [email protected] & @xmath77 + 2mass 1237 + 6526 & [email protected]/t@xmath397 & [email protected] & @xmath78 + & & [email protected] & @xmath79 + 2mass 0727 + 1710 & t@xmath387/t@xmath398 & @xmath80 & @xmath81 + 2mass 1217@xmath30311 & [email protected]/t@xmath397 & @xmath82 & @xmath83 + gliese 570d & t@xmath388/t@xmath397 & @xmath84 & @xmath83 + & & @xmath85 & @xmath68 + 2mass 0415@xmath30935 & t@xmath388/t@xmath398 & @xmath63 7.9 & @xmath74 + lcccl csi(a ) & @xmath86+@xmath87 & @xmath88 & 8521 cs i & 1,2 + csi(b ) & @xmath89+@xmath90 & @xmath91 & 8943 cs i & 1,2 + h@xmath0o & @xmath92 & @xmath93 & 9250 h@xmath0o & 2 + crh(a ) & @xmath94 & @xmath95 & 8611 crh & 3 + crh(b ) & @xmath96 & @xmath97 & 9969 crh & 2,3 + feh(a ) & @xmath94 & @xmath98 & 8692 feh & 3 + feh(b ) & @xmath96 & @xmath99 & 9896 feh &
2,3 + color - e & @xmath100 & @xmath101 & spectral slope & 2 + llccccl 2mass 1632 + 1904 & l8 & 1.70 & 1.02 & 1.11 & 1.88 & 1 + sdss 1254@xmath30122 & t@xmath392 & 2.01 & 0.78 & 1.13 & 4.02 & 2,3 + 2mass 0559@xmath31404 & t@xmath395 & 1.77 & 0.63 & 1.37 & 4.24 & 2 + sdss 1624 + 0029 & t@xmath396 & 1.68 & 0.47 & 1.15 & 3.83 & 4 + 2mass 0415@xmath30935 & t@xmath398 & 1.19 & 0.25 & 0.94 & 4.20 & 2,3 + lccccllll sd 0423@xmath30414 & 1.95 ( t@xmath392 ) & 1.20 ( @xmath63 l8 ) & 1.44 ( @xmath63 t@xmath395 ) & 2.21 ( l8 ) & l8 : & @xmath63 l8 & t@xmath380 & 1 + sd 0837@xmath30000 & 1.80 ( l8/t@xmath392 ) & 0.89 ( l8/t@xmath392 ) & 0.90 ( @xmath63 t@xmath395 ) & 3.35 ( l8/t@xmath392 ) & t@xmath390@xmath102 & t@xmath392 & t@xmath381 & 1 + sd 1021@xmath30304 & 1.98 ( t@xmath392 ) & 0.67 ( t@xmath392/5 ) & 1.37 ( t@xmath395 ) & 3.85 ( @xmath103 t@xmath392 ) & t@xmath394@xmath102 & t@xmath392 & t@xmath383 & 1 + 2 m 1503 + 2525 & 1.61 ( t@xmath396 ) & 0.50 ( t@xmath396 ) & 1.30 ( t@xmath395/6 ) & 4.75 ( @xmath103 t@xmath392 ) & t@xmath396 & t@xmath395/6 & [email protected] & 2 + 2 m 1534@xmath32952ab & 1.70 ( t@xmath396 ) & 0.54 ( t@xmath395/6 ) & 1.38 ( t@xmath395 ) & 4.47 ( @xmath103 t@xmath392 ) & t@xmath396 & t@xmath395/6 & [email protected] & 2 + 2 m 0755 + 2212 & 1.50 ( t@xmath396/8 ) & 0.52 ( t@xmath395/6 ) & 1.33 ( t@xmath395 ) & 3.78 ( @xmath103 t@xmath392 ) & t@xmath396 & t@xmath395/6 & [email protected] : & 2 + 2 m 1225@xmath32739ab & 1.82 ( t@xmath395 ) & 0.41 ( t@xmath396 ) & 1.17 ( t@xmath396 ) & 4.28 ( @xmath103 t@xmath392 ) & t@xmath396 & t@xmath396 & t@xmath386 & 2 + sd 1346@xmath30031 & 1.43 ( t@xmath396/8 ) & 0.35 ( t@xmath396/8 ) & 1.01 ( t@xmath396/8 ) & 5.01 ( @xmath103 t@xmath392 ) & t@xmath397 & t@xmath396/8 & t@xmath386 & 3 + 2 m 0937 + 2931 & 1.55 ( t@xmath396/8 ) & 0.38 ( t@xmath396/8 ) & 1.09 ( t@xmath396/8 ) & 5.99 ( @xmath103 t@xmath392 ) & t@xmath397 & t@xmath396/8 p & t@xmath386 p & 2 + 2 m 1047 + 2124 & 1.46 ( t@xmath396/8 ) & 0.37 ( t@xmath396/8 ) & 1.00 ( t@xmath396/8 ) & 4.41 ( @xmath103 t@xmath392 ) & t@xmath397 & t@xmath396/8 & [email protected] & 2 + 2 m 1237 + 6526 & 1.58 ( t@xmath396 ) & 0.37 ( t@xmath396/8 ) & 0.92 ( t@xmath398 ) & 4.93 ( @xmath103 t@xmath392 ) & t@xmath397 & t@xmath396/8 & [email protected] & 2,3 + 2 m 0727 + 1710 & 1.28 ( t@xmath398 ) & 0.36 ( t@xmath396/8 ) & 0.92 ( t@xmath398 ) & 3.84 ( @xmath103 t@xmath392 ) & t@xmath398 & t@xmath396/8 & t@xmath387 & 2 + 2 m 1217@xmath30311 & 1.71 ( t@xmath395/6 ) & 0.34 ( t@xmath396/8 ) & 0.92 ( t@xmath398 ) & 3.77 ( @xmath103 t@xmath392 ) & t@xmath397 & t@xmath398 & [email protected] & 2 + gliese 570d & 1.48 ( t@xmath396/8 ) & 0.34 ( t@xmath396/8 ) & 0.87 ( t@xmath398 ) & 4.39 ( @xmath103 t@xmath392 ) & t@xmath397 & t@xmath398 & t@xmath388 & 1,2 + | we present 630010100 spectra for a sample of 13 t dwarfs observed using lris mounted on the keck i 10 m telescope .
a variety of features are identified and analyzed , including pressure - broadened k i and na i doublets ; narrow cs i and rb i lines ; weak cah , crh , and feh bands ; strong h@xmath0o absorption ; and a possible weak ch@xmath1 band .
h@xmath2 emission is detected in three of the t dwarfs , strong in the previously reported active t dwarf 2mass 1237 + 6526 and weak in sdss 1254@xmath30122 and 2mass 1047 + 2124 .
none of the t dwarfs exhibit li i absorption .
guided by the evolution of optical spectral features with near - infrared spectral type , we derive a parallel optical classification scheme , focusing on spectral types t5 to t8 , anchored to select spectral standards .
we find general agreement between optical and near - infrared types for nearly all of the t dwarfs so far observed , including two earlier - type t dwarfs , within our classification uncertainties ( @xmath41 subtype ) .
these results suggest that competing gravity and temperature effects compensate for each other over the 0.62.5 @xmath5 spectral region .
we identify one possible means of disentangling these effects by comparing the strength of the k i red wing to the 9250 h@xmath0o band .
one of our objects , 2mass 0937 + 2931 , exhibits a peculiar spectrum , with a substantial red slope and relatively strong feh absorption , both consequences of a metal - deficient atmosphere . based on its near - infrared properties and substantial space motion , this object may be a thick disk or halo brown dwarf . |
recently , the scaling behaviors of the transverse momentum distributions for particles produced in high energy collisions attract more and more attentions .
the search on the scaling behaviors of particle spectrum is significantly important for understanding the evolution of high energy heavy ion collisions and particle production mechanisms , because the produced particle distribution carries information about the dynamics of the system and is one of the most important observables in high energy collisions . in @xcite , with the experimental data from star , phenix and brahms , we found there is a scaling behavior of transverse momentum @xmath0 spectrum for pions .
the scaling function is independent of the colliding system , the colliding energy , the centrality and ( pseudo)rapidity . for protons and anti - protons , there also exists a similar scaling behavior which is independent of the centrality and rapidity in au+au collisions at @xmath1 @xcite . for strange particles , such as @xmath3 , @xmath4 and @xmath5 ,
the scaling behaviors for their distributions are under consideration . in high energy heavy ion collisions , a lot of the initial kinetic energy of the bombarding nuclei is converted into those of the produced particles , including longitudinal and transverse ones .
different from momentum , kinetic energy is a scalar and is directly associated with the temperature of the hot medium created in the collisions . for different species of particles , the same momentum corresponds to different kinetic energy because of mass effect . as well - known , the kinetic energy , rather than momentum , of particles in a thermalized system satisfies the boltzmann distribution .
thus the distribution of kinetic energy of particles produced in ultra - relativistic heavy ion collisions is more effective in revealing the thermal properties of the system . with above consideration in mind
, one can ask whether the transverse kinetic energy @xmath2 distribution has similar scaling behavior .
this is our motivation of investigating the scaling properties of distributions of transverse kinetic energy @xmath2 of particles .
we hope the mass effect can be suppressed in the new scaling function . in this paper the scaling @xmath2 distribution of protons in the mid - rapidity region
is studied and compared with the scaling @xmath0 distribution .
it is very essential to ask why the scaling behaviors exist for different particles and what is the potential universal dynamics . in @xcite
we found that the string overlap mechanism can not explain data for pion and proton simultaneously . in this paper
we will consider another mechanism based on the parton percolation theory @xcite , and hope to get more information about particle production mechanism in nuclear collisions .
this paper is organized as follows : in sec .
ii , we discuss relation between distributions for @xmath2 and @xmath0 , and the scaling property of @xmath2 distribution of protons produced in au+au collisions at @xmath1 .
universal scaling functions for pion and proton are given . in sec .
iii we discuss the universal scaling functions in the framework of cluster decay .
the conclusion is drawn in sec .
from the definition of transverse kinetic energy @xmath6 , with @xmath7 mass of the particle , the invariant @xmath2 distribution is @xmath8 from the definition of @xmath2 , one can see that when @xmath9 or @xmath10 , @xmath2 is approximately equal to @xmath0 , and @xmath11 so it is feasible to only consider the scaling behavior of @xmath0 distribution in the range @xmath12 gev/@xmath13 for pions , whose mass is only @xmath14 . actually , the results obtained in @xcite are excellent .
but if @xmath7 is large , as for kaons , protons and anti - protons , etc , we must consider the mass effect and investigate the scaling behaviors of @xmath2 distributions in the same range of @xmath0 . in @xcite
we have summarized the method for searching the scaling behavior of the particle s @xmath0 spectrum .
now we can use the similar method to investigate the scaling behavior of the transverse energy distribution of final state particles , with transverse kinetic energy @xmath2 instead of transverse momentum @xmath0 .
first , define a scaled variable @xmath15 and the scaled spectrum @xmath16 with @xmath17 and @xmath3 free parameters chosen to fit the scaled distributions to the same curve .
of course values of @xmath17 and @xmath3 are different for distributions at different centralities , given that for most central collisions they are set to be 1 . then @xmath18 can be regarded as a parameterization of the @xmath2 distribution in most central au+au collisions . to obtain a universal scaling function for all centralities , independent of the arbitrary in choosing values of @xmath17 and @xmath3 for the most central collisions , we introduce
another scaled variable @xmath19 and the normalized scaling function @xmath20 here @xmath21 is defined as @xmath22 and one can easily check that @xmath23 and @xmath24 . with above steps , one can investigate whether there exists a scaling behavior of transverse energy distribution for protons produced at mid - rapidity in au+au collisions at @xmath1 , which has a wide rang @xmath0 coverage @xcite . as shown in fig .
[ protonfd ] , all data points for different centralities can be shifted to the same curve within errorbar , and the corresponding values of parameters @xmath17 and @xmath3 shown in table i. to parameterize the curve , we define @xmath25 , and the scaling function is @xmath26 after normalization , @xmath27 with redefined @xmath28 .
spectra for protons produced at mid - rapidity in au+au collisions at rhic .
feed - down corrections are considered in the data .
the solid curve is from eq .
( [ protondis ] ) .
the data are taken from @xcite after calculation with eq .
( [ ekpt]).,scaledwidth=45.0% ] .parameters for coalescing all data points to the same curves in figs .
[ protonfd ] for @xmath29 and [ pion ] for @xmath30 . [ cols="^,^,^,^,^",options="header " , ]
in this paper we discussed the scaling properties of the transverse kinetic energy distributions for pions and protons in au+au collisions at 200 gev .
we showed that the @xmath2 scaling behavior can describe experimental data better than that for the @xmath0 case .
although the cluster formation and decay mechanism can describe spectra of either pion or proton at different centralities , it can not give consistent description , with the same parameters in the cluster distribution , for both species of particles at the same time .
thus it can be excluded as a consistent particle production mechanism .
other particle production mechanism must be in effect for the scaling behavior .
this work was supported in part by the national natural science foundation of china under grant nos .
10635020 and 10775057 , by the ministry of education of china under grant no .
306022 and project irt0624 . | with the experimental data from star on the centrality dependence of transverse momentum @xmath0 spectra of pions and protons in au+au collisions at @xmath1 , we investigate the scaling properties of transverse energy @xmath2 distributions at different centralities . in the framework of cluster formation and decay mechanism for particle production ,
the universal transverse energy distributions for pion and proton can be described separately but not simultaneously . |
leptogenesis @xcite provides a simple and elegant explanation of the cosmological matter - antimatter asymmetry .
a beautiful aspect of this mechanism is the connection between the baryon asymmetry and neutrino properties . in its simplest version leptogenesis
is dominated by the @xmath0 violating interactions of the lightest of the heavy majorana neutrinos , the seesaw partners of the ordinary neutrinos .
the requirement of successful baryogenesis yields stringent constraints on the masses of light and heavy neutrinos .
in particular , all light neutrino masses have to be smaller than @xmath6 @xcite .
leptogenesis is closely related to classical gut baryogenesis @xcite , where the deviation of the distribution function of some heavy particle from its equilibrium distribution provides the necessary departure from thermal equilibrium .
the non - equilibrium process of baryogenesis is usually studied by means of boltzmann equations @xcite . in the same way ,
leptogenesis has been studied during the past years , with increasing sophistication @xcite-@xcite .
the goal of the present paper is to provide an analytical description of the leptogenesis process such that the dependence on the neutrino mass parameters becomes transparent . as we shall see this is important to understand the size of corrections to the simplest boltzmann equations , which have to be taken into account to arrive eventually at a ` theory of leptogenesis ' .
we will first consider the simplest case where the initial temperature @xmath7 is larger than @xmath1 , the mass of the lightest heavy neutrino @xmath8 .
we will also neglect decays of the two heavier neutrinos @xmath9 and @xmath10 , assuming that a generation of @xmath11 asymmetry from their decays either does not occur at all or that it does not influence the final value of @xmath11 .
further , we restrict ourselves to the non supersymmetric case , and we assume that the lightest heavy neutrino @xmath8 is the only relevant degree of freedom beyond the standard model particle species . within this minimal framework the boltzmann equations can be written in the following form , @xmath12 where @xmath13 .
the number density @xmath14 and the amount of @xmath11 asymmetry , @xmath15 , are calculated in a portion of comoving volume that contains one photon at temperatures @xmath16 , so that the relativistic equilibrium @xmath8 number density is given by @xmath17 .
there are four classes of processes which contribute to the different terms in the equations : decays , inverse decays , @xmath18 scatterings and @xmath19 processes mediated by heavy neutrinos .
the first three all modify the @xmath8 abundance and try to push it towards its equilibrium value @xmath20 . denoting by @xmath21 the hubble expansion rate , the term @xmath22 accounts for decays and inverse decays , while the scattering term @xmath23 represents the @xmath18 scatterings .
decays also yield the source term for the generation of the @xmath11 asymmetry , the first term in eq .
( [ lg2 ] ) , while all other processes contribute to the total washout term @xmath24 which competes with the decay source term .
the expansion rate is given by @xmath25 where @xmath26 is the total number of degrees of freedom , and @xmath27 is the planck mass .
note that we have not included the @xmath8 degrees of freedom since , as we will see , in the preferred strong washout regime , the heavy neutrinos are non - relativistic when the baryon asymmetry is produced .
the two terms @xmath28 and @xmath29 depend on the _ effective neutrino mass _ @xcite , defined as @xmath30 which has to be compared with the _ equilibrium neutrino mass _ @xmath31
the decay parameter @xmath32 introduced in the context of ordinary gut baryogenesis @xcite , controls whether or not @xmath8 decays are in equilibrium . here
@xmath33 is the @xmath8 decay width . the washout term @xmath34 has two contributions , @xmath35 ; the first term only depends on @xmath2 , while the second one depends on the product @xmath36 , where @xmath37 is the sum of the light neutrino masses squared @xcite .
the solution for @xmath15 is the sum of two terms @xcite , @xmath38 where the second term describes @xmath11 production from @xmath8 decays .
it is expressed in terms of the _ efficiency factor _
@xmath39 @xcite which does not depend on the cp asymmetry @xmath40 . in the following sections we shall use two integral expressions for the efficiency factor , [ ef ] ( z)&=&43 _ z_i^zdz d ( n_n_1-n_n_1^eq)e^-_z^zdzw(z ) + & = & - 43_z_i^zdz dd+sdn_n_1dze^-_z^zdzw(z ) .
here @xmath14 and @xmath41 are the solution of the first kinetic equation ( [ lg1 ] ) and its derivative , respectively .
the efficiency factor @xmath42 is normalized in such a way that its final value @xmath43 approaches one in the limit of thermal initial abundance of the heavy neutrinos @xmath8 and no washout ( @xmath44 ) . in general , for @xmath45 , one has @xmath46 . the first term in eq .
( [ solution ] ) accounts for the possible generation of a @xmath11 asymmetry before @xmath8 decays , e.g. from decays of the two heavier neutrinos @xmath9 and @xmath10 , or from a completely independent mechanism . in the following
we shall neglect such an initial asymmetry @xmath47 . in @xcite
it was shown that for values @xmath48 even large initial asymmetries are washed out for initial temperatures @xmath49 .
the predicted baryon to photon number ratio has to be compared with the value @xmath50 measured at recombination .
it is related to @xmath51 by @xmath52 . here
@xcite is the fraction of @xmath11 asymmetry converted into a baryon asymmetry by sphaleron processes , and @xmath54 is the dilution factor calculated assuming standard photon production from the onset of leptogenesis till recombination . using eq .
( [ ef ] ) , one then obtains @xmath55 in the following sections we will study analytically the solutions of the kinetic equations , focusing in particular on the final value of the efficiency factor .
we start in sect .
2 with the basic framework of decays and inverse decays . in the two regimes of weak ( @xmath56 ) and
strong ( @xmath48 ) washout the efficiency factor is obtained analytically , which then leads to a simple interpolation valid for all values of @xmath2 .
@xmath18 scatterings are added in sect . 3 , and the resulting lower bounds on the heavy neutrino mass @xmath1 and on the initial temperature @xmath7 are discussed . in sect . 4 an analytic derivation of the upper bound on the light neutrino masses is given , and in sect .
5 various corrections are described which have to be taken into account in a theory of leptogenesis . in appendix a a detailed discussion of the @xmath19 processes in the resonance region
is presented . in the case of
maximal @xmath0 violation the entire @xmath19 scattering cross section can be expressed in terms of @xmath1 , @xmath2 and @xmath3 .
the resulting boltzmann equations are compared with previously obtained results based on exact kadanoff - baym equations . in appendix
b various useful formulae are collected .
recently , two potentially important , and usually neglected , effects on leptogenesis have been discussed : the @xmath18 processes involving gauge bosons @xcite and thermal corrections at high temperature @xcite .
further , the strength of the @xmath19 washout term has been corrected @xcite compared to previous analysis .
however , the reaction densities for the gauge boson processes are presently controversial @xcite .
also the suggestion made in @xcite to include thermal masses as kinematical masses in decay and scattering processes leads to an unconventional picture at temperatures @xmath57 , which differs qualitatively from the situation at temperatures @xmath58 .
if thermal corrections are only included as propagator effects @xcite their influence is small .
this issue remains to be clarified .
fortunately , both effects are only important in the case of weak washout , i.e. for @xmath56 , where the final baryon asymmetry is strongly dependent on initial conditions in any case . in the strong washout regime , @xmath48 , which appears to be favored by the present evidence for neutrino masses
, they do not affect the final baryon asymmetry significantly . in the following
we will therefore ignore gauge boson processes and thermal corrections .
these questions will be addressed elsewhere .
the main results of this paper are summarized in the figures 6 , 9 and 10 .
illustrates that for the basic processes of decays and inverse decays the analytical approximation for the efficiency factor agrees well with the numerical result .
the figure also demonstrates that scatterings lead to a departure from this basic picture only for values @xmath59 , where the final baryon asymmetry depends strongly on the initial conditions .
9 shows the dependence of the efficiency factor on initial conditions and on @xmath18 scatterings for different values of the effective higgs mass @xmath60 .
again , for @xmath48 this dependence is small and , within the theoretical uncertainties , the efficiency factor is given by the simple power law _
f = ( 2 1 ) 10 ^ -2 ( 0.01ev)^1.10.1 . knowing the efficiency factor , one obtains from eqs .
( [ etab ] ) and ( [ cpmax ] ) the maximal baryon asymmetry .
10 shows the lower bound on the initial temperature @xmath7 as function of @xmath2 . in the most interesting mass range favored by neutrino oscillations it is about one order of magnitude smaller than the lower bound on @xmath1 .
the smallest temperature @xmath61 gev is reached at @xmath62 ev . in sect .
4 an analytic expression for the light neutrino mass bound is derived , which explicitly shows the dependence on the involved parameters .
it is very instructive to consider first a simplified picture in which decays and inverse decays are the only processes . for consistency , also
the real intermediate state contribution to the @xmath63 processes has to be included .
the kinetic equations ( [ lg1 ] ) and ( [ lg2 ] ) then reduce to @xmath64 where @xmath65 is the contribution to the washout term due to inverse decays . from eqs .
( [ ef ] ) and ( [ dlg1 ] ) one obtains for the efficiency factor , @xmath66 as we shall see , decays and inverse decays are sufficient to describe qualitatively many properties of the full problem . after a discussion of several useful analytic approximations we will study in detail the two regimes of weak and strong washout .
the insight into the dynamics of the non - equilibrium process gained from the investigation of these limiting cases will then allow us to obtain analytic interpolation formulae which describe rather accurately the entire parameter range .
all results will be compared with numerical solutions of the kinetic equations .
let us first recall some basic definitions and formulae .
the decay rate takes the form @xcite , @xmath67 where the thermally averaged dilation factor is given by the ratio of the modified bessel functions @xmath68 and @xmath69 , @xmath70 and @xmath71 is the decay width , @xmath72 with the higgs vacuum expectation value @xmath73 gev . the decay term @xmath28 is conveniently written in the form @xcite
@xmath74 the inverse decay rate is related to the decay rate by @xmath75 where @xmath76 is the equilibrium density of lepton doublets .
since the number of degrees of freedom for heavy majorana neutrinos and lepton doublets is the same , @xmath77 , one has @xmath78 the contribution of inverse decays to the washout term @xmath34 is therefore [ wid1 ] w_id(z ) = 12 _
id(z)h(z)z= 14 k z^3 k_1(z ) , which , together with eqs .
( [ g ] ) , ( [ d ] ) and ( [ neq ] ) , implies [ wid2 ] w_id(z ) = 12 d(z)n_n_1^eq(z)n_l^eq .
all relevant quantities are given in terms of the bessel functions @xmath68 and @xmath69 , whose asymptotic limits are well known . at high temperatures
one has , @xmath79 whereas at low temperatures , @xmath80 accurate interpolating functions for @xmath81 and @xmath82 for all values of @xmath83 are @xmath84 note , that for @xmath85 this approximation gives @xmath86 rather than the exact asymptotic form @xmath87 .
however , the high temperature domain is not so important for baryogenesis and the approximation ( [ approx ] ) is rather precise in the more relevant regime around @xmath88 .
( [ approx ] ) yields very simple expressions for the dilation factor and the decay term , @xmath89 as fig .
[ dil ] shows these analytical approximations are rather precise .
the relative error is always less than @xmath90 .
the washout term ( [ wid1 ] ) becomes in the approximation ( [ approx ] ) , @xmath91 it is useful to define a value @xmath92 , corresponding to a ` decay temperature ' @xmath93 below which decays are _ in equilibrium _ , by @xmath94 .
the value of @xmath92 is determined by @xmath95 , and from eq .
( [ dapprox ] ) one obtains @xmath96 for @xmath97 , this yields @xmath98 , whereas @xmath99 for @xmath100 . at @xmath101 one
has @xmath102 .
inverse decays are _ in equilibrium _ if @xmath103 . from eq .
( [ wid ] ) one easily finds that @xmath104 reaches its maximal value @xmath105 at @xmath106 .
hence , for @xmath107 , there exists an interval @xmath108 , where inverse decays are _ in equilibrium_. for @xmath109 no such interval exists and inverse decays are always _ out of equilibrium_. in the regime far out of equilibrium , @xmath97 , decays occur at very small temperatures , @xmath110 , and the produced @xmath11 asymmetry is not reduced by washout effects . in this case the integral for the efficiency factor ( [ dlg3 ] ) becomes simply , @xmath111 for @xmath112 no asymmetry is generated because the heavy neutrinos do not decay .
they also can not be produced since inverse decays are switched off as well .
hence , in this regime the dynamics is completely frozen .
for @xmath113 the equilibrium abundance is negligible , and from eq .
( [ dlg1 ] ) one finds , @xmath114 note that we have neglected the small neutrino abundance which for @xmath115 is produced before the neutrinos decay .
[ out ] shows the evolution of @xmath116 and @xmath117 for @xmath118 and @xmath119 , with @xmath120 , comparing the numerical solutions with the analytical expressions .
the final value of the efficiency factor @xmath43 is proportional to the initial @xmath8 abundance . if @xmath121 , then @xmath122 .
but if the initial abundance is zero , then @xmath123 as well .
therefore in this region there is the well known problem that one has to invoke some external mechanism to produce the initial abundance of neutrinos .
moreover the assumption that the initial asymmetry is washed out does not hold .
thus in the regime @xmath97 the results strongly depend on the initial conditions and the picture is not self - contained . in order to obtain the efficiency factor in the case of vanishing initial @xmath8-abundance , @xmath124
, one has to calculate how heavy neutrinos are dynamically produced by inverse decays .
this requires solving the kinetic equation ( [ dlg1 ] ) with the initial condition @xmath125 .
let us define a value @xmath126 by the condition @xmath127 eq . ( [ dlg1 ] ) implies that the number density reaches its maximum at @xmath128 .
an approximate solution can be found by noting that for @xmath129 inverse decays dominate and thus @xmath130 a straightforward integration yields for @xmath131 ( cf .
( [ g ] ) , ( [ d ] ) , ( [ neq ] ) ) , @xmath132 in the case @xmath133 , this implies for the number density , @xmath134 where the small dependence on @xmath135 has been neglected .
we can now calculate the corresponding approximate solution for the efficiency factor @xmath42 .
for @xmath131 the efficiency factor @xmath136 is negative since , @xmath137 . from eqs .
( [ dlg3 ] ) and ( [ low ] ) one obtains @xmath138 as expected , for @xmath139 the efficiency factor is proportional to @xmath14 , up to corrections which correspond to washout effects .
for @xmath140 , @xmath141 is reduced by washout effects . for @xmath142 , there is an additional positive contribution to the efficiency factor , @xmath143 the total efficiency factor is the sum of both contributions , @xmath144 for @xmath142 we now have to distinguish two different situations , the weak and strong washout regimes , respectively . consider first the case of weak washout , @xmath97 , which implies @xmath145 . from eq .
( [ nprod ] ) one then finds , @xmath146 a solution for @xmath116 , valid for any @xmath83 , is obtained by using in eq .
( [ nprod ] ) the useful approximation @xmath147^{1/c}}\ , , \ ] ] with @xmath148 .
this yields an interpolation of the two asymptotic regimes ( cf .
( [ nn1 < ] ) and ( [ nkweak ] ) ) , which is in excellent agreement with the numerical result , as shown in fig .
3a for @xmath149
. for @xmath150 decays dominate over inverse decays , such that @xmath151 in this way one easily obtains @xmath152 moreover , for @xmath153 , @xmath104 is exponentially suppressed and washout effects can be neglected in first approximation . for the negative part of the efficiency factor one then has ( cf .
( [ kminus ] ) ) @xmath154 from eq .
( [ oodec ] ) one obtains for the positive contribution , @xmath155 the final efficiency factor is then given by @xmath156 to first order in @xmath157 the final efficiency factor vanishes .
this corresponds to the approximation where washout effects are completely neglected .
as discussed above , @xmath158 is then proportional to @xmath159 and therefore zero . to obtain a non - zero asymmetry
the washout in the period @xmath131 is crucial .
it reduces the absolute value of the negative contribution @xmath160 , yielding a positive efficiency factor of order @xmath161 , @xmath162 ^ 2 \simeq { 9\pi^2\over 64}\,k^2 \ ; .\ ] ] such a reduction of the generated asymmetry has previously been observed in the context of gut baryogenesis @xcite .
note , that for @xmath163 eq .
( [ reduction ] ) does not hold , since in this case @xmath126 becomes small and washout effects for @xmath164 are also important . in fig .
[ prod]a the analytical solutions for @xmath116 and @xmath165 are compared with the numerical results for @xmath166 . a residual asymmetry survives after @xmath167 as remnant of the cancellation between the negative and the positive contributions to the efficiency factor .
the second one is prevalent because washout suppresses @xmath168 more efficiently .
as one can see in fig .
[ prod]a , the analytical solution for the asymmetry slightly underestimates the final numerical value .
this is because for @xmath169 the approximation of neglecting washout for @xmath142 becomes inaccurate . as @xmath95 increases
, @xmath126 decreases , and at @xmath170 the maximal number density @xmath171 reaches the equilibrium density @xmath172 at @xmath173 . for @xmath174 ,
one obtains from eq .
( [ nn1 < ] ) , @xmath175 .
a more accurate description for @xmath176 has to take into account decays in addition to inverse decays , i.e. one has to solve eq .
( [ dlg1 ] ) . since @xmath177
, one can use @xmath178 , and one then easily finds , @xmath179 which correctly reproduces eq .
( [ nn1 < ] ) for @xmath180 .
an example , with @xmath181 , is shown in fig .
[ prod]b which illustrates how well the analytical expression for neutrino production agrees with the numerical result .
consider now the efficiency factor .
for @xmath174 we can neglect the negative contribution @xmath182 , assuming that the asymmetry generated at high temperatures is efficiently washed out .
this is practically equivalent to assuming thermal initial abundance .
we will see in the next section how to describe the transition from the weak to the strong washout regime .
for @xmath183 , inverse decays are in equilibrium in the range @xmath184 , with @xmath185 . in the strong washout regime , @xmath100 ,
the efficiency factor can again be calculated analytically . for @xmath186^{1/3}$ ]
decays are not effective in tracking the equilibrium distribution . for the difference
@xmath187 with @xmath188 , one has , @xmath189 the corresponding efficiency factor is given by ( cf .
( [ ef ] ) ) @xmath190 on the other hand , for @xmath113 the neutrino abundance tracks closely the equilibrium behavior .
since @xmath191 , one can solve eq .
( [ dlg1 ] ) systematically in powers of @xmath192 , which yields @xmath193 using the properties of bessel functions , eq .
( [ neq ] ) yields for the derivative of the equilibrium density ,
@xmath194 we can now calculate the efficiency factor . from eqs .
( [ dlg3 ] ) and ( [ dneq ] ) one obtains @xmath195 the integral is dominated by the contribution from a region around the value @xmath196 where @xmath197 has a minimum .
the condition for a local minimum @xmath198 , the vanishing of the first derivative , yields @xmath199 since the second derivative at @xmath198 is positive one has @xmath200 .
the integral ( [ efd ] ) can be evaluated systematically by the steepest descent method ( cf .
alternatively , a simple and very useful approximate analytical solution can be obtained by replacing in the exponent of the integrand @xmath104 by @xmath201 the efficiency factor then becomes @xmath202 it is now easy to understand the behavior of @xmath42 . for @xmath203 ,
one has @xmath204 , while for @xmath205 the efficiency factor gets frozen at a final value @xmath206 , up to a small correction @xmath207 .
one can also easily find global solutions for all values of @xmath83 by interpolating the asymptotic solutions for @xmath208 and @xmath209 , respectively . from eqs .
( [ d ] ) , ( [ del1 ] ) , ( [ del ] ) and ( [ dneq ] ) one obtains for the difference between @xmath8-abundance and equilibrium abundance , @xmath210 similarly , an interpolation between the expressions ( [ kzsmall ] ) and ( [ kz ] ) for the efficiency factor is given by @xmath211 a typical example of strong washout is shown in fig . [ strong ] for the value @xmath181 , as in fig .
[ prod]b , but now for thermal initial abundance . in this figure
we also show the decay , inverse decay and washout terms . instead of the neutrino abundance the deviation @xmath212 is shown . the dotted , short dashed , dot - dashed and dashed lines are the approximations eq .
( [ dapprox ] ) for the @xmath28 term , eq .
( [ wid ] ) for the @xmath65 term , eq .
( [ delta ] ) for @xmath212 and eq .
( [ kstrong ] ) for @xmath42 , respectively .
the thin solid lines are the numerical results which agree well with the analytical approximations .
the behavior @xmath213 for @xmath113 and the freeze - out of @xmath15 at @xmath214 are clearly visible .
let us now focus on the final value of the efficiency factor @xmath43 .
note , that for @xmath174 also @xmath215 , and the condition ( [ eqz0 ] ) becomes approximately @xmath216 .
this means @xmath217 .
hence , the asymmetry produced for @xmath218 is essentially washed out , while for for @xmath219 washout is negligible ( @xmath220 . this simple picture will have some interesting consequences and applications .
the integral in eq .
( [ kz ] ) is easily evaluated , @xmath221 using the approximations ( [ d ] ) and ( [ wid ] ) , the condition ( [ eqz0 ] ) for @xmath222 becomes explicitly @xmath223 @xmath224 approaches one as @xmath95 goes to zero of eq .
( [ eqz0 ] ) approaches asymptotically 1.33 for @xmath225 . however , in the strong washout regime and also for the extrapolation @xmath225 this difference is irrelevant for @xmath158 . ] . for @xmath100
the solution of eq .
( [ z0a ] ) is given by @xmath226 where @xmath227 is one of the real branches of the lambert w function @xcite .
this result can be approximated by using the asymptotic expansion of @xmath227 @xcite , @xmath228 ^ 5\right ) + { \cal o}\left({\mbox{ln}(k)\over\mbox{ln}(\mbox{ln}(k))}\right)\;. \label{zb2}\ ] ] a rather accurate expression for @xmath224 for all values of @xmath95 is given by the interpolation ( cf .
fig . 5 ) ,
@xmath229 ^ 5\right)\;. \label{interpol}\ ] ] note the rapid transition from strong to weak washout at @xmath170 .
the final value of the efficiency factor takes the simple form @xmath230 this analytical expression for the final efficiency factor , combined with eq . ( [ interpol ] ) for @xmath224 , provides an accurate description of @xmath231 , as shown in fig .
[ deckf ] .
( [ kfan ] ) can also be extrapolated into the regime of weak washout , @xmath97 , where one obtains @xmath232 corresponding to thermal initial abundance , @xmath233 .
it turns out that also in the transition region eqs .
( [ kfan ] ) and ( [ interpol ] ) provide a rather accurate description , as is evident from fig .
[ deckf ] .
the largest discrepancy between analytical and numerical results is about 30% around @xmath234 . for comparison also the numerical result including scatterings is shown .
the difference with respect to the basic ` decay - plus - inverse - decay ' picture becomes significant only for @xmath235 .
the above analysis is easily extended to the case where the strength of the washout term @xmath65 is modified to @xmath236 .
for instance , in the model considered in @xcite , @xmath11 number changes by two in heavy particle decays , corresponding to @xmath237 . on the other hand ,
the heavy particle abundance is not affected by this change .
the final efficiency factor is therefore given by @xmath238 where @xmath224 is again given by eq .
( [ z0a ] ) . in the regime
@xmath174 our expression for the final efficiency factor can be approximated by @xmath239 which is very similar to the result was noted in @xcite , which is mostly related to the definition of the decay parameter @xmath95 ( cf .
( [ decpar ] ) ) . ] obtained by kolb and turner @xcite . comparing the efficiency factor ( [ kfan ] ) with the solution of the boltzmann equations including scatterings , as shown in fig . 6
, one arrives at the conclusion that the simple decay - plus - inverse - decay picture represents a very good approximation for leptogenesis in the strong washout regime .
as we will see , the difference is essentially negligible within the current theoretical uncertainties . given the results of the previous sections it is straightforward to obtain an expression for the efficiency factor for all values of @xmath95 also in the case of dynamical initial abundance .
we first introduce a number density @xmath240 which interpolates between the maximal number densities @xmath241 and @xmath242 ( cf .
( [ nkweak ] ) ) for strong and weak washout , respectively , @xmath243 the efficiency factor is in general the sum of a positive and a negative contribution , @xmath244 here @xmath245(k ) is given by ( [ kmweak ] ) for @xmath246 .
a generalization , accounting for washout also for @xmath142 , reads @xmath247 this expression extends the validity of the analytical solution to values @xmath248 in the case of a dynamical initial abundance .
@xmath158 is exponentially suppressed for @xmath174 .
an interpolation , satisfying the asymptotic behaviors at small and large @xmath95 , is given by @xmath249 on the other hand , the expression ( [ kfan ] ) for @xmath250 , which is valid for @xmath174 , has to approach @xmath251 for @xmath246 .
these requirements are fulfilled by @xmath252 equations ( [ nkall ] ) , ( [ km ] ) and ( [ kp ] ) , together with the interpolation ( [ interpol ] ) for @xmath224 , yield an accurate description of the efficiency factor @xmath231 for all values of @xmath95 , as demonstrated by fig . [ deckf ] .
this result is a good starting point for obtaining an analytic description of the efficiency factor for the full problem . in
the following sections we shall go beyond the simple decay - and - inverse - decay picture and include other processes step by step .
the scattering term @xmath29 and the related washout contribution @xmath253 arise from two different classes of higgs and lepton mediated inelastic scatterings involving the top quark ( @xmath254 ) and gauge bosons ( @xmath255 ) , @xmath256 their main effect is to enhance the neutrino production and thus the efficiency factor for @xmath56 .
further , they also contribute to the washout term , which leads to a correction of the efficiency factor for @xmath48 , i.e. in the strong washout regime .
since the scattering processes are specific to leptogenesis we shall use in this section mostly the variable @xmath257 @xcite rather than @xmath95 .
top quark and gauge boson scattering terms are expected to be of similar size
. however , the reaction densities for the gauge boson processes are presently controversial @xcite .
we shall therefore discuss these processes in detail elsewhere .
we shall also neglect the scale dependence of the top - yukawa coupling , which reduces the size of @xmath258 , since this decrease of @xmath29 will be partially compensated by @xmath259 .
the term @xmath260 is again the sum of two terms arising from the s - channel processes @xmath261 and the t - channel processes @xmath262 , @xmath263 , @xmath264 the scattering terms are defined as usual in terms of scattering rates and expansion rate , [ srates ] s_,s(t ) = _ , s(t)^(n_1)hz , and introducing the functions @xmath265 ( cf .
appendix b ) it is possible to write [ st ] s^t = k_s6 ( f_,s(z)+2f_,t(z ) ) ; here we have introduced the ratio @xmath266 with @xmath267 at high temperatures , @xmath268 , the functions @xmath269 have the following asymptotic form , f_,s(z ) & & 2 , + f_,t(z ) & & 2 . in fig .
[ s ] we have plotted the rates @xmath270 and @xmath28 as function of @xmath83 for @xmath271 , i.e. @xmath272 . for values @xmath273 the sum @xmath270 is dominated by the scattering rate @xmath29 , while for @xmath274 the decay rate @xmath275 dominates .
a simple analytic approximation for the sum @xmath276 is given by @xmath277\;,\ ] ] where @xmath278 here we have introduced the higgs mass @xmath60 to cut off the infrared divergence of the t - channel process . as fig .
[ s ] illustrates , the approximation ( [ s+d ] ) agrees well with the numerical result for @xmath279 as well as @xmath280 .
note that the latter value corresponds to the thermal higgs mass , @xmath281 , at the baryogenesis temperature @xmath5 in the strong washout regime .
the washout term induced by the @xmath18 scattering processes is again the sum of s- and t - channel contributions , @xmath282 the washout rates are directly related to the scattering rates ( [ srates ] ) , w_,t & = & ^l_,thz = n_n_1^eqn_l^eq s_,t , + w_,s & = & n_n_1n_n_1^eq^l_,shz = n_n_1^eqn_l^eqn_n_1n_n_1^eq s_,s . using eq . ( [ wid2 ] ) for the equilibrium number densities one obtains w_l=1 = 2 w_id1d ( n_n_1n_n_1^eq s_,s + 2 s_,t ) .
the washout rate including inverse decays is then given by [ w0 ] w_0 & = & w_id+w_l=1 + & = & w_id(1 + 1d(2 n_n_1n_n_1^eqs_,s + 4 s_,t ) ) .
this is the total washout rate as long as @xmath283 , the off - shell contribution from heavy neutrinos , can be neglected .
this is justified for sufficiently small values of @xmath1 ( cf .
we can now calculate the production of heavy neutrinos and study how the efficiency factor gets enhanced by the presence of the scattering term .
we again define a value @xmath126 by the condition ( [ condeq ] ) , n_n_1(z_eq)=n_n_1^eq(z_eq ) . for @xmath131
the number density can be obtained by integrating the equation @xmath284 the result is given by the expression ( @xmath285 ) @xmath286 \;.\ ] ] here the first integral is given by @xmath287^{1/c } } + z^3\,k_2(z)\ ; , \ ] ] where we have used the approximation eq .
( [ z2i1 ] ) .
the second integral can be expressed as [ ib ] i_b(z ) & = & _ 0^zdzz^4(1+az)k_2(z ) + & & 2 _ 0 ^ 1dzz^2(1+az )
+ a_1^zdzz^3k_2(z ) + & & 23 ( ( 1+a^3)(1+a ) - a^3 - a^2 + 12 a ) + ak_3(1)- az^3k_3(z ) .
the value @xmath126 can now be determined by setting @xmath14 , as determined from eqs .
( [ n < ] ) , ( [ ia ] ) and ( [ ib ] ) , equal to @xmath20 . using an approximate form for @xmath288 ,
one obtains an equation similar to eq .
( [ z0a ] ) , as described in appendix b. this yields a good approximation for @xmath126 in the case @xmath56 .
consider now the case of weak washout , @xmath289 , which implies @xmath145 .
for @xmath140 , decays dominate over inverse decays , [ dn > ] dn_n_1dz -(d+s)n_n_1 < 0 .
using @xmath290 , valid for @xmath291 ( cf .
( [ s+d ] ) and fig . [ s ] ) , this yields for the number density the simple expression [ n > ] n_n_1(z ) & & n_n_1^eq(z_eq ) e^-_z_eq^zdz(s+d ) + & & n_n_1^eq(z_eq)e^-k2(z^2-z^2_eq ) . in fig .
[ prods ] the solution @xmath116 is shown for @xmath292 .
the analytical solution agrees well with the numerical result .
we also make a comparison with the result already displayed in fig .
[ prod ] , where the s term is neglected .
as expected , the presence of the s term enhances the density @xmath14 at @xmath128 .
moreover the comparison illustrates the strong sensitivity of the efficiency factor in the case @xmath293 , not only to the initial conditions , but also to the theoretical description . a difference in @xmath8 abundance by less than a factor of two at @xmath126 corresponds to final efficiency factors which differ by two orders of magnitude .
this is due to delicate cancellations between the positive and the negative contribution to the efficiency factor and is a source of large theoretical uncertainties in the small @xmath2 regime .
[ figsprod ] we can now calculate the efficiency factor . in sect .
2.3.1 we have seen that in the absence of scatterings the inclusion of the small washout term was necessary to create an asymmetry between the negative contribution , @xmath294 , and the positive one , @xmath295 , in order to have a non - zero final value @xmath296 .
given the @xmath29 term one can neglect washout to first approximation . from eq .
( [ ef ] ) one then obtains @xmath297 where ( cf . (
[ d ] ) , ( [ s+d ] ) ) j(z ) = d+sd ( 1 + 158z ) . due to the s term we now have @xmath298 ,
although washout is neglected .
the reason is quite clear : as in the case without scatterings , the asymmetry is changed only by decays and inverse decays ; however , the number of decaying neutrinos at @xmath126 is now larger because of the additional production due to scatterings .
to first approximation we can thus calculate the efficiency factor neglecting washout . for @xmath299
one obtains ( cf .
( [ z2i1 ] ) ) , [ k < ] ^-(z)= -43_z_i^zdzdn_n_1^eq -3kz^3 ^ 1/c . for @xmath300 one
has @xmath301 ( cf .
[ s ] ) ; eq .
( [ efs ] ) the yields the simple result , [ k > ] ( z ) = 43(n_n_1(z_eq)j^-1(z_eq ) - n_n_1(z)j^-1(z))+ ^-(z_eq ) , which is shown in fig .
[ prods ] for @xmath302 and @xmath303 ( short dashed line ) ; it agrees reasonably well with the corresponding numerical solution ( solid line ) .
the analytical solution somewhat overestimates @xmath304 ; correspondingly , the final value @xmath296 is underestimated .
the analytical solution explains why the final value of the efficiency factor , @xmath158 , is proportional to @xmath2 , [ skfnow ] _ f. where [ ntildek ] ( k ) = 2n(k)z^3_eq ( ( 9)^c + ( 2z^3_eq)^c)^1/c , with @xmath242 .
contrary to the case discussed in sect .
2.3.1 , for which @xmath305 and @xmath306 , @xmath307 and @xmath308 are now different .
hence , there is no cancellation of terms @xmath309 between @xmath295 and @xmath182 .
the expression ( [ skfnow ] ) for the final efficiency factor fails for effective neutrino masses @xmath310 . including washout
mainly reduces the negative contribution @xmath182 and thereby enhances the final value of the efficiency factor .
( [ k < ] ) is then changed into [ kfmeno ] ^-_f = -43_z_i^dzd n_n_1^eq e^-_z^dzw_0(z )
. for @xmath129 one has @xmath137 , and the washout rate becomes ( cf .
( [ w0 ] ) ) w_0(z)w_id(z)d+4s_,td . from the expression ( [ s+d ] ) for @xmath276 one obtains d+4s_,t & & k_s(23 + ( m_1m_h ) z^2(1+az ) ) + & & 23k_s+kz ; here the last approximation requires @xmath311 .
together with eq .
( [ d ] ) this yields w_0(z)w_id(z)(1 + z ) , where the coefficient @xmath312 is given by = 2k_s3k + 158 . since @xmath313 is a total derivative , one obtains for the efficiency factor , ^-_f & = & -2 _
z_i^dzw_id(z)e^-14 k^2 k_2(z ) e^-_z^dzw_id(z ) + & & -2e^-12k _
z_i^dzw_id(z ) e^-_z^dzw_id(z ) , where we have used @xmath314 for @xmath315 . except for the exponential pre - factor
, this yields the result obtained in sect .
2.3.1 for decays and inverse decays ( cf . ( [ kmweak ] ) ) when the wash - out at @xmath300 is neglected [ wwws ] ^-_f = -2e^-12k(1-e^-23n(k ) ) .
the case without scatterings is recovered for @xmath316 .
in the case of strong washout , @xmath174 , the density of heavy neutrinos follows closely the equilibrium abundance , and one can obtain an analytical solution repeating the discussion in sect .
the efficiency factor is now given by [ efsw ] ( z)= - 43_z_in^zdzj^-1 dn_n_1^eqdze^-_z^zdzw_0(z ) .
using @xmath317 in the washout rate one obtains ( cf .
( [ w0 ] ) ) , [ w0 ] w_0 w_idj . in this way
one finds for the efficiency factor [ psis ] ( z ) & = & 2k_z_i^zdz1zj(z ) w_id(z)e^- _ z^zdzw_id(z ) j(z ) + & & _
z_i^zdze^-_s(z,z ) . as in sect .
2.3.2 , the dominant contribution to the integral arises from a region around a value @xmath318 where @xmath319 has a minimum . since @xmath320 for large @xmath83 , the value @xmath321 is again given by eq .
( [ eqz0 ] ) up to corrections @xmath322 .
replacing now @xmath104 by @xmath323 in the exponent of the integrand and by @xmath324 in the pre - factor , respectively , one obtains for the final efficiency factor the approximate solution , @xmath325 this extends eq .
( [ betakf ] ) to the case where scatterings are included .
note , that at small @xmath95 the efficiency factor ( [ skfstrong ] ) does not approach one , the value corresponding to thermal initial abundance .
however , any initial abundance can be reproduced by adjusting the exponent in eq .
( [ skfstrong ] ) . replacing @xmath326 by @xmath327
leaves eq .
( [ skfstrong ] ) essentially unchanged at large @xmath95 , whereas at small @xmath95 one has @xmath328 corresponding to thermal initial abundance .
the result is shown in fig .
[ skf ] ( short - dashed line ) and compared with numerical results for different values of @xmath329 . in the strong washout regime , @xmath330 , and for @xmath303 ,
the analytical and numerical results agree within @xmath331 .
since the strong washout regime is most interesting with respect to neutrino mass models , this is one of the most relevant results of this paper . for practical purposes
it is interesting to note that , within the current theoretical uncertainties , the efficiency factor for @xmath48 is given by the simple power law ,
@xmath332 the quoted uncertainties represent , approximately , the range visible in fig . 9 for large @xmath2 .
it is limited from above by the thin solid line corresponding to decays plus inverse decays and from below by the dot - dot - dashed line where scatterings are included with the extremely small ratio @xmath333 .
note that @xmath334 corresponds to a thermal higgs mass , @xmath281 , at the baryogenesis temperature @xmath335 . as in sect .
2.3.3 we can now obtain an expression for the efficiency factor for all values of @xmath95 by interpolating between the two regimes of small @xmath95 and large @xmath95 .
we shall use the number density @xmath308 ( cf .
( [ ntildek ] ) ) and the interpolation ( [ interpol ] ) for @xmath224 , which is related to the baryogenesis temperature by @xmath336 .
the efficiency factor is the sum of a positive and a negative contribution , _
f(k ) = _ f^+(k ) + _ f^-(k ) . here
@xmath337 differs from eq .
( [ km ] ) just by the exponential pre - factor induced by the scatterings ( cf .
( [ wwws ] ) ) , which yields @xmath338 the expression ( [ skfstrong ] ) for @xmath250 , which is valid for @xmath174 , has to approach eq .
( [ skfnow ] ) for @xmath246 .
an interpolating function is given by @xmath339 eq .
( [ kp ] ) is recovered for @xmath306 and @xmath340 .
the sum @xmath341 is shown in fig .
[ skf ] for @xmath303 .
the agreement with the numerical result is very good . including washout yields a description which correctly interpolates between the weak washout regime , @xmath293 and the strong washout regime , @xmath330 . the results for the efficiency factor are easily translated into theoretical predictions for the observed baryon - to - photon ratio using the relation ( [ etab ] ) .
the theoretical prediction has to be compared with the results from wmap @xcite combined with the sloan digital sky survey @xcite , @xmath342 , corresponding to @xmath343 the comparison yields the required @xmath0 asymmetry in terms of the baryon - to - photon ratio and the efficiency factor @xmath158 ( cf .
( [ etab ] ) ) , @xmath344 the @xmath0 asymmetry @xmath40 can be written as product of a maximal asymmetry and an effective leptogenesis phase @xmath345 @xcite , @xmath346 the connection between the leptogenesis phase and other @xmath0 violating observables is an important topic of current research @xcite . the maximal @xmath0 asymmetry @xmath347 depends in general on @xmath1 , @xmath2 and , via the light neutrino masses @xmath348 , on the absolute neutrino mass scale @xmath3 @xcite . for given light neutrino masses , i.e. fixed @xmath349 and @xmath350 ,
@xmath40 is maximized in the limit @xmath351 , for which one obtains @xcite , @xmath352 this expression reaches its maximum for fully hierarchical neutrinos , corresponding to @xmath353 and @xmath354 .
neutrino oscillation experiments give for atmospheric neutrinos @xcite @xmath355 and for solar neutrinos @xcite @xmath356 implying @xmath357 thus , apart the small experimental error , @xmath358 is a fixed parameter and the resulting maximal asymmetry depends uniquely on @xmath1 @xcite , @xmath359 the maximal @xmath0 asymmetry , or equivalently the maximal leptogenesis phase , corresponds to a maximal baryon asymmetry @xmath360 .
the cmb constraint @xmath361 , together with eq .
( [ cpmax ] ) , then yields a lower bound on the heavy neutrino mass @xmath1 , [ lbm1 ] m_1 > m_1^min & = & 1d163v^2m_atm _ b^cmb_f + & & 6.410 ^ 8gev ( _ b^cmb610 ^ -10 ) ( 0.05evm_atm)_f^-1 . note that the bound depends on the combination @xmath362
whose error , after the wmap result , receives a similar contribution both from @xmath363 and @xmath358 such that @xmath364 with the last inequality indicating the @xmath365 lower bound . for values of @xmath366
@xcite one can use the results of this section for the efficiency factor , neglecting the @xmath19 washout term .
( [ lbm1 ] ) then provides a lower bound on @xmath1 which depends only on @xmath2 . fig .
[ figbounds ] shows the analytical results for @xmath367 , based on eq .
( [ skfstrong ] ) for thermal initial abundance ( thin lines ) and the sum of eqs .
( [ skf- ] ) and ( [ skf+ ] ) for zero initial abundance ( thick lines ) . for comparison also the numerical results ( solid lines ) are shown .
the absolute minimum for @xmath1 is obtained for _ thermal initial abundance _ in the limit @xmath368 , for which @xmath122 .
the corresponding lower bound on @xmath1 can be read off from eq .
( [ lowm1 ] ) and at @xmath369 one finds @xmath370 this result is in agreement with @xcite and also with the recent calculation @xcite . note that the lower bound on @xmath1 becomes much more stringent in the case of only two heavy majorana neutrinos @xcite .
the bound for thermal initial abundance is model independent .
however , it relies on some unspecified mechanism which thermalizes the heavy neutrinos @xmath8 before the temperature drops considerably below @xmath1 .
further , the case @xmath371 is rather artificial within neutrino mass models , and in this regime a pre - existing asymmetry would not be washed out @xcite . for _ zero initial abundance _
the lower bound is obtained for @xmath372 , corresponding to @xmath373 . in this case one
obtains from eq .
( [ lowm1 ] ) @xcite , @xmath374 particularly interesting is the lower bound on @xmath1 in the favored neutrino mass range @xmath375 .
this range lies in the strong washout regime where the simple power law scaling ( [ powerlaw ] ) holds .
one thus obtains from eq .
( [ lowm1 ] ) @xmath376 which at @xmath369 implies @xmath377 where the last range corresponds to values @xmath378 .
note that these bounds are fully consistent with neglecting the @xmath19 washout term , which is justified for @xmath379 . in the case of near mass degeneracy between the lightest and the next - to - lightest heavy neutrino @xmath9 , @xmath380 , the upper bound on the @xmath0 asymmetry @xmath40
is enhanced by a factor @xmath381 @xcite .
the @xmath0 asymmetry of @xmath9 can also be maximal .
since the number of decaying neutrinos is essentially doubled , one obtains for the reduced lower bound on @xmath1 ( @xmath382 ) , @xmath383 where @xmath384 is given by eq .
( [ lbm1 ] ) .
it has been suggested that for extreme degeneracies a resonant regime @xcite is reached where @xmath385 . in this case
there is practically no lower bound on @xmath1 from leptogenesis .
it is usually assumed that the lower bound on the initial temperature @xmath7 roughly coincides with @xmath384 , the lower bound on the heavy neutrino mass @xmath1 . here
@xmath7 can be thought of as the temperature after reheating , below which the universe is radiation dominated @xcite .
however , in the following we will show that this is only the case in the weak washout regime , i.e. for @xmath386 , whereas in the more interesting strong washout regime @xmath7 can be about one order of magnitude smaller than @xmath1 . in general ,
the maximal baryon asymmetry is a function of both , @xmath1 and @xmath387 , @xmath388 in a rigorous procedure one would have to treat @xmath1 and @xmath7 as independent variables and to determine the values @xmath389 as well as @xmath384 for which the cmb constraint @xmath390 is satisfied .
this will yield a value @xmath391 somewhat larger than @xmath392 . for simplicity
, we shall use the approximation @xmath393 in the following .
we then define the value @xmath394 , and the corresponding temperature @xmath395 , by requiring that the final asymmetry @xmath396 agrees with observation within @xmath397 relative error of the quantity @xmath362 which controls @xmath384 . in the weak washout regime ,
i.e. @xmath56 , one has @xmath398 . at temperatures smaller than @xmath1 , the predicted asymmetry rapidly decreases .
either , there is not enough time to produce neutrinos ( for zero initial abundance ) or the thermal abundance is boltzmann suppressed ( for thermal initial abundance ) . as fig .
[ fig : zin ] illustrates , for @xmath56 the final efficiency factors for @xmath399 and @xmath400 differ by only 10% .
hence , in the weak washout regime one has @xmath401 . in the strong washout regime
the baryon asymmetry is predominantly produced around @xmath198 .
the value of @xmath394 is thus given by @xmath402 , where @xmath403 is the width of the gaussian which approximates the function @xmath404 ( cf .
( [ psis ] ) ) peaked at @xmath405 and whose integral between @xmath394 and infinity gives the final efficiency factor minus the small error that is tolerated . in fig .
[ figpsi ] the function @xmath404 is shown for @xmath406 and @xmath407 , respectively .
the width of the peak is given by @xmath408 for @xmath330 .
one can easily write down an approximate expression for @xmath409 which interpolates between the two regimes of weak and strong washout , @xmath410 the importance of the quantity @xmath321 becomes apparent by comparing figs .
[ z0 ] and [ fig : zin ] .
for instance , for @xmath411 one has @xmath412 and thus @xmath413 , while for @xmath414 one has @xmath415 and thus @xmath416 .
clearly , for values @xmath417 the suppression of the efficiency factor becomes significant . from eq .
( [ zinmax ] ) one immediately obtains a lower bound on the initial temperature @xmath7 , @xmath418 the result is shown in fig .
[ figbounds ] . for small
@xmath419 one has @xmath398 and consequently @xmath420 .
hence , in particular , the @xmath369 bounds ( [ m1lbth ] ) and ( [ m1lbza ] ) apply also to @xmath7 .
more interestingly , in the favored region @xmath421 ( dashed box ) the @xmath369 bound ( [ m1lbfr ] ) gets relaxed by a factor @xmath422 to @xmath423 and thus t_i^min=(410 ^ 9 210 ^ 10)gev therefore , in the favored region of @xmath2 ( dashed box ) , @xmath424 is only one order of magnitude higher than the absolute minimum @xmath425 at @xmath426 ( zero initial abundance ) and less than two orders of magnitude higher than the asymptotical minimum for @xmath293 ( thermal initial abundance ) .
this is important in view of the ` gravitino problem ' which yields an upper bound on @xmath7 for some supersymmetric extensions of the standard model . comparing our results with those of @xcite ,
where the additional @xmath11 asymmetry has been calculated which is produced during the reheating period at temperatures above @xmath427 and below some maximal temperature @xmath428 , we find the same amount of relaxation of the bound on @xmath429 .
this indicates that the relaxation is a consequence of @xmath430 in the case of strong washout rather than the existence of a non radiation dominated regime above @xmath427 .
we now want to study the effect of the contribution @xmath283 to the total washout .
this term originates from the @xmath431 processes @xmath432 and @xmath433 with the heavy neutrinos @xmath8 , @xmath9 and @xmath10 in s- and t - channel , respectively .
@xmath283 is the only term in the kinetic equations which is not proportional to @xmath2 but instead to the heavy neutrino mass @xmath1 . at low temperatures the washout term @xmath283 reads ,
@xmath434 where @xmath3 is the absolute neutrino mass scale , and the dimensionless constant @xmath435 is given by @xmath436 @xmath283 is compared in fig .
[ truth ] of appendix a with the total washout term @xmath437 as discussed in the appendix , there is a sharp transition to a low temperature regime where @xmath283 dominates over @xmath438 .
this transition occurs for a value @xmath439 , which is determined by @xmath440 from eqs .
( [ w0 ] ) and ( [ dwmax ] ) one easily obtains , @xmath441 in the case @xmath442 , the values of @xmath198 and of the efficiency factor at @xmath443 are not affected by @xmath283 . since for @xmath444
no asymmetry is produced , the total efficiency factor is simply given by @xmath445 where the second factor describes the modification due to the presence of @xmath446 .
note that @xmath447 depends on @xmath2 also via @xmath198 . for @xmath448 we can use the low temperature limit ( [ dwmax ] ) for @xmath283 , which yields @xmath449 given
the solar and atmospheric mass squared differences and a neutrino mass pattern , i.e. @xmath450 or @xmath451 , the dependence of @xmath350 on @xmath349 is fixed . in ref .
@xcite the absolute neutrino mass scale @xmath3 was used as variable . in the following we prefer to use instead the lightest neutrino mass @xmath349 . in the case of normal hierarchy , with @xmath452 and @xmath453 ,
one has @xmath454 in the case of inverted hierarchy analogous relations hold .
consider now the maximal baryon asymmetry ( cf .
( [ etab ] ) ) , @xmath455 in the case @xmath353 the maximal @xmath0 asymmetry was depending only on @xmath1 ( cf .
( [ cpmax ] ) ) . if @xmath456 this is suppressed by a function @xmath457 depending both on @xmath349 and on @xmath2 @xcite such that @xmath458 the maximal value @xmath459 is obtained in the case @xmath353 .
the function @xmath460 is conveniently factorized , @xmath461 the first factor , @xmath462 is the maximal value of @xmath460 for fixed @xmath349 , which decreases @xmath463 for @xmath464 .
the function @xmath465 contains the dependence on @xmath2 , @xmath466 this expression describes correctly the behavior of the maximal @xmath0 asymmetry in the limits @xmath467 and @xmath468 .
however , it has recently been pointed out @xcite that eq .
( [ funf ] ) underestimates the maximal @xmath0 asymmetry in particular in the regime of quasi - degenerate neutrinos ) has been obtained using eq .
( 22 ) in ref .
@xcite and assuming @xmath469 , which is valid only in the limit @xmath467 .
note , however , that eq .
( [ funf ] ) approximates the maximal @xmath0 asymmetry within about 20% also in the quasi - degenerate case for the relevant values of @xmath2 . for quasi - degenerate neutrinos , with @xmath470 ,
one easily finds that the maximal @xmath0 asymmetry is reached for @xmath471 . ] . for simplicity
, we shall first calculate the neutrino mass bound using eq .
( [ funf ] ) and then discuss the correction .
let us now calculate the value @xmath1 that maximizes @xmath396 . in the @xmath472-plane
this defines a trajectory @xmath473 along which @xmath396 is maximal with respect to @xmath1 .
the corresponding condition , @xmath474 yields the relation , @xmath475 where the quantity @xmath476 is a function of @xmath2 .
it is now easy to see that the ratio @xmath477 , maximized with respect to @xmath1 , can be expressed in the following form , @xmath478 where @xmath479 is the constant @xmath480 and @xmath481 .
the parameter @xmath482 is the product @xmath483 it accounts for various factors affecting the ratio @xmath477 : ( 1 ) the maximal @xmath0 asymmetry , @xmath484 ; ( 2 ) the atmospheric neutrino mass scale , @xmath485 ; ( 3 ) the observed baryon asymmetry , @xmath486 ; ( 4 ) the variation @xmath487 of the strength of the @xmath19 washout term , and ( 5 ) the variation @xmath488 of the efficiency factor at small @xmath489 .
this parametrization of the maximal asymmetry is useful to study the dependence of the neutrino mass bound on the various parameters involved . in order to determine the absolute maximum of the asymmetry
@xmath360 we also have to find the extremum with respect to @xmath2 and , finally , the maximum with respect to @xmath349 or , equivalently , the absolute neutrino mass scale .
comparison with the observed asymmetry @xmath490 then yields the leptogenesis neutrino mass bound . anticipating again that the maximum falls in the region of large @xmath2
, we can use the analytical expression ( [ skfstrong ] ) for @xmath158 in the strong washout region . since @xmath491
, one has @xmath492 further , for large @xmath2 the function @xmath493 can be approximated by @xmath494 with this simplified expression it is easy to see that the peak is reached for @xmath495 corresponding to @xmath496 .
the peak value of the asymmetry is given by @xmath497 anticipating @xmath498 , one has to zeroth order in @xmath499 , @xmath500 imposing now the cmb constraint @xmath501 we find the leptogenesis bound on the absolute neutrino mass scale ( cf .
@xcite ) , @xmath502 with @xmath503 in this last equation we used the fact that in a standard thermal history the dilution factor , contained in @xmath504 , is given by @xmath505 with @xmath506 the number of the ( entropy ) degrees of freedom at recombination .
combining eqs .
( [ m ] ) and ( [ mt ] ) one finds @xmath507 which is consistent with the approximation of strong washout used in eq .
( [ zbk ] ) . from fig .
[ z0 ] one then reads off @xmath508 .
together with ( [ m1peak ] ) this yields for the peak value of @xmath1 , @xmath509 it is straightforward to go beyond the zeroth order in @xmath499 . in the case of normal hierarchy the lightest neutrino mass bound is given by @xmath510 whereas in the case of inverted hierarchy one has @xmath511 which yields @xmath512 . in order to obtain numerical results for the upper bounds on the light neutrino masses one has to specify the baryon asymmetry and the neutrino mass squared differences .
for @xmath363 we use the wmap plus sdss result ( [ etanew ] ) , while the value for @xmath358 is given by the eq .
( [ matm ] ) . since @xmath513 , the experimental error on @xmath514 is about @xmath515 . setting all other parameters @xmath516 ,
one finds for the central value @xmath517 .
we then obtain @xmath518 the corresponding @xmath365 upper bounds on the neutrino masses are for normal hierarchy , @xmath519 and correspondingly in the case of inverted hierarchy , @xmath520 these analytical bounds are consistent with the numerical results obtained in @xcite , if one accounts for the different parameters , @xmath521 and @xmath522 ( @xmath523 ) , and the over - estimate of the washout term @xmath438 by 50% .
the bound on the maximal @xmath0 asymmetry derived in @xcite corresponds in the relevant range of large @xmath349 and @xmath2 to the function ( cf .
( [ funf ] ) ) , @xmath524 repeating the above analysis one finds that the peak of the asymmetry is shifted to @xmath525 , with @xmath526 . from eqs .
( [ zbk ] ) , ( [ m ] ) and ( [ max ] ) one then concludes that the neutrino mass bound is relaxed by the factor @xmath527 , i.e. 7% , corresponding to an increase of the neutrino mass bound by @xmath528 .
an important correction arises from the dependence of the neutrino masses on the renormalization scale @xmath529 .
the only low energy quantity upon which @xmath530 depends is the atmospheric neutrino mass scale @xmath358 .
hence , there are two competing effects : the first one is the running of @xmath358 from the fermi scale @xmath531 to the high scale @xmath532 ( @xmath533 ) , the second one is the running of @xmath530 from @xmath532 down to @xmath534 . in the standard model
the light neutrino masses scale uniformly under the renormalization group . since @xmath535 , the first effect then gives a correction that is only half of the second one .
renormalization group effects make the bound more restrictive @xcite . in order to have an upper bound
, we have to choose those values of the parameters that produce the smallest effect .
this corresponds to choosing the lowest higgs mass compatible with positive higgs self - coupling at @xmath533 , which is about @xmath536 .
the atmospheric neutrino mass scale is then increased by about @xmath537 @xcite and the bound gets @xmath538 weaker at @xmath532 , but @xmath538 more restrictive at @xmath534 .
combining the effect of radiative corrections and the larger @xmath0 asymmetry ( eq . ( [ funfcor ] ) ) , we finally obtain from eq .
( [ bnor ] ) and ( [ binv ] ) at @xmath365 , @xmath539 which , thanks to cancellations among different corrections , agrees with the bound obtained in @xcite .
it is important to realize , however , that there are corrections of the same order as those discussed above which can not be treated within the present framework .
it is usually assumed that in leptogenesis first a lepton asymmetry is produced , which is then partially transformed into a baryon asymmetry by sphaleron processes . however , this picture is incorrect @xcite .
the duration of leptogenesis is about two orders of magnitude larger than the inverse hubble parameter when it starts . since many processes in the plasma , in particular the sphaleron processes , are much faster , the generated asymmetry is ` instantaneously ' distributed among quarks and leptons .
hence , the chemical potentials of quarks and leptons are changed already during the process of leptogenesis .
a complete analysis has to take into account how the contributing ` spectator processes ' , which are in thermal equilibrium , change with decreasing temperature ( cf .
, e.g. , @xcite ) . in @xcite
is has been estimated that spectator processes reduce the generated baryon asymmetry by about a factor of two .
hence , there is presently a theoretical uncertainty of at least @xmath540 on the neutrino mass bound . finally , it has to be kept in mind that our whole analysis is based on the simplest version of the seesaw mechanism with hierarchical heavy majorana neutrinos .
the leptogenesis neutrino mass bound can be relaxed if the heavy majorana neutrinos are , at least partially , quasi - degenerate in mass . in this case
the @xmath0 asymmetry can be much larger @xcite than the upper bound used in the above discussion .
this possibility has to be discussed in the context of realistic models of neutrino masses .
further , if higgs triplets contribute significantly to neutrino masses the connection between baryon asymmetry and neutrino masses disappears entirely .
different relations between neutrino masses and baryon asymmetry are also obtained in non - thermal leptogenesis @xcite .
the goal of leptogenesis is the prediction of the baryon asymmetry , given neutrino masses and mixings .
the consistency of present calculations with observations is impressive , but so far it is not possible to quote a rigorous theoretical error on the predicted asymmetry , which is a necessary requirement for a ` theory of leptogenesis ' .
the generation of a baryon asymmetry is a non - equilibrium process which is generally treated by means of boltzmann equations .
this procedure has a basic conceptual problem : the boltzmann equations are classical equations for the time evolution of phase space distribution functions ; the involved collision terms , however , are zero temperature s - matrix elements which involve quantum interferences in a crucial manner .
clearly , a full quantum mechanical treatment is necessary to understand the range of validity of the boltzmann equations and to determine the size of corrections .
a first step in this direction has been made in @xcite where a perturbative solution of the exact kadanoff - baym equations has been constructed . to zeroth order , for non - relativistic heavy neutrinos
, the non - equilibrium green functions have been obtained in terms of distribution functions satisfying boltzmann equations .
note that in the favoured strong washout regime the decaying heavy neutrinos are indeed non - relativistic .
it is instructive to recall the various corrections .
there are off - shell contributions , ` memory effects ' related to the derivative expansion of the wigner transforms , relativistic corrections and higher - order loop corrections .
all these correction terms are known explicitly , but their size during the process of baryogenesis and , in particular , their effect on the final baryon asymmetry have not yet been worked out .
recently , thermal corrections have been studied @xcite .
they correspond to loop corrections involving gauge bosons and the top quark . at large temperatures , @xmath57 , the processes in the plasma and
the @xmath0 asymmetries change significantly if thermal masses are treated as kinematical masses in the evaluation of scattering matrix elements @xcite . on the contrary ,
thermal corrections are small if they are only included as propagator effects @xcite . to clarify this issue
is of general importance for the treatment of non - equilibrium processes at high temperatures .
the effect of all these corrections on the final baryon asymmetry depends crucially on the value of the neutrino masses .
large thermal corrections would modify the asymmetry at temperatures above @xmath1 .
this affects the final baryon asymmetry only in the case of small washout , i.e. @xmath56 . in the strong washout regime , @xmath48 , which appears to be favored by the current evidence for neutrino masses ,
the baryon asymmetry is generated at a temperature @xmath541 . in this case
thermal corrections are small .
correspondingly , the recently obtained bounds on light and heavy neutrino masses @xcite are all very similar .
the final value of the baryon asymmetry is significantly affected by ` spectator processes ' @xcite which can not be treated based on the simple boltzmann equations discussed in this paper . it has been estimated that this effect changes the baryon asymmetry by a factor of about two , leading to a theoretical uncertainty of the leptogenesis neutrino mass bound of about @xmath540 . clearly , to obtain a more accurate prediction for the baryon asymmetry requires a considerable increase in the complexity of the calculation .
an important step towards the theory of leptogenesis would be a systematic evaluation of all corrections to the simple boltzmann equations in the ` easy regime ' of strong washout where @xmath48 .
one could then see where this approach breaks down as @xmath2 decreases and @xmath5 approaches @xmath1 . on the experimental side , information on the absolute neutrino mass scale @xmath3 , and therefore on @xmath542 , is of crucial importance .
maybe , we are lucky , and nature has chosen neutrino masses in the strong washout regime where leptogenesis works best .
+ * acknowledgments * + we would like to thank l. covi , k. hamaguchi , j. pati and m. ratz for helpful discussions , and g. kane and the michigan center for theoretical physics ( mctp ) for hospitality during the _ baryogenesis workshop_. p.d.b . thanks the desy and cern theory groups for their kind hospitality .
the work of p.d.b . has been supported by the eu fifth framework network supersymmetry and the early universe " ( hprn - ct-2000 - 00152 ) .
a crucial and delicate point in setting up the boltzmann equations for leptogenesis is the subtraction of the real intermediate state contribution ( ris ) from the @xmath63 scattering processes @xcite . without this subtraction ,
decays and inverse decays lead to the generation of a lepton asymmetry in thermal equilibrium , in contradiction with general theorems . in order to explicitly split the @xmath63 scattering processes into ris and remainder one has to calculate the @xmath19 processes , including one - loop self - energy and vertex corrections in the resonance region .
this calculation has been carried out in @xcite where the relevant results are given in eqs .
( 68 ) - ( 85 ) .
let us consider for simplicity the case @xmath543 ( @xmath544 , @xmath254 and @xmath545 are the usual mandelstam variables ) . for our purposes
it is sufficient to study the averaged matrix element squared , |m(|l|l)|^2_av = _ -s^0 du |m(|l|l)|^2 , where the integral over @xmath545 corresponds to the integral over the final state lepton angle , i.e. a partial phase space integration . in order to study the resonance region
the diagonal part of the self - energy was re - summed in @xcite whereas the off - diagonal part was treated as perturbation in the yukawa couplings @xmath546 . in the @xmath63 scattering amplitude
the free propagator is then replaced by a breit - wigner propagator , = 1m_1 ^ 2(1d_1(x ) -i 1r_1(x ) ) , where = x-1(x-1)^2 + c^2 , = c(x-1)^2 + c^2 , and [ unit ] 1d_1(x)^2 + 1r_1(x)^2 = 1c1r_1(x ) , with x = sm_1 ^ 2 , c = _ 1m_1 = 18k_11 , k_ij = ( h^h)_ij ; here @xmath547 is the yukawa coupling of @xmath548 to @xmath549 .
the averaged matrix elements are then given by the following expression @xcite , |m(|l|l)|^2_av & = & 2 s^2 _ ij(a_ij - b_ij - c_ij - 4s _ k(d_ijk + _ ijk ) ) , + where we have only shown terms contributing to the subtraction of the ris part as well as the leading order off - shell part , e.g. @xmath0 conserving one - loop corrections have not been included . @xmath550 and @xmath551 represent the various @xmath552-@xmath548 s - channel interference terms . up to terms @xmath553 they are ( @xmath554 ) a_11 & = & k_11 ^ 2 1m_1 ^ 2(1d_1 ^ 2 + 1r_1 ^ 2 ) , + a_1i + a_i1 & = & -2 re\{k^2_1i}1m_1 m_i 1d_1 , + a_ij & = & re\{k^2_ij}1m_i m_j , + b_11 & = & b_ij + b_ji = 0 , + b_1i + b_i1 & = & 2 im\{k^2_1i}1m_1 m_i 1r_1 ; the @xmath8-@xmath8 s - channel terms with self - energy and vertex corrections , respectively , read ( @xmath555 ) c_1k & = & - 2x im\{k_1k^2 } cm_1 m_k ( 1d_1 ^ 2 + 1r_1 ^ 2 ) , + d_11k & = & 12 im\{k_1k^2 } cm_1 ^ 4 f(m_k^2s ) ( 1d_1 ^ 2 + 1r_1 ^ 2 ) , where f(y ) & = & ( 1-(1+y ) ) + & = & - 12 + o(1y ) ; finally , the s - u - channel interference term is _ 11k = - d_11k . from these equations one reads off d_11k
+ _ 11k = 0 , and , for @xmath556 ,
@xmath557 hence , the @xmath0 asymmetry of the full @xmath63 cross section vanishes to @xmath558 .
the ` pole terms ' , corresponding to @xmath8-@xmath8 s - channel contributions , are cancelled by on-/off - shell s - channel interferences ( self - energy ) and s - channel / u - channel interference ( vertex correction ) .
off - shell , the corresponding cancellations take place to @xmath553 @xcite .
as an unstable particle the heavy neutrino @xmath8 is defined as pole in the @xmath63 scattering amplitude ( |l|l ) l|n_1 n_1||l| .
the residue yields the decay amplitude and , in particular , the @xmath0 asymmetry .
the ris term can then be identified as the squared matrix element in the zero - width limit , & = & 16 ^2 k_11 m_1 ^ 2 ( 1 + 2_1^m + 2_1^v ) ( x-1 ) , where @xmath559 and @xmath560 are the familiar @xmath0 asymmetries due to mixing and vertex correction , respectively ( @xmath561 ) , _ 1^m & = & 18 _ k im\{k_1k^2}k_11m_1m_k , + _ 1^v & = & - 18 _ k im\{k_1k^2}k_11 f(m_k^2m_1 ^ 2 ) . in @xcite
it has been shown that the subtraction of the ris term , which corresponds to the replacement @xmath562 , leads to boltzmann equations with the expected properties , which have the equilibrium solution @xmath563 , @xmath564 .
it is now straightforward to write down the subtracted matrix element squared .
keeping only terms @xmath558 , where the zero - width limit can be taken for @xmath565 and @xmath566 , one obtains the simple expressions ( @xmath554 ) , |m(|l|l)|^2_sub & = & |m_l=2|^2_+ + |m_l=2|^2_- , + 2s^2\{k_11 ^ 2m_1 ^ 2 + & & -6_ire(k_1i^2)1m_1m_i + & & .+3_i , jre(k_ij^2)1m_im_j } , [ correct1 ] + & = & -32 ^ 2 k_11 m_1 ^ 2(_1^m + _
1^v ) ( x-1 ) .
[ correct2 ] note that the subtracted squared matrix element , contrary to the unsubtracted one , violates @xmath0 . to leading order in the coupling this part contributes only on - shell , and it is @xmath567 suppressed with respect to the leading born term .
away from the pole , for @xmath568 , one has |m(|l|l)|^2_sub & = & & = & 6 s^2 _
ijre\{k^2_ij}1m_i m_j + & = & 6 s^2v^4 tr(m_^m _ ) , [ lowenergy1 ] which is the crucial term leading to the upper bound on the neutrino masses @xcite .
we also have to take into account the @xmath19 process @xmath569 .
the corresponding matrix elements reads |m_l=2,t|^2 & = & 2s^2\ { k_11 ^ 2m_1 ^ 2 .
+ & & .+6_ire(k_1i^2)1m_1 m_i1x ( x+1 ) + 3_i , jre(k_ij^2)1m_i m_j}. for small center of mass energies one again obtains @xmath570 for the derivation of the upper bound on the light neutrino masses one needs the maximal @xmath0 asymmetry for given @xmath1 , @xmath2 and @xmath3 . in this case
also the complete @xmath571 matrix element depends just on these three variables .
this is easily seen in the flavor basis where the yukawa matrix @xmath572 connects light and heavy neutrino mass eigenstates .
the matrix [ ortho ] _ij = v_ij , is then orthogonal , @xmath573 @xcite , which implies _ k _ ik_jk1m_k = v^2 _ ij .
using @xmath574 , this implies for the interference term appearing in eq .
( [ correct1 ] ) , _ i1re\{k^2_1i}1m_i = -k_11 ^ 2m_1 + _ j=1 ^ 3 m_jv^2re\{^2_j1 } .
the conditions [ cond ] re\{^2_21 } = re\{^2_31 } = 0 , \{^2_11 } = m_1 m_1v^2 , yield a good approximation for the maximal @xmath0 asymmetry @xcite .
the difference to the maximal @xmath0 asymmetry @xcite can then be treated as a perturbation , as discussed in sect .
( [ cond ] ) then implies for the interference term _ i1re\{k^2_1i}1m_i = - m_1v^4(^2 - m_1 ^ 2 ) .
[ int ] inserting this expression in eq .
( [ correct1 ] ) one obtains for the @xmath19 matrix element in the case of maximal @xmath0 asymmetry , |m_l=2|^2_+ & = & 2s^2v^4\{_1 ^ 2 + & & .+6(_1 ^ 2-m_1 ^ 2)+3 ^ 2}. for the process @xmath569 one obtains in the case of maximal @xmath0 asymmetry , |m_l=2,t|^2 & = & 2s^2v^4\{_1 ^ 2 . + & & .+6(_1 ^ 2-m_1 ^ 2)+3 ^ 2 } . for small energies , @xmath575 , these matrix elements again reduce to @xmath576 whereas for intermediate energies @xmath577 one finds @xmath578 following @xcite , it has been standard practice @xcite-@xcite to determine @xmath579 by computing the born diagrams for the @xmath63 process with breit - wigner propagator and dropping @xmath580 , the imaginary part of the propagator , since in the zero width limit & = & 1m_1 ^ 2(1d_1(x ) -i 1r_1(x ) ) + & & - i ( s - m_1 ^ 2 ) .
recently , it has been pointed out that this procedure is not correct @xcite . in a toy model
, the same conclusion has been reached in @xcite .
indeed , the described procedure leads to a subtracted squared matrix element which contains terms @xmath581 @xcite , implying @xmath582 on - shell , in contradiction with eq .
( [ correct1 ] ) .
the zero - width limits of the squared amplitude and the squared imaginary part are different .
this was overlooked in the past , leading to an overestimate of the washout rate due to inverse decays by 50% @xcite .
the ris term has to be subtracted from the full @xmath63 cross section , not just from the born cross section , in order to obtain the crucial @xmath0 violating contribution proportional to @xmath40 is @xmath0 violating , using a given @xmath0 asymmetry @xmath583 not determined by the @xmath63 processes .
it is different from the on - shell part of @xmath584 which , as a tree - level rate , conserves @xmath0 .
as discussed above , the correct subtraction term is obtained from the full reaction rate including vertex and self - energy corrections to @xmath558 by separating the on - shell part from the interference terms .
this procedure automatically yields the correct @xmath0 asymmetry @xmath583 . ] .
let us now consider the boltzmann equation for the density of lepton doublets , assuming kinetic equilibrium , @xmath585 & & + \ ; { n_{\bar{l } } \over n^{\rm eq}_l } \gamma^{\rm eq}_{\rm sub}(\bar{l}\bar{\phi}\to l\phi ) - { n_l \over n^{\rm eq}_l } \gamma^{\rm eq}_{\rm sub}(l\phi\to\bar{l}\bar{\phi } ) \nonumber\\[1ex ] & & + \ ; \gamma^{\rm eq}(\bar{\phi}\bar{\phi}\to ll ) - \left({n_l \over
n^{\rm eq}_l}\right)^2 \gamma^{\rm eq}(ll\to\bar{\phi}\bar{\phi } ) \nonumber\;.\end{aligned}\ ] ] here , @xmath586 are the usual reaction densities in thermal equilibrium and we have assumed that the higgs doublets @xmath587 are in thermal equilibrium , neglecting their chemical potential .
the @xmath0 asymmetry @xmath588 is defined in such a way that @xmath589 \gamma^{\rm eq}(n_1\to \bar{l}\bar{\phi } ) = \gamma^{\rm eq}(l\phi\to n_1 ) & = & { 1-\varepsilon_1\over 2}\gamma_{n_1}\;.\end{aligned}\ ] ] further , for the @xmath590 processes we have @xmath591 \gamma^{\rm eq}_{\rm sub}(l\phi\to\bar{l}\bar{\phi } ) & = & \gamma^{\rm eq}_{\delta l=2,+ } + { 1\over2}\varepsilon_1\gamma_{n_1}\;,\\[1ex ] \gamma^{\rm eq}(\bar{\phi}\bar{\phi}\leftrightarrow ll)&= & \gamma^{\rm eq}(\bar{l}\bar{l}\leftrightarrow\phi\phi ) = \gamma^{\rm eq}_{\delta l=2,t}\;.\end{aligned}\ ] ] introducing a lepton , or @xmath11 asymmetry , n_l = n_l^eq - 12
n_b - l , n_|l = n_l^eq + 12 n_b - l , assuming @xmath592 , and keeping only terms @xmath593 , one obtains the kinetic equation for the @xmath11 asymmetry [ boltzm ] dn_b - ldt + 3 h n_b - l = -_1(-1 ) _ n_1 - n_b - l n_l^eq ( 12 _ n_1 + _
l=2 ) , where @xmath594 the @xmath0 violating part of @xmath595 yields the term @xmath596 which guarantees that for @xmath597 no asymmetry is generated .
note that the old procedure for subtracting the ris part of the @xmath63 process would have led to a contribution @xmath598 in the washout term rather than @xmath599 @xcite .
neglecting the off - shell contribution @xmath600 , using the relation _
n_1 = n_n_1^eq z h d = 2 n_l^eq z h w_id , and changing variables from @xmath254 to @xmath13 and from number densities to particle numbers in a comoving volume ( cf .
@xcite ) , one obtains the boltzmann equation ( [ dlg2 ] ) . in order to obtain the kinetic equation ( [ boltzm ] )
the correct identification of the ris term is essential .
it is therefore of crucial importance to derive this equation from first principles . in the case of non - relativistic heavy neutrinos ,
i.e. @xmath58 , this has been done in @xcite .
note that in the strong washout regime , where @xmath541 , the decaying neutrinos are indeed non - relativistic .
( 49 ) of @xcite gives the analogue of ( [ boltzm ] ) for boltzmann distribution functions rather than the integrated number densities .
the starting point of this derivation are the kadanoff - baym equations which describe the full quantum mechanical problem .
leptogenesis is then studied as a process close to thermal equilibrium . as a consequence ,
the deviations of distribution functions from equilibrium distribution functions , @xmath601 and @xmath602 appear from the beginning . here
we have assumed that due to the other interactions in the standard model @xmath603 . ] . for simplicity , in eq .
( 49 ) of ref .
@xcite the contribution to the washout term from interferences with the heavy neutrinos @xmath9 and @xmath10 has been neglected .
otherwise the result is identical to eq .
( [ boltzm ] ) .
in particular , the relative size of the driving term for the asymmetry , which is proportional to @xmath40 , and the washout term due to inverse decays agrees with ( [ boltzm ] ) .
it is important to derive the boltzmann equations and the reaction densities within a full quantum mechanical treatment also for relativistic heavy neutrinos , in particular in the resonance region @xmath604 .
it is instructive to discuss the different contributions to @xmath605 , the reaction density corresponding to the averaged matrix element squared @xmath606 & = & { 4s^2\over v^4}\left\{\mt^2\left({1\over d_1 ^ 2}+{1\over r_1 ^ 2}\right ) -32\pi^2{\mt v^2\over m_1}{1\over x^2}\delta(x-1 ) + 24 \mb^2 \right .
\label{deltalfinal}\\[1ex ] & & + \;\left.\mt^2\left[6+{8\over x}+{2\over x+1 } + { 1\over d_1}\left(3+{2\over x}\right)\right.\right .
\nonumber\\[1ex ] & & \left.-\;{2\over x^2}\left(6x+3+{2\over x+2}+{x+1\over d_1}\right ) { \rm ln}(x+1)\right]\nonumber\\[1ex ] & & \left.-\;6\,m_1 ^ 2\left[2+{1\over x } + { 1\over 2d_1}-{(2x+1)\over x^2 } \ln(x+1)\right]\right\}\;,\nonumber\end{aligned}\ ] ] where we have again assumed the relation ( [ int ] ) . the different contributions to the washout term @xmath34 are shown in fig .
[ truth ] .
the term proportional to @xmath607 , as well as the contributions from the last three lines at high temperatures , @xmath608 , give a simple power law behavior , corresponding to eqs .
( [ lowenergy ] ) and ( [ intermediate ] ) . at low temperatures ,
the term proportional to @xmath609 rapidly approaches zero and becomes negligible .
it can be seen clearly that for @xmath610 the contributions from the first two terms in eq .
( [ deltalfinal ] ) cancel each other to a very good approximation , corresponding to the subtraction of ris contributions . however , the term proportional to @xmath611 has a different low temperature limit than the delta function and cancels against the term in the second and third lines of eq .
( [ deltalfinal ] ) for @xmath612 .
finally , this discussion is only applicable if off - shell and ris contributions can be separated .
this is related to the usual approximation that the right handed neutrinos can be considered as asymptotic free states , i.e. that one can write down a boltzmann equation for them , which is the case if their width is small , i.e. , @xmath613 where @xmath614 is some constant smaller than one .
this translates into the following condition for @xmath615 : @xmath616 we have checked numerically that the separation of on - shell and off - shell contributions works well , as long as eq .
( [ bound ] ) with @xmath617 is fulfilled .
the scattering rates are expressed through the reaction densities rates and these , in turn , through the reduced cross sections , _ , t(s)^(n_1)= _ , t(s)n_n_1^eq = m_132g_n_1 ^ 2 i_,t(s)(z)k_2(z)z^3 , where we introduced the following integrals @xmath618 the reduced cross sections can be written in the following form @xcite : @xmath619 with @xmath620 , @xmath621 and where we defined the following functions : @xmath622 \,\ , , \ ] ] @xmath623 with @xmath624 .
the functions @xmath625 are then defined as : @xmath626 and in this way the eq .
( [ st ] ) for @xmath627 follows . similarly to the eq .
( [ approx ] ) for @xmath82 , the modified bessel function @xmath628 can be approximated by the analytical expression k_3(z ) e^-z ( 945128 + 358z+z^2 ) . for @xmath629 and small @xmath95 , @xmath126
is well described by z_eq = 0.4 + 1.3(1+k^-0.88 ) . | during the process of thermal leptogenesis temperature decreases by about one order of magnitude while the baryon asymmetry is generated .
we present an analytical description of this process so that the dependence on the neutrino mass parameters becomes transparent . in the case of maximal @xmath0 asymmetry
all decay and scattering rates in the plasma are determined by the mass @xmath1 of the decaying heavy majorana neutrino , the effective light neutrino mass @xmath2 and the absolute mass scale @xmath3 of the light neutrinos . in the mass range suggested by neutrino oscillations , @xmath4 ,
leptogenesis is dominated just by decays and inverse decays .
the effect of all other scattering processes lies within the theoretical uncertainty of present calculations .
the final baryon asymmetry is dominantly produced at a temperature @xmath5 which can be about one order of magnitude below the heavy neutrino mass @xmath1 .
we also derive an analytical expression for the upper bound on the light neutrino masses implied by successful leptogenesis . |
numerous studies have convincingly established that at [email protected] rich clusters of galaxies have a higher fraction of photometrically blue galaxies than is found in present - epoch clusters of similar richness ( e.g. , butcher & oemler 1978 , 1984 ; couch & newell 1984 ; macclaren 1988 ; dressler & gunn 1992 ; rakos & schombert 1995 ; belloni 1995 ) .
in addition , follow - up spectroscopy ( e.g. , dressler & gunn 1982 , 1983 , 1992 ; lavery & henry 1986 ; henry & lavery 1987 ; couch & sharples 1987 ; soucail 1988 ; pickles & van der kruit 1991 ) has revealed that many galaxies in distant clusters exhibit strong emission lines , indicating either ongoing star formation or starburst ( sb ) activity , while others show strong balmer absorption lines with little or no emission , indicating either recently truncated star formation or poststarburst ( psb ) activity .
the fraction of the strong balmer line galaxies which have experienced a major starburst as opposed to sudden truncation of star formation is a matter of current debate ( cf .
barger 1996 ; balogh 1997 ; morris 1998 ) .
recently , morphological studies carried out with the _ hubble space telescope _ have shown that a larger fraction of the star forming galaxies are classified as spiral , irregular , or interactions / mergers than in nearby clusters ( dressler 1994a,1994b ; couch 1994 ; wirth 1994 ; couch 1998 ) , a result that confirms the ground - based imaging studies of thompson ( 1988 ) , lavery & henry ( 1988 ) , and lavery ( 1992 ) .
as mentioned above , at z=0 there are relatively few blue galaxies in rich clusters , i.e. , with b - [email protected] mags bluer than the c - m sequence of the red galaxies in the cluster ( e.g. , bower 1992 ) .
furthermore , in present - epoch clusters only a small fraction of galaxies are classified as spirals , at least within the morphological framework that has been developed from mostly isolated nearby spirals .
in fact , even those cluster galaxies classified as spirals are usually found to be depleted in hi content ( e.g. , haynes 1984 and references therein ; giovanelli & haynes 1985 ; gavazzi 1989 ) .
nevertheless , recent work has indicated that the differences in evolutionary processes between nearby and distant rich clusters may not be as extreme as previously considered . moss & whittle ( 1993 ) and moss ( 1995 ) found that a higher fraction of spirals in nearby clusters that are morphologically classified as early type spirals show elevated , and centrally concentrated , star formation levels than in a field sample .
gavazzi ( 1995 ) also found evidence for elevated star formation rates in cluster spirals .
in addition , caldwell et al ( 1993 , 1996 ) and caldwell & rose ( 1997 , 1998 ) have shown that a substantial fraction of early - type galaxies ( i.e. , those morphologically classified as e or s0 ) in nearby rich clusters show signs of psb or sb activity .
finally , koopmann & kenney ( 1998 ) have pointed out that in general spiral galaxies in clusters do not fit consistently within the morphological framework of isolated spirals .
specifically , many cluster spirals with low disk star formation rates that would indicate early - type spiral morphologies in fact have concentrations of light typical of late type spirals .
thus there may exist many late type spirals in clusters at low redshift which have such suppressed star formation rates in their disks ( due to the gas depletion processes present in rich clusters ) , that they are lumped together with early type spirals . in short ,
a perhaps more accurate picture of nearby clusters is that environmentally driven evolutionary processes in galaxies are still ongoing .
these environmental processes may be qualitatively similar to those observed in distant clusters , although they are occurring at a reduced level in intensity and frequency .
the reduced intensity in nearby clusters may be due in part to the depletion in hi reservoir in the general galaxy population since z=0.3 , due to star formation or other removal processes .
more generally , the reduction in intensity and frequency has been discussed within the context of a hierarchical clustering scenario for structure formation by bower ( 1991 ) and kauffmann ( 1995 ) . given that evolutionary processes in cluster galaxies , perhaps similar in nature to those occurring in distant clusters , are indeed ongoing at the present epoch , then the proximity of nearby rich clusters , such as coma , provides an excellent prospect for studying these processes in detail .
in particular , the star formation bursts can be studied at high enough spatial resolution that their cause(s ) might be revealed .
for example , how centrally concentrated are the starbursts ?
are there any young star clusters , such as are found in merging galaxies ( e.g. , ngc 1275 , holtzmann 1992 ; ngc 7252 , miller 1997 ) .
are there morphological peculiarities associated with the starbursts that can be unambiguously linked to tidal disturbances or other external triggering mechanisms ? and what are the morphological differences between sb and psb galaxies ? to that end , we have otained wfpc2 images with the _ hubble space telescope _ ( _ hst _ ) of a number of sb and psb galaxies in the coma and dc2048 - 52 nearby rich clusters .
the morphological characteristics of these galaxies , as revealed by wfpc2 imaging , forms the subject of this paper . the characteristics of the _ hst _ images are described in 2 . in
3 the morphologies and inner luminosity and color profiles of the coma and dc2048 - 52 sb / psb galaxies are presented . in
4 we discuss which mechanism can best account for the sb / psb activity and morphological disturbances revealed by the _ hst _ images .
before describing the _ hst _ wfpc2 images , we briefly summarize how our sample of nine galaxies was selected .
we originally obtained multi - fiber spectroscopy for many galaxies in coma and dc2048 - 52 ( and in three other clusters ) which we believed to be early - type based on morphologies from ground - based images and/or colors ( caldwell 1993 ; caldwell & rose 1997 ) . for a substantial fraction ( typically @xmath115% ) of these galaxies the spectra indicate either current or recent star formation .
the nine galaxies chosen were then selected from among those with spectroscopic evidence for unusual star formation histories .
furthermore , they were chosen to cover a representative range of current and past star formation . finally , while in coma the sb / psb galaxies are primarily located in the sw region of the cluster , in the case of dc2048 - 52 they are much more uniformly spread over the main cluster and the northern subcluster .
wfpc2 images were obtained in the b ( f450w filter ) and i ( f814w filter ) passbands with _ hst _ during cycle 6 for four fields in the coma cluster and three fields in dc2048 - 52 . in each case
a sb or psb galaxy was centered in the pc frame , except in the case of one field in coma , where both the psb galaxy dressler 99 and the sb galaxy dressler 100 were included on a single pc frame .
two 600 second exposures were obtained for each galaxy in the f450w filter , and two 400 second exposures in the f814w filter .
the images were processed using the standard _ hst
_ pipeline routines . to remove cosmic rays , the pairs of exposures were coadded using the ` gcombine ' routine in iraf / stsdas with the ` crreject ' option .
in addition , images of the psb galaxy dressler 216 in coma were extracted from the _ hst _ archive . in this case , the galaxy is located on one of the wf chips , and the passbands are v ( f555w ) and i ( f814 ) .
there are seven 900 second exposures in v and eight 900 second exposures in i. the same processing steps were used for these images .
radial light profiles in both colors were extracted for these galaxies using a program that fits elliptical isophotes ( caldwell & rose 1997 ) .
the elliptical isophotes were found on the i filter image ; intensity data falling on these ellipses were then collected from the bluer image ( b or v ) , after adjusting for a change in center .
calibrations for the f450w , f555w , and f814w frames were those provided in holtzmann ( 1995 ) .
error estimates in the surface brightnesses , magnitudes , and colors were calculated from the photon statistics , the readout noise , and the error in determining the mean sky .
color profiles were found by subtracting the surface brightness profiles of the two filters .
no attempt at matching the point - spread functions of the two filters was made .
the mean color quoted in table [ tab1 ] refers to the integrated color within the radius where the i surface brightness falls to 23 mag arcsec@xmath2 . for reference ,
normal luminous elliptical galaxies have b - [email protected] .
in addition to the _ hst _ images , ground - based images for eight of the nine galaxies have been obtained , either with the 1.2-m telescope of the f. l. whipple observatory for the coma galaxies or with the ctio 0.9-m telescope for the dc2048 - 52 galaxies .
the characteristics of these ground - based images are given in caldwell ( 1993 ) and caldwell & rose ( 1997 ) .
the _ hst _ and ground - based images are displayed in figs . [ pic2 ] , [ pic1 ] , and [ pic3 ] . in fig .
[ pic2 ] we have assembled those galaxies with clear signs of active star formation , while galaxies with relatively smooth morphologies , i.e. , little or no sign of ongoing star formation , are shown in fig . [ pic1 ] .
galaxies which appear to be weak late - type spirals are shown in fig .
each row in these figures shows the _ hst _ images at low and high contrast in the first two panels , followed by the ground - based image in the third panel . as mentioned earlier , the sb or psb nature of the nine galaxies
was originally determined from multi - fiber spectroscopy with either the hydra spectrograph on the kpno 4-m ( for coma ) or the argus spectrograph on the ctio 4-m ( for dc2048 - 52 ) .
the one exception is dressler 61 in coma , for which the psb nature was determined from a spectrum obtained with the mmt .
all of these observations are described in caldwell ( 1993 ) and caldwell & rose ( 1997 ) .
in addition , the sb or psb nature was shown to be centrally concentrated , although spatially extended , from follow - up long - slit spectroscopy of coma galaxies described in caldwell & rose ( 1996 ) .
for d61 in coma , an additional spectrum was obtained with the 1.5-m telescope of the f. l. whipple observatory and fast spectograph ( fabricant 1998 ) at 1.5 / pixel and covering the wavelength region @xmath3@xmath33635 - 7585 .
these spectroscopic observations are referred to in 3 below , where the properties of the nine sb / psb galaxies are summarized .
the morphologies and luminosity profiles of the sb and psb galaxies for which we have hst images are quite diverse . as a result
, we begin with an individual discussion of each galaxy , and in 4 draw general conclusions about the starburst nature of these galaxies . for convenience of organization
, we begin with the sb galaxies of fig .
[ pic2 ] , then proceed to the psb galaxies of fig .
[ pic1 ] , and finish with the weak late - type spirals of fig .
the basic parameters for each galaxy discussed below are summarized in table [ tab1 ] . for the purposes of the following discussion
, it may be useful to know that in coma , which we assume to be 100 mpc distant ( corresponding to = 70 mpc@xmath4 , and farther than the 86 mpc we assumed in previous papers ) , 1= 490 pc , and for dc2048 - 52 ( at 200 mpc ) , 1= 980 pc .
hence the multi - fiber spectra , which cover 3 in diameter on the coma cluster galaxies and 2diameter on the galaxies in dc2048 - 52 , provide integrated spectra of typically the central @xmath12 - 3 kpc in diameter of these galaxies . _
d15 in coma : _ + the spectrum of dressler 15 in coma ( hereafter referred to as coma - d15 ) reported in caldwell ( 1993 ) and caldwell ( 1996 ) reveals strong emission lines that are a mixture of an hii region spectrum and an agn spectrum .
the emission is very centrally concentrated , but spatially resolved out to about 2 , and has a velocity gradient with a full amplitude of 100 .
thus coma - d15 is currently undergoing a central starburst and also harbors an agn .
a nearby apparent `` companion '' galaxy is in fact a background galaxy with redshift of z=0.127 . on the basis of ground - based images ,
coma - d15 has been classified as an s0 galaxy ( dressler 1980 , caldwell et al .
1993 , andreon et al .
the wfpc2 images in fig .
[ pic2 ] , however , reveal that the central bright `` bulge '' is resolved into a spiral disk , where the distance from the nucleus to the inner bright spiral arms is @xmath11.5 , or @xmath10.75 kpc .
the spirality extends to only 3(1.5 kpc ) , which explains why it was missed on the ground - based images .
this bright inner spiral structure is similar to that seen with _
hst _ of some other s0 galaxies , e.g. ngc 524 and ngc 3599 ( lauer 1995 ) , though the spiral in coma - d15 is stronger , and better resembles that seen in the merger product galaxy ngc 7252 ( whitmore et al .
in fact , the radial extent of the spiral arms in both galaxies is very similar ( @xmath5 kpc ) .
there are however no large tidal tails in coma - d15 , as are seen in ngc 7252 , nor is there a companion galaxy evident . the radial luminosity profile of coma - d15 is shown in fig .
[ sb_pro ] . where the luminosity profiles in b and i are plotted versus radius ( @xmath6 ) and @xmath7 .
the b - i color profile is also plotted .
the thin solid lines represent @xmath81 @xmath9 errors for the b light and color profiles .
an additional systematic error from the calibration may be present at the level of 5% , which has no effect on our results .
the radial luminosity profile shows a strongly peaked light distribution inside the central 1 in radius , then a bright `` bump '' in the profile caused by the spiral arms , and then a basically exponential disk beyond a radius of @xmath12 .
the radial color profile reveals a blue nucleus surrounded by a substantially redder ( by @xmath10.4 mag in b - i ) disk , which then gives way to the blue spiral arms at @xmath11.5 . _
d100 in coma : _
+ dressler 100 in coma ( hereafter coma - d100 ) is an apparent nearby companion to coma - d99 which has been classified as an irregular and an sba from ground - based images ( dressler 1980 , andreon et al .
1996 , respectively ) .
although coma - d99 and coma - d100 are projected to be only 17(8.5 kpc ) apart , they have a large velocity difference of 4700 .
thus , even if they are indeed experiencing a close passage , they can not be a strongly interacting pair in the traditional sense of low encounter velocity interacting / merging galaxies .
the long - slit spectrum of coma - d99 from caldwell ( 1996 ) was oriented at a position angle of 121@xmath10 e of n , in order to pass through the center of coma - d100 as well .
this spectrum reveals that coma - d100 has an active star formation spectrum in the central @xmath12(@xmath11 kpc ) in radius , i.e. , [ oiii]@xmath35007 and h@xmath11 emission can be seen out to this radius ; strong underlying balmer absorption can also be seen in the higher order balmer lines . at @xmath13radius ,
the spectrum is characteristic of a psb state , i.e. , no emission is detected but strong balmer absorption lines are clearly evident .
we have applied the ca ii h + h@xmath12 versus h@xmath13/fe i age dating technique that was developed in leonardi & rose ( 1996 ) and applied to other coma cluster psb galaxies in caldwell ( 1996 ) and caldwell & rose ( 1998 ) .
this technique , which decouples the mostly degenerate effects of psb age from burst strength ( for a fading burst superposed on an underlying `` old '' galaxy ) , relies on the high sensitivity of the relative strength of h@xmath12 versus the ca ii h and k lines to the contribution from a stars . using this age dating method
, we find that the age of the outer psb region of coma - d100 , summed over both sides of the slit , is @xmath10.25 gyr post starburst ( see fig . [ age_ps ] ) .
thus , while star formation is still proceeding vigorously in the central 1 kpc of coma - d100 , it has already subsided at a radius of @xmath11.5 kpc .
such is seen in other nearby sb galaxies such as ngc5253 ( caldwell & phillips 1989 ) .
our wfpc2 images certainly support an unusual morphology for coma - d100 ( see fig .
[ pic2 ] ) . a strong two - armed spiral pattern is visible as well as highly visible dust lanes .
a bright nucleus is partially obscured by heavy dust in the central @xmath10.5 in radius .
the two - armed spiral pattern is visible out to 3(1.5 kpc ) in radius .
a set of parallel dust lanes project out on the east side of coma - d100 , extending out to @xmath13.4 .
dust predominates on the west side in the central @xmath10.5 in radius , but is more prominent on the east side at larger radii .
the dust shows up prominently in the color profiles , causing a redder color around 1 - 1.5than the colors inward and outward of that location . finally , from an mmt long - slit spectrum we measured a velocity difference of only 14 between the nw and se side of the slit at a distance of 3from the nucleus , which is within the measurement errors of @xmath1@xmath820 .
thus we can conclude that there is little rotation at the 121@xmath10 position angle .
this is smaller than we would expect to see for a disk galaxy measured @xmath163@xmath10 away from the galaxy s major axis , though we hestitate to call it unusual . _
d45 in coma : _
+ dressler 45 in coma ( hereafter coma - d45 ) was originally determined to be a starburst galaxy in caldwell ( 1993 ) , because of its strong emission lines characteristic of an hii region spectrum as well as enhanced balmer absorption that is evident in the higher order balmer lines .
the kinematics of the ionized gas was studied in caldwell ( 1996 ) , where an asymmetric rotation curve was discovered from long - slit mmt spectra at four position angles .
coma - d45 was classified as sa from ground - based images , and an r band image presented in caldwell ( 1996 ) tended to confirm that morphology , while revealing an asymmetric structure to the outer isophotes .
we must correct an error in the color reported in caldwell et al .
the b - r color of coma - d45 from the gmp paper was thought to be quite red , unusual for an sa galaxy ( figure 12 in caldwell et al . 1993 ) .
a coding error was the cause of this statement : in fact gmp did not report a color for coma - d45 .
a similar problem occured for coma - d15 ; there were no other such errors .
the colors for both coma - d45 and coma - d15 are bluer than previously reported , substantially so in the former case , as can be surmised from table [ tab1 ] .
the wfpc2 images of coma - d45 reveal that the central `` bulge '' of this galaxy is actually a highly irregular structure , with several bright knots ( presumably hii regions ) forming a partial ring with a diameter of @xmath11 , or about 0.5 kpc . as can be seen in fig .
[ pic2 ] , the structure of coma - d45 is irregular on all scales , including the fainter isophotes mentioned above in the ground - based images . the radial luminosity profile ( fig .
[ sb_pro ] ) shows two exponentials of different scale lengths , with the break occurring at @xmath12 - 3 .
the highly irregular morphology of coma - d45 revealed in the wfpc2 images may clarify some of the asymmetric nature of the rotation curve reported in caldwell ( 1996 ) , in that the kinematic center of the galaxy may be well displaced from the emission line peak intensity . in that case , the rotation curve can be seen as reasonably symmetric around a displaced center within the central @xmath12 in radius . outside of that radius , however , the gas appears to be counter - rotating . _
d99 in coma : _
+ high s / n mmt spectroscopy of dressler 99 in coma ( hereafter coma - d99 ) shows that it is a psb galaxy seen @xmath11.0 gyr after the starburst .
the long - slit mmt spectra showed that the strong balmer absorption lines are confined to the inner 1 in radius . on ground - based images
the galaxy has been classified as s0 or e / s0 .
( dressler 1980 , caldwell 1993 ) . the wfpc2 images ( see fig .
[ pic1 ] for the poststarburst galaxies ) show a smooth light distribution , with no indications of spiral structure .
the bright nuclear `` bulge '' , which is @xmath10.5 , or @xmath10.25 kpc , in radius , is surrounded by what appears to be a smooth lower surface brightness envelope .
the radial luminosity profile ( fig .
[ psb_pro ] ) confirms the bulge / disk character of the galaxy .
the radial color profile reveals the central area ( but outside of 0.2 ) to be 0.3 - 0.5 mag bluer in b - i than the surrounding disk , thus confirming the earlier conclusions from ground - based imaging and spectroscopy that the starburst was highly concentrated in the center of coma - d99 .
the inner few pixels ( @xmath140.2 ) of the b image show some structure that may indicate the presence of an off - nuclear star cluster ( this feature appears on both b images , and thus is not an artifact ) .
interestingly , the i image shows a more peaked nucleus than the b image , which is unresolved in a deconvolved image
. a nuclear star cluster may be present , somewhat reddened , for the color of the nucleus alone ( within 0.1 ) is b - i=1.6 , substantially redder than the surrounding area .
interestingly , the central surface brightness of coma - d99 is quite low , at around b=18.3 mag arcsec@xmath2 , or an estimated v=17.6 mag arcsec@xmath2 .
the expected value for a moderate luminosity e / s0 like coma - d99 is much higher , around v=15 ( as is found for an archival sample of e / s0 s observed with hst ) .
only high luminosity galaxies such as ngc 4889 tend to have similar central surface brightnesses .
the fact that a nucleus shows up only in the i band may indicate that reddening is partly responsible for the low observed central b surface brightness , though high reddening is not indicated by the overall blue color of the central regions .
the low observed central surface brightness must for the most part be real , and may be a clue to the mechanism for the starburst in this galaxy . _
d216 in dc2048 - 52 : _
+ on the basis of ground - based images , dressler 216 in dc2048 ( hereafter dc2048-d216 ) is classified as an s0 .
it was classified as psb by caldwell & rose ( 1997 ) on the basis of strong balmer absorption lines and only a small and uncertain amount of [ oii]@xmath33727 emission .
the wfpc2 images reveal a very smooth and mostly symmetric morphology ( see fig
. [ pic1 ] ) .
the only asymmetric structure is a slightly curved barlike ( i.e. , integral sign ) structure that starts outside the bright nucleus and can be traced out to @xmath11.2 in radius ( 1.2 kpc ) . beyond 1.2 in radius ,
the morphology is smooth with no asymmetry . the radial luminosity profile ( fig .
[ psb_pro ] ) is that of a disk galaxy with a bulge within 1 .
the color profile indicates that the disk becomes slightly bluer towards the center , while the bulge may be yet bluer , though the different resolutions in the b and i frames may be affecting the color profile there .
there is a companion spiral to this galaxy , not shown in the figure , located about 11away , but for which we have no redshift .
this galaxy is normal in appearance , without any signs of tidal interactions , so we may presume it did not interact with dc2048-d216 and cause the latter s starburst . _
d61 in coma : _
+ high s / n spectroscopy of dressler 61 in coma ( hereafter coma - d61 ) with the mmt reveals a very strong psb spectrum , which is extended in radius out to at least 3 , and which has a small ( 4 equivalent width ) component of [ oii]@xmath33727 emission . from an integrated spectrum obtained with the 1.5-m telescope at mt .
hopkins over the central 36 in diameter , and covering h@xmath15 and [ nii]@xmath36548,6584 , we found that the equivalent width of the combined [ nii]@xmath36548 and @xmath36584 emission is @xmath17.5 .
h@xmath15 absorption is seen to be exactly balanced by emission . from the equivalent widths of the higher order balmer absorption lines ( which are relatively free of emission ) we estimate that the underlying h@xmath15 absorption is @xmath15 .
thus the ratio of [ nii ] to h@xmath15 emission is slightly greater than 1 , which is typical of liner emission line spectra , and definitely in excess of even the most metal - rich hii regions ( e.g. , baldwin 1981 ; veilleux & osterbrock 1987 ) .
thus the weak emission spectrum is due to nuclear activity , and not to a small rate of residual star formation , which strengthens the case that coma - d61 is indeed a psb galaxy . using the previously mentioned leonardi & rose ( 1996 )
age dating technique , we have determined the psb age of coma - d61 to be @xmath10.5 gyr , as can be seen in fig . [ age_ps ] .
this is a considerably shorter psb time than the typically 1 gyr for other psb galaxies in coma to which we have applied this technique . from ground - based imaging coma - d61
has been classified as s0 .
the wfpc2 images ( fig .
[ pic1 ] ) reveal it as an edge - on disk galaxy with a very bright nucleus .
two radial dust lanes are apparent about 1to the east of the center .
coma - d61 has an unusual bulge : it is boxy or peanut - shaped , and additionally has an x morphology that is either part of the box shape , or superposed on that . a popular parameter to characterize
the degree of boxiness is the a4 parameter , the cos ( 4@xmath16 ) deviation from a perfect ellipse , normalized to the semimajor axis and local intensity gradient .
a4 is about -0.02 in the bulge area on coma - d61 .
after deconvolving the image using the lucy - richardson technique ( lucy 1974 ; richardson 1972 ) , it is apparent that the nucleus is unresolved ( see fig . [ d61.nuc ] ) , with the surface brightness steeply falling from b=12.8 mag arcsec@xmath2 to 16.1 within 0.09 .
it is not clear how much the agn continuum adds to the central luminosity , but if the light comes purely from a nuclear star cluster , its luminosity is m@xmath17 within 3 pixels ( 0.14or 66 pc ) , which represents about 7% of the galaxy s total light .
this luminosity is higher than any of the star clusters formed in the merging galaxies studied by whitmore et al .
( 1997 ) , or by carlson et al .
( 1998 ) , and is also higher than the nuclear star clusters found in normal early - type galaxies by lauer et al .
the color of the nucleus is b - i=1.20 , about 0.2 mag bluer than that of the surrounding bulge .
the x - shape is much more clearly seen in fig .
[ d61 ] , where the overall axisymmetric component of the bulge light profile has been subtracted off , by fitting ellipses to the isophotes ( the disk light was excluded from the fit ) .
the x - shape can be traced out to at least 1.8(0.9 kpc ) outside the nucleus .
the x is not affected by diffraction spikes from the bright nucleus because the opening angle of the x is 110 , not the 90expected from the orthogonal secondary support vanes . moreover ,
the x is better aligned with the disk of coma - d61 than it is with the known orientation of the diffraction spikes on wfpc2 images .
the disk is also unusual , in that it does not extend into the center .
this feature may be what is referred to as a lens ( kormendy 1992 ) , in that a shelf , or constant surface brightness area in the profile is present , at least in the unprocessed profile ( fig .
[ d61_pro ] ) . in the image in which the bulge has been removed
, the profile along the major axis does not show a shelf for the disk , rather the area shows a rise and fall in surface brightness with radius .
the maximum brightness occurs at the same radius on both sides of the nucleus , so the feature is most likely a ring rather than a spiral arm .
this `` ring '' has about the same color as the bulge at that radius ( b - i=1.5 ) . on the other hand
, the x feature appears to be somewhat bluer , with b - i=@xmath18 , though there is a strong systematic error in this value because of the difficulty in subtracting off the elliptical bulge in exactly the same way for the images in the different filters .
still , the blue color supports the idea that the stars contributing to the x - like distribution are young .
finally , the mmt spectrum mentioned above was oriented along the major axis of coma - d61 , and even though the spectrum was not very deep , we have extracted both stellar and gas kinematics from it .
the stars rotate with a velocity of about 65 out to 3(1.5 kpc ) from the nucleus .
similarly , the gas shows a rotation of about 50
. it would be interesting to pursue the kinematics of this object with a more detailed study . _
d216 in coma : _
+ multifiber spectroscopy of dressler 216 in coma ( hereafter coma - d216 ) indicates a weak psb nature for this galaxy , which has been classified as sa by dressler ( 1980 ) from ground - based images .
the archival wfpc2 v ( see fig . [ pic1 ] ) and i band images reveal a large bulge as expected for its classification , and well - defined smooth spiral structure out to a radius of nearly 14(7 kpc ) .
the radial luminosity profiles indicate that the disk component only dominates outside a radius of 4(see fig .
[ psb_pro ] ) .
in addition , the bulge is considerably redder than the disk , which appears to be consistent with its weak psb status .
a final comment regarding coma - d216 is that the archival _ hst _ image of the bright `` e+a '' galaxy in a665 at z=0.18 studied by franx ( 1993 ) bears a strong morphological resemblance to coma - d216 .
an image of this galaxy is shown in franx ( 1997 ) , revealing it to have spiral arms surrounding a large bulge . _
d148 in dc2048 - 52 : _
+ multi - fiber spectroscopy of dressler 148 in dc2048 - 52 ( heferafter dc2048-d148 ) revealed a psb spectrum .
however , the s / n in the region of [ oii]@xmath33727 is sufficiently low that we can not exclude a small equivalent width in that emission line , hence a small amount of ongoing star formation .
dc2048-d148 is classified as s0 on ground - based images .
the wfpc2 images ( see fig .
[ pic3 ] ) reveal a notably different morphology .
a central light cusp is surrounded by a low surface brightness spiral structure out to a radius of @xmath17 , or @xmath17 kpc .
faint hii regions can be seen within a largely flocculent spiral arm pattern .
the radial luminosity profile is dominated by the central light cusp in the inner @xmath11 in radius , where an r@xmath19 law profile is evident ( see fig . [ psb2_pro ] ) . for radii
greater than @xmath13 , an exponential disk is the primary contributor .
the color is essentially constant in the central 1 , and then reddens by about 0.2 mag in b
i out to a radius of @xmath14 .
the mean b - i color is red for a late - type spiral , as is the case for d192 below . _
d192 in dc2048 - 52 : _ + multi - fiber spectroscopy of dressler 192 in dc2048 - 52 ( hereafter dc2048-d192 ) reveals strong balmer absorption lines with weak ( @xmath145 equivalent width ) emission in [ oii]@xmath33727 . we do not have a spectrum in the red that can distinguish whether this emission is due to a small amount of ongoing star formation or to an agn
dc2048-d192 was classified as psb in caldwell & rose ( 1997 ) , due to the very strong balmer absorption and only weak [ oii]@xmath33727 emission .
however , the wfpc2 images indicate that star formation is still ongoing in the central @xmath11.25(@xmath11.25 kpc ) in radius . as can be seen in fig .
[ pic3 ] a central bright nuclear component is surrounded by a relatively bright star forming region , which is punctuated by hii regions and/or bright star clusters and dust lanes .
a fainter asymmetric spiral pattern can be traced out to @xmath13.75 , or @xmath13.75 kpc in radius .
the luminosity profile for dc2048-d192 ( fig .
[ psb2_pro ] ) is dominated by an exponential disk beyond @xmath13 , while a more centrally concentrated component is the primary contributor at smaller radii .
the radial color map ( fig . [ psb2_pro ] ) shows the nucleus to be slightly redder than the surrounding region , except for a red area at a radius of @xmath12from the nucleus , at the location of an apparent major dust lane in the image .
at this point we can summarize our knowledge of the sb and psb galaxies in nearby clusters as follows . from ground - based multi - fiber and long - slit spectroscopy
we have found that some galaxies , which from ground - based images are classified as early - type galaxies , have signs of unusual star formation histories , both in terms of the star formation being elevated in the recent past and that star formation being centrally concentrated .
the ground - based images indicate that all of these galaxies are disk galaxies . from the hst images , we find that the sb galaxies have spiral structure . in some cases ( coma - d15 ) the spiral structure
is restricted to the inner couple of kpc , while in others ( coma - d45 and coma - d100 ) it is global , but in all cases the star formation is centrally concentrated .
most of the psb galaxies have smooth morphologies , perhaps because their time since the cessation of the starbursts is already @xmath11 gyr ( caldwell et al . 1996 ) .
however one psb galaxy ( coma - d61 ) exhibits an x - shaped bulge , which is interesting since the starburst age of this galaxy is relatively young , at @xmath10.5 gyr .
none of the observed galaxies show large collections of young star clusters such as are seen in some merger products ( holtzmann 1992 , miller 1997 ) , which may be an indication that major mergers are not the mechanism involved here .
the two late - type spiral galaxies are a cause for uncertainty .
these might be related to anemic galaxies , in which a starburst has not actually occured ; rather , the normal disk star formation in such galaxies has been truncated .
more detailed spectroscopy than is currently available is needed to sort out this problem .
although only nine sb / psb have been studied so far , it seems reasonable to propose an evolutionary sequence , for example , from coma - d15 to coma - d61 to coma - d99 .
thus the starburst which is centrally located but extended would eventually fade , resulting in a galaxy with smoother appearance , but with spectroscopic evidence for the past starburst .
we caution that precursors might well have a large range in morphologies , and thus the resultants might also have a range in appearances .
the central issue we now address is what these observed morphologies allow us to conclude about the source of the sb / psb activity in nearby rich clusters .
the mechanisms that have been proposed to explain the rapid evolution in galaxies taking place in distant clusters tend to fall into two basic categories : external gravitational perturbation of some kind , ( e.g. , tidal interaction or merging ) or a gas removal mechanism ( e.g. , ram pressure stripping or ablation ) .
ram pressure stripping is undoubtedly an important factor in the evolution of cluster galaxies , and probably plays a major role in the observed hi depletion of spiral disks ( e.g. , gunn & gott 1972 ; cowie & songaila 1977 ; dressler 1985 ) .
however , given the central concentration of the star formation and the non - axisymmetric morphologies present in some cases , which are difficult to understand in terms of a `` continuous '' gas removal process , we naturally tend to consider the external gravitational perturbation mechanism as the more likely of the two possibilities as a source of the observed central star formation .
such a perturbation could arise from a single strong tidal interaction and/or merger from either an equal mass perturber or a minor mass galaxy .
alternatively , the tidal perturbation could be produced by the global cluster gravitational potential ( byrd & valtonen 1990 ; valluri 1993 ; henriksen & byrd 1996 ) , or , in the case of `` galaxy harassment '' , from the combined effects of many discrete minor disturbances and of the global cluster potential ( moore 1996 , 1998 ) .
we first consider strong tidal interactions and/or mergers of equal mass galaxies .
observations of galaxies in distant clusters have revealed the signatures of major encounters / mergers in the form of long tidal tails in a substantial number of cases ( e.g. , lavery & henry 1988 ; couch 1998 and references therein ) .
these signatures can be taken as evidence for such mergers , although not all the observed cases are clearly due to mergers ( couch 1998 ) . however , among the sb / psb galaxies that we have imaged in nearby clusters , there are no signs of such major interactions / mergers ; we find no cases where long tidal tails are evident , nor do we find signs of a major companion with similar radial velocity . in the three sb / psb cases where nearby `` companions '' do appear to be present , two `` companions '' have in fact very different velocities .
coma - d99 and coma - d100 have a radial velocity difference of 4700 km / sec and ( 2 ) the nearby `` companion '' to coma - d15 is a background galaxy with z=0.127 . in the third case , dc2048-d216 , the `` companion '' has no measured redshift , but that galaxy itself is a normal looking late - type spiral , with no indication of tidal interactions . as well , dc2048-d216 has a smooth appearance indicating that a significant time has elapsed since the putative tidal iteraction would have occured .
thus we do not consider strong tidal interactions or mergers between equal mass galaxies to be the primary mechanism for inducing starbursts in the nearby rich clusters .
mergers between a galaxy and a low - mass companion ( hereafter referred to as a minor merger ) provide a more promising prospect .
mihos & hernquist ( 1994 ) and hernquist & mihos ( 1995 ) have carried out numerical simulations of minor mergers which include gas dynamics and star formation .
a key result of their simulations is that gas in the main galaxy loses angular momentum due to torques exerted by non - axisymmetric disturbances in the stellar component which are generated by the tidal passage and infall of the satellite galaxy . as a result , a great deal of gas piles up in the central region of the main galaxy , leading to a burst of star formation there .
the central concentration of the star formation and the magnitude and duration of the star formation burst depend on the bulge - to - disk ratio of the main galaxy .
however , in general the star formation is concentrated within the central @xmath11 kpc of the main galaxy , and the starburst lasts for typically @xmath1100 150 myr ( mihos & hernquist 1994 ) .
some of their simulations bear a striking resemblance to d100 and d15 in coma .
specifically , in fig .
4 ( for simulation of the gas ) and fig .
3 ( for simulation of the stars ) of hernquist & mihos ( 1995 ) , at a time step of 43.2 the main galaxy looks very similar to d100 . at time steps greater than 80 , there is considerable resemblance to d15 . in addition , both whitmore & bell ( 1988 ) and mihos et al .
( 1995 ) have discussed x shapes in the bulges of galaxies and have proposed that they are produced by the disturbance created when a minor merger takes place .
the mihos et al .
( 1995 ) simulation of a minor merger viewed edge - on ( see their fig .
3 ) is a close match to our _ hst _ image of d61 in coma . finally , as can be seen from the edge - on views of minor merger simulations published in fig . 3 of hernquist & mihos ( 1995 ) , minor mergers heat the disk of the main galaxy , but do not lead to its disruption , which is in agreement with the fact that the psb galaxies in coma and dc2048 - 52 still exhibit surviving disks .
given that the minor merger simulations of mihos & hernquist are capable of generating the kind of morphological signatures and central star formation that characterize some of our _ hst _ images , we evaluated the possibility that minor mergers are the _ primary _ source of the star forming early - type galaxies in nearby clusters .
specifically , we are interested in reproducing the situation in the sw region of the coma cluster , where many psb galaxies were found by caldwell ( 1993 ) , a spatial / kinematic subcluster has been clearly identified ( colless & dunn 1996 ; biviano 1996 ) , and a secondary peak in the cluster x - ray emission has been discovered ( watt 1992 ; briel 1992 ; white 1993 ) .
while there is much current debate concerning whether this subcluster , with its associated x - ray and starburst phenomena , is currently falling into the main coma cluster or has already passed through ( e.g. , burns 1994 ; colless & dunn 1996 ) , for the purposes of this discussion we will assume the latter case .
thus we speculate that the starbursts in the coma psb galaxies were triggered by minor mergers which occurred when the subcluster encountered the numerous small galaxies in the central part of the coma cluster .
we tested this hypothesis by running n - body simulations of colliding clusters of galaxies using the publicly available nemo stellar dynamical software ( teuben 1995 ) .
the simulations are similar to those described in caldwell & rose ( 1997 ) , in which the evolution of a subcluster falling into a main cluster is followed .
as before , we are only concerned with the gravitational interactions of the particles , so the hydrodynamics of the icm are not considered . a detailed description of the simulations , and of how we identified the minor merger events , is given in the appendix .
the results of the simulations , which are also described in the appendix and summarized in table a.1 , show that the frequency of minor mergers is far too low to explain the large fraction of sb / psb galaxies in the sw region of coma . as mentioned earlier ,
an alternative way to produce morphological disturbances in cluster galaxies through external gravitational influence is via the `` galaxy harassment '' mechanism of moore ( 1996 , 1998 ) . here
the combined effects of frequent impulsive perturbations from small galaxies in high - speed encounters and the changing global tidal field of the cluster produce similar disturbances to the minor merger scenario . given
that harassment is a continual process for any cluster galaxy , we can expect that the morphological and star formation signatures of the harassment process should be evident in those galaxies which still have disks containing significant amounts of hi .
thus galaxies in subclusters which are infalling for the first time into main clusters , and hence have not yet ( completely ) lost their gas supply in the hostile cluster environment , are good candidates to produce visible reaction to the harassment process .
in fact , fujita ( 1998 ) has recently shown from a simple model of pressure - enhanced star formation that galaxy harassment does indeed lead to elevated rates of star formation .
in short , galaxy harassment could provide a natural explanation for the appearance of the many sb / psb galaxies in the sw region of coma , where the ngc 4839 subcluster appears to be interacting with the main cluster .
of course , one can also expect to see isolated examples of harassed galaxies , since any disk galaxy with a gas supply that is falling into the cluster can in principle go through the evolutionary stages that appear to be present among the sb and psb galaxies in coma and dc2048 - 52 .
several questions remain about the applicability of galaxy harassment to the sb / psb galaxies in nearby clusters .
this process has been used to primarily describe the evolution of the small galaxies in clusters , whereas the coma and dc2048 - 52 galaxies that we have imaged are quite substantial , typically only a magnitude fainter than l*. it is important to evaluate the process for these more massive galaxies , with particular regard to the end state produced .
specifically , our hst images and ground - based images indicate that the psb galaxies still have a prominent disk , thus observationally the disks do survive the perturbation .
but do simulated harassed disks survive ? as well , the hst images indicate that the morphologies of the sb / psb galaxies are quite heterogeneous , indicating that a range in precursor morphologies are being processed into a still considerable range in resultant morphologies .
it is important to simulate the effect of harassment on a reasonably wide range of input disk morphologies .
the effect of harassment on a wide range of initial disk luminosity ( mass ) must be considered as well . in caldwell & rose ( 1998 ) we found two low - luminosity ( b@xmath20 ) coma galaxies in a recent psb state .
they are clearly s0 s , as indicated by their light profiles , hence harassment must not in general lead to the destruction of disks in smaller galaxies if it is to explain these small galaxies in coma .
finally , can harassment produce the x morphology that is so evident in coma - d61 ?
to summarize , _ hst
_ wfpc2 images of nine sb / psb galaxies in the coma and dc2048 - 52 nearby rich clusters have revealed that the morphologies of the galaxies are rather heterogeneous .
however , the sb galaxies tend to have ongoing highly centralized star formation in a region typically @xmath11 - 2 kpc in radius that appears as a bright `` bulge '' on ground - based images . in contrast
, the psb galaxies for the most part have substantially smoother morphologies than their sb counterparts , but with blue colors in their central @xmath11 - 2 kpc in radius .
these findings , coupled with the ground - based spectrosocpy reported in earlier papers , are generally consistent with a picture in which the sb galaxies are presently experiencing enhanced star formation in their central regions , and ultimately will fade into the the smoother morphologies represented by the psb galaxies , with lingering evidence for past central starbursts . while the sample of nine galaxies is still small for drawing general conclusions about the mechanism that triggers the starbursts , in several galaxies we see evidence for an external tidal perturbation as the likely cause of the centralized starburst .
perhaps the `` galaxy harassment '' mechanism of moore ( 1996 ) is capable of providing such tidal perturbations .
the fact that some galaxies in nearby clusters appear to be normal s0 s or ellipticals on ground - based images , but with the increased spatial resolution of _ hst _ are revealed to have centrally concentrated recent or ongoing star formation , has implications for morphological studies of distant galaxies . in the typical [email protected] cluster
for which _ hst _ images can now provide comparable physical resolution as ground - based images of coma cluster galaxies , one may now question whether some galaxies classified as early - type from _ hst _ images actually harbor ongoing or recent starbursts in their central regions . finally , to establish a clearer picture of how a galaxy is excited into a starburst and then fades through the psb phase into a `` normal '' early - type galaxy , we need to increase the sample of sb / psb galaxies beyond the nine observed for this paper to several dozen .
a larger sample is particularly necessary if we are to assess how evolution through the sb and psb phase depends on the initial morphology of the precursor galaxy .
such information would establish an observational framework for testing the effects of tidal perturbations on galaxies within the dense cluster environment .
this research was supported by grant no .
go-06773.01 - 95a ( cycle 6 ) from the space telescope science institute to the university of north carolina and by nsf grant ast-9320723 to the university of north carolina .
the numerical simulations used in 4.2.1 to assess the frequency of minor mergers in clusters were carried out using the routine `` hackcode1 '' , which is the standard nemo n - body integrator based on the barnes & hut ( 1989 ) hierarchial tree algorithm .
the nemo stellar dynamical software is publicly available on the world wide web . because we would like to compare the simulations to our observations , it is important to know to what physical units the nemo units correspond .
the nemo units are defined as follows : 1 ) the gravitational potential of each cluster is @xmath21 , 2 ) the kinetic energy is @xmath22 ( therefore the cluster is virialized ) , 3 ) the total mass of the cluster is @xmath23 and 4 ) the gravitational constant is @xmath24 . noting that @xmath25 where @xmath26 is the gravitational radius , we find that @xmath27 , and from the kinetic energy requirements , @xmath28 , where @xmath29 is the rms velocity dispersion of the galaxies . observations of the x - ray gas ( fusco - femiano & hughes 1994 ) and of the galaxy distribution ( kent & gunn 1982 ) show that the core radius of coma is @xmath30 kpc and @xmath31 kpc respectively . using the fact that for a plummer model ( which we adopt here for its mathematical convenience ) , @xmath32 and an estimated distance to the coma cluster of 100 mpc , we find an average @xmath26 on the order of 1 mpc implying that a nemo unit of distance is 500 kpc .
the inferred 3-dimensional rms velocity dispersion for coma is approximately 1700 km / s ( colless & dunn 1996 ; biviano et al 1996 ) implying that a nemo unit of velocity is 2404 km / s . a unit of time in nemo is defined as 0.4 crossing time , which , for the coma cluster with a crossing time of 1 gyr , corresponds to 0.4 gyr . in general , the simulations consisted of two clusters : a main cluster with @xmath15000 particles and a less massive subcluster with 256 particles .
the point of the simulations was to assess how large a fraction of the galaxies in the infalling subcluster would experience a minor merger with one of the small galaxies in the main cluster .
the mihos & hernquist ( 1994 ) and hernquist & mihos ( 1995 ) simulations used a mass ratio between large galaxy and satellite galaxy of 10:1 . while it is not entirely clear what is the minimum mass ratio that can still produce a large tidal effect on the larger galaxy , we adopted a ratio of 20:1 as a reasonable estimate .
consequently , our simulations are designed to look for minor mergers between `` large '' galaxies in the subcluster and `` small '' galaxies in the main cluster . given that the typical psb galaxy in the sw region of the coma cluster has b = 16.5 , we expect the typical minor merger to be with a main cluster galaxy with b = 19.75 . from recent analyses of the coma cluster luminosity function
( bernstein 1995 ; lobo 1997 ; secker 1997 ) we estimate there to be approximately 1200 galaxies with b @xmath33 19.75 . hence in simulating the coma cluster with @xmath15000 particles we overestimate the frequency of minor mergers by a factor of approximately 4 .
however , by using such a large number of particles in both cluster and subcluster we obtain better statistics on the merger frequency and then can adjust the merger rates accordingly . in our simulations each cluster
was constructed from a virialized plummer model with a softening parameter for each particle of 0.05 nemo units ( equivalent to 25 kpc ) to account for the extended mass distributions of the individual galaxies .
the two structures were initially placed 4 distance units apart with a variety of purely radial subcluster infall velocities ranging from 1/4 parabolic to 1/16 parabolic . in most cases , the mass ratio of the two clusters was 4:1 .
the subcluster was constructed of only one type of particle each with a mass of @xmath34x@xmath35 nemo mass units , so that the total mass of the subcluster was 1 . for the main cluster we used 5120 particles each with a mass of @xmath36x@xmath37 , i.e.
, approximately 20 times less massive than the particles in the subcluster . then to bring the total mass of the main cluster to four times larger than the subcluster , we inserted 100 particles each of mass @xmath38x@xmath39 .
all simulations were run for 10 time units ( i.e. 4 gyr total ) with 15 increments per each 0.4 gyr time unit . at times near 4.5 , the subcluster is just passing through the center of the main cluster and at times near 9 , the subcluster has almost halted and is about to fall back into the main cluster for the second time . because we are only interested in the first passage of the subcluster
, the simulations were not continued beyond this point . to increase the statistical reliability of the results
, we ran each simulation four times , with different random initial positions and velocities for the individual particles within the two virialized plummer models . for each iteration , nemo outputs the mass , the x - y - z spatial coordinates and the x - y - z velocity coordinates of each particle .
the particle data is then analyzed to determine if a minor merger has taken place .
the specific conditions under which galaxy mergers occur are still somewhat uncertain .
analytical work carried out under the impulsive approximation ( e.g. , alladin 1965 ; alladin 1975 ; white 1982 and references therein ) have indicated that tidal interactions will dissipate sufficient orbital energy to effect capture if the original encounter velocity between the two galaxies is slightly hyperbolic or less . more specifically , the energy dissipation from tidal interaction scales as : @xmath40 where @xmath41 and @xmath42 are the velocity and distance separation of the two galaxies at closest approach ( spitzer 1958 ; alladin 1975 ; white 1982 ) . in more general terms ,
white ( 1982 ) concluded that if a pair of equal mass galaxies have a relative velocity and impact parameter at infinity given by @xmath43 and @xmath44 , and with relative velocity @xmath41 and distance @xmath42 at closest approach , then merging will occur for @xmath45 and @xmath46 , where @xmath47 and @xmath26 are the rms velocity dispersion and gravitational radius ( as defined through the virial theorem ) of each galaxy .
the encounter parameters at infinity and at closest approach are related by @xmath48 and @xmath49 , where @xmath50 is the mass of each galaxy . for the specific case of a pair of galaxies with @xmath47 = 180 and m = 10@xmath51 ,
the maximum relative velocity at infinity that can result in a merger is then @xmath43 = 255 , and for a closest approach of @xmath42 = 25 kpc , the orbital kinetic energy term @xmath52 is approximately twice the gravitational potential energy term @xmath53 , since @xmath54 .
hence , it appears that as a general rule a merger will occur if the oribital kinetic energy at closest approach is less than twice the orbital potential energy .
this is clearly a very crude approximation , given that the actual tidal encounter is dependent on the mass distribution within the galaxies and their internal kinematics ( e.g. , elliptical versus disk galaxy ) and the angle of encounter .
unfortunately , the above assumption of an impulsive encounter is actually invalid in the slow encounters that lead to a merger .
however , numerical simulations ( e.g. , toomre 1977 ; white 1982 ) have tended to confirm the general results obtained from the above analytic approximations . as a specific case , in the hernquist & mihos ( 1995 ) simulations
a minor merger was achieved when the satellite galaxy was initially placed at 6 disk scale lengths from the large galaxy , or @xmath120 25 kpc for a major disk galaxy like the milky way ( e.g. , mihalas & binney 1981 ) .
we therefore adopt as the first criterion for a minor merger that the two galaxies must come within 25 kpc ( or 0.05 nemo distance units ) of each other . for our second criterion ,
i.e. , that the galaxies must have a relatively slow encounter speed , we specify that the encounter velocity at closest approach must be low enough that the orbital kinetic energy of the main galaxy plus satellite be less than twice the gravitational potential energy , i.e. , the constraint developed above from the impulsive approximation studies .
finally , since we are interested in the frequency of minor mergers , the third criterion is that the encounter must take place between an infalling subcluster member and a low mass particle in the main cluster .
the general results of the n - body simulations are that typically less than 2% of the subcluster galaxies experience a minor merger . given that we overestimated the number of small galaxies in coma by more than a factor of 4
, we thus predict that only 0.5% of the galaxies in a merging subcluster would experience a minor merger .
in contrast , @xmath140% of the early - type galaxies in the sw region of coma are either sb or psb ( caldwell et al 1993 )
. furthermore , those mergers that do take place in our simulations are almost exclusively among particles in the subcluster that are stripped off during the subcluster passage and thus remain behind with the main cluster .
as was discussed in caldwell & rose ( 1997 ) , about 60% of the subcluster particles become bound to the main cluster during the subcluster passage .
it is among these particles that the minor mergers occur in our simulations .
moreover , roughly 85% of the observed mergers take place between times 6 and 9 , i.e. , at late times .
what is clearly happening is that some of the particles left behind in the main cluster eventually experience a minor merger , which has little to do with the initial pasage of the subcluster through the main cluster .
this is not what is seen in the case of the sw region of coma , where the sb / psb galaxies are associated both spatially and kinematically ( in the mean ) with the subcluster .
in addition , the age dating of the psb galaxies indicates that they underwent starbursts within a fairly narrow time frame , thus implying that the starbursts were triggered in a well - defined event , such as passage of the subcluster through the main cluster . to further test these results , we changed the merger distance criterion from 25 kpc to 50 kpc .
while the number of minor mergers did increase with this change , the number only doubled to about 4% , with the mergers occuring in the same location ( i.e. , in the main cluster center ) and at the same times .
it may seem puzzling that the number of mergers did not quadruple when the distance was doubled , as might be expected since the cross sectional area goes as the square of the distance .
however , one must consider the energy criterion as well .
for a galaxy to satisfy both the distance and the energy requirements after the distance criterion is doubled , it must initially have a smaller velocity so that it still meets the energy condition at closest approach . for completeness
, we also tested the effects of changing the energy criterion so that the kinetic energy of the particles at closest approach must be less than 2.5 times the gravitational potential energy , as opposed to the factor of 2 previously considered . even under these conditions
, the number of minor mergers that occured was only near 3% .
likewise , there was no significant increase in the number of minor mergers when the infall parameters were changed .
thus , from our n - body simulations , it seems very unlikely that minor mergers are the main source of the star formation in the early type galaxies in nearby clusters .
a summary of the results of the simulations is given in table a.1 .
alladin , s. m. 1965 , , 141 , 768 alladin , s. m. , potdar , a. , & sastry , k. s. 1975 , in dynamics of stellar systems , edited by a. hayli , ( reidel , dordrecht ) , 167 andreon , s. , davoust , e. , michard , r. , nieto , j .- l . , & poulain , p. 1996
, , 116 , 429 baldwin , j. a. , phillips , m. m. , & terlevich , r. 1981 , , 93 , 5 balogh , m. l. , morris , s. l. , yee , h. k. c. , carlberg , r. g. , & ellingson , r. 1997 , , 488 , l75 barger , a. j. , aragn - salamanca , a. , ellis , r. s. , couch , w. j. , smail , i. , & sharples , r. m. 1996 , , 279 , 1 barnes , j. e. , & hut , p. 1989
, , 70 , 389 belloni , p. , bruzual , g. , thimm , g. -j . , & roser , h .- j .
1995 , , 297 , 61 bernstein , g. m. , nichol , r. c. , tyson , j. a. , ulmer , m. p. , & wittman , d. 1995 , , 110 , 1507 biviano , a. , durret , f. , gerbal , d. , le fevre , o. , lobo , c. , mazure , a. , & slezak , e. 1996 , a&a , 311 , 95 bower , r. g. 1991 , , 248 , 332 bower , r. g. , lucey , j. r. , & ellis , r. s. 1992 , , 254 , 601 briel , , u. g. , henry , j. p. , & bohringer , h. 1992 , , 259 , l31 bruzual a. g. , & charlot , s. 1995 , , in preparation burns , j. o. , roettiger , k. , ledlow , m. , & klypin , a. 1994 , , 427 , l87 butcher , h. & oemler , a. 1978 , , 219 , 18 butcher , h. & oemler , a. 1984 , , 285 , 426 byrd , g. , & valtonen , m. 1990 , , 350 , 89 caldwell , n. , & phillips , m. m. 1989 , , 338 , 789 caldwell , n. , & rose , j. a. 1997 , , 113 , 492 caldwell , n. , & rose , j. a. 1998 , , 115 , 1423 caldwell , n. , rose , j. a. , franx , m. , & leonardi , a. 1996 , , 111 , 78 caldwell , n. , rose , j. a. , sharples , r. m. , ellis , r. s. , & bower , r. g. 1993 , , 106 , 473 carlson , m. n. , et al .
1998 , , 115 , 1778 colless , m. & dunn , a. 1996 , , 458 , 435 couch , w. j. , barger , a. j. , smail , i. , ellis , r. s. , & sharples , r. m. 1998 , , 497 , 188 couch , w. , ellis , r.s . , sharples , r.m . , & smail , i. 1994 , , 430 , 121 couch , w. j. & newell , e. b. 1984 , , 56 , 143 couch , w. & sharples , r.m . 1987 , , 229 , 423 cowie , l. l. , & songaila , a. 1977 , nature , 266 , 501 dressler , a. 1980 , , 42 , 565 dressler , a. 1985 , , 22 , 185 dressler , a. & gunn , j. e. 1982 , , 263 , 533 dressler , a. & gunn , j. e. 1983 , , 270 , 7 dressler , a. & gunn , j. e. 1992 , , 78 , 1 dressler , a. , oemler , a. , butcher , h. r. , & gunn , j. e. 1994a , , 430 , 107 dressler , a. , oemler , a. , sparks , w. b. , & lucas , r. a. 1994b , , 435 , l23 fabricant , d. , cheilets , p. , caldwell , n. , & geary , j. 1998 , , 110 , 79 franx , m. 1993 , , 407 , l5 franx , m. , kelson , d. , van dokkum , p. illingworth , g. , & fabricant , d. 1997 , in asp conference series `` the second stromlo symposium : the nature of elliptical galaxies '' , eds . m. arnaboldi , g. s. da costa , & p. saha , 116 , p. 512
fujita , y. 1998 , astro - ph/9807120 fusco - femiano , r. , & hughes , j. p. 1994
, , 429 , 545 gavazzi , g. 1989 , , 346 , 59 gavazzi , g. , randone , i. , & branchini , e. 1995 , , 438 , 590 giovanelli , r. , & haynes , m. p. 1985
, , 292 , 404 gunn , j. e. , & gott , j. r. 1972 , , 176 , 1 haynes , m. p. , giovanelli , r. , & chincarini , g. l. 1984 , , 22 , 445 henriksen , m. j. , & byrd , g. 1996 , , 459 , 82 henry , j. p. , & lavery , r. j. 1987 , , 323 , 473 hernquist , l. & mihos , j. c. 1995 , , 448 , 41 holtzman , j. a. , faber , s. m. , shaya , e. j. , lauer , t. r. , grothe , j. , hunter , d. a. , baum , w. a. , ewald , s. p. , hester , j. f. , light , r. m. , lynds , c. r. , oneil , e. j. , jr .
, & westphal , j. a. 1992 , , 103 , 691 kauffmann , g. 1995 , , 274 , 153 kent , s. m. , & gunn , j. e. 1982 , , 87 , 945 koopmann , r. a. , & kenney , j. d. p. 1998
, , 497 , l75 kormendy , j. 1982 , in morphology and dynamics of galaxies , edited by l. martinet & m. mayor , ( geneva observatory ) , 113 lauer , t. r. , ajhar , e. a. , byun , y .-
i . , dressler , a. , faber , s. m. , grillmair , c. , kormendy , j. , richstone , d. , & tremaine , s. 1995 , , 110 , 2622 lavery , r. j. , & henry , j. p. 1986
, , 304 , l5 lavery , r. j. , & henry , j. p. 1988 , 330 ,
596 lavery , r. j. , pierce , m. j. , & mcclure , r. d. 1992 , , 104 , 2067 leonardi , a. j. , & rose , j. a. 1996 , , 111 , 182 lobo , c. , biviano , a. , durret , f. , gerbal , d. , le fevre , o. , mazure , a. , & slezak , e. 1997 , , 317 , 385 lucy , l. b. 1974 , , 79 , 745 maclaren , i. , ellis , r. s. , & couch , w. j. 1988 , , 230 , 249 mihalas , d. , & binney , j. 1981 , in galactic astronomy : structure and kinematics , second edition , ( w. h. freeman and co. , san francisco ) , 326 mihos , j. c. , & hernquist , l. 1994 , , 425 , l13 mihos , j. c. , walker , i. r. , hernquist , l. , de oliveira , c. m. , & bolte , m. 1995 , , 447 , l87 miller , b. w. , whitmore , b. c. , schweizer , f. , & fall , s. m. 1997 , , 114 , 2381 moore , b. , katz , n. , lake , g. , dressler , a. , & oemler , a. 1996 , nature , 379 , 613 moore , b. , lake , g. , & katz , n. 1998 , , 495 , 139 morris , s. l. , hutchings , j. b. , carlberg , r. g. , yee , h. k. c. , ellingson , e. , balogh , m. l. , abraham , r. g. , & smecker - hane , t. a. , in press , astro - ph/9805216 moss , c. , & whittle , m. 1993 , , 407 , l17 moss , c. , whittle , m. , pesce , j. e. , & socas - navarro , h. 1995 , astro lett . and communications , 31 , 215 pickles , a. j. & van der kruit , p. c. 1991 , , 91 , 1 rakos , k. d. & schombert , j. m. 1995 , , 439 , 47 richardson , w. h. 1972 , josa , 62 , 52 secker , j. , harris , w. e. , & plummer , j. d. 1997 , , 109 , 1377 soucail , g. , mellier , y. , fort , b. , & cailloux , m. 1988 , , 73 , 471 spitzer , l. 1958 , , 127 , 17 teuben , p. j. 1995 , in astronomical data analysis software and systems iv , edited by r. shaw , h. e. payne , & j. j. e. hayes , pasp conf .
77 ( asp , san francisco ) , 398 thompson , l. 1988 , , 324 , 112 toomre , a. 1977 , in evolution of galaxies and stellar populations , edited by r. b. larson & b. m. tinsley , ( yale university observatory , new haven ) , 401 valluri , m. 1993 , , 408 , 57 veilleux , s. & osterbrock , d. e. 1987 , , 63 , 295 watt , m. p. , ponman , t. j. , bertram , d. et al .
1992 , , 2 58 , 738 white , s. d. m. 1982 , in morphology and dynamics of galaxies , edited by l. martinet & m. mayor , ( geneva observatory ) , 291 white , s. d. m. , briel , u. g. , & henry , j. p. 1993
, , 2 61 , l8 whitmore , b. c. , & bell , m. 1988 , , 324 , 741 whitmore , b. c. , schweizer , f. , leitherer , c. , borne , k , & robert , c. 1993 , , 106 , 1354 whitmore , b. c. , miller , b. w. , schweizer , f. , & fall , s. m. 1997 , , 114 , 1797 wirth , g. d. , koo , d. c. , & kron , r. g. 1994 , , 435 , l105 | _ hst _ wfpc2 images in b ( f450w ) and i ( f814w ) have been obtained for three starburst ( sb ) and two post - starburst ( psb ) galaxies in the coma cluster , and for three such galaxies in the cluster dc2048 - 52 .
v ( f555w ) and i images for an additional psb galaxy in coma have been extracted from the _ hst _ archive .
seven of these galaxies were previously classified as e / s0 on the basis of ground - based images , one as sa , and the other as an irregular .
the _ hst _ images reveal these sb / psb galaxies to be heterogeneous in morphology .
nevertheless a common theme is that many of them , especially the sb galaxies , tend to have centralized spiral structure that appears simply as a bright `` bulge''on ground - based images .
in addition , while some psb galaxies exhibit distinct spiral structure , on the whole they have smoother morphologies than the sb galaxies .
the morphologies and luminosity profiles are generally consistent with substantial starbursts in the form of centralized spiral structure ( the sb galaxies ) which fade into smoother morphologies ( the psb galaxies ) , with lingering spectroscopic evidence for past central starbursts .
an important point is that the psb galaxies retain disks , i.e , they have not evolved into spheroidal systems .
while the morphologies revealed in the _ hst _ images are heterogeneous , and thus may not fit well into a single picture , we see evidence in several cases that the morphologies and centralized star formation have been driven by external tidal perturbations . we discuss several physical mechanisms for inducing star formation in cluster galaxies with a view towards explaining the particular morphologies seen in the _ hst _ images . |
@xmath2 searches for physics beyond the standard model ( bsm ) of particle physics are ongoing at high energy and high intensity , high precision experiments , and much theoretical effort is devoted to computing the rates expected in the standard model ( sm ) or to bsm scenarios accommodating whatever anomalies exist at the time and also dark matter and neutrino masses / mixing .
many anomalies " arose over the last decades , but no single clear conflict with the sm presently exists ( and this includes the discovery of the higgs particle at the lhc ) . some anomalies such as the high energy @xmath3 asymmetry at fermi - lab were not supported by the higher energy lhc and by further calculations of the sm expected values . in this context
the excess in semitaunic @xmath4 meson decays seems to be rather unique .
it has been seen in @xmath5 by both high intensity babar@xcite and belle@xcite electron positron colliders , and the @xmath6 excess was recently confirmed by lhcb@xcite in a @xmath7 times higher energy proton collider .
also the heavy quark effective field theory ( hqeft ) developed over the last decades@xcite reliably predicts the differential rates for these and other @xmath4 decays in the ( qcd corrected ) sm .
in particular the @xmath8 form factors can be extracted from the measured @xmath9 with @xmath10 or @xmath11 to predict the tauonic decay rates with the hqeft fixing the form factors in terms of few model parameters@xcite . the above and
the prospect of more lhcb results for decays of non-@xmath12 meson @xmath13 flavored hadrons prompt us to further consider this anomaly .
recently we became aware of the paper @xcite that addresses the issues with greater precision depth and detail and also provides an extensive bibliography and we will keep referring to this paper as we go along . in view of the importance of the general subject , the simpler version of the inclusive decay rates that we present and the different bsm model involving a charged uncolored new exchange ( rather than a leptoquark exchange ) that we emphasize , our present paper may still be useful .
in the next section we present a crude yet transparent estimate suggesting that the sm is unlikely to reproduce the large measured ratios @xmath14 and @xmath15 of the semitaunic and semileptonic modes . in sec .
iii we comment on explanations of the anomaly using new particles exchanged in the various channels of the semitaunic process . in sec .
iv we focus on a charged uncolored @xmath16 particle mediating only @xmath17 generation mixing .
@xmath2 we do not attempt to reproduce the accurate sm calculations of the ratios of semitauonic and semileptonic @xmath4 decays @xmath18 @xmath19 .
we believe that these calculations , building on the hqeft , are correct , though further verification by using presently much advanced lattice calculations would clearly be of great importance .
it is easy to see why at the nonrecoil point the @xmath20 semileptonic transition becomes universal if the @xmath13 and @xmath21 quarks were infinitely heavy . for @xmath22 = 0 $ ]
, the @xmath23 transition changes nothing for the spectator quark " or other light degrees of freedom of qcd which still see the same static color charge at the origin , possibly with a flipped spin .
this does not change energies in this limit and the relevant form factor becomes 1 .
however , to extend beyond this special point using a systematic expansion in @xmath24 , one needs the complete machinery of hqeft . while briefly described in particle data group ( pdg ) minireviews , most high energy physicists may find the actual calculations of @xmath25 shrouded in technical details and , like other monumental calculations of lattice qcd , simply take them on faith .
because of their extreme importance , the following crude yet easy to follow estimate of the @xmath26 ratios may be of some use . rather than addressing separately the @xmath27 and @xmath28 cases
, we will consider the ratio of the sum of the two modes in the semileptonic and semitauonic cases : @xmath29 + [ \gamma(b\rightarrow d^ { * } + l + \bar{\nu}_{l})]}.\ ] ] consistent values of the ckm matrix element@xmath30 were obtained by analyzing the exclusive @xmath27 and @xmath28 decay channels and independently via the inclusive semileptonic decay .
the latter calculation used quark hadron duality formally referred to as the short distance expansion .
it is based on the separation of time scales between the fast quark level @xmath31 transition of duration @xmath32 and the longer stage lasting @xmath33 during which the charmed quark combines with the spectator @xmath34 to form the final hadronic state .
the hadronization does not affect the total inclusive decay rate which is modified , however from the zeroth order , partonic calculation by qcd corrections .
these were computed to various orders and included in the above - mentioned work of freytsis - ligeti and ruderman @xcite . while these authors also used the inclusive calculation as further and less model dependent evidence , for the tension between sm calculation and the experimental measured @xmath35 and @xmath36 , our approach is simpler and different .
the idea is that the inclusive @xmath37 ( where @xmath38 is a charmed hadron ) or rather the ratio @xmath39 can provide an upper bound on @xmath40 .
the bound is obtained by assuming that , in the @xmath41 case , the complete inclusive decay is channeled into the @xmath27 and @xmath28 final states only .
the main factor fixing @xmath42(inclusive ) , where the zero index denotes the zeroth order partonic calculation of @xmath43 , is phase space which is significantly smaller for the semitauonic decay .
@xmath44 or @xmath45 $ ] is the only kinematic variable relevant for the hadronization .
lower @xmath44 i.e. smaller @xmath21 recoil momenta favors the formation of the ground states @xmath27 and @xmath28 . as evident from fig .
[ 1b ] , the distribution of @xmath46 in the decays to final states with the heavier @xmath41 , peaks at lower values than in the @xmath47 case .
this suggests a larger fraction of @xmath27 and @xmath28 in the inclusive tauonic decays ( we have used the values of @xmath48 gev and @xmath49 gev taken from ref.@xcite ) .
quark as a function of @xmath50 .
the color orange represents the leptonic @xmath47 cases and the blue one represents the tauonic decays.,title="fig:",width=415 ] + the ratio between inclusive semitauonic and semileptonic @xmath51 decays for the above @xmath52 and @xmath53 was very small , @xmath54 .
motivated by the comments at the end of the previous paragraph , we make the extreme assumption that the semitauonic inclusive decay goes _ only _ into the @xmath27 and @xmath28 final states .
this _ is not _ the case for the other semileptonic @xmath4 decays where from the pdg we find that only @xmath55 70% of the decays are into @xmath27 or @xmath28 . with the remaining 30% corresponding to higher
@xmath56 , nonresonant @xmath57 or even baryonic @xmath58 states .
since 100% of all the inclusive semitauonic decays go into @xmath27 or @xmath28 , this enhances @xmath40 to @xmath59 which is still much less than @xmath60 . to test for the sensitivity of the results to @xmath52 and @xmath53 , we varied @xmath52 between @xmath61 and @xmath62 gev and @xmath53 between @xmath63 and @xmath64 gev .
we obtained @xmath65 values ranging from @xmath66 to @xmath67 and corresponding maximal @xmath68 between @xmath69 and @xmath70 . except for the last case ,
all the @xmath68 are in disagreement with the experimental values .
we note a certain part of the @xmath71 corrections is particularly pertinent to the present approach , further decreases @xmath43 .
in addition to the @xmath72 process , we have the distinct four body perturbative final state , where an extra gluon has been radiated , @xmath73 .
we estimate that this enhancement is @xmath74 of the total inclusive decay . due to the much more dramatic decrease of the four body @xmath75 phase space with increasing lepton masses
, we expect that the latter part is almost completely absent in the inclusive @xmath76 decay to charm and the massive @xmath41 lepton . thereby decreasing @xmath43 and also the maximal @xmath68 by up to an extra 10% .
@xmath2 a new anomaly which appear only when all the quarks and leptons involved are members of the third generation is attractive in the framework of extradimensional theories allowing partial compositeness but only for third generation fermions@xcite .
the present anomaly involves beside the @xmath13 , @xmath41 , and @xmath77 also the @xmath21 quark which up to the @xmath78 mixing is a member of the second generation .
still it comes closer to the ideal third generation fermions only than all previously studied cases , in particular the @xmath79 decay involving a first generation @xmath80 quark and in which no anomaly was reported .
existing and forthcoming lower bounds on masses of new particles and/or on deviations from the sm in other processes , constrain new physics interpretations of the @xmath35 , @xmath36 anomaly .
we next proceed with a general phenomenological discussion ( which is less detailed and somewhat different from that of ref .
the anomaly can be due to a new particle @xmath81 which interacts with the @xmath13 and @xmath21 quarks , @xmath41 and @xmath82 leptons . _
a priori _ this can happen in three distinct ways : * the @xmath81 particle could be color and charge neutral and exchanged between the @xmath13 and @xmath21 quarks and between the @xmath83 and @xmath41 ( case a ) .
* the @xmath81 particle could be a leptoquark exchanging between the @xmath84 and a @xmath85 vertex or between a @xmath86 and a @xmath87 vertex ( case b ) . *
the @xmath16 particle could be exchanged just like the ordinary sm @xmath88 in the @xmath89 or @xmath90 , i.e. the @xmath91 channel ( case c ) . in case
a the @xmath81 particle exchange is a new radiative correction of the ordinary weak current hadronic vertex and of the weak leptonic vertex . to explain the big @xmath35 and @xmath36 , these radiative corrections should be large , exceeding the strong qcd corrections of the hadronic vertex due to a gluon exchange .
this new exchange should not occur between the first family generation .
this scenario is strongly constrained by the well - established lepton universality in @xmath92 decays .
a modified @xmath93 vertex will also change the rate of tau decays .
thus , attributing the observed anomaly a diagonal @xmath81 coupling , which otherwise minimally affects the flavor structure of the theory , is practically excluded .
case b , with the leptoquark @xmath94 particle , was discussed in great detail in ref .
@xcite which found it to be the most likely explanation . as correctly pointed by the authors
, this would require relatively light -o(tev ) @xmath95 and the discovery of such particles in future lhc experiments would be sensational .
gev leptoquarks first arose in the framework of grand unified theories .
it will be very remarkable if the eventual discovery of a far lighter leptoquark will be motivated by the much more humble @xmath13 flavor physics . ] here we consider case c , namely the possibility of a @xmath91 channel exchange of a new charged , uncolored @xmath81 particle .
this was addressed before in particular by e.g. refs @xcite and @xcite .
it is known @xcite that a charged scalar particle in the two higgs doublet ( 2hdm ) type ii extensions of the sm fails to reproduce _ both _ @xmath35 and @xmath96 for any combination of masses and couplings . on the other hand , in ref.@xcite
it was shown that it is possible to fit @xmath35 and @xmath96 with to the 2hdm type iii with some set of parameters .
reference @xcite suggested that the aligned 2hdm can fit both @xmath35 and @xmath96 , but the ranges of parameters used to do this fit are in conflict with the constraints from leptonic charm decays , so they analyze the predictions of the model for observables ( @xmath41 spin asymmetry , forward - backward asymmetries , and differential decay rates ) sensitive to the charged - scalar contributions .
here we focus on a vector @xmath81 particle with @xmath97 couplings to quarks and leptons .
as noted by ref.@xcite this simplifies the analysis since the new bsm amplitude has the same form as the ordinary @xmath98 exchange and generates the same four - fermi effective lagrangian with a new fermi constant , @xmath99 the large electron positron collier ( lep ) bound ( @xmath100 gev from pair productions ) is likely to be improved by the lhc .
it is important to note that @xmath101 pair production rates are smaller than those of pairs of leptoquarks ( @xmath102 ) by a factor of @xmath103 .
therefore , we expect that the new bounds on @xmath104 which at present are of order 200 gev will be significantly weaker than those for leptoquarks .
this may allow such particles to play the large role required to explain @xmath68 and is an important motivation to consider the following toy model .
@xmath2 we next proceed to discus a toy model involving second and third generations only , which is based on the exchange of a charged , color neutral , vector @xmath105 particle in the @xmath91 ( @xmath106 ) channel .
many bsm extensions modifying the flavor structure lead to excessive fcnc transitions and in particular to large @xmath107 and @xmath108 mixing . to avoid an @xmath81 particle tree - level exchange from generating this most acute mixing problem
, we endow our @xmath109 particles with conserved generation numbers .
specifically we denote the required @xmath81 boson as @xmath110 .
this refers to the second and third generations each including a quark and lepton @xmath111 doublets .
thus , we denote @xmath112 , @xmath113 , @xmath114 and @xmath115 and also @xmath116 , @xmath117 , @xmath118 , and @xmath119
. our proposed @xmath81 particle can generate the quark @xmath120 transition and its conjugate can generate the leptonic @xmath121 transitions .
thus , in our model the new bsm contribution involves a different final state @xmath122 rather than @xmath123 which is emitted in standard model @xmath98 exchange .
clearly the experiments where the @xmath124 anomalies were discovered can not distinguish neutrinos with different flavors .
however , an important consequence is that in the present approach the sm and bsm amplitudes do not interfere and the ratio of their contributions is @xmath125^{4},\ ] ] so that @xmath126^{4}$ ] . in order to explain the observed anomaly
, the last ratio should be about @xmath127 , and finally we obtain the condition @xmath128 the fact that the present lower bounds on @xmath104 are very low is very helpful in satisfying this constraint , e.g by choosing @xmath129 gev and @xmath130 .
the weak @xmath131 symmetry that should be respected implies that our charged @xmath110 is accompanied by a neutral @xmath132 .
this neutral @xmath133 particle is the middle member of an @xmath134 triplet .
its third member is @xmath110 , @xmath135 the linear combination @xmath136 is orthogonal to the middle member of the triplet and is a singlet of weak isospin .
a key point is that the neutral @xmath133 do not mediate @xmath137 mixing as can be readily verified by following the 2,3 generation indices .
however , they do generate fcnc transitions .
in particular the neutral member of the triplet @xmath138 will generate both @xmath139 and the dramatic flavor changing @xmath140 transitions which will be @xmath141 . using the pdg upper bound @xmath142 on the branching of @xmath143 and of @xmath144
, we estimate the inclusive @xmath145 to be less than @xmath146 .
this implies the constraint @xmath147 notice that once @xmath148 , the last condition is satisfied .
further , if the bound on @xmath139 is saturated we expect a similar slightly smaller rate due to the phase space reduction for the inclusive @xmath140 which should be experimentally looked for .
thus we predict an inclusive decay rate @xmath149 which is proportional to @xmath150 .
weak isospin symmetry implies the same @xmath151 as in the previous equations . up to now
we have considered the minimal set of extra particles that @xmath152 requires once we have the @xmath153 boson mediating the bsm contribution explaining the observed excess of the @xmath154 . while we imposed the necessary equality of the couplings of these we had to assume a mass ratio of @xmath155 the neutral to charged member of the resulting @xmath156 triplet .
one may want to extend the set of new particles by imposing a @xmath157 family symmetry and use the @xmath158 generator operating on our @xmath153 boson to predict two new particles of the form @xmath159 @xmath160 . these generation diagonal bosons of mass @xmath161 do not generate new fcnc transitions . however , they add a coherent contribution to the charged @xmath162 of size @xmath163 with @xmath164 the sm , @xmath88 exchange contribution . by assumption
there is no analog contribution to the leptonic @xmath41 decay to electron and neutrinos . using the ratio @xmath165 required to fit the @xmath26 anomaly
, we find that a ratio of @xmath166 of the masses is enough to avoid violation of the relatively well experimentally verified lepton universality of the @xmath41 decay in the first and second generation of families . for completeness
we indicate in fig [ fd1 ] the set of 16 bosons of the form @xmath167 that are generated by repeating operations like the above and which form a @xmath168 @xmath169 representation . as indicated no new types of transitions
are then generated except for the completely ( generation and weak isospin ) diagonal transition due to the 4 mesons , which we refer to as @xmath170 , in the center of our figure .
since the latter are of the same type as those considered in case a , we need to impose on their masses an even stronger constraint than above namely @xmath171 tev .
so far we have only addressed possible difficulties associated with the newly introduced @xmath81 particle sector on its own . to avoid difficulties with other precision measurements we had to impose a 1:3:6:10 hierarch between the masses of the required @xmath105 , its weak isospin @xmath172 partner , the generation neutral @xmath173 and the completely generation and @xmath111 diagonal @xmath174 .
it seems difficult to explain the large @xmath156 breaking manifest via the @xmath175 mass difference mass difference does not violate @xmath156 because it originates from higgs coupling between left and right quarks where the latter are @xmath156 singlets . to have a remotely related scenario here
we should involve also right - handed @xmath81 particles as well . ] .
we also need to address the all important issue of the breaking of flavor symmetries by the higgs couplings .
these were reflected in the quark and lepton spurions in the paper of ref.@xcite .
while it is conceivable that our model when fully implemented with its own new higgs sector will be able to generate the magic of gim - like mechanism avoiding _ all _ the possible dangerous resulting fcnc , this seems rather unlikely . discussing this alternative , less dramatic than the leptoquark explanation , is still very useful @xcite
. combinations .
the particle at the upper left corner of mass @xmath176 is the one we need to mediate the extra incoherent contribution @xmath177 .
the @xmath172 particle of mass @xmath178 at the middle of the left edge is obtained from the initial @xmath105 by a weak isospin lowering operation .
to each particle corresponds a conjugate particle mediating the reverse reaction obtained by a @xmath179 and @xmath180 reflection through the origin of the square .
as indicated by the arrows corresponding to @xmath81 emission and absorption , the @xmath172 particle mediates the decays @xmath181 and @xmath182 .
the @xmath183 particle of mass @xmath184 in the top middle was obtained by operation with the generation shifting operator @xmath185 .
it adds coherently to the standard model weak decay @xmath162 but not to the corresponding electronic tau decay which puts a lower limit on @xmath186.,title="fig:",width=552 ] +
the @xmath35 and @xmath36 which @xcite and we discussed are very tantalizing .
if further lhcb and belle-2 experiments will verify the high ratios and exclude a very subtle systematic error plaguing all existing experiments , we will have to incorporate the anomaly into the theoretical bsm framework .
the fact that it seems almost impossible to do so makes it only all the more challenging .
the authors thanks andreas crivellin and martin jung for useful comments . e.c.f.s.f .
thanks university of maryland and nasa goddard space flight center for the hospitality while this work was being completed and fapesp for full support under contracts no .
14/05505 - 6 and 11/21945 - 8 .
s. fajfer , j. f. kamenik and i. nisandzic , aip conf .
proc . * 1492 * , 82 ( 2012 ) .
a. crivellin , c. greub and a. kokulu , phys .
d * 86 * , 054014 ( 2012 ) .
a. crivellin , j. heeck and p. stoffer , arxiv:1507.07567 .
a. celis , m. jung , x. q. li and a. pich , j. high energy phys .
01 ( 2013 ) | the anomalously large experimentally measured ratios of the semitauonic decay @xmath0 and the corresponding semileptonic @xmath1 disagree with the predictions of the standard e.w + qcd model(s.m ) .
we briefly comment on this disagreement and on possible new physics explanations which are rather constrained and difficult to implement . |
a _ nonclassical linear boltzmann equation _ has been recently proposed to address particle transport problems in which the particle flux experiences a nonexponential attenuation law @xcite .
this nonexponential behavior arises in certain inhomogeneous random media in which the locations of the scattering centers are spatially correlated , such as in a pebble bed reactor core @xcite .
independent of these developments , a similar kinetic equation has been rigorously derived for the periodic lorentz gas in a series of papers by golse ( cf .
@xcite ) , and by marklof and str " ombergsson @xcite .
related work has also been performed by grosjean @xcite , considering a generalization of neutron transport that includes arbitrary path - length distributions . for the case of monoenergetic particle transport with isotropic scattering ,
the nonclassical linear boltzmann equation is written as @xmath1 , \end{split}\ ] ] where @xmath2 is the nonclassical angular flux , @xmath3 is the scattering ratio ( probability of scattering ) , and @xmath4 is a source . here , the total cross section @xmath5 is a function of the path length @xmath6 ( distance traveled by the particle since its previous interaction ) , such that the path length distribution @xmath7 does not have to be exponential . if @xmath0 is exponential , eq .
( [ eq1.1 ] ) reduces to the classical linear boltzmann equation @xmath8 for the classical angular flux @xmath9 it has been shown @xcite that , by selecting @xmath10 in a proper way , eq . can be converted to an integral equation for the scalar flux @xmath11 that is identical to the integral equation that can be constructed for certain diffusion - based approximations to eq . in the hierarchy of the @xmath12 equations @xcite .
the work in this paper shows that this is also the case for the _ nonclassical diffusion equation _
@xcite @xmath13 which is an asymptotic approximation of eq .
( [ eq1.1 ] ) when @xmath14 . here , @xmath15 specifically , we find @xmath0 and the corresponding @xmath10 such that the integral equation for eq .
( [ eq1.6 ] ) is identical to the integral equation for eq .
( [ eq1.1 ] ) .
we also show that the second moment of @xmath0 for nonclassical diffusion preserves the _ true _
mean - squared free path @xmath16 , which gives a new insight on the results obtained in @xcite for the classical diffusion approximations .
the remainder of this paper is organized as follows . in section [ sec2 ]
we convert eq . to an integral equation for the scalar flux given in eq . .
in section [ sec3 ] we convert eq . to an integral equation for the scalar flux , and find the correct choice of @xmath0 such that the integral equation obtained in section [ sec2 ] is identical to this nonclassical diffusion integral equation .
we also present a numerical example illustrating the differences on @xmath0 and @xmath10 for classical transport , classical diffusion , and nonclassical diffusion .
the paper concludes with a discussion in section [ sec4 ] .
let @xmath17 be given by @xmath18 where @xmath19 and @xmath20 we can now write eq . as the initial value problem [ eq2.2 ] @xmath21 following the steps presented in @xcite and @xcite , we : ( i ) use the method of characteristics to calculate the solution of eqs .
; ( ii ) operate on this solution by @xmath22 ; and ( iii ) perform the change of spatial variables from the 3-d spherical @xmath23 to the 3-d cartesian @xmath24 defined by @xmath25 .
this yields @xmath26 where @xmath27 and @xmath17 are given by eqs . and , respectively .
the nonclassical diffusion formulation presented in eq . is an asymptotic approximation of eq . in the case of @xmath28 . in this formulation ,
@xmath29 and @xmath16 as given by eq .
represent the first and second moments of the _ true _ path length distribution @xmath0 .
if @xmath0 is exponential , then @xmath30 , @xmath31 , and eq .
reduces to the classical diffusion equation @xmath32 for the general case in which @xmath0 is _ not _ assumed to be an exponential , we define @xmath33 and rewrite eq . as : [ eq3.2 ] @xmath34 where @xmath35 the green s function for the operator on the left hand side of eq .
is : @xmath36 therefore , we can transform eq . into an integral equation for @xmath37 by taking @xmath38 bearing in mind that @xmath39 represents the mean free path of a particle ( i.e. the average distance between collisions ) , the collision - rate density can be written as @xmath40 , such that @xmath41 this result agrees with eq .
iff @xmath42 it is easy to verify that @xmath43 which shows that eq . is a distribution function .
furthermore , @xmath10 is given by @xmath44 this shows that the nonclassical transport equation reproduces the nonclassical diffusion approximation given by eq . if @xmath0 and @xmath10 are defined by eqs . and .
moreover , if @xmath0 is exponential , this results agrees with the @xmath0 and @xmath10 obtained for the classical diffusion equation in @xcite .
we also point out that [ eq3.9 ] @xmath45 the first moment of @xmath0 only approximates the mean free path @xmath29 , as can be seen in eq . .
however , eq . shows that the _ true _ mean - squared free path @xmath16 is preserved .
this sheds new light on the results presented in @xcite , where the second moments of the path length distributions obtained for all ( classical ) diffusion approximations give the exact ( classical ) transport value @xmath46 .
figures [ fig1 ] and [ fig2 ] show the functions @xmath10 and @xmath0 for the transport of neutrons taking place in the interior of a homogenized 3-d pebble bed reactor core system , as described in @xcite .
the system consists of fuel spheres ( pebbles ) randomly packed in a background void with packing fraction @xmath47 .
the parameters for the material of the fuel pebbles are : diameter @xmath48 ; total cross section @xmath49 ; and scattering ratio @xmath50 . as discussed in detail in @xcite , the _ classically homogenized _ system ( using the atomic mix model ) has total cross section @xmath51 , and its _ true _ mean - squared free path is numerically calculated to be @xmath52 .
we note that nonclassical transport occurs due to the spatial correlations of the fuel pebbles in the system , and therefore @xmath53 .
we have shown that the nonclassical diffusion equation for an infinite homogeneous medium can be represented exactly by the nonclassical linear boltzmann equation with the correct choice of @xmath10 and @xmath0 .
we derived explicit expressions for these quantities and showed that , while the first moment of the path length distribution @xmath0 only approximates the true mean free path @xmath39 , its second moment _ preserves _ the true mean - squared free path @xmath16 .
this result provides a deeper understanding of the results presented in @xcite for the second moment of the path - length distributions for classical diffusion approximations .
the work on this paper allows us to construct @xmath0 and @xmath10 that yield the correct solution for the nonclassical boltzmann equation in a diffusive system .
the only parameter necessary for this is the true mean - squared free path @xmath16 .
this paves the road to the possibility of using this easy - to - obtain @xmath0 to approximate the solutions of the nonclassical boltzmann equation as the system moves away from the diffusive limit .
further work needs to be done to investigate how well such approach would perform ; this task , however , must be left for future work .
this paper was prepared by richard vasques under award number nrc - hq-84 - 14-g-0052 from the nuclear regulatory commission .
the statements , findings , conclusions , and recommendations are those of the author and do not necessarily reflect the view of the us nuclear regulatory commission .
r. vasques . estimating anisotropic diffusion of neutrons near the boundary of a pebble bed random system .
_ proceedings of the international conference on mathematics and computational methods applied to nuclear science & engineering - m&c 2013 [ cd - rom]_. la grange park , il : american nuclear society ; 2013 .
17361747 . c. grosjean , the exact mathematical theory of multiple scattering of particles in an infinite medium .
vlaamsche akad .
schoone kunsten belgi " e _ 36 , paleis der academi " en , brussels , 1951 . | we show that , by correctly selecting the probability distribution function @xmath0 for a particle s distance - to - collision , the nonclassical diffusion equation can be represented exactly by the nonclassical linear boltzmann equation for an infinite homogeneous medium .
this choice of @xmath0 preserves the _ true _
mean - squared free path of the system , which sheds new light on the results obtained in previous work . |
although the standard model ( sm ) @xcite is considered as an effective low energy theory that should be embedded into a more fundamental theory , many of the sm predictions have been successfully tested by precision measurements .
the latter impose strong restrictions to new physics contributions associated to any extension of the sm @xcite .
thus , small deviations between the experimental data and the sm predictions allow to set stringent limits on new physics from a more fundamental theory that contains new types of matter and interactions at the tev scale . it will be explored with the new generation of accelerators and detectors like the forthcoming large hadron collider ( lhc ) @xcite . among the possible extensions of the sm , the models with gauge symmetry @xmath6
, also called 3 - 3 - 1 models @xcite , arise as an interesting alternative with new physics content and some motivating features .
first of all , from the cancellation of chiral anomalies @xcite and asymptotic freedom in qcd , the 3 - 3 - 1 models can explain why there are three fermion families .
secondly , since the third family is treated under a different representation , the large mass difference between the heaviest quark family and the two lighter ones may be understood third - family .
thirdly , the models have a scalar content similar to the two higgs doublet model ( 2hdm ) , which allow to predict the quantization of electric charge and the vectorial character of the electromagnetic interactions @xcite .
also , these models contain a natural peccei - quinn symmetry , necessary to solve the strong - cp problem @xcite .
finally , the model introduces new types of matter relevant to the next generations of colliders at the tev energy scales , which do not spoil the low energy limits at the electroweak scale . in the sm
, the flavor changing neutral currents ( fcnc ) are strongly suppresed with respect to the charged - current weak interactions , which follows from the experimental data on neutral meson decays and the mass difference in meson systems exhibiting particle - antiparticle mixing @xcite . in particular
, some extensions of the sm produce new fcnc contributions at tree level , as for example some models with an extra neutral @xmath1 boson , which represents an stringent limit for new physics .
although not all models with new neutral @xmath1 bosons exhibit additional fcnc contributions @xcite , many interesting ones contain fcnc effects at tree level @xcite . in 3 - 3 - 1 models , the contributions in meson systems
have been considered before @xcite in @xmath7 and @xmath8 systems , which induce the flavor changing transitions @xmath9 and @xmath10 , respectively , while no information other than a lower bound associated to the @xmath11 transition was available for the @xmath12 system .
however , the @xmath13 sector was recently confirmed in the @xmath14 mixing by both cdf and do/ @xcite : @xmath15 ref .
@xcite use the following averages @xmath16 for the experimental and sm prediction , respectively . since the study of @xmath17 physics has been an important tool to extract information on cp violation and new physics @xcite
, we will use the above data for the mass difference of the @xmath14 system to explore the fcnc contribution induced by the @xmath18 boson in the two most popular 3 - 3 - 1 models .
however , since fcnc contribution in these models are very sensitive to the rotations of the fermionic spectrum to mass eigenstates , it is neccesary to implement some criterion to fix the values of the components of the rotation matrices , and to get numerical predictions on the meson mass diference .
in contrast to other studies in @xmath19 , @xmath20 and @xmath21 systems @xcite , we will consider various cases for the rotation matrix , including the texture - zero approches , where an ansatz on the texture of the femion mass matrices is adopted in agreement with the measured masses and mixing angles of the cabibbo - kobayashi - maskawa matrix . an additional motivation to study the @xmath14 system comes from the fact that the @xmath13 sector induces the maximum flavor - changing contribution , as will be confirmed in this work .
this offers a good opportunity to extract information on new physics at low energy .
( [ massdiff ] ) shows good agreement between the experimental data and the sm one - loop prediction of @xmath22 , however , due to the hadronic parameters , the sm prediction contains a large uncertainty which we use to find allowed regions for the mass of the @xmath18 boson and the weak angle associated to the @xmath5 coupling by assuming four specific forms in the rotation matrix of the quark mass .
the fermionic structure is shown in tab . [ tab : espectro ] where all leptons transform as @xmath23 and @xmath24 under the @xmath25 sector , with @xmath26 and @xmath27 the @xmath28 generators associated with the left- and right - handed leptons , respectively ; while the quarks transform as @xmath29 , @xmath30 for the first two families , and @xmath31 , @xmath32 for the third family , each one with its @xmath28 values for the left- and right - handed quarks .
the quantum numbers @xmath33 for each representation are given in the third column from tab .
[ tab : espectro ] , where the electric charge is defined by @xmath34 with @xmath35diag@xmath36 , @xmath37diag@xmath38 and @xmath39 and @xmath40 , where the first case contains the foot - long - truan model ( flt ) @xcite and the second contains the pisano - pleitez - frampton model ( ppf ) @xcite .
@xmath41 for the scalar sector , we introduce the triplet field @xmath42 with vacuum expectation value ( vev ) @xmath43 , which provides the masses of the third fermionic components . in the second transition
, it is necessary to introduce two triplets@xmath44 and @xmath45 with vev @xmath46 and @xmath47 , in order to give masses to the quarks of up- and down - type , respectively @xcite . in the gauge boson spectrum associated with the group @xmath48 we are just interested in the physical neutral sector that corresponds to the photon , @xmath49 , and @xmath50 which are written in terms of the electroweak basis for @xmath51 and @xmath40 as @xcite @xmath52 where the weinberg angle is defined as @xcite @xmath53 and @xmath54 @xmath55 correspond to the coupling constants of the groups @xmath56 and @xmath28 , respectively .
using the fermionic content in weak eigenstates from tab . [ tab : espectro ] , we obtain the neutral coupling for the sm quarks @xcite
@xmath57 , \label{l - neutro}\ ] ] where @xmath58 for up- and down - type quarks , respectively .
the vector and axial - vector couplings of the @xmath49 boson are @xmath59 with @xmath60 the electric charge of each quark given by tab .
tab : espectro ; while the corresponding couplings to @xmath18 are given by @xmath61 , \notag \\ \widetilde{g}_{v , a}^{d^{0 } } & = & \frac{g_{x}c_{w}}{2g_{l}t_{w}}\left [ \frac{1}{% \sqrt{3}}\left ( diag\left ( 1,1,-1\right ) + \frac{\beta t_{w}^{2}}{\sqrt{3}}% \right ) \pm 2q_{d^{0}}\beta t_{w}^{2}\right ] , \label{zprima - couplings}\end{aligned}\ ] ] which are written for @xmath51 and @xmath40 .
in particular , for the @xmath62 coupling in the neutral lagrangian in eq .
( [ l - neutro ] ) , we can write @xmath63 , \label{l - neutro2}\ ] ] where @xmath64 and @xmath65 the chiral projectors . using the neutral @xmath1-couplings from eq .
( [ zprima - couplings ] ) , the new chiral couplings @xmath66 are written as follows @xmath67 , \notag \\ \widetilde{\epsilon } _ { r}^{u^{0},d^{0 } } & = & \frac{g_{x}c_{w}}{g_{l}t_{w}}% \left [ q_{u^{0},d^{0}}\beta t_{w}^{2}\right ] .
\label{acoples - quirales}\end{aligned}\ ] ] on the other hand , we will consider linear combinations among the three families of quarks to obtain couplings in mass eigenstates @xmath68 where @xmath69 denotes the quarks in mass eigensates , @xmath70 in weak eigensates and @xmath71 the rotation matrix that diagonalize the yukawa mass terms .
thus , we can write the eq .
( [ l - neutro2 ] ) as @xmath72 , \label{l - mass}\ ] ] where the chiral couplings in mass eigenstates are defined as @xmath73 because of the fact that @xmath74 in eq .
( acoples - quirales ) is family independent , the right - handed couplings remain flavor - diagonal in the mass eigenbasis , such that @xmath75 . however , due to the @xmath76 term from eq .
( [ acoples - quirales ] ) ( family dependent couplings ) , we obtain non - diagonal components in the left - handed couplings @xmath77 in eq .
( [ chiral - mass ] ) , which is sensitive to the form of the rotation matrix @xmath71 . in order to have a predictive model
, we adopt a different ansatz on the texture of the quark mass matrices in agreement with the six quark physical masses and the four physical parameters of the ckm matrix .
the @xmath78 lagrangian for the yukawa interaction between quarks is @xmath79 \notag \\ & & + \overline{q_{3l}}\left [ \gamma _ { \rho } ^{3d}\rho d_{r}^{0}+\gamma _ { \eta } ^{3u}\eta u_{r}^{0}+\gamma _ { \chi } ^{3j}\chi j_{3r}^{0}\right ] + h.c , \label{yukawa-1}\end{aligned}\ ] ] with @xmath80 and @xmath81 being the two scalar triplets neccesary to give masses to the sm fermion spectrum from table tab : espectro , and @xmath42 the scalar triplet that gives masses to the new extra fermions @xmath82 , as explained in sec .
2 . thus , we are not interested in the couplings of @xmath83 .
@xmath84 are the yukawa interaction matrices .
taking into account only the @xmath85 sector ( which lies in the two upper components of each scalar triplet ) , and omitting the couplings of @xmath86 , the mass eigenstates of the scalar sector can be written as beta - arbitrary @xmath87where @xmath88 denotes the conjugate representation of @xmath89 @xmath90 and @xmath91 thus , after some algebraic manipulation , the neutral couplings of the yukawa lagrangian can be written as @xmath92 \left ( 1+% \frac{h_{3}^{0}\mp i\phi _ { 3}^{0}}{\nu } \right ) \notag \\ & & + \left [ \overline{d_{l}^{0}}\left ( \gamma _ { d^{0}}\right ) d_{r}^{0}+\overline{% u_{l}^{0}}\left ( \gamma _ { u^{0}}\right ) u_{r}^{0}\right ] \left ( h_{4}^{0}\pm ih_{1}^{0}\right ) + h.c , \label{yukawa-2}\end{aligned}\ ] ] where the fermion masses and yukawa coupling matrices are given by @xmath93 where @xmath94 and @xmath95 .
the lagrangian from eq .
( [ yukawa-2 ] ) is equivalent to the two - higgs - doublet model ( 2hdm ) lagrangian @xcite , which exhibits fcnc due to the non - diagonal components of @xmath96 in the literature , there are various approches on the zero - textures of the quark mass matrices @xmath97 from eq .
( [ yukawa - matrix ] ) , where the most popular are listed as follows : in the basis @xmath98 the quark mass matrices in the fritzsch ansatz are defined as @xcite @xmath99 with @xmath100 @xmath101 and @xmath102 where @xmath103 corresponds to the physical mass of the quarks .
the above ansatz is diagonalized by the following rotation matrices for both the up- and down - type quarks @xmath104 : this texture takes the same form as eq .
( [ texture - fritzsch ] ) , but with @xmath105 in the ( 2,2 ) component of the mass matrix 4zeros .
this form have the following rotation matrices @xmath106 the above ansatz was reconsidered by the authors in ref .
fitmatsuda , where the parameter @xmath107 in the @xmath108 component is taken as a free parameter . in particular , they define the ratio @xmath109 , such that the experimental values of the ckm matrix are derived by fine tuning of the parameter @xmath110 .
thus , the non - zero components of the mass matrix takes the form @xmath111 in the @xmath112 components , @xmath113 in the @xmath114 component , and @xmath115 in the @xmath116 components , where the hierarchy @xmath117 is required .
the rotation matrix is @xmath118 the authors in ref .
fitmatsuda obtain the values @xmath119 and @xmath120 . :
another consistent possibility is to consider different texture assignment for the up- and down - type quarks , as follows @xcite @xmath121 with @xmath122 @xmath123 and @xmath124 for the up sector , while for the down sector the structure is @xmath125 @xmath126 and @xmath127 the above textures are diagonalized by @xcite @xmath128 where @xmath129
the left - handed coupling in eq .
( [ chiral - mass ] ) contains non - diagonal components , which induce mixing between the neutral @xmath1 boson and quarks from different families .
this will produce new physics contributions to the mass difference in neutral meson systems as for example in kaons @xmath7 bottom @xmath131 and bottom - strange @xmath130 mesons , each one induced by the @xmath132 @xmath133 and @xmath134 transition , respectively .
in particular , we take the most recent data of the @xmath135 difference mass given by eq .
( [ massdiff ] ) in order to constraint new physics induced by the @xmath1 interaction .
the ratio between the experimental value and the sm prediction in eq .
( [ massdiff ] ) is @xcite @xmath136 with @xmath137 the @xmath138 component of @xmath139 defined by eq .
( [ chiral - mass ] ) , and @xmath140 the weak phase .
the above data constrain the values of the @xmath1 mass and the weak phase assuming different ansatz in the texture of the mass matrices of the quarks , as discussed in sec .
3 . for the rotation matrix @xmath141 in the down sector
, we consider the fritzsch ansatz ( @xmath142 ) in eq .
( [ fritzsch - biunitary ] ) , the matsuda - nishihura ansatz ( @xmath143 ) in eq .
( [ matsuda2-biunitary ] ) , and the matsuda ansatz ( @xmath144 ) in eq .
( [ matsuda - biunitary ] ) . in order to achieve a complete comparison
, we also consider the flavor - changing contribution assuming that @xmath145 , with @xmath146 the @xmath147 component of the ckm matrix , where we use the values @xmath148 @xcite .
we use the notation @xmath149 for this last case . figs .
[ fig - bs ] show plots of the contours at @xmath150 c.l for both ( a ) @xmath151 and ( b ) @xmath152 models , and for each ansatz of the rotation matrix , where we use the following data at the @xmath49 scale @xmath153 c.l .
, title="fig : " ] c.l .
, title="fig : " ] the regions below each curve correspond to excluded points in the @xmath154 plane .
we can also see excluded regions in the center of each curve , as shown in the plots .
the minimum values for @xmath155 are found when @xmath156 in particular , the ansatz @xmath149 induces the lowest bounds with values @xmath157 tev , while the matsuda ansatz @xmath144 leads high values with bounds from @xmath158 tev for @xmath151 models , and @xmath159 tev for @xmath160 although the flavor - changing contribution is very sensitive to the rotation matrix , we find points that overlap regions from different ansatz in the curves of the plots .
for example , in fig .
[ fig - bs]-(a ) , we find the same value @xmath161 tev for @xmath162 and @xmath163 if @xmath164 and @xmath165 , respectively .
the differences exhibited by each curve in the above figures arise from the size of the mixing components of the couplings @xmath166 for each ansatz . in tab .
[ tab : left - handed - couplings ] , we compare the non - diagonal components of the left - handed coupling in the down sector .
we also compare the fcnc contribution for the @xmath167 component in the up sector , as shown in the last line from tab .
[ tab : left - handed - couplings ] .
first of all , we observe that the maximum mixing resides in the @xmath13 sector , which is about one order of magnitude bigger than other flavor - changing transitions , like for example in @xmath7 @xmath8 and @xmath168 systems , which induce the flavor changing transitions @xmath9 , @xmath10 , and @xmath169 , respectively .
secondly , the matsuda ansatz yields the biggest couplings , so that in order to control the low energy limits exhibit by eq .
( [ meson - mezcla ] ) , it is necessary to impose stronger restrictions to the new physics contribution induced by this ansatz , such as seen in figs .
[ fig - bs ] . .
[ cols="^,^,^,^,^,^,^,^,^",options="header " , ] in addition to the low energy differences shown by the plots in the above figures , the different sizes of the coupling @xmath166 leads to different predictions of the decay width of the @xmath1 boson into quarks .
in particular , the flavor - changing width can be written as @xmath170=\frac{% g_{l}^{2}m_{z^{\prime } } } { 8\pi c_{w}^{2}}\left [ \left ( \widetilde{b% } _ { l}^{qq^{\prime } }
\right ) ^{2}\right ] , \label{z'-width - fcnc}\ ] ] from tab .
[ tab : left - handed - couplings ] it is evident that the main source of flavor - changing decay is @xmath171 , with a decay probability of about @xmath172 bigger than others flavor - changing decays .
a detailed study of fcnc decays is carried out in ref .
@xcite in the matsuda ansatz .
other anstze , as the ones considered here , will yield lower values in the width .
in the framework of the 3 - 3 - 1 models , we have described the contribution to the mass difference @xmath22 in @xmath173 meson systems .
these models behave as a purely left - handed neutral flavor - changing model . using the recent experimental data and the sm one - loop prediction of @xmath22
, we found bounds for the mass of the @xmath18 boson in the foot - long - truan model ( flt ) and the pisano - pleitez - frampton model ( ppf ) .
the lowest values of @xmath174 are found when the weak angle associated to the @xmath5 coupling is @xmath156 by assuming four different anstze in the texture of the mass matrices of the quarks , we obtained plots of the allowed regions in the @xmath154 plane .
we considered the fritzsch ansatz ( @xmath142 ) , the matsuda - nishihura ansatz ( @xmath143 ) , and the matsuda ansatz ( @xmath144 ) for the rotation matrix @xmath141 in the down sector .
we also assumed another alternative , where @xmath145 .
lower bounds from @xmath174@xmath175 tev to @xmath176 tev in the flt model , and from @xmath174@xmath177 tev to @xmath178 tev in the ppf model are found for each ansatz of the rotation mass matrix . since the matsuda ansatz leads to the biggest size of the left - handed couplings as shown by tab .
[ tab : left - handed - couplings ] , this ansatz exhibit stronger low energy limits than the other anstze .
also , the matsuda texture yields a bigger probability of flavor - changing decay for the @xmath18 boson than the other anstze .
this work was partially supported by fundacin banco de la repblica and by alfa - ec funds through the helen programme .
f. ochoa would like to thank f. schrempp for helpful discussions and hospitality at desy , where part of this work was developed in the framework of the helen programme .
99 s. glashow , nucl .
22 , 579 ( 1961 ) ; s. weinberg , phys .
19 , 1264 ( 1967 ) ; a. salam , in elementary particle theory : relativistic groups and analyticity ( nobel symposium no .
8) , edited by n. svartholm ( almqvist and wiksell , stockholm , 1968 ) , p. 367 .
bennet and c.e .
wieman , phys .
82 , 2484 ( 1999 ) ; aleph , delphi , l3 , opal , sld , lep electroweak working group , sld electroweak and heavy flavor groups : s. schael et al . , hep - ex/0509008 ; ibid .
j. alcarez et al . , hep - ex/0511027 ; p.l .
anthony et.al .
[ slac e158 collaboration ] , phys .
95 , 081601 ( 2005 ) ; r.d .
young , r.d .
carlini , a.w .
thomas , j. roche , arxiv : hep - ph/0704.2618 ( 2007 ) .
f. pisano and v. pleitez , phys .
* d46 * , 410 ( 1992 ) ; r. foot , o.f .
hernandez , f. pisano , v. pleitez , phys .
d47 , 4158 ( 1993 ) ; v. pleitez and m.d .
tonasse , phys .
d48 , 2353 ( 1993 ) ; nguyen tuan anh , nguyen anh ky , hoang ngoc long , int .
a16 * , 541 ( 2001 ) .
frampton , phys .
* 69 * , 2889 ( 1992 ) ; p.h .
frampton , p. krastev and j.t .
liu , mod .
phys . lett . *
9a * , 761 ( 1994 ) ; p.h .
frampton et .
9a * , 1975 ( 1994 ) j.s .
bell , r. jackiw , nuovo cim . * a60 * 47 ( 1969 ) ; s.l .
adler , phys . rev . * 177 * , 2426 ( 1969 ) ; d.j .
gross , r. jackiw , phys.rev . * d6 * 477 ( 1972 ) .
h. georgi and s. l. glashow , phys . rev .
* d6 * 429 , ( 1972 ) ; s. okubo , phys . rev .
* d16 * , 3528 ( 1977 ) ; j. banks and h. georgi , phys . rev . * 14 * 1159 ( 1976 ) . paul langacker , m. plumacher , phys . rev . *
d62 * , 013006 ( 2000 ) ; v. barger , cheng - wei chiang , p. langacker , hye - sung lee , phys . lett .
* b580 * , 186 - 196 ( 2004 ) ; s. fajfer , p. singer , phys .
rev . * d65 * , 017301 ( 2002 ) ; m.a .
perez , m.a .
soriano , phys .
* d46 * , 284 ( 1992 ) ; d.l .
anderson and m. sher , phys . rev .
* d72 * , 095014 ( 2005 ) ; j. a. rodriguez and m. sher , phys . rev . *
d70 * 117702 ( 2004 ) ; c. promberger , s. schatt , f. schwab , arxiv : hep - ph/0702169 ( 2007 ) .
dumm , f. pisano , v. pleitez , mod .
a9 * , 1609 ( 1994 ) ; h. n. long and v. t. van , j. phys .
g : nucl . part .
25 * , 2319 ( 1999 ) ; c. promberger , s. schatt , f. schwab , phys . rev . *
d75 * , 115007 ( 2007 ) .
r. foot , h.n .
long and t.a .
tran , phys .
rev . * d50 * , r34 ( 1994 ) ; h.n .
long , ibid . * 53 * , 437 ( 1996 ) ; _ ibid _ , * 54 * , 4691 ( 1996 ) ; mod .
a * 13 * , 1865 ( 1998 ) .
rodolfo a. diaz , r. martinez , f. ochoa , phys .
rev . * d69 * , 095009 ( 2004 ) .
rodolfo a. diaz , r. martinez , f. ochoa , phys .
* d72 , * 035018 ( 2005 ) .
j. gunion , et .
al . , the higgs hunter s guide ( addison - wesley , new york , 1990 ) ; s. glashow and s. weinberg , phys .
* d15 * , 1958 ( 1977 ) ; d. atwood , l. reina and a. soni , phys .
lett . * 75 * , 3800 ( 1993 ) ; phys . rev . * d53 * , 1199 ( 1996 ) ; phys . rev . * d54 * , 3296 ( 1996 ) ; phys . rev . *
d55 * , 3156 ( 1997 ) ; m. sher and yao yuan , phys . rev . * d44 * , 1461 ( 1991 ) ; t.p .
cheng and m. sher , phys . rev . * d35 * , 3490 ( 1987 ) .
h. fritzsch , phys .
73b , 317 ( 1978 ) ; t. hagiwara et al .
phys . lett .
76b , 602 ( 1978 ) ; m. de crombrugghe , phys .
80b , 365 ( 1979 ) ; h. fritzsch , nucl .
b155 , 189 ( 1979 ) ; h. fritzsch , p. minkowski , phys .
73 , 67 ( 1981 ) .
d. du and z - z .
xing , phys .
d48 , 2349 ( 1993 ) ; l.j . hall and a. rasin , phys .
b315 , 164 ( 1993 ) ; h. fritzsch and d. holtmanspotter , phys . lett .
b 338 , 290 ( 1994 ) ; h. fritzsch and z - z .
xing , phys .
b353 , 114 ( 1995 ) ; 413 , 396 ( 1997 ) ; phys .
d57 , 594 ( 1998 ) ; nucl .
b556 , 49 ( 1999 ) ; g.c .
branco , d. emmanuel - costa and r.g .
felipe , phys .
b477 , 147 ( 2000 ) ; r. rosenfeld and j.l .
rosner , phys .
b516 , 408 ( 2001 ) ; j.l .
chkareuli and c.d .
froggatt , nucl .
b626 , 307 ( 2002 ) ; h. fritzsch and z - z .
xing , phys .
b555 , 63 ( 2003 ) . | comparing the theoretically predicted and measured values of the mass difference of the @xmath0 system , we estimate the lower bound on the mass of the @xmath1 boson of models based on the @xmath2 gauge group . by assuming zero - texture approaches of the quark mass matrices
, we find the ratio of the measured value to the theoretical prediction from the standard model and the @xmath1 contribution from the 331 models of the mass difference of the @xmath0 system .
we find lower bounds on the @xmath1 mass ranging between @xmath3 tev and @xmath4 tev for the two most popular 331 models , and four different zero - textures anstze .
the above results are expressed as a function of the weak angle associated to the @xmath5 couplings . |
the alexander polynomial of singular curves in @xmath12 provides an effective way to relate the fundamental group of the complement of such curves to the topology and geometry of their singularities . in this paper
we show that the mordell - weil group of certain elliptic threefolds with constant @xmath13-invariant equal to either zero or 1728 is closely related to the alexander module of plane curves associated with the threefolds . as a byproduct of this relation
we obtain a bound on the degree of the alexander polynomials of plane curves . the bound is linear in the degree of the curve and gives a new restriction on the groups which can be fundamental groups of the complements to plane curves .
let @xmath14 be a curve in @xmath15 of degree @xmath1 .
its alexander polynomial is defined in terms of purely topological data : @xmath16 , where @xmath10 is a line at infinity , and a surjection @xmath17 ( cf .
section [ defalpol ] for a definition ) . on the other hand , dependence of the alexander polynomial of @xmath18 , on its degree
, the local type of its singularities , and their _ position _ , relating the latter to the topology , has been known for some time ( cf .
for example , for irreducible curves having nodes and cusps as the only singularities , the degree of the alexander polynomial @xmath8 equals @xmath19 where @xmath20 and @xmath21 are respectively the first and second commutators of @xmath22 ( in this case there is only one choice , up to sign , of @xmath23 ) .
moreover , @xmath8 is not trivial only if @xmath24 , in which case @xmath25 , where @xmath26 is the superabundance of the curves of degree @xmath27 passing through the cusps of @xmath18 . for a given curve
, this provides a purely geometric method for calculation of the alexander polynomial .
however , how big can this superabundance be for a special cuspidal curve is still not known ( cf .
the largest known value of @xmath26 for irreducible cuspidal curves , to our knowledge , is 3 .
this occurs for the dual curve to a non - singular cubic , that is a sextic with 9 cusps . in this paper
we give an example of an irreducible curve with nodes and cusps as only singularities , for which the superabundance of the set of cusps is equal to 4 .
one of our main results is the inequality ( cf .
corollary [ inequality ] ) : @xmath28 for reducible curves and general @xmath23 , the explicit relation with the fundamental group comes from the equality ( cf .
@xcite ) : @xmath29
where @xmath30 $ ] , @xmath31 ( if @xmath18 is a curve in @xmath32 then @xmath33 is the number of irreducible components of @xmath18 ) , @xmath34 is an epimorphism , and @xmath35 represents the class of a meridian around the @xmath36-th irreducible component of @xmath18 , i.e. its abelian image in @xmath37 is @xmath38 .
the main idea in this note is to relate the degree of the alexander polynomial to the rank of the mordell - weil group over the field @xmath39 of elliptic curves with @xmath13-invariant equal to either zero or 1728 .
this is done by relating the alexander module of @xmath22 to rational pencils of elliptic type corresponding to @xmath11 ( cf . definition [ def - orb ] ) .
the relationship between alexander invariants and pencils , in the case of line arrangements and reducible curves is discussed in @xcite : the positive dimensional components of characteristic varieties induce maps between their complements and the complements to @xmath40 points in @xmath41 . here
we show that non - vanishing of alexander polynomial of a curve yields existence of special pencils with non - reduced fibers .
these pencils are such that they induce rational maps from @xmath15 onto @xmath41 with an orbifold structure and take the curve @xmath18 onto a finite set of points .
the pencils are orbifold elliptic pencils in the following sense ( cf .
definitions [ def - orb ] and [ dfn - group - orb ] ) : each map @xmath42 has three non - reduced fibers of the form @xmath43 , where @xmath44 divides the equation of @xmath18 .
the multiplicities @xmath45 are such that @xmath46 and thus the orbifold structure is given by assigning multiplicities @xmath45 to those three points in @xmath41 .
the possible orbifold structures @xmath47 depend on the local type of singularities of @xmath18 .
for example for irreducible curves with only nodes and cusps as singularities , we relate the global alexander polynomial of @xmath18 to the following functional relation : @xmath48 where @xmath11 is an equation of @xmath18 and @xmath49 are homogeneous polynomials .
such a relation is equivalent to the existence of a rational map from @xmath15 onto @xmath41 with the orbifold structure @xmath50 .
this orbifold structure can be obtained considering the global orbifold in the usual sense ( cf .
@xcite ) corresponding to the action of a cyclic group of order 6 on an elliptic curve with non - trivial stabilizers at three points ; their orders being equal to @xmath51 and @xmath52 respectively .
the correspondence between the alexander modules and orbifold elliptic pencils is established in two rather different steps .
on one hand , the alexander module of @xmath18 can be related to the mordell - weil group of the elliptic threefold : @xmath53 over the field @xmath39 of rational functions in two variables having @xmath13-invariant equal to zero .
we have the following ( cf .
theorem [ theoremonwf ] ) : let @xmath18 be an irreducible curve in @xmath15 having ordinary nodes and cusps as the only singularities .
let @xmath54 be a ( reduced ) equation of the affine part of @xmath18 .
then the @xmath55-rank of the mordell - weil group of the elliptic threefold ( [ threefold-2 ] ) is equal to the degree of the alexander polynomial of the curve @xmath18 .
the mordell - weil group here is the group of rational sections of the elliptic threefolds ( see for instance @xcite and section [ sec - mordellweil ] for further discussion ) .
the rank of the mordell - weil group of the threefold was recently studied in @xcite for the case @xmath56 using different methods ( cf . also @xcite ) .
these results follow immediately from the correspondence between alexander polynomials and mordell - weil groups in this paper since the alexander polynomials of sextic curves considered in @xcite are readily available . on the other hand ,
each element of the mordell - weil group of the aforementioned elliptic threefold defines a functional relation of the type .
this can be summarized as follows ( cf .
theorem [ cor - kulikov ] ) : for any irreducible plane curve @xmath57 whose only singularities are nodes and cusps the following statements are equivalent : 1 .
@xmath18 admits a quasi - toric relation of elliptic type @xmath50 , 2 .
@xmath18 admits an infinite number of quasi - toric relations of elliptic type @xmath50 , 3 .
@xmath8 is not trivial ( @xmath58 ) .
, up to sign , i.e. @xmath59 is independent of it ; for other types of quasi - toric relations see section [ sec - examples ] ] moreover , the set of quasi - toric relations of @xmath18 @xmath60 ^ 3\mid f^2+g^3+h^6f=0\}$ ] has a group structure and it is isomorphic to @xmath61 , where @xmath62 .
also , @xmath18 admits an infinite number of primitive quasi - toric relations , unless @xmath63 , in which case @xmath18 only has one primitive quasi - toric relation .
we also consider here other singularities which will define relations of the form : @xmath64 where @xmath65 is either @xmath66 or @xmath67 .
such relations in turn correspond to orbifold rational pencils with respective orbifold structures ( cf .
theorem [ cor - kulikov - gen ] ) . while commonly the existence of irrational pencils on surfaces is obtained by an extension of the defranchis method ( from 1-form with vanishing wedge product , cf .
@xcite ) , rational orbifold pencils are obtained rather differently : here the pencils are a byproduct of the splitting of albanese varieties of cyclic multiple planes into a product of elliptic curves which we derive either from roan s decomposition theorem for abelian varieties with an automorphism ( cf .
@xcite ) or directly , using a hodge theoretical refinement of the argument used in the proof of divisibility theorem in @xcite .
for example for the cyclic multiple planes branched over a reduced curve with nodes and cusps as the only singularities , the albanese variety splits as a product of elliptic curves with @xmath13-invariant equal to zero .
one may contrast this with the case of jacobians of curves where the situation is different ( cf . @xcite ) .
in particular , the relation between alexander polynomials and mordell - weil groups allows us to give bounds on the degree of alexander polynomials .
these follow from the bounds on the rank of mordell - weil groups obtained from the connection with the mordell - weil groups of certain elliptic surfaces and from the shioda - tate formula ( cf .
@xcite ) . however , the correspondence between the alexander polynomials and the ranks of the mordell - weil group should be of independent interest ( cf .
@xcite ) .
the curves with the largest known values of @xmath68 are given in section [ sec - examples ] .
the bound presented here is sharp for sextics , however , the inequality is apparently far from being sharp in general .
perhaps a better understanding of the mordell - weil rank of can yield a better estimate .
also , note that corollary [ inequality ] provides a partial answer to ( * ? ? ?
* problem 2.1 ) .
another application of the results presented in this paper is an alternative argument to confirm oka s conjecture of sextic curves having a non - trivial alexander polynomial ( i.e. that equations of such curves have form @xmath69 ) .
the answer to oka s conjecture was first obtained by a. degtyarev ( cf .
@xcite ) . for the sake of clarity
we often start our discussions focusing on the case of irreducible curves having only nodes and cusps as singularities .
the results , however , are obtained , as was already mentioned , for curves with a wider class of singularities , which we call @xmath70-essential and @xmath70-partial ( cf .
[ deltapartial ] ) . moreover the results are applicable to reducible and non - reduced curves as well . from the point of view of fundamental groups ,
non - reduced curves correspond to homomorphisms @xmath23 that are more general than those given by the linking number of loops with @xmath18 .
the condition of being @xmath70-essential is purely local , meaning that the local alexander polynomial of the link of the singularity considered w.r.t .
the restriction of @xmath23 on the local fundamental group is divisible by the cyclotomic polynomial of degree @xmath70 .
as we shall see , this is the natural class of curves leading to the elliptic pencils .
the content of the paper is as follows . in section [ sec - settings ]
we give definitions for alexander polynomials w.r.t .
any homomorphism @xmath23 ( as mentioned above ) , for orbifold surfaces and morphisms , and for quasi - toric relations . in section [ sec - mordellweil ]
we relate the alexander polynomial to the mordell - weil group of the threefold ( [ threefold-2 ] ) associated with @xmath11 . in section [ sec - qt - rels ] , the @xmath39-points of the threefolds
are presented as quasi - toric relations of @xmath11 , i.e. functional equations of the form @xmath71 , over the ring @xmath72 $ ] .
the correspondence between the points of and the quasi - toric relations where @xmath11 fits in , is explained at the end of section [ sec - mordellweil ] . in section [ sec - delta - curves ]
we generalize the results to a larger class of curves .
finally , in section [ sec - examples ] a list of explicit examples and applications is given as to show curves that fit into quasi - toric relations of all rational orbifolds of elliptic type .
also one of these examples provides the largest values known to date of the degree of the alexander polynomial of an irreducible curve with nodes and cusps .
@xmath11 a ( possibly reducible or even non - reduced ) non - zero homogeneous polynomial in @xmath72 $ ] such that @xmath11 is not a power , that is @xmath73 for @xmath74 $ ] implies @xmath75 .
@xmath18 the set of zeroes of @xmath11 , hence a reduced curve . @xmath76 a non singular model of a cyclic multiple plane @xmath77 @xmath78 a model of cyclic multiple plane in @xmath79 .
@xmath80 an affine model of the elliptic threefold corresponding to a curve @xmath81 .
@xmath82 a smooth projective birational model of @xmath80 .
@xmath83 singular projective model of @xmath82 ( cf .
proposition [ equationforw ] ) .
@xmath84 the elliptic curve with j - invariant zero .
@xmath85 a model of @xmath84 in @xmath86 .
@xmath87 a split elliptic threefold @xmath88 .
@xmath89 a birational model of split elliptic threefold @xmath88
in this section we will review several results on alexander invariants which appear in the literature and extend them to the generality required in this paper .
we shall consider reducible , not necessarily reduced curves in @xmath15 .
let @xmath18 be a plane curve given as the set of zeroes of a homogeneous polynomial @xmath90 , which is not a power ( see notation [ ssec - notation ] ) , which means @xmath91 .
consider @xmath92 its decomposition into irreducible components and let @xmath93 , @xmath94 be the collection of multiplicities of each irreducible factor @xmath44 in the equation of @xmath18 .
so if @xmath95 denotes the degree of @xmath96 , then the total degree of @xmath97 is given by @xmath98 .
let @xmath10 be a line transversal to @xmath18 which we shall view as the line at infinity and let @xmath99 . recall that @xmath100 is a free abelian group with @xmath33 generators having canonical identification with @xmath101 ( cf .
the isomorphism is given by mapping the class of the boundary of a small holomorphic 2-disk transversal to the component @xmath96 to @xmath102 ( with @xmath103 appearing as the @xmath36-th component ) .
let @xmath23 be the epimorphism @xmath104 , given by @xmath105 .
let @xmath106={\mathbb q}[t , t^{-1}]$ ] denote the group ring over @xmath107 , @xmath108 , and @xmath109 $ ] be the commutator of @xmath110 . by the hurewicz theorem @xmath111
can be identified with the homology of infinite cyclic cover of @xmath112 corresponding to @xmath23 and hence can be viewed as a module over the group ring @xmath113 $ ] .
homomorphisms ( resp
. epimorphism ) @xmath114 such that @xmath115 for all @xmath36 , are in one - to - one correspondence with polynomials , ( resp .
non - power polynomials ) considered up to a scalar factor , having @xmath18 as the zero set .
indeed if @xmath44 is an irreducible polynomial having @xmath96 as its set of zeros , then one defines the polynomial corresponding to @xmath23 as @xmath116 vice versa , given a polynomial @xmath11 having @xmath18 as its zero set one defines the homomorphism @xmath117 using @xmath118 and extends it as the composition @xmath119 .
this homomorphism is in fact an epimorphism if @xmath11 is not a power .
it will also be denoted by @xmath120 .
@xcite ) the alexander polynomial @xmath8 of @xmath18 relative to a surjection @xmath121 is a generator of the order of the torsion of the @xmath106={\mathbb q}[t , t^{-1}]$]-module @xmath122 normalized in such a way that it is a polynomial in @xmath123 satisfying @xmath124 . by the discussion in the previous paragraph , a ( not necessarily reduced ) equation @xmath11 of @xmath18 defines both the set of zeroes @xmath18 and a surjection @xmath120 .
hence @xmath125 will also denote @xmath126 .
the alexander polynomial @xmath59 can be expressed in terms of characteristic varieties studied in @xcite or in terms of @xmath127 viewed as a module over @xmath128={\mathbb z}[t_1^{\pm 1},\dots , t_r^{\pm 1}]$ ] as follows .
the polynomial @xmath59 is the order of the torsion of @xmath129 up to a power of @xmath130 , viewed as a @xmath131$]-module in the obvious way .
the zeroes of @xmath59 can also be seen as the intersection of the characteristic variety @xmath132 with the 1-dimensional torus of equation @xmath133 ( cf .
* theorem 2.26 ) ) .
, case @xmath134 was considered for example by oka in @xcite as _ @xmath135-alexander polynomials_. ] we shall also need the local version of the polynomials @xmath59 defined similarly .
let @xmath136 be a singular point of @xmath18 .
the epimorphism @xmath17 induces a homomorphism @xmath137 of the local fundamental group of @xmath18 to @xmath55 and hence the alexander polynomial of the link of @xmath136 w.r.t . the homomorphism @xmath137 .
has index @xmath138 in @xmath55 , then the alexander polynomial @xmath139 is @xmath140 where @xmath141 is the alexander polynomial relative to the surjection @xmath142 .
recall that this represents the order of the 1-dimensional homology of the cyclic cover corresponding to @xmath23 which is this case has @xmath138 connected components cf .
@xcite ] in other words , if @xmath143 is the inclusion from a sufficiently small sphere around @xmath136 in the total space and @xmath144 is a meridian around a component of the link , then @xmath145 these polynomials will be denoted by @xmath139 .
we have the following : [ prop - divisibility ] let @xmath18 be a plane curve and @xmath93 denote the multiplicities of its irreducible .
then @xmath8 divides the product of the local alexander polynomials : @xmath146 in particular , if the local alexander polynomials have only roots of unity of degree @xmath70 as their roots , then : @xmath147 where @xmath148 is the cyclotomic polynomial of the @xmath149-th roots of unity .
moreover , if @xmath150 , then @xmath151 where @xmath152 is the total degree .
details of the arguments are similar to those used in the proof of the divisibility theorem ( cf .
@xcite and lemma [ albaneseeigenvalues ] below ) .
the starting point is the surjection of the fundamental group of a regular neighborhood of @xmath18 in @xmath153 onto @xmath154 ( which is a consequence of the lefschetz hyperplane section theorem ) . on the other hand
, one uses the mayer - vietoris sequence for the @xmath23-cyclic cover of this neighborhood in order to split it into a union of cyclic covers : those of the local singularities and the neighborhood of the non - singular part of @xmath18 .
this shows that the alexander polynomial of the neighborhood of @xmath18 is equal to the product of the alexander polynomials of singularities and a divisor of second product in .
these divisors come as contributions of @xmath155 and @xmath37 of terms of the mayer - vietoris sequence corresponding to the intersections of the complements to the links of the singularities with the mentioned @xmath156-bundle of the non - singular part of @xmath18 .
this yields the divisibility .
the second divisibility relation is a generalization of the divisibility at infinity ( cf .
@xcite ) and follows from the calculation of the alexander polynomial of the link at infinity with multiplicities .
it is equal to @xmath157 as a consequence of the torres relation ( cf .
@xcite ) applied to the hopf link with multivariable alexander polynomial @xmath158 .
now we recall some basic definitions needed here referring for more details to ( see @xcite ) .
[ def - orb ] an _ orbifold _
curve @xmath159 is ( an open or closed ) riemann surface @xmath160 with a function @xmath161 whose value is @xmath103 outside a finite number of points .
a point @xmath162 for which @xmath163 is called an _
orbifold point_. one may think of a neighborhood of a point @xmath164 with @xmath165 as the quotient of a disk ( centered at @xmath136 ) by a rotation of angle @xmath166 .
a loop around @xmath136 is considered to be trivial in @xmath159 if its lifting bounds a disk .
following this idea , orbifold fundamental groups can be defined as follows .
[ dfn - group - orb ] for an orbifold @xmath159 , let @xmath167 be the orbifold points , @xmath168 . then
, the _ orbifold fundamental group _ of @xmath159 is @xmath169 where @xmath170 is a meridian of @xmath171 .
we will denote @xmath159 simply by @xmath172 . in this paper
, we will mostly consider orbifold groups of @xmath41 with three orbifold points .
the groups @xmath173 are subgroups of index two of full triangle groups . in particular , they can be identified with the orientation - preserving isometries of a plane tiled with triangles with angles @xmath174 , @xmath175 , and @xmath176 ( cf .
* corollary 2.5 ) ) .
a dominant algebraic morphism @xmath177 defines an _ orbifold morphism _
@xmath178 if for all @xmath162 , the divisor @xmath179 is a @xmath180-multiple .
one has the following result regarding orbifold morphisms .
[ prop - orb ] let @xmath181 define an _ orbifold morphism _ @xmath178
. then @xmath182 induces a morphism @xmath183 .
moreover , if the generic fiber is connected , then @xmath184 is surjective . proposition [ prop - orb ] will be applied systematically throughout this paper .
we will show a typical example of this .
suppose @xmath185 fits in a functional equation of type @xmath186 where @xmath187 , and @xmath188 are polynomials .
note that @xmath44 are not necessarily irreducible or reduced .
also note that induces a pencil map @xmath189 given by @xmath190)=[h_1 ^ 3f_1:h_2 ^ 2f_2]$ ] .
consider @xmath97 .
since @xmath191 has three multiple fibers ( over @xmath192 $ ] , @xmath193 $ ] , and @xmath194 $ ] ) one has an orbifold morphism @xmath195 . in particular , if the pencil @xmath182 is primitive ( in the sense that it coincides with its stein factorization ) then by proposition [ prop - orb ] , there is an epimorphism @xmath196 where @xmath197 are as in definition [ dfn - group - orb ] and @xmath198 according to .
finally , note that @xmath199 , and the elements @xmath200 are roots of @xmath8 , where @xmath201}={\mathbb z}_6\ ] ] is induced by @xmath202 on the abelianizations of the groups .
therefore one has that @xmath8 is of the form @xmath203 where @xmath204 has no 6-th roots of unity as zeroes and @xmath205 are non - negative integers .
using this technique one can show the following result needed in the sequel and which we shall prove for completeness ( cf .
@xcite , ch 2 , theorem 2.3 ) .
[ prop - bound - pencil ] the number of multiple members in a primitive pencil of plane curves ( with no base components ) is at most two .
let us assume that the pencil is generated by two multiple fibers , that is , @xmath189 , given by @xmath190)=[f^p : g^q]$ ] , where @xmath206 ( otherwise , the pencil is not primitive ) .
assume there is a third multiple member , that is , @xmath207 . according to proposition [ prop - orb ] , one obtains an orbifold morphism with connected fibers , and thus an epimorphism @xmath208 i.e. both groups are trivial .
the groups @xmath209 are virtually torsion free ( cf .
* theorem 2.7 ) ) and they are subgroups of order two of the schwartz group , which is infinite ( cf .
* corollary 2.4 ) ) .
in particular , they are non - trivial , which contradicts the triviality of @xmath210 . for pencils other than pencils of plane curves , using logarithmic transforms , one can obtain elliptic fibrations with any number of multiple fibers ( cf .
@xcite ) .
in this paper we are interested in a particular type of orbifold morphisms .
we say a 2-dimensional orbifold @xmath159 is a _ rational orbifold curve of elliptic type _ if @xmath211 ( that is @xmath159 is a compact riemann sphere ) and @xmath212 where @xmath213 is the number of orbifold points .
[ lemma - elliptic - type ] the possible rational orbifold curves of elliptic type are : 1 .
@xmath214 2 .
@xmath215 3 .
@xmath216 and 4 .
@xmath217 [ rem - elliptic ] the reason to call such orbifolds of elliptic type is the following .
consider @xmath218 the regular covering of order @xmath219 ( the least common multiple of the orbifold multiplicities ) ramified with index @xmath45 at the @xmath220 preimages of @xmath221 ( the orbifold point of order @xmath45 ) of @xmath159 .
in general , if @xmath222 , then @xmath218 is a riemann surface of genus @xmath223 thus , according to lemma [ lemma - elliptic - type ] , for any rational orbifold curve @xmath218 of elliptic type the associated covering @xmath218 is an elliptic curve ( that is , a complex compact curve of genus 1 ) .
the results in this paper require that the least common multiple of the orbifold indices be @xmath224 .
therefore type @xmath225 from the list above will be disregarded .
a _ quasi - toric relation of type @xmath65 _ is a sextuple @xmath226 of non - zero homogeneous polynomials in @xmath72 $ ] satisfying the following functional relation @xmath227 the _ support _ of a quasi - toric relation @xmath228 as above is the zero set @xmath229 . in this context
, we may also refer to @xmath18 as a curve that _ satisfies ( or supports ) a quasi - toric relation _ of type @xmath65 .
we will say a quasi - toric relation of type @xmath65 is _ of elliptic type _ if @xmath65 is of the form @xmath230 in lemma [ lemma - elliptic - type ] .
given a quasi - toric relation @xmath231 as above , note that @xmath232 and hence @xmath233 therefore , if @xmath228 is of elliptic type , then @xmath234 special classes of curves satisfying quasi - toric relations have already been considered , namely , the class of _ curves of torus type _ @xmath235 ,
i.e. quasi - toric relations of type @xmath236 of the form @xmath237 ( cf .
@xcite ) and the class of _ curves having quasi - toric decompositions _
@xmath238 , i.e. quasi - toric relations of type @xmath238 of the form @xmath239 ( cf .
also , functional equations of a similar sort appear in @xcite .
recall that a _ quasi - toric decomposition of @xmath11 _ is a collection of homogeneous polynomials @xmath240 $ ] such that the following identity holds : @xmath241 for two co - prime positive integers @xmath242 .
analogously , a curve is of torus type @xmath243 if it admits a quasi - toric decomposition as in where @xmath244 .
to each quasi - toric relation @xmath228 of elliptic type satisfying there corresponds a map from a certain cyclic multiple plane branched over the curve supporting @xmath228 to the elliptic curve @xmath245 ( see remark [ rem - elliptic ] ) as follows .
let @xmath246 and @xmath247 ( note that according to one has @xmath248 ) .
let @xmath249 be the following surface given in the weighted projective space @xmath250 @xmath251 then let @xmath245 be the elliptic curve in @xmath15 given by the equation @xmath252 .
to each quasi - toric relation there corresponds the following map @xmath253 ( note that the map is well defined by and ) .
we will say that two quasi - toric relations , @xmath254 and @xmath255 of the same elliptic type @xmath256 are _ equivalent _ iff the corresponding maps coincide i.e. there exists a non - trivial rational function @xmath257 such that @xmath258 [ ex-236 ] note that any quasi - toric relation @xmath259 of elliptic type @xmath50 is equivalent to one of the form @xmath260 since @xmath261 , @xmath262 , @xmath263 , and @xmath264 satisfy . in other words ,
any @xmath50 quasi - toric relation is equivalent to one of the form : @xmath265 this stresses the idea that non - reduced components are indeed required .
in this section , let us fix @xmath267 $ ] , an irreducible homogeneous polynomial of degree @xmath268 , whose set of zeroes in @xmath15 is a curve @xmath18 that has only nodes and cusps as singularities ( i.e. @xmath18 has either @xmath269 or @xmath270 as a local equation around each singular point ) .
the homomorphism @xmath271 , which is given by the total linking number with @xmath18 in the affine plane @xmath153 , in this case corresponds to the trivial morphism @xmath272 for any meridian around @xmath18 .
consider a threefold @xmath82 containing , as a zariski open subset , the affine threefold @xmath80 given in @xmath273 by the following equation : @xmath274 the projection onto the @xmath275-plane exhibits @xmath82 as an elliptic threefold whose fibers over generic points have @xmath13-invariant equal to zero , the actual choice of @xmath82 is not important .
nevertheless , a particular biregular model of @xmath82 will be given ( cf .
lemma [ equationforw ] ) ] the main result of this section is a calculation of the mordell - weil group of the @xmath39-points of @xmath82 in terms of the classical alexander polynomial @xmath276 of @xmath18 ( that is , the alexander polynomial w.r.t .
the homomorphism @xmath23 described above ) .
[ theoremonwf ] the @xmath55-rank of the mordell - weil group of @xmath82 over @xmath39 is equal to the degree @xmath277 of the alexander polynomial of @xmath18
. let @xmath278 ( or simply @xmath76 ) denote a smooth model the 6-fold cyclic cover of @xmath15 branched along @xmath81 corresponding to the surjection @xmath279 which is defined using the composition of the homomorphism as above and the reduction modulo @xmath52 .
@xmath278 contains , as an open subset , the affine hypersurface in @xmath280 given by the equation @xmath281 we shall assume that @xmath278 supports a holomorphic action of the group @xmath282 of roots of unity of degree 6 , extending the action of this group affine surface given by @xmath283 where @xmath284 is selected primitive root of unity of degree 6 .
such smooth model @xmath76 can be obtained for example using an equivariant resolution of singularities of projective closure of ( [ multipleplane ] ) .
recall that the degree of the alexander polynomial @xmath276 of @xmath18 is equal to @xmath285 , where @xmath286 is the irregularity @xmath287 of @xmath76 ( cf .
let @xmath84 denote the elliptic curve with @xmath13-invariant equal to zero .
as its biregular model one can take the projective closure of affine curve @xmath288 .
we shall start the proof of theorem [ theoremonwf ] by describing the albanese varieties of six - fold cyclic multiple planes @xmath76 .
such a description will be based on the following decomposition theorem due to s. roan ( see ( * ? ? ? * theorem 3.2 ) ) .
[ roanth ] let @xmath289 be an abelian variety of dimension @xmath5 and @xmath290 an automorphism of @xmath289 of order @xmath291 .
let @xmath292 be the collection of the eigenvalues of the automorphism @xmath293 induced by @xmath290 on the universal cover of @xmath289 and let @xmath294 be the union of fixed points of powers @xmath295 for @xmath296 .
assume that @xmath294 is finite and @xmath297 , where @xmath182 is the euler function .
then @xmath298 and @xmath299 is a cm - field ( cf .
* section 3 ) ) .
moreover , there are @xmath300 abelian varieties @xmath301 of cm - type @xmath299 s.t .
@xmath289 factors as @xmath302 and @xmath290 decomposes into a product of automorphisms of @xmath303 induced by primitive @xmath1-th roots of unity .
recall ( cf .
@xcite ) that the albanese variety of a smooth projective variety @xmath289 is defined as @xmath304 and the albanese map @xmath305 is given by @xmath306 where @xmath307 is a base point and the integral in ( [ albanesemap ] ) is viewed as a linear function on @xmath308 , well defined up to the periods ( i.e. the integrals over loops representing classes in @xmath309 ) . the map @xmath310 is well defined up to translation ( i.e. a choice of @xmath307 ) . a choice of a positive line bundle on @xmath289 yields a polarization of @xmath311 making it into an abelian variety . in the case of the cyclic multiple plane @xmath76 we select a point @xmath307 in the locus with maximal ramification index of its projection onto @xmath15 .
the map @xmath310 is universal in the sense that for any map @xmath312 into an abelian variety @xmath313 , there exists a factorization : @xmath314 it follows from ( [ defalb ] ) that when @xmath289 carries a biholomorphic action of a group @xmath22 fixing the base point @xmath307 , the action of @xmath22 on @xmath308 , given by @xmath315 , induces the action on @xmath311 . for this action
the map @xmath310 is equivariant : @xmath316 for example , for @xmath317 one has @xmath318 .
we will need also a local version of the above construction corresponding to the mixed hodge structure associated with a germ of plane curve singularity ( with assumptions stated below ) : @xmath319 while there are several constructions of a mixed hodge structure associated with a germ of singularity ( [ planesing ] ) ( cf .
@xcite ) , we shall consider only the case when the monodromy action on the cohomology of milnor fiber is semi - simple ( e.g. the ordinary cusps , nodes and more generally the singularities appearing in tables [ tab : d - ess ] , [ tab : d-4-ess ] and [ tab : d-3-ess ] in section [ sec - delta - curves ] ) . in this case one
can identify the ( co)homology of the milnor fiber with the ( co)homology of the link of the surface singularity @xmath320 where @xmath321 is the order of the monodromy of the cohomology of milnor fiber ( cf . for example @xcite for a similar discussion ) .
more precisely we have the following : [ comparemhs ] let ( [ planesing ] ) be a germ of a plane curve ( possibly reducible and non - reduced ) with semi - simple monodromy of order @xmath321 and the milnor fiber @xmath322 .
let @xmath323 be link of the corresponding surface singularity ( [ localcover ] ) .
then there is the isomorphism of the mixed hodge structures : @xmath324 where the mixed hodge structure on the left is the tate twist of the mixed hodge structure constructed in @xcite or @xcite and the one on the right is the mixed hodge structure on vanishing cohomology constructed in @xcite recall that the construction of the mixed hodge structure on the cohomology of a link of an isolated singularity is based on the identification of the latter with the cohomology of the punctured regular neighborhood of the exceptional set in a resolution for ( [ localcover ] ) .
alternatively , one can use the mixed hodge structure on the local cohomology of supported at the singularity of .
dualizing one obtains the mixed hodge structure on the homology as well .
consider @xmath325 as a holomorphic function on the germ @xmath326 of the surface singularity @xmath327 ( cf .
( [ localcover ] ) ) and its milnor fiber @xmath328 i.e. the subset of @xmath326 given by the equation @xmath329 for fixed @xmath123 .
it has canonical identification with the milnor fiber of @xmath330 over @xmath331 .
monodromy of the milnor fiber of @xmath325 coincides with @xmath332 where @xmath333 is the monodromy of @xmath334 .
denote by @xmath335 the subset of @xmath326 given by @xmath336 and consider the wang exact sequence of the mixed hodge structures ( cf . for example @xcite section 1.7 ) ) : @xmath337 the last term of this sequence is zero since the milnor fiber has the homotopy type of a 1-complex . since @xmath338 and
since @xmath339 has the mixed hodge structure on its cohomology is constructed via the identification with the mixed hodge structure on @xmath340 we obtain isomorphism : @xmath341 it provides an identification of the cohomology of punctured neighborhood @xmath342 with the cohomology of milnor fiber i.e. @xmath343 moreover , the exact sequence of pair : @xmath344 yields the isomorphism @xmath345 this together with ( [ partialisomorphism ] ) yields the claim of the lemma .
construction of the mixed hodge structure on either side of ( [ lemmaisomorphism ] ) yields that the type of this mixed hodge structure on corresponding _ homology _ is @xmath346 in fact with assumption of semi - simplicity made in lemma [ comparemhs ] the mixed hodge structure on the homology of milnor fiber is pure of type @xmath347 .
the equivalence of categories of such mixed hodge structures without torsion and the categories of 1-motifs constructed in @xcite allows to extract the abelian variety @xmath313 which is the part of structure of 1-motif .
we shall refer to this abelian variety as the local albanese variety of the singularity .
the polarization of the hodge structure on @xmath348 , required for such equivalence is the standard polarization of such graded component associated with the milnor fiber . in the case of the cusp @xmath270 the assumptions of the lemma [ comparemhs ]
are fulfilled .
one can also describe the construction of corresponding local albanese variety as taking the quotient of @xmath349 by the homology lattice of the closed riemann surface which is the compactified milnor fiber .
in particular the hodge structure on the cohomology of its milnor fiber @xmath350 is the hodge structure of the elliptic curve @xmath84 with @xmath266 .
the above discussion yields : the eigenvalues of the generator @xmath351 of the group of covering transformations ( [ localcover ] ) acting on the cohomology of the singularity link coincide with the eigenvalues of the action of the monodromy on the cohomology of the milnor fiber ( cf .
@xcite or @xcite ) .
the eigenvalues of the above generator of the group of deck transformations of the germ [ localcover ] ) acting on @xmath352 have the form @xmath353 where @xmath290 runs through the elements the spectrum of the singularity which belongs to the interval @xmath354 ( cf .
@xcite ) .
for example , in the case of the cusp @xmath270 the monodromy , corresponding to the path @xmath355 ( @xmath356 ) in the positive direction given by the complex structure yields the monodromy of the milnor fiber @xmath328 in lemma [ comparemhs ] given by @xmath357 in order to apply theorem [ roanth ] to @xmath358 , we shall need the following : [ albaneseeigenvalues ] let , as above , @xmath76 be a smooth @xmath359-equivariant model of a cyclic 6-fold covering space of @xmath15 branched over a curve with only nodes and cusps as singularities .
let @xmath333 be a generator of covering group of @xmath76 .
the automorphism of @xmath358 induced by @xmath333 has only one eigenvalue ( which is a primitive root of unity of degree six ) .
it follows from @xcite that the alexander module of a cyclic multiple plane branched over a curve with nodes and cusps , up to summands @xmath131/(t-1)$ ] , is isomorphic to a direct sum @xmath360/(t^2-t+1)]^s.\ ] ] more precisely , the proof of the divisibility theorem @xcite shows that the alexander module is a quotient of the direct sum of the alexander modules of all singularities or equivalently the direct sum of the homology of milnor fibers of singularities with module structure given by the action of the monodromy ( in the present case of the cusp @xmath270 the local alexander module is just one summand in ) .
since the divisibility result is stated in @xcite with assumption of irreducibility of the ramification locus @xmath18 of we shall review the argument .
denote by @xmath361 a tubular neighborhood of @xmath18 in @xmath15 and consider the surjection @xmath362 induced by embedding ( the surjectivity is a consequence of the surjectivity , for @xmath363 , of the map @xmath364 , in turn , following from the weak lefschetz theorem ) .
the surjection ( [ surjectivityone ] ) induces the surjection of the homology of 6-fold cyclic coverings : @xmath365 on the other hand , the covering space @xmath366 decomposes as @xmath367 where @xmath368 are small regular neighborhoods of all singular points @xmath136 of @xmath18 , @xmath369 is the regular neighborhood of the smooth locus of @xmath18 and the subscript designates the 6-fold cyclic cover .
the corresponding mayer vietoris sequence yields a surjection : @xmath370 one can view the cohomology of the 6-fold cover of @xmath371 as the cohomology of the punctured neighborhood of the part of exceptional curve in resolution of 6-fold cover of @xmath372 branched over @xmath373 outside of proper preimage of @xmath18 . for any cusp @xmath136 the deck transformation acting on @xmath374
has as eigenvalue the same primitive root of unity of degree six ( corresponding to the element in the spectrum of @xmath270 in the interval @xmath354 ; in the case of the action on cohomology this part contains only @xmath375 ) .
each of the other spaces @xmath376 appearing in can be viewed as a punctured neighborhood of a quasi - projective variety and as such also supports the canonical mixed hodge structure ( cf .
@xcite ) moreover , the sequence is a sequence of mixed hodge structures .
it is shown in @xcite that the map @xmath377 is surjective and that the composition of and takes @xmath378 to zero . both sequences and are equivariant with respect to the deck transformations .
the sequence yields that any eigenvalues of @xmath333 acting on @xmath379 must be an eigenvalue of @xmath333 acting on @xmath380 which is different from 1 i.e. is the eigenvalue of monodromy acting on milnor fiber of cusp and corresponding to the part of spectrum in @xmath354 .
this implies that the action of the deck transformation on @xmath358 is the multiplication by the same root of unity of degree @xmath52 as well ( exponent of the only element of the spectrum belonging to @xmath354 ) .
also note that an @xmath36-th power of this automorphism ( @xmath381 ) has zero as the only fixed point , since the existence of a fixed point for such @xmath36 would yield an eigenspace of the monodromy corresponding to an eigenvalue which is not a primitive root of degree 6 .
theorem [ roanth ] , when applied to @xmath382 with @xmath383 , and @xmath384 , shows ( since the condition on the fixed point sets follows from the explicit form of the action on @xmath385 ) that @xmath386 and that each component @xmath387 is the elliptic curve with an automorphism of order six i.e. the curve @xmath84 with @xmath266 . hence we obtain the following : [ mapstoe ] let @xmath76 be a 6-fold cyclic multiple plane with branching curve having only nodes and cusps as singularities with irregularity @xmath286 . then @xmath388 in particular for the set of morphisms taking @xmath307 to the zero of @xmath84 one has : @xmath389 next we shall reformulate this proposition in terms of the mordell - weil group of split elliptic threefold @xmath390 viewed as elliptic curve defined over @xmath391 ( and which can be viewed as a cover of @xmath82 cf .
section [ pencilsonmultiple ] ) .
recall that given an extension @xmath392 and an abelian variety @xmath313 over @xmath393 , one has an abelian variety @xmath394 over @xmath138 ( called the _ chow trace _ ) and homomorphism @xmath395 defined over @xmath138 such that for any extension @xmath396 disjoint from @xmath393 and abelian variety @xmath397 over @xmath398 and @xmath399 over @xmath400 exist @xmath401 such that @xmath402 ( cf .
@xcite ) . in particular , to @xmath87 over @xmath391 , one can associate chow trace which is the elliptic curve @xmath394 over @xmath403 such that quotient of the group of @xmath391-points of @xmath87 by the subgroup of @xmath403-points of @xmath394 is a finitely generated abelian group .
( mordell - weil theorem cf .
@xcite and ( * ? ? ? * theorem 1 ) ) .
the chow trace @xmath394 of @xmath87 is @xmath84 and the group @xmath404 is the subgroup in @xmath391 of points the corresponding to constant maps @xmath405 .
we shall denote by @xmath406 the quotient of the group of @xmath391 points by the subgroup of torsion point and the points of chow trace .
as already mentioned , @xmath406 is a finitely generated abelian group .
proposition [ mapstoe ] can be reformulated in terms of the mordell - weil group as follows : [ cor - mw ] the mordell - weil group of @xmath391-points of @xmath88 is a free @xmath407 $ ] module having rank @xmath286 ( where @xmath286 is the irregularity of @xmath76 ) .
non zero elements of this mordell weil group are represented by the classes of non - constant sections of @xmath408 .
those corresponds to the maps @xmath405 up to translation .
hence the corollary follows from propositions [ mapstoe ] .
next we shall return to the threefold @xmath82 which is a smooth birational model of ( [ equationwf ] ) .
we shall view it as an elliptic curve over field @xmath409 .
it splits over the field @xmath410 .
we want to have an explicit correspondence between the elliptic pencils on @xmath76 and @xmath15 points of @xmath82 . ]
this is used in theorem [ irregularitymw ] to obtain the relation between the mordell weil group of @xmath15-points of @xmath82 and mordell weil group of @xmath391-points of split threefold @xmath390 and is based on lemma [ equationforw ] below .
before we state it we shall introduce several notations .
compactifications @xmath82 of the threefold ( [ equationwf ] ) have several useful biregular models in weighted projective spaces and their products which we shall derive .
we can view @xmath84 as the curve given by equation @xmath411 in weighted projective plane @xmath412 .
let @xmath85 be surface in @xmath86 given by @xmath413 projections to the factors we shall denote @xmath414 and @xmath415 resp
. clearly , the map @xmath416 takes @xmath417 to the surface in @xmath86 given by @xmath418 ( no dependence on @xmath419 ) , which is isomorphic to @xmath420 .
the action of @xmath282 on @xmath85 corresponding to the standard action on @xmath84 and trivial action on @xmath41 is given by @xmath421 .
we denote by @xmath78 the biregular model of the cyclic cover of @xmath15 branched over @xmath81 and @xmath336 , which is the surface in @xmath79 given by @xmath422,[m : n])\in { \mathbb p}^2 \times { \mathbb p}^1 \mid z^{6k}m^6=n^6f(x , y , z)\}.\ ] ] the action of the deck transformation is given by @xmath423 .
the projection on the first ( resp .
the second ) factor will be denoted @xmath424 ( resp .
@xmath425 ) .
we shall consider the threefold @xmath426 with the fibered product taken relative to the maps @xmath425 and @xmath415 respectively with coordinates of respective copies of @xmath41 identified using the relation : @xmath427 the birational equivalence between @xmath85 and @xmath420 yields the birational isomorphism : @xmath428 i.e. @xmath429 is a projective model of @xmath87 .
[ equationforw ] the threefold @xmath83 in @xmath430 given by @xmath431 is birationally equivalent to @xmath432 with the diagonal action of @xmath282 .
it contains the hypersurface ( [ equationwf ] ) as an open set i.e. is a model of @xmath82 .
the equations of @xmath429 in @xmath433 are ( [ eqeq ] ) , ( [ v6eq ] ) and ( [ identification ] ) .
hence @xmath429 is biregular to complete intersection in @xmath434 given by @xmath435 projection of this complete intersection on @xmath436 has as its image the set of points @xmath437 for which the determinant of the system ( [ completeint ] ) in @xmath438 is zero i.e. is the hypersurface given by : @xmath439 clearly this projection is a cyclic @xmath282-covering of the hypersurface ( [ intermideq ] ) .
alternatively it is the quotient of ( [ completeint ] ) by the action of @xmath282 given by @xmath440 .
moreover it shows that ( [ intermideq ] ) is the quotient of @xmath429 by the _ diagonal _ ( as follows from ( [ identification ] ) ) action on @xmath282 .
finally both , the hypersurface in the statement of lemma [ equationforw ] and ( [ intermideq ] ) have @xmath82 as zariski open subset which yields the statement .
next we compare the @xmath391-points of @xmath84 and @xmath15 points of elliptic threefold @xmath82 . for @xmath76 , which is a non - singular model of @xmath78 , the @xmath403-split elliptic threefold @xmath390 , as above ,
is the elliptic curve over @xmath391 obtained by field extension @xmath441 .
@xmath391-points of @xmath87 correspond to rational maps @xmath405 which also can be viewed as sections of the projection @xmath442 by associating with a map its graph in @xmath87 and vice versa .
the corresponding mordell - weil group was calculated in corollary [ cor - mw ]
. the group @xmath282 acts ( diagonally ) on @xmath87 and hence also on @xmath406 .
we denote the invariant subgroup as @xmath443 . to a @xmath282-invariant element @xmath444 of @xmath406
corresponds @xmath282-invariant @xmath76-point @xmath445 in the sense that @xmath446 .
its graph @xmath447 is @xmath282-invariant and taking the @xmath282-quotient yields the map : @xmath448 hence we obtain a @xmath15-point of @xmath82 .
vice versa a section @xmath449 lifts to a birational map of cyclic covers i.e. the map @xmath450 ( which follows from comparison of the complements to the branching loci of both coverings ) .
this yields an equivariant ( i.e. commuting with @xmath282-action ) elliptic pencil .
[ irregularitymw ] the correspondence @xmath451 induces an isomorphism @xmath452 . in particular @xmath453 .
it is enough to check that for two equivariant maps @xmath454 in the same coset of the chow trace the map @xmath455 given by @xmath456 is constant with image @xmath457 .
indeed this is a map to a point which due to equivariance should be the @xmath282-fixed point of @xmath84 i.e. zero . to see the second part , since @xmath458 $ ] , we infer from proposition [ mapstoe ] or corollary [ cor - mw ] , that @xmath459^q\ ] ] the action of the group @xmath282 on @xmath358 is via multiplication by @xmath460 ( i.e. as is in the case of local albanese of the cusp ) . since all elements in @xmath461 commute with complex multiplication the action of @xmath282 on @xmath461 is trivial and one obtains @xmath285 as the @xmath55 rank of @xmath462 .
( of theorem [ theoremonwf ] ) .
it follows immediately from theorem [ irregularitymw ] and the well - known relation between the degree of the alexander polynomial and the irregularity of cyclic multiple planes ( cf .
@xcite ) .
@xcite ) for a degree 6 curve with 6,7,8 , and 9 cusps , the @xmath55-ranks of @xmath462 are equal to 0,2,4 , and 6 respectively .
[ boundcusps ] if @xmath268 is the degree of a homogeneous polynomial @xmath267 $ ] , for which the corresponding curve @xmath97 has only nodes and cusps as singularities then the @xmath55-rank of the mordell - weil group of @xmath82 satisfies : @xmath463 let @xmath464 be a generic line in the base of the elliptic threefold @xmath465
. then @xmath466 is injective ( cf .
@xcite ) and a bound on @xmath467 therefore yields a bound on rank of @xmath462 . on the other hand : @xmath468 the surface @xmath469 is a hypersurface in the weighted projective space @xmath470 , which is a quotient of the surface in @xmath471 given by the equation : @xmath472 with the action of the group @xmath473 ( @xmath474 are the coordinates of @xmath475 ) .
since the fixed points are outside of @xmath476 the surface @xmath469 is non singular .
in particular the elliptic surface @xmath469 gives an elliptic fibration with @xmath477 each degenerate fiber being isomorphic to a cubic curve with one cusp .
i.e. by noether formula : @xmath479 hence , using the calculation of the hodge diamond of an elliptic surface ( cf .
@xcite ) and tate - shioda formula ( cf .
* cor . 6.13 ) ) we derive : @xmath480 which gives the claim .
this immediately yields : [ inequality ] the degree of the alexander polynomial of an irreducible curve @xmath18 of degree @xmath268 , whose singularities are only nodes and cusps satisfies : @xmath481
in this section we shall present an explicit relation between the quasi - toric relations introduced in section [ quasitoricsection ] and the elements of the mordell weil group of @xmath82 .
such a relation is expected since it was shown earlier that quasi - toric relations correspond to elliptic pencils on cyclic multiple planes and orbifold pencils ( cf . section [ quasitoricsection ] ) . on the other hand in the past section
such pencils were related to the mordell - weil groups .
let us consider map of ( [ formulaequationforw ] ) onto @xmath15 induced by the projection centered at @xmath482 @xmath483\in & { \mathbb p}(2,3,1,1,1)\setminus \{[u , v,0,0,0]\}\\ \downarrow & \downarrow\\ [ x , y , z ] & { \mathbb p}^2 \end{matrix}\ ] ] note that a rational section of this projection is given by @xmath484\ ] ] where @xmath485 and @xmath486 , satisfying @xmath487 and hence @xmath488 since @xmath11 can be chosen not to be divisible by @xmath325 , by the unique factorization property of @xmath72 $ ] , @xmath489 $ ] , and hence @xmath490 is a quasi - toric relation of @xmath11 .
conversely , given any quasi - toric relation @xmath491 , for certain @xmath492 $ ] such that @xmath493 and @xmath494 , the map @xmath495 $ ] results in a section of the projection ( [ eq - proj ] ) . as a consequence ,
we obtain the following : [ prop - number - qt ] the degree of the alexander polynomial is equal to the number of equivalence classes of quasi - toric relations which correspond to independent elements ( over @xmath407 $ ] ) of the mordell - weil group of @xmath82 given by equation ( [ formulaequationforw ] ) . also , using theorem [ irregularitymw ] and projection ( [ eq - proj ] ) one can give formulas for the additive structure in @xmath462 . by corollary [ cor - mw ] ,
it is enough just to give the action of @xmath407 $ ] on the sections in @xmath462 and the addition .
consider sections @xmath496 and @xmath497 of @xmath80 and assume for simplicity that @xmath498 .
then one has the following . under the above conditions , @xmath499 also , if @xmath500 @xmath501 otherwise @xmath502 since @xmath503 , one obtains ( [ eq - action ] ) .
the formulas can be easily obtained from the well - known formulas of the group law in @xmath84 ( cf .
* chapter iii.3 ) ) .
next we shall describe a geometric property of generators or the mordell weil group viewed as elliptic pencils .
an alternative way to see that every quasi - toric relation of @xmath11 contributes to the degree of its alexander polynomial comes from the theory of orbifold surfaces and orbifold morphisms reviewed in section [ orbifoldcurves ] .
as mentioned before , the elements of the mordell - weil group @xmath462 are represented by @xmath282-equivariant surjective maps @xmath405 ( which we called elliptic pencils ) . to end this section
, we will relate those which have irreducible generic members to generators of the mordell - weil group .
we call an elliptic pencil @xmath405 _ primitive _ if its generic fiber is irreducible
. one has the following result .
[ prop - primitive ] an elliptic pencil @xmath504 is non - primitive if and only if there exists an elliptic non - injective homomorphism @xmath505 such that @xmath506 .
it is enough to check that the stein factorization of a non - primitive pencil @xmath504 factorizes through an elliptic curve @xmath506 , where @xmath507 and @xmath182 is a primitive elliptic pencil . in order to see this
, one can use the cyclic order - six action @xmath282 and obtain a pencil @xmath508 , where @xmath509 is @xmath41 with an orbifold structure 2 , 3 , 6 . using the stein factorization on @xmath510 one obtains @xmath511 , where @xmath512 . since @xmath513 is a rational surface ,
one obtains that @xmath514 with an orbifold structure given by at least three orbifold points of orders 2,3 , and 6 .
the orbifold points of order 2 and 3 correspond to double and triple fibers , whereas the order 6 point corresponds to a non - reduced ( but not multiple ) fiber of type @xmath515 where @xmath11 is the branching locus of the 6-fold cover of @xmath15 . by proposition [ prop - bound - pencil ]
, pencils of curves can not have more than two multiple members , hence @xmath516 .
the 6-fold cover @xmath84 of @xmath160 ramified with orders 2 , 3 , and 6 on the three orbifold points of @xmath160 allows for the existence of a factorization of @xmath4 , say @xmath517 induced by @xmath518 and @xmath519 . since @xmath518 is primitive , the map @xmath182 is also primitive .
note that the elliptic pencil obtained by @xmath520 should be non - primitive , since it factors through @xmath521 , given by the degree 4 map @xmath522 ( see proposition [ prop - primitive ] ) .
in fact , @xmath520 produces the following quasi - toric relation @xmath523 where @xmath524 , @xmath525 ( from ( [ eq - double ] ) ) , and @xmath526 is non - irreducible since @xmath527 which decomposes into a product of four factors of type @xmath528 , since @xmath529 is a homogeneous polynomial of degree four in @xmath530 .
note that @xmath531 and @xmath532 are the union of four members ( counted with multiplicity ) of the pencil @xmath533 .
also , note that @xmath534 is a non - primitive pencil whose generic member also consists of four members of @xmath533 . in particular , the 4:1 map @xmath535 , given by @xmath536):=[x(9x-20y)^3:(27x^2 - 36xy+8y^2)^2]\ ] ] is such that @xmath512 is the stein factorization of @xmath510 , where @xmath537)=[f_2 ^ 3(x , y , z):-f_3 ^ 2(x , y , z)]$ ] and @xmath538)=[k_2 ^ 3(x , y , z):k_3 ^ 2(x , y , z)]$ ] . summarizing the previous results , one obtains .
[ cor - gen - primitive ] let @xmath285 be the degree of the alexander polynomial of @xmath11 , then there exist @xmath286 primitive quasi - toric relations @xmath539 of @xmath11 such that @xmath540{\sigma}_1 \oplus \dots \oplus { \mathbb z}[\omega_6]{\sigma}_q$ ] , where the action of @xmath407 $ ] on @xmath541 is described in ( [ eq - action ] ) , ( [ eq - sum ] ) , and ( [ eq - double ] ) .
by proposition [ prop - number - qt ] there exist @xmath286 quasi - toric relations @xmath539 of @xmath11 such that @xmath540{\sigma}_1 \oplus \dots \oplus { \mathbb z}[\omega_6]{\sigma}_q$ ] .
the only question left to be proved is whether or not the quasi - toric relations @xmath541 can be chosen to be primitive . by proposition [ prop - primitive ] ,
if @xmath541 was not primitive , then @xmath542 for a certain @xmath543={\text{\rm hom}}(e_0,e_0)$ ] and @xmath544 primitive . since @xmath545 can not be a unit , then @xmath546 { \sigma}_1 \oplus \dots \oplus { \mathbb z}[\omega_6]{\sigma}_q$ ] , which contradicts our assumption .
[ cor - kulikov ] for any irreducible plane curve @xmath57 whose only singularities are nodes and cusps the following statements are equivalent : 1 . [ cor - kulikov-1 ] @xmath18 admits a quasi - toric relation of elliptic type @xmath50 , 2 .
[ cor - kulikov-2 ] @xmath18 admits an infinite number of quasi - toric relations of elliptic type @xmath50 , 3 .
[ cor - kulikov-3 ] @xmath2 is not trivial ( @xmath547 ) . moreover , the set of quasi - toric relations of @xmath18 @xmath60 ^ 3\mid f^2+g^3+h^6f=0\}$ ] has a group structure and it is isomorphic to @xmath61 , where @xmath62 .
also , @xmath18 admits an infinite number of primitive quasi - toric relations , unless @xmath63 , in which case @xmath18 only has one primitive quasi - toric relation . for the first part , @xmath548 is an immediate consequence of the group structure of the set of quasi - toric relations , namely , once a quasi - toric relation @xmath549 is given , the set @xmath407 { \sigma}$ ] provides an infinite number of such relations .
also @xmath548 is a consequence of proposition [ prop - number - qt ] .
the _ moreover _ part is a consequence of proposition [ prop - primitive ] .
theorem [ cor - kulikov ] is related to @xcite .
also this result has been recently noticed by kawashima - yoshizaki in ( * ? ? ?
* proposition 3 ) . in this paper , a quasi - toric relation @xmath26 is considered and then the series @xmath550 is used ( it is an easy exercise ) .
the resulting quasi - toric relations @xmath551 are not primitive and the generic member of the pencil associated with @xmath551 is the product of @xmath552 members of the original pencil associated with @xmath26 , which incidentally is the degree of @xmath521 , @xmath553 .
in this section we extend the results of previous sections to singularities more general than nodes and cusps .
let us consider the general situation described in section [ defalpol ] , that is , let us fix @xmath267 $ ] , a homogeneous polynomial of degree @xmath1 which is not a power , whose set of zeroes in @xmath15 is the curve @xmath18 . by this
is equivalent to fixing a projective plane curve @xmath554 and a list of multiplicities @xmath555 ( such that @xmath556 ) or a surjection @xmath557 ( where @xmath10 is a line at infinity transversal to @xmath18 ) such that @xmath558 , @xmath35 a meridian around @xmath96 .
recall that @xmath559 corresponds to the multiplicity of each irreducible component of @xmath11 .
[ deltapartial ] let @xmath561 be a germ of a singular point of @xmath18 .
we call @xmath561 a _ @xmath70-essential singularity _
_ @xmath70-co - prime singularity _ ) w.r.t .
@xmath23 if and only if the roots of @xmath139 are all @xmath70-roots of unity ( resp .
no root of @xmath139 is a @xmath70-root of unity except for @xmath562 ) .
we say that a curve @xmath563 _ has only @xmath70-essential singularities _ if there exists an epimorphism @xmath564 such that @xmath561 is a @xmath70-essential singularity w.r.t .
@xmath137 for all @xmath565 ( see to recall the construction of @xmath137 ) .
a curve @xmath18 is called _
w.r.t . a homomorphism @xmath23 if any singularity @xmath136 of @xmath18 , is either @xmath70-essential or @xmath70-co - prime .
we also call a curve _ @xmath70-total _ w.r.t . to a homomorphism @xmath23 if it is @xmath70-partial and all the roots of the global alexander polynomial @xmath8 w.r.t .
@xmath23 are roots of unity of degree @xmath70 ( not necessarily primitive ) .
[ rem-6ess ] the curves whose only singularities are ( reduced ) nodes and cusps necessarily have @xmath52-essential singularities . as another example one can consider @xmath566 , where @xmath567 is the tricuspidal quartic and @xmath10 is its bitangent and the epimorphism @xmath23 mapping the meridian of @xmath10 to @xmath568 and the meridian of @xmath567 to @xmath103 ( i.e. the homomorphism of type @xmath569 .
the local alexander polynomial of @xmath18 at a tacnode w.r.t .
@xmath23 is given by @xmath570 , whereas at a cusp it is simply @xmath571 .
therefore all singularities of @xmath18 w.r.t .
homomorphism of type @xmath569 .
are 6-essential .
in particular , by proposition [ prop - divisibility ] , @xmath18 is a @xmath52-total curve .
also note that tacnodes with respect to the homomorphism of type @xmath572 are 4-essential singularities as well as nodes with respect to the homomorphism of type @xmath573 . as for 3-essential singularities , one has @xmath574-singularities w.r.t .
@xmath23 of type @xmath572 and nodes w.r.t .
@xmath23 of type @xmath575 .. -singularities ( resp .
@xmath576-singularities ) are locally equivalent to @xmath577 ( resp .
@xmath578 ) . ]
[ prop - essential - local - global ] a curve with only @xmath70-essential singularities is @xmath579-total for some @xmath580 .
the result is an immediate consequence of proposition [ prop - divisibility ] .
note that the converse of proposition [ prop - essential - local - global ] is not true ( see section [ sec - examples ] ) .
also , proposition [ prop - essential - local - global ] can be sharpened using @xcite , so that not all singularities are required to be @xmath70-essential .
[ rem - total - partial ] note that there are examples of @xmath70-partial curves that are not @xmath70-total , for instance a union of two cuspidal cubics intersecting each other at three smooth points with multiplicity of intersection 3 , that is , a sextic @xmath18 with singularities @xmath581 and three points of type @xmath574 .
similar notations will be used in the rest of the paper .
] has alexander polynomial @xmath582 ( cf .
therefore @xmath18 is 3-partial but not 3-total .
this curve will be studied in more detail in example [ ex-2a2 - 3a5 ] .
[ prop - delta - alexander ] let @xmath561 be a germ of curve singularity and let @xmath583 be a birational morphism such that the support of the total pull - back @xmath584 is a normal crossing divisor on @xmath585 .
let @xmath586 be the multiplicity of @xmath587 where @xmath588 is the local equation of the @xmath13-th branch of @xmath18 at @xmath136 along the exceptional curve @xmath589 of @xmath590 .
let @xmath591 be the value of @xmath23 on the meridian corresponding to @xmath588 .
then the local alexander polynomial of @xmath18 at @xmath136 w.r.t .
@xmath23 is given by : @xmath592 where @xmath593 and @xmath594 is the topological euler characteristic .
indeed , the infinite cyclic cover of the complement to the zero set of the germ of @xmath18 at @xmath136 corresponding to the homomorphism @xmath23 can be identified with the milnor fiber of non - reduced singularity : @xmath595 now the claim follows from acampo s formula ( cf .
@xcite ) .
proposition [ prop - delta - alexander ] allows one to compile a complete list of @xmath70-essential singularities with @xmath596 . in table
[ tab : d - ess ] a list of the possible 6-essential singularities is given .
the first column shows the number of local branches of the singularity .
the second column contains the reduced type of the singularity .
the third column shows a list of all possible multiplicities for the branches .
it is a consequence of the divisibility conditions on the multiplicities of each branch imposed by proposition [ prop - delta - alexander ] which follows from the requirement to have a 6-essential singularity : see the proof below .
the forth column gives the list of multiplicities @xmath597 of the irreducible factors of the alexander polynomial in @xmath598 $ ] @xmath599 finally the fifth column shows that the set of logarithms belonging to the interval @xmath354 of the eigenvalues of ( the semi - simple part of ) the monodromy acting on @xmath600 of the cohomology of the milnor fiber .
[ tab : d - ess ] [ proofoftable1 ] table [ tab : d - ess ] contains a complete list of @xmath52-essential singularities .
we will use notation from proposition [ prop - delta - alexander ] .
the contributing exceptional divisors ( i.e. those with @xmath601 ) appearing in the end of resolution of singularity @xmath561 have maximal multiplicity @xmath602 among components preceding it .
hence the primitive root corresponding to the component with maximal such multiplicity wo nt cancel in .
moreover , the multiplicities of the contributing exceptional divisors have to divide @xmath52 .
in the case @xmath603 this forces the singularity to be a cusp . also , for @xmath604 this forces the irreducible branches to be smooth .
the rest of the list can be worked out just keeping in mind that since @xmath605 , in particular removing any branch from a valid singularity with @xmath606 branches , one should obtain a valid singularity with @xmath33 branches . finally , if @xmath607 is a @xmath70-essential singularity type of @xmath33 branches whose only valid homomorphism is @xmath608 , then there is no @xmath70-essential singularity @xmath609 of @xmath610 branches such that @xmath611 results from removing a branch from @xmath609 . for example , for reduced singularity @xmath612 the proposition [ prop - delta - alexander ] yields that a collection of multiplicities of branches @xmath613 yields a 6-essential non - reduced singularity if and only if @xmath614
this is satisfied only by the pair @xmath615 .
similarly , for singularity @xmath616 the divisibility condition is @xmath617 which is satisfied by three triplets indicated in the table , etc .
note that the possible morphisms @xmath23 shown in the third column are all up to action of the permutation group except for the case @xmath618 , where the last branch is not interchangeable with the others .
so , whereas @xmath619 in the @xmath616 case represents six possible morphisms , @xmath620 in the @xmath618 case only represents one possible morphism . in order to check the last column , it is enough to compute the constants of quasi - adjunction ( cf .
@xcite ) different from @xmath621 , since @xmath138 is a constant of quasi - adjunction different from @xmath621 if and only if @xmath622 is an element of the spectrum of the singularity corresponding to the part of weight 1 ( cf .
@xcite ) .
constants of quasi - adjunction can be found in terms of a resolution of the singularity as follows .
if @xmath623 is an equation of the germ of plane curve singularity at the origin , @xmath624 is its resolution ( i.e. the proper preimage of the pull - back is a normal - crossing divisor ) , @xmath625 are the exceptional components of the resolution @xmath590 , @xmath626 ( resp .
@xmath627 , resp .
@xmath628 ) is the multiplicity along @xmath625 of @xmath629 ( resp .
@xmath630 , resp .
@xmath631 , where @xmath632 is a germ of a function at the origin ) then for a fixed @xmath633 , @xmath634 is the minimal solution to the system of inequalities ( the minimum is taken over @xmath138 ) : @xmath635 ( cf .
* ( 2.3.8 ) ) ) . for example , for @xmath636-singularity ( i.e. the cusp which can be resolved by three blow ups ) one has for the last exceptional component : @xmath637 and hence @xmath638 .
therefore for the spectrum ( i.e. the last column in table [ tab : d - ess ] ) one obtains the well - known constant @xmath639 . similarly , for @xmath616-singularity @xmath640
( which is resolved by the single blow up ) with the multiplicities of the components @xmath641 one obtains only one inequality @xmath642 hence for various choices of @xmath633 one obtains @xmath643 .
thus @xmath644 are the only elements of the spectrum in @xmath354 corresponding to the part of weight 1 . in the following tables
the notations are the same as in the table [ tab : d - ess ] and in table [ tab : d-4-ess ] the factorization of the alexander polynomial is the following : @xmath645 [ tab : d-4-ess ] [ tab : d-3-ess ] table [ tab : d-4-ess ] ( resp .
[ tab : d-3-ess ] ) classifies the 4-essential ( resp .
3-essential ) singularities .
the proof is same as the proof of proposition [ proofoftable1 ] .
[ oneroot ] for simple @xmath70-essential singularities ( i.e. of type @xmath646 or @xmath647 ) there is at most one eigenvalue of order @xmath648 ( @xmath649 ) for the action of the monodromy on the hodge component @xmath600 of the milnor fiber
. inspection on tables [ tab : d - ess]-[tab : d-3-ess ] shows that except for @xmath562 , the multiplicity of each root of the characteristic polynomial of the monodromy is equal to one .
since the conjugation of an eigenvalue of the monodromy acting on @xmath600 is an eigenvalue of its action on @xmath650 it follows that @xmath651 is not an eigenvalue of the monodromy action on @xmath600 .
again , by direct inspection of tables [ tab : d - ess]-[tab : d-3-ess ] the result follows
. the following auxiliary results will be useful in the rest of the arguments .
[ deltaessentialalexander ] let @xmath18 be a @xmath70-partial curve with @xmath33 irreducible components .
then the polynomial @xmath652 divides the alexander polynomial @xmath59 of @xmath18 w.r.t .
@xmath23 , where @xmath205 above denotes the multiplicity of the primitive root of unity of degree @xmath36 in the alexander polynomial of @xmath18 .
moreover , let @xmath653 be the cyclic cover of @xmath15 of degree @xmath70 branched over a curve @xmath18 according to the multiplicities @xmath654 .
then , the characteristic polynomial of the deck transformation acting on @xmath655 equals @xmath656 if @xmath18 is @xmath70-total , then @xmath657 the first ( resp .
last ) assertion about the alexander polynomial follows directly from the definitions of @xmath70-partial ( resp .
@xmath70-total ) curves and the well - known fact that the multiplicity of the root @xmath562 for a curve with @xmath33 irreducible components in the alexander polynomial is equal to @xmath658 .
the _ moreover _ part follows from the relation between the homology of branched and unbranched covers ( cf .
@xcite ) .
[ localdecomposition ] for a plane curve singularity @xmath136 , denote by @xmath659 the local jacobian ( cf .
section [ subsec - mw ] ) let @xmath18 be a @xmath70-partial curve and let @xmath653 be cyclic cover of degree @xmath70 of @xmath15 branched over @xmath18 .
then @xmath660 is isogenous to a quotient of the product of local jacobians of singularities of @xmath18 .
it follows from the proof of lemma [ albaneseeigenvalues ] .
finally , we will give a description of the multiplicities @xmath205 in terms of the albanese variety of the ramified coverings of @xmath15 .
denote by @xmath84 ( resp .
@xmath661 ) the elliptic curve with @xmath13-invariant zero ( resp .
@xmath662 ) . then one has the following . [ thm - main - alb ] let @xmath18 be a @xmath70-partial curve ( @xmath663 or @xmath52 ) with @xmath70-essential singularities of type @xmath646 and @xmath647 .
then the albanese variety @xmath660 corresponding to the curve @xmath18 can be decomposed as follows : @xmath664 where @xmath205 is the multiplicity of the @xmath36-th primitive root of unity in @xmath8 , @xmath313 is an abelian variety of dimension @xmath665 , `` @xmath666 '' means isomorphic , and `` @xmath667 '' means _
the deck transformation of @xmath653 induces the action of the cyclic group @xmath668 on the albanese variety of @xmath653 .
it follows from roan s decomposition theorem ( cf .
* theorem 2.1 ) ) that @xmath660 is isogenous to a product @xmath669 where @xmath670 is the number of orders of eigenvalues of @xmath290 .
we shall show that abelian varieties @xmath303 can be decomposed further to yields isogenies [ isogeniesdeltaalbanese ] . by lemma [ deltaessentialalexander ]
, one has that @xmath103 is not an eigenvalue of @xmath290 , and hence each component @xmath303 supports an automorphism whose order is a non - trivial divisor of @xmath70 .
we denote by @xmath313 the component supporting an automorphism of order 2 . in the cases of @xmath646 or @xmath647 singularities , corollary [ oneroot ] yields that the action of @xmath290 on each component @xmath303 has at most one eigenvalue .
indeed this is the eigenvalue of the monodromy on the component @xmath600 of the milnor fiber of the singularities whose local alexander polynomial contains such eigenvalue as a root .
each eigenvalue of the monodromy on @xmath600 for a @xmath70-essential singularity does not appear on @xmath650 even if for different @xmath70-essential singularities . in particular even different @xmath70-essential singularities contribute at most to the same root of the global alexander polynomial as follows from the table in corollary [ oneroot ] .
this allows one to apply the second decomposition theorem ( cf .
* theorem 3.2 ) ) as was done above in the case of cuspidal curves .
the result follows . 1 .
if @xmath18 has singularities of type @xmath671 then the above argument shows that @xmath358 has , up to isogeny , the factors which are the jacobians of the projection model of @xmath672 and the elliptic curves of @xmath661 and @xmath84 .
2 . whenever @xmath673 , the local jacobians ( cf . section [ subsec - mw ] ) depend on the moduli describing the local algebraic analytic type of singularities . for example , in the case of singularities @xmath674 the local jacobian is the jacobian of the 4-fold cyclic cover branched over 4 points . the quotient by an involution of such cover may yield an arbitrary elliptic curve in the legendre family of elliptic curves . hence , the albanese variety of the cyclic cover @xmath653 is isogenous to the quotient of abelian varieties whose moduli a priori depends on the analytic type of the singularities . it would be interesting to see if one can find examples showing that such variations can take place .
now we shall relate the mordell - weil group of threefolds corresponding to @xmath18 to @xmath59 .
let @xmath87 denote an elliptic threefold birational to the affine hypersurface given by the equation : @xmath675 one has the following : [ thm - main - delta ] let @xmath57 be a @xmath70-partial curve as in theorem [ thm - main - alb ] , then @xmath676 in addition , if @xmath18 is a @xmath70-total curve , then @xmath677 as in the proof of theorem [ irregularitymw ] one has the identification of @xmath462 and @xmath197-invariant elements of the mordell - weil group of @xmath82 over the extension @xmath678 of @xmath39 with the galois group @xmath197 , where @xmath679 for @xmath680 and @xmath681 for @xmath682 .
the threefold @xmath683 splits over the field @xmath684 since it is isomorphic to a direct product threefold using @xmath685 , @xmath686 .
the inequalities follow from : @xmath687 , where @xmath688 for @xmath689 , @xmath690 for @xmath691 and @xmath692 for @xmath682 .
[ thm - bound - delta ] let @xmath57 be a @xmath70-partial curve for @xmath663 or @xmath52 , then : 1 .
[ thm - bound - delta-1 ] there is a one - to - one correspondence between quasi - toric relations corresponding to @xmath11 and the @xmath39-points of the threefold @xmath82 . if , in addition , the singularities of @xmath18 are as in theorem [ thm - main - alb ] then : 1 .
the multiplicities of the factors of the global alexander polynomial of @xmath70-partial curves satisfy the following inequalities : @xmath693 the argument is the same as in the proof of theorem [ boundcusps ] since the rank of the mordell - weil group of the threefolds for each @xmath70 is bounded by the rank of mordell - weil group of the elliptic surface with @xmath1 degenerate fibers each isomorphic to a cubic curve with a single cusp as follows from equations ( [ threefold3 ] ) . for each @xmath70
we obtain a bound on the rank of the submodule of the alexander module corresponding to the action of the deck transformation on the subspaces generated by the eigenvalues which are roots of unity of degrees @xmath694 and @xmath52 . as an immediate consequence of theorem [ thm - bound - delta ]
one has the following generalization of theorem [ cor - kulikov ] : [ cor - kulikov - gen ] let @xmath57 be a curve , then the following statements are equivalent : 1 .
[ cor - kulikov - gen-1 ] @xmath18 admits a quasi - toric relation of elliptic type @xmath66 ( resp .
@xmath67 , or @xmath50 ) , 2 .
[ cor - kulikov - gen-2 ] @xmath18 admits an infinite number of quasi - toric relations of elliptic type @xmath66 ( resp .
@xmath67 , or @xmath50 ) . also , and
above imply 1 .
[ cor - kulikov - gen-3 ] @xmath18 is a @xmath70-partial curve ( @xmath695 ) for @xmath689 ( resp .
@xmath696 , or @xmath52 ) .
moreover , if the singularities of @xmath18 are as in theorem [ thm - main - alb ] , then , , and are equivalent .
first of all note that this result generalizes theorem [ cor - kulikov ] since curves with only nodes and cusps as singularities and non - trivial alexander polynomial automatically satisfy that @xmath697 for a primitive 6th - root of unity , and hence they are @xmath70-partial curves ( in fact , they are @xmath70-total ) .
statement @xmath548 follows from the group structure of quasi - toric relations as exhibited in theorem [ thm - bound - delta ] .
also @xmath698 is immediate since a quasi - toric relation of @xmath11 induces an equivariant map @xmath699 , which induces a map @xmath700 , where @xmath701 , @xmath67 , resp .
@xmath50 according to @xmath702 , resp . @xmath52 .
this implies that @xmath703 , where @xmath23 is defined as in . finally , @xmath698 under the conditions of theorem [ thm - main - alb ] is an immediate consequence of theorem [ thm - main - delta ] , since implies @xmath704 .
[ rem-6l ] note that @xmath698 in theorem [ cor - kulikov - gen ] is not true for general singularities as the following example shows .
consider @xmath18 the curve given by the product of 6 concurrent lines .
generically , @xmath18 does not satisfy a quasi - toric relation of type @xmath50 or @xmath66 ( cf .
[ sixlines ] ) , but its alexander polynomial is not trivial , namely , @xmath705 .
the theorem [ thm - main - delta ] has implications for the structure of the characteristic variety @xmath706 of a plane curve @xmath18 .
characteristic varieties of curves extend the notion of alexander polynomials of curves to non - irreducible curves ( for a definition see @xcite ) . in (
* theorem 1.6 ) ( resp .
* theorem 1 ) ) , structure theorems for the irreducible components of @xmath706 are given in terms of the existence of maps from @xmath707 onto riemann surfaces ( resp .
orbifold surfaces ) .
the following result sharpens ( * ? ? ?
* theorem 1 ) for the special case of torsion points of order @xmath560 on @xmath706 .
let @xmath18 be a curve whose singularities are as in theorem [ thm - main - alb ] , @xmath708 , and consider @xmath709 the first characteristic variety of @xmath289 .
for any @xmath710 torsion point @xmath711 of order @xmath70 of @xmath709 there exists an admissible orbifold map @xmath712 such that @xmath713 .
let @xmath714 be a torsion point of order @xmath70 in @xmath709 .
note that the homomorphism induced by @xmath715 is such that the cyclotomic polynomial @xmath716 of the @xmath70-roots of unity divides @xmath717 .
hence the hypotheses of theorem [ cor - kulikov - gen ] are satisfied and therefore there exists an orbifold riemann surface @xmath159 such that @xmath716 divides @xmath718 and a dominant orbifold morphism @xmath719 .
after performing a stein factorization we may assume that the induced homomorphism @xmath719 is surjective .
finally note that any such a surjection induces an inclusion of characteristic varieties , that is , @xmath720 .
moreover , since @xmath716 divides @xmath718 , then @xmath713 .
theorem [ cor - kulikov - gen ] can not be generalized to @xmath70-essential curves outside of the range @xmath560 .
for instance , note that in @xcite an example of an irreducible affine quintic @xmath721 with @xmath722 as affine singularities is shown .
the curve @xmath723 is 10-total since @xmath724 .
however , there is no quasi - toric decomposition of @xmath725 of type @xmath726 .
this seems to contradict ( * ? ? ?
* theorem 1 ) .
[ rem - okas - conj ] since no irreducible sextic has an alexander polynomial of the form @xmath727 ( cf .
@xcite ) , theorem [ cor - kulikov - gen ] is another way to show that any irreducible sextic ( with simple singularities ) has a non - trivial alexander polynomial if and only if it is of torus type ( note that an irreducible sextic admits a quasi - toric decomposition if and only if it is of torus type as shown in example [ ex-236 ] ) .
this is a weaker version of what is now known as _
oka s conjecture _
@xcite ) .
[ sixlines ] consider the case when @xmath18 is union of @xmath52 concurrent lines .
to these @xmath52 lines @xmath728 correspond @xmath52 points @xmath729 in @xmath41 parametrizing lines in the pencil containing @xmath730 .
the 3-fold cover @xmath731 has as compactification surface in the weighted projective space @xmath732 , which after a weighted blow - up has a map onto the 3-fold cyclic cover @xmath733 of @xmath41 branched over the points @xmath729 and having rational curves are fibers .
hence @xmath734 is isomorphic to the jacobian of @xmath733 .
if @xmath729 form a collection of roots of a polynomial over @xmath107 having as the galois group over the later the symmetric ( or alternating ) group then the results of @xcite show that the above jacobian is simple and hence does not have maps onto the elliptic curve @xmath84 . on the other hand
the jacobian of the 3-fold cover of @xmath41 branched over the roots of the polynomial @xmath735 has @xmath84 as a factor , since @xmath84 is a quotient of the former by the involution . note that the characteristic polynomial of the deck transformation on @xmath736 is @xmath737 .
note that oka s conjecture can not be generalized to general reducible sextics as shown in remark [ rem-6l ] .
however , in light of theorem [ cor - kulikov - gen ] it seems that one can ask the following version of oka s conjecture to non - reduced curves , namely : [ conj - oka - gen ] is it true that a ( possibly non - reduced ) sextic @xmath57 ( with @xmath738 ) whose singularities are as in theorem [ thm - main - alb ] has a non - trivial alexander polynomial if and only if it admits a quasi - toric relation ? according to theorem [ cor - kulikov - gen ] , one has the following rewriting of question [ conj - oka - gen ] .
the answer to question [ conj - oka - gen ] is affirmative if and only if @xmath739 for any non - reduced curve @xmath18 of total degree @xmath52 whose singularities are as in theorem [ thm - main - delta ] .
the purpose of this section is to exhibit the different examples of elliptic quasi - toric relations . in example
[ ex-6 - 9 ] ( resp .
[ ex-4 - 3 ] ) we present quasi - toric relations of type @xmath50 and describe generators for the mordell - weil group of elliptic sections , which is of rank 3 ( resp . 2 ) .
in example [ ex-12 - 39 ] we present a cuspidal curve whose alexander polynomial has the largest degree known to our knowledge . finally , examples [ ex-4 - 4 - 2 ] ( resp .
[ ex-3 - 3 - 3 ] ) show examples of quasi - toric relations of type @xmath67 ( resp .
@xmath66 ) and example [ ex-2a2 - 3a5 ] shows an example of a curve with both @xmath50 and @xmath66 quasi - toric relations .
[ ex-6 - 9 ] consider the sextic curve @xmath740 with 9 cusps .
one easy way to obtain equations is as the preimage of the conic @xmath741 by the kummer abelian cover @xmath742\mapsto [ x^3:y^3:z^3]$ ] .
it is well known that the alexander polynomial of @xmath740 is @xmath743 .
note that @xmath744 belongs to the following pencils : @xmath745 therefore @xmath746 has the following 3 quasi - toric relations : @xmath747 in addition , note that @xmath748 leads to another quasi - toric relation .
if we consider @xmath749 as elements of the elliptic curve @xmath750 over @xmath39 , then @xmath407 { \sigma}_1 \oplus { \mathbb z}[\omega_6 ] { \sigma}_2 \oplus { \mathbb z}[\omega_6 ] { \sigma}_3 $ ] is the group of all quasi - toric relations of @xmath740 .
for instance , @xmath751}.\ ] ] one can obtain the above relation using , , and .
this example was first considered in this context by tokunaga ( * ? ? ?
* theorem 0.2 ) .
note that the author exhibits 12 primitive quasi - toric relations of @xmath740 such that @xmath244 . whether or not those decompositions are
the only ones satisfying @xmath244 remains open .
[ ex-4 - 3 ] consider the tricuspidal quartic @xmath752 .
since @xmath753 is the dual of a nodal cubic , it should contain a bitangent , which is the dual of the node . in our case
one can check that @xmath754 is the bitangent at @xmath755 $ ] and @xmath756 $ ] , where @xmath757 is a root of @xmath758 .
note that @xmath759 is a non - reduced sextic whose singularities are @xmath52-essential ( see remark [ rem-6ess ] ) .
its twisted alexander polynomial w.r.t .
the multiplicities @xmath573 is given by @xmath760 .
in addition we show some interesting primitive quasi - toric relations coming from the combinations of the generators @xmath766}:=\omega_6^i { \sigma}_1 + \omega_6^j { \sigma}_2 $ ] .
such primitive quasi - toric relations have the following form : @xmath767 } & \equiv & c_{4,3}l_0 ^ 2x^6 & = & 4 c_{4,[1,1]}^3 + c_{6,[1,1]}^2\\ { \sigma}_{[1,\omega_6 ] } & \equiv & c_{4,3}l_0 ^ 2(x - z)^6 & = & 4 c_{4,[1,\omega_6]}^3 + c_{6,[1,\omega_6]}^2\\ { \sigma}_{[1,\omega_6 ^ 2 ] } & \equiv & c_{4,3}l_0 ^ 2z^6 & = & 4 c_{4,[1,\omega_6 ^
2]}^3 + c_{6,[1,\omega_6 ^ 2]}^2\\ { \sigma}_{[1,\omega_6 ^ 3 ] } & \equiv & c_{4,3}l_0 ^
2(z - y)^6 & = & 4 c_{4,[1,\omega_6 ^ 3]}^3 + c_{6,[1,\omega_6 ^ 3]}^2\\ { \sigma}_{[1,\omega_6 ^ 4 ] } & \equiv & c_{4,3}l_0 ^ 2y^6 & = & 4 c_{4,[1,\omega_6 ^
4]}^3 + c_{6,[1,\omega_6 ^ 4]}^2\\ { \sigma}_{[1,\omega_6 ^ 5 ] } & \equiv & c_{4,3}l_0 ^ 2(x - y)^6 & = & 4 c_{4,[1,\omega_6 ^ 5]}^3 + c_{6,[1,\omega_6 ^ 5]}^2 , \endarray\ ] ] where @xmath768}$ ] denotes a homogeneous polynomial of degree @xmath138 corresponding to the quasi - toric relation @xmath766}$ ] . for simplicity we only show the first pair of polynomials , @xmath769}$ ] and @xmath770}$ ] , respectively .
@xmath771 the equations above can easily be obtained using , , and .
[ ex-12 - 39 ] consider the tricuspidal quartic @xmath753 as above , the bitangent @xmath772 , and two tangent lines , say @xmath773 ( at @xmath774 $ ] ) , and @xmath775 ( at @xmath776 $ ] ) .
the kummer cover @xmath742\mapsto [ \ell_0 ^ 3:\ell_1 ^ 3:\ell_2 ^ 3]$ ] produces a curve @xmath777 of degree 12 with 39 cusps .
we compute the superabundance of @xmath777 as follows .
let @xmath778 be the ideal sheaf supported on the 39 cusps and such that @xmath779 the maximal ideal at any @xmath780 cusp of @xmath777 .
the superabundance of @xmath777 is @xmath781 and it coincides with the multiplicity of the 6-th root of unity as a root of the alexander polynomial @xmath782 of @xmath777 . note that @xmath783 . since @xmath784 , @xmath785 , and
@xmath786 one has that @xmath787 .
note that @xmath786 since the only curve of degree 7 passing through all cusps is @xmath788 , where @xmath789 is the preimage by the kummer map of the conic passing through the three cusps , @xmath790 , and @xmath160 .
therefore , @xmath791 .
this is , to our knowledge , the first example of a cuspidal curve whose alexander polynomial has a non - trivial root with multiplicity greater than 3 .
as was discussed already in example [ ex-6 - 9 ] as a cuspidal curve for which the alexander polynomial has factors of multiplicity three one can take the dual curve of a smooth cubic ( its fundamental group was calculated by zariski @xcite ) .
for other examples cf .
@xcite .
[ ex-4 - 4 - 2 ] consider the moduli space of sextics with three singular points @xmath136 , @xmath792 , @xmath790 of types @xmath793 , @xmath612 , and @xmath794 respectively .
such a moduli space has been studied in @xcite and it consists of two connected components @xmath795 , @xmath796 . both have as representatives reducible sextics which are the product of a quartic @xmath797 and a smooth conic @xmath744 intersecting at the point @xmath136 ( of type @xmath793 ) and hence @xmath798 .
there is a geometrical difference between sextics in @xmath795 and @xmath796 . for one kind of sextics ,
say @xmath799 , the tangent line at @xmath136 also contains @xmath792 , whereas for the other sextics , say @xmath800 , it does not .
the alexander polynomial of both kinds is trivial ; however , if we consider the homomorphism @xmath801 , where @xmath559 is the image of a meridian around @xmath96 , then @xmath802 note that @xmath803 is a @xmath696-total curve , but not all of its singularities are @xmath696-essential , since one can check that @xmath793 has a local alexander polynomial @xmath804 .
one can also use degtyarev s divisibility criterion @xcite to prove that the factors coming from @xmath805 do not contribute . therefore , theorem [ thm - main - delta ] can be applied and hence @xmath803 has a quasi - toric relation of elliptic type @xmath67 , whereas @xmath806 does not .
in particular , these are the equations for the irreducible components of @xmath807 : @xmath808 which fit in the following quasi - toric relation : @xmath809 where @xmath810 , @xmath811 , and @xmath812 . [
ex-3 - 3 - 3 ] as an example of a quasi - toric relation of elliptic type @xmath66 we can present the classical example @xmath813 .
the classical alexander polynomial of @xmath97 is @xmath814 and it is readily seen that it is a @xmath815-total curve , since its singularities ( besides the nodes ) are ordinary triple points , which are @xmath815-essential singularities
. hence one can apply theorem [ cor - kulikov - gen ] and derive that @xmath11 fits in a quasi - toric relation of elliptic type @xmath66 .
for instance : @xmath816 however , according to theorem [ thm - main - alb ] there should exist another relation independent from , namely @xmath817 where @xmath818 @xmath757 is a third - root of unity , and @xmath819 [ ex-2a2 - 3a5 ] in remark [ rem - total - partial ] we have presented a @xmath52-total sextic curve @xmath18 which is also @xmath815-partial .
in particular , according to theorem [ cor - kulikov - gen ] , @xmath18 admits quasi - toric relations both of type @xmath50 and @xmath66 .
we will show this explicitly .
note that @xmath18 is the union of two cuspidal cubics ( cf .
* corollary 1.2 ) ) @xmath820 and @xmath821 given by the following equations : @xmath822 one can check that the curve @xmath18 satisfies the following relations : @xmath823 where @xmath824 , @xmath825 , and @xmath244 .
birkenhake , v. gonzlez - aguilera , h. lange , _ automorphisms of 3-dimensional abelian varieties_. complex geometry of groups ( olmu , 1998),2547 , contemp .
, * 240 * , amer . math .
soc . , providence , ri , 1999 .
r. friedman and j. w. morgan , _ smooth four - manifolds and complex surfaces _ , ergebnisse der mathematik und ihrer grenzgebiete ( 3 ) [ results in mathematics and related areas ( 3 ) ] , vol .
27 , springer - verlag , berlin , 1994 . , _ first order deformations for rank one local systems with a non - vanishing cohomology _ , topology appl . * 118 * ( 2002 ) , no . 1 - 2 , 159168 , arrangements in boston : a conference on hyperplane arrangements ( 1999 ) .
j. milnor , _ on the @xmath815-dimensional brieskorn manifolds @xmath827 _ , knots , groups , and 3-manifolds ( papers dedicated to the memory of r. h. fox ) , princeton univ . press ,
princeton , n. j. , 1975 , pp .
175225 . ann . of math .
studies , no .
84 .
j.steenbrink , vanishing cohomology steenbrink , j. h. m. mixed hodge structure on the vanishing cohomology .
real and complex singularities ( proc .
ninth nordic summer school / navf sympos .
oslo , 1976 ) , pp .
525 - 563 .
o. zariski , _ algebraic surfaces_. second supplemented edition . with appendices by s. s. abhyankar , j. lipman , and d. mumford .
ergebnisse der mathematik und ihrer grenzgebiete , band * 61*. springer - verlag , new york - heidelberg , 1971 . | we show that the degree of the alexander polynomial of an irreducible plane algebraic curve with nodes and cusps as the only singularities does not exceed @xmath0 where @xmath1 is the degree of the curve . we also show that the alexander polynomial @xmath2 of an irreducible curve @xmath3 whose singularities are nodes and cusps is non - trivial if and only if there exist homogeneous polynomials @xmath4 , @xmath5 , and @xmath6 such that @xmath7 .
this is obtained as a consequence of the correspondence , described here , between alexander polynomials and ranks of mordell - weil groups of certain threefolds over function fields .
all results also are extended to the case of reducible curves and alexander polynomials @xmath8 corresponding to surjections @xmath9 , where @xmath10 is a line at infinity .
in addition , we provide a detailed description of the collection of relations of @xmath11 as above in terms of the multiplicities of the roots of @xmath8 .
this generalization is made in the context of a larger class of singularities i.e. those which lead to rational orbifolds of elliptic type . |
in this paper , a _ convex lattice polygonal line _ @xmath1 is a piecewise linear path on the plane , starting at the origin @xmath3 , with vertices on the integer lattice @xmath20 , and such that the inclination of its consecutive edges strictly increases staying between @xmath21 and @xmath22 .
let @xmath23 be the set of all convex lattice polygonal lines with finitely many edges , and denote by @xmath24 the subset of polygonal lines @xmath25 whose right endpoint @xmath26 is fixed at @xmath27 .
we are concerned with the problem of _ limit shape _ of `` typical '' @xmath6 , as @xmath5 , with respect to some probability measure @xmath28 on @xmath0 . here
the `` limit shape '' is understood as a planar curve @xmath29 such that , with overwhelming @xmath28-probability for large enough @xmath30 , properly scaled polygonal lines @xmath31 lie within an arbitrarily small neighborhood of @xmath29 .
more precisely , for any @xmath32 it should hold that @xmath33 where @xmath34 is some metric on the path space for instance , induced by the hausdorff distance between compact sets ( in @xmath35 ) , @xmath36 where @xmath37 is the euclidean vector norm . of course
, the limit shape and its very existence may depend on the probability law @xmath28 .
with respect to the _ uniform distribution _ on @xmath0 , the problem was solved independently by vershik @xcite , brny @xcite and sinai @xcite , who showed that , under the scaling @xmath38 , the limit shape @xmath29 is given by a parabola arc defined by the cartesian equation @xmath39 more precisely ( cf .
( [ eq : lln ] ) ) , if @xmath40 so that @xmath41 then , for any @xmath32 , @xmath42 ( here and in what follows , @xmath43 denotes the number of elements in a set . )
the proofs in papers @xcite involved a blend of combinatorial , variational and geometric arguments and were based on a direct analysis of the corresponding generating function via a multivariate saddle - point method for a cauchy integral @xcite or a suitable tauberian theorem @xcite . extending some of these ideas and using large deviations techniques , vershik and zeitouni
@xcite developed a systematic approach to the limit shape problem for the uniform measure on more general ensembles of convex lattice polygonal lines with various geometric restrictions .
sinai @xcite proposed an alternative , probabilistic method essentially based on randomization of the right endpoint of the polygonal line @xmath6 ; we will comment more on this approach in section [ sec1.3 ] .
let us point out that the paper @xcite contained the basic ideas but only sketches of the proofs .
some of these techniques were subsequently elaborated by bogachev and zarbaliev @xcite and also by zarbaliev in his phd thesis @xcite , however a complete proof has not been published as yet . [
rm : strict ] a polygonal line @xmath6 can be viewed as a vector sum of its consecutive edges , resulting in a given integer vector @xmath4 ; due to the convexity property , the order of parts in the sum is uniquely determined .
hence , any such @xmath1 represents an integer _ vector partition _ of @xmath44 or , more precisely , a _ strict _ vector partition ( i.e. , without proportional parts ; see @xcite ) .
this observation incorporates the topic of convex lattice polygonal lines in a general theory of integer partitions . for ordinary , one - dimensional partitions ,
the problem of limit shape can also be set up , but for a special geometric object associated with partitions , called _ young diagrams _
@xcite .
vershik @xcite pointed out that it would be interesting to study asymptotic properties of convex lattice polygonal lines under other probability measures @xmath28 on @xmath0 , and conjectured that the limit shape might be universal for some classes of measures . independently , a similar hypothesis was put forward by prokhorov @xcite . in the present paper , we prove vershik prokhorov s universality conjecture for a parametric family of probability measures @xmath45 ( @xmath46 ) on @xmath0 defined by @xmath47 with @xmath48 where the product is taken over all edges @xmath49 of @xmath6 , @xmath50 is the number of lattice points on the edge @xmath49 except its left endpoint , and @xmath51 note that for @xmath8 the measure ( [ eq : p - r ] ) is reduced to the uniform distribution on @xmath0 .
qualitatively , formulas ( [ eq : b - gamma ] ) , ( [ eq : b - k - r ] ) introduce certain probability weights for random edges on @xmath1 by encouraging ( @xmath52 ) or discouraging ( @xmath53 ) lattice points on each edge as compared to the reference case @xmath8 .
assume that @xmath54 , and consider the standard scaling transformation @xmath55 , @xmath56 .
it is convenient to work with a sup - distance between the scaled polygonal lines @xmath57 ( @xmath6 ) and the limit curve @xmath29 , based on the _ tangential parameterization _ of convex paths ( see the appendix , section [ a-1 ] ) .
more specifically , for @xmath58 $ ] denote by @xmath59 the right endpoint of that part of @xmath60 where the tangent slope ( wherever it exists ) does not exceed @xmath61 .
similarly , the tangential parameterization of the parabola arc @xmath29 ( see ( [ eq : gamma0 ] ) ) is given by the vector function @xmath62 the _ tangential distance _ between @xmath60 and @xmath29 is then defined as @xmath63 where , as before , @xmath37 is the euclidean vector norm in @xmath35 ( cf .
general definition ( [ eq : dt ] ) of the metric @xmath64 in the appendix , section [ a-1 ] ) .
we can now state our main result about the universality of the limit shape @xmath29 under the measures @xmath65 ( cf.theorem [ th:8.2a ] ) .
[ th : main1 ] for each @xmath66 and any @xmath32 @xmath67 it can be shown ( see the appendix , section [ a-1 ] ) that the hausdorff distance @xmath68 ( see ( [ eq : dh ] ) ) is dominated by the tangential distance @xmath69 defined in ( [ eq : dt ] ) ( however , these metrics are not equivalent ) . in particular , theorem [ th : main1 ] with @xmath8 recovers the limit shape result ( [ eq : vbs ] ) for the uniform distribution on @xmath0 . as was mentioned above , in the original paper by sinai @xcite the proof of the limit shape result
was only sketched , so even in the uniform case our proof seems to be the first complete implementation of sinai s probabilistic method ( which , as we will try to explain below , is far from straightforward ) .
let us also point out that theorem [ th : main1 ] is a _ non - trivial _ extension of ( [ eq : vbs ] ) since the measures @xmath10 ( @xmath70 ) are _ not close _ to the uniform distribution @xmath71 in total variation distance ( denoted by @xmath72 ) , and in fact @xmath73 as @xmath5 ( see theorem [ th : lim(r ) ] in the appendix ) .
the result of theorem [ th : main1 ] for `` pure '' measures @xmath65 readily extends to mixed measures .
[ th : main2 ] let @xmath74 be a probability measure on @xmath75 and set @xmath76 then for any @xmath32 @xmath77 follows from equation ( [ eq : rho ] ) and theorem [ th : main1 ] by lebesgue s dominated convergence theorem .
theorem [ th : main2 ] shows that the limit shape result holds true ( with the same limit @xmath29 ) when the parameter @xmath78 specifying the distribution @xmath45 is chosen at random . using the terminology designed for settings with random environments , theorems [ th : main1 ] and [ th : main2 ]
may be interpreted as `` quenched '' and `` annealed '' statements , respectively .
the universality of the limit shape @xmath29 , established in theorem [ th : main1 ] , is not a general rule but rather an exception , holding for some , but not all , probability measures on the polygonal space @xmath0 .
in fact , as was shown by bogachev and zarbaliev @xcite , _ any _ @xmath79-smooth , strictly convex curve @xmath80 started at the origin may appear as the limit shape with respect to _ some _ probability measure @xmath81 on @xmath0 , as @xmath5 .
the main results of the present paper have been recently reported ( without proofs ) in a brief note @xcite .
our proof of theorem [ th : main1 ] employs an elegant probabilistic approach first applied to convex lattice polygonal lines by sinai @xcite .
this method is based on randomization of the ( fixed ) right endpoint @xmath82 of polygonal lines @xmath6 , leading to the interpretation of a given ( e.g. , uniform ) measure @xmath28 on @xmath0 as the conditional distribution induced by a suitable probability measure @xmath83 ( depending on an auxiliary `` free '' parameter @xmath16 ) defined on the `` global '' space @xmath15 of _ all _ convex lattice polygonal lines ( with finitely many edges ) . to make the measure @xmath83 closer to @xmath28 on the subspace @xmath84 specified by the condition @xmath82 , it is natural to pick the parameter @xmath85 from the asymptotic equation @xmath86 ( @xmath5 ) .
then , in principle , asymptotic properties of polygonal lines @xmath1 ( e.g. , the limit shape ) can be established first for @xmath87 and then transferred to @xmath88 via conditioning with respect to @xmath0 and using an appropriate local limit theorem for the probability @xmath89 .
a great advantage of working with the measure @xmath83 is that it may be chosen as a `` multiplicative statistic '' @xcite ( i.e. , a direct product of one - dimensional probability measures ) , thus corresponding to the distribution of a sequence of independent random variables , which immediately brings in insights and well - developed analytical tools of probability theory . sinai s approach in @xcite was motivated by a heuristic analogy with statistical mechanics , where similar ideas are well known in the context of asymptotic equivalence , in the thermodynamic limit , of various statistical ensembles ( i.e. , _ microcanonical _ , _ canonical _ , and _ grand canonical _ ) that may be associated with a given physical system ( e.g. , gas ) by optional fixing of the total energy and/or the number of particles ( see ruelle @xcite ) . in particular , khinchin @xcite has pioneered a systematic use of local limit theorems of probability theory in problems of statistical mechanics .
deep connections between statistical properties of quantum systems ( where discrete random structures naturally arise due to quantization ) and asymptotic theory of random integer partitions are discussed in a series of papers by vershik @xcite ( see also the recent work by comtet et al .
@xcite and further references therein ) .
note also that a general idea of randomization has proved instrumental in a large variety of combinatorial problems ( see , e.g. , @xcite and the vast bibliography therein ) .
the probabilistic method is very insightful and efficient , as it makes the arguments heuristically transparent and natural .
however , the practical implementation of this approach requires substantial work , especially in the two - dimensional context of convex lattice polygonal lines as compared to the one - dimensional case exemplified by integer partitions and the corresponding young diagrams @xcite .
to begin with , evaluation of expected values and some higher - order statistical moments of random polygonal lines leads one to deal with various sums over the set @xmath90 of points @xmath91 with co - prime coordinates ( see section [ sec2.1 ] ) .
sinai @xcite was able to obtain the limit of some basic sums of such a kind by appealing to the known asymptotic density of the set @xmath90 in @xmath92 ( given by @xmath93 ) ; however , this argumentation is insufficient for more refined asymptotics . in the present paper ,
we handle this technical problem by using the mbius inversion formula ( see section [ sec3 ] ) , which enables one to reduce sums over @xmath90 to more regular sums .
as already mentioned , another crucial ingredient required for the probabilistic method is a suitable local limit theorem that furnishes a `` bridge '' between the global distribution @xmath83 and the conditional one , @xmath28 .
analytical difficulties encountered in the proof of such a result are already significant in the case of ordinary integer partitions ( for more details and concrete examples , see @xcite and further references therein ) .
the case of convex lattice polygonal lines , corresponding to two - dimensional strict vector partitions ( see remark [ rm : strict ] ) , is notoriously tedious , even though the standard method of characteristic functions is still applicable . to the best of our knowledge , after the original paper by sinai @xcite where the result was just stated ( with a minor error in the determinant of the covariance matrix @xcite ) , full details have not been worked out in the literature ( however , see @xcite ) .
we prove the following result in this direction ( cf.theorem [ th : lclt ] ) .
[ th : intro - lclt ] suppose that the parameter @xmath85 is chosen so that @xmath94 . then as @xmath5 @xmath95 where @xmath96 is the covariance matrix of the random vector @xmath97 with respect to the probability measure @xmath98 .
the quantities @xmath99 and @xmath100 , obtained via the measure @xmath14 , depend in general on the parameter @xmath78 as well .
for the sake of notational convenience the latter is omitted , which should cause no confusion since @xmath78 is always fixed , unless stated explicitly otherwise .
one can show that the covariance matrix @xmath100 is of the order of @xmath101 , and in particular @xmath102 and @xmath103 . from the right - hand side of ( [ eq : intro - lclt ] ) , it is then clear that one needs to refine the error term in the asymptotic relation @xmath104 by estimating the deviation @xmath105 to at least the order of @xmath106 .
we have been able to obtain the following estimate ( cf.theorem [ th:5.1 ] ) .
[ th : intro - a ] under the conditions of theorem @xmath107 the proof of this result is quite involved .
the main idea is to apply the mellin transform and use the inversion formula to obtain a suitable integral representation for the difference @xmath108 of the form ( @xmath109 ) @xmath110 where @xmath111 happens to be of the order of @xmath112 ( according to the `` optimal '' choice of @xmath85 as explained at the beginning of section [ sec1.3 ] ; cf .
theorem [ th : delta12 ] ) , @xmath113 is an explicit function analytic in the strip @xmath114 , and @xmath115 is the riemann zeta function . as usual , to obtain a better estimate of the integral one has to shift the integration contour in ( [ eq : intro - m ] ) as far to the left as possible , and it turns out that to get an estimate of order @xmath116 one needs to enter the critical strip @xmath117 , which requires information about zeroes of the zeta function in view of the denominator @xmath115 in ( [ eq : intro - m ] ) .
the rest of the paper is organized as follows . in section [ sec2 ] , we explain the basics of the probability method in the polygonal context and define the parametric families of measures @xmath14 and @xmath65 ( @xmath46 ) . in section [ sec3 ] , we choose suitable values of the parameter @xmath16 ( theorem [ th : delta12 ] ) , which implies convergence of `` expected '' polygonal lines to the limit curve @xmath29 ( section [ sec4 ] , theorems [ th:3.2 ] and [ th:8.1.1a ] ) . the refined error estimate ( [ eq : intro - a ] ) is proved in section [ sec5 ] ( theorem [ th:5.1 ] ) .
higher - order moment sums are analyzed in section [ sec6 ] ; in particular , the asymptotics of the covariance matrix @xmath100 is obtained in theorem [ th : k ] .
section [ sec7 ] is devoted to the proof of the local central limit theorem ( theorem [ th : lclt ] ) .
finally , the limit shape result , with respect to both @xmath14 and @xmath65 , is proved in section [ sec8 ] ( theorems [ th:8.2 ] and [ th:8.2a ] ) .
the appendix includes necessary details about the tangential parameterization and the tangential metric @xmath69 on the space of convex paths ( section [ a-1 ] ) , as well as a discussion of the total variation distance between the measures @xmath65 ( @xmath70 ) and the uniform distribution @xmath71 ( section [ a-2 ] , theorems [ th : vasser ] and [ th : lim(r ) ] ) .
let us fix some general notations frequently used in the paper . for a row - vector @xmath118 , its euclidean norm ( length )
is denoted by @xmath119 , and @xmath120 is the corresponding inner product of vectors @xmath121 .
we denote @xmath122 , @xmath123 , and simialrly @xmath124 , @xmath125 .
as was observed by sinai @xcite , one can encode convex lattice polygonal lines via suitable integer - valued functions . more specifically , consider the set @xmath90 of all pairs of co - prime non - negative integers , @xmath126 where @xmath127 stands for the greatest common divisor of two integers .
( in particular , the pairs @xmath128 and @xmath129 are included in this set , while @xmath130 is not . )
let @xmath131 be the space of functions on @xmath90 with non - negative integer values , and consider the subspace of functions with _ finite support _
, @xmath132 where @xmath133 .
it is easy to see that the space @xmath134 is in one - to - one correspondence with the space @xmath135 of all ( finite ) convex lattice polygonal lines : @xmath136 indeed , let us interpret points @xmath137 as radius - vectors ( pointing from the origin to @xmath138 ) .
now , for any @xmath139 , a finite collection of nonzero vectors @xmath140 , arranged in the order of increase of their slope @xmath141 $ ] , determines consecutive edges of some convex lattice polygonal line @xmath25 .
conversely , vector edges of a lattice polygonal line @xmath25 can be uniquely represented in the form @xmath142 , with @xmath137 and integer @xmath143 ; setting @xmath144 for such @xmath138 and zero otherwise , we obtain a function @xmath145 .
( the special case where @xmath146 for all @xmath137 corresponds to the `` trivial '' polygonal line @xmath147 with coinciding endpoints . ) that is to say , each @xmath148 determines the _ direction _ of a potential edge , only utilized if @xmath149 , in which case the value @xmath150 specifies the _ scaling factor _ , altogether yielding a vector edge @xmath151 ; finally , assembling all such edges into a polygonal line is uniquely determined by the fixation of the starting point ( at the origin ) and the convexity property .
note that , according to the above construction , @xmath152 has the meaning of the number of lattice points on the edge @xmath151 ( except its left endpoint ) .
the right endpoint @xmath26 of the polygonal line @xmath25 associated with a configuration @xmath145 is expressed by the formula @xmath153 in what follows , we shall identify the spaces @xmath23 and @xmath134 .
in particular , any probability measure on @xmath23 can be treated as the distribution of a @xmath154-valued random field @xmath155 on @xmath90 with almost surely ( a.s . ) finite support .
let @xmath157 be a sequence of non - negative numbers such that @xmath158 ( without loss of generality , we put @xmath159 ) and not all @xmath160 vanish for @xmath161 , and assume that the generating function @xmath162 is finite for @xmath163 .
let @xmath16 be a two - dimensional parameter , with @xmath164 . throughout the paper , we shall use the multi - index notation @xmath165 we now define the `` global '' probability measure @xmath156 on the space @xmath166 as the distribution of a random field @xmath167 with mutually independent values and marginal distributions of the form @xmath168 [ pr : f0 ] for each @xmath169 , the condition @xmath170 is necessary and sufficient in order that @xmath171 . according to ( [ q ] ) , @xmath172 ( @xmath137 ) . since the random variables @xmath152 are mutually independent for different @xmath137 , borel
cantelli s lemma implies that @xmath173 if and only if @xmath174 in turn , the latter inequality is equivalent to ( [ n ] ) . that is to say , under condition ( [ n ] ) a sample configuration of the random field @xmath155 belongs ( @xmath156-a.s . ) to the space @xmath134 and therefore determines a ( random ) finite polygonal line @xmath175 . by the mutual independence of the values @xmath152 ,
the corresponding @xmath83-probability is given by @xmath176 where @xmath177 is the right endpoint of @xmath1 ( see ( [ eq : xi ] ) ) , and @xmath178 note that the infinite product in ( [ eq : b ] ) contains only finitely many terms different from @xmath179 , since for @xmath180 we have @xmath181 .
hence , expression ( [ eq : b ] ) can be rewritten in a more intrinsic form ( cf.([eq : b - gamma ] ) ) @xmath182 where the product is taken over all edges @xmath49 of @xmath6 and @xmath50 is the number of lattice points on the edge @xmath49 except its left endpoint ( see section [ sec2.1 ] ) . in particular , for the trivial polygonal line @xmath183 formula ( [ q1 ] ) yields @xmath184 note , however , that @xmath185 since , due to our assumptions , ( [ eq : b ] ) implies @xmath186 for @xmath187 and hence , according to ( [ n ] ) , @xmath188 . on the subspace @xmath24 of polygonal lines with the right endpoint fixed at @xmath4 , the measure @xmath156 induces the conditional distribution @xmath189 provided , of course
, that @xmath190 ( i.e. , there is at least one @xmath6 with @xmath191 , cf.([q1 ] ) ) .
the parameter @xmath85 may be dropped from the notation for @xmath28 due to the following fact .
[ pr : noz ] the measure @xmath28 in does not depend on @xmath85 . if @xmath192 then @xmath82 and hence formula ( [ q1 ] ) is reduced to @xmath193 accordingly , using ( [ n ] ) and ( [ pn ] ) we get the expression @xmath194 which is @xmath85-free .
let us consider a special parametric family of measures @xmath197 , determined by formula ( [ q ] ) with the coefficients @xmath160 of the form @xmath198 ( note that @xmath199 , in accordance with our convention in section [ sec2.2 ] ) . by the binomial expansion formula ,
the generating function ( [ eq : b ] ) of the sequence ( [ b_k ] ) is given by @xmath200 and from ( [ q ] ) it follows that under the law @xmath98 the random variable @xmath152 has the probability generating function @xmath201 consequently , formula ( [ q ] ) specializes to @xmath202 that is to say , with respect to the measure @xmath203 the random variable @xmath152 has a negative binomial distribution with parameters @xmath78 and @xmath204 ( * ? ? ?
* , p.165 ) ; in particular , its expected value and variance are given by ( see ( * ? ? ?
* , p.269 ) ) @xmath205=\frac{rz^x}{1-z^x}{\mbox{$\;\!$}},\qquad { \mathrm{var}}[\nu(x)]=\frac{rz^x}{(1-z^x)^2}{\mbox{$\;\!$}}\qquad ( x\in{\mathcal{x}}).\ ] ] according to formulas ( [ pn ] ) and ( [ condp ] ) , the corresponding conditional measure @xmath206 is expressed as @xmath207 where @xmath208 is given by the general formula ( [ eq : b1 ] ) specialized to the coefficients @xmath209 defined in ( [ b_k ] ) . in the special case @xmath8 , we have @xmath210 so that ( [ qb ] ) is reduced to the geometric distribution ( with parameter @xmath204 ) @xmath211 whereas the conditional measure ( [ condp1 ] ) specifies the uniform distribution on @xmath0 ( cf .
@xcite ) : @xmath212 [ rm : r ] since @xmath213 , the sequence @xmath214 is strictly increasing or decreasing in @xmath215 according as @xmath52 or @xmath53 , respectively . that is to say , the measures @xmath216 and @xmath65 encourage ( @xmath52 ) or discourage ( @xmath53 ) lattice points on edges , as compared to the reference case @xmath8 . it is easy to see that condition ( [ n ] ) is satisfied and , by proposition [ pr : f0 ] , @xmath217 indeed , using ( [ g ] ) we have @xmath218 whenever @xmath219 , and the latter condition is fulfilled since @xmath220
in what follows , the asymptotic notation of the form @xmath221 ( where @xmath4 ) means that @xmath222 we also use the standard notation @xmath223 for @xmath224 as @xmath225 . throughout the paper , we shall work under the following convention about the limit @xmath40 .
[ as : c ] the notation @xmath5 signifies that @xmath225 in such a way that @xmath226 .
in particular , this implies that @xmath227 as @xmath5 , and @xmath228 from the parameter @xmath85 ( see proposition [ pr : noz ] ) in order to better adapt the measure @xmath203 to the subspace @xmath84 determined by the condition @xmath82 ( where @xmath229 is defined in ( [ eq : xi ] ) ) . to this end , it is natural to require that the latter condition be satisfied ( at least asymptotically ) for the _ expected _ value of @xmath97 ( cf .
more precisely , we will seek @xmath16 as a solution to the following asymptotic equations : @xmath230 where @xmath231 denotes expectation with respect to the distribution @xmath203 . from ( [ eq : xi ] ) ,
using the first formula in ( [ eq : nu ] ) , we obtain @xmath232 let us represent the parameters @xmath233 , @xmath234 in the form @xmath235 where the quantities @xmath236 ( possibly depending on the ratio @xmath237 ) are presumed to be bounded from above and separated from zero .
hence , ( [ e_i ] ) takes the form @xmath238 [ th : delta12 ] conditions are satisfied if @xmath239 in are chosen to be @xmath240 where @xmath241 and @xmath242 is the riemann zeta function .
let us prove the first of the asymptotic relations ( [ calibr1 ] ) .
set @xmath243 and @xmath244 then we can rewrite ( [ e_i ] ) in projection to the first coordinate as @xmath245 let us also consider the function @xmath246 ( adding terms with @xmath247 does not affect the sum , since @xmath248 vanishes at the origin ) .
recalling the definition of the set @xmath90 ( see ( [ eq : x ] ) ) , we note that @xmath92 can be decomposed as a _
union of multiples of @xmath90 : @xmath249 .
hence , the double sum in ( [ eq : f ] ) is reduced to @xmath250 & = \frac{rh{\mbox{$\;\!$}}{{\mathrm e}}^{-h\alpha_1}}{(1-{{\mathrm e}}^{-h\alpha_1})^2(1-{{\mathrm e}}^{-h\alpha_2})}{\mbox{$\;\!$}}. \end{aligned}\ ] ] by the mbius inversion formula ( see ( * ? ? ?
* theorem 270 , p.237 ) ) , @xmath251 where @xmath252 ( @xmath253 ) is the _ mbius function _ defined as follows : @xmath254 , @xmath255 if @xmath256 is a product of @xmath257 different primes , and @xmath258 if @xmath256 has a squared factor ( * ? ? ?
* , p.234 ) ; in particular , @xmath259 .
a sufficient condition for ( [ mebius_f ] ) is that the double series @xmath260 should be convergent , which is easily verified in our case : @xmath261 and , according to ( [ eq : f ] ) and ( [ f ] ) , @xmath262 using ( [ f ] ) and ( [ mebius_f ] ) , we can rewrite ( [ e_1f ] ) as @xmath263 note that ( [ alpha ] ) and ( [ delta12 ] ) imply @xmath264 where @xmath265 is defined in theorem [ th : delta12 ] .
hence , we can rewrite ( [ e_1 ] ) in the form @xmath266 we now need an elementary estimate , which will also be instrumental later on .
[ lm:2.3 ] for any @xmath143 @xmath267 there exists @xmath268 such that for all @xmath269 @xmath270 set @xmath271 and note that @xmath272 by continuity , the function @xmath273 is bounded on @xmath75 , and ( [ eq : exp_bound ] ) follows . by lemma [ lm:2.3 ] ,
the general term of the series ( [ e_1 ] ) is estimated , uniformly in @xmath215 and @xmath256 , by @xmath274 .
hence , by lebesgue s dominated convergence theorem one can pass to the limit in ( [ e_1 ] ) termwise : @xmath275 here the expression for the second sum ( over @xmath256 ) is obtained using the mbius inversion formula ( [ mebius_f ] ) with @xmath276 , @xmath277 ( cf .
* theorem 287 , p.250 ) ) .
similarly , we can check that , as @xmath5 , @xmath278 the theorem is proved .
the term @xmath279 appearing in formula ( [ zeta32 ] ) and the like , equals the asymptotic density of co - prime pairs @xmath280 among all integer points on @xmath92 ( see ( * ? ? ?
* theorem 459 , p.409 ) ) .
[ as : z ] throughout the rest of the paper , we assume that the parameters @xmath281 are chosen according to formulas ( [ alpha ] ) , ( [ delta12 ] ) . in particular , the measure @xmath14 becomes dependent on @xmath4 , as well as all @xmath14-probabilities and mean values .
for @xmath25 , denote by @xmath282 ( @xmath58 $ ] ) the part of @xmath1 where the slope does not exceed @xmath283 . hence , the path @xmath284 serves as a tangential parameterization of the scaled polygonal line @xmath31 , where @xmath285 ( see section [ sec1.2 ] , and also section [ a-1 ] below ) .
consider the set @xmath286.\ ] ] according to the association @xmath287 described in section [ sec2.1 ] , for each @xmath58 $ ] the polygonal line @xmath282 is determined by a truncated configuration @xmath288 , hence its right endpoint @xmath289 is given by @xmath290.\ ] ] in particular , @xmath291 , @xmath292 ( cf.([eq : xi ] ) ) .
similarly to ( [ e_i ] ) and ( [ e_i ] ) , @xmath293= r\sum_{k=1}^\infty\sum_{x\in { \mathcal{x}}(t)}x{\mbox{$\;\!$}}{{\mathrm e}}^{-k\langle\alpha , x\rangle } , \qquad t\in[0,\infty].\ ] ] let us also set ( cf . ( [ eq : g * ] ) ) @xmath294.\ ] ] as will be verified in the appendix ( see section [ a-1 ] ) , the vector - function @xmath295 gives a tangential parameterization of the parabola @xmath29 defined in ( [ eq : gamma0 ] ) .
the goal of this section is to establish the convergence of the ( scaled ) expectation @xmath296 $ ] to the limit @xmath297 , first for each @xmath58 $ ] ( section [ sec4.1 ] ) and then uniformly in @xmath58 $ ] ( section [ sec4.2 ] ) .
[ th:3.2 ] for each @xmath58 $ ] @xmath298=g^*_j(t)\qquad ( j=1,2).\ ] ] theorem [ th : delta12 ] implies that ( [ sh ] ) holds for @xmath299 .
assume that @xmath300 and let @xmath301 ( the case @xmath302 is considered in a similar manner ) . setting for brevity @xmath303 and arguing as in the proof of theorem [ th : delta12 ] ( see ( [ e_i ] ) , ( [ e_1f ] ) and ( [ e_1 ] ) ) , from ( [ eq : e(t ) ] ) we obtain @xmath304 & = r\sum_{k,{\mbox{$\:\!$}}{}m=1}^\infty m{\mbox{$\:\!$}}\mu(m)\sum_{x_1=1}^\infty x_1{{\mathrm e}}^{-km\alpha_1x_1}\sum_{x_2=0}^{\hat x_2 } { { \mathrm e}}^{-km\alpha_2x_2 } \notag\\ \label{z1 } & = r\sum_{k,{\mbox{$\:\!$}}{}m=1}^\infty \frac{m{\mbox{$\:\!$}}\mu(m)}{1-{{\mathrm e}}^{-km\alpha_2 } } \sum_{x_1=1}^\infty x_1{{\mathrm e}}^{-km\alpha_1x_1}\bigl(1-{{\mathrm e}}^{-km\alpha_2(\hat{x}_2 + 1)}\bigr),\end{aligned}\ ] ] where @xmath305 denotes the integer part of @xmath306 , so that @xmath307 aiming to replace @xmath308 by @xmath306 in ( [ z1 ] ) , we recall ( [ eq : alpha2:1 ] ) and rewrite the sum over @xmath309 as @xmath310 where @xmath311 using that @xmath312 ( see ( [ eq : x * ] ) ) and applying lemma [ lm:2.3 ] , we obtain , uniformly in @xmath313 and @xmath58 $ ] , @xmath314 substituting this estimate into ( [ z1 ] ) , we see that the error resulting from the replacement of @xmath308 by @xmath306 is dominated by @xmath315 returning to representation ( [ z1 ] ) and computing the sum in ( [ eq : sum+delta ] ) , we find @xmath316 = r\sum_{k,{\mbox{$\:\!$}}{}m=1}^\infty \left.\frac{m{\mbox{$\:\!$}}\mu(m)}{1-{{\mathrm e}}^{-km\alpha_2}}\cdot \frac{{{\mathrm e}}^{-km\alpha_1y}}{(1-{{\mathrm e}}^{-km\alpha_1y})^2}\ , \right|_{\,y=1+t}^{\,y=1}\\[.2pc ] + o\bigl(\alpha_1^{-2}\ln\alpha_1\bigr).\end{gathered}\ ] ] passing to the limit by lebesgue s dominated convergence theorem , similarly to the proof of theorem [ th : delta12 ] ( cf.([zeta32 ] ) ) we get , as @xmath5 , @xmath317\to \frac{1}{\kappa^3}\sum_{k=1}^\infty \frac{1}{k^{3}}\sum_{m=1}^\infty \frac{\mu(m)}{m^{2 } } \left(1-\frac{1}{(1+t)^2}\right ) = \frac{t^2 + 2t}{(1+t)^2}{\mbox{$\;\!$}},\ ] ] which coincides with @xmath318 , as claimed .
there is a stronger version of theorem [ th:3.2 ] .
[ th:8.1.1a ] convergence in is uniform in @xmath58 $ ] that is @xmath319-g^*_j(t)\bigr|=0\qquad ( j=1,2).\ ] ] for the proof , we need the following general lemma . [ lm:8.1 ]
let @xmath320 be a sequence of nondecreasing functions on a finite interval @xmath321 $ ] such that for each @xmath322 $ ] @xmath323 where @xmath324 is a continuous nondecreasing function on @xmath321 $ ] .
then the convergence @xmath325 as @xmath5 is uniform on @xmath321 $ ] . since @xmath326 is continuous on a closed interval @xmath321 $ ] , it is uniformly continuous
. therefore , for any @xmath32 there exists @xmath327 such that @xmath328 whenever @xmath329 .
let @xmath330 be a partition such that @xmath331 .
since @xmath332 for each @xmath333 , there exists @xmath334 such that @xmath335 for all @xmath336 . by monotonicity of @xmath337 and @xmath326
, this implies that for any @xmath338 $ ] and all @xmath336 @xmath339 similarly , @xmath340 .
therefore , @xmath341}|f_n(t)-f(t)|\le 2\varepsilon$ ] , and the uniform convergence follows .
suppose that @xmath301 ( the case @xmath302 is handled similarly ) .
note that for each @xmath30 the function @xmath342=\frac{1}{n_1 } \sum_{x\in { \mathcal{x}}(t ) } x_1{\mbox{$\:\!$}}{e}_{z}^{r}[\nu(x)]\ ] ] is nondecreasing in @xmath61 .
therefore , by lemma [ lm:8.1 ] the convergence ( [ sh ] ) is uniform on any interval @xmath343 $ ] .
furthermore , since @xmath344\to g^*_1(\infty)$ ] and the function @xmath318 is continuous at infinity ( see ( [ u_12 ] ) ) , for the proof of the uniform convergence on a suitable interval @xmath345 $ ] it suffices to show that for any @xmath32 one can choose @xmath346 such that , for all large enough @xmath347 and all @xmath348 , @xmath349 on account of ( [ ue_1 ] ) we have @xmath350 = { \mbox{$\;\!\!$}}\sum_{k,{\mbox{$\:\!$}}{}m=1}^\infty { \mbox{$\:\!\!$}}\frac{rm{\mbox{$\:\!$}}\mu(m)}{1-{{\mathrm e}}^{-km\alpha_2}}\cdot \frac{{{\mathrm e}}^{-km\alpha_1(1+t)}}{\bigl(1-{{\mathrm e}}^{-km\alpha_1(1+t)}\bigr)^2}\\ + o\bigl(\alpha_1^{-2}\ln\alpha_1\bigr).\end{gathered}\ ] ] note that by lemma [ lm:2.3 ] , uniformly in @xmath313 , @xmath351 returning to ( [ eq : ez ] ) , we obtain , uniformly in @xmath348 , @xmath352 & = \frac{o(1)}{(1+t)^{2 } } \sum_{k=1}^\infty \frac{1}{k^{3}}\sum_{m=1}^\infty \frac{1}{m^{2}}=\frac{o(1)}{(1+t^*)^{2}}{\mbox{$\;\!$}},\end{aligned}\ ] ] whence by ( [ alpha ] ) we get ( [ eq : e < epsilon ] ) .
for future applications , we need to refine the asymptotic formulas ( [ calibr1 ] ) by estimating the error term .
the following theorem is one of the main technical ingredients of our work .
[ th:5.1 ] suppose that the parameter @xmath85 is chosen according to formulas so that @xmath104 see theorem .
then @xmath353 as @xmath5
. for the proof of this theorem , some preparations are needed .
let a function @xmath354 be continuous and absolutely integrable on @xmath355 , together with its partial derivatives up to the second order . set @xmath356 ( as verified below , the series in ( [ f0 ] ) is absolutely convergent for all @xmath357 ) , and assume that for some @xmath358 @xmath359 consider the mellin transform of @xmath360 ( see , e.g. , ( * ? ? ?
* ch.vi , 9 ) ) , @xmath361 [ lm : delta1 ] under the above conditions the function @xmath362 is meromorphic in the strip @xmath363 with a single simple pole at @xmath364 .
moreover @xmath362 satisfies the identity @xmath365 where @xmath366 identity ( [ muntz2 ] ) is a two - dimensional analogue of mntz s formula for univariate functions ( see ( * ? ? ?
* , pp.2829 ) ) . let a function @xmath367 be continuous and continuously differentiable , and suppose that both @xmath368 and @xmath369 are absolutely integrable on @xmath370 .
it follows that @xmath371 ; indeed , note that @xmath372 hence @xmath373 exists and , since @xmath368 is integrable , the limit must equal zero .
then the well - known euler
maclaurin summation formula states that @xmath374 where @xmath375 - 1 $ ] ( cf .
a.4 , p.254 ) ) .
applying formula ( [ em1 ] ) twice to the double series ( [ f0 ] ) , we obtain @xmath376 since @xmath377 , the above conditions on the function @xmath326 imply that all integrals in ( [ eq : fi ] ) exist , hence @xmath360 is well defined for all @xmath357 .
moreover , from ( [ eq : fi ] ) it follows that @xmath378 the estimates ( [ beta ] ) and ( [ h1 ] ) imply that @xmath362 as defined in ( [ mel ] ) is a regular function for @xmath379 .
let us now note that for such @xmath380 we can rewrite ( [ mel ] ) as @xmath381 according to condition ( [ beta ] ) , the first term on the right - hand side of ( [ m(s ) ] ) , as a function of @xmath380 , is regular for @xmath382 , whereas the last term is regular for @xmath383 by ( [ h1 ] ) .
hence , formula ( [ m(s ) ] ) furnishes an analytic continuation of the function @xmath362 into the strip @xmath384 , where it is meromorphic and , moreover , has a single ( simple ) pole at point @xmath364 . finally , observing that @xmath385 and rearranging the terms in ( [ m(s ) ] ) using ( [ delta ] ) , we obtain ( [ muntz2 ] ) .
[ lm : delta2 ] under the conditions of lemma @xmath386 from ( [ delta ] ) , ( [ eq : fi ] ) we have @xmath387 as @xmath388 . combined with the estimate ( [ h1 ] ) established in the proof of lemma [ lm : delta1 ]
, this implies that the integral ( [ muntz2 ] ) converges absolutely in the strip @xmath389 . the representation ( [ eq : inverse ] ) then follows from ( [ muntz2 ] ) by the inversion formula for the mellin transform ( see ( * ? ? ?
* theorem 9a , pp.246247 ) ) .
let us consider the first coordinate , @xmath390 ( for @xmath391 the proof is similar ) .
the proof consists of several steps .
according to ( [ e_1 ] ) we have @xmath392 where ( see ( [ f+ ] ) , ( [ f ] ) ) @xmath393 note that @xmath394 moreover , using ( [ eq : alpha2:1 ] ) we have @xmath395 ( cf .
( [ zeta32 ] ) ) . subtracting ( [ eq : int(f ) ] ) from ( [ e_z ] ) , we obtain the representation @xmath396 where @xmath397 is defined in ( [ delta ] ) . clearly , the functions @xmath326 and @xmath398 satisfy the hypotheses of lemma [ lm : delta1 ] ( with @xmath399 ) . setting @xmath303 and using ( [ eq : alpha2:1 ] ) , the mellin transform of @xmath360 defined by ( [ mel ] )
can be represented as @xmath400 where @xmath401 as a result , applying lemma [ lm : delta2 ] we can rewrite ( [ dif1 ] ) as @xmath402 it is not difficult to find explicitly the analytic continuation of the function @xmath403 into the domain @xmath389 . to this end , let us represent the integral ( [ tildem ] ) as @xmath404 where @xmath405 the last two integrals in ( [ tildem1 ] ) are easily evaluated : @xmath406 where @xmath407 is the gamma function , and similarly @xmath408 substituting expressions ( [ in_1 ] ) and ( [ in_2 ] ) into ( [ tildem1 ] ) , we obtain @xmath409 since the expression in the parentheses in ( [ j ] ) is @xmath410 as @xmath411 and @xmath412 as @xmath413 , the integral ( [ j ] ) is absolutely convergent ( and therefore the function @xmath414 is regular ) for @xmath415 .
furthermore , it is well known that @xmath416 is analytic for @xmath415 ( * ? ? ?
* , p.148 ) , while @xmath115 has a single pole at point @xmath417 ( * ? ?
* , p.152 ) .
hence , the right - hand side of ( [ tildem2 ] ) is meromorphic in the semi - plane @xmath415 with poles at @xmath417 and @xmath364 .
let us estimate the function @xmath418 as @xmath419 .
firstly , by integration by parts in ( [ j ] ) it is easy to show that , uniformly in a strip @xmath420 , @xmath421 the gamma function in such a strip is known to satisfy a uniform estimate @xmath422 ( see ( * ? ? ?
* , p.151 ) ) .
furthermore , the zeta function is obviously bounded in any semi - plane @xmath423 : @xmath424 we also have the following bounds , uniform in @xmath425 , on the growth of the zeta function as @xmath419 ( see ( * ? ? ?
* theorem 1.9 , p.25 ) ) : @xmath426 as a result , by ( [ gfe ] ) , ( [ zeta_2 ] ) and ( [ zeta_123 ] ) the second and third summands on the right - hand side of ( [ tildem2 ] ) give only exponentially small contributions as compared to ( [ j(it ) ] ) , so that @xmath427 in view of ( [ m(it ) ] ) , for @xmath428 there is an absolute convergence on the right - hand side of ( [ dif2 ] ) , @xmath429 hence , the summation and integration in ( [ dif2 ] ) can be interchanged to yield @xmath430 while evaluating the sum over @xmath256 here , we used the mbius inversion formula ( [ mebius_f ] ) with @xmath431 , @xmath432 ( cf .
( [ zeta32 ] ) ; see also ( * ? ?
* theorem 287 , p.250 ) ) . substituting ( [ tildem2 ] ) into ( [ dif3 ] ) , we finally obtain @xmath433 where @xmath434\ ] ] and the function @xmath414 is given by ( [ j ] ) . by the la valle poussin theorem ( see ( * ? ?
* , theorem 5 , p.69 ) ) , there exists a constant @xmath435 such that @xmath436 in the domain @xmath437 moreover , it is known ( see ( * ? ? ?
* eq.(3.11.8 ) , p.60 ) ) that in the domain ( [ eta ] ) the following uniform estimate holds : @xmath438 without loss of generality , one can assume @xmath439 , so that ( see ( [ eta ] ) ) @xmath440 therefore , @xmath441 ( see ( [ psi ] ) ) is regular for all @xmath442 such that @xmath443 ( @xmath444 ) .
let us show that the integration contour @xmath445 in ( [ dif4 ] ) can be replaced by the curve @xmath446 ( @xmath444 ) . by the cauchy theorem , it suffices to check that @xmath447 we have @xmath448 in view of the remark after formula ( [ zeta_1 ] ) , we have @xmath449 , hence application of the estimate ( [ zeta_2 ] ) gives , for @xmath450 , @xmath451 , @xmath452 ( since @xmath453 , we may assume that @xmath454 ) . to estimate the expression in the square brackets in ( [ psi ] ) , we use the estimates ( [ j(it ) ] ) , ( [ gfe ] ) , ( [ zeta_123 ] ) and ( [ zeta_1 ] ) . as a result , we obtain @xmath455 which implies that the right - hand side of ( [ it ] ) tends to zero as @xmath456 , as required .
therefore , the integral in ( [ dif4 ] ) can be rewritten in the form @xmath457 it remains to estimate the integral ( [ lambda ] ) as @xmath5 .
let us set @xmath458 \end{aligned}\ ] ] ( see ( [ psi ] ) ) , then equation ( [ lambda ] ) is rewritten as @xmath459 using that @xmath460 ( see ( [ alpha ] ) ) , we get @xmath461 since by ( [ eta ] ) @xmath462 let us now note that , as @xmath5 , the integrand function in ( [ |psi| ] ) tends to zero for each @xmath61 , because @xmath453 and @xmath463 ( see ( [ eta ] ) ) . finally ,
eligibility of passing to the limit under the integral sign follows from lebesgue s dominated convergence theorem .
indeed , the integrand function in ( [ |psi| ] ) is bounded by @xmath464 , and integrability of the latter is easily checked by applying the estimates ( [ j(it ) ] ) , ( [ gfe ] ) , ( [ zeta_123 ] ) and ( [ zeta_1 ] ) to the expression ( [ psi0 ] ) , which yields ( cf . ( [ psi(it ) ] ) )
@xmath465 thus , we have shown that the integral in ( [ |psi| ] ) is @xmath466 as @xmath5 , hence @xmath467 . substituting this estimate into ( [ dif4 ] )
, we obtain the statement of theorem [ th:5.1 ] .
the proof is complete .
according to the second formula in ( [ eq : nu ] ) , we have @xmath468=\frac{rz^x}{(1-z^x)^2}=r\sum_{k=1}^\infty kz^{kx}.\ ] ] let @xmath96 be the covariance matrix ( with respect to the measure @xmath14 ) of the random vector @xmath469 .
recalling that the random variables @xmath152 are independent for different @xmath148 and using ( [ d_z ] ) , we see that the elements @xmath470 of the matrix @xmath100 are given by @xmath471= r\sum_{k=1}^\infty \sum_{x\in { \mathcal{x } } } k x_i
x_j z^{kx},\ \quad i , j\in\{1,2\}.\ ] ] [ th : k ] as @xmath5 @xmath472 where @xmath265 is defined in theorem and the matrix @xmath473 is given by @xmath474 1&2{\mbox{$\:\!$}}n_2/n_1{\mbox{$\;\!\!$}}\end{array } \right)\!.\ ] ] let us consider @xmath475 ( the other elements of @xmath100 are analyzed in a similar manner ) . substituting ( [ alpha ] ) into ( [ d_zxi ] ) , we obtain @xmath476 using the mbius inversion formula ( [ mebius_f ] ) , similarly to ( [ e_1 ] ) expression ( [ d_zxi+ ] ) can be rewritten in the form @xmath477 note also that @xmath478 returning to the representation ( [ d1_1(t ) ] ) and using ( [ eq : neweq ] ) , we obtain @xmath479 by lemma [ lm:2.3 ] , the general term in the series ( [ eq : neweq1 ] ) admits a uniform estimate @xmath480 .
hence , by lebesgue s dominated convergence theorem one can pass to the limit in ( [ eq : neweq1 ] ) to obtain @xmath481 using ( [ alpha ] ) and ( [ delta12 ] ) , this yields @xmath482 as required ( cf .
( [ eq : sigma ] ) , ( [ eq : sigma1 ] ) )
. denote @xmath483,\qquad x\in { \mathcal{x}},\ ] ] and for @xmath484 set @xmath485,\qquad \mu_k(x):={e}_{z}^{r}{\mbox{$\;\!\!$}}\bigl|\nu_0(x)^k\bigr|\ ] ] ( for notational simplicity , we suppress the dependence on @xmath78 and @xmath85 ) . [
lm : mu < m ] for each @xmath484 and all @xmath148 @xmath486 omitting for brevity the argument @xmath138 , by newton s binomial formula and lyapunov s inequality we obtain @xmath487&=\sum_{i=0}^k \binom{k}{i } { \mbox{$\:\!$}}m_i{\mbox{$\;\!$}}m_1^{k - i}\\ & \le\sum_{i=0}^k \binom{k}{i}{\mbox{$\:\!$}}m_k^{i / k}{\mbox{$\;\!$}}m_k^{(k - i)/k}=2^k m_k{\mbox{$\:\!$}},\end{aligned}\ ] ] and ( [ eq : mu < m ] ) follows . [ lm:6.2 ] for each @xmath484 there exist positive constants @xmath488 and @xmath489 such that for all @xmath148 @xmath490 fix @xmath137 and let @xmath491 $ ] be the characteristic function of the random variable @xmath152 with respect to the measure @xmath203 . from ( [ phi ] )
it follows that @xmath492 let us first prove that for any @xmath484 @xmath493 where @xmath494 . indeed , if @xmath495 then differentiation of ( [ p.f ] ) yields @xmath496 which is in accordance with ( [ f ] ) if we put @xmath497 .
assume now that ( [ f ] ) is valid for some @xmath215 .
differentiating ( [ f ] ) once more , we obtain @xmath498 where we have set @xmath499 j{\mbox{$\:\!$}}c_{j,{\mbox{$\:\!$}}{}k}+(r+j-1){\mbox{$\:\!$}}c_{j-1,{\mbox{$\:\!$}}{}k}{\mbox{$\:\!$}},\ \ & 2\le j\le k,\\[.2pc ] ( r+k){\mbox{$\:\!$}}c_{k,{\mbox{$\:\!$}}{}k}{\mbox{$\:\!$ } } , & j = k+1 .
\end{array } \right.\ ] ] hence , by induction , formula ( [ f ] ) is valid for all @xmath215 .
now , by ( [ f ] ) we have @xmath500 since @xmath501 .
hence , inequalities ( [ t=0 ] ) hold with @xmath502 , @xmath503 .
[ lm : sum(z ) ] for any @xmath484 and @xmath267 @xmath504 ) , by lemma [ lm:2.3 ] we have @xmath505 where @xmath506 . on the other hand , @xmath507 since @xmath508 and @xmath509 ,
from ( [ eq : z^k1 ] ) and ( [ eq : z^k2 ] ) we see that for the proof of ( [ eq : z^k ] ) it remains to show @xmath510 using the mbius inversion formula ( [ mebius_f ] ) , similarly as in sections [ sec3 ] and [ sec4 ] we obtain @xmath511 \label{eq : m } & = \sum_{m=1}^\infty \mu(m)\,\frac{{{\mathrm e}}^{-m\alpha_1}+{{\mathrm e}}^{-m\alpha_2}- { { \mathrm e}}^{-m(\alpha_1+\alpha_2)}}{(1-{{\mathrm e}}^{-m\alpha_1})(1-{{\mathrm e}}^{-m\alpha_2})}{\mbox{$\;\!$}}.\end{aligned}\ ] ] by lemma [ lm:2.3 ] , the general term of the series ( [ eq : m ] ) is @xmath512 ( uniformly in @xmath256 ) .
hence , by lebesgue s dominated convergence theorem , the right - hand side of ( [ eq : m ] ) is asymptotically equivalent to @xmath513 and ( [ eq : z^k3 ] ) follows .
[ lm:6.3 ] for any @xmath484 @xmath514 readily follows from the estimates ( [ t=0 ] ) and lemma [ lm : sum(z ) ] .
[ lm : muk ] for any integer @xmath515
@xmath516 an upper bound follows ( for all @xmath161 ) from the inequality ( [ eq : mu < m ] ) and lemma [ lm:6.3 ] . on the other hand , by lyapunov s inequality and formula ( [ d_z ] ) , for any @xmath517 we have @xmath518\bigr)^{k/2}= \frac{r^{k/2}z^{kx/2}}{(1-z^x)^k}{\mbox{$\;\!$}},\ ] ] and
a lower bound follows by lemma [ lm : sum(z ) ] .
[ lm:6.6 ] for each @xmath484 and @xmath109 @xmath519^{2k}\ !
= o\bigl(|n|^{4k/3}\bigr),\qquad n\to\infty.\ ] ] let @xmath301 ( the case @xmath302 is considered similarly ) .
using the notation ( [ eq : nu0 ] ) , we obtain , by the multinomial expansion , @xmath520^{2k } = { e}_{z}^{r}\biggl({\mbox{$\;\!\!$}}\sum_{x\in { \mathcal{x } } } x_1{\mbox{$\:\!$}}\nu_0(x)\biggr){\vphantom{\biggr)}}^{2k}\\ \label{eq : sum - prod } & = \sum_{\ell=1}^{2k}\ ! \sum_{\mbox{\scriptsize $ \begin{array}{c } k_1,\dots,{\mbox{$\:\!$}}k_\ell\ge1,\\ k_1{\mbox{$\;\!\!$}}+\cdots+k_\ell=2k \end{array } $ } } \!\!\!\!c_{k_1{\mbox{$\;\!\!$}},\dots,{\mbox{$\:\!$}}k_\ell } \sum_{\{x^{1}{\mbox{$\:\!\!$}},{\mbox{$\:\!$}}\dots,{\mbox{$\;\!$}}x^{\ell}\}\subset { \mathcal{x } } } { \mbox{$\:\!$}}\prod_{i=1}^\ell ( x_1^i)^{k_i } { e}_{z}^{r}{\mbox{$\;\!\!$}}\bigl[\nu_0(x^i)^{k_i}\bigr],\end{aligned}\ ] ] where @xmath521 are combinatorial coefficients accounting for the number of identical terms in the expansion . using that @xmath522=0 $ ]
, we can assume that @xmath523 for all @xmath524 .
since @xmath525 , this implies that @xmath526 .
hence , recalling the notation ( [ eq : m - mu ] ) and using lemma [ lm : muk ] , we see that the internal sum in ( [ eq : sum - prod ] ) ( over @xmath527 ) is bounded by @xmath528 and the lemma is proved .
as was explained in the introduction ( see section [ sec1.3 ] ) , the role of a local limit theorem in our approach is to yield the asymptotics of the probability @xmath529 appearing in the representation of the measure @xmath10 as a conditional distribution , @xmath530 , @xmath531 ( see ( [ condp1 ] ) ) . as before ,
we denote @xmath532 , @xmath533 , where the random vector @xmath229 is defined in ( [ eq : xi ] ) . from ( [ d_zxi ] ) , it is easy to see ( e.g. , using the cauchy
schwarz inequality together with the characterization of the equality case ) that the matrix @xmath100 is positive definite ; in particular , @xmath534 and hence @xmath100 is invertible .
let @xmath535 be the ( unique ) square root of the matrix @xmath536 ( see , e.g. , ( * ? ? ?
* ch.6 , 5 , pp .
9394 ) ) , that is , a symmetric , positive definite matrix such that @xmath537 .
denote by @xmath538 the density of a standard two - dimensional normal distribution @xmath539 ( with zero mean and identity covariance matrix ) , @xmath540 then the density of the normal distribution @xmath541 ( with mean @xmath99 and covariance matrix @xmath100 ) is given by @xmath542 with these notations , we can now state our local limit theorem .
[ th : lclt ] uniformly in @xmath543 @xmath544 theorem [ th : lclt ] is a two - dimensional local central limit theorem for the sum @xmath177 with independent terms whose distribution depends on a large parameter @xmath4 ; however , the summation scheme is rather different from the classic one , since the number of non - vanishing terms is not fixed in advance and , moreover , the summands actually involved in the sum are determined by sampling .
one implication of theorem [ th : lclt ] will be particularly useful .
[ cor : q ] as @xmath5 @xmath545 where @xmath546 . before proving the theorem
, we have to make some ( quite lengthy ) technical preparations , collected below in sections [ sec7.2][sec7.4 ] . the matrix norm induced by the euclidean vector norm @xmath37
is defined by @xmath547 .
it is well known that for a ( real ) square matrix @xmath548 its norm is given by @xmath549 where @xmath550 is the spectral radius of a matrix , defined to be the largest modulus of its eigenvalues ( see , e.g. , ( * ? ? ?
* , pp.210211 ) ) .
we need some general facts about the matrix norm @xmath551 . even though they are mostly well - known , specific references are not easy to find ( cf . ,
e.g. , @xcite ) . for the reader s convenience ,
we give neat proofs of the lemmas below based on the spectral characterization ( [ eq : norm ] ) .
[ lm:7.1.1 ] if @xmath548 is a real matrix then @xmath552 .
the matrix @xmath553 is symmetric and non - negative definite , hence , using ( [ eq : norm ] ) , we obtain @xmath554 , as claimed .
[ lm:|a| ] if @xmath555 is a real @xmath556 matrix then @xmath557 note that @xmath558 , where @xmath559 denotes the trace , and furthermore @xmath560 since @xmath561 by ( [ eq : norm ] ) , this implies ( [ eq : m - norm ] ) .
the following simple fact pertaining to dimension @xmath562 seems to be less known .
[ lm:7.1.2 ] let @xmath548 be a symmetric @xmath563 matrix with @xmath564 .
then @xmath565 let @xmath566 and @xmath567 ( @xmath568 ) be the eigenvalues of @xmath548 , then @xmath569 which makes the equality ( [ eq : d=2 ] ) obvious .
in this section , we collect some information about the asymptotic behavior of the matrix @xmath570 .
the next lemma is a direct consequence of theorem [ th : k ] .
[ lm : detk ] as @xmath5 @xmath571 let us now estimate the norms of the matrices @xmath100 and @xmath535 .
[ lm : k_z ] as @xmath5 one has @xmath572 .
lemma [ lm:|a| ] and theorem [ th : k ] imply @xmath573 and the required estimate follows .
[ nv ] for the matrix @xmath535 one has @xmath574 and [ lm:7.1.2 ] we have @xmath575 and an application of lemmas [ lm : detk ] and [ lm : k_z ] completes the proof .
we also need to estimate the so - called _ lyapunov coefficient _
@xmath576 where @xmath577 ( see ( [ eq : m - mu ] ) ) .
[ lm:7.1 ] as @xmath5 one has @xmath578
. follows from ( [ l3 ] ) using lemmas [ nv ] and [ lm : muk ] ( with @xmath579 ) .
recall from section [ sec2.1 ] that , with respect to the measure @xmath203 , the random variables @xmath580 are independent and have negative binomial distribution with parameters @xmath78 and @xmath204 .
in particular , @xmath152 has the characteristic function ( see ( [ p.f ] ) ) @xmath581 hence the characteristic function @xmath582 of the vector @xmath583 is given by @xmath584 [ lm:7.2_f ] let @xmath585 be the characteristic function of the random variable @xmath586 $ ] .
then @xmath587 where @xmath588 see .
let a random variable @xmath589 be independent of @xmath590 and have the same distribution , and set @xmath591 .
note that @xmath592=0 $ ] and @xmath593= 2{\mbox{$\;\!$}}{\mathrm{var}}[\nu(x)]=2{\mbox{$\:\!$}}\mu_2(x)$ ] .
we also have the inequality @xmath594 ( see ( * ? ? ?
* lemma 8.8 , pp.6667 ) ) .
hence , by taylor s formula , the characteristic function of @xmath595 can be represented in the form @xmath596 where @xmath597 .
now , using the elementary inequality @xmath598 and the fact that @xmath599 , we get @xmath600 and the lemma follows by ( [ x.f_2 ] )
. the characteristic function of the vector @xmath601 is given by @xmath602 [ lm:7.2_f ] if @xmath535 then for all @xmath603 @xmath604 using ( [ x.f_5 ] ) and ( [ x.f_4 ] ) , we obtain @xmath605 the first sum in ( [ x.f_6 * ] ) is evaluated exactly as @xmath606 & = \lambda v_z k_z{\mbox{$\:\!$}}v_z\lambda^{{\mbox{$\;\!\!$}}{{\mbox{\tiny$\top$}}}}\!= |\lambda|^2 , \end{aligned}\ ] ] since @xmath607 . for the second sum in ( [ x.f_6 * ] ) , by the cauchy
schwarz inequality and on account of ( [ l3 ] ) we have @xmath608 now , substituting ( [ dl0 ] ) , ( [ dl ] ) into ( [ x.f_6 * ] ) , we get ( [ x.f_6 ] ) .
[ lm:7.2 ] if @xmath609 then @xmath610 let us first suppose that @xmath611 .
then @xmath612 , so ( [ x.f_6 ] ) implies @xmath613 .
hence , @xmath614 in accord with ( [ 16l ] ) .
suppose now that @xmath615 .
taylor s formula implies @xmath616 where @xmath617 . by lyapunov
s inequality , @xmath618 , so @xmath619 , we have from ( [ dl ] ) @xmath620 and so ( [ eq:1/2 ] ) yields @xmath621 similarly , using the elementary inequality @xmath622 , from ( [ eq:1/2 ] ) we obtain @xmath623 whence , in view of ( [ eq : m3 ] ) and a general bound @xmath624 , it follows that @xmath625 consider the function @xmath626 of complex variable @xmath627 , choosing the principal branch of the logarithm ( i.e. , such that @xmath628 ) .
taylor s expansion implies @xmath629 for @xmath630 , where @xmath597 . by ( [ eq : f-1 ] ) ,
( [ eq:|f-1| ] ) and ( [ |f-1|^2 ] ) this yields @xmath631 where @xmath632 . substituting this into ( [ x.f_5 ] ) , due to ( [ dl0 ] ) and ( [ dl ] )
we obtain @xmath633 using the elementary inequality @xmath634 , which holds for any @xmath635 , we have @xmath636 and the proof is completed .
[ lm:7.3 ] for all @xmath603 @xmath637 where @xmath638 according to ( [ x.f_5 ] ) , we have @xmath639 using ( [ f_nu ] ) , for any @xmath640
we can write @xmath641 utilizing this estimate under the sum in ( [ eq : j_1 ] ) ( with @xmath642 ) and recalling the notation ( [ alpha ] ) , we arrive at ( [ f_j ] ) .
let us first deduce the corollary from the theorem . according to theorem [ th:5.1 ] , @xmath643 .
together with lemma [ nv ] this implies @xmath644 hence , by lemma [ lm : detk ] we get @xmath645 and ( [ sim ] ) follows from ( [ eq : lclt ] ) . by definition ,
the characteristic function of the random vector @xmath646 is given by the fourier series @xmath647 hence the fourier coefficients are expressed as @xmath648 where @xmath649 ( see ( [ eq : phi1 ] ) ) is given by @xmath650 so by the fourier inversion formula @xmath651 note that if @xmath652 then , according to lemmas [ nv ] and [ lm:7.1 ] , @xmath653 which implies that @xmath654 .
using this observation and subtracting ( [ f_o ] ) from ( [ l_1 ] ) , we get , uniformly in @xmath543 , @xmath655 where @xmath656 by the substitution @xmath657 , the integral @xmath658 is reduced to @xmath659 on account of lemmas [ lm : detk ] , [ lm:7.1 ] , and [ lm:7.2 ] .
similarly , again putting @xmath657 and passing to the polar coordinates , we get , due to lemmas [ lm : detk ] and [ lm:7.1 ] , @xmath660 estimation of @xmath661 is the main part of the proof . using lemma [ lm:7.3 ]
, we obtain @xmath662 where @xmath663 is given by ( [ j_0 ] ) .
the condition @xmath664 implies that @xmath665 for a suitable ( small enough ) constant @xmath666 and hence @xmath667 for otherwise from lemmas [ lm : k_z ] , [ lm:7.1 ] and [ lm:7.2 ] it would follow @xmath668 hence , the estimate ( [ i3 ] ) is reduced to @xmath669 to estimate the first integral in ( [ i3.1 ] ) , by keeping in the sum ( [ j_0 ] ) only pairs of the form @xmath670 , @xmath671 , we obtain @xmath672 because @xmath673 for any @xmath674 . since @xmath675
, we have @xmath676 substituting this estimate into ( [ j_1 ] ) , we conclude that @xmath677 is asymptotically bounded from below by @xmath678 , uniformly in @xmath675 . thus , the first integral in ( [ i3.1 ] ) is bounded by @xmath679 similarly , the second integral in ( [ i3.1 ] ) ( where @xmath680 ) is estimated by reducing summation in ( [ j_0 ] ) to that over @xmath681 only . as a result
, we obtain that @xmath682 . substituting this estimate together with ( [ i1 ] ) and ( [ i2 ] ) into ( [ i ] ) , we get ( [ eq : lclt ] ) , and so the theorem is proved .
recall the notation ( see ( [ eq : x_n ] ) , ( [ xi(t ) ] ) ) @xmath683 , where @xmath684 , @xmath58 $ ] .
as stated at the beginning of section [ sec4 ] , the tangential parameterization of the scaled polygonal line @xmath31 is given by @xmath685,\ ] ] whereas the limit shape @xmath29 determined by equation ( [ eq : gamma0 ] ) is parameterized by the vector - function @xmath686 defined in ( [ u_12 ] ) ( see more details in the appendix , section [ a-1 ] ) .
the goal of this section is to use the preparatory results obtained so far and prove the uniform convergence of random paths @xmath687 to the limit @xmath688 in probability with respect to both @xmath14 ( section [ sec8.1 ] ) and @xmath689 ( section [ sec8.2 ] ) .
let us point out that , in view of ( [ eq : tilde - xi ] ) , theorems [ th:8.2 ] and [ th:8.2a ] below can be easily reformulated ( cf .
theorem [ th : main1 ] stated in the introduction ) using the tangential distance @xmath690 ( see ( [ eq : d - t ] ) ; cf .
general definition ( [ eq : dt ] ) in section [ a-1 ] below ) .
let us first establish the universality of the limit shape under the measures @xmath14 , which in conjunction with the next theorem [ th:8.2a ] illustrates the asymptotic `` equivalence '' of the probability spaces @xmath17 and @xmath18 .
[ th:8.2 ] for each @xmath32 @xmath691 by theorems [ th:3.2 ] and [ th:8.1.1a ] , the expectation of the random process @xmath692 uniformly converges to @xmath693 as @xmath5 .
therefore , we only need to check that for each @xmath32 @xmath694\bigr|>\varepsilon\biggr\}=0.\ ] ] note that the random process @xmath695 $ ] has independent increments and zero mean , hence it is a martingale with respect to the natural filtration @xmath696 , @xmath58 $ ] . from the definition of @xmath697 ( see ( [ xi(t ) ] ) ) , it is also clear that @xmath698 is a cdlg process , that is , its paths are everywhere right - continuous and have left limits .
therefore , applying the kolmogorov
doob submartingale inequality ( see , e.g. , ( * ? ? ? * corollary 2.1 , p.14 ) ) and using theorem [ th : k ] , we obtain @xmath699 and the theorem is proved . we are finally ready to prove our main result about the universality of the limit shape under the measures @xmath65 ( cf.theorem [ th : main1 ] ) .
[ th:8.2a ] for any @xmath32 @xmath700 similarly as in the proof of theorem [ th:8.2 ] , the claim of the theorem is reduced to the limit @xmath701 where @xmath702 $ ] . using ( [ pn ] )
we get @xmath703 applying the kolmogorov
doob submartingale inequality and using lemma [ lm:6.6 ] ( with @xmath579 ) , we obtain @xmath704 ^ 6}{(\varepsilon n_j)^{6 } } = o\bigl(|n|^{-2}\bigr).\ ] ] on the other hand , by corollary [ cor : q ] @xmath705 in view of these estimates , the right - hand side of ( [ a1 ] ) is dominated by a quantity of order of @xmath706 , and the theorem is proved .
let @xmath707 be the space of paths in @xmath355 starting from the origin and such that each path @xmath708 is continuous , piecewise @xmath709-smooth ( i.e. , everywhere except a finite set ) , bounded and convex , and , furthermore , its tangent slope ( where it exists ) is non - negative , including the possible value @xmath710 .
convexity implies that the slope is non - decreasing as a function of the natural parameter ( i.e. , the length along the path measured from the origin ) . for @xmath708 ,
let @xmath711 denote the right endpoint of the ( closure of the ) part of @xmath80 where the tangent slope does not exceed @xmath58 $ ] .
note that the functions @xmath712 , @xmath713 are cdlg ( i.e. , right - continuous with left limits ) , and @xmath714 more precisely , equation ( [ eq : slope ] ) holds at points where the tangent exists and its slope is strictly growing ; corners on @xmath80 correspond to intervals where both functions @xmath715 and @xmath716 are constant , whereas flat ( straight line ) pieces on @xmath80 lead to simultaneous jumps of @xmath715 and @xmath716 .
the canonical limit shape curve @xmath29 ( see ( [ eq : gamma0 ] ) ) is determined by the parametric equations ( cf .
( [ eq : g * ] ) , ( [ u_12 ] ) ) @xmath717.\ ] ] indeed , it can be readily seen that the functions ( [ u_12 _ ] ) satisfy the cartesian equation ( [ eq : gamma0 ] ) for @xmath29 ; moreover , it is easy to check that @xmath61 in equations ( [ u_12 _ ] ) is the tangential parameter : @xmath718 and hence ( cf .
( [ eq : slope ] ) ) @xmath719 the tangential distance @xmath69 between paths in @xmath720 is defined as follows : @xmath721 [ lm : a1 ] the hausdorff distance @xmath68 defined in is dominated by the tangential distance @xmath69 @xmath722 first of all , note that any path @xmath708 can be approximated , simultaneously in metrics @xmath68 and @xmath69 , by polygonal lines @xmath723
( for instance , by inscribing polygonal lines with refined edges in the arc @xmath80 ) so that @xmath724 this reduces inequality ( [ eq : dhdt ] ) to the case where @xmath725 , @xmath726 are polygonal lines . moreover , by symmetry it suffices to show that @xmath727 note that if a point @xmath728 can be represented as @xmath729 with some @xmath730 ( i.e. , @xmath138 is a vertex of @xmath725 ) , then @xmath731 and inequality ( [ eq : t*1a ] ) follows .
suppose now that @xmath728 lies on an edge say of slope @xmath346between two consecutive vertices @xmath732 and @xmath733 , then @xmath734 with some @xmath735 and @xmath736 note that for all @xmath737 we have @xmath738 , hence @xmath739 if @xmath740 is continuous at @xmath741 , then inequality ( [ eq : t*0 ] ) extends to @xmath741 : @xmath742 substituting this inequality into the right - hand side of ( [ eq : t * * ] ) , we see that @xmath743 , which implies ( [ eq : t*1a ] ) .
if @xmath744 then @xmath346 coincides with the slope of some edge on @xmath726 ( with the endpoints , say , @xmath745 and @xmath746 ) , which is thus parallel to the edge on @xmath725 where the point @xmath138 lies ( i.e. , with the endpoints @xmath732 , @xmath733 ) .
setting @xmath747 , we have @xmath748 , @xmath749 .
to complete the proof , it remains to observe that the shortest distance from a point on a base of a trapezoid to the opposite base does not exceed the maximum length of the two lateral sides .
hence , @xmath750\bigr\}\\[-.25pc ] & \le \max\bigl\{|g_{\gamma_1}(s_*)-g_{\gamma_2}(s_*)| , & \le d_{\mathcal t}(\gamma_1,\gamma_2),\end{aligned}\ ] ] and the bound ( [ eq : t*1a ] ) follows .
note , however , that the metrics @xmath68 and @xmath751 are _ not equivalent_. for instance , if @xmath708 is a smooth , strictly convex curve with curvature bounded below by a constant @xmath752 , then for an inscribed polygonal line @xmath753 with edges of length no more than @xmath32 , its tangential distance from @xmath80 is of the order of @xmath754 , but the hausdorff distance is of the order of @xmath755 : @xmath756 moreover , in the degenerate case where the curvature may vanish , the difference between the two metrics may be even more dramatic .
for instance , it is possible that two polygonal lines are close to each other in the hausdorff distance while their tangential distance is quite large . for an example , consider two straight line segments @xmath757 of the same ( large ) length @xmath758 , both starting from the origin and with very close slopes , so that the euclidean distance @xmath759 between their right endpoints is small ; then @xmath760 whereas @xmath761 .
note that if probability measures @xmath762 and @xmath28 on the polygonal space @xmath0 are asymptotically close to each other in _ total variation _
( _ tv _ ) , that is , @xmath763 as @xmath5 , where @xmath764 then the problem of universality of the limit shape is resolved in a trivial way , in that if a limit shape @xmath29 exists under @xmath28 then the same curve @xmath29 provides the limit shape under @xmath762 .
indeed , assuming that the event @xmath765 satisfies @xmath766 as @xmath5 , we have @xmath767 however , the family @xmath768 , defined by formula ( [ condp ] ) with the coefficients ( [ b_k ] ) , is _ not _ close to @xmath12 in total variation , at least uniformly in @xmath66 .
[ th : vasser ] for every fixed @xmath30 the limiting distance in total variation between @xmath65 and @xmath71 as @xmath769 or @xmath770 is given by @xmath771 to obtain a lower bound for @xmath772 in the case @xmath770 , consider the polygonal line @xmath773 consisting of two edges , horizontal ( from the origin to @xmath774 ) and vertical ( from @xmath774 to @xmath4 ) .
the corresponding configuration @xmath775 is determined by the conditions @xmath776 , @xmath777 and @xmath778 otherwise .
note that @xmath779 as @xmath770 ( see ( [ b_k ] ) ) , hence @xmath780 , where @xmath781 is the total number of integer points on @xmath1 ( excluding the origin ) .
we have @xmath782 , so @xmath783 as @xmath770 . on the other hand , for any @xmath6 ( @xmath784 ) one
has @xmath785 .
indeed , for @xmath148 we have @xmath786 and , moreover , @xmath787 unless @xmath788 or @xmath789 .
hence , for any @xmath6 ( @xmath784 ) , @xmath790 so that @xmath791 and @xmath792 as @xmath770
. therefore , from ( [ condp ] ) we get @xmath793 and it follows that @xmath794 for the case @xmath795 , consider the polygonal line @xmath796 consisting of one edge leading from the origin to @xmath4 .
that is , @xmath797 and @xmath798 otherwise , where @xmath799 . clearly , @xmath800 ,
while for any other @xmath6 ( i.e. , with more than one edge ) , by ( [ b_k ] ) we have @xmath801 as @xmath802 . therefore , according to ( [ condp ] ) , @xmath803 and similarly to ( [ b*1 ] ) we obtain @xmath804 the upper bound for @xmath772 ( uniform in @xmath78 ) follows from the known fact ( see @xcite ) that the total variation distance can be expressed in terms of a certain vasershtein ( kantorovich rubinstein , cf .
@xcite ) distance : @xmath805.\ ] ] here the infimum is taken over all pairs of random elements @xmath806 and @xmath807 defined on a common probability space @xmath808 with values in @xmath0 and the marginal distributions @xmath65 and @xmath71 , respectively ; @xmath809 denotes expectation with respect to the probability measure @xmath810 , and the function @xmath811 on @xmath812 is such that @xmath813 if @xmath814 and @xmath815 if @xmath816 ( therefore defining a discrete metric in @xmath0 ) . choosing @xmath806 and @xmath807 so that they are independent of each other , we obtain @xmath817= 1-{\mathsf{p}}\{x = y\}\\[.3pc ] & = 1-\sum_{\vargamma\in{{\varpi}}_n}{\mathsf{p}}\{x=\vargamma , y=\vargamma\ } = 1-\sum_{\vargamma\in{{\varpi}}_n}{p}_n^r(\vargamma)\cdot { p}_n^1(\vargamma)\\ & = 1-\frac{1}{\#({{\varpi}}_n)}\sum_{\vargamma\in{{\varpi}}_n}{p}_n^r(\vargamma ) = 1-\frac{1}{\#({{\varpi}}_n)}{\mbox{$\;\!$}}.\end{aligned}\ ] ] combining this estimate with ( [ b*1 ] ) and ( [ b*2 ] ) , we obtain ( [ eq : limvar ] ) . in the limit
@xmath5 , theorem [ th : vasser ] yields @xmath818 that is to say , in the successive limit @xmath819 , @xmath5 , the measures @xmath65 and @xmath71 become singular with respect to each other .
moreover , one can show that the distance @xmath772 is not small even for a fixed @xmath70 . to this end
, it suffices to find a function on @xmath0 possessing a limiting distribution ( possibly degenerate ) under each @xmath65 , with the limit depending on the parameter @xmath78 .
recalling remark [ rm : r ] , it is natural to seek such a function in the form referring to integer points on @xmath6 . indeed , for the statistic @xmath820 introduced in the proof of theorem
[ th : vasser ] , the following law of large numbers holds ( see bogachev and zarbaliev ( * ? ? ?
* theorem 3 ) and zarbaliev @xcite ) .
[ lm : a3 ] under assumption for each @xmath66 and any @xmath32 @xmath821 the result ( [ eq : a ] ) implies that , for any @xmath70 and all @xmath32 small enough , @xmath822 for every fixed @xmath70 we have @xmath823 hence the measures @xmath65 and @xmath71 on @xmath0 are asymptotically singular with respect to each other as @xmath5 .
we would like to express our deep gratitude to ya .
g. sinai for introducing us to the beautiful area of random partitions and for stimulating discussions over the years .
our sincere thanks are also due to yu.v.prokhorov and a.m.vershik for useful remarks and discussions at various stages of this work .
we are also thankful to the anonymous referees for the careful reading of the manuscripts and useful suggestions that have helped to improve the presentation of the paper .
bogachev , l.v . and zarbaliev , s.m .
approximation of convex curves by random lattice polygons .
preprint ni04003-igs , isaac newton inst .
, cambridge .
available at http://www.newton.cam.ac.uk/preprints/ni04003.pdf comtet , a. , majumdar , s.n .
, ouvry , s. and sabhapandit , s. ( 2007 ) .
integer partitions and exclusion statistics : limit shapes and the largest parts of young diagrams .
_ j. stat .
theory exp .
_ , p10001 .
freiman , g.a . and
granovsky , b.l .
clustering in coagulation - fragmentation processes , random combinatorial structures and additive number systems : asymptotic formulae and limiting laws . _
soc . _ * 357 * 24832507 .
freiman , g.a .
and yudin , a.a .
the interface between probability theory and additive number theory ( local limit theorems and structure theory of set addition ) . in _
representation theory , dynamical systems , and asymptotic combinatorics .
ser . _ * 217 * ( v.kaimanovich and a.lodkin , eds . ) , _ advances in the mathematical sciences _ * 58 * 5172 .
amer . math .
soc . , providence , ri .
hardy , g.h . and
wright , e.m .
( 1960 ) . _
an introduction to the theory of numbers _
oxford univ . press , oxford .
horn , r.a . and johnson , c.r .
_ matrix analysis _
, corrected reprint from the 1st edition ( 1985 ) .
cambridge univ . press , cambridge .
vershik , a.m. ( 1995 ) .
asymptotic combinatorics and algebraic analysis . in _ proceedings of the international congress of mathematicians
zrich , _ , _ vol . 2 _ ( s.d .
chatterji , ed . ) 13841394 .
birkhuser , basel . | let @xmath0 be the set of convex polygonal lines @xmath1 with vertices on @xmath2 and fixed endpoints @xmath3 and @xmath4 .
we are concerned with the _ limit shape _ , as @xmath5 , of `` typical '' @xmath6 with respect to a parametric family of probability measures @xmath7 on @xmath0 , including the uniform distribution ( @xmath8 ) for which the limit shape was found in the early 1990s independently by a.m.vershik , i.brny and ya.g.sinai .
we show that , in fact , the limit shape is _ universal _ in the class @xmath9 , even though @xmath10 ( @xmath11 ) and @xmath12 are asymptotically singular .
measures @xmath10 are constructed , following sinai s approach , as conditional distributions @xmath13 , where @xmath14 are suitable product measures on the space @xmath15 , depending on an auxiliary `` free '' parameter @xmath16 .
the transition from @xmath17 to @xmath18 is based on the asymptotics of the probability @xmath19 , furnished by a certain two - dimensional local limit theorem .
the proofs involve subtle analytical tools including the mbius inversion formula and properties of zeroes of the riemann zeta function . . |
thomson scattering of laser beam photons by moderately relativistic electrons ( with energy on the order of tens mev ) has been proposed as a source of hard x - ray radiation in the range of 10100 kev @xcite .
the x - ray radiation from such a source ( laser electron x - ray generator
lexg ) is expected to be tightly collimated and therefore to possess a high spectral brilliance thus making it suitable for a wide range of applications @xcite including the spectroscopy ( exafs and xanes ) @xcite , ordinary and phase x - ray tomography @xcite , x - ray structural analysis @xcite as well as in biology and medicine @xcite .
theoretical and numerical analysis of properties of the lexg radiation has attracted significant attention @xcite .
however one aspect has not , in our opinion , been sufficiently addressed yet .
this is the polarization of the outgoing x - ray radiation and its dependence on the laser beam polarization and the properties of electron beam especially on its emittance .
some formulas for the polarization of the compton back - scattered gamma radiation were presented in @xcite but the authors focused their analysis primarily on the gamma radiation flux angular dependence and its spectral brilliance while giving polarization only a precursory treatment .
it should be noted that the polarization of back - scattered gamma radiation is important in nuclear physics and it has been experimentally measured @xcite .
in addition , it was also used as a tool to measure the polarization of electron and positron beams @xcite . from general considerations
it follows that the finite emittance of electron beam should cause degradation of the polarization degree of x - ray radiation changing it from unity ( full polarization ) , which is expected for the illumination by a fully polarized laser beam .
moreover , all three stocks parameters of the x - ray radiation can change significantly if the electron beam has a finite emittance . in the present paper
we derive general formulas for the polarization of the x - ray radiation emitted by lexg . using the specially written program code
we conduct a number of numerical experiments for a range of laser and electron beam parameters .
we demonstrate how the finite emittance of electron beam degrades and generally affects the x - ray radiation polarization .
in the case considered in this paper of x - ray radiation in the range of tens kiloelectronvolts produced by thomson scattering of an optical laser beam on an electron beam with a modest electron energy on the order of tens megaelectronvolts the following approximations hold : photon electron scattering can be treated as a classical scattering of zero mass particles ( photons ) by finite mass particles ( electrons ) .
this approximation neglects both quantum and wave effects .
the process is characterized by a differential cross - section , which depends in the general case on momentums and energies of both photons and electrons and on the polarization of photons . in the lexg x - ray radiation
is assumed to arise as a result of collisions of trains of electron bunches with synchronized trains of optical ( laser ) pulses .
the electron bunches can circulate in a storage ring or be produced by a linear accelerator .
the polarization of both laser light and x - rays can be characterized by a hermitian polarization matrix of dimension @xmath0 , which can be represented in the following form @xmath1 where @xmath2 is the radiation intensity and @xmath3 , @xmath4 @xmath5 are the stocks parameters ( @xmath6 )
. the polarization degree of the radiation is defined as @xcite @xmath7 where the intensity of the radiation can be expressed as @xmath8 .
it is well known that a general expression for the number of scattering events of a particle with 4d momentum @xmath9 by a particle with 4d momentum @xmath10 in the direction @xmath11 ( @xmath12 ) , in the energy interval @xmath13 and into a unit of four dimensional volume @xmath14 in this paper .
] can be written as @xcite @xmath15 where the summing over index @xmath16 is implied , @xmath17 is a solid angle element in the direction @xmath11 , @xmath18 is the differential scattering cross - section of the particle with 4d momentum @xmath19 on the particle with 4d momentum @xmath20 in the energy interval @xmath13 .
the currents of particles @xmath19 and @xmath20 are denoted as @xmath21 and @xmath22 , respectively . to take the photon polarization into account
the scalar variables @xmath23 and @xmath24 in ( [ q00 ] ) should be replaced with @xmath0 matrices similar to one defined in ( [ q000 ] ) . from expression ( [ q00 ] ) taking into account the radiation polarization and assuming that particle @xmath25 represents photons with zero mass it is possible to obtain that the number of photons with momentum @xmath26 ( @xmath27 ) in a fixed polarization state scattered by the electrons with momentum @xmath28 in the direction @xmath11 from a unit of four dimensional volume @xmath29 in the frequency interval @xmath30 @xcite is @xmath31 where @xmath32 is the four dimensional electron current with momentum @xmath33 , @xmath34 is the four dimensional photon current with wave vector @xmath35 ( its square is zero ) , @xmath36 is the electron @xmath37 factor for energy @xmath38 , @xmath39 is the matrix of differential scattering cross - sections of a photon on an electron in frequency interval @xmath40 and @xmath41 and @xmath42 are the spatial densities of the electrons and photons in respective beams
. the cross - section matrix can be expressed through the polarization matrix of laser radiation @xmath43 as follows @xmath44 where @xmath45 is a symmetrical non - hermitian transformation matrix for the radiation field amplitudes , which is defined below , and @xmath43 is defined as @xmath46 where @xmath47 , @xmath48 @xmath49 are the stocks parameters of laser radiation .
the number and polarization of emitted photons defined in ( [ q01 ] ) should be integrated over electron and laser photon momentums as well as over time to obtain the x - ray radiation flux and polarization ( in the matrix form ) emitted from an element of the spatial volume @xmath50 where @xmath51 is the circulation frequency of electron bunches in the storage ring of lexg or pulse frequency in the trains produced by the linear accelerator .
expression ( [ q04 ] ) gives the flux and polarization of x - ray photons averaged over a large number of laser pulses and electron bunches from a 3d spatial volume element @xmath52 into a solid angle @xmath17 and frequency interval @xmath40 .
using formula ( [ q04 ] ) the average flux and polarization of x - ray photons in solid angle @xmath17 and the frequency interval @xmath40 can be calculated as @xmath53 whereas the average spectral brilliance of the x - ray radiation in the direction @xmath11 and at the frequency @xmath54 is @xmath55 where @xmath56 is position of the observation point and @xmath57 is a narrow spectral interval usually assumed to be equal to @xmath58 .
let s now introduce two stationary coordinate frames : @xmath59 and @xmath60 .
the first is related to the electron beam with its z - axis parallel to the average velocity of electron bunches and the second to the laser beam with its @xmath61-axis parallel to the average direction of laser pulses .
for such a choice of coordinate systems the collision angle between the laser and electron beams is close to @xmath62 .
the general geometry of the electron beam laser beam interaction used in this paper is shown in fig.[scheme ] . to calculate the x - ray radiation polarization , as was noted above , a field amplitude transformation matrix @xmath63
should be used instead of a scalar cross - section ( see ( [ q02 ] ) ) .
such a matrix can be generally expressed as @xmath64 where @xmath65 is an orthogonal matrix representing the rotation by the angle @xmath66 , which is the angle between the plane of vectors @xmath67 and @xmath11 and the plane of axes @xmath68 and @xmath69 .
diagonal matrix @xmath70 is @xmath71 where @xmath72 and @xmath73 are scalar scattering cross - sections for s- and p - polarized laser beams , respectively
. it can be demonstrated using ( [ q02 ] ) and ( [ q101 ] ) that elements of the cross - section matrix @xmath39 @xmath74 can be expressed as @xmath75 ) ([q186 ] ) ) as well as angles , vectors and shifts are shown . ] from expressions ( [ q02])([q03 ] ) and ( [ q101])([q102 ] ) it follows that the polarization degree of x - ray radiation ( [ q001 ] ) ( assuming that the angular spread of electron momentums in the electron beam is zero ) is transformed as @xmath76 where @xmath77 and intensity @xmath78 .
the value of @xmath2 is equal to the x - ray radiation intensity produced by scattering of a naturally polarized laser beam . in particular , from ( [ q109 ] ) it can be seen that the fully polarized laser radiation ( @xmath79 ) produces the fully polarized x - ray radiation if the angular spread of electron momentums in the electron beam is neglected .
expressions ( [ q105])([q108 ] ) correspond to formulas ( 35 ) from @xcite .
coefficients @xmath72 and @xmath73 can be obtained from the differential cross - sections for linearly polarized light . for our purpose
we can neglect all corrections to this cross - section related to collision angle @xmath80 between the laser and electron beams .
such corrections are proportional to its difference @xmath81 from @xmath62 squared : @xmath82 and therefore they are very small .
this allows one to use the cross - section for a head - on collision of an electron with a photon when @xmath83 . as it was shown in @xcite , the differential cross - section for linearly polarized laser beam
can be written as @xmath84,\label{q061}\\ \frac{\omega'}{\omega_l}&=\frac{1+v}{1-v\cos\theta},\quad v=|\mathbf{v}|,\quad \mathbf{vn}=v\cos\theta , \label{q062}\end{aligned}\ ] ] where frequency @xmath85 is related to the laser frequency by formula ( [ q062 ] ) , @xmath86 is the angle between the electric field vector of laser beam and axis @xmath68 , @xmath87 is the thomson cross - section .
coefficients @xmath72 and @xmath73 are obtained from ( [ q061 ] ) when @xmath88 and @xmath89 , respectively .
let s consider the spatial distributions of electrons and photons .
since the betatron and synchrotron oscillations of electrons in a storage ring can be considered approximately independent from each other @xcite the volume densities @xmath90 and @xmath91 can be factorized into a transversal part ( @xmath92 ) and a longitudinal part ( @xmath93 ) , which determine the distribution of electrons and photons by the transversal momentum and coordinate and by the energy and longitudinal coordinate , respectively .
so , the electron and photon distributions can be written as @xmath94 where @xmath95 @xmath96 are the total number of electrons and photons in an electron bunch and a laser pulse , respectively , @xmath97 is the temporal delay between an electron bunch and a laser pulse and @xmath98 is a vector accounting for the spatial shift between an electron bunch and a laser pulse .
the longitudinal component of photon distribution function in ( [ q112 ] ) is assumed to have a gaussian form in space @xmath99 where @xmath100 is the laser pulse length and @xmath101 is the average frequency of laser radiation . in formula ( [ q12 ] ) the laser pulse
is considered to be monochromatic , that is usually justified for sufficiently long ( @xmath102 ps ) pulses in the case of lexg .
the longitudinal component of electron distribution function in ( [ q111 ] ) by coordinate @xmath69 and the energy ( @xmath37 ) is also gaussian @xmath103 where @xmath104 is the electron bunch length , @xmath105 is the average gamma factor of electron bunch and @xmath106 is the gamma factor distribution width .
for transversal distribution functions in ( [ q111 ] ) and ( [ q112 ] ) the common form used in literature @xcite is that of a gaussian beam @xmath107 where @xmath108 and @xmath109 are the transversal electron momentum spreads along @xmath110 and @xmath68 axes , respectively and @xmath111 is a unit vector in the direction of laser beam .
it is well known @xcite that @xmath112 and functions @xmath113 , @xmath113 and @xmath114 are @xcite @xmath115 where @xmath116 is the beta function of electron beam at the interaction point in direction of @xmath110(@xmath68 ) axis , @xmath117 is the transversal emittance of electron beam in direction of @xmath110(@xmath68 ) axis and @xmath118 is the rayleigh length of laser beam . in ( [ q152 ] )
the photon transversal momentum distribution is in the form of a delta function , which corresponds to the approximation considered above where the scattering cross - section does not depend on the collision angle between an electron and a photon . in ( [ q13 ] ) ,
( [ q151 ] ) and ( [ q152 ] ) the electron energy distribution as well as the electron transversal momentum distribution are considered to be independent from the spatial coordinates
an assumption , which is justified in case of lexg because the transversal sizes of both beams are relatively large .
one can note that time is only present in the longitudinal components of electron and photon densities @xmath119 and @xmath120 .
therefore the integral over time in ( [ q04 ] ) can be computed analytically @xmath121 we are now ready to write the final expression for polarization matrix @xmath122 from ( [ q04 ] ) using the cross - sections and distributions obtained above .
so , taking into account integral ( [ q14 ] ) and averaging the cross - section matrix ( [ q104 ] ) over @xmath37 , expression ( [ q04 ] ) can be transformed into @xmath123 where @xmath124 is the cross - section polarization matrix averaged over the distribution of @xmath37 , coordinates @xmath59 and @xmath60 are related to each other by the following expressions ( see fig.[scheme ] ) consisting of two sequential rotations around @xmath110 and @xmath69 axes ( taking into account the axial symmetry of laser beam ) @xmath125 here @xmath126 is the observation angle and angle @xmath127 corresponds to the rotation around axis @xmath69 . in formulas ( [ q181 ] ) and ( [ q182 ] ) the collision angle @xmath128 ( @xmath129 ) is not assumed to be equal to @xmath62 as in differential cross - section matrix ( [ q104 ] ) .
.the lexg parameters used for numerical experiments [ cols=">,^,<",options="header " , ] taking into account that cross - section ( [ q062 ] ) contains a delta - function , the averaging over @xmath37 in @xmath39 ( see ( [ q105])([q108 ] ) ) can be performed analytically .
then the averaged ( over @xmath37 ) diagonal matrix elements @xmath130 , @xmath131 and @xmath132 can be obtained @xmath133 ^ 2}{\displaystyle 4\delta\gamma^2},\label{q251}\\ \overline{m_{12}}&=\sqrt{\overline{m_{11}}\,\overline{m_{22}}},\label{q252}\\ \theta^2&=2(1-\mathbf{vn}).\nonumber\end{aligned}\ ] ] and the corresponding electron beam energies .
the angles and energies were chosen to produce the radiation peaks at the same energy of 30 kev . ] as it is clear from ( [ q19 ] ) , to calculate the x - ray polarization one should average the polarization matrix @xmath124 from ( [ q104 ] ) over the transversal spread of electron momentums .
this operation is equivalent to averaging of expressions ( [ q105])([q108 ] ) and in a general case leads to a partial depolarization of initially fully polarized x - ray radiation .
the averaging can not be performed analytically for an arbitrary observational angle . however , in the special case of zero observational angle @xmath126 the matrix elements @xmath72 and @xmath73 does not depend on the angle @xmath66 , which in this case is equal to the azimuth angle between an electron velocity and the average velocity of electron bunches .
now expressions ( [ q105])([q108 ] ) can be analytically averaged over angle @xmath66 , which results in the following expressions @xmath134 to obtain polarization matrix @xmath135 expressions ( [ q26])([q29 ] ) need to be further averaged over the angle between the velocity of an electron and the average velocity of electron bunch .
the result of such an averaging depends on the ratio of angular width of x - ray distribution in @xmath130 and @xmath131 to the width of angular spread of the electron momentums ( [ q151 ] ) .
it can be noted that formulas ( [ q26])([q29 ] ) are the same as formulas ( 38 ) in @xcite . and the corresponding electron energies as function of the x - ray radiation energy .
the transversal spread of electron momentums was taken into account .
the angles and energies were chosen to produce the radiation peaks at the same energy of 30 kev , as in fig.[f1 ] . ]
to calculate the x - ray radiation characteristics of lexg a program code was written in the java programming language @xcite .
the code is based on the model outlined in this paper .
it is capable of computing the full polarization matrix of x - ray radiation for arbitrary lexg parameters . and two x - ray energies as function of the observation angle .
the transversal spread of electron momentums is taken into account . ] to demonstrate how the transversal emittance and other beam parameters influence the polarization of the x - ray radiation we conducted a number of numerical experiments .
the main parameters of lexg used in the calculations are shown in table [ tab1 ] . in the figures below the spectral brilliance and polarization of x - ray radiation of lexg with parameters from table [ tab1 ]
are depicted as functions of various parameters . and two x - ray energies as function of the electron and laser beam waist radii , which were considered equal .
the waist radii were varied by adjusting the beta - function and rayleigh length .
the transversal spread of electron momentums was taken into account as well . ] as function of the relative electron energy spread @xmath136 .
the transversal spread of electron momentums was taken into account as in fig.[f2 ] . ] in fig.[f1 ] the spectral brilliance of lexg radiation for a naturally polarized laser beam ( @xmath137 ) is shown for three different observation angles @xmath126 and electron beam energies @xmath38 .
the collision angle of the electron and laser beams @xmath128 is set at 50 mrad . for convenience .
] as it was noted above , the intensity of the naturally polarized light serves as the denominator in the calculation of the stocks parameters and degree of polarization of the x - ray radiation . from fig.[f1
] one can see that , in particular , the width of spectral brilliance curves significantly increases and the peak spectral brilliance significantly decreases as the observation angle grows .
the curves are asymmetrical with respect to their maxima with long exponential tails in the direction of lower x - ray energies .
they become progressively more symmetrical as the observational angle is increased .
this asymmetry arises because at zero observational angle the x - ray radiation of higher energies is emitted by higher energy electrons moving strictly in the direction of the electron beam .
there are only a few electrons due to the relatively narrow electron energy spectrum of ( [ q13 ] ) .
the long exponential tail is created mainly by the electrons moving non - collinearly to the electron beam direction . the number of such electrons decreases exponentially as function of the angle according to ( [ q151 ] ) . in fig.[f2 ] the x - ray radiation polarization degree for three observation angles @xmath126 and the electron energies @xmath38 as function of the x - ray radiation energy are shown .
the transversal spread of the electron momentums is taken into account
the polarization degree would be unity without it .
the collision angle @xmath128 is 50 mrad and the laser beam is linearly polarized with @xmath138 and @xmath139 ( the polarization plane was rotated by @xmath140 around @xmath69 axis relative to @xmath68 axis ) .
it can be seen that the transversal electron momentum spread ( directly related to the transversal emittance ) causes a depolarization of the emitted x - ray radiation , which increases with the observation angle .
it is clear that for the parameters listed in table [ tab1 ] depolarization of the x - ray radiation by the electron beam emittance is rather small a few percent only .
the polarization degree also approaches unity as the x - ray energy grows .
significant depolarization is observed only for the x - ray energy less than its peak value in fig.[f1 ] .
this can be explained by the fact that the higher energy x - rays arise as a result of scattering by higher energy electrons , which move strictly in the direction of the electron beam , while the lower energy x - rays are produced by electrons moving at some angles to the electron beam direction . in fig.[f3][f5 ]
the degree of polarization for the same three electron energies as in fig.[f2 ] is plotted against the observation angle , beam waist diameter and relative electron energy spread @xmath136 for two x - ray energies 24 and 26 kev .
the latter energies are less than the peak energies in fig.[f1 ] . in fig.[f4][f5 ] observation angles are the same as in fig.[f2 ] for the same electron beam energies .
other parameters are are also the same as in fig.[f2 ] .
it can be seen that the polarization degree can exhibit rather complex behavior with local minima and maxima , see fig.[f3 ] .
the polarization degree at 26 kev is generally higher than at 24 kev as in fig.[f2 ] .
the origin of the non - monotonic behavior can be traced back to the interaction between the finite angular momentum spread of electron beam with the polarization parameters in ( [ q105])([q108 ] ) during the averaging over electron momentums . in fig.[f4 ]
the polarization degree looks like a shelf when the beam s waist is small and the observational angle is zero ( solid curves ) .
this behavior is explained by sort of a saturation of the polarization degradation when the transversal electron momentum spread exceeds x - ray beam angular width from ( [ q24])([q251 ] ) .
the saturated polarization degree in this case is determined by average parameters ( [ q26])([q29 ] ) .
in fig.[f5 ] the polarization degree increases with @xmath136 at smaller observation angles and weakly decreases at larger observation angles .
such a behavior is explained by the fact that at smaller angles as the energy spread increases relatively more x - rays are emitted by electrons moving in the direction of electron beam , which reduces the angular spread of the contributing electrons . at larger angles
the behavior switches in the opposite direction because a larger contribution from the electrons moving in the direction of electron beam means a larger angular spread of the contributing electrons . in fig.[f6 ]
stocks parameters as function of the x - ray energy are shown for observation angle of 5.6 mrad .
the electron energy is set at 46.10 mev .
the remaining parameters are the same as in fig.[f2 ] for @xmath141 mrad . .5 and @xmath5 as well as the polarization degree as function of the x - ray energy .
the observational angle is 5.6 mrad .
the transversal electron momentum spread is zero ( top ) or not ( bottom).,title="fig : " ] .5 and @xmath5 as well as the polarization degree as function of the x - ray energy .
the observational angle is 5.6 mrad .
the transversal electron momentum spread is zero ( top ) or not ( bottom).,title="fig : " ] from fig.[f6 ] it is evident that the transversal electron momentum spread radically changes the x - ray polarization . in the top graph x - ray radiation
is scattered only by the electrons with momentums parallel to the electron beam direction whereas in the bottom graph higher energy x - ray radiation is scattered mainly by electrons moving in the direction of observer .
this makes the higher energy x - rays polarization in the bottom graph the same as for zero observational angle ( not shown ) whereas in the top graph its polarization remains rotated by about @xmath140 as for lower energy x - rays .
in the this paper a working model for the calculation of polarization of x - ray radiation produced by a laser - electron x - ray generator is presented .
it allows one to simulate the full polarization state of x - ray radiation for an arbitrary initial polarization state of laser beam .
the model takes into account the transversal momentum spread of electron beam , which has significant influence over the final polarization state of x - ray radiation .
the model has been implemented in a specially written program code in java programming language . using a number of numerical experiments we demonstrated that the transversal momentum spread of electron beam causes partial depolarization of the scatted x - ray radiation .
however for typical lexg parameters this depolarization was rather small a few percent only . in addition , the orientation of the x - ray radiation polarization can change dramatically when the transversal electron moment spread is taken into account .
the model of the x - ray radiation polarization of lexg presented in this work may be useful for evaluation of the lexg applications where polarization of the x - ray radiation is important .
such applications include , for example , the study of surface characteristics of the materials by the x - ray ellipsometry @xcite and the polarization sensitive spectroscopy for instance , investigation of x - ray magnetic circular dichroism near absorption edges in magnetic materials @xcite .
the authors of the paper want to thank prof .
v.i shvedunov for fruitful discussions concerning the physics of electron storage rings and injectors .
this work was supported by the program of fundamental investigations of the russian academy of science presidium `` fundamental and applied problems of photonics and physics of new optical materials '' . | in this paper we analyze the polarization of the x - ray radiation coming from laser electron x - ray generator ( lexg ) .
we obtain general relations connecting the polarization state of outgoing x - ray radiation to the polarization state of laser beam as well as to the parameters of electron beam .
we demonstrate that finite electron beam emittance causes a partial depolarization of initially fully polarized x - ray radiation even when the laser beam is fully polarized .
we demonstrate with a number of numerical experiments that finite electron beam emittance can in some cases fundamentally alter the polarization state of x - ray radiation as compared to the polarization state of x - ray radiation scattered by electron beam with a zero emittance .
possible applications of polarized lexg s radiation are discussed . , |
we use the tight - binding parametrization of our first - principles band structure of the antibonding band created by hybridization between cu @xmath78 and o @xmath79 orbitals as our starting point .
the fs reconstruction is modelled due to sdw which coexists with the @xmath69wave superconductivity.@xcite while we choose a @xmath80modulation of the sdw , the results are general and are reproduced by charge density wave , @xmath50-density wave , or flux phase with similar modulation.@xcite furthermore , a two - dimensional stripe model with incommensurate modulation along the diagonal direction also predicts the coexistence of electron and hole - pockets in addition to other open fss.@xcite our obtained results of the qpi , neutron and arpes spectra are equivalent and reproducible as long as an electron pocket is present at @xmath81points irrespective of its microscopic origin .
the hubbard - bcs hamiltonian in momentum space is@xcite @xmath82 where @xmath83 is the electronic creation ( destruction ) operator with momentum @xmath84 and spin @xmath85 .
@xmath86 is the free particle dispersion .
@xmath87 is the hubbard onsite interaction term chosen to be 0.86ev .
the sdw order parameter @xmath88 is treated within self - consistent mean - field approximation .
the bcs superconducting gap is @xmath89/2.$ ] , where @xmath90=30mev is the experimental gap parameter for ybco@xmath27 taken from arpes data.@xcite the corresponding single particle @xmath32 green s function is constructed from eq .
[ eq1 ] which includes umklapp part from spin density wave and anomalous term coming from the sc gap.@xcite the qpi maps and non - interacting susceptibility are calculated as convolutions of the green s function @xmath91_{11}$ ] and @xmath92 .
we calculate the self - energy due to all components of fluctuations along the spin and charge degrees of freedom within self - consistent gw - approach@xcite @xmath93 @xmath94 is the fluctuation potential obtained within random - phase approximation ( rpa ) as @xmath95 , where @xmath96 is the imaginary part of the rpa susceptibility of transverse spin ( @xmath97 ) , longitudinal spin ( @xmath98 ) and charge ( @xmath98 ) correlations functions .
finally , the vertex correction is approximated within ward s identity as @xmath99 .
the calculation is performed in real frequency space using analytical continuation @xmath100 .
[ fig5 ] schematically demonstrates the evolution of the electron pocket as a function of doping in hole doped cuprates . at half - filling ,
strong sdw order opens up an insulating gap as shown in fig .
[ fig5](b ) .
doping reduces the strength of the pseudgap interaction , see fig .
[ fig5](a ) . with hole doping ,
the doped holes accumulate at the top of the lower sdw band [ red line in fig .
[ fig5](c ) ] , which give rise to a hole pocket at the nodel point as shown in fig .
[ fig5](e ) . near optimal doping where the pseudogap is very small
, the bottom of the upper sdw band drops below the fermi level around @xmath101 in fig .
[ fig5](d ) .
thus an electron pocket appears as shown in fig .
[ fig5](f ) . in this doping range
the pseudogap opening shifts its location to the ` hot - spot ' region between the hole pocket and electron pocket .
it should be noted that the magnitude of the pseudogap can be so small in this region that it may be overlooked due to the large superconducting gap for materials like bi2212 or ybco . with sufficiently large magnetic field
when the superconducting gap is suppressed , the electron pocket becomes visible in quantum oscillation or hall effect probes .
in fig . [ fig6 ] , we compare the evolution of the qpi patterns in the case of a paramagnetic ground state , hole - pocket , and coexisting electron+hole pockets .
as mentioned in the main text , there are several distinguishing features to unambiguously identify the electron pocket that will show up collectively in the dispersion , intensity , and constant energy profile of the qpi pattern .
( 1 ) in a paramagnetic state , all qpi vectors show continuous energy dependence with no kink or non - dispersive pattern . for the case of a hole pocket without any electron pocket on the fs , all the dispersion features stop at the energy where the ` bright - spots ' reach to the top of the dispersion .
no new qpi vector appears above this energy and along the @xmath108-direction , the qpi vectors do not extend to @xmath24 while along the diagonal it does not cross the @xmath22-boundary .
on the other hand , in the case of coexisting electron and hole pockets both these features should be present .
( 2 ) the associated intensity of all qpi vectors also reflects the presence of an electron pocket . in a hole - pocket ,
a peak in the intensity occurs at the tip of the hole - pocket which is less than the sc gap amplitude .
the peak extends to the sc gap amplitude in the case of a paramagnetic ground state .
for the electron and hole pocket , both peaks will be present and can be used to identify the presence of an electron - pocket .
( 3 ) finally , the constant energy cuts of the qpi pattern can help distinguish the electron pocket from a hole - pocket , but the former can not be separated from a paramagnetic ground state . as discussed in point ( 1 ) above , if @xmath109 as well as @xmath110 at some energy , that will be an unambiguous signature of the presence of an antinodal fs .
i. dimov , _ et al .
_ , _ competing order , fermi surface reconstruction , and quantum oscillations in underdoped high temperature superconductors .
b * 78 * , 134529 ( 2008 ) .
a.j . millis , m. r. norman , _ antiphase stripe order as the origin of electron pockets observed in 1/8-hole - doped cuprates .
b * 76 * , 220503(r ) ( 2007 ) . r.k .
kaul , y. b. kim , s.sachdev , t. senthil _ algebraic charge liquids . _ nature physics * 4 * , 28 ( 2007 ) .
f. laliberte _
_ , _ fermi - surface reconstruction by stripe order in cuprate superconductors .
_ preprint available at http://arxiv.org/abs/1102.0984 .
a. hackl , m. vojta , s. sachdev _ quasiparticle nernst effect in stripe - ordered cuprates .
b * 81 * , 045102 ( 2010 ) .
d. leboeuf , _ et al .
_ , _ electron pockets in the fermi surface of hole - doped high-@xmath77 superconductors .
_ nature * 450 * , 533 ( 2007 ) .
n. doiron - leyraud , _ et al .
_ , _ quantum oscillations and the fermi surface in an underdoped high-@xmath77 superconductor .
_ nature * 447 * , 565 ( 2007 ) .
s. sebastian _
et al . _ , a multi - component fermi surface in the vortex state of an underdoped high-@xmath77 superconductor .
nature * 454 * , 200 ( 2008 ) . in ybco
, the presence of electron - like fs may have other band - structure origin .
tanmoy das , unpublished .
wen , p. a. lee , _ theory of underdoped cuprates _ ,
lett . * 76 * , 503 ( 1996 ) .
m. r. norman _ et al .
_ , _ destruction of the fermi surface in underdoped high-@xmath77 superconductors . _
nature * 392 * , 157 ( 1998 ) .
t. pereg - barnea , h. weber , g. refael , and m. franz , _
quantum oscillations from fermi arcs _
, nature physics * 6 * , 44 ( 2009 ) .
n. harrison , _ spin - density wave fermi surface reconstruction in underdoped yba@xmath2cu@xmath3o@xmath111 . _
lett . * 102 * , 206405 ( 2009 ) .
t. das , r. s. markiewicz , a. bansil _ competing order scenario of two - gap behavior in hole - doped cuprates .
b * 77 * , 134516 ( 2008 ) .
t. das , r. s. markiewicz , and a. bansil , _ optical model - solution to the competition between a pseudogap phase and a mott - gap phase in high - temperature cuprate superconductors .
b * 81 * , 174504 ( 2010 ) .
t. das , r. s. markiewicz , and a. bansil , _ reconstructing the bulk fermi surface and the superconducting gap properties from neutron scattering experiments .
_ preprint avaiable at http://arxiv.org/abs/1110.0756
. t. das , unpublished ( 2012 ) .
t. hanaguri , y. kohsaka , j. c. davis , c. lupien , i. yamada , m. azuma , m. takano , k. ohishi , m. ono , h. takagi , _ quasi - particle interference and superconducting gap in a high - temperature superconductor ca@xmath112na@xmath113cuo@xmath2cl@xmath2 _ , nature physics * 3 * , 865 ( 2007 ) .
ar abanov , a. v. chubukov , _ a relation between the resonance neutron peak and arpes data in cuprates _ ,
lett . * 83 * , 165 ( 1999 ) .
m. eschrig , m. r. norman , _ effect of the magnetic resonance on the electronic spectra of high - tc superconductors . _
phys . rev .
b * 67 * , 144503 ( 2003 ) .
i. eremin , d. k. morr , a. v. chubukov , k. h. bennemann , m. r. norman , _ novel neutron resonance mode in @xmath69wave superconductors .
lett . * 94 * , 147001 ( 2005 ) .
k. mcelroy , _ et al .
_ _ destruction of antinodal state coherence via ` checkerboard ' charge ordering in strongly underdoped superconducting bi@xmath2sr@xmath2cacu@xmath2o@xmath114 .
_ preprint available at http://arxiv.org/abs/cond-mat/0406491 ( 2004 ) .
_ spectroscopic fingerprint of phase - incoherent superconductivity in the cuprate pseudogap state .
_ science * 325 * , 1099 ( 2009 ) .
t. das , and a. v. balatsky , _ two energy scales in the magnetic resonance spectrum of electron and hole doped pnictide superconductors .
lett . * 106 * , 157004 ( 2011 ) .
s. huefner , m. a. hossain , a. damascelli , g. a. sawatzky , _ two gaps make a high - temperature superconductor ?
71 * , 062501 ( 2008 ) .
_ , _ doping dependence of the @xmath22 shadow band in la - based cuprates studied by angle - resolved photoemission spectroscopy .
_ new j. phys . * 13 * , 013031 ( 2011 ) .
j. chang , _ et al .
_ , _ electronic structure near the 1/8-anomaly in la - based cuprates .
_ new j. phys . * 10 * , 103016 ( 2008 ) .
m. plate , _ et al .
_ , _ fermi surface and quasiparticle excitations of overdoped tl@xmath2ba@xmath2cuo@xmath115 by arpes .
lett . * 95 * , 077001 ( 2005 ) .
y. kohsaka _
_ , _ how cooper pairs vanish approaching the mott insulator in bi@xmath2sr@xmath2cacu@xmath2o@xmath114 . _
nature * 454 * , 1072 ( 2008 ) . | fingerprint of the electron - pocket in cuprates has been obtained only in numerous magneto - transport measurements , but its absence in spectroscopic observations pose a long - standing mystery .
we develop a theoretical tool to provide ways to detect electron - pockets via numerous spectroscopies including scanning tunneling microscopy ( stm ) spectra , inelastic neutron scattering ( ins ) , and angle - resolved photoemission spectroscopy ( arpes ) .
we show that the quasiparticle - interference ( qpi ) pattern , measured by stm , shows additional 7 @xmath0 vectors associated with the scattering on the electron - pocket , than that on the hole - pocket .
furthermore , the bogolyubov quasiparticle scatterings of the electron pocket may lead to a second magnetic resonance mode in the ins spectra at a higher resonance energy . finally , we reanalyze some stm , ins , and arpes experimental data of several cuprate compounds which dictates the direct fingerprints of electron pockets in these systems .
copper - oxide high - temperature superconductors evolve from a mott insulator to the superconducting state through an unknown ` pseudogap ' phase .
many competing order origins of the ` pseudogap ' have been proposed , some of which lead to a fermi surface ( fs ) reconstruction into hole - pocket and electron - pockets.@xcite hole - pockets are detected in many experiments . on the other hand
the existence of electron pockets has been overlooked for the past twenty years and only recently has been proposed by hall - effect , quantum oscillation at high magnetic field , nernst and seebeck measurements.@xcite in particular , hall - effect measurements have revealed a negative sign in the low - temperature hall coefficients which is taken as a signature of electron - like quasiparticles on the fs.@xcite the hall - coefficient in fact changes sign from negative to positive with increasing temperature but below @xmath1 , suggesting the coexistence of both electron and hole - pockets on the fs .
shubnikov - de - haas ( sdh ) experiments in yba@xmath2cu@xmath3o@xmath4 and yba@xmath2cu@xmath5o@xmath6 ( ybco ) also argue for the presence of closed fs pockets,@xcite with slope suggestive of electron - pockets.@xcite this observation received further supports from the nernst and seebeck measurements which have been shown theoretically to be consistent with the coexistence of electron and hole - pockets.@xcite the question arises , _ if an electron pocket is present on the fs , are there spectroscopic fingerprints that can detect it directly ? _ for example , arpes which directly measures the single - particle spectral weight , has so far been unable to convincingly separate out the presence of an electron - pocket from a full paramagnetic fs .
many theoretical proposals have been put forward to explain the fs topology in cuprates,@xcite however , a consistent picture to describe both the bulk measurements and the spectroscopies has yet not been achieved . within a strong coupling scenario ,
the holes , doped into the parent mott insulator , create in - gap states at the fermi level without a well defined quasiparticle dispersion.@xcite again in the pre - formed sc pairing theory of the ` pseudogap ' , one would predict that a single large hole - like fs persists at all dopings , with sc fluctuations suppressing spectral weight in the antinodal regions , leaving a fermi arc.@xcite such a model would predict a hole - like sign of the hall coefficient at all temperatures,@xcite incompatible with the observed sign changing hall - effect@xcite and nernst and seebeck measurements.@xcite an alternative approach using a density wave picture of the pseudogap has been successful in explaining many aspects like the behavior of quantum oscillations , hall , nernst and seebeck effects,@xcite and arpes , stm and neutron scattering.@xcite of course , in ybco there are other band - structure properties that can serve as electron - like fs.@xcite to find signatures of electron pockets , we model the pseudogap as a spin - density wave ( sdw ) state which leads to the fs reconstruction into hole and electron pockets.@xcite . using this model
, we find that ( i ) the qpi pattern seen in stm exhibits 7 new @xmath7vectors which evolve in a qualitatively different way than the ones expected for a hole - pocket ; ( ii ) similarly , the ins measurements also display an additional resonance peak in the spin - excitation spectrum in the sc state coming from the electron - pocket ; ( iii ) furthermore , in some doping regions arpes fs spectral weight data reveal two peaks at the nodal and antinodal points with a dip between them which suggests reconstruction of the fs into hole and electron pockets , respectively ; ( iv ) we also demonstrate several key properties of these three spectroscopies which quantitatively and unambiguously can establish the presence of an electron - pocket on the fs . the development of electron pocket in hole doped cuprates is doping ( and material ) dependent . in the overdoped region ,
the fs consists of a large hole - like fs centered at @xmath8 . at strong underdoping , a pseudogap opens in the region of momentum space near @xmath9 and @xmath10 , leaving a hole - pocket or ` fermi arc ' at the nodal point @xmath11 .
these hole pockets are observed directly by arpes and are consistent with stm and many other experiments .
with increasing doping , as the pseudogap correlation weakens but remains finite , the bottom of the conduction band at @xmath9 and @xmath10 drops down below the fermi level producing an electron pocket at some critical doping , even without the application of any external magnetic field ( see appendix [ app1 ] for the details of the evolution of the fs ) .
a small gap persists in the regions where the bare fs crosses the magnetic brillouin zone [ marked by a dashed line in fig .
1(a ) ] .
as the electron - pockets are expected to form in the doping range where the fs crosses over from small pocket to large fs , spectroscopies need guidance to distinguish a pocket from a full hole - like fs .
therefore , we provide a careful analysis of the spectroscopic details to illustrate how to observe the electron - pocket . in the superconducting state , the @xmath12wave pairing restricts the coherent bogoliubov quasiparticles to move on the @xmath13space of the electron and hole pockets , see fig .
[ fig1](a ) .
the scattering process of these particle - hole excitations leads to many observable features , like the elastic scatterings of the cooper pairs seen as a qpi pattern in stm.@xcite similarly , inelastic scattering between particle and hole bogoliubov quasiparticles leads to a scattering profile as revealed by ins.@xcite the qpi and ins patterns generated by the hole - pocket are well studied in cuprates.@xcite here we study how these patterns evolve naturally to include contributions of the electron pocket .
* scanning tunneling microscopy@xmath14 * figures .
[ fig1](b ) and [ fig1](c ) contrast the qpi patterns at two representative quasiparticle energies at which the cooper pair resides on the hole - pocket ( lower energy ) and electron pocket ( higher energy ) , respectively .
there is a qualitative difference in the overall qpi pattern at these two energy scales .
first , since scattering is purely elastic , appearance of an electron pocket leads to new features in qpi that correspond to 7 additional electron - electron scattering @xmath15 vectors in addition to hole - hole scattering @xmath16 vectors .
no elastic scattering features connect electron and hole pockets as they have different quasiparticle energies .
the definitive distinction between the two pockets can be marked by the values of two high symmetry vectors @xmath17 and @xmath18 .
@xmath19 connects equivalent energy points on two hole - pockets along the diagonal direction . as the hole - pocket terminates at the magnetic zone boundary at which @xmath20 , therefore , if @xmath21 continues to grow above @xmath22 , it must come from the electron pocket .
similarly , @xmath23 [ along the ( 100)-direction ] will attain its maximum value equal to the reciprocal lattice vector of @xmath24 and @xmath25 at the highest energy of the qpi pattern . in addtion ,
one requires to pay attention to the energy dependence of the qpi vectors as well as their associated intensities . due to the van - hove singularity at the antinodal point as well as the discontinuous jump from the hole - pocket to the electron - pocket fs ,
one expects a ` kink ' in the energy dependence of each qpi vector , see figs .
[ fig1](d ) and [ fig1](e ) . as the @xmath16 vectors reach the top of the hole - pocket [ i.e. when @xmath26 , the bogoluybov scattering of these vectors vanishes and they become merely fs nesting .
therefore , all @xmath16 vectors shoot almost vertically upward but with diminishing intensity . nearly at the same energy , the bogolyubov scattering on the electron pocket turns on and @xmath15 vectors appear on the qpi pattern . unlike @xmath16s , @xmath15s
disperse slowly with energy but the associated intensity begins to rise again .
therefore , not only the magnitude of the @xmath15 vectors as discussed above , but also the expected ` kink ' in their dispersion and their associated intensity will serve as quantitative and unambiguous marks for the presence of electron - pockets . to demonstrate how the electron - pocket leads to a different set of qpi patterns , we calculate the qpi spectra in a coexisting uniform phase of sdw induced pseudogap and @xmath12wave superconductivity.@xcite we concentrate on ybco@xmath27 where the band dispersion is obtained by the tight - binding fitting to the first - principle calculations .
based on this ground state , the self - energy correction due to spin and charge fluctuations is computed within a self - consistent @xmath28-model [ see appendix [ app1]].@xcite the lifetime broadening due to the imaginary part of the self - energy helps create ` bright - spots ' on the constant energy single - particle spectra . at any energy in the sc state , we have 8 ` bright spots ' due to @xmath12wave symmetry as shown in figs .
[ fig2](f1 ) , [ fig2](f2 ) , [ fig2](g1 ) , [ fig2](g2 ) at four representative energy cuts below the fermi level . at @xmath29 ,
the ` bright - spots ' are concentrated at the nodal points ( not shown ) and with increasing energy , they move towards the antinodal direction .
the locus of the ` bright - spot ' is always restricted to move on the normal state fs and takes the form of well - known ` banana - shape ' in the low - energy region ( on the ` hole - pocket ' ) ; see figs .
[ fig2](f1 ) and [ fig2](f2 ) .
as the ` bright - spots ' hit the magnetic zone boundary [ green dashed line ] , they move to the electron - pocket region , see figs .
[ fig2](g1 ) and [ fig2](g2 ) .
we calculate the qpi pattern as a convolution of the constant energy single particle spectra @xmath30 $ ] , where @xmath31 is the @xmath32 single particle green s function in the sdw - sc state , see appendix [ appqpi ] . at @xmath29 , @xmath33 and @xmath34
are the same vector connecting the nodal points . with increasing @xmath35
, @xmath33 gradually shrinks whereas @xmath34 grows@xmath14both very much linearly with energy , coming from the linear dispersion of the nodal bogoliubov quasiparticles .
a similar linear dispersion is evident in the behavior of @xmath36 .
@xmath37 starts from @xmath38 at @xmath29 and increases to a maximum value less than @xmath22 in all underdoped cuprates while @xmath39 starts at a finite vector slightly below @xmath22 and reaches @xmath22 at the edge of the hole pocket .
the resulting two dimensional qpi pattern in this energy scale is shown in figs . [ fig2](d1 ) and [ fig2](d2 ) , and agrees qualitatively with the experimental results of bi2212.@xcite above this energy , all @xmath16 vectors become normal - state fs nesting , which is not related to the bogoliybov scattering , and bend backward with much less dispersion while the associated intensity gradually diminishes .
therefore , in the absence of an electron pocket , one can expect the qpi pattern to remain very much same as a function of energy but with much broadened peaks due to the lack of bogolyubov coherence peaks .
the other weak - intensities apart from the leading 7 @xmath0-vectors are associated with the shadow bands , which are not relevant for the present study .
the most interesting feature of the qpi happens above the pseudogap energy scale which separates the electron pocket from the hole pocket .
new @xmath7vectors develop due to the bogoluibov scattering of the electron pocket .
these @xmath15 vectors are practically the continuation of the @xmath16s above the magnetic zone boundary but with different slope and intensity which are related to the curvature of the electron pocket and the associated van - hove singularity .
the resulting constant energy qpi maps are shown in figs .
[ fig2](e1)-[fig2](e2 ) with very distinct interference patterns compared to the hole pocket [ compare with figs .
[ fig2](d1 ) and [ fig2](d2 ) , respectively ] . in the electron pocket regions ,
only @xmath40 disperses toward @xmath38 , whereas the others disperse away from the magnetic zone boundary to the reciprocal unit cell boundary .
@xmath41 approach each other forming a squarish profile centered at @xmath42 which is present at all energies .
also , @xmath43 approach @xmath44 and its equivalent @xmath45-points .
we emphasize that the most robust features signaling the presence of the electron pocket will be the values of @xmath46 , @xmath47 in that @xmath48 at the beginning of electron pocket whereas @xmath47 reach the zone boundary ( @xmath49 [ and its equivalent points ] at the top of the electron - pocket .
the intensity of each @xmath16 and @xmath15 vector follows closely to the density of states ( doss ) as shown in fig .
[ fig2](c ) .
both the dos and the qpi intensity grow linearly with @xmath35 , demonstrating @xmath50-wave pairing symmetry and the particle - hole symmetry in the bogoliubov quasiparticles even in the pseudogap state and also under the influence of many - body effects . above the tip of the hole - pocket , the intensity drops in the pseudogap energy region and then it rises again sharply up to the tip of the electron pocket .
experimentally the first peak in intensity is well documented for underdoped samples while some evidence of the second peak is seen in overdoped bi2212 [ see for example ref . ] .
we summarize three robust signatures that help unambiguously differentiate the presence of the electron pocket from the hole pocket or paramagnetic full fs ( see appendix [ app3 ] for details ) .
( 1 ) in the dispersion relation of the qpi vectors as shown in figs .
[ fig2](a ) and [ fig2](b ) , all the @xmath0 vectors stop dispersing at the tip of the hole - pocket . only for the case of an electron pocket ,
new qpi vectors appear which extend to @xmath44 and its equivalent points along the ( 100)-direction or above @xmath42 along the diagonal direction .
furthermore , to differentiate an electron pocket from a paramagnetic full fs , one needs to pay attention to the break in the slope of the qpi vectors going from the hole - pocket to the electron pocket .
( 2 ) for constant energy scans , the qpi profile becomes essentially energy independent above the tip of the hole - pocket ; for example , when @xmath51 stops dispersing .
in contrast , in the present case of an electron - pocket , the new qpi pattern forms with two distinguishing marks that @xmath48 and @xmath52 at the tip of the electron - pocket .
( 3 ) the intensity of the qpi vectors as a function of energy shows two distinct peaks in the case when both electron and hole pocket are present on the fs .
lower energy peak occurs at the tip of the hole pocket at an energy @xmath53 while the second peak happens at the tip of the electron pocket exactly at @xmath54 . in the absence of an electron pocket ,
only the first peak will be present whereas in a paramagnetic ground state only the second peak will show up .
* inelastic neutron scattering spectroscopy-*we turn next to the low - energy ins spectra in fig .
[ fig3 ] , mainly in the region below @xmath55 where bogoliubov quasiparticle scattering dominates in the spin - excitation dispersion.@xcite ins measures the imaginary part of the susceptibility whose non - interacting part is @xmath56 $ ] , where @xmath57 is the matsubara frequency index , see ref . .
in the sc state , @xmath58 arises from the inelastic scattering of the cooper pairs ( many body effects which are incorporated in the random - phase approximation shift the energy scale of the spectra to a slightly lower energy ; nevertheless the overall shape of the spectrum is not greatly changed ) .
therefore , the spectrum is dominated by scattering by bright spots , similar to qpi but connecting features above and below the fermi level . among 7 @xmath59 vectors in the qpi pattern discussed above only four vectors
participate in the ins spectra , see fig .
[ fig3](a ) . furthermore , owing to the selection rule associated with elastic scatterings in stm , @xmath16 and @xmath15 are always energy resolved .
but in the ins spectra the separation between the two energy scales becomes obscured due to the turning on of inter pocket inelastic scattering .
we denote the corresponding electron to hole pocket scattering channel by @xmath60 as shown by dashed lines of the same color in fig .
[ fig3](a ) .
the resulting constant energy ins profile in the sc region is sketched in fig .
[ fig3](b ) .
in addition to the energy and momentum conservation principles associated with the inelastic bogoliubov quasiparticle scattering , the coherence factors of both the superconducting state and sdw state play a major role here.@xcite the sign change of the superconducting order at the ` hot - spot ' @xmath0 , i.e. @xmath61 , is a crucial for finding non - vanishing contributions to the ins spectra .
sdw order with a modulation vector @xmath62 provides an additional coherence factor which leads to a gradual increase of intensity of the ins spectra as @xmath0 approaches @xmath63 . in the hole pocket region
, our calculation correctly reproduces the magnetic resonance peak at @xmath64 ( magenta arrows in figs .
[ fig3](e ) , [ fig3](f ) and [ fig3](g ) and both the downward and upward dispersions of the ` hour - glass ' pattern.@xcite below the resonance , the magnetic scattering of the cooper pairs also yields the maximum intensity in the bond direction , see fig .
[ fig3](c1 ) . in the absence of the electron - pocket ,
the ins intensity maxima rotate by 45@xmath65 towards the diagonal direction above the resonance energy , again consistent with the hourglass phenomenology . in the presence of the electron pocket ,
the ins pattern exhibits several distinguishing characteristics which can be separated from the usual hourglass pattern of the hole - pocket : ( 1 ) the intensity profile in the constant energy surface hosts peaks both along the bond direction as well as along the diagonal direction above the @xmath22-resonance , see figs .
[ fig3](d1 ) , [ fig3](d2 ) and [ fig3](d3 ) .
( 2 ) an additional resonance energy @xmath66 is observed along the bond direction in fig .
3(f1 ) , and in the integrated ins intensity in fig .
[ fig3](g ) .
the presence of two resonances is also theoretically calculated for iron - pnictide superconductors , although the the differences in the fs topology and the pairing symmetry between these two classes of superconductors make the details of the resonance spectra look very different.@xcite ( 3 ) more resonance branches appear in the ins spectra although weak in intensity , in figs .
[ fig3](f1 ) , [ fig3](f2 ) .
the experimental results of ybco@xmath27 shows clear evidence for the second peak as shown in figs .
[ fig3](e ) and [ fig3](g ) .
the energy scale of both the resonances are set by the sc gap amplitude as @xmath67 where @xmath68/2 $ ] is the structure factor of the @xmath69wave pairing .
we have used the arpes value of sc gap magnitude @xmath70 mev from refs .
@xcite .
the two energy scales are determined by the position of the corresponding ` hot - spot ' momentum value on the fs .
the first resonance occurs at @xmath63 where the @xmath39 vector connects the ` bright - spots ' at the tip of the hole pocket which gives @xmath71mev .
the second resonance occurs when @xmath72 touches the brillouin zone boundary which yields @xmath73mev .
note that in our calculation , the two spin resonance energies have a direct relation to the peaks in the qpi , shown in fig .
[ fig2](c ) , where the spin resonance peak occurs at twice the energy of the peak in the qpi intensity .
* angle - resolved photoemission spectroscopy- * the same information on the presence of the electron pocket can be directly obtained from arpes .
arpes measures the single particle spectral weight @xmath74 . in fig .
[ fig4 ] , we provide some evidence for the presence of the electron pockets in the arpes data . in the strongly underdoped cuprates where the pseudogap is large , it gaps out the whole antinodal region above the magnetic zone boundary .
thus electron pockets disappear from the fs and only the hole - pocket is present , as shown in fig .
[ fig4](a ) .
it is interesting to notice that even in the theoretical spectra , there is a finite incoherent spectral weight present away from the hole - pocket which traces the underlying ungapped fs .
this is the effect of the imaginary part of the self - energy correction which is calculated to be quasi - linear in the low - energy region . as a result
the residual spectral weight gradually decreases from the nodal to the antinodal regions as the pseudogap grows along the same direction but spreads over a larger energy and momentum region .
this result is consistent with experiment in a deeply underdoped sample of lsco as shown in fig .
[ fig4](c ) . therefore , in order to identify the electron pocket at the antinodal point , one needs to pay attention to the spectral weight . in the near - optimal region close to the quantum critical point
, the electron pocket appears at the fermi level , leading to coherent spectral weight at the antinodal point as shown in fig .
[ fig4](b ) .
looking at the experimental data for dopings @xmath75 of lsco in figs .
[ fig4](d)-[fig4](e ) and in an overdoped sample of tbco in fig .
[ fig4](f ) , we see that both the hole and the electron pockets are present in this doping range . especially , the spectral weight maps of the fs in figs . [ fig4](d ) and [ fig4](f ) have peaks of comparable magnitude at both nodal point and antinodal point while it is suppressed in between these two points . in the case of an ungapped full fs , the spectral weight is expected to be coherent and similar at each fermi momentum , whereas as discussed above , when only a hole pocket is present the spectral weight gradually decreases from the nodal point to the antinodal point . therefore , the experimental results in figs . [ fig4](d)-[fig4](f ) convincingly establish the presence of the electron pocket in the vicinity of optimal doping for these two materials . the procedure of inverting the qpi data to reconstruct the single - particle fs is well known@xcite and following the conventional procedure , we find that the fs constructed from the experimental data of qpi maps used in fig . [ fig2 ] lies reasonably on top of the theoretical data .
note that the existing experimental data has not yet been analyzed with the notion to identify the electron pocket .
similarly , we extract the fs from the ins data of ybco shown in fig .
[ fig3 ] and the result agrees well with the picture of coexisting hole and electron pockets as shown in fig .
[ fig4](b ) .
finally we comment on the difficulties of arpes to observe the electron pocket .
as mentioned earlier , electron - pockets are expected near the quantum critical point of the pseudogap at which the sc gap still survives . in this doping region
, a typical phase diagram shows that the pseudogap transition temperature @xmath76 for most of the cuprates.@xcite furthermore , in the sc state , the electron pockets , being positioned at the antinodal point are fully gapped by @xmath12wave superconductivity and therefore , arpes can not detect it . when temperature is increased above @xmath77 to close the sc gap , the small pseudogaps are also nearly closed , so the hole pocket and electron pocket disappears , and a full metallic fs forms . on the other hand ,
quantum oscillations are performed in high magnetic fields at which superconductivity is suppressed , and the electron pocket becomes exposed .
additional complications can arise since arpes and stm are sensitive to the surface states as well as to so - called ` matrix - element ' effects , which could also explain failure to see certain portions of the fs at particular experimental conditions . in conclusion , we present the detailed spectroscopic analysis of the electron pocket that will allow both single particle ( arpes ) and two particle spectroscopies ( stm and ins ) to detect electron pockets that are posited to be present near optimal doping .
these simple qualitative features provide a sharp contrast to a simple hole pocket models and hence offer a direct test of their presence .
the simplest model that has electron pockets is the sdw state with or without coexisting superconducting order .
even with this simplified model we find significant spectroscopic features that allow qualitative and quantitative determination of the electron pockets in cuprates . |
a reasonable expression would be as follows .
let @xmath18 be nonnegative , continuous , differential and strictly monotone on [ a , b ] ( @xmath3 ) , it s inverse function is @xmath1 , @xmath19 $ ] .
then , @xmath20 ^ 2dy}=\mbox{sgn}(f(b)-f(a))\left\{{\pi[{b^2f(b)-a^2f(a)}]-2\pi\int_a^b { xf(x)dx}}\right\}\ ] ] @xmath21 ^ 2dx}=\mbox{sgn}(g(d)-g(c))\left\{{\pi[{d^2g(d)-c^2g(c)}]-2\pi\int_c^d { yg(y)dy}}\right\}\ ] ] where @xmath22 is the sign function .
\(i ) let @xmath18 be strictly monotone increasing(fig1),then @xmath15 , @xmath16 , @xmath23 , and @xmath24 ^ 2dy}=\pi \int_a^b { x^2df(x)}=\pi \left[{x^2f(x)\vert_a^b - \int_a^b { f(x)dx^2}}\right ] \\ & = \displaystyle \pi \left [ { b^2f(b ) -a^2f(a ) } \right ] - 2\pi \int_a^b { xf(x)dx } \\ \end{array}\ ] ] is strictly monotone decreasing.,title="fig:",width=226,height=151 ] [ fig1 ] is strictly monotone decreasing.,title="fig:",width=226,height=151 ] [ fig2 ] \(ii ) let @xmath18 be strictly monotone decreasing ( fig 2 ) , then @xmath25 , @xmath13 , ( @xmath23 ) , and @xmath26 ^ 2dy}=\pi \int_b^a { x^2df(x)}=-\pi \int_a^b{x^2df(x)}\\ & = \displaystyle -\pi\left[{x^2f(x)\vert_a^b -\int_a^b { f(x)dx^2 } } \right]\\ & = \displaystyle 2\pi \int_a^b { xf(x)dx } - \pi \left[{b^2f(b ) - a^2f(a ) } \right ] \\ \end{array}\ ] ] similarly , we can prove the other two cases of volume of solids of revolution with integration by parts as follows .
\(iii ) if @xmath1 is strictly monotone increasing ( fig 1 ) , then @xmath27 , @xmath28 , ( @xmath29 ) , and @xmath30 ^ 2dx}=\pi \int_c^d { y^2dg(y)}=\pi \left[{y^2g(y)\vert_c^d-\int_c^d { g(y)dy^2 } } \right ] \\ & = \displaystyle \pi \left [ { d^2g(d ) -c^2g(c ) } \right]-2\pi \int_c^d { yg(y)dy } \\
\end{array}\ ] ] \(iv ) if @xmath1 is strictly monotone decreasing ( fig 2 ) , then @xmath31 , @xmath32 , ( @xmath29 ) , and @xmath33 ^ 2dx}=\pi \int_d^c { y^2dg(y)}=- \pi \int_c^d { y^2dg(y)}\\ & = \displaystyle -\pi \left [ { y^2g(y)\vert_c^d -\int_c^d { g(y)dy^2}}\right ] \\ & = \displaystyle 2\pi \int_c^d { yg(y)dy } - \pi \left[{d^2g(d ) - c^2g(c ) } \right ] \\ \end{array}\ ] ] hence , this ends the proof
. if @xmath18 is continuous and not strictly monotone on @xmath34 $ ] , and @xmath34 $ ] can be divided into several strictly monotone intervals of @xmath0 , we have the following lemma
. * lemma 1 . *
suppose that @xmath18is continuous and piecewise strictly monotone on @xmath34 $ ] , if @xmath35 , @xmath36 , @xmath37 , and @xmath0 is not strictly monotone on @xmath34 $ ] , the number of local extreme value points in @xmath38 is even .
we prove the lemma in two cases , @xmath39 , and @xmath40 . in case of @xmath39 , since @xmath41 , @xmath37 , the nearest extreme value point to @xmath8 is a local maximum value point . since @xmath42 , @xmath37 , the nearest extreme value point to @xmath9 is a local minimum value point .
as @xmath18 is continuous and piecewise strictly monotone on @xmath34 $ ] , the number of extreme value points should be even . or
, using method of reduction to absurdity , if the number of extreme values was odd , the last extreme value point ( nearest extreme value point to @xmath9 ) might be a local maximum extreme value point , it is a contradiction .
similarly , we can prove the other case .
hence , we complete the proof .
then , the volume of curvilinear trapezoidal composed by curved edge @xmath18 in term of y - axis , and rotating with y - axis can reach the following conclusion .
let @xmath18 be continuous , piecewise strictly monotone and differential on @xmath34 $ ] .
@xmath43is the curve determined by @xmath18 on @xmath34 $ ] .
denote @xmath44 , @xmath45 .
suppose that both @xmath46 and @xmath47 @xmath48 intersect @xmath18 in only one point on @xmath34 $ ] respectively . then the volume of the plane graph along @xmath49 , @xmath46 , @xmath50 , @xmath47 rotating with @xmath51-axis is @xmath52 - 2\pi\int_a^b { xf(x)dx}}\right\}\ ] ] according to lemma 1 , since @xmath18 is continuous , and strictly piecewise monotone on @xmath34 $ ] , assume that the strictly piecewise monotone intervals of @xmath18 are @xmath53 $ ] , @xmath54 , @xmath55 , @xmath56 , @xmath57 , and @xmath34=\bigcup\limits_{i= 0}^n { [ x_i , x_{i + 1 } ] } $ ] , where @xmath58 are extreme value points of @xmath11 .
is continuous and piecewise strictly monotone with @xmath59.,title="fig:",width=226,height=151 ] [ fig3 ] is continuous and piecewise strictly monotone with @xmath59.,title="fig:",width=226,height=151 ] [ fig4 ] \(1 ) if @xmath60 ( fig3 ) , as both @xmath46 and @xmath47 @xmath48 intersect @xmath61 only once on @xmath34 $ ] respectively , @xmath18 is strictly increasing on both @xmath62 $ ] and @xmath63 $ ] , where @xmath64 is even ( if @xmath65 , theorem 2 degrades to theorem 1 ) and @xmath15 , @xmath16 . denote @xmath66 $ ] , and @xmath67 is the volume of curved trapezoid @xmath68 with @xmath51 -axis rotating with @xmath51-axis .
since @xmath18 is strictly monotone and differential on @xmath69 $ ] , theorem 1 would be employed in the volume of solids of revolution on each curvilinear trapezoidal determined by piecewise strictly monotone and differential curve . then , @xmath70 - 2\pi \int_{x_{2k } }
^{x_{2k + 1}}{xf(x)dx}\\ v_{\gamma_{2k+1 } } & = \displaystyle 2\pi \int_{x_{2k+1 } } ^{x_{2k+2}}{xf(x)dx}-\pi [ x_{2k+2}^2 f(x_{2k+2 } ) -x_{2k+1}^2 f(x_{2k+1 } ) ] \\ & \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad k=0,1,\cdots , n/2 . \\ \end{array}\ ] ] using mathematical induction , we can obtain , @xmath71 - 2\pi\int_{x_i } ^{x_{i + 1}}{xf(x)dx}}\right\ } } \\ & = \displaystyle \pi [ x_{n + 1}^2 f(x_{n+1 } ) - x_0 ^ 2 f(x_0 ) ] - 2\pi \int_{x_0 } ^{x_{n+1 } } { xf(x)dx } \\ & = \displaystyle \pi [ b^2f(b ) -a^2f(a ) ] - 2\pi \int_a^b { xf(x)dx } \\
\end{array}\ ] ] \(2 ) if @xmath40 ( fig4 ) , as both @xmath46 and @xmath47 @xmath48 intersect @xmath18 one point on @xmath34 $ ] respectively .
then @xmath18 is strictly decreasing on @xmath72 $ ] and @xmath73 $ ] , where @xmath64 is even , and @xmath14 , @xmath13 .
denote @xmath74$],and @xmath67 is the volume of curvilinear trapezoid of @xmath68 in term of @xmath51-axis rotating with @xmath51-axis .
then , @xmath75\\ v_{\gamma_{2k+1 } } & = \displaystyle \pi [ x_{2k+2}^2 f(x_{2k+2})-x_{2k+1}^2 f(x_{2k+1 } ) ] -2\pi \int_{x_{2k+1 } } ^{x_{2k+2}}{xf(x)dx}\\ & \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad k=0,1,\cdots , n/2 .
\\ \end{array}\ ] ] using mathematical induction , we can obtain , @xmath76 } \right\ } } \\ & = \displaystyle 2\pi \int_{x_0}^{x_{n+1}}{xf(x)dx } - \pi [ x_{n+1}^2 f(x_{n+1 } ) - x_0 ^ 2 f(x_0 ) ] \\ & = \displaystyle 2\pi \int_a^b { xf(x)dx } - \pi [ b^2f(b ) - a^2f(a ) ] \\ \end{array}\ ] ] to sum up formulae(9)(10 ) , the theorem is completely proved .
similarly , we can obtain the following theorem : let @xmath1 be continuous , piecewise strictly monotone and differential on @xmath77 $ ] .
@xmath43 is the curve determined by @xmath78 on @xmath77 $ ] .
denote @xmath79 , @xmath80 .
suppose that both @xmath8 and @xmath9 @xmath81 intersect @xmath1 in only one point on @xmath77 $ ] respectively .
then the volume of the plane graph along @xmath7 , @xmath8 , @xmath43 and @xmath9 rotating with @xmath82-axis is @xmath83 - 2\pi \int_c^d { yg(y)dy}}\right\}\ ] ]
* example 1 * for kepler equation [ 3 ] : @xmath84 , @xmath85 , @xmath86 $ ] .
( fig 5 ) ( i ) to calculate the volume of solids of revolution of curved trapezoid composed by kepler curve , @xmath87 and @xmath51-axis rotating with @xmath51-axis . ( ii ) to calculate the volume of solids of revolution of curved trapezoid composed by kepler curve , @xmath88 and @xmath82-axis rotating with @xmath82-axis .
* solution * ( i ) for @xmath89,@xmath85 , @xmath90 ^ 2dy}=\pi \int_0^{2\pi}{[y -\varepsilon \sin y]^2dy } \\ & = \displaystyle [ \frac{y^3}{3 } + 2\varepsilon ( y\cos y - \sin y ) + \varepsilon ^2(\frac{y}{2 } - \frac{\sin2y}{4})]\vert_0^{2\pi}\\ & = \displaystyle \frac{8}{3}\pi ^3 + ( 4\varepsilon + \varepsilon ^2)\pi \\ \end{array}\ ] ] \(ii ) since @xmath91 , the curve determined by kepler function @xmath89 is strictly increasing in term of @xmath51 on @xmath92 $ ] . according to inverse function theorem of monotone continuous function , it s inverse function @xmath18 exists in term of @xmath82 on @xmath92 $ ] and is strictly increasing . then , the function @xmath18 determined by kepler function is an implicit function and also transcendental equation . using formula(3 ) ,
@xmath93 , @xmath94 , @xmath95 , @xmath96 , we obtain @xmath97 ^ 2dx}=8\pi ^4 - 2\pi \int_0^{2\pi}{yg(y)dy } \\ & = \displaystyle 8\pi ^4 - 2\pi \int_0^{2\pi}{y(y - \varepsilon \sin y)dy } \\ & = \displaystyle 8\pi ^4 - 2\pi [ \frac{y^3}{3 } + \varepsilon (
y\cos y - \sin y)]\vert_0^{2\pi}\\ & = \displaystyle \frac{8}{3}\pi ^4 + 4\varepsilon \pi ^2 \\ \end{array}\ ] ] the volumes of curved trapezoids of @xmath108 in term of @xmath51-axis rotating with @xmath51-axis are denoted as @xmath109,@xmath110,@xmath111 , @xmath112,@xmath113,and the volume of curved trapezoid @xmath18 ( @xmath114 ) , @xmath7 , @xmath115 , @xmath49 rotating with @xmath51-axis is @xmath116.then , @xmath117 according to theorem 1 , @xmath118\vert_0^{x_1}\\ & = \displaystyle \pi x_1 ^2f(x_1 ) - 2\pi [ \frac{x_1 ^3}{3\pi}+ ( \sin x_1 - x_1 \cos x_1 ) ] \\ \end{array}\ ] ] @xmath119 \\ & = \displaystyle 2\pi [ \frac{x^3}{3\pi}+(\sin x - x\cos x)]\vert_{x_1 } ^{x_2}- \pi [ x_2 ^ 2 f(x_2 ) - x_1 ^ 2 f(x_1 ) ] \\ & = \displaystyle 2\pi [ \frac{x_2 ^3}{3\pi } + ( \sin x_2 - x_2 \cos x_2 ) ] - 2\pi [ \frac{x_1 ^3}{3\pi}+ ( \sin x_1 - x_1
\cos x_1 ) ] - \pi [ x_2 ^ 2 f(x_2 ) - x_1 ^ 2 f(x_1 ) ] \\ \end{array}\ ] ] @xmath120 - 2\pi \int_{x_2 } ^{x_3}{x(\frac{x}{\pi}+ \sin x)dx } \\ & = \displaystyle \pi [ x_3 ^ 2 f(x_3 ) - x_2 ^ 2 f(x_2 ) ] -2\pi [ \frac{x^3}{3\pi}+ ( \sin x - x\cos x)]\vert_{x_2 } ^{x_3}\\ & = \displaystyle \pi [ x_3 ^ 2 f(x_3 ) - x_2 ^ 2 f(x_2 ) ] - 2\pi [ \frac{x_3 ^3}{3\pi}+ ( \sin x_3 - x_3 \cos x_3 ) ] + 2\pi [ \frac{x_2 ^ 3}{3\pi}+ ( \sin x_2- x_2 \cos x_2 ) ] \\ \end{array}\ ] ] substituting @xmath113,@xmath111,@xmath112,@xmath121 into formula ( 12 ) , we obtain @xmath122 \\ & = \displaystyle 8\pi ^3 - 2\pi[\frac{8\pi ^2}{3 } - 2\pi ] = \displaystyle \frac{8\pi ^3}{3 } + 4\pi ^2 \\
\end{array}\ ] ] * solution 2 * : using formula ( 8) , the volume of solids of revolution is @xmath123 - 2\pi\int_a^b { xf(x)dx}}\right\}\\ & = \displaystyle \left\ { { \pi \left [ { ( 2\pi ) ^2f(2\pi ) - 0 } \right ] - 2\pi \int_0^{2\pi } { x(\frac{x}{\pi}+ \sin x)dx}}\right\ } \\ & = \displaystyle 8\pi ^3 - 2\pi [ \frac{x^3}{\pi}+ \sin x - x\cos x]\vert_0^{2\pi}\\ & = \displaystyle 8\pi ^3 - 2\pi [ \frac{8\pi ^2}{3 } - 2\pi ] = \frac{8\pi ^3}{3 } + 4\pi ^2 \\
\end{array}\ ] ] 1 j. stewart . _ calculus _ ( seventh edition ) .
brooks / cole , cengage learning . 2012 .
343 - 494 .
v. a. zorich .
_ mathematical analysis i _ ( 4th edition ) .
springer - verlag berlin heidelberg .
. a. hall .
kepler s problem .
_ annals of mathematics_.*10 * ( 3 ) ( 1883 ) 6566 . | a calculation formula of volume of revolution with integration by parts of definite integral is derived based on monotone function , and extended to a general case that curved trapezoids is determined by continuous , piecewise strictly monotone and differential function . and
, two examples are given , ones curvilinear trapezoids is determined by kepler equation , and the other curvilinear trapezoids is a function transmuted from kepler equation .
school of mathematics and statistics , beijing institute of technology , beijing 100081,p.r china + email : [email protected] .
school of mathematics and system sciences , beihang university , beijing , 100083,p.r china + email : [email protected] . in [ 1 ] ( p494,ex71 ) , if @xmath0 is one - to - one and therefore has an inverse function @xmath1 , where @xmath2$],@xmath3 and @xmath4 $ ] .
let s be the solid obtained by rotating about the y - axis the region bounded by @xmath5 [ where @xmath6 , @xmath7 , @xmath8 and @xmath9 ( which is also called curvilinear trapezoid [ 2 ] ) .
using cylindrical shells and slicing method , the volume of the given solid is @xmath10 ^ 2dy}\ ] ] this formula can also be proved by integration by parts with the increasing case of @xmath11 .
however , there is slight divarication in formula ( 1 ) . 1 .
if we guarantee the condition @xmath12 $ ] , formula ( 1 ) does not hold in case of decreasing @xmath11 , as under this condition @xmath13 , @xmath14 .
2 . if we ensure that the condition @xmath15 , @xmath16 is established regardless of increasing or decreasing of @xmath11 , the condition @xmath4 $ ] does not hold in case of decreasing of @xmath11 , as @xmath17 .
however , formula ( 1 ) still holds in mathematical style . |
over the past 82 years , the study of dipole moments of elementary particles has provided a wealth of information on subatomic physics . from the pioneering work of stern@xcite through the discovery of the large anomalous magnetic moments of the proton@xcite and neutron@xcite
, the ground work was laid for the discovery of spin , of radiative corrections and the renormalizable theory of qed , of the quark structure of baryons and the development of qcd .
a charged particle with spin @xmath1 has a magnetic moment @xmath2 where @xmath3 is called the gyromagnetic ratio .
the expression in the middle is the quantity one finds listed in the particle data tables.@xcite the quantity @xmath4 is the anomalous magnetic dipole moment ( or simply the anomaly ) which is related to the @xmath5-factor in the right - hand equation .
the dirac equation tells us that @xmath6 for spin angular momentum , and is unity for orbital angular momentum ( the latter having been verified experimentally@xcite ) .
this can be seen from the non - relativistic reduction for an electron in a weak magnetic field : @xmath7 \psi , \ ] ] and the subscript on @xmath5 is dropped in the following discussion . for point particles ,
the anomaly arises from radiative corrections , three examples of which are shown in fig .
[ fg : radcor ] .
the `` vertex '' correction and vacuum polarization are important in the lepton anomaly , while the self - energy term is included in the dressed mass .
the situation for baryons is quite different , since their internal quark structure gives them large anomalies .
the vertex correction in lowest - order gives the famous schwinger@xcite result , @xmath8 , which was verified experimentally by foley and kusch.@xcite to lowest order , the s - matrix element@xcite for a charged particle in a magnetic field is given by @xmath9 u(p ) a^{\lambda}(q)\ ] ] in general @xmath4 ( or @xmath5 ) is an expansion in @xmath10 , @xmath11 with 1 diagram for the schwinger ( second - order ) contribution , 5 for the fourth order , 40 for the sixth order , 891 for the eighth order .
the qed contributions to electron and muon 2 have now been calculated through eighth order , @xmath12 and the the tenth - order contribution has been estimated.@xcite the first few orders are shown schematically below in fig . [
fg : aexpan ] . .
] while magnetic dipole moments ( mdms ) are a natural property of charged particles with spin , electric dipole moments ( edms ) are forbidden both by parity and by time reversal symmetry .
interestingly enough , purcell and ramsey@xcite suggested in 1950 that a measurement of the neutron edm would be a good way to search for parity violation , well in advance of the paper by lee and yang . after the discovery of parity violation , landau@xcite and ramsey@xcite pointed out that an edm would violate both _
t _ symmetries .
this can be seen by examining the hamiltonian for a spin one - half particle in the presence of both an electric and magnetic field , @xmath13 .
the transformation properties of @xmath14 , @xmath15 , @xmath16 and @xmath17 are given in the table [ tb : tranprop ] , and we see that while @xmath18 is even under all three , @xmath19 is odd under both _
_ t_. thus the existence of an edm implies that both _ p _ and _ t _ are violated . in the context of _ cpt _ symmetry ,
an edm implies _ cp _ violation .
the standard model value for the electron and muon edm is @xmath20 @xmath21-cm , well beyond the reach of experiments ( which are at the @xmath22 @xmath21-cm level ) .
observation of a non - zero @xmath21 or @xmath23 edm would be a clear signal for new physics .
.transformation properties of the magnetic and electric fields and dipole moments . [ cols="^,^,^,^ " , ] a new experiment to search for a permanent edm of the muon with a design sensitivity of @xmath24 @xmath21-cm is now being planned for j - parc.@xcite this sensitivity lies well within values predicted by susy models.@xcite feng , et al.,@xcite have calculated the range of @xmath25 available to such an experiment , which is shown in fig .
[ fg : phicp ] .
available to a dedicated muon edm experiment.@xcite the two bands show the one and two standard - deviation ranges if @xmath26 differs from the standard model value by @xmath27 . ] with an edm present , the spin precession relative to the momentum is given by @xmath28 + { e \over m}\left [ { \eta \over 2 } \left ( { \vec e \over c } + \vec \beta \times \vec b \right ) \right ] \label{eq : omegawedm}\ ] ] where @xmath29 and @xmath30 . for reasonable values of @xmath31 , the motional electric field
@xmath32 is much larger than electric fields that can be obtained in the laboratory , and the two vector frequencies are orthogonal to each other .
the edm has two effects on the precession : the magnitude of the observed frequency is increased , and the precession plane is tipped relative to the magnetic field .
e821 was operated at the magic @xmath33 so that the focusing electric field did not cause a spin precession .
the edm signal in e821 is very difficult to observe , since the tipping of the precession plane is very small ( @xmath34 mrad ) .
the dedicated experiment will be operated at 500 mev / c , off of the magic @xmath33 , and will use a radial electric field to stop the @xmath0 precession . thus the edm would cause a steady build - up of the spin out of the plane with time .
detectors would be placed above and below the storage region , and a time - dependent up - down asymmetry would be the signal of an edm .
measurements of the muon and electron anomalies played an important role in our understanding of sub - atomic physics in the 20th century .
the electron anomaly was tied closely to the development of qed .
the subsequent measurement of the muon anomaly showed that the muon was indeed a `` heavy electron '' which obeyed qed.@xcite with the sub - ppm accuracy now available for the muon anomaly,@xcite there may be indications that new physics is beginning to appear .
marciano@xcite has pointed out that using a few very well measured standard model parameters , rather than a global fit to all electroweak measurements , predicts a higgs mass which is much smaller than the present experimental limit from lep .
if one argues that any discrepancy between the standard model value of @xmath26 and the experimental one is an indication that the hadronic contribution has been underestimated , and uses this discrepancy to `` determine the hadronic contribution '' , the higgs mass limit gets even smaller .
marciano concludes that `` hints of ` new physics ' may be starting to appear in quantum loop effects . ''
the non - observation of an electron edm is becoming an issue for supersymmetry , just as the non - observation of a neutron edm implies such a mysteriously ( some would say un - naturally ) small @xmath35-parameter for qcd .
the search for edms will continue , and if one is observed , the motivation for further searches in other systems will be even stronger .
the muon presents a unique opportunity to observe an edm in a second - generation particle , where the _ cp _ phase might be different from the first generation , or the scaling with mass might be quadratic rather than linear .
it is clear that the study of lepton moments ( and neutron edm searches ) will continue to be a topic of great importance in the first part of the 21st century .
both the theoretical and experimental situations are evolving .
stay tuned for further developments .
i wish to thank my colleagues on the muon 2 experiment , as well as m. davier , e. de rafael , g. gabrielse , w. marciano , and b. odum for helpful discussions .
special thanks go to r. carey and d. hertzog for their helpful comments on this manuscript .
m. davier , s. eidelman , a. hcker , and z. zhang , eur .
phys j. * c 27 * 497 ( 2003 ) .
see k. hagiwara , a.d .
martin , d. nomura , and t. teubner phys . lett . *
b557 * , 69 ( 2003 ) for an independent analysis of the @xmath36 data .
marc knecht , andreas nyffeler , phys .
* d65 * , 073034 ( 2002 ) ; m. knecht , a. nyffeler , m. perrottet , e. de rafael , phys .
88 * , 071802 ( 2002 ) , j. bijnens , e. pallante , and j. prades , nucl .
* b626 * , 410 ( 2002 ) ; m. hayakawa and t. kinoshita hep - ph/0112102 , ( 2002 ) , i. blokland , a. czarnecki and k. melinkov phys .
88 * , 071803 ( 2002 ) , and references therin . | from the famous experiments of stern and gerlach to the present , measurements of magnetic dipole moments , and searches for electric dipole moments of `` elementary '' particles have played a major role in our understanding of sub - atomic physics . in this
talk i discuss the progress on measurements and theory of the magnetic dipole moments of the electron and muon .
i also discuss a new proposal to search for a permanent electric dipole moment ( edm ) of the muon and put it into the more general context of other edm searches .
2@xmath0 [email protected] + department of physics + boston university + boston , ma 02215 usa |
multiferroics are used in various applications for energy production , transmission of high voltage lines , data storage devices , and sensors @xcite . they will help to replace a number of currently used lead - based materials which contain lead toxic and harmful to the environment .
one of the most promising candidates for applications , bismuth ferrite bifeo@xmath0 , is actively studied because of the coupling between ferroelectric and magnetic order around room temperature in this compound .
recent investigations of bifeo@xmath0 at high pressures up to 60 gpa reveal metal - insulator @xcite and high - spin to low - spin ( hs - ls ) transitions @xcite in fe@xmath4 at room temperature in a relatively wide pressure range 40 - 55 gpa accomplished by the structural phase transition @xcite .
the problem becomes even more complicated taking into account that bifeo@xmath0 has a rich phase diagram . at ambient pressure and up to 1100 k
it has rhombohedral ( @xmath1 ) crystal structure@xcite .
increasing temperature and pressure , the structure of bifeo@xmath0 transforms into orthorhombic @xmath2 @xcite and then cubic @xmath3 @xcite . for many years
, metal - insulator transition ( mit ) in @xmath5 or @xmath6 metal compounds @xcite is one of the central issues in condensed matter physics .
the most spectacular examples are pressure - driven transitions from a wide gap mott insulator to metallic state in transition metal oxides . for mno and fe@xmath7o@xmath0 ( @xmath8 configuration ) ,
the metal - insulator transition is accompanied by the high - spin to low - spin ( hs - ls ) transition .
recently , these and other mits and spin transitions were successfully described theoretically @xcite employing the method combining density functional approximations ( like , gga or lda ) with dynamical mean - field theory @xcite .
pressure - driven mit correlated with magnetic collapse could be treated as a delocalization of magnetic electrons or structural phase transition into new phase with nel point below room temperature @xcite .
another mott - type mechanism controlled by dramatic @xmath9 decrease due to hs - ls transition was proposed in ref.@xcite .
mott - type mit driven by a broadening of the @xmath10 states was confirmed by the lda+@xmath11 calculations which revealed the hs - ls transition at 36 gpa and a transition to the metallic phase with no localized moment above 72 gpa @xcite . in this paper
we investigate the properties of bifeo@xmath0 under pressure employing the method accounting for dynamical electronic correlations and considering three experimental crystal structures .
the gga+dmft method @xcite is realized in a computational scheme constructed in the following way : first , a hamiltonian @xmath12 is produced using converged gga results for a compound under investigation , then the many - body hamiltonian is set up , and finally the corresponding self - consistent dmft equations are solved . in this work
the hamiltonians @xmath12 are constructed in a wannier function ( wf ) basis @xcite using the projection procedure described in detail in ref . .
initial ab - initio calculations of the electronic structure are done within the pseudo - potential plane - wave method , as implemented in quantum espresso @xcite .
the wfs are defined by the choice of bloch functions hilbert space and by a set of trial localized orbitals that will be projected on these bloch functions .
the basis set includes all bands that are formed by o 2@xmath13 and fe 3@xmath5 states and correspondingly the full set of the o 2@xmath13 and fe 3@xmath5 atomic orbitals to be projected on bloch functions for these bands .
that would correspond to the extended model where in addition to the @xmath5 orbitals all @xmath13 orbitals are also included .
the resulting @xmath14 hamiltonian to be solved by dmft has the form @xmath15 where @xmath16 is the coulomb interaction matrix , @xmath17 is the occupation number operator for the @xmath5 electrons with orbitals @xmath18 or @xmath19 and spin indices @xmath20 or @xmath21 on the @xmath22-th site .
the term @xmath23 stands for the @xmath5-@xmath5 interaction already accounted for in gga , so called double - counting correction . in the present calculation
the double - counting was chosen in the following form @xmath24 . here
@xmath25 is the self - consistent total number of @xmath5 electrons obtained within gga+dmft , @xmath26 is the average coulomb parameter for the @xmath5 states , and @xmath27 is unit operator .
the elements of @xmath28 matrix are parameterized by @xmath11 and @xmath29 according to the procedure described in @xcite .
the values of coulomb repulsion parameter @xmath11 and hund exchange parameter @xmath29 were calculated by the constrained lda method @xcite on wannier functions @xcite .
the values @xmath11=6 ev and @xmath29=0.93 ev obtained in these calculations are close to the previous estimations @xcite .
the effective impurity problem in dmft was solved by the hybridization expansion continuous - time quantum monte - carlo method ( ct - qmc ) @xcite .
the calculations for all volumes were performed in the paramagnetic state for inverse temperature @xmath30=15 ev@xmath31 corresponding to 770 k. according to the phase diagram , no long - range ordering in bifeo@xmath0 was found in this temperature region .
the spectral functions on real energies were calculated employing maximum entropy method ( mem)@xcite .
in all structures under investigation iron is surrounded by oxygen ions forming an octahedron . in the high pressure
cubic @xmath3 phase , the fe @xmath5 band is split by crystal field into threefold degenerate @xmath10 and twofold degenerate @xmath32 sub - bands .
a trigonal distortion in the @xmath1 structure lowers the point symmetry group @xmath33 to @xmath34 , so that threefold degenerate @xmath10 becomes split into twofold degenerate @xmath35 and non - degenerate @xmath36 band .
the @xmath2 structure has the lowest symmetry and all orbitals become nonequivalent but still three groups of orbitals could be considered : @xmath37 and @xmath38 , @xmath39 and @xmath40 and @xmath36 . for simplicity
we will use cubic @xmath10 and @xmath32 orbitals notations for analysis hereafter .
previous calculations demonstrated that lda fails to describe an insulating character of the bifeo@xmath0 ground state at ambient pressure and for all volumes it is metallic in all structures .
however , a gap of the afm origin was obtained in the gga calculation made for several magnetic structures @xcite .
states for different volumes related to the ambient pressure volume @xmath41 obtained in the gga+dmft ( ct - qmc ) calculations for the low pressure @xmath1 phase at 770 k.,scaledwidth=42.5% ] our gga+dmft calculations produced for the @xmath1 phase of bifeo@xmath0 show that taking into account coulomb correlation effects results a wide - gap mott insulator and high - spin state for ambient pressure in agreement with the experimental data .
the obtained spectral functions for different cell volumes are shown in fig .
[ fig : r3c_sf ] .
the calculated energy gap value of about 1.2 ev agrees well with the experimental value of 2.4 ev for direct optical gap @xcite and 1.8 ev indirect optical gap at ap @xcite at room temperature and indirect gap value 1.3 ev measured in thin films @xcite , 1.5 ev at 820k @xcite .
the magnitude of magnetic moment @xmath42 is 4.6@xmath43 at ap .
this number agrees very well with the high - spin state of the fe@xmath44 ion ( d@xmath45 configuration ) in cubic crystal field : 2 electrons in the @xmath46 band and 3 electrons in the @xmath10 band with the magnetic moment value of 5 @xmath43 .
then the magnetic moment decreases and at 0.76@xmath41 drops down to 2.5 @xmath43 , and it equals to 2.0 @xmath43 when the volume is 0.72@xmath41 ( 63 gpa ) .
the later value is close to the one expected for the ls state with 5 electrons in the @xmath10 band but strong hybridization between iron and oxygen leads to the non zero occupation of the @xmath32 orbitals .
this mechanism was obtained in all three crystal structures and is considered in detail for the @xmath2 case .
phase at 770k.,scaledwidth=42.5% ] the most interesting phase is @xmath2 since both the hs - ls and metal - insulator transitions occur simultaneously with the @xmath2 to @xmath3 phase transition .
the gga+dmft spectral functions for different cell volumes are shown in fig .
[ fig : pbnm_sf ] .
note , that the @xmath2 phase does nt exist to 2 gpa at 770 k. but bifeo@xmath0 in the @xmath2 structure and experimental with cell volume of @xmath1 at ambient pressure is an insulator with a gap of @xmath471 ev which is close to the experimental data @xcite .
when the cell volume is 0.85v@xmath48 the gap is closed but iron is still in the hs state ( see fig .
[ fig : mom ] ) that contradicts to the mechanism proposed in ref . .
then together with the volume decrease the @xmath10 occupancy increases and both magnetic moment and a number of @xmath46 electrons shrink . both transitions do nt appear to happen instantly but have a crossover region 0.85 - 0.82v@xmath48 that agrees with the experimental observations @xcite .
states ( black circles ) , @xmath10 ( blue diamonds ) , and @xmath32 ( green triangles ) states obtained in the gga+dmft ( ct - qmc ) calculations for the @xmath2 phase .
, scaledwidth=42.5% ] mit at room temperature where long range order exists occurs in the pressure range 45 - 55 gpa . in order to reproduce the temperature dependence of critical pressure value the paramagnetic gga+dmft calculation for @xmath19=40ev@xmath31 ev was carried out .
the results obtained show that at 33 gpa bifeo@xmath0 is already in the metallic ls state as well as at higher temperature .
hence increase of the critical pressure is nt a temperature - driven effect .
then we have carried out magnetic gga+dmft calculation for the same cell volumes of @xmath2 structures .
small external magnetic field 0.01 ev was applied to each fe atom so that direction of the field on sites corresponds to experimentally observed g - type afm .
the gap value increases slightly but more important magnetic solution was stabilized and mit occurs between 0.82 - 0.8 v@xmath48 that corresponds to 33 - 43 gpa .
this value is close to transition pressure 45 - 55 gpa measured at 300 k in afm phase . .
experimental data taken from from ref . are shown with solid lines .
insulator and metallic solutions obtained in gga+dmft calculation are shown with black squares and blue triangles , correspondingly .
critical pressure obtained in paramagnetic calculation is 25 gpa .
spin polarized gga+dmft results at 300 k results in imt in a range 33 - 43 gpa.,scaledwidth=42.5% ] probabilities of charge and orbital configurations measured in the gga+dmft method within impurity solver are shown in fig.[fig : prob ] .
the hybridization between iron and oxygen and hence the @xmath5 states occupation increases with the volume contraction . at ambient pressure ,
the @xmath8 configuration has the larger impact but the @xmath49 probability is also sizable . with the volume reduction probabilities of @xmath49 and @xmath50 configurations
grow up and the probability of the @xmath8 configuration shrinks , see the upper panel of fig.[fig : prob ] .
more detailed picture of orbital configuration impacts is shown in fig .
[ fig : prob ] ( lower panel ) .
one can see that the probabilities of the hs configuration with any number of electrons ( @xmath8 , @xmath49 and @xmath50 ) which dominate at ambient pressure goes down and after the cell volume become smaller than 0.85v@xmath48 drops and ls and is configuration probability increase .
( upper panel ) and orbital configurations of fe ion ( lower panel ) vs. pressure obtained in the gga+dmft ( ct - qmc ) calculations for @xmath2 phase ( shown only largest ones ) : @xmath51-hs ( up : @xmath52 , down : @xmath53 ) , @xmath51-ls ( up : @xmath54 , down : @xmath55 ) , @xmath56-hs@xmath57 ( up : @xmath52 , down : @xmath58 ) , @xmath56-hs@xmath7 ( up : @xmath52 , down : @xmath59 ) , @xmath56-is ( up : @xmath60 , down : @xmath55 ) , @xmath56-ls ( up : @xmath54 , down : @xmath61 ) , @xmath62-hs ( up : @xmath52 , down : @xmath63 ) , @xmath62-is@xmath57 ( up : @xmath60 , down : @xmath54 ) , @xmath62-is@xmath7 ( up : @xmath60 , down : @xmath61).,title="fig:",scaledwidth=42.5% ] ( upper panel ) and orbital configurations of fe ion ( lower panel ) vs. pressure obtained in the gga+dmft ( ct - qmc ) calculations for @xmath2 phase ( shown only largest ones ) : @xmath51-hs ( up : @xmath52 , down : @xmath53 ) , @xmath51-ls ( up : @xmath54 , down : @xmath55 ) , @xmath56-hs@xmath57 ( up : @xmath52 , down : @xmath58 ) , @xmath56-hs@xmath7 ( up : @xmath52 , down : @xmath59 ) , @xmath56-is ( up : @xmath60 , down : @xmath55 ) , @xmath56-ls ( up : @xmath54 , down : @xmath61 ) , @xmath62-hs ( up : @xmath52 , down : @xmath63 ) , @xmath62-is@xmath57 ( up : @xmath60 , down : @xmath54 ) , @xmath62-is@xmath7 ( up : @xmath60 , down : @xmath61).,title="fig:",scaledwidth=42.5% ] the calculation carried out for the high temperature cubic @xmath3 phase demonstrates that for all experimental volumes bifeo@xmath0 has a metallic behavior .
the high spin to low spin state transition was also obtained in the calculation for the @xmath3 phase at 0.8v@xmath48 ( 33 gpa ) .
we have performed the gga+dmft calculations for bifeo@xmath0 at 700 k where it has no long range order for a number of different cell volumes corresponding to the whole range of experimentally applied pressures ( from ambient pressure up to 60 gpa ) . in agreement with experiments ,
the spectral functions for bifeo@xmath0 in the @xmath1 phase as well as in @xmath2 at ambient pressure demonstrates an energy gap of about 1.2 ev in a good agreement with experiment . when the unit cell volume is 0.85 - 0.82 v@xmath48 that corresponds to 25 - 33
gpa bifeo@xmath0 in the @xmath2 crystal structure becomes metallic , and , simultaneously , its magnetic moments shrink . in the low pressure @xmath1 structure ,
the same transition occurs in cell with the smaller volume 0.8 - 0.76 v@xmath48 ( 40 - 50 gpa ) . the high pressure cubic @xmath3 phase is metallic in the whole range of the investigated pressures ( cell volumes ) but the hs - ls transition also takes place when the cell volume is smaller than 0.82 v@xmath48 ( 33 gpa ) . the higher critical pressure of 43 gpa was obtained in the afm gga+dmft calculations for the @xmath2 structure in good agreement with the experimental phase diagram .
the authors thank p. werner for the ct - qmc impurity solver .
the gga+dmft calculations were performed on the supercomputing center of imm urb ras .
this work was partially supported by rfbr ( project no .
13 - 02 - 00050 ) .
this publication is based on work supported by a grant from the u.s . civilian research & development foundation ( crdf global ) via a joint grant no .
rup1 - 7077-ek-12 ( 12-cd-2 ) . | a metal insulator transition ( mit ) in bifeo@xmath0 under pressure was investigated by a method combining generalized gradient corrected local density approximation with dynamical mean field theory ( gga+dmft ) .
our paramagnetic calculations are found to be in agreement with experimental phase diagram : magnetic and spectral properties of bifeo3 at ambient and high pressures were calculated for three experimental crystal structures @xmath1 , @xmath2 and @xmath3 . at ambient pressure in the @xmath1 phase ,
an insulating gap of 1.2 ev was obtained in good agreement with its experimental value .
both @xmath1 and @xmath2 phases have a metal insulator transition that occurs simultaneously with a high spin ( hs ) to low
spin ( ls ) transition .
the critical pressure for the @xmath2 phase is 2533 gpa that agrees well with the experimental observations .
the high pressure and temperature @xmath3 phase exhibits a metallic behavior observed experimentally as well as in our calculations in the whole range of considered pressures and undergoes to the ls state at 33 gpa where a @xmath2 to @xmath3 transition is experimentally observed .
the antiferromagnetic gga+dmft calculations carried out for the @xmath2 structure result in simultaneous mit and hs - ls transitions at a critical pressure of 43 gpa in agreement with the experimental data . |
the idea that the dark matter in the universe may contain both a ` cold ' as well as a ` hot ' component @xcite first arose as a serious possibility @xcite within the framework of grand unified theories ( which also inspired inflation , cosmic strings , baryogenesis , etc . ) . in the original models the cold component was the axion ,
although nowadays the lightest supersymmetric particle ( lsp ) often plays that role .
lightly massive neutrinos are the hot component , and it was usually assumed that the neutrinos would display the hierarchical mass spectrum characteristic of the quarks and charged leptons , with a single neutrino ( _ e.g. _ , @xmath0 ) dominating the mass density of the hot component .
however , it was noted from the beginning @xcite that because of their special nature , nearly degenerate massive neutrinos is a logical , albeit not necessarily the simplest , possibility .
the cosmological implications of neutrinos that are closely degenerate in mass has received much recent attention @xcite , inspired to some extent by the remarkable series of neutrino oscillation experiments ( solar , atmospheric , and more recently accelerator ) which suggest that two or more neutrino flavors may contribute to the hot dark matter in the universe . before discussing the impact of neutrino oscillation experiments on the cold plus hot dark matter ( c@xmath7hdm ) scenario ,
let us briefly recall why the latter has attracted so much recent attention . from the mid to the late 1980s evidence was mounting that the so - called standard " cold dark matter ( cdm ) scenario had trouble providing a consistent explanation of small ( @xmath8 galactic ) as well as large scale structure
. on the contrary , a c@xmath7hdm model ( with @xmath9 ) provided a far more consistent fit to the data @xcite . in order to provide additional tests ,
the quadrupole anisotropy of the microwave background expected in c@xmath7hdm models was estimated in 1989 @xcite and compared with the cdm prediction .
normalized to the ` small ' scale data , it was found that the c@xmath7hdm prediction exceeded the cdm value by a factor of about two , and this was dramatically verified when the cobe team @xcite made its announcement in 1992 @xcite . in the last two years , particularly after extensive numerical simulations of galaxy and cluster formation @xcite and more detailed analytical work @xcite , additional support for the c@xmath7hdm model has emerged .
it appears that this model provides the simplest ( consistent ) realization of an inflationary scenario for large scale structure formation .
it has recently been pointed out @xcite , ( and our results are in agreement ) that with two or more neutrino flavors contributing to the hot dark matter , somewhat better fits to the present data may be possible . for fixed @xmath10 there is also some dependence both on the spectral index as well as the hubble constant .
it must be admitted , though , that within the framework of grand unified theories , it may not be easy to realize scenarios with two or three nearly degenerate ( in mass ) neutrinos .
furthermore , in some of the scenarios we consider , one needs to invoke a fourth ( sterile ) neutrino which is nearly degenerate in mass with the electron neutrino .
precisely why this singlet state can be so light is an important question for future research .
the layout of the paper is as follows . in section 2.1
we list three sets of observations related to solar , atmospheric , and dark matter neutrinos which provide hints for non - zero neutrino masses . for completeness
, we also summarize the results of the lsnd experiment at los alamos , although we will not include them among the constraints to be satisfied .
we then proceed to a discussion of three distinct schemes which are consistent with the observed solar and atmospheric @xmath11 oscillations , and which also allow for neutrino dark matter .
we discuss the implications of these schemes for a variety of experiments , particularly the @xmath12 and @xmath4 oscillation experiments at accelerators . in section 3
we address the main issues raised by neutrino oscillations for cosmological structure formation within the c@xmath7hdm framework .
we review the physics of the differences between having one and more than one massive neutrino flavor , giving some analytic formula to estimate the scale ( section 3.1 ) and amplitude ( section 3.2 ) of the effects on the growth of primordial density fluctuations . in 3.3
we give results of some more detailed numerical computations of the cosmological models based on the solutions to the oscillation data , including some models , in 3.4 , motivated by the lsnd results .
finally , in section 3.5 we discuss the dependence on the cosmological parameters ( hubble constant , spectral index @xmath13 , and baryonic density fraction ) of conclusions drawn from comparing specific models to large scale structure data .
one of the greatest challenges in particle physics today pertains to the issue of the neutrino masses .
there are indications from a variety of experiments that one ( or more ) of the known neutrinos may possess non
zero mass(es ) . in this section
we briefly summarize the relevant observations based on solar and atmospheric neutrinos , as well as present arguments for neutrino dark matter .
[ for completeness , we also summarize the recent findings of the lsnd experiment at los alamos . ] in section 2.2 , we discuss scenarios for neutrino masses and mixings which can simultaneously accommodate these observations .
we find that there are essentially three viable schemes , and remarkably enough , all three will be tested in ongoing and planned neutrino oscillation experiments ( particularly los alamos lsnd , rutherford karmen , the cern chorus / nomad and fermilab e803 experiments ) . 1 .
_ solar neutrino puzzle : _ the apparent deficit in the flux of solar @xmath14 s @xcite that has persisted in the chlorine experiment for over two decades @xcite has been confirmed in the last several years by three independent experiments : the kamiokande water cerenkov detector @xcite , and the sage @xcite and gallex @xcite radiochemical experiments .
these four observations , which probe different energy regimes in the solar neutrino spectrum , can all be simultaneously explained in terms of two flavor neutrino oscillations .
the matter enhanced msw @xcite oscillation for a two flavor ( @xmath15 ) system admits two branches @xcite : ( i ) the small angle non adiabatic solution , which requires sin@xmath16 and @xmath17 .
( ii ) the large angle msw solution requiring sin@xmath18 and @xmath19 . here
@xmath20 could be @xmath21 or a sterile neutrino @xmath22 for the small angle msw solution , while @xmath23 in the large angle msw solution .
there is also the @xmath24 vacuum oscillation solution @xcite for @xmath25 , with sin@xmath26 and @xmath27 .
atmospheric neutrino anomaly : _ the ratio of muon to electron neutrinos from atmospheric cascades measured by the kamiokande @xcite and the imb @xcite experiments appear to suggest a deficit of about a factor of 2 when compared to monte carlo simulations @xcite . for sub
gev neutrinos ( i.e. , neutrinos with energy less than @xmath28 1 gev ) , kamiokande finds for this ratio of ratios ( which is expected to be 1 in the absence of neutrino oscillations ) @xmath29 which is in good agreement with the imb value of @xmath30 . recently kamiokande has also reported results for the multi
gev neutrinos @xcite where the ratio is @xmath31 , in agreement with the sub - gev data .
this apparent deficit of @xmath32 s or the excess of @xmath14 s can be attributed to either @xmath33 oscillations with sin@xmath34 and @xmath35 , or to @xmath36 oscillations with sin@xmath37 and @xmath38 .
the resolution of the anomaly in terms of @xmath39 oscillations , where @xmath22 is a sterile neutrino , will run into difficulty with primordial nucleosynthesis calculations which require that @xcite @xmath40 .
( note that the solar @xmath41 oscillation parameters satisfy this constraint . ) among the other atmospheric neutrino experiments , only soudan ii @xcite sees an anomaly with a preliminary value of @xmath42 .
the frejus @xcite results are marginally in conflict with the combined kamiokande and imb data .
considering the low statistics of frejus experiment compared with the high statistical significance of kamiokande and imb data , we shall ignore this marginal discrepancy in our theoretical discussions .
_ hot component of dark matter : _ as already noted in the introduction , and further explained in the next section , a combined fit to the cobe data as well as the data on distribution of galaxies on the large and small angular scales is difficult to achieve within the cold dark matter scenario .
the simplest consistent scenario requires a significant hot component ( 15 - 30)% in the dark matter @xcite leading to the c@xmath7hdm scheme , with neutrinos being the natural candidates for the hot component .
the mass of the neutrino comprising the hot dark matter should be in the few ev range .
@xmath43 oscillation at accelerators : _ recently the lsnd experiment at the lampf facility in los alamos has reported positive evidence for @xmath43 oscillations @xcite .
if these results survive further scrutiny and there is independent verification ( e.g , at the karmen experiment @xcite ) , it certainly will have a strong impact on particle and nuclear physics as well as cosmology and astrophysics .
the initial lsnd data , if interpreted in terms of two neutrino oscillations suggest a mixing probability of @xmath44 , when oscillation constraints from karmen and bnl - e776 @xcite accelerator searches as well as reactor neutrino constraints from bugey facility @xcite are folded in . in the discussion that follows
, we shall take into considerations the sensitivity of the lsnd and other accelerator and reactor experiments in searching for @xmath43 oscillations , but we shall not demand that the theoretical scenarios discussed correctly reproduce the precise numerical values reported by the lsnd collaboration .
we feel that this position is justified at this time , especially in view of the fact that a different analysis of the lsnd data has yielded considerably weaker limits on the mixing parameters @xcite .
we find it remarkable however , that independent of the lsnd data , two of the three scenarios which can accommodate points ( 1)(3 ) above , imply an observable signal at accelerator @xmath45 oscillation experiments .
* 2.2 theoretical schemes : * let us focus on the solar , atmospheric and hot dark matter neutrinos , leaving out the lsnd results which will not play a role in determining the viable scenarios .
it is clear that the mass splittings required for explaining these three observations do not overlap ( @xmath46 for solar neutrinos , @xmath47 for atmospheric neutrinos and @xmath48 ( 1 to few ) ev for hot dark matter ) .
assuming that there are no additional light neutrinos , one concludes that the three neutrinos @xmath49 should be nearly degenerate in mass .
( note that we often identify a mass eigenstate through its dominant flavor .
the respective masses are @xmath50 and @xmath51 . )
this leads us to our first scenario .
assume that @xmath52 and @xmath0 have a nearly common mass of about ( 1 to 3 ) ev . their masses are split by small amounts , such that @xmath53 and @xmath54 .
phenomenological neutrino mass matrices that are consistent with these assumptions are easily constructed @xcite .
this scheme will account for the solar neutrino data via @xmath36 msw oscillations and the atmospheric neutrino anomaly in terms of @xmath33 oscillations .
the required mixing angles are free parameters and can be adjusted to the suggested values . the @xmath55 mixing angle can be arbitrarily small , but it can also be as large as about 0.05 . according to this scenario , neutrinoless double beta decay ( @xmath56 )
is at ( or even above ? ) the present experimental limit ( modulo nuclear matrix element uncertainties ) , and so that it can be ruled out if @xmath57 is not observed in the near future . also according to this scheme the @xmath1 oscillation experiments at accelerators ( chorus / nomad , e803 @xcite ) , as well as @xmath58 oscillation searches at reactors , should fail to find anything interesting . in both cases the relevant @xmath59 , which is not in the range accessible in these experiments .
however , this statement is strictly true only to the extent that the leptonic mixing matrix is assumed to be unitary .
deviation from unitarity can produce oscillation signals , although there should be no spatial dependence .
this situation can arise if the ordinary neutrinos mix with a neutral heavy lepton with mass on the order of the electroweak scale .
let us consider this in more detail . for definiteness ,
let us assume that the three light neutrinos have an admixture of a heavy fourth generation neutrino with mass greater than @xmath60 .
writing @xmath61 and similarly for the weak eigenstates @xmath21 , one sees that muon neutrinos produced in @xmath62 decays will be the weak eigenstates with the heavy @xmath63 component removed .
the @xmath14 and @xmath32 states are no longer orthogonal , leading to an apparent oscillation probability given by @xcite @xmath64 here we have set the @xmath65s to zero since they are all @xmath66 and thus negligible for accelerator neutrino experiments . if the mixing parameters @xmath67 , then the oscillation probability at the lsnd experiment will be in the range of @xmath68 with no position dependence .
the data presented in ref .
@xcite is consistent with such an interpretation , as can be inferred from the large @xmath65 region of their fig .
various other constraints on such leakage to a heavy fourth generation neutrino has been studied in @xcite , where it is shown that the best existing limit is from neutrino oscillation experiments .
@xmath1 oscillation probability of @xmath69 can also be achieved by a similar mechanism , but this may be more difficult to measure . if the heavy neutrino is a standard model singlet , there are stringent limits from @xmath70boson decays which make the non unitary oscillations unobservable at accelerators . to see this , consider the mixing of ordinary neutrinos with a heavy isospin singlet neutral lepton @xmath71 of mass @xmath72 .
in the presence of such mixings , the invisible decay width of @xmath70 will be modified so that the number of effective neutrino species coupling to @xmath70 is given by @xmath73\ ] ] where @xmath74 and @xmath75 are phase space factors given by @xmath76 with @xmath77 .
it is to be understood here that for @xmath78 , all terms in eq .
( 3 ) will contribute to @xmath79 , while for @xmath80 , the @xmath74 and @xmath75 terms do not contribute . if @xmath81 , the term proportional to @xmath75 should be set to zero , while @xmath74 is non zero .
the mixing parameter @xmath82 is equal to the ( 4,4 ) element of the unitary matrix @xmath83 which diagonalizes the neutrino mass matrix .
it is related to the leptonic mixing matrix @xmath84 appearing in the charged current via @xmath85 although @xmath84 , now a @xmath86 matrix is not unitary , the @xmath58 transition probability is still given by eq .
( 2 ) , since each row of @xmath84 is normalized to 1 .
it becomes apparent from eq .
( 2 ) and ( 5 ) that @xmath87 .
if the mass of @xmath71 is greater than @xmath88 , then @xmath89 , so that @xmath90 ^ 2/4 \equiv [ \delta n_\nu/(1+|\alpha|^2)]^2/4 $ ] .
@xmath91 can be as large as 0.042 at 1 sigma @xcite , which when combined with @xmath92 yields @xmath93 , too small to be observable .
( even with @xmath94 , the 2 sigma value , @xmath95 . ) if the mass of @xmath71 is less than @xmath96 , then both the @xmath74 and @xmath75 terms in eq .
( 3 ) will be relevant , making @xmath79 more consistent with 3 .
( as @xmath97 , @xmath98 .
) however , such a scenario is ruled out by direct search limits for a heavy neutral lepton decaying into the usual charged leptons @xmath99 , as should happen here .
( the limit on the mass of such a neutral lepton is @xmath100 ) . ) if the mass of @xmath71 obeys @xmath81 , then @xmath101 , but @xmath102 in eq .
( 3 ) . in this case , noting that @xmath74 is at most 11/16 ( for @xmath103 , we see that @xmath104 can be as large as @xmath105 , which is close to present experimental sensitivity .
however , since the branching ratio @xmath106 \sim 2 \times 10^{-4}$ ] , few hundred such events should have been observed at lep with the @xmath71 decaying subsequently into @xmath107 .
this possibility is also excluded based on the non observation of such events .
we conclude that the case of isosinglet neutrino does not lead to any observable deviation from unitarity in the accelerator neutrino oscillation experiments .
note that there is no such constraint on a sequential fourth generation neutrino ( or neutrinos in vector like families ) .
if there is no degeneracy in the neutrino masses and we assume some kind of a mass hierarchy , accounting for the solar , atmospheric and hot dark matter neutrinos would require the introduction of a light ( sterile ) neutrino @xmath22 .
there is then just one consistent scheme in this case .
as far as we are aware , this possibility has not been discussed in the literature , so we shall elaborate on it . by assumption , in this scenario the known neutrinos are non degenerate , and indeed have hierarchical masses .
the solar neutrino puzzle is resolved via @xmath41 msw oscillations such that @xmath108 .
the atmospheric neutrino anomaly is explained via @xmath109 oscillations with @xmath110 .
the tau neutrino with a few ev mass constitutes the ` hot ' component of dark matter .
as far as @xmath45 oscillations at accelerators are concerned , the `` direct '' transition is not possible because @xmath47 .
however , it has recently been pointed out @xcite that `` indirect '' @xmath45 transition via a virtual @xmath111 in the few ev mass range can still occur in such a scheme .
for this to be in the experimental reach , both @xmath55 oscillations ( at reactors ) and @xmath33 oscillations ( at accelerators ) should be in the observable range . as shown in @xcite , the present limits from bugey reactor @xcite on @xmath112 oscillations , and the charm - ii , fermilab e531 and cdhs limits @xcite on @xmath33 oscillations translates into an observable `` indirect '' @xmath36 oscillations for a @xmath0 mass in the few ev range , which can account for the lsnd results .
let us point out that in this second scenario , the allowed msw parameter region will be somewhat shifted because of the rapid @xmath36 oscillations from the sun to earth , leading to a further depletion of @xmath14 s by a factor of @xmath113 .
this prediction will be tested in the forthcoming solar neutrino experiments .
this scheme may be on a somewhat better theoretical footing than the first one because no mass degeneracy is assumed .
a light ` sterile ' neutrino state , following @xcite , may arise from the hidden sector of some fundamental theory .
if the hidden sector contains @xmath114 , one can define an unbroken ` parity ' at the lagrangian level .
the @xmath115 s needed for the see saw mechanism , having no gauge quantum numbers can freely mix from the observable and hidden sectors .
this would lead to mixing between the light neutrinos and their mirror partners ( which are also light ) .
gravitationally induced interactions could also mix the neutrinos from the two sectors .
this scheme also requires a light sterile neutrino @xmath22 .
in addition , an approximate lepton number symmetry such as @xmath116 is necessary to make @xmath32 and @xmath0 nearly degenerate in mass @xcite .
when this symmetry breaks by a small amount , maximal @xmath33 mixing occurs , facilitating a resolution of the atmospheric neutrino problem . the solar neutrino puzzle is explained via @xmath41 oscillations .
the common mass of @xmath32 and @xmath0 is assumed to be @xmath8 few ev , so that they constitute a ` two flavor ' hot dark matter . for accelerator neutrinos ,
there is direct @xmath117 oscillations at an observable level . in particular ,
the lsnd data can be accommodated .
however , @xmath33 oscillations will be beyond the reach of chorus / nomad and e803 . aside from the two flavor
hot dark matter component , this scenario has other special cases relevant for cosmology .
the solar neutrino data requires the mass splitting @xmath65 between @xmath14 and @xmath22 to be @xmath118 , but the masses themselves may be in the cosmologically interesting range of a few ev . since this pair behaves like dirac neutrinos , there is no conflict with @xmath119 constraints .
for cosmology however , the hot dark matter may be a linear combination of @xmath14 and @xmath120 .
if the total mass of @xmath121 pair exceeds that of @xmath14 , we call it a ( 2 + 1 ) c@xmath7hdm scheme , and if the @xmath14 mass dominates over @xmath121 , it is a ( 1 + 2 ) scheme . of course , one could also recover the one flavor case by making the mass of @xmath33 pair much smaller than 1 ev .
( this case will correspond to an inverted mass hierarchy @xcite . )
in studies of large scale structure , it has become clear that a critical density cold dark matter ( cdm ) universe with density perturbations that have roughly a harrison - zeldovich spectrum can not simultaneous fit observations of structure on large and small scales .
normalized on large scales to fit the cobe observations , cdm produces too many clusters of galaxies and galactic pairwise velocities far in excess of observations @xcite .
this basic problem is neatly resolved in a cold plus hot dark matter scenario because the growth of the density perturbations on small scales is damped by the presence of the hot dark matter component @xcite . since the neutrino oscillation experiments may be telling us that the neutrino component is spread among several flavors we would like to understand , in a quantitative way , how this changes the usual ( one flavor ) cold plus hot dark matter model .
we will begin with a general description of the effects of a hot component on the growth of density fluctuations .
first we will give some analytic formulae for estimating the size of the effects , then we will show some results of more detailed calculations .
finally we conclude with some remarks about the uncertainties in cosmological parameters which currently prevent one to use cosmological data to pin down the values of the neutrino masses .
neutrinos retain large velocities from the time before primordial nucleosynthesis when they were in thermal equilibrium .
this means that if one initially sets up density perturbations , the neutrinos will rearrange themselves to a different pattern at some later time , erasing neutrino density fluctuations on length scales over which the neutrinos can have traveled .
free streaming " or jeans " length scale @xmath122 is given at time @xmath123 by @xmath124 where @xmath125 is the scale factor which describes the expansion of the universe , @xmath126 is the conformal time ( @xmath127 ) , so @xmath128 is the physical horizon size , and @xmath129 is the average neutrino velocity .
an analytic fit for @xmath130 is given by @xcite @xmath131^{-1/2},\ ] ] where @xmath132 is the scale factor when the neutrinos become non - relativistic , and we use units where the velocity of light is unity .
we have @xmath133 where @xmath134 ev is the present cosmic background photon temperature .
the value of the jeans length at the current time @xmath135 is @xmath136 @xmath137 is the parametrization of our ignorance of the true value of the hubble constant @xmath138 km s@xmath5 mpc@xmath5 . to relate this length to the masses of known astrophysical structures
we convert to the total mass contained in the volume @xmath139 , ( the jeans mass ) . currently , the jeans mass @xmath140 is @xmath141 where @xmath142 is the mass contained in the current hubble volume .
interestingly enough , the mass in ( 10 ) is comparable to a galactic mass for neutrinos in the range @xmath143 ev .
note that in the formulae presented here we assume a critical density of matter , and only a fraction is in the form of hot dark matter the remainder is comprised of cold dark matter and baryons . during the matter dominated epoch , @xmath144 , @xmath145
( while the neutrinos are non - relativistic ) and @xmath146 , so the jeans mass @xmath147 and decreases with time . when the neutrinos are relativistic , @xmath148 , and the jeans mass increases with time .
the maximum in the jeans mass occurs when the neutrinos are just becoming non - relativistic , _ i.e. _ , when @xmath149 : @xmath150 this relation is approximate , because the time when neutrinos become relativistic is quite close to the time when matter begins to dominate over radiation in the universe , so the horizon size does not scale exactly as @xmath151 .
the exact formula must be determined numerically for each value of the neutrino mass .
the formula in ( 11 ) is accurate to within an order of magnitude , which is good enough for estimating the size of the effects under study here . in the older models of structure formation in which the dark matter is totally composed of neutrinos of mass @xmath8 30 ev ,
this maximum jeans mass is somewhat larger than a typical mass of a cluster of galaxies . in studies of single massive flavor c@xmath7hdm models ,
a more typical value of the neutrino mass @xmath152 ev ( for @xmath153 ) was preferred , yielding a maximum jeans mass of @xmath154 , which is slightly smaller than the so called great attractor " @xcite , the largest concentration of mass which is pulling galaxies gravitationally on scales of order @xmath155 mpc .
if we assume , as in scenario i ) , that the same @xmath156 is now composed of 3 nearly degenerate neutrinos of mass @xmath157 ev , @xmath158 , which is somewhat larger than the great attractor " mass . we will proceed to examine how these two mass scales ( @xmath159 and @xmath160 ) are incorporated in predictions of structure formation . in the early universe , when matter begins to dominate the energy density of the universe , the cdm and c@xmath7hdm mass fluctuation spectra are identical .
the only feature imprinted on the initial spectrum at this time is a decrease of amplitude on length scales smaller than the horizon at matter domination .
these scales experience a smaller amount of growth during the radiation dominated era due to the effects of radiation pressure . after matter domination
the fluctuation amplitudes on these scales can grow , driven by their own self gravitation .
if a volume of space contains a mass @xmath161 which is slightly different than the average @xmath162 for that size , and the dark matter is strictly cdm , the fluctuation amplitude will grow proportional to the scale factor @xmath163 @xmath164 growth in the c@xmath7hdm model is more interesting .
when a mass fluctuation contains a mass @xmath165 at time @xmath123 , the neutrinos will stream out of the fluctuation and change the local gravitational potential .
this smaller potential also reduces the growth rate of the mass fluctuations to @xcite @xmath166 the fact that the jeans mass decreases with time means that if @xmath167 at some early time , sometime in the future @xmath168 will be true .
once @xmath168 , the growth rate returns to the cdm growth rate , as the neutrinos will then effectively be cold " on this mass scale .
this leads to scale dependent changes in the mass fluctuation spectrum .
the equations for the jeans mass ( [ mjmax ] ) and ( [ mj0 ] ) , and the growth rate ( [ chdmgrowth ] ) also illustrate the effect of having multiple flavors of massive neutrinos with comparable masses .
the jeans mass depends on the individual neutrino mass , while the growth rate depends only on the _ sum _ of the neutrino masses , as @xmath169 so increasing the number of degenerate flavors means that the jeans mass will be increased for a given value of @xmath156 .
the growth of fluctuations on mass scale @xmath162 in the intermediate region @xmath170 will be damped " relative to their value in a universe dominated only by cdm .
the amount of damping between the time when the neutrinos become non relativistic and when the jeans mass becomes smaller than scale @xmath162 is given by @xmath171^{(2p -2)/3}\ ] ] the maximum amount of damping occurs for mass scales @xmath172 which is a constant given by @xmath173^{(2p-2)/3 } = ( a_{nr})^{1-p}\ ] ] the difference , then , between having the mass equally spread among @xmath174 flavors as opposed to 1 flavor results in an additional small scale damping of @xmath175 , which is between 10% - 20% for @xmath176 .
the combination of hot and cold components of dark matter thus has the following effects , when compared to the a model with only a cold component .
first of all , the growth of density perturbations with masses in the intermediate range @xmath177 is reduced by the factor given in eq .
( [ damping ] ) which depends mainly on @xmath156 .
below @xmath178 the mass fluctuation spectrum follows the scale dependence of a cdm model but with an amplitude reduced by the factor in eq .
( [ dampmax ] ) . splitting the neutrino density among @xmath174 flavors of approximately degenerate neutrinos produces an additional damping of @xmath179 below a somewhat higher value of @xmath160 .
this additional damping may be important for reconciling observations of clusters of galaxies with a cobe normalized spectrum of density fluctuations . in order to do a more careful comparison of these models to observed large scale structure , a more accurate evaluation of the effects expressed above is required .
we achieve this by integrating the linearized general relativistic evolution equations for the photons , neutrinos ( massive and effectively massless flavors ) , baryons and cdm particles .
the procedure we use is described in detail in ref .
the models have a baryon mass fraction @xmath180 , which is reasonably consistent with nucleosynthesis @xcite .
the cdm fraction is then given by @xmath181 ( since here we are only considering critical density universes ) .
in figures 1 and 2 , we use a harrison zeldovich ( scale free ) spectrum normalized to cobe @xcite with @xmath153 . after smoothing the density field with a low pass gaussian filter , we arrive at the curves shown in figures 1 and 2 .
the masses on the x - axis are @xmath182 , as appropriate for a gaussian filter radius of @xmath183 . in the three panels a ) , b ) , and
c ) , we show curves for @xmath184 , @xmath185 , and @xmath186 respectively . within each panel
we show curves for 1 , 2 , and 3 , degenerate neutrino flavors . in panel
a ) , we also display the analogous cdm model curve for comparison .
the models under consideration are now known to fit a large amount of data on large scale structure @xcite , but here we will concentrate on the two toughest constraints for c@xmath7hdm models to satisfy .
the first constraint is on the amplitude of mass fluctuations which form clusters .
it has been argued @xcite that the number density of rich clusters requires that on a scale @xmath187 , @xmath188 ( 2 times their error bar " ) . as stated in ref .
@xcite the errors are hard to estimate because of possible systematic uncertainties in the masses assigned to clusters . indeed ,
the mass scales for two different determinations of the cluster mass function in refs @xcite and @xcite differ by almost a factor of 2 . ref .
@xcite argues that the cluster masses could be overestimated , while measurements of the gravitational lensing by the clusters suggest @xcite that the usual mass estimates ( used to determine ( @xmath189 ) ) may be low by as much as a factor of 3 . in order to allow for this uncertainty
, we also give an estimated horizontal error bar which is a factor of two ( in cluster mass ) from the value @xmath190 , to reflect the size of the possible systematic error . at the far left end of the figures
we have also drawn a line which corresponds to the lower limit of the mass fluctuation amplitude implied by the measurements of the abundance of damped lyman alpha systems " as indicated by the recent survey of ref .
@xcite ; see also ref . @xcite .
while early work @xcite indicated that c@xmath7hdm models ( with @xmath191 ) predictions were far below the observations , it now seems as though the assumptions used may have been too restrictive .
there is little information about the exact nature of these damped lyman alpha systems , other than their neutral hydrogen column densities and spectral line widths .
a more general analysis @xcite shows how to properly estimate number abundances of objects on these scales @xcite .
we have used this latter procedure to estimate the amplitude of the constraint line in figures 1 and 2 .
it is still uncertain what mass is to be identified with these damped lyman alpha systems .
it seems reasonable that we should associate dark matter halo masses of between @xmath192 and we have placed our constraint line there in our figures .
( the curves should pass above the constraint line somewhere in the range . )
the mixed dark matter models are more compatible with the lower end of this mass range . however , with @xmath193 and @xmath153 one has to take very small halo masses even to make this model work marginally . from figure 1 then it is apparent that @xmath194 if @xmath153 with a pure harrison
zeldovich spectrum .
figure 1 also points out an interesting aspect of spreading the neutrino mass density among a number of massive neutrino flavors @xmath174 . with @xmath195 ,
the amplitude of the density fluctuations on cluster scales seems to be slightly larger than what is required ; increasing @xmath174 improves the fit .
this result depends only weakly on @xmath156 on this mass scale .
[ eq.([damping ] ) predicts less than 10% difference in the cluster scale @xmath196 amplitude over the range @xmath197 .
] however , because of the @xmath156 dependence in the @xmath179 damping factor ( via @xmath198 ) , @xmath199 works best for @xmath200 and @xmath201 works best for @xmath202 . on the other hand increasing the number of flavors makes the disagreement with lyman alpha systems datum worse for @xmath200 , so @xmath203 seems to work best for models with @xmath204 .
we should point out here that the conclusions we draw from figure 1 hold only for the spectrum , hubble constant and baryon fraction used in the models , and altering any of these parameters changes the conclusions .
we will explore this further later .
the purpose of figure 1 is mainly to illustrate the effects of modifying the hot dark matter composition . in the above
we have considered the consequences of the three neutrino scenarios in which we have @xmath174 flavors with roughly equal mass . however , if we take the lsnd results as a serious constraint , then scenario ( iii ) admits the possibility that all three known neutrino flavors can have masses in the ev range , although the electron neutrino mass can still be significantly different from the @xmath205 and @xmath206 neutrino masses ( which form a degenerate pair ) . since
only the square of the mass difference is measured , we have two cases to consider : a ) @xmath207 , ( 2 + 1 " ) and b ) @xmath208 ( 1 + 2 " ) . in a
@xmath209 " model , where @xmath32 and @xmath0 are each heavier than @xmath14 , the lightest neutrino ( @xmath14 ) will have a very large maximum jeans mass , ( @xmath210 ) , compared to the other two heavier neutrino jeans masses ( @xmath211 ) .
@xmath212 " model , where@xmath32 and @xmath0 are lighter than @xmath14 , the lightest neutrinos ( @xmath32 and @xmath0 ) will have very large @xmath210 , compared to the remaining heavier @xmath14 neutrino @xmath211 . for @xmath213 , there will be an amount of damping which would be there if only the lightest neutrino comprised @xmath156 . for @xmath214 the damping will be intermediate between the one and three flavor massive neutrino case .
to illustrate these two cases , we will use values of @xmath215 and 20 ev@xmath216 , for the @xmath209 " and @xmath212 " models respectively in the context of an @xmath217 , @xmath153 , @xmath218 cosmology . in figure 2
we plot the curves for the @xmath209 model ( @xmath219 ev and @xmath220 ev ) , and the @xmath212 model ( @xmath221 ev and @xmath222 ev . for comparison
we also plot the curves for @xmath195 , 2 , and 3 flavors for the same @xmath223 .
the curve for @xmath209 ( @xmath212 ) lies in between the curves for the 2 and 3 ( 1 and 2 ) degenerate flavors .
we can estimate the difference between having 3 flavors with equal and unequal masses as follows . in the 1 + 2 " and 2 + 1 " models the lightest flavor mass @xmath224 is lighter than the three equal flavor mass @xmath225 . the heavier flavor mass @xmath226 is greater than @xmath225 . consider a mass scale @xmath227 .
after some time , when @xmath228 , the amount of damping is the same in both the equal and unequal mass cases .
after that time that scale will experience an additional damping of @xmath229^{(2p_l -2)/3}$ ] ( eq . [ damping ] ) in the unequal mass case , and @xmath230^{(2p-2)/3}$ ] in the 3 flavor case .
here @xmath231 is @xmath198 in equation ( [ chdmgrowth ] ) calculated assuming @xmath156 is only made from the lightest neutrino(s ) .
the ratio of the damping in the 1 + 2 ( and 2 + 1 ) cases to the 3 equal mass case is ( for scales @xmath232 ) @xmath233 the ratio in ( [ dratio ] ) is always larger than 1 , indicating that the damping is greatest in the 3 equal flavor case .
however , if the ratio of the masses of heavy to light flavors is @xmath234 the difference between damping in the various cases differs from the 3 flavor case by @xmath235 % .
this is the reason for choosing the particular @xmath65 for plotting .
had we plotted , for example , the 2 + 1 and 1 + 2 models using @xmath236 ev@xmath216 , ( the same value used in ref .
@xcite , but for @xmath202 ) both of the curves would be nearly indistinguishable from the 3 degenerate neutrino case for masses @xmath237 . when the light and heavy masses differ by an order of magnitude or more , as in figure 2
, the @xmath209 model goes to the 2 degenerate flavor model and the @xmath212 goes to the single flavor model .
we note that because the lightest neutrino weighs less than the 3 degenerate flavor neutrino , on the very largest scales where @xmath238 , there will be some damping which is absent in the 3 degenerate flavor case .
we see that these additional models offer now a continuum of @xmath239 values between the 1 and 3 degenerate mass neutrino cases .
the sole motivation for considering a case where the neutrinos have masses _ and _ mass differences of order a few ev in scenario iii ) is the possibility of detecting oscillations with @xmath65 in the few ev@xmath216 range in the lsnd ( and also karmen ) experiments . in the framework of gut theories however
, such a mass spectrum will not be easy to understand . in the previous sections we have plotted results for specific values of the hubble constant , initial density fluctuation spectrum , and @xmath240 .
we have not drawn strong conclusions about the neutrino mass scenarios , because the results are quite sensitive to the specific values used for these cosmological parameters .
we will discuss the effects of these parameters on the cosmological structure formation and give specific examples of models from each of the three mass scenarios .
1 . @xmath241 : the value of the hubble constant has been the subject of a long and ongoing observational campaign ( see @xcite for some recent measurements ) .
the experiments find that @xmath242 .
this uncertainty is quite large when seen from the point of view of the required neutrino masses . rewriting equation ( [ omeganu ] ) for
degenerate neutrinos we find @xmath243 which emphasizes that the neutrino masses scale as @xmath244 . in figures 1 and 2
we have used @xmath153 .
some of the recent measurements @xcite seem to imply a larger value of @xmath137 .
in particular , if @xmath245 , then the masses used in the previous examples are increased by a factor of @xmath246 .
for example , ref .
@xcite advocates using @xmath247 and @xmath248 with @xmath249 ev@xmath216 , @xmath153 .
if instead @xmath245 with the other parameters fixed , then the oscillation signal at los alamos would be consistent with @xmath250 ev@xmath216 .
+ the scale which corresponds to the horizon size at matter domination also depends on the hubble constant . if we use a larger value of @xmath137 , then matter domination occurs earlier , when the horizon size was smaller , which means that the radiation dominated era is less effective at arresting the growth of small scale fluctuations .
such an effect exacerbates the small scale problems in cdm models and favors using a larger value of @xmath156 @xcite .
+ are attributed 2 .
initial mass fluctuation spectrum + it has long been known that inflation predicts a power spectrum of density fluctuations with a spectral index @xmath13 close to unity .
since the amount of deviation is strongly model dependent , many investigators are content to use @xmath218 . in figures 1 and 2
we have also used the value of @xmath218 for the spectral index . however
, even small deviations from @xmath218 can lead to significant changes in the conclusions . to see this we give the initial spectrum in terms of rms mass fluctuations ( @xmath251 ) :
@xmath252 in the simplest models of inflation , particularly those based on guts ( see , _
e.g. , _ ref .
@xcite ) , @xmath253 , although other values of @xmath13 are certainly possible @xcite . from eq . ( [ masspect ] )
we see that decreasing ( increasing ) @xmath13 from unity decreases ( increases ) the small scale power .
since the models are normalized at large scales to cobe observations the mass fluctuation curves in figures 1 and 2 will pivot " around a very large mass scale @xmath254 .
the pivot " mass scale here is somewhat smaller than the horizon mass , because the best fit cobe quadrupole anisotropy scales as @xmath255 @xcite .
( we have fit our spectra by normalizing to the 7th multipole moment of the sachs - wolfe anisotropy , as recommended in ref .
@xcite ) 3 . @xmath256 in the past few years comparisons of the primordial light element abundances inferred from observations with those calculated have led to strict limits @xcite on the amount of baryonic material in the universe , @xmath257 .
we have used the central value of this range for the models in figures 1 and 2 .
improvements in the observations of deuterium and @xmath258he and determinations of the neutron lifetime have now led to a situation where all of the light element abundances are not consistent in a universe where the three known neutrino flavors have the standard number densities ( derived from thermal equilibrium ) at nucleosynthesis @xcite .
if , however , the @xmath259he abundances derived from observations have been systematically underestimated @xcite by about 5%-10% then big bang nucleosynthesis would be made consistent with the three flavors of neutrinos , provided that the baryon density is about 50% larger than previously thought .
this would go in the right direction to explain why the baryon to dark matter ratio in cluster cores is so much larger than @xmath240 @xcite .
+ if we allow for larger baryon fractions , this will also change the amplitude of the mass fluctuations for a given @xmath156 and @xmath13 .
we can understand this through the following .
baryons ( mostly protons ) after nucleosynthesis are electrically charged and so are strongly coupled to the photon field .
baryonic density fluctuations can not grow until the photon temperature decreases to allow the stable formation of neutral hydrogen . because photon
baryon decoupling does not happen until after matter domination , the baryons are prevented from falling into the gravitational wells supplied by the cold dark matter .
this leads to a damping in the growth of density fluctuations ( relative to a case with no baryons ) .
the amount of damping ( on scales smaller than the horizon size at photon baryon decoupling ) in the final amplitude of density fluctuations is constant , so the effect is quite similar to changing the number of degenerate neutrino flavors .
these uncertainties mean that one can not currently use the cosmological data to determine which of our three scenarios is correct , or even what the value of @xmath156 is .
one must do a systematic study of the available parameter space taking into account the full range of these parameters .
studies of the interplay between these cosmological parameters have been done in various contexts @xcite , with more in progress @xcite . as a graphic illustration of this uncertainty , and also to demonstrate use of the formulae presented in sections 3.1 - 3.3 , we show four models in figure 3 that give nearly identical predictions of structure formation , but use quite different solutions of the neutrino oscillation data .
the choices of parameters are as follows . 1 . the solid curve . [ scenario ( iii ) ] .
this model with @xmath199 , @xmath202 , @xmath218 , and @xmath153 is advocated as a good fit to cosmological observations by ref .
@xcite consistent with the los alamos lsnd experiment .
we present this model for comparison .
2 . the dot - dashed curve .
[ also scenario ( iii ) ] .
we increase @xmath137 to @xmath260 in this @xmath199 model , but now there will be too much small scale power . to compensate for this ,
we increase @xmath156 to 0.25 , and decrease @xmath13 to 0.95 , a value which actually is more in line with standard inflationary predictions .
the @xmath261 experiments would then be expected to see a signal corresponding to @xmath262 ev@xmath216 .
the baryon fraction has been scaled from the first model as @xmath244 consistent with the nucleosynthesis constraints .
3 . the short dashed curve . [ scenario ( i ) ] . again using @xmath245 and @xmath156 ,
we now consider @xmath201 degenerate flavors . because the extra flavor provides an additional damping factor over the two flavor model of @xmath263 , ( eq . [ dampmax ] ) , to get a curve similar to the previous one we can increase @xmath13 to 0.98 , which increases the amplitude at the cluster mass by @xmath264 , ( eq . [ masspect ] ) , so we get very nearly the same fit at the cluster scale .
the value @xmath265 happens to be the value predicted in a particular susy gut inflation model @xcite .
4 . the long dashed curve .
[ scenario ( ii ) ] . here
we return to @xmath153 , @xmath266 , and @xmath218 as in the solid curve . to compensate for the increase in amplitude by going back to 1 flavor @xmath267 in the solid curve , we increase @xmath240 to 0.10 .
this value of @xmath240 is consistent with 3 relativistic neutrino flavors during nucleosynthesis if the @xmath259he abundances have been systematically underestimated .
we have now shown that allowing for more than one degenerate ( in mass ) neutrino state , as indicated by solar and atmospheric oscillation experiments , produces a degeneracy " in the predictions of structure formation for c@xmath7hdm models , given the uncertainty in cosmological parameters . in order to break this latter
degeneracy " we need a convincing detection of neutrino oscillations in an accelerator experiment , or an improvement in the determination of cosmological parameters , or both . in the meantime , we just note that there is a very rich structure contained in c@xmath7hdm models of structure formation , which still allows considerable flexibility in fitting astrophysical data .
we have investigated the impact of neutrino oscillations , as indicated by a number of solar and atmospheric neutrino experiments , on the ` cold plus hot ' dark matter scenario of large scale structure formation .
we are led to three distinct scenarios for neutrino masses and mixings with interesting predictions for the ongoing / planned experiments .
we note in particular the expectations for @xmath268 and @xmath1 oscillations being currently searched for .
the cosmological implications of the three scenarios are explored in some detail .
for some choices of the cosmological parameters ( particularly @xmath153 , @xmath218 , and @xmath269 ) , the two and three neutrino flavors scenarios provide a somewhat better fit to the present data than the single ( neutrino ) flavor case .
however , as we show , this need not hold for a different parameter choice .
it is too early to pick out the best model but it is remarkable that taking account of the oscillation experiments , the c@xmath7hdm models can still provide a good fit to the large scale structure data within the context of the simplest inflation models .
abc d. pogosyan and a. starobinsky , 11th potsdam workshop on rel .
, potsdam , ( 1994 ) .
j. r. primack , j. holtzman , a. a. klypin , d. o. caldwell , _ phys . rev
_ , * 74 * , 2160 , ( 1995 ) .
there are a large number of papers on this subject , and we refer interested readers to general treatments : a. linde , _ particle physics and inflationary cosmology _ , harwood academic , n.y . , 1990 ; e. kolb and m. turner , _ the early universe _ , addison - wesley , new york , ( 1989 ) . c. copi , d.n .
schramm , m.s .
turner , 1994 , _ science _
( in press ) .
white , j.f .
navarro , a.e .
evrard , and c.s .
frenk , 1993 , _ nature _ , * 366 * , 429 .
* fig . 1*.
we show the difference in the rms mass fluctuation amplitude ( @xmath239 ) between having the hot dark matter ( neutrino ) component distributed among 1 , 2 , or 3 flavors of degenerate neutrinos which correspond to our scenarios ii ) , iii ) and i ) respectively , for fitting the neutrino oscillation data . here
@xmath153 , @xmath269 , and @xmath218 , with the spectrum normalized to cobe .
the data points are : 1 ) the mass fluctuation estimate based on cluster formation @xcite and 2 ) the lower limit on the mass fluctuations consistent with the formation of lyman alpha cloud abundances at high redshift @xcite .
( the mass at which this lower limit is to be imposed should be somewhere along the constraint line . )
panels a ) , b ) , and c ) correspond to @xmath270 and 0.30 , respectively .
it can be seen that increasing the number of neutrino flavors marginally improves the fit to the cluster data . in panel
a ) we also show the curve for a pure cold dark matter model . *
2*. the rms mass fluctuation amplitude in a special case of neutrino scenario iii , which has been made consistent with the lsnd best fit of @xmath271 ev@xmath216 in a model with @xmath272 , @xmath153 , @xmath218 . for comparison
we also show ( from top to bottom ) the same @xmath156 with 1 , 2 , or 3 degenerate neutrino flavors .
the @xmath209 " and @xmath212 " model curves are intermediate to the degenerate neutrino cases .
the data are the same as in figure 1 . *
3*. the rms mass fluctuation in some selected models with varying cosmological parameters .
we plot curves for two models from scenario iii with @xmath199 for ( @xmath202 , @xmath153 , @xmath218 ) - as advocated in ref .
@xcite - and for ( @xmath217 , @xmath245 , @xmath273 ) .
these two models would predict @xmath274 and 18 ev@xmath216 at the lsnd and karmen experiments .
the third curve is for a model with @xmath201 ( @xmath217 , @xmath245 , @xmath265 ) .
lastly we show a curve from scenario ii ) with @xmath174 ( @xmath202 , @xmath153 , @xmath275 ) . here
we have increased @xmath240 to reduce the ampitude at cluster scales . as can be seen
, it is difficult to determine which neutrino mass scenario is correct based on current cosmological data alone .
the data are the same as in figure 1 . | we explore the implications of possible neutrino oscillations , as indicated by the solar and atmospheric neutrino experiments , for the cold plus hot dark matter scenario of large scale structure formation .
we find that there are essentially three distinct schemes that can accommodate the oscillation data and which also allow for dark matter neutrinos .
these include ( i ) three nearly degenerate ( in mass ) neutrinos , ( ii ) non - degenerate masses with @xmath0 in the ev range , and ( iii ) nearly degenerate @xmath1 pair ( in the ev range ) , with the additional possibility that the electron neutrino is cosmologically significant .
the last two schemes invoke a ` sterile ' neutrino which is light ( @xmath2 ev ) .
we discuss the implications of these schemes for @xmath3 and @xmath4 oscillation , and find that scheme ( ii ) in particular , predicts them to be in the observable range .
as far as structure formation is concerned , we compare the one neutrino flavor case with a variety of other possibilities , including two and three degenerate neutrino flavors .
we show , both analytically and numerically , the effects of these neutrino mass scenarios on the amplitude of cosmological density fluctuations . with a hubble constant of 50 km s@xmath5
mpc@xmath5 , a spectral index of unity , and @xmath6 , the two and three flavor scenarios fit the observational data marginally better than the single flavor scheme .
however , taking account of the uncertainties in these parameters , we show that it is premature to pick a clear winner .
= 8.0 in = -0.15 in ba-95 - 19 ; astro - ph/9507006 + k.s .
babu , r.k .
schaefer , and q. shafi + pacs numbers : 98.65-r , 14.60-st , 95.35+d + |
we now prove part ( i ) of lemma 1 ; part ( ii ) follows along the same line of proof .
suppose that @xmath33 .
recalling that @xmath130 , the corresponding supremum expectation value of the test operator is @xmath131 .
now we prove that under the condition @xmath33 the optimal point @xmath132 for the set @xmath23 must satisfy @xmath37 .
for this , assume the contrary , that is , @xmath133 and thus , @xmath134 .
define the separable state @xmath135 ( @xmath136 $ ] ) . since the two points @xmath132 and @xmath137 are on two different sides of the hyperplane defined by @xmath16 , it is possible to find a @xmath138 such that @xmath139 lies on @xmath16 , i.e. , @xmath140 and @xmath141 .
now , let us consider the position of @xmath139 with respect to @xmath25 .
we have @xmath142 the last inequality is a result of the convexity properties , as the supremum of a nonconstant convex ( here , linear ) function on a nonconstant convex set is strictly larger than its supremum over any subset of that convex set not equal to the set itself .
equation , however , implies a contradiction to the witnessing property of @xmath25 for the set @xmath23 .
thus , @xmath143 , as claimed . we now prove theorem 1 in the case where condition ( i ) of lemma 1 holds true .
the same kind of proof can be used for the case where condition ( ii ) holds .
first , without loss of generality , we restrict ourselves to the finite @xmath129-dimensional hilbert space , as entanglement can always be detected in finite dimensions @xcite .
we also require the following lemmas .
any separable mixed quantum state @xmath4 , which satisfies @xmath152 can be decomposed as @xmath153 for some @xmath154 such that @xmath137 is a pure product state satisfying @xmath94 .
the state @xmath155 is the remainder .
a. if either @xmath156 or @xmath157 for every @xmath158 , then the corresponding hyperplane is tangent to the set of separable states . as a result
, @xmath144 belongs to a face of @xmath0 and thus , it can be decomposed in terms of extremal points of that face , i.e. , pure separable states satisfying @xmath94 .
b. assume that @xmath159 and @xmath160 for some @xmath161 .
thus , there are infinitely many pure product states @xmath10 , such that @xmath162 , and infinitely many pure product states @xmath163 , such that @xmath164
. choose an arbitrary pair @xmath10 and @xmath163 from those states .
there necessarily exists a single parameter continuous family of unitary operators @xmath165\}$ ] such that @xmath166 and @xmath167 and @xmath168 such that @xmath169 for some characteristic time @xmath170 .
hence , one can define @xmath171 and @xmath172 . as a result of the continuity of @xmath173 ,
the function @xmath174 is continuous , @xmath175 , and @xmath176 corresponds to a continuous path on the surface of the set of separable states connecting the points @xmath137 and @xmath177 . ] .
therefore , there must exist a value of the parameter @xmath178 such that @xmath179 , and thus there exists a pure product state @xmath180 for which @xmath181 .
now , since there are infinitely many @xmath10 s , @xmath163 s , and one parameter families @xmath182 , there must exist infinitely many points @xmath145 .
let us call the set of all such states @xmath183 .
thus , one can always choose a complete set of vectors @xmath184 from the set @xmath183 such that @xmath185 for all @xmath186 , because @xmath187 .
+ the final step is to show that there must always exist at least one decomposition of @xmath144 which contains at least one @xmath188 .
equivalently , using lemma 1 above , there must exist one @xmath188 for which @xmath146 . to see this ,
assume that there is no @xmath145 such that @xmath146 , this means that @xmath189 for all elements of @xmath190 , and thus @xmath191 , i.e. , @xmath144 can not be a quantum state , implying a contradiction .
we can now prove theorem 1 .
suppose that @xmath192 where each @xmath193 is a pure product state and @xmath194 .
using lemma 1 , we know that @xmath195 where @xmath196 . thus , we can write @xmath197 the first inequality follows from the fact that @xmath193 probably is not the optimal separable state over @xmath198 .
the third equality is obtained by assuming that the optimal state of @xmath198 is @xmath199 .
the second inequality then follows by noticing that @xmath200 .
note that , we put no specific condition on the decomposition of @xmath205 in eq . . as a result , we use the fact that @xmath205 is necessary inside the set @xmath0 and we can use the lemma above and say that we can always decompose @xmath205 in the form @xmath206 for some @xmath154 , @xmath137 being a pure product state satisfying @xmath94 , and @xmath155 being the remainder state . using the previous result , this implies @xmath207 , i.e. , the optimal point can always be chosen to be a pure state .
the commutativity of @xmath16 and @xmath6 together with their separability imply that there exists a basis set of product operators @xmath208 that diagonalizes both of them as @xmath209 and @xmath210 .
suppose that there exists an entangled state @xmath1 such that @xmath93 and @xmath211 , with @xmath212 being the supremum value attainable by separable states .
consider the separable state @xmath213 it is clear that @xmath214 and @xmath215 , implying a contradiction .
recall that , by neumark s theorem , two operators @xmath16 and @xmath6 are jointly measurable if and only if they correspond to two operators @xmath216 and @xmath217 in an extended hilbert space @xmath218 , respectively , satisfying @xmath219=0 $ ] such that @xmath220 and @xmath221 for an orthogonal projection operator @xmath222 @xcite .
then , using theorem 2 , if @xmath49 and @xmath50 are jointly measurable for either alice or bob , for each @xmath19 there exists a separable state which maximizes the expectation value of @xmath217 ( and hence , @xmath6 ) over _ all _ quantum states making @xmath16 and @xmath6 unsuitable for uew .
here we derive the optimal separable state for the simplest multipartite version of our protocol using the fixed three - outcome measurement scheme of eq . , and the test and constraint operators in eqs . and , respectively . first , the optimal separable states with respect to the partitioning @xmath99must be of the form @xmath223 where each @xmath224 can be _ entangled _ for party @xmath100 .
it is clear that @xmath225 with @xmath226 , and thus , for @xmath108 , it is only required that one of the @xmath227 s equals zero .
similarly , @xmath228 where @xmath229 . by extrapolating the bipartite case , it turns out that each vector @xmath224 has the following form : @xmath230,\text{~ for some party~}\mathcal{i}_u,\\ & |\psi_v\rangle = f_v^{-\frac{1}{2}}\bigotimes_{i=1}^{m_v}|\chi^{+}\rangle,\text{~ for all parties~}\mathcal{i}_v\neq \mathcal{i}_u , \end{split}\ ] ] where @xmath231 and @xmath232 are appropriate normalizations .
the form of @xmath233 guarantees that @xmath234 and thus @xmath108 , while delivering the maximum value for @xmath235 within that partition .
the form of @xmath236 guarantees that for each party @xmath237 the maximum @xmath238 will be obtained . to determine which partition must be chosen as party @xmath239 , we first calculate the following values . @xmath240 it can be easily verified that for @xmath241 , @xmath242 and @xmath243 as @xmath244 .
hence , to obtain the maximum value @xmath245 , we chose the party @xmath246 to be the party with the maximum number of agents ( subsystems ) , i.e. , @xmath247 in the notation from the main text , where @xmath104 . consequently , @xmath248 a normalization coefficient , and @xmath249.\ ] ] note that , for @xmath112 , i.e. , when no partitioning has been made , we find that the state @xmath250 in eq
. gives the highest expectation value for the test operator over the _ full _ state space .
that is , the resulting state is genuinely entangled .
j. sperling and w. vogel , phys .
a * 79 * , 052313 ( 2009 ) .
g. cassinelli , e. de vito , and a. levrero , j. math .
appl . * 210 * , 472 ( 1997 ) .
p. kruszynski and w. m. de muynck , j. math . phys .
* 28 * , 1761 ( 1987 ) . | entanglement witnesses are invaluable for efficient quantum entanglement certification without the need for expensive quantum state tomography . yet
, standard entanglement witnessing requires multiple measurements and its bounds can be elusive as a result of experimental imperfections . here
we introduce and demonstrate a novel procedure for entanglement detection which seamlessly and easily improves any standard witnessing procedure by using additional available information to tighten the witnessing bounds .
moreover , by relaxing the requirements on the witness operators , our method removes the general need for the difficult task of witness decomposition into local observables .
we experimentally demonstrate entanglement detection with our approach using a separable test operator and a simple fixed measurement - device for each agent .
finally we show that the method can be generalized to higher - dimensional and multipartite cases with a complexity that scales linearly with the number of parties .
quantum entanglement provides many advantages beyond classical limits , including quantum communication , computation , and information processing @xcite . yet , determining whether a given quantum state is entangled or not is a theoretically and experimentally challenging task @xcite .
in particular , the ideal approach of reconstructing the full quantum state via quantum tomography is practically infeasible for all but the smallest systems .
an elegant solution to this problem , known as entanglement witnessing , relies on the geometry of the set of all non - entangled ( separable ) quantum states @xcite .
since these states form a convex set , it is always possible to find a hyperplane such that a given entangled state lies on one side of the hyperplane , while all separable states are on the other side , see fig .
[ fig : wpc ] .
this hyperplane is a so - called _ entanglement witness _ ( ew ) and corresponds to a joint observable that has a bounded expectation value over all separable quantum states .
any quantum state that produces a value beyond the bound must therefore be entangled .
this simplification , however , comes at a cost : first , different entangled states in general require different entanglement witnesses to be detected ; second , not every ew can be practically realized , i.e. , can be decomposed in terms of operators corresponding to available local measurement devices ( see also refs .
@xcite for examples of the reverse procedure : constructing ews from local observables ) ; third , when such a decomposition is possible , it might require multiple measurement devices ( with multiple settings ) to be implemented ; and fourth , witnessing bounds can be elusive in the presence of experimental imperfections .
consequently , one would ideally like to construct ews that have a simple decomposition and , at the same time , detect a large set of entangled states .
there are three main techniques to improve ews .
first , to use nonlinear ews by addition of nonlinear terms to the original witness operator @xcite .
second , by using collective measurement of ews on multiple copies of quantum states @xcite .
third , by optimizing a given witness to tighten the bound on the statistics of separable states as much as possible @xcite .
the latter , which we refer to as standard entanglement witnessing ( sew ) , is the most common procedure and enjoys the property that it can be used as a complementary procedure to the first two techniques . in sew one
first evaluates the supremum ( infimum ) expectation value of the witness observable for all separable states .
the witness operator is then decomposed into local measurements , such that the expectation value can be computed from the combined local measurement statistics .
a comparison against the corresponding upper ( lower ) bound for separable states establishes the entanglement of the tested state .
crucially , a significant amount of information from these measurements remains unused by combining the statistics . in this letter , we introduce and demonstrate a novel approach for witnessing quantum entanglement that makes use of this otherwise unused additional information to seamlessly and easily enhance any existing witnessing protocol .
our method , which we call _ ultrafine entanglement witnessing _ ( uew ) , relaxes the requirements on the test operators , which facilitates entanglement detection of a wide range of states , even if no witness decomposition in the common sense is provided .
moreover , our approach makes it possible to detect entanglement using only a simple measurement device for each party with the minimal number of measurements .
we implement this technique experimentally on two - qubit entangled states shared by alice and bob , each of whom has access to a fixed three - outcome measurement device .
finally , we show that uew can straightforwardly be extended to multipartite scenarios , with an experimental complexity that scales linearly with the number of parties involved .
standard entanglement witnessing relies on the fact that the set of all separable states , @xmath0 , is the collection of all convex combinations of pure product states . as a fruitful consequence of this convexity
, we can identify quantum states outside @xmath0 ( i.e. , entangled states ) as follows .
the hahn - banach theorem implies that for every entangled state @xmath1 there exists a hyperplane cutting through the state space that separates @xmath1 from the set of separable states , see fig .
[ fig : wpc ] .
mathematically , there exists a hermitian operator @xmath2 such that @xmath3 for all @xmath4 , while @xmath5 @xcite .
the operator @xmath2 is called an _
entanglement witness_. powerful ews are most commonly constructed by optimizing a hermitian ( and possibly completely positive ) _ test _ operator @xmath6 over the set of separable states as @xcite @xmath7 where @xmath8 is the identity operator , and @xmath9 .
indeed , it is sufficient to optimize only over pure product states @xmath10 @xcite .
one can also employ a similar recipe using the infimum value @xmath11 .
this optimization procedure can be geometrically understood as translating the hyperplane corresponding to the test operator until it is tangent to the set of separable states , as illustrated in fig .
[ fig : wpc ] .
hence , there exists a pure product state , called the _ optimal point _ ,
for which @xmath12 @xcite .
the resulting ew is said to be the _
finest _ witness in the sense that any further shift of its corresponding hyperplane will lead to an operator whose expectation value becomes negative for some separable states , thus , violating the proper witnessing conditions @xcite .
it is , however , possible to significantly increase the detection power of any test operator by taking into account additional constraints and information about the states under investigation , which effectively reduces the size of the set of viable separable states , see fig .
[ fig : wpc ] .
these constraints reflect physical restrictions on the measurement statistics that can be produced by separable states in certain situations .
similar considerations have previously been applied in the context of non - gaussianity detection @xcite .
consider , for example , a system composed of two spin-@xmath13 particles , which , in a measurement along the @xmath14-axis , are always found either both with spin up , or both with spin down .
there is a large number of separable states that can not produce such statistics and can thus be excluded from the optimization procedure for any witness aiming to detect the potential entanglement .
crucially , the required information about the state is already available , but not used , in almost every standard witnessing experiment . consider a hermitian operator @xmath15 corresponding to some physical observable .
just like a witness , @xmath16 corresponds to a hyperplane cutting through the space of all quantum states , splitting it into two half - spaces @xmath17 and @xmath18 , where @xmath19 is a real - valued free parameter . depending on the choice of constant @xmath19 , the hyperplane @xmath16 may or may not cut through the set of separable states , defining the two closed convex subsets @xmath20 and @xmath21 .
clearly , whenever one of the sets is empty the other one coincides with the set of all separable states , and hence , our method reduces to sew .
therefore , in the following , we will consider parameter values for which both @xmath22 and @xmath23 are nonempty .
using the test operator @xmath6 one can now construct two ews , @xmath24 and @xmath25 , optimal to the sets @xmath22 and @xmath23 , respectively , by replacing @xmath26 of eq . with @xmath27 and @xmath28 .
consequently , a state @xmath1 is entangled if @xmath29 [ [ lemma1 . ] ] lemma 1 .
+ + + + + + + + given a test operator @xmath6 with optimal points to the sets @xmath0 and @xmath30 as @xmath10 and @xmath31 for @xmath32 , respectively , a. if @xmath33 , then @xmath34 and @xmath35 , i.e. , @xmath36 .
furthermore , @xmath37 . b. if @xmath38 , then @xmath39 and @xmath40 , i.e. , @xmath41 .
furthermore , @xmath42 .
we point the interested reader to the supplemental material @xcite for the proof of lemma 1 .
lemma 1 shows that the optimal point from sew remains optimal for one of the two sets @xmath30 , while for the other set the optimal point lies on the hyperplane @xmath16 , as visualized in fig .
[ fig : wpc ] .
this in particular implies that for a given @xmath19 one of the conditions in eq . is advantageous over sew .
in addition , eq . together with lemma 1 imply that uew and sew are equivalent only in the special case that the constraint value @xmath19 is chosen exactly to match the expectation value of the constraint operator in the sew optimal point , i.e. , for @xmath43 .
this , in turn , implies that our strategy never performs worse than sew .
accordingly , we also obtain the following useful results the proofs of which are provided in the supplemental material @xcite .
[ [ theorem1 . ] ] theorem 1 .
+ + + + + + + + + + for a given constraint value @xmath19 , the optimal state @xmath44 to the test operator @xmath6 is a pure state with @xmath45 .
[ [ theorem2 . ] ] theorem 2 .
+ + + + + + + + + + the necessary condition for the separable operators @xmath16 and @xmath6 to detect entanglement via uew is that @xmath46\neq0 $ ] .
[ [ corollary1 . ] ] corollary 1 .
+ + + + + + + + + + + + if @xmath47 and @xmath48 are product operators , then @xmath49 and @xmath50 ( @xmath51,@xmath52 ) must be jointly non - measurable operators . in sew it
is necessary that the test operator @xmath6 has an entangled eigenspace , since otherwise the supremum ( and infimum ) expectation values could be obtained by separable eigenstates .
theorem 2 and corollary 1 show that our approach relaxes this requirement on the test operators and can be implemented with two separable ( or even product ) hermitian operators .
notably , corollary 1 implies that each party must use a measurement device with at least three outcomes , independent of the hilbert space dimension of the system .
this property makes uew very efficient for experimental applications by significantly reducing the number of measurements required for detection of particular classes of bipartite and multipartite entangled states .
consider two parties , alice and bob , each of whom has access to a single measurement device @xmath53 , and @xmath54 , respectively . in the simplest scenario ,
alice and bob can use their measurement devices to detect a range of entangled states using the following protocol .
a. choose a constraint operator of the form @xmath55 .
b. choose a test operator of the form @xmath56 for @xmath57 ; note that the conditions of theorem 2 and corollary 1 must apply to @xmath16 and @xmath6 .
c. for each value @xmath19 , compute @xmath58 constrained to @xmath59 .
d. the result is the concave curve of @xmath60 versus @xmath19 , called the _
separability curve_. any point corresponding to a quantum state above this curve indicates either of the conditions in eq . , and thus implies its entanglement .
corollary 1 implies that @xmath61 and @xmath62 must not be jointly measurable and thus , @xmath63 and @xmath64 can not be dichotomic measurements . in the case where alice and bob use the same three - outcome measurement device with @xmath16
defined as above , there are six different separability curves .
one of these curves is shown in fig .
[ fig : results ] together with the expectation and experimental observations for a family of entangled states .
we emphasize here that , in general , arbitrary povm elements can be combined to form complex constraints and test operators .
moreover , one might consider using multiple constraints , which would lead to _
separability hypersurfaces_. hence , there is a large number of different possible ways to implement uew with a single measurement device .
is implemented directly by a partially polarizing beam splitter ( ppbs ) with reflection coefficients @xmath65 for horizontal and @xmath66 for vertical polarization .
the other povm elements @xmath67 and @xmath68 are implemented using a set of quarter - waveplate ( qwp ) , half - waveplate ( hwp ) and a polarizing beam splitter ( pbs ) .
( b ) visualization of our three - outcome povm in the xz - plane of the bloch sphere . ]
experimentally we consider two - qubit states encoded in the polarization of single photons , shared between alice and bob .
they are both equipped with a three - outcome measurement device as shown in fig .
[ fig : setup ] , which implements the povm elements @xmath69 where @xmath70 with an arbitrary phase @xmath71 and @xmath72 .
alice and bob then choose the test and constraint operators as @xmath73 which implies that @xmath74 . the corresponding separability curve for the case @xmath75 is shown in fig .
[ fig : results ] , together with a density plot of @xmath76 separable states , randomly sampled from the uniform distribution of pure states on alice s and bob s local bloch spheres .
equations and imply that in our experiment the constraint corresponds to a limit on the vertical polarization component . and @xmath77 . the separability curve , which represents the largest expectation values of the test operator obtainable from separable states , is shown in orange , while the maximal values obtainable with entangled states are represented by the blue curve .
the blue data points correspond to 21 equispaced entangled states and include @xmath78 error bars .
the black dots , obtained from randomly sampled pure product states with uniform distributions over the local bloch spheres , illustrate the density of separable states with respect to the test and constraint operators .
note that the point where the orange and blue curves meet is the optimal point one would obtain for @xmath6 in sew .
this highlights that sew is unable to detect any entanglement with our choice of a separable test operator . ]
starting from a general pure state we find that the maximal violation of the bound is obtained by states of the form @xmath79 satisfying @xmath80 .
equation thus implies that @xmath81 for a given value of @xmath19 .
maximizing the expectation value of the test operator is then equivalent to maximizing the overlap @xmath82/2 $ ] .
since the last term is independent of the chosen state , we can assume @xmath83 which reduces the problem to maximizing @xmath84 constrained to @xmath85
. for @xmath75 one then obtains @xmath86 , @xmath87 , and the maximum expectation value of the test operator @xmath88 note that these values are independent of @xmath71 .
figure [ fig : results ] shows the theoretical maximal violation curve , together with our experimental results for @xmath77 .
as we have seen , our approach allows for entanglement detection using separable test and constraint operators , which is not possible in sew .
in addition , uew can be used to seamlessly and easily improve any existing standard witnessing experiment : consider two parties implementing a standard ew using the test operator @xmath6 , decomposed into local povm elements as @xmath89 with @xmath90 .
their aim is to violate the inequality @xmath91 using an entangled state @xmath1 .
in such a scenario uew allows us to tighten the bound , thus making the entanglement detection more experimentally robust .
this is achieved by choosing an arbitrary pair ( or a subset ) of povm elements as a constraint , say @xmath92 , and computing @xmath93 using already measured data . then , one re - optimizes @xmath6 over the set of pure product states with @xmath94 to obtain the new tighter bound @xmath95 and tests the inequality @xmath96 .
notably , there are many ( in fact , infinitely many ) constraints that could be constructed from the initially measured povm elements .
we now show how the simple procedure for uew outlined above can be directly extended to the multipartite scenario where a quantum state is shared between multiple parties .
sew for this case has been demonstrated in theory and experiment in refs .
@xcite .
consider a @xmath97-qubit system shared between @xmath97 agents , each of them having a three - outcome measurement device with povm elements given by eq . .
moreover , suppose that an arbitrary @xmath98-partitioning of the system has been chosen as @xmath99 , where each party @xmath100 is a subset of the index set @xmath101 , containing @xmath102 agents ( and hence , subsystems ) , so that @xmath103
. moreover , the list of parties is ordered such that @xmath104 .
now , the agents chose the test and constraint operators as @xmath105 and @xmath106 which implies that @xmath107 .
as a proof - of - principles , suppose that @xmath108 .
in the supplemental material @xcite , we prove that the maximum separable bound for a partition @xmath109 is given by @xmath110 since the test and constraint operators are invariant under the exchange of agents between different parties , so is the bound @xmath111 .
three cases are of particular interest .
first , if @xmath112 , then no partitioning has been made and @xmath113 represents the maximum expectation value of the test operator @xmath6 over _ all _ @xmath97-partite quantum states and hence , this bound can not be violated by any quantum state .
second , if @xmath114 the resulting bound corresponds to the bipartitions for which there is one subsystem in one party and @xmath115 subsystems in the other .
one can easily see that for any bipartition with @xmath116 , @xmath117 .
consequently , any state violating this bound is entangled within all bipartitions and thus , genuinely @xmath97-partite entangled . finally , if @xmath118 each party constitutes one agent corresponding to the partition with the highest resolution , i.e. , @xmath119 .
thus , any state violating the bound @xmath120 is partially entangled . versus the cardinality of the largest party @xmath121 for @xmath122 , where each @xmath97 is shown in a different color . for a state of interest with @xmath108 , a violation of
the bound @xmath123 proves its ( partial ) entanglement , while a violation @xmath124 proves its genuine @xmath97-partite entanglement .
the inset shows the case @xmath125 with the four black dots representing the four subsystem and the purple shaded boxes visualize the maximally allowed entangled subset . ]
figure [ fig : multipartite ] shows @xmath126 versus the cardinality of the largest party @xmath121 for @xmath122 . as @xmath97 increases it becomes increasingly harder to detect genuine @xmath97-partite entanglement with the simplest version of our approach .
however , detecting partial entanglement by violating the bound @xmath123 remains experimentally feasible for larger @xmath97 .
this example shows that our approach can be extended to the multipartite case , where it allows for simple entanglement detection with a number of measurements that scales as @xmath127 with the number of agents @xmath97 , as opposed to the tomographic methods or bell tests for which the number of measurements scales exponentially with the number of qubits @xcite .
in fact , current ews require at least @xmath128 measurements for each agent , where @xmath129 is the minimum hilbert space dimension of the subsystems @xcite , while our technique provides the possibility of detecting entangled states using only three - outcome measurements independent of the hilbert space dimensionality . in conclusion ,
we have introduced a novel procedure for witnessing quantum entanglement using additional information that is typically already available in a standard witnessing experiment .
our ultrafine entanglement witnessing relaxes the requirements on the test operators and allows for entanglement detection with a much smaller number of measurements compared to the standard entanglement witnessing .
in particular , our procedure makes it possible to witness entanglement using a single fixed measurement - device for each party .
this is a considerable experimental simplification which potentially allows for faster and more precise detection of entanglement compared to the existing protocols .
we have demonstrated this in practice for a family of two - qubit entangled states using two fixed three - outcome povms .
we also showed that our method always performs at least as well as the standard procedure and seamlessly and easily improves it .
we have described a scalable experimental protocol that generalizes to higher dimensional and multipartite quantum systems , and showed that , in its simplest form , the number of measurements required for this protocol scales linearly with the number of agents . [ [ acknowledgements . ] ] acknowledgements. + + + + + + + + + + + + + + + + + + the authors gratefully acknowledge a. g. white for helpful discussions and t. vulpecula for experimental assistance .
this project was supported by the arc centres for quantum computation and communication technology ( ce110001027 ) and engineered quantum systems ( ce110001013 ) .
100 m. a. nielsen and i. l. chunang , _ quantum computation and quantum information _ , ( cambridge university press , cambridge , 2000 ) .
r. horodecki , p. horodecki , m. horodecki , and k. horodecki , rev .
mod .
phys . * 81 * , 865 ( 2009 ) .
l. gurvits , in _ proceedings of the thirty - fifth annual acm symposium on theory of computing , 2003 _ , ( acm , new york , 2003 ) , p. 10 .
s. gharibian , quantum info . comput . * 10 * , 343 ( 2010 ) .
m. horodecki , p. horodecki , and r. horodecki , phys .
lett .
a * 223 * , 1 ( 1996 ) .
m. horodecki , p. horodecki , and r. horodecki , phys .
lett .
a * 283 * , 1 ( 2001 ) .
d. chruciski and g. sarbicki , j. phys .
a : math . theor . * 47 * , 483001 ( 2014 ) .
s. yu and n. liu , phys .
rev .
lett .
* 95 * , 150504 ( 2005 ) .
h. gholipour and f. shahandeh , phys .
rev .
a. * 93 * , 062318 ( 2016 ) .
o. ghne and n. ltkenhaus , phys .
rev .
lett . * 96 * , 170502 ( 2006 ) .
p. horodecki , phys .
rev .
a * 68 * , 052101 ( 2003 ) .
m. lewenstein , b. kraus , j. i. cirac , and p. horodecki , phys .
rev .
a * 62 * , 052310 ( 2000 ) .
j. sperling and w. vogel , phys .
rev .
a * 79 * , 022318 ( 2009 ) .
f. shultz , j. math .
phys . * 57 * , 015218 ( 2016 ) .
r. filip and l. mita , ladislav , phys .
rev .
lett . * 106 * , 200401 ( 2011 ) .
see the supplemental material at ... for detailed proofs of lemma 1 , theorem 1 , theorem 2 , corollar 1 , and derivation of eq . .
j. sperling and w. vogel , phys .
rev .
lett . * 111 * , 110503 ( 2013 ) .
s. gerke , j. sperling , w. vogel , y. cai , j. roslund , n. treps , and fabre , c. , phys .
rev .
lett . * 114 * , 050501 ( 2015 ) .
g. tth and o. ghne , phys .
rev .
lett . * 94 * , 060501 ( 2005 ) .
o. ghne , p. hyllus , d. bruss , a. ekert , m. lewenstein , c. macchiavello , and a. sanpera , j. mod . opt . * 50 * , 1079 ( 2003 ) . |
inside rotationally excited baryons , qcd leads to the formation of diquarks well separated from the remaining quark . at this separation
the scalar , spin independent , confining part of the effective qcd potential is dominant .
since qcd forces are also flavor independent , the force between the quark ( @xmath3 ) and the diquark ( @xmath4 ) inside an excited baryon is essentially the same as the one between @xmath3 and the antiquark @xmath5 inside an excited meson .
thus , the approximate spin - flavor independence of hadronic physics expressed by @xmath0 symmetry gets extended to the @xmath6 supersymmetry through a symmetry between @xmath5 and @xmath7 , resulting into the parallelism of mesonic and baryonic regge trajectories .
various aspects and implications of this approximate effective supersymmetry and its breaking are discussed in the first section . in the next section
we discuss the skyrmion approach to effective hadronic lagrangians . in the early @xmath8 s skyrme@xmath9 proposed a model in which baryons can be understood as solitons in an effective lagrangian for bosons .
he added a fourth term to the non - linear sigma model@xmath10 and proposed that the topological soliton solution of the meson field could be identified as a nucleon , and that the winding number is the baryon number .
this pioneering work of skyrme was largely ignored until two decades later when pak and tze@xmath11 , and gipson and tze@xmath12 re - expressed skyrme s ideas in the modern context of qcd and studied their implications for weak interactions . in an attempt to verify one of the skyrme
s conjectures that the winding numbers are the baryon numbers , balachandran and his collaborators@xmath13 considered a system of interacting quarks and solitons .
using some results of goldstone and wilczek@xmath14 , they showed that , in the presence of the classical soliton field , the dirac sea of quarks carries a baryonic current which is same as the topological current .
the possible connections of qcd to effective meson theories are due to t hooft@xmath15 and witten@xmath16 .
they showed that in the limit of large number of colors @xmath17 , qcd can be reduced to a theory of weakly interacting mesons and glue balls . using @xmath18 as an expansion parameter
, witten argued that to leading order , baryons emerge as solitons in the weakly coupled phase of mesons if some topological properties and stability conditions are assumed .
later witten@xmath19 showed that it is necessary to introduce a wess - zumino term in the effective chiral theory in order to predict some observed physical processes allowed by the non - abelian anomaly in qcd .
he also demonstrated that the gauged w - z action can lead to the identification of the topological current with the baryonic current . to verify another skyrme conjecture about topological solitons being fermions , he rotated a skyrmion adiabatically through @xmath20-radians and found that the skyrmion had picked up a phase of @xmath21 .
hence a skyrmion is a fermion if @xmath17 is odd .
afterwards , adkins , nappi and witten@xmath22 systematically studied some static properties of nucleons in an explicit model in which the skyrme lagrangian is invariant under @xmath23 .
these results agreed with experiment to 30 % .
since then , there has been a tremendous interest in the skyrme model .
a good number of modifications to the model have been proposed , such as including the effect of the pion mass@xmath24 , adding higher order terms@xmath25 , extending the symmetry to @xmath26@xmath27 , including vector mesons@xmath28 , hidden symmetry approach@xmath29 , bound state approach to strangeness@xmath30 , and non linear supersymmetric models and relation to skyrmions@xmath31 .
a recent review of supersymmetric chiral models was given by perelomov@xmath32 .
although the skyrme model provides a reasonably good description of baryons , there are many problems which remain to be solved . in the skyrme model ,
if the masses of the low lying baryons @xmath33 and @xmath34 are taken as input parameters , the predicted value of the pion decay constant @xmath35 is found to be too high ; or if the experimental value of @xmath35 is taken as an input , the predicted masses of the nucleons turn out to be too high . because the value @xmath35 is too large , many other predictions which depend on @xmath35
disagree with experiment .
the model also predicts a tower of baryon states with equal spin and isospin ( @xmath36 ) which do not exist in nature .
they are considered artifacts of the model . quark model , on the other hand , is regarded as an effective theory when the gluon degrees of freedom are integrated out . if both the skyrme model and the quark model are effective models of qcd , they should have common symmetry of hadrons , such as chiral symmetry , the hadronic approximate @xmath0 symmetry@xmath37 and @xmath6 symmetry@xmath38 have been established for the quark model but ways to restore these symmetries for the skyrme model have yet to be found . in this paper
we address some of these problems , give partial solutions to some of the difficulties by improving the skyrmion model for hadrons and suggest a strategy toward building a full fledged quantitatively viable skyrmionic description of hadronic physics .
supersymmetry is a symmetry between fermions of half odd integer spin and bosons of integer spin . in supergravity ,
for example , we have an action that is invariant under operations transforming fields of spin @xmath39 , @xmath40 , @xmath41 , @xmath42 , and @xmath43 among themselves@xmath44 . here
we recall that low lying hadrons have the same range of spin values with @xmath45 and @xmath42 baryons interlaced with @xmath46 and @xmath43 mesons .
groups that transform @xmath47 and @xmath40 mesons into each other and simultaneously mix @xmath45 and @xmath42 baryons have been proposed long ago@xmath48 .
this kind of symmetry had to be broken symmetry since no degeneracy is observed between particles of different spin .
however , a cursory glance at the chew - frautschi plot of hadronic trajectories@xmath49 reveals the following features : * \(a ) all trajectories with @xmath50 ( hadronic spin ) versus @xmath51 ( @xmath52 hadronic mass ) are approximately linear . *
\(b ) leading mesonic trajectories associated with lowest spin @xmath53 and @xmath43 are parallel among themselves . *
\(c ) leading baryonic trajectories associated with @xmath45 , @xmath42 are also parallel among themselves .
* \(d ) mesonic and baryonic trajectories are approximately parallel to each other with a universal slope of @xmath54 ( gev)@xmath55 .
* \(e ) the separation between mesonic trajectories is nearly the same as the one between baryonic trajectories .
now , properties ( b ) and ( c ) suggest the existence of a phenomenological symmetry between mesons of different spin , which also operates on baryons with different spin . for hadrons composed of light quarks @xmath56 , @xmath57 and @xmath58 , this symmetry is expressed by the group @xmath59 where @xmath60 describes the rotational excitations on the leading trajectories , and the spin - flavor group @xmath0 classifies the lowest elements of the trajectories into particle multiplets .
the property ( d ) on the other hand , tells us that there is a new kind of symmetry ( supersymmetry ) between the bosonic mesons and fermionic baryons .
the universal regge slope for hadrons is a supersymmetric observable .
the meaning of property ( e ) is that , the physical mechanism that breaks the @xmath0 symmetry must also be responsible for breaking of its supersymmetric extension .
finally , from property ( a ) we infer that the potential binding the quarks is approximately linear and we have to apply relativistic quantum mechanics appropriate to light quarks while the schroedinger nonrelativistic theory is sufficient for the description of quarkonium systems for heavy quarks .
according to the quark model of gell - mann and zweig , mesons and baryons are described respectively by bound ( @xmath61 ) and @xmath62 systems .
any symmetry between mesons and baryons must correspond at the quark level to a supersymmetry between @xmath5 ( antiquarks ) and bound @xmath7 states ( diquarks ) .
now @xmath5 , with @xmath45 and unitary spin associated with the triplet @xmath63 representation of the flavor @xmath64 belongs to the @xmath65 representation of @xmath0 .
the low lying baryons are in its @xmath66 representation .
since @xmath66 is contained in @xmath67 , the diquark with @xmath47 or @xmath41 must be in the symmetric @xmath68 representation of @xmath0 .
hence the hadronic supersymmetry we are seeking must transform the @xmath69 and @xmath68 @xmath0 multiplets , both color antitriplets into each other and therefore must be @xmath70 dimensional with @xmath71 fermionic and @xmath72 bosonic states .
this supergroup is @xmath6 .
it was first introduced by miyazawa@xmath73 as a generalization of the hadronic @xmath0 symmetry , following earlier attempts by hwa and nuyts@xmath74 . the @xmath75 symmetry that also implies @xmath76 symmetry will transform the meson ( @xmath77 in general not only to baryons @xmath62 and antibaryons @xmath78 but also to exotic mesons @xmath79 ( @xmath80 , and @xmath81 ) that belong to the @xmath0 representations @xmath82 and @xmath83 .
the @xmath84 and @xmath85 are @xmath86 and @xmath87 mesons while the ( @xmath83 ) also includes mesons with spin @xmath88 and isospin @xmath43 .
all the low lying hadrons will now be in the adjoint representation of @xmath6 with both spin and isospin taking values @xmath89 and @xmath43 .
the next introduction of supersymmetry into physics was within the context of dual resonance models that evolved later into string models .
these theories due to ramond@xmath90 and neveu and schwarz@xmath91 lead naturally to linear baryon and meson trajectories that are parallel .
the string models were not local and they were not relativistic in four dimensions .
examples of renormalizable relativistic local quantum field theories involving @xmath92 and @xmath41 fields with interaction were first constructed by wess and zumino@xmath93 following initial attempts by other authors@xmath94 .
wess and zumino based their work on the super - poincar algebra that is a supersymmetric generalization of the infinitesimal poincar group and sits in the superconformal algebra .
the final step of incorporating @xmath95 and @xmath96 fields in a local relativistic field theory was taken by the discoverers of supergravity@xmath44 and extended supergravity who were able to supersymmetrize einstein s general relativity and kaluza - klein theories .
all these local supersymmetric quantum field theories have fascinating convergence and symmetry properties but they are far from describing properties of hadrons or even quarks and leptons .
if these fundamental fields have any physical reality at all , they may be associated with preons ( or haplons ) that would be hypothetical constituents of quarks , leptons and fundamental gauge bosons .
returning to the more concrete world of hadrons , we may try to see if the phenomenological approximate group @xmath0 and its supersymmetric extension can be justified within the standard theory of colored quarks interacting through gluons that are associated with the color gauge group @xmath97 .
the justification of @xmath0 and the derivation of its breaking was given by georgi , glashow and de rujula@xmath98 .
on the other hand , a phenomenological supersymmetry in nuclear physics between odd and even nuclei was discovered and formulated by means of supergroups@xmath99 @xmath100 by balantekin , bars and iachello@xmath101 .
it was natural to see if such groups could also describe hadronic supersymmetry .
in the first part of this paper we propose to extend the approach of georgi et al . to see
how far it can provide a basis for the existence of an approximate hadronic supersymmetry .
it turns out that most of the ingredients for this approach are already in the literature .
the key concepts are : * \(a ) * the approximate validity of the string theory * as an approximation to qcd .
this was shown by the lattice gauge theory as a strong coupling approximation to qcd , following the pioneering work of wilson@xmath102 and also by the elongated bag model of johnson and thorn@xmath103 following the bag model@xmath104 approximation to qcd .
* \(b ) * the emergence of the diquark structure * in qcd through the string approximation .
this was done by eguchi@xmath105 and also by johnson and thorn@xmath103 . *
\(c ) the vector ( spin dependent ) nature of the coulomb part together with the scalar ( spin independent ) nature of the confining part of the @xmath106 potential .
these properties were collectively worked out by many authors using both perturbation theory and lattice gauge theory methods@xmath107 . *
\(d ) the necessity for introducing exotic @xmath79 mesons in qcd through the bag approximation@xmath108 , the string picture@xmath109 or the confining potential model@xmath110 . *
\(e ) deviations from linearity of regge trajectories within the context of qcd@xmath49 .
once all the parts of this jigsaw puzzle is put together , a rather simple picture emerges .
the diquark behaves very much like the antiquark in the strong coupling regime because qcd forces are flavor independent and the confining part of qcd potential is spin independent .
this immediately leads to a @xmath111 effective supersymmetry at large separation . at short distances ,
spin is approximately conserved because of asymptotic freedom but spin independence gets broken through one - gluon exchange in the perturbation theory regime .
the diquark structure also disappears at short distance , leading to the breaking of both @xmath0 and @xmath6 at low energies .
there is another difficulty associated with a supersymmetric extension of @xmath0 noted by salam and sthrathdee@xmath112 . because of the anticommutativity of the grassmann numbers and also of supercharges , antisymmetrical representations of @xmath0 like @xmath113 for diquarks and @xmath114 for baryons will occur in hadronic supermultiplets . in the case of colored quarks , however , it is possible to introduce non associative grassmann numbers @xmath115 and @xmath116 constructed out of octonion units@xmath117 .
they transform like a triplet under the @xmath64 subgroup of the automorphism group @xmath118 of octonions . then the octonionic quarks @xmath119 and @xmath120 will commute unlike @xmath121 and @xmath122 that are anticommutative ( @xmath123 and @xmath124 are combined spin - flavor indices ) .
this procedure which is the basis of color algebra not only converts antisymmetrical spin - flavor group representations into symmetrical ones , but suppresses the color sextet @xmath65 representations for diquarks that would otherwise be allowed under the rule for the preservation of the pauli principle for the colored quark states . with the introduction of octonionic quarks , diquarks , mesons and baryons can all be viewed as elements of an octonionic superalgebra@xmath38 .
the simplest supersymmetric hamiltonian is obtained starting from semirelativistic dynamical models of quarks and diquarks already used by lichtenberg et al.,@xmath125 for an approximate calculation of baryonic masses . in the following subsections
we will first give a derivation of hadronic mass formulae using semi - relativistic and relativistic formulation .
later , in section 2 using the skyrmion model we will show that we obtain the same mass formulae .
. in qcd ,
both @xmath106 and @xmath127 forces are attractive , the color factor of the latter being half of that of the former .
thus , the formation of diquarks inside a baryon is a definite possibility .
various strong coupling approximations to qcd , like lattice gauge theory@xmath102 @xmath107 , t hooft s @xmath128 approximation ( when @xmath17 , the number of colors is very large ) , or the elongated bag model@xmath103 all give a linear potential between widely separated quarks and an effective string that approximates the gluon flux tube .
in such a theory , eguchi@xmath105 has shown that it is energetically favorable for three quarks in a baryon to form a linear structure with a quark in the middle and two at the ends or , for high rotational excitation , a bilocal linear structure with one quark at one end and a diquark at the other end .
this was reconfirmed by johnson and thorn independently in the bag model when the bag is deformed and elongated in a rotationally excited baryon@xmath103 . thus if we move along a leading baryon regge trajectory to a region of high @xmath50 , we are likely to find a baryon as a bilocal object consisting of a quark and a diquark instead of a trilocal object that represents a ground state baryon more accurately . on the other hand
a meson is a bilocal object consisting of a quark and an antiquark interacting via a linear potential at large separation .
consider then a bilocal object with constituents with respective masses @xmath129 , @xmath130 , spins @xmath131 , @xmath132 and color representations @xmath133 and @xmath134 .
one of the constituents can be a quark @xmath3 or the anti - diquark @xmath135 ( both color triplets ) while the other can be an antiquark @xmath5 or a diquark @xmath136 which are both color antitriplets .
the qcd force between the two constituents will be flavor independent and it will consist of two parts : a coulomb like part @xmath137 which transforms like the time component of a @xmath138-vector ( due to the exchange of single @xmath139 gluons at short separations ) and a confining part @xmath140 which is largely a relativistic scalar , is spin independent and is due to the exchange of a great many gluons that form a flux tube or an elongated bag at large separation of the constituents .
let @xmath141 and @xmath142 be the center of mass momenta of the two constituents .
the quantity which is canonically conjugate to the relative coordinate @xmath143 is @xmath144 ignoring the center of mass motion , following lichtenberg et al.@xmath38 , we can write a semi - relativistic wave equation for the wave function @xmath145 of the bilocal object with energy eigenvalues @xmath146 , namely @xmath147^{\frac{1}{2 } } + [ ( m_2 + \frac{1}{2}~v_s)^2 -\nabla^2)]^{\frac{1}{2 } } \ } \psi_{12 } \label{eq : bir}\ ] ] the scalar and vector potentials are given by @xmath148 where @xmath149 is the color factor , @xmath150 is the strong coupling constant at the energy @xmath146 , and the spin dependent part of the vector potential is the hyperfine structure correction due to gluon exchange with @xmath151 .
we see that at large @xmath152 , neglecting the mass difference @xmath153 , we find the same equation for both the @xmath154 and the @xmath155 system , except for the presence of the hyperfine term that breaks the symmetry between @xmath5 and @xmath136 . to this approximation
, we can transform the second constituent @xmath5 into @xmath136 and vice - versa without changing the energy eigenvalue @xmath156 .
this means that the system admits the approximate @xmath6 supersymmetry transformation @xmath157 @xmath158 in addition to the @xmath0 transformation @xmath159 @xmath160 the breaking of both @xmath0 and @xmath6 is due to the hyperfine term while the supersymmetry is further broken by the quark - diquark mass difference @xmath161 .
we could also have brought into play the wave functions @xmath162 and @xmath163 of the diquark and antidiquark at point @xmath164 in the field of the quark at point @xmath164 .
the masses @xmath129 and @xmath130 must then be replaced by the reduced masses @xmath165 in this case , the wave function belongs to the fundamental @xmath70 dimensional representation of @xmath6 and the hamiltonian commutes with the supersymmetry transformation eq.([eq : iki ] ) of the wave function except for the difference in the reduced masses @xmath166 and @xmath167 and the hyperfine structure term in the hamiltonian .
the breaking of supersymmetry results in the mass difference @xmath168 at high energies both for baryons and mesons . at low energies ,
the baryon mass becomes a trilocal object with three quarks and the mass splitting is given by @xmath169 where @xmath129 , @xmath130 and @xmath170 are the masses of the three different quark constituents .
going back to the first formulation which is more symmetrical , we can write the hamiltonian associated with eq.([eq : bir ] ) in a @xmath6 covariant form : we have four cases for @xmath171 , namely @xmath172 , @xmath173 , @xmath174 and @xmath175 that represents mesons , exotic mesons , baryons and antibaryons with respective dimensions @xmath176 , @xmath177 , @xmath178 , and @xmath179 that all fit into the following @xmath180 adjoint representation of @xmath6 : @xmath181 the wave equation for the hadronic wavefunction @xmath182 can now be written as @xmath183+\mbox{\boldmath$s$ } \psi \mbox{\boldmath$s$ } \label{eq : sekiz}\ ] ] where @xmath184 is the diagonal matrix @xmath185 with @xmath186^{\frac{1}{2 } } \nonumber\ ] ] @xmath187^{\frac{1}{2}}\ ] ] and @xmath188 the second , spin dependent term on the right hand side of eq.([eq : sekiz ] ) is a symmetry breaking term for both @xmath0 and @xmath6 .
the first term is @xmath0 symmetrical .
it also preserves the supersymmetry in the limit @xmath189 .
hence if the quark - diquark mass difference and the spin dependent terms are neglected , eq.([eq : sekiz ] ) is invariant under the @xmath6 infinitesimal transformation @xmath190 \label{eq : onbir}\ ] ] where @xmath191 where @xmath192 and @xmath33 have elements @xmath193 and @xmath194 respectively in eq.([eq : uc ] ) while the rectangular matrices @xmath195 and @xmath196 have elements @xmath197 and @xmath198 occurring in eq.([eq : iki ] ) . in that limit ,
the hamiltonian @xmath199 commutes with the generators of the transformation eq.([eq : onbir ] ) .
consider the hamiltonian of eq.([eq : bir ] ) .
we can write @xmath200 where @xmath201 is associated with the orbital excitation of the system . for high rotational excitations , the expectation value of @xmath152 is large , corresponding to a stretched string .
the angular momentum @xmath201 is also large .
the value of the centrifugal energy which is proportional to @xmath202 has a similarly large value .
since @xmath140 that is proportional to @xmath152 will also have a high absolute value , the constituent masses become negligible in the high relativistic limit . on the other hand ,
the radial excitation term @xmath203 can be neglected on the leading trajectory associated with the lowest radial energy .
the ground state energy eigenvalue @xmath204 of the hamiltonian can be estimated by using the heisenberg uncertainty principle .
this leads to the replacement of @xmath152 by @xmath205 and @xmath206 by @xmath207 then @xmath204 as a function of @xmath205 is minimized for the value of @xmath208 of @xmath205 .
the @xmath208 corresponds to the bohr radius for the bound state .
the confining energy associated with this bohr radius is obtained from the linear confining potential @xmath209 , so that the effective masses of the constituents become @xmath210 for a meson @xmath211 and @xmath212 are the current quark masses while @xmath213 and @xmath214 can be interpreted as the constituent quark masses .
note that even in the case of vanishing quark masses associated with perfect chiral symmetry , confinement results in non zero constituent masses that spontaneously break the @xmath215 symmetry of the @xmath56 , @xmath57 quarks .
let us illustrate this method on the simplified spin free hamiltonian involving only the scalar potential . in the center of mass system , @xmath216 , or @xmath217 .
the semi - relativistic hamiltonian of the system is then given by @xmath218 taking @xmath219 for the quark - antiquark system , we have @xmath220 where we have written the momentum part in spherical coordinates . putting @xmath221 for the @xmath106 system
we find @xmath222 by minimizing the function @xmath223 for @xmath56 and @xmath57 quarks , @xmath224 is small and can be neglected so that @xmath225\ ] ] which has a minimum for @xmath226 giving @xmath227 thus , we obtain a linear regge trajectory with @xmath228 also @xmath229 where @xmath230 arises from the quark spins .
experimentally @xmath231 for mesons giving the value @xmath232 gev for @xmath233 . a more accurate calculation ( see [ 4 ] )
gives @xmath234 the constituent quark mass can be defined in two ways @xmath235 or @xmath236 the first definition gives for @xmath237 , @xmath238 in the case of @xmath56 and @xmath57 quarks .
when the coulomb like terms are introduced in the simplified hamiltonian ( [ eq : e ] ) with negligible quark masses one obtains @xmath239\ ] ] with @xmath240 in the energy range around @xmath41 gev , @xmath241 is of order of unity .
estimates range from @xmath242 to @xmath243 .
minimization of @xmath204 gives @xmath244 where @xmath245 @xmath246 the minimum @xmath247 is obtained for @xmath248 , giving to second order in @xmath124 : @xmath249 linear regge trajectories are obtained if @xmath250 is negligible .
then for mesons @xmath251 the @xmath250 is negligible for small @xmath201 only if we take the lowest estimate for @xmath241 , giving @xmath252 for @xmath253 in the @xmath254 case . for mesons with @xmath56 , @xmath57 constituents , incorporating their spins through the breit term
we obtain approximately @xmath255 where @xmath256 is the constituent quark mass .
this gives @xmath257 the regge slope being of the order of @xmath258 an average meson mass of the same order is obtained from eq.([eq : ez ] ) in the linear trajectory approximation . to this approximation @xmath253
should be treated like a parameter rather than be placed by its value derived from qcd under varying assumptions .
using eq.([eq : alp ] ) for @xmath233 one gets a better fit to the meson masses by taking @xmath259 .
turning now to baryon masses , we must first estimate the diquark mass .
we have for the @xmath4 system @xmath260 that is slightly higher than the average meson mass @xmath261 here we note that @xmath204 is not very sensitive to the precise value of the qcd running coupling constant in the @xmath262 range .
taking @xmath263 changes @xmath264 from @xmath265 to @xmath266 .
note that eq.([eq : dq ] ) gives @xmath267 . for excited @xmath106 and @xmath155 systems
if the rotational excitation energy is large compared with @xmath233 , then both the @xmath268 and the coulomb term @xmath269 ( same for @xmath155 and @xmath106 systems ) can be neglected .
thus , for both ( @xmath155 ) [ excited baryon ] and @xmath106 [ excited meson ] systems we have eq.([eq : min ] ) , namely @xmath270 giving again eq.([eq : mina ] ) , i.e. @xmath271 as an explanation of hadronic supersymmetry in the nucleon and meson regge spectra .
we also have , extrapolating to small @xmath201 : @xmath272 for @xmath273 we find @xmath274 this relationship is same as the one proved in our earlier paper@xmath275 through the assumption that @xmath276 symmetry is broken by an operator that behaves like @xmath277 , @xmath278 member of @xmath279 representations of @xmath0 , which is true to @xmath280 .
it corresponds to a confined quark approximation with @xmath281 .
the potential model gives a more accurate symmetry breaking ( @xmath259 ) : @xmath282 with an accuracy of @xmath283 .
this mass squared formula arises from the second order iteration of the @xmath155 , @xmath106 dirac equation .
the factor @xmath284 comes from @xmath285 at this point it is more instructive to derive a first order mass formula . since the constituent quark mass @xmath256 is given by eq.([eq : mc ] ) @xmath286 , we have @xmath287 so that @xmath288 when the baryon is regarded as a @xmath155 system , each constituent gains an effective mass @xmath289 which was approximately the effective mass of the quark in the meson . hence , the effective masses of @xmath3 and @xmath136 in the baryon are @xmath290 the spin splittings for the nucleon @xmath33 and the @xmath34 are given by the breit term @xmath291 for the nucleon with spin @xmath40 the term @xmath292 gives @xmath293 while it has the value @xmath40 for @xmath34 with spin @xmath42 .
using the same @xmath184 for mesons and baryons which are both considered to be a bound state of a color triplet with a color antitriplet we can relate the baryon splitting @xmath294 to the meson splitting @xmath295 for which @xmath296 takes the values @xmath297 and @xmath298 .
hence we find @xmath299 which leads to a linear mass formula @xmath300 which is well satisfied , and has been verified before using the three quark constituents for the baryon@xmath275 .
the formation of diquarks which behave like antiquarks as far as qcd is concerned is crucial to hadronic supersymmetry and to quark dynamics for excited hadrons .
the splittings in the mass spectrum are well understood on the basis of spin - dependent terms derived from qcd .
this approach to hadronic physics has led to many in depth investigations recently . for extensive references
we refer to recent papers by lichtenberg and collaborators@xmath301 and by klempt@xmath302 . to see the symmetry breaking effect ,
note that the mass of a hadron will take the approximate form @xmath303 where @xmath304 and @xmath305 ( @xmath306 ) are respectively the constituent mass and the spin of a quark or a diquark .
the spin - dependent breit term will split the masses of hadrons of different spin values .
if we assume @xmath307 , where @xmath224 is the constituent mass of @xmath56 or @xmath57 quarks , and denote the mass of a diquark as @xmath308 , then this approximation gives @xmath309 @xmath310 @xmath311 eliminating @xmath224 , @xmath308 and @xmath184 , we obtain a mass relation @xmath312 which agrees with experiment to 13% .
let us consider a quark - antiquark system in the approximation that the potential is only a scalar . in the center of mass system , @xmath216 , or @xmath217 the semi - relativistic hamiltonian of the system
is then given by @xmath313 taking @xmath219 for the quark - antiquark system , we have @xmath314 where we have written the momentum part in spherical coordinates . for @xmath56 and @xmath57 quarks ,
@xmath224 is small and can be neglected so that @xmath315\ ] ] similar equations for a @xmath106 meson system were already proposed and solved numerically@xmath316 or exactly@xmath317 by several authors .
the eigenfunction @xmath318 for @xmath319 has eigenvalue @xmath320 such that @xmath321\psi = e^{2}\psi\ ] ] the differential equation can be solved exactly , and and the normalized eigenfunction is found to be @xmath322^{1/2 } r^{\ell}\exp[-\frac{b}{4}r^{2}]\times \nonumber \\ & & f(-n_{r}+1,{\textstyle \ell+\frac{3}{2},\frac{b}{2}r^{2 } } ) y_{\ell}^{m}(\theta,\phi ) \end{aligned}\ ] ] where @xmath323 is the confluent hypergeometric function , and @xmath324 , is the radial quantum number .
the eigenvalue is given by @xmath325 \label{eq : regge}\ ] ] therefore we obtain linear regge trajectories of slope @xmath326 when we make plots of @xmath201 versus @xmath327 .
the case of @xmath328 corresponds to the leading regge trajectory , and cases of @xmath329 , correspond to the parallel daughter trajectories .
note that iachello and his collaborators@xmath330 have obtained a similar mass formula based on algebraic methods .
starting from a spectrum generating algebra @xmath331 , we can write a chain of subalgebras @xmath332 they proposed that the hamiltonian can be expanded in terms of invariants of the chain of subalgebras @xmath333 where @xmath334 denotes one of the invariants of @xmath331 .
in a two - body problem , the spectrum generating algebra is @xmath335 , and in the total symmetric representation , one of the two chains of subalgebras is @xmath336 followed by @xmath337 with respective casimir invariants @xmath338 and @xmath339 , where @xmath340 corresponds to the vibrational mode . in order to derive the relativistic mass formula like eq.([eq : regge ] ) , one has to write a formula for the square of the mass , rather than the mass itself .
furthermore , without violating the dynamic symmetry , the hamiltonian can be written in terms of non - linear functions of the casimir invariants .
therefore , we can have @xmath341 comparing this mass formula with eq.([eq : regge ] ) , we find that the relativistic quark model suggests a relation of @xmath342 in eq.([eq : iachello ] ) .
however , the above algebraic method treats @xmath124 and @xmath343 as two independent parameters , and so the extra degree of freedom allows eq.([eq : iachello ] ) a better fit to the experimental data . the fitted parameters for mesons are found to be : @xmath344 ( gev)@xmath345 and @xmath346 ( gev)@xmath345 .
the inclusion of the coulomb - like term causes deviations from the linear trajectories at low energies and changes the relation between the parameters @xmath124 and @xmath343 .
it is important to carry out a calculation based on a better approximation to see if the empirical values of these parameters are compatible with our model .
the supergroup @xmath6 acts on a quark and antidiquark situated at the same point @xmath347 . at the point
@xmath348 we can consider the action of this supergroup with the same parameters , or one with different parameters . in the first case we have a global symmetry . in the second case ,
if we only deal with bilocal fields the symmetry will be represented by @xmath349 , doubling the supergroup . on the other hand ,
if any number of points are considered , with different parameters attached to each point , we are led to introduce a local supersymmetry @xmath6 to which we should add the local color group @xmath350 .
since it is not a fundamental symmetry , we shall not deal with the local @xmath6 group here .
however , the doubling the supergroup is useful for bilocal fields since the decomposition of the adjoint representation of the 728-dimensional @xmath6 group with respect to @xmath351 gives @xmath352 a further decomposition of the double supergroup into its field with respect to its center of mass coordinates , as will be seen below leads to the decomposition of the 126-dimensional cosets @xmath353 and @xmath354 into @xmath355 of the diagonal @xmath0 .
we would have a much tighter and more elegant scheme if we could perform such a decomposition from the start and be able to identify @xmath356 part of the fundametal representation of @xmath6 with the @xmath72-dimensional representation of the @xmath0 subgroup , which means going beyond the @xmath6 supersymmetry to a smaller supergroup having @xmath0 as a subgroup .
full description of a bilocal treatment and a minimal scheme where the bilocal treatment gets carried over unchanged into it will be a subject of another publication . setting up of an effective relativistic theory based on the dirac equation for the quark and the klein - gordon equation for the antiquark , exhibiting invariance under relativistic supersymmetry is needed .
this can be achieved by means of an effective wess - zumino type lagrangian constructed out of a gluon field interacting with a vector superfield formed by a quark and an antidiquark .
in what follows we will present some preliminary work in this direction .
low - lying baryons occur in the symmetric @xmath357 representation@xmath358 of @xmath0 , whereas the pauli principle would have led to the antisymmetrical @xmath359 representation .
this was a crucial fact for the introduction of color degree of freedom@xmath360 based on @xmath350 .
since the quark field transforms like a color triplet and the diquark like a color antitriplet under @xmath350 , the color degrees of freedom of the constituents must be included correctly in order to obtain a correct representation of the q - d system .
hadronic states must be color singlets .
these are represented by bilocal operators @xmath361 in the bilocal approximation@xmath362 that gives @xmath363 for mesons and @xmath364 for baryons . here
@xmath365 represents the antiquark situated at @xmath366 , @xmath367 the quark situated at @xmath368 , and @xmath369 the diquark situated at @xmath366 .
if we denote the c.m . and the relative coordinates of the consituents by @xmath370 and @xmath371 , where @xmath372 and @xmath373 with @xmath211 and @xmath212 being their masses , we can then write @xmath374 for the operator that creates hadrons out of the vacuum . the matrix element of this operator between the vacuum and the hadronic state @xmath375 will be of the form @xmath376 where @xmath377 is the free wave function of the hadron as a function of the c.m . coordinate and @xmath378 is the bound - state solution of the @xmath276 invariant hamiltonian describing the @xmath106 mesons , @xmath155 baryons , @xmath379 antibaryons and @xmath79 exotic mesons , given by @xmath380 \psi_{\alpha \beta}\ ] ] here @xmath381 in the c.m . system and @xmath224 and @xmath58 denote the masses and spins of the constituents , @xmath241 the strong - coupling constant , @xmath382 is the scalar potential with @xmath152 being the distance between the constituents in the bilocal object , and @xmath383
. the operator product expansion@xmath384 will give a singular part depending only on @xmath371 and proportional to the propagator of the field binding the two constituents .
there will be a finite number of singular coefficients @xmath385 depending on the dimensionality of the constituent fields .
for example , for a meson , the singular term is proportional to the progagator of the gluon field binding the two constituents .
once we subtract the singular part , the remaining part @xmath386 is analytic in r and thus we can write @xmath387 now @xmath388 creates a hadron at its c.m .
point @xmath370 equivalent to a @xmath389 , s - state of the two constituents . for a baryon
this is a state associated with @xmath3 and @xmath390 at the same point @xmath370 , hence it is essentially a 3-quark state when the three quarks are at a common location .
the @xmath391 can create three @xmath392 states with opposite parity to the state created by @xmath388 .
hence , if we denote the nonsingular parts of @xmath363 and @xmath393 by @xmath394 $ ] and @xmath395 $ ] , respectively , we have @xmath396|0 > = |m({\bf r } ) > + { \bf r } \cdot |{\bf m}^{'}({\bf r } ) > + o(r^{2})\ ] ] @xmath397|0 > = |b({\bf r } ) > + { \bf r } \cdot |{\bf b}^{'}({\bf r } ) > + o(r^{2})\ ] ] and similarly for the exotic meson states @xmath398 .
here m belongs to the @xmath399-dimensional representation of @xmath0 corresponding to an @xmath389 bound state of the quark and the antiquark .
the @xmath400 is an orbital excitation @xmath401 of opposite parity , which are in the @xmath402 representation of the group @xmath403 , @xmath60 being associated with the relative angular momentum of the constituents .
the @xmath404 states contain mesons like @xmath405 , @xmath406 , @xmath407 and scalar particles . on the whole ,
the @xmath389 and @xmath392 part @xmath363 contain @xmath408 meson states .
switching to the baryon states , the requirement of antisymmetry in color and symmetry in spin - flavor indices gives the @xmath409 representation for @xmath410 .
the @xmath392 multiplets have negative parity and have mixed spin - flavor symmetry .
they belong to the representation @xmath411 of @xmath403 and are represented by the states @xmath412 which are @xmath413 in number . on the whole , these @xmath414 states account for all the observed low - lying baryon states obtained from @xmath415 .
a similar analysis can be carried out for the exotic meson states @xmath398 , where the diquark and the antidiquark can be bound in an @xmath389 or @xmath392 state with opposite parities .
the exact , unbroken color group @xmath97 is the backbone of the strong interaction .
it is worthwhile to understand its role in the diquark picture more clearly .
two of the colored quarks in the baryon combine into an anti - triplet @xmath416 , @xmath417 .
the @xmath418 partner of the diquark and the @xmath419 partner of the nucleon do not exist . in hadron dynamics
the only color combinations to consider are @xmath420 and @xmath421 .
these relations imply the existence of split octonion units through a representation of the grassmann algebra @xmath422 , @xmath423 .
what is a bit strange is that operators @xmath424 , unlike ordinary fermionic operators , are not associative .
we also have @xmath425=\epsilon_{ijk}~u_k^{*}$ ] .
the jacobi identity does not hold since @xmath426=- i e_7 \neq 0 $ ] , where @xmath427 , anticommute with @xmath424 and @xmath428 .
the behavior of various states under the color group are best seen if we use split octonion units defined by@xmath117 @xmath429 @xmath430 the automorphism group of the octonion algebra is the 14-parameter exceptional group @xmath431 . the imaginary octonion units @xmath432 fall into its 7-dimensional representation . under the @xmath350 subgroup of @xmath431 that leaves @xmath433 invariant , @xmath434 and @xmath435
are singlets , while @xmath436 and @xmath437 correspond , respectively , to the representations @xmath133 and @xmath438 .
the multiplication table can now be written in a manifestly @xmath350 invariant manner ( together with the complex conjugate equations ) : @xmath439 @xmath440 @xmath441 @xmath442 where @xmath443 is completely antisymmetric with @xmath444 for @xmath445 = @xmath446 , @xmath447 , @xmath448 , @xmath449 , @xmath450 , @xmath451 , @xmath452 .
here , one sees the virtue of octonion multiplication .
if we consider the direct products @xmath453 @xmath454 for @xmath350 , then these equations show that octonion multiplication gets rid of @xmath455 in @xmath456 , while it gets rid of @xmath457 in @xmath458 .
combining eq.([eq : oct ] ) and eq.([eq : octa ] ) we find @xmath459 thus the octonion product leaves only the color part in @xmath456 and @xmath460 , so that it is a natural algebra for colored quarks . for convenience
we now produce the following multiplication table for the split octonion units : [ cols="<,^,^,^,^",options="header " , ] we see that most everything is improved ( some spectacularly ) . except for the magnetic properties of the nucleon , the predictions derived from the model are in excellent agreement with the experimental values .
inclusion of the effect of vector mesons could change and improve the results significantly since the virtual photon can make a transition to the @xmath461 and @xmath462 mesons in the vector dominance scenarios .
possible improvement of the predicted values of the magnetic properties will be discussed below . in the generalized model ( as is also in @xmath463 case ) we find that the calculated nucleon charge densities oscillate within the nucleons , a peculiar future of nucleon substructure .
it is well known , from the deep inelastic scattering experiments , that nucleons have point - like constituents .
since the quark degrees of freedom are integrated out in the skyrme model , we expect the substructure of the nucleons not to be point - like .
therefore the oscillatory behavior of the charge densities may be due to interference effect of the constituents , which is similar to the quark interference effect found in the bag model@xmath104 .
excellent agreement with the experimental value of @xmath35 ( = 186 mev ) suggests that @xmath464 model could serve as a better description of the nucleons .
the reason might be that composite model associated with the lagrangian @xmath465 describes an effective hadronic theory based on qcd with the color group @xmath466 .
the baryon consisting of @xmath340 quarks are in a spatially symmetric ground state . in that case the current @xmath467 could be interpreted as a quark number current , so that the baryon has quark number 3 for @xmath468 . under spin and isospin transformations ,
the skyrmion matrix @xmath469 has same transformation properties as the nucleon matrix of spin states @xmath470 and @xmath471 ( @xmath472 ) @xmath473 with left and right multiplications being respectively associated with spin and isospin rotations
. now @xmath33 and @xmath34 which form the 20-dimensional symmetric representation of the spin - isospin group @xmath474 are cubic functions of the elements of the underlying quark matrix @xmath475 that can be taken to be proportional to @xmath476 .
when the color group is @xmath64 and @xmath477 is quantized fermionically , it is possible to obtain the symmetric @xmath478 representation for the @xmath479 system that is cubic in @xmath477 .
four of these twenty cubic functions , which are eigenstates of the hamiltonian , are associated with @xmath33 .
they have both spin and isospin @xmath480 and can be put in a @xmath481 matrix form @xmath482 , where @xmath483 with @xmath484 @xmath485 being a singlet in spin and isospin .
the remaining 16 cubic functions correspond to the wave functions of the @xmath34 resonances .
since the quark has a baryon number @xmath486 and the matrix @xmath487 gives baryon number @xmath41 , it is natural to interpret the @xmath488 dependence of the quark in the nucleon as being given by @xmath489 .
then the quark matrix could be split into the form @xmath490 in a static approximation .
the @xmath33 and @xmath34 wave functions would then be associated with the cubic functions of the elements of @xmath491 as well as being cubic in @xmath476 .
the fermionic quantization is preferable for eliminating the higher excited states of the skyrmion represented by higher order polynomials in @xmath476 that are eigenstates of the hamiltonian . a convincing derivation of @xmath468 composite skyrme lagrangian from qcd theory of quarks and gluons with color group @xmath64 is still lacking . for the moment one can regard it as a phenomenological model that gives reasonable quantitative predictions .
we shall deal with this issue and the wave functions of the @xmath34 resonances written in terms of the cubic functions in a forthcoming publication@xmath492 .
for now we give a brief introduction to this subject .
the baryon states can be written in terms of @xmath493 , for example @xmath494 @xmath495 note that we can write @xmath496 so that @xmath497 giving @xmath498 with @xmath499 note that the left column of @xmath477 is @xmath500 while the right column is @xmath501 so that @xmath477 can be written in compact form as @xmath502 now the @xmath503 state is @xmath504 and @xmath505 state is given by @xmath506 in general we can construct the @xmath34 resonances in this way .
we can write down the @xmath482 ( @xmath96 ) part in a matrix form @xmath507 this term arises as part of a general quark matrix ( in terms of @xmath56 s and the @xmath57 s ) as follows : let us write down the following matrices @xmath508 and @xmath509 each expression giving @xmath510 triplet quark combinations as expected . in component form @xmath511 and @xmath512 we can now write the full matrix @xmath156 as @xmath513 where @xmath514
@xmath515 @xmath516 @xmath517 we can now embed @xmath96 part into the center of the @xmath156 matrix .
we get the @xmath96 part from @xmath518q= \nonumber\ ] ] @xmath519 similarly we can write down the @xmath45 parts as @xmath520q= [ u_1 d_2-u_2 d_1 ] \left ( \begin{array}{cc } d_1 & 0 \\ 0 & u_2 \end{array } \right)\ ] ] and @xmath521q= [ u_1 d_2 -u_2 d_1 ] \left ( \begin{array}{cc } 0 & u_1 \\
d_2 & 0 \end{array } \right)\ ] ] finally the full matrix containing all @xmath359 terms become : @xmath522 the expression shown in eq.([eq : centrali ] ) then corresponds to the following term : @xmath523 note that the first column of @xmath524 contains all the @xmath525 @xmath503 , @xmath526 , @xmath527 , and @xmath528 terms , the second column contains @xmath529 , third column contains @xmath530 , and the last column contains @xmath531 expressions .
we now give these terms explicitly .
for the @xmath503 states we have @xmath532 @xmath533 @xmath534 @xmath535 for the @xmath526 state we have @xmath536 @xmath537 @xmath538 @xmath539 for the @xmath527 states we have @xmath540 @xmath541 @xmath542 @xmath543 finally , for the @xmath528 states we have @xmath544 @xmath545 @xmath546 @xmath547 action of the hamiltonian operator given above on the @xmath527 , @xmath526 , @xmath528 and @xmath503 states will be dealt with in detail in a forthcoming publication .
restoration of symmetries .
one can now have a supersymmetry algebra generated by the creation operators for @xmath392 ( fermionic ) and those for @xmath548 ( bosonic ) .
the @xmath549 cubic functions will be a representation of this superalgebra .
construction of the enlarged hamiltonian that will be approximately invariant under @xmath474 and its supersymmetric extension @xmath550 ( subalgebra of @xmath6 ) is the next step .
finding its eigenvalues and linear ( quark ) , quadratic ( diquark ) and cubic ( @xmath33 and @xmath34 ) eigenstates is the following step .
final step will be recalculation of masses , charge radii , meson - nucleon coupling constants and magnetic moments , pion decay constant , meson - meson couplings , etc .
in the large @xmath17 expansion baryons emerge as solitons within an effective theory of infinitely many weakly interacting mesons .
a more realistic model should include not just the pseudoscalar meson field but also other heavier vector mesons .
if we assume the @xmath551 s , @xmath461 s and @xmath552 s form an @xmath474 multiplet , they should be treated in a unified manner .
accordingly , we now consider an @xmath553 matrix configuration of these mesons @xmath554 where @xmath555 will be substituted back into eq.([eq : v1 ] ) later . here
@xmath556 and @xmath557 are the couplings , and @xmath558 and @xmath559 are the masses of @xmath461 and @xmath552 mesons .
the @xmath553 pauli and dirac matrices are formed by making the direct products @xmath560 also @xmath561\ ] ] first let us write down the simplest chiral model @xmath562 + { \rm h.c . }
\nonumber \\ & = & -8~ \partial_\mu \mbox{\boldmath$\phi$ } \cdot \partial^\mu \mbox{\boldmath$\phi$ } + 4~(\partial_\mu \omega_{\alpha\beta } ~\partial^\mu \omega^{\alpha\beta}+ \partial_\mu \mbox{\boldmath$\rho$}_{\alpha\beta } \cdot \partial^\mu \mbox{\boldmath$\rho$}^{\alpha\beta } ) + o(4 ) \end{aligned}\ ] ] which is not sophisticated enough to contain much interesting physics .
since isospin space and spin space do not mix , there is no interaction among mesons . to connect the isospin space and the spin space
while keeping the isospin symmetry intact we can insert @xmath563- and @xmath564-matrices between the configuration @xmath565 . to write down the more realistic lagrangian we first evaluate the following terms : @xmath566 = 32~ \partial_\mu \mbox{\boldmath$\phi$ } \cdot \partial^\nu \mbox{\boldmath$\phi$ } + o(4 ) \label{eq : v3}\ ] ] and and we can now combine eqs.(@xmath569 ) : now using eq.(@xmath571 ) in @xmath565 and eq.(@xmath572 ) we have @xmath573 and since the masses of vector mesons are significant in comparison to the baryonic masses , we must include the meson masses in a more realistic lagrangian .
two mass terms comes from which gives masses to the ( spin triplets ) vector mesons @xmath461 and @xmath552 while leaving the @xmath551 massless ( spin singlet ) . using eq.(@xmath571 )
we get we now evaluate the term @xmath577= -32 ~\mbox{\boldmath$\phi$ } \cdot \mbox{\boldmath$\phi$ } + 16~\mbox{\boldmath$\rho$}^{\alpha\beta } \cdot \mbox{\boldmath$\rho$}_{\alpha\beta } + o(4)\ ] ] and again using eq.(@xmath578 ) we arrive at @xmath579= -\frac{1}{2}~m_\pi^2~ ~\mbox{\boldmath$\phi$ } \cdot \mbox{\boldmath$\phi$ } + \frac{f_\pi^2~ g_\rho^2~ m_\pi^2}{16~m_\rho^4}~\mbox{\boldmath$\rho$}^{\alpha\beta } \cdot \mbox{\boldmath$\rho$}_{\alpha\beta } + o(4)\ ] ] a term which shifts the masses of the isospin triplets @xmath551 and @xmath552 with respect to the isospin singlet @xmath461 .
we now combine all these into an approximate meson lagrangian , expanded to third order , and containing @xmath580 , @xmath581 and @xmath582 interactions : @xmath583 note that some of the total divergences have been subtracted . from the kinetic and mass terms of the lagrangian we get @xmath584 and @xmath585 which is the ksrf@xmath586 relation , and substituting @xmath587 we have mass relations for the mesons @xmath588 in perfect agreement with experiment .
if we identify the @xmath461 and @xmath552 fields as @xmath589 and @xmath590 up to second order , lagrangian only of free fields ( @xmath461 mesons ) @xmath591 take @xmath592 so the equations of motion from @xmath593 are @xmath594 @xmath595 @xmath596
gauge potentials can be obtained from the decomposition of a larger algebra into a subalgebra and a coset .
gauge bosons can be combined with the fundamental mesons to form a single multiplet so that all mesons are treated equally in the larger symmetry group .
we can write @xmath597 in terms of the new variables @xmath598 and @xmath599 @xmath600 under the group @xmath601 @xmath602 transform as @xmath603 @xmath604 so that the lagrangian @xmath605 is invariant under the global symmetry @xmath606 , with @xmath607 , @xmath608 , and @xmath609 .
since @xmath597 transforms as @xmath610 without any explicit @xmath611 dependence , the symmetry is said to be `` hidden '' . the associated gauge field @xmath612 transforms as @xmath613 we can now construct a lagrangian that respects this gauge symmetry . writing the covariant derivative @xmath614 where @xmath615 is the gauge coupling , we can write two independent second order terms that are gauge invariant : @xmath616 ^ 2\ ] ] @xmath617 ^ 2\ ] ] so that a combination of the form @xmath618 is equivalent to the non - linear sigma model eq.([eq : v12 ] ) . here
@xmath619 is an arbitrary parameter . using unitary gauge @xmath620 we have @xmath621 @xmath622\}^2\ ] ]
we see that when we use the equation of motion for @xmath623 @xmath624\}\ ] ] @xmath623 vanishes and @xmath593 becomes identical to @xmath625 . making the identification of vector fields @xmath626 with the @xmath552-meson fields @xmath627 ,
we add by hand the kinetic term of @xmath552 , so that @xmath628 again , expanding to third term , using the weak field expansion we get the lagrangian @xmath629 the mass of the @xmath552 meson is given by @xmath630 which is identical to earlier value when @xmath631 . overall result of this approach agrees with experiment very well . even though this description we just gave works very well with vector mesons , it would be more useful if we could find a way of including scalars and vectors in a unified manner . toward achieving such a goal
, we can now consider the genesis of covariant derivatives through a decomposition of a larger algebra involving spin or dirac operators , into a subalgebra and a coset .
the @xmath564-model obtained from the decomposition is valued in the subalgebra of the larger algebra , while the gauge fields are valued in the coset space .
once we identify the derivative of the coset elements as vector gauge potentials , the derivative of the larger algebra configuration can be written as a covariant derivative , i.e. , a derivative of the subalgebra element plus a gauge potential .
let us rewrite @xmath565 as @xmath632 where @xmath633 is a configuration of pion fields only , and @xmath634 is a configuration of vector meson fields only .
we also note that @xmath635 \nonumber \\ & & u^\dag ( \partial_\mu + v^{-1 } \partial_\mu v)u = u^\dag d_\mu u\end{aligned}\ ] ] therefore , the covariant derivative @xmath636 arises naturally with vector mesons being the gauge potentials @xmath637 . under this decomposition
, the @xmath593 of the non - linear @xmath564-model becomes @xmath638 \rightarrow tr[u^\dag ( d_\mu u ) u^\dag ( d^\mu u)]\ ] ] we see that the ordinary derivatives are replaced by the covariant derivatives , and the field configuration @xmath565 is replaced by the field configuration @xmath597 that is valued in the subalgebra .
but there is one problem : we note that @xmath639 where @xmath640 is a function of the vector meson fields , has a spin matrix @xmath641 dependence but @xmath642 has none .
therefore the trace of the matrix with scalar - vector meson interaction terms is zero . to generate non - zero values of such terms
it is necessary to add @xmath564- and @xmath343-matrices to @xmath593 . for simplicity , we now take @xmath643 to be a configuration of @xmath552-mesons only . the covariant derivative is @xmath644 if we now define @xmath645 and @xmath646 such that they satisfy the same dirac algebra as the original @xmath563- and @xmath564-matrices , namely @xmath647 and @xmath648 & = & v^{-1}~[\sigma_{\alpha\beta } , \sigma_{\mu\nu } ] ~v \nonumber\\ & & = 2i~v^{-1}~(\eta_{\alpha\mu } \sigma_{\beta\nu } + \eta_{\beta\nu } \sigma_{\alpha\mu } - \eta_{\alpha\nu } \sigma_{\beta\mu } - \eta_{\beta\mu } \sigma_{\alpha\nu } ) v \nonumber \\ & & = 2i~v^{-1}~(\eta_{\alpha\mu } \sigma_{\beta\nu } + \eta_{\beta\nu } \sigma_{\alpha\mu } - \eta_{\alpha\nu } \sigma_{\beta\mu } - \eta_{\beta\mu } \sigma_{\alpha\nu } ) \end{aligned}\ ] ] using these in eq.([eq : v8 ] ) after some lengthy algebra we arrive at a weak field expansion of the lagrangian @xmath649 where we have made the identification @xmath650 , and obtained the ksrf relation .
it is easy to see that this result agrees with hidden symmetry approach we discussed .
therefore , a realistic connection between covariant derivatives and the coset elements of the larger algebra has been demonstrated through this correspondence .
another example of obtaining covariant derivatives is from the consideration of a chiral lagrangian that includes the meson nucleon interactions .
the lagrangian @xmath651 is invariant under the chiral transformation .
there is no mass @xmath652 for the fermions because such a term is not chiral invariant , hence the fermion field @xmath182 can not represent a realistic physical nucleon , which has to be massive .
however , by performing a unitary transformation involving the meson field we can define a field @xmath653 @xmath654 such that the new nucleon mass term @xmath655 becomes chiral invariant . since @xmath656 we have @xmath657 . then
, the transformed lagrangian becomes @xmath658 where @xmath659 and the covariant derivative @xmath660 with @xmath661 we see that the lagrangian in eq.([eq : v13 ] ) gives a dirac equation for the nucleons ( @xmath662 , which contains both the vector and the pseudoscalar meson - nucleon interactions through the covariant derivatives .
these examples show that gauge potentials can be obtained from the decompositions of larger algebra into a subalgebra and a coset .
gauge bosons , which take on special roles in the model originally , can be combined with the fundamental mesons to form a single multiplet so that all mesons are treated equally in the larger symmetry group .
earlier we quantized the skyrmion by rotating the soliton solution by the @xmath663 time - dependent matrix @xmath476 .
the canonical conjugate matrix of @xmath493 is given by @xmath664 and the hamiltonian is obtained from @xmath665 although at the classical level @xmath666 and @xmath667 are commuting variables , after quantization they become operators and are no longer commuting .
we can either perform bosonic quantization @xmath668 = i\delta_\alpha^\beta \delta_\beta^\nu\ ] ] @xmath669=[{\cal{a } } , { \cal{a}}]=0\ ] ] or fermionic quantization @xmath670 @xmath671 that give the same algebra of the spin operators formed by the @xmath493 s and @xmath672 s . in the bosonic quantization , operators @xmath666
commute among themselves , allowing particle states be described in terms of power functions in @xmath493 of arbitrary orders . on the other hand , in the case of fermionic quantization , the @xmath666 anticommute among themselves so that states with two or more identical @xmath666 do not exist , and the powers in @xmath493 of the state functions are limited . in the case when we take @xmath493 to be fermionic , then the cubic functions for @xmath503 vanishes . to overcome this difficulty , we consider the same problem which occurred in the quark model when the problem arose of putting three spin up @xmath56 quarks together to form @xmath673 @xmath503 state . to conform with pauli exclusion principle , three colors had to be introduced in the quark model to make the state antisymmetric in permuting any two quarks . in a similar sense ,
we now triple the number of variables in the skyrme model by introducing two more matrices @xmath405 and @xmath674 in addition to the matrix @xmath493 , and require the new lagrangian to be invariant under the permutations of @xmath493 , @xmath405 and @xmath674 .
the color group @xmath64 is replaced by the discrete subgroup @xmath675 . in general for @xmath676 expansion , the color group @xmath677 is replaced by @xmath678 , and there will be @xmath33 different matrices .
we propose a new lagrangian that accommodates the new degrees of freedom we note that the time dependent parts of @xmath593 remain unchanged , so we only have to concentrate on the time derivatives of @xmath680 writing @xmath681 in the same way as before , after some tedious algebra , we find that the time - dependent parts of the non - linear sigma model terms are given by @xmath682 @xmath683 @xmath684 the time - dependent part of the skyrme term can also be similarly calculated .
after some more lengthy algebra , we obtain the lagrangian @xmath685 \end{aligned}\ ] ] where @xmath192 and @xmath686 are given by eq.([eq : m ] ) and eq.([eq:2s ] ) , respectively , and @xmath687 is the size of the @xmath243-space to be set equal to infinity after the legendre transformation of the hamiltonian . defining the canonical conjugates
@xmath688 and the corresponding spin operators @xmath689 we obtain the hamiltonian finally , letting @xmath691 , we arrive at the hamiltonian @xmath692 whose eigenstates will be functions of the variables @xmath493 , @xmath405 and @xmath674 . in analogy to the @xmath474 quark model , we can describe the nucleons and the @xmath34 s as @xmath474 multiplets and write their states in terms of functions trilinear in @xmath493 , @xmath405 and @xmath674 .
for example , we can write the state function of spin @xmath693 @xmath503 as @xmath694 and the spin up proton as @xmath695 the problem of the infinite towers of particle states disappears because the pauli exclusion principle does not allow any more states with @xmath696 .
in a recent paper cheung and grsey@xmath697 showed that generators for @xmath474 are constructed from dynamical skyrmion operators .
in their paper they gave a fermionic realization of @xmath23 algebra of the skyrme model .
they imbedded the @xmath23 algebra into @xmath335 by introducing new generators which are simple modifications of the @xmath215 generators .
the particle wave functions are written in powers of the collective coordinates .
the infinite tower of states , which produces one of the problems for the skyrme model can now be eliminated by restricting the values of these powers .
we observe that if the collective coordinates are quantized fermionically , arbitrary functions of @xmath493 ( where @xmath698 with @xmath699 ) can be written as polynomials in @xmath493 of finite degrees , and therefore the height of the tower of particle states will have an upper bound .
another motivation for introducing fermionic quantization is that the predicted baryon masses from the skyrme model are too high if the experimental value of @xmath35 is taken as an input parameter.modification of the model by addition of new terms will usually increase the particle mass further since usually the new terms give positive contributions to the hamiltonian . since in the renormalization procedure of a supersymmetric field theory
the contributions coming from fermion loops are negative , such supersymmetric theory will have a better convergence .
it is conceivable that the masses may be lowered if one can construct a supersymmetric skyrme model , and fermionic quantization is a first step in this direction .
note that the spin and the isospin operators we introduced form the @xmath71 generators of the @xmath23 algebra . to imbed the algebra into @xmath335
, we introduce the operators @xmath700 and @xmath701 the form of the operators @xmath702 and @xmath703 are easy to obtain as they differ from the operators @xmath704 and @xmath705 only by a factor of a pauli matrix .
one can easily verify that the 15 operators @xmath704 , @xmath705 and @xmath703 form the generators of @xmath474 .
if we include the operator @xmath702 , which commutes with all the @xmath474 generators , we obtain a @xmath335 algebra .
the @xmath474 casimir operator is given by @xmath706 the fermionic quantization prescription allows us to expand the algebra into superalgebra .
the supersymmetric extension of the @xmath707 algebra and the operators forming the @xmath23 subalgebra can now be obtained .
let @xmath708 be an @xmath33-vector , with @xmath709 its canonical cojugate ( @xmath710 . together with their complex conjugates they obey the following commutation relations : @xmath711 @xmath712 @xmath713 @xmath714 defining the operators @xmath715 as @xmath716 they generate the @xmath707 algebra @xmath717=\delta_{\beta\mu } j_{\alpha\nu } - \delta_{\alpha\nu } j_{\mu\beta}\ ] ] for the case of @xmath718 , we keep the track of the imbedding by noting that the operators @xmath719 @xmath720 @xmath721 @xmath722 @xmath723 @xmath724 form an @xmath23 subalgebra of the skyrme model .
so far we have only the even algebraic operators @xmath715 that are products of two fermionic operators @xmath493 and @xmath672 . to make a supersymmetric extension of the algebra , we need some odd algebraic operators that are made of products of odd number of fermionic operators .
the even and odd operators are expected to form a superalgebra . to obtain the superalgebra we now introduce the ( odd ) supercharge operators @xmath725 and @xmath726 where @xmath686 is an arbitrary complex parameter . using the commutation relations we see that @xmath715 , @xmath727 and @xmath728 with ( @xmath729 ) form the superalgebra @xmath730 @xmath731 @xmath732=-\delta_{\alpha\gamma } \overline{q}_\beta\ ] ]
the quadratic invariant of the superalgebra is @xmath733\ ] ] which can be obtained by taking the supertrace of @xmath734 , where @xmath735 is a supertraceless matrix configuration with even and odd operator matrix elements .
other invariants of the superalgebra are @xmath736 . among the irreducible representations of @xmath737 ,
there is a 15-dimensional boson multiplet ( @xmath551 , @xmath552 , @xmath461 ) and a 20-dimensional fermion multiplet ( @xmath33 , @xmath34 ) . with the supersymmetric extension of @xmath335 it
is possible to extend lagrangians which contain only the @xmath551 , @xmath552 ( and @xmath461 ) to the ones that contain mesons , nucleons and @xmath34 s .
the new lagrangian must necessarily contain the dirac and rarita - schwinger lagrangians for @xmath33 s and @xmath34 s as well as the usual meson lagrangian .
the hamiltonian to be obtained then should resemble the casimir invariant @xmath738 shown above , and the supersymmetry breaking will bring the hamiltonian to the form eq.([eq:5s ] ) , and the subsequent breaking of @xmath335 into @xmath215 will bring the hamiltonian to the form of eq.([eq:3s ] ) . in order to avoid the problem of predicting too high soliton masses when taking physical value of @xmath35 as an input parameter
, many people subtract from the hamiltonian a zero point energy without explaining its origin .
we suggest that the subtraction of the zero point energy without explaining its origin .
we suggest that the subtraction of the zero point energy may come from the remnant effect of supersymmetry after the supersymmetric theory has been broken down to a non - supersymmetric theory .
it is known that in a field theory where supersymmetry is exact , the ground state energy is zero .
the positive contribution from bosons is exactly canceled by the negative contributions from the fermions . in a theory with approximate hadronic supersymmetry , although we do not expect exact cancellation of bosonic and fermionic parts of the soliton energy , it is probable that there is a good balance of both sides so that the baryonic masses can be lowered significantly to give a reasonable value of the pion decay constant .
quark model with potentials derived from qcd , including the quark - diquark model for excited hadrons gave mass formulae in very good agreement with experiment and went a long way in explaining the approximate symmetries and supersymmetries of the hadronic spectrum , including the symmetry breaking mechanism .
near success of the minimal @xmath663 skyrme model was modified to a generalized composite skyrme model in which baryons are considered as composite objects made up of many @xmath663 configurations . in the special case of three @xmath663 configuration , the static properties of the soliton had excellent agreement with the experimental data .
according to the large @xmath17 expansion , a more realistic model should include the vector mesons and the pions .
inclusion of low lying vector mesons @xmath461 and @xmath552 in a modified model , where the lagrangian of the model was constructed by inserting @xmath343- and @xmath564- matrices that allowed us to get the usual meson physics , including the meson mass relations in perfect agreement with experiment .
all the mesons were treated on equal footing so that they belong to a single meson multiplet .
the simplified model we built , in which a soliton ansatz to the lagrangian @xmath593 has not yet been found , gives us a good indication of the salient features of any soliton solutions to the full model .
magnetic properties are improved by adding vector mesons to the model . with the static approximation ,
the magnetic part of the configuration vanishes and the model is then becomes reducible to an @xmath474 skyrme model that produces a variety of mass spectra for the solitons .
fermionic quantization for the collective coordinates of the skyrmion is introduced .
it has the advantage that through the pauli exclusion principle it it chops off the unphysical infinite tower of ( @xmath739 particle states . with this approach , nucleons and @xmath34 s
can all be expressed as trilinear functions of the collective coordinates as we have shown in this paper , and the fermionic quantization approach allows a simple incorporation of supersymmetry into the model .
such a supersymmetric extension makes it possible to lower the soliton mass since fermions contribute negatively to the hamiltonian .
imbedding of the @xmath215 symmetry into @xmath335 and @xmath740 ( @xmath741 in our case ) is the first step towards a supersymmetric extension .
@xmath707 algebra is extended to a superalgebra for arbitrary @xmath33 by the introduction of supercharges , and the casimir invariants are easily obtained by taking the supertrace of the powers of the configuration @xmath156 .
the fermionic degrees of freedom must be introduced in such a way that an approximate symmetry under @xmath550 holds .
it is to be seen if after the suggested modifications one will obtain a quantitatively viable skyrmionic description of hadronic physics at least as good as the 2-body potential model@xmath742 inspired by qcd and reflecting its fundamental symmetries .
it is highly desirable to find an explicit relationship between the relativistic quark model we built and the skyrme model .
these two models should have common hadronic symmetries .
the approximate @xmath0 symmetry and @xmath6 supersymmetry have been established for the quark model , but not yet for the skyrme model .
we can build a supersymmetric skyrme model based on the superalgebra we developed .
some preliminary work has been done in that direction , which will be the subject of another publication .
it will be very interesting to compare the two different supersymmetry realizations of the quark model and the skyrme model .
strangeness should also be incorporated into the model .
if we include strangeness in our model while treating all mesons equally , we would expect the mesons to form an @xmath0 multiplet . with such modifications we would expect that the predictions of the static properties of the solitons can be brought into much better agreement with experiment .
also , it should in principle be possible to construct fermi fields from bose fields in @xmath743 dimensions in the light of these recent developments .
the picture that one would like to see emerge is a duality between a field theory written in terms of quark and gluon fields that would accurately reflect the hilbert space structure at very small distances on one hand , and a dual theory expressed in terms of non - linear meson couplings whose soliton solutions would correspond to the structure of the hilbert space as seen from a larger distance perspective on the other .
this would require two boson - fermion connections .
first the quarks and gluons must give rise to the non - linear meson fields as composites and these latter must be reinterpreted in terms of baryons .
it is not clear how fundamental this phenomenologically hoped for connection can be .
other applications such as when one considers the pion fluctuations around a soliton one can study the meson - nucleon scatterings . in the many - skyrmion system , one can study dibaryon states , nucleon - nucleon interactions and properties of nuclei .
skyrme model is a viable model of baryons .
recent developments of the model give strong support to the belief that baryons can be regarded as topological solitons of a theory of interacting meson fields , and the low energy physics should be described non - perturbatively by an effective lagrangian based on the underlying symmetries of qcd .
greenberg , _ phys .
* 13 * ( 1964 ) 598 _ ; y. nambu in `` preludes in theoretical physics '' eds .
a. de shalit , h. fesbach and l. van hove , p.133 ( amsterdam , 1966 ) ; h. fritzsch , m. gell - mann and h. leutwyler , _ phys . lett . * 47b * ( 1973 ) 365_. | an approximate hadronic symmetry based on spin and flavor independence and broken by spin and mass dependent terms is shown to follow from qcd .
this symmetry justifies the @xmath0 classification scheme , but is more general in allowing its supersymmetric extension based on a diquark - antiquark symmetry .
it will be shown that the same supersymmetry is also implied in the skyrmion type effective lagrangian which could be extracted from qcd .
predictions of the skyrme model is improved by using different realizations of the chiral group . *
effective hadronic lagrangians based on qcd : potential models and skyrmions * + sultan catto@xmath1 + _ physics department + the graduate school and university center + 365 fifth avenue + new york , ny 10016 - 4309 + and + center for theoretical physics + the rockefeller university + 1230 york avenue + new york ny 10021 - 6399 _ + pacs numbers : 12.40.aa , 12.40.qq , 11.30.pb @xmath2 work supported in part by doe contracts no .
de - ac-0276 er 03074 and 03075 , and psc - cuny research awards . |
primordial non - gaussianity has been attracting attention as a powerful probe to discriminate many existing inflation models ( for example , see @xcite ) .
statistics of zero - mean gaussian fluctuations can be characterized by a variance which corresponds to two - point function .
hence , the non - gaussianity of fluctuations can be linked to the higher - order correlation functions .
a lot of studies have been done for the bispectrum , i.e. three - point function of the primordial perturbations .
for the so - called local type perturbation , the bispectrum is completely characterized by a single non - linearity parameter @xmath2 @xcite .
the current bound on @xmath2 is @xmath3 at @xmath4 confidence level @xcite . in the last few years
, trispectrum is becoming an important observable as well .
for the local type non - gaussianity , the trispectrum is specified by two non - linearity parameters @xmath5 and @xmath6 @xcite .
the current bounds on @xmath5 and @xmath6 are respectively @xmath7 and @xmath8 at @xmath4 confidence level @xcite . an interesting fact that boosts the importance of the study of the trispectrum
is that @xmath5 has a minimum determined by @xmath2 @xcite , @xmath9 it was shown in ref .
@xcite that planck satellite can measure @xmath5 down to @xmath10 .
therefore , future detection of @xmath11 which is the central value of the current bound implies that we should also detect non - vanishing @xmath5 .
detection of both @xmath2 and @xmath5 ( and possibly @xmath6 as well ) enables us to constrain or even pin down the inflation model and the origin of the primordial fluctuations .
in fact , in ref . @xcite the authors have discussed an observational constraint on the ratio between @xmath2 and @xmath5 which is defined as @xmath12 .
there exist a few studies that extend the inequality ( [ ft ] ) .
the authors of refs .
@xcite considered one - loop corrections to ( [ ft ] ) and found the one - loop effect appears as the scale dependent modification of the coefficient in front of @xmath13 . in ref .
@xcite , we have discussed that how many parameters one needs in order to characterize higher order correlation functions of primordial curvature fluctuations and also introduced the non - linearity parameters for the five - point function .
the non - linearity parameters up to the six - point function were introduced in @xcite and two inequalities between the six- and four - point function were derived @xcite . a potential importance and usefulness of the higher order correlation functions motivate us to look for the similar inequalities like ( [ ft ] ) for the higher order correlation functions . in this paper , we provide a general formalism to construct the inequalities among the non - linearity parameters for the higher order correlation functions . to this end , we will adopt the diagrammatic approach developed in refs .
@xcite that turns out to be a convenient way to study the higher order correlation functions . by using this method , in principle
, we can systematically derive all the inequalities up to any - point function .
this paper is organized as follows . in the next section
, we briefly review the higher order correlation functions of primordial curvature fluctuations , based on @xmath14 formalism and the diagrammatic approach . in section [ sec : inequality ] , we define the non - linearity parameters to characterize the higher order functions and construct the series of inequalities by applying the cauchy - schwarz inequality . we also present all inequalities up to the six- and eight - point functions .
section [ sec : con ] is spent on the conclusion of this paper .
according to the @xmath14 formalism @xcite , the curvature perturbation on the uniform energy density hypersurface on super - horizon scales at @xmath15 is equal to the perturbation of the e - folding number evaluated from the flat hypersurface at @xmath16 to the uniform energy density hypersurface at @xmath17 at the same point : @xmath18 the initial time @xmath19 can be chosen arbitrary . if we chose @xmath19 as the time slightly after the scale we are interested in leaves the hubble horizon during inflation , then @xmath14 in eq .
( [ deltan1 ] ) is sourced by the scalar field fluctuations .
thus , @xmath20 can be taylor - expanded in terms of the scalar field fluctuations as @xmath21 where @xmath22 run from @xmath23 to @xmath24 ( @xmath24 is a number of light scalar fields that acquire super - horizon scale fluctuations during inflation ) . in the following , we assume that @xmath25 are gaussian variables . the two - point function of @xmath20 can be written as @xmath26 to leading order in the field fluctuations , @xmath27 is given by @xmath28 where summation over @xmath29 is assumed .
the three - point function of @xmath20 can be written as @xmath30 to leading order in the field fluctuations , @xmath31 is given by @xmath32 following literatures , the constant coefficient in front of the square of the power spectrum is written as @xcite @xmath33 the four - point function of @xmath20 can be written as @xmath34 to leading order in the field fluctuations , @xmath35 is given by @xmath36 where @xmath37 .
note that there appear two distinct terms that exhibit different wavenumber dependence . as a consequence ,
we need two parameters to specify the tri - spectrum .
following literatures , the two constant parameters are defined by @xcite @xmath38 we can further proceed to higher order correlation functions as follows .
the leading order of the @xmath39-point function is given by a sum of several distinct terms which are products of @xmath40 power spectra and exhibit different wavenumber dependence .
according to the diagrammatic method @xcite , each of these leading terms has a corresponding connected tree diagram that consists of @xmath39 vertices and @xmath40 lines connecting two vertices .
tree diagrams for the two- , three- and four - point functions are shown in fig .
[ fig : diagram_2.eps ] . in reverse , given a tree diagram with @xmath39 vertices , we can reconstruct the corresponding term that constitutes the @xmath39-point function as follows .
first , we assign a different wavenumber @xmath41 to each vertex of the diagram , where @xmath41 are the arguments of the @xmath39-point function with the constraint @xmath42 .
next , we assign a wavenumber to each line in the diagram , too .
removing a line from the diagram yields two respectively connected sub - diagrams .
then , one assigns to the removed line the sum of the vectors associated with all vertices in one of the two sub - diagrams . any use of the two sub - diagrams yields the same answer because of the constraint @xmath42 . after associating the wavenumbers with all lines
, we can assign the corresponding factors to the vertices and the lines . as for
the vertex with p lines attached , assign the factor @xmath43 to it . as for the lines ,
assign @xmath44 , where the argument of @xmath44 is set to the length of the wavenumber associated with each line . by multiplying all these factors assigned to vertices and lines ,
we obtain the corresponding leading term . any term constructed in this way
is given by a product of two terms .
one is the constant term which is given by contracting all the pairs of indices of the expansion coefficients assigned to each vertex of the tree diagram .
the other term is the @xmath40 product of the power spectra of the scalar field plus its permutations .
if we replace each power spectrum of the scalar fields by that of the curvature perturbation , i.e. @xmath45 , the constant part gets a factor @xmath46 in the denominator . following the cases for the bispectrum and the trispectrum
, we define the constant parameter corresponding to this diagram as a coefficient in front of the @xmath40 product of @xmath47 . by taking the sum over
all the possible ( connected ) tree diagrams with @xmath39 vertices , we obtain the @xmath39-point function . formalism .
we find that we need two parameters in order to characterize the trispectrum . ]
it was shown in ref .
@xcite that the functions constructed from two tree diagrams that are not isomorphic each other always yield different wavenumber dependence .
therefore , the number of independent constant parameters required to specify the @xmath39-point function is equal to the one of all the possible connected tree diagrams with @xmath39 vertices that are not isomorphic to each other . a formula for a number @xmath48 of the independent connected tree diagrams with @xmath39 vertices is given in refs .
@xcite . according to refs .
@xcite , we can formally construct a function @xmath49 as an infinite series , @xmath50 then @xmath49 is given by @xmath51 where @xmath52 is a function that satisfies , @xmath53.\ ] ] from the last equation , we can recursively obtain the taylor - expansion coefficients of @xmath52 . then from eq .
( [ eqfort ] ) , we can recursively obtain @xmath48 as well .
for example , @xmath54 .
as we mentioned in the last section , wavenumber dependence of the @xmath39-point function is completely specified by a set of constant parameters each of which has a corresponding connected tree diagram with @xmath39 vertices . for the purpose of constructing the general inequalities among the non - linearity parameters ,
it is convenient to adopt a normalization condition , @xmath55 without a loss of generality , this condition can be always imposed by rescaling the scalar field fluctuation . with this condition ,
the denominator of the non - linearity parameters for the @xmath39-point function , which is given by @xmath46 , becomes unity . for any connected tree diagram with @xmath39 vertices ,
if we cut a line appearing in the diagram , the diagram splits into two sub - diagrams .
if the resulting two sub - diagrams are isomorphic to each other , we call the parent diagram symmetric diagram . obviously , in order for a diagram to be a symmetric diagram , @xmath39 must be even .
let us denote by @xmath56 a number of all the symmetric diagrams among @xmath57 tree diagrams with @xmath58 vertices .
for example , @xmath59 .
our aim is to construct the inequalities of the cauchy - schwarz type for the non - linearity parameters up to the @xmath39-point functions .
as it will be clear later , this is possible when @xmath39 is even . for the case of odd @xmath39
, we can not construct the inequalities .
for example , there are inequalities among the non - linearity parameters up to the six - point function .
but there are no inequalities in which the highest order of the non - linearity parameters is five . since whether @xmath39 is
even or odd is important , let us first write the non - linearity parameters for the @xmath58-point function as @xmath60 here , @xmath61 and @xmath62 are the non - linearity parameters for the symmetric diagram and the non - symmetric diagram , respectively . cases . ] in fig .
[ fig : symmetric_diagram.eps ] , we show the symmetric diagrams and the non - symmetric diagrams for @xmath63-point functions . for the four - point function ,
a diagram corresponding to @xmath5 is a symmetric diagram while the one corresponding to @xmath6 is a non - symmetric diagram .
therefore , relations between @xmath64 and the conventional non - linearity parameters @xmath65 are given by @xmath66 for @xmath67-point function , we write the non - linearity parameters as @xmath68 in fig .
[ fig : three_five.eps ] , we show the diagrams for @xmath69-point functions . cases . ] a relation between @xmath70 and the conventional non - linearity parameter @xmath2 is given by @xmath71 let us suppose that the highest order of the non - linearity parameters appearing in the inequalities is @xmath72 .
we can construct the series of inequalities in a following way .
since @xmath61 is a non - linearity parameter corresponding to the symmetric diagram , it can be written as @xmath73 where @xmath74 is a vector associated with a diagram @xmath75 which is one of the sub - diagrams generated by dividing the symmetric diagram into two diagrams that are isomorphic to each other . therefore , @xmath75 is a tree diagram with @xmath76 vertices and with one external line .
next , let us consider a quantity @xmath77 for @xmath78 .
this must be a non - linearity parameter whose corresponding diagram is a diagram with @xmath58 vertices and is given by connecting the external line of @xmath75 with that of @xmath79 . since this diagram is a non - symmetric diagram , there exists some @xmath80 such that @xmath81 applying the cauchy - schwarz inequality to eqs .
( [ highest1 ] ) and ( [ highest2 ] ) yields an inequality , @xmath82 a number of the independent inequalities of this kind is @xmath83 .
these are the inequalities among the non - linearity parameters for @xmath58-point function .
let us then lower the order by one and consider a quantity , @xmath85 this is a non - linearity parameter whose corresponding diagram is obtained by connecting the external line of @xmath75 with that of @xmath86 .
therefore , there exists some @xmath80 such that @xmath87 meanwhile , since a quantity @xmath88 is a non - linearity parameter whose corresponding diagram is obtained by connecting the external lines of the same two @xmath86 , we have @xmath89 applying again the cauchy - schwarz inequality to eqs .
( [ nhighest1 ] ) and ( [ nhighest2 ] ) yields @xmath90 a number of the independent inequalities of this kind is @xmath91 .
these are the inequalities among the non - linearity parameters for @xmath92 and @xmath93-point functions . let us further lower the order by one and consider a quantity @xmath94 this is a non - linearity parameter
whose corresponding diagram is obtained by connecting the external line of @xmath75 with that of @xmath95 .
since this diagram can be both symmetric or non - symmetric , we write it as @xmath96 applying the same reasoning as before , we obtain @xmath97 a number of the independent inequalities of this kind is @xmath98 .
these are the inequalities among the non - linearity parameters for @xmath99 and @xmath100-point functions . by repeating the above procedures until the order reduces to one ,
we can obtain all the cauchy - schwarz type inequalities that involve the non - linearity parameters for the @xmath58-point function as the highest order .
in particular , at the final procedure , we end up with getting a set of inequalities : @xmath101 a number of the independent inequalities of this kind is @xmath56 . combining all these results ,
a number @xmath102 of independent inequalities that involve the non - linearity parameters for the @xmath58-point function as the highest order is given by @xmath103 for example , @xmath104 .
there are two different tree diagrams for the four - point function .
one of them is a symmetric diagram which corresponds to @xmath5 and the other one is a non - symmetric diagram corresponding to @xmath6 .
according to the general argument we developed in the last subsection , there is no inequality between @xmath5 and @xmath6 because @xmath105 .
then , let us lower the order by one . at this stage , we obtain one inequality @xmath106 in terms of the conventional non - linearity parameters , this can be written as @xmath107 since we have reduced to the lowest order , there are no other inequalities . since @xmath108 , there exists a single inequality among the non - linearity parameters for the six - point function : @xmath109 where the correspondence between the diagrams and the non - linearity parameters is given in fig .
[ fig : symmetric_diagram.eps ] . by lowering the order by one ,
we obtain @xmath110 inequalities : @xmath111 reducing further the order by one , we obtain @xmath112 inequalities : @xmath113 the last two inequalities were also provided in @xcite . since we have reduced to the lowest order , there are no other inequalities .
total number of inequalities is five . since @xmath114
, there exist six inequalities among the non - linearity parameters for the eight - point function : @xmath115 where the correspondence between the diagrams and the non - linearity parameters is given in fig .
[ fig : n8.eps ] .
-point function.,height=566 ] by lowering the order by one , we obtain @xmath116 inequalities : @xmath117 where the correspondence between the diagrams and the non - linearity parameters is given in fig .
[ fig : n7.eps ] . -point function . ]
reducing further the order by one , we obtain @xmath118 inequalities : @xmath119 reducing further the order by one , we obtain @xmath120 inequalities : @xmath121 since we have reduced to the lowest order , there are no other inequalities . total number of inequalities is 22 .
the primordial non - gaussianity has been focused on by many authors as a new probe of the inflation dynamics .
in addition to the bispectrum , use of the higher order correlation functions will become useful in the future .
in particular , the inequality between @xmath5 and @xmath6 shows the importance of the trispectrum to look for the non - gaussianity in the primordial perturbations . in this paper , we developed a general formalism to construct the cauchy - schwarz type inequalities among the non - linearity parameters for the higher order correlation functions .
this method enables us to derive all the inequalities up to any - point function . as an application , we explicitly derived all the inequalities up to the four , six and eight - point functions .
we first confirmed that there is just one inequality , which is eq .
( [ ft ] ) , up to the four - point function .
up to the six - point function , there are five new inequalities two of which were given in ref . @xcite .
up to the eight - point function , there are further 22 new inequalities
. + * acknowledgments : * t. s. is supported by a grant - in - aid for jsps fellows
s. y. is partially supported by the grant - in - aid for scientific research from the ministry of education , science , sports , and culture , japan , no .
22340056 . s.
y. also acknowledges support from the grant - in - aid for the global coe program `` quest for fundamental principles in the universe : from particles to the solar system and the cosmos '' from mext , japan .
the authors also thank the yukawa institute for theoretical physics at kyoto university , where this work was discussed during the yitp - t-10 - 05 on `` cosmological perturbation and cosmic microwave background '' .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl . * 192 * , 18 ( 2011 ) [ arxiv:1001.4538 [ astro-ph.co ] ] .
l. boubekeur and d. h. lyth , phys .
d * 73 * , 021301 ( 2006 ) [ arxiv : astro - ph/0504046 ] .
j. smidt , a. amblard , c. t. byrnes , a. cooray , a. heavens and d. munshi , phys .
d * 81 * , 123007 ( 2010 ) [ arxiv:1004.1409 [ astro-ph.co ] ] .
t. suyama and m. yamaguchi , phys .
d * 77 * , 023505 ( 2008 ) [ arxiv:0709.2545 [ astro - ph ] ] . n. kogo and e. komatsu , phys .
d * 73 * , 083007 ( 2006 ) [ arxiv : astro - ph/0602099 ] .
t. suyama , t. takahashi , m. yamaguchi and s. yokoyama , jcap * 1012 * , 030 ( 2010 ) [ arxiv:1009.1979 [ astro-ph.co ] ] .
n. s. sugiyama , e. komatsu and t. futamase , arxiv:1101.3636 [ gr - qc ] . c. lin and y. wang , jcap * 1007 * , 011 ( 2010 ) [ arxiv:1004.0461 [ astro-ph.co ] ] .
j. meyers and n. sivanandam , arxiv:1104.5238 [ astro-ph.co ] . c. t. byrnes , k. koyama , m. sasaki and d. wands , jcap * 0711 * , 027 ( 2007 ) [ arxiv:0705.4096 [ hep - th ] ] .
a. a. starobinsky , jetp lett .
* 42 * , 152 ( 1985 ) [ pisma zh .
* 42 * , 124 ( 1985 ) ] .
m. sasaki and e. d. stewart , prog .
phys . * 95 * , 71 ( 1996 ) [ arxiv : astro - ph/9507001 ] .
m. sasaki and t. tanaka , prog .
phys . * 99 * , 763 ( 1998 ) [ arxiv : gr - qc/9801017 ] .
d. h. lyth , k. a. malik and m. sasaki , jcap * 0505 * , 004 ( 2005 ) [ arxiv : astro - ph/0411220 ] .
d. h. lyth and y. rodriguez , phys .
lett . *
95 * , 121302 ( 2005 ) [ arxiv : astro - ph/0504045 ] .
j. riordan , _ `` introduction to combinatorial analysis '' , dover publications , inc . , mineola , new york _ | for the local - type primordial perturbation , it is known that there is an inequality between the bispectrum and the trispectrum . by using the diagrammatic method
, we develop a general formalism to systematically construct the similar inequalities up to any order correlation function . as an application , we explicitly derive all the inequalities up to six and eight - point functions .
.5 in * extension of local - type inequality for the higher order correlation functions * .45 in teruaki suyama@xmath0 and shuichiro yokoyama@xmath1 .45 in _
@xmath0 research center for the early universe , graduate school of science , the university of tokyo , tokyo 113 - 0033 , japan + @xmath1department of physics and astrophysics , nagoya university , aichi 464 - 8602 , japan _
.4 in |
the local standard of rest ( lsr ) is defined a point in space that is moving on a perfectly circular orbit around the galactic centre at the solar position .
an accurate estimate of the lsr , namely measuring the peculiar velocities of the sun , is a starting point for galactic kinematic and dynamic studies since all measurements are made relative to the sun .
it is needed for the derivations of basic parameters such as the oort constants , which describe the local spatial variations of the stellar velocity field ( feast & whitelock 1997 , olling & dehnen 2003 ) , as well as the orbital eccentricity distributions of different stellar populations which can be used to constrain the various scenarios of galactic disk formation ( sales et al .
2009 , wilson et al .
2011 , lee et al .
2011 ) . as a fundamental task of galactic astronomy , an accurate determination of the three galactic space velocity components of the solar motion with respect to the lsr
, @xmath3 , @xmath0 and @xmath4 , has been a long - standing challenge .
similar to the ones given in francis & anderson ( 2009 ) and cokunolu et al .
( 2011 ) , table1 presents an updated summary of determinations ( including results from the current work ) of the solar motion from the literature of over a century .
as the table 1 shows , in spite of utilizing a variety of methods and tracers , the measurements have hitherto failed to converge , especially for the component in the direction of galactic rotation ( @xmath0 ) . in principle , the radial ( @xmath3 ) and vertical ( @xmath4 ) components of the solar motion can be directly obtained by taking the negative of the mean heliocentric velocities of various stellar populations in the solar - neighbourhood that are well relaxed .
the situation for the component in the direction of galactic rotation ( @xmath0 ) is more complicated , due to the non - negligible values of asymmetric drift , @xmath6 , of stars with respect to the lsr
motions that are related to the velocity dispersions of the individual stellar populations of concern . the classical method to derive @xmath0
is based on strmberg s equation ( strmberg 1946 ) , which assumes a linear correlation between the mean negative heliocentric azimuthal velocity , @xmath7 , and the squared radial velocity dispersion , @xmath8 , of any stellar sample .
then a linear extrapolation to @xmath9 for an idealized population yields @xmath10 . theoretically , for any well - relaxed stellar population , the jeans equation , i.e. eq .
( 4.228 ) of binney & tremaine ( 2008 ) , links the asymmetric drift , @xmath6 , with the radial velocity dispersion and the square bracket term ( radial scale lengths and the properties of velocity ellipsoid ) in axisymmetric equilibrium .
the physical underpinning of strmberg s equation is that the square bracket term is a constant for the different stellar populations and the stellar populations used are in axisymmetric equilibrium .
while the assumptions sound reasonable , the canonical value of @xmath11 , derived by dehnen and binney ( 1998 , hereafter db98 ) based on strmberg s equation , has recently been found to be @xmath12 smaller than those determined using open clusters ( piskunov et al .
2006 ) and masers in massive star forming regions ( reid et al . 2009 ; bobylev & bajkova 2010 ; mcmillan & binney 2010 ) as tracers . the possible causes of discrepancies have been addressed recently by a number of investigators .
schnrich et al .
( 2010 ) show that strmberg relation could underestimate @xmath1 due to the presence of a metallicity gradient , which breaks the linear relation between the mean rotation velocity and the squared velocity dispersion of stars binned by colour .
this implies that the square bracket term may vary from one population to another population as represented binned by the individual colour bins .
in addition , the axisymmetric equilibrium of galactic potential , a basic assumption underlying strmberg s equation as mentioned above , can be broken due to the presence of significant non - axisymmetric structures in our galaxy ( e.g. the spiral arms , the central bar ; francis & anderson 2009 ; binney 2010 ) . apart from @xmath0
, estimates of the radial and vertical components of the solar motion , @xmath3 and @xmath4 , can also be biased by the presence of non - axisymmetric structures .
to minimize the effects of asymmetric drifts when estimating @xmath0 , two different methods have been developed that avoid the assumption of a constant square bracket term for the different stellar populations underlying strmberg s equation .
one is to determine the solar peculiar velocity from the offset of the modelled velocity distribution that best matches the observed data ( binney 2010 ; schnrich et al . 2010 ; schnrich & binney 2012 , hereafter sb12 ) .
alternatively , the lsr can also be determined utilizing cold populations of thin disc stars as tracers , under the assumption that they suffer from negligible asymmetric drifts ( francis & anderson 2009 ; cokunolu et al .
compared to the method based on strmberg s equation , the latter approach has the advantage that it is simple and straight forward , and the results do not depend on the assumption that underpins the former method .
the key to a successful application of this technique is to select and define a well representing sample of cold populations of thin disc stars .
it should be emphasised that all the approaches outlined above assume axisymmetric equilibrium .
thus it is essential to select a sample of stars that are least affected by those known non - axisymmetric structures .
more discussions of this issue will be made below . [
cols="^,^,^,^,^,^ " , ] after corrected differential galactic rotation assuming a flat rotation curve of constant @xmath13kms@xmath2 .
after corrected differential galactic rotation by fitting eq.(5 ) ( see also fig.9 ) .
the value of each component is the mean of results deduced from the three sets of initial lsr listed in table 5 .
the value of @xmath0 obtained in the current work by the vdf method is 12 km s@xmath2 smaller than those of binney ( 2010 ) and schnrich et al .
the small differences might be caused by bulk motions that affect the samples used in these work , as discussed earlier .
as fig.6 shows , the presence of the pleiades and/or hyades streams moves the peak of @xmath14 distribution of the gcs local sample toward negative values , leading to their estimates of @xmath1 being12 km s@xmath2 larger than obtained with the lss - gac sample , which shows a much smoother azimuthal velocity distribution and is less affected by the bulk motions of stellar streams .
for the ctds method , we have determined the lsr based on two assumptions .
firstly , we assume that cold populations of thin disc stars selected by applying an orbital eccentricity cut suffer from negligible asymmetric drift effects .
the asymmetric drifts of cold populations of thin disc stars selected from the orbital eccentricity cut can be estimated approximately using eq .
( 4.228 ) of binney & tremaine ( 2008 ) from the velocity dispersions ( @xmath15 ) , velocity ellipsoid , stellar density ( @xmath16 ) and circular speed ( @xmath17 ) .
assume that the tilt angle is toward the galactic center ( binney et al .
2014 ; bdenbender et al .
2014 ) and both @xmath16 and @xmath18 are exponential functions of @xmath19 , @xmath20 @xmath21 where @xmath22 and @xmath23 are the scale lengths of @xmath16 and @xmath18 , respectively . under those assumptions ,
( 4.228 ) of binney & tremaine ( 2008 ) becomes , @xmath24\,,\ ] ] where the intrinsic velocity dispersions of the selected thin disc stars are @xmath25 , @xmath26 and @xmath27 kms@xmath2 , respectively . adopting @xmath28 kpc from sharma et al .
( 2014 ) , @xmath29 kpc , @xmath30 kms@xmath2 and a typical position @xmath31 kpc of the selected stars , we have @xmath32 kms@xmath2 . if using instead a larger scale length @xmath33 kpc from the recent work of chang et al .
( 2011 ) , the estimated value of @xmath6 decrease to 1.68 kms@xmath2 .
the results show that our assumption that cold populations of thin disk stars selected with an orbital eccentricity cut of @xmath34@xmath35 have small asymmetric drifts is a reasonable one .
this is corroborated by the close agreement of values of @xmath1 deduced from the two independent analyses , i.e. those based on the vdf and ctds methods , respectively .
the effects of asymmetric drifts of the current sample of thin disc stars on @xmath1 estimation , if any , are likely to be quite small deduced from the ctds method is only 0.62 kms@xmath2 _ larger _ than that deduced from the vdf method ( which fully accounts for the effects of asymmetric drifts ) , the value of @xmath6 of the current sample of thin disc stars is likely to be smaller than 1 kms@xmath2 . ] . as a further test of the assumption
, we note that stars of different spectral types ( populations ) suffer from different levels of the asymmetric drift effects . for sub - samples of dwarfs of f , g and k spectral type
, the level of those effects can vary from a few to tens of km s@xmath2 ( db98 , schnrich et al .
the fact that all the three of f , g and k sub - samples give consistent results suggest again the assumption is a reasonable one .
as described earlier , the second assumption underlying our analysis with the ctds method is that the sample suffers from little systematic kinematic effects , induced by , for example , non - axisymmetric structures , accretion events or asymmetric mass infall . as already pointed out in section 5.1
, the current sample does not seem to be affected significantly by any moving group .
the validity of this assumption also seems to be supported by the consistent results yielded by the individual sub - samples of different spectral type , considering that those sub - samples probe different depths ( because of their different intrinsic , absolute magnitudes ) .
fig.11 plots the results from the three sub - samples for the case of the db98 initial lsr .
the values of solar motion derived from the sub - samples are presented in table6 . fig.11 and
table6 show that for all the three velocity components of lsr , the values yielded by all the three sub - samples agree within the fitting errors . as a by - product
, the results also suggest that the local disk seems to be radially and vertically stable and circularly symmetric .
we note all the errors quoted for the estimates of lsr presented here are purely formal .
there are potential systematic errors for both the methods of vdf and ctds . for the vdf method
, potential biases may come from the underlying assumptions and approximations , as well as from the parameters adopted to construct the analytical azimuthal velocity distribution . as for the ctds method , we believe that the effects of asymmetric drift are minor , although effects on the level of @xmath362 km s@xmath2 could be present as estimated above , considering that the thin disc stars selected do not have exactly zero eccentricity , i.e. they do not move in perfect circular orbits . although there might be a variety of potential sources of systematic errors affecting the results from both methods , it is reassuring that both methods yield consistent values of @xmath0 within 2@xmath37 .
our final , adopted values of lsr are presented in table 7 , obtained by taking the mean of individual estimates , weighted by the inverse square of the formal errors of individual estimates .
the traditional method based on strmberg s equation fails to produce accurate estimates of the lsr , especially for the @xmath0 component , due to the presence of kinematic and/or metallicity biases addressed recently by a number of investigators . in this work ,
two alternative and independent methods have been used to derive the lsr using a completely new sample of stars selected from the lss - gac dr1 .
the first method derives @xmath0 by fiting the observed azimuthal velocity distribution with an analytical distribution function which has been shown to reproduce well reults from rigorous torus - based dynamics modeling .
we obtain @xmath38kms@xmath2 .
the value is about 12 kms@xmath2 smaller than those derived by binney ( 2010 ) and schnrich et al .
( 2010 ) by applying the same method to the gcs local sample .
we suggest that the small differences are caused by the presence of stellar streams that significantly plagues the gcs sample , but to a much less extent the lss - gac dr1 sample .
the second method derives the lsr directly using cold populations of thin disc stars as tracers .
the method relies on the definition of a proper sample of cold populations of thin disc stars . in this paper
, we propose that the orbital eccentricity serves as a natural quantity to select cold populations of thin disc stars . after applying an orbital eccentricity cut ,
two approaches are adopted to correct for the effects of differential galactic rotation and determine the lsr .
the first approach is applicable to the @xmath39 component only .
the analysis yields @xmath40 km s@xmath2 . the second approach , under the assumption of a flat rotation curve and negligible asymmetric drifts ,
is based on gaussian fitting to the measured velocity distributions , after correcting for the effects of differential galactic rotation .
values of @xmath3 derived from both approaches are consistent with each other and in agreement with that deduced from maser measurements ( bobylev & bajkova 2010 , 2014 ) . from the second approach ,
we find @xmath41 km s@xmath2 , which is consistent within 2@xmath37 with the value deduced by the method of vdf .
our newly derived value of @xmath42 km s@xmath2 is very close to the recent estimate of schnrich ( 2012 ) . both for vdf and ctds methods
, the whole sample is further divided into sub - samples of f , g and k spectral types .
results derived from those sub - samples all yield consistent results , as listed in table 4 and 6 .
the results corroborate the robustness of the lsr determined by both methods
. benefited from the lss - gac dr1 and the ucac4 proper motion catalog , we have built the largest local sample of fgk main - sequence stars within 600 pc of the sun with accurate distances and velocities . by controlling the random errors discussed in section 4 and correcting systematic errors both in radial velocities and in proper motions as discussed in section 3.2 , the sample has enabled us to determine the lsr with the smallest uncertainties hitherto ( see table 1 ) .
our final recommended values of lsr are : ( @xmath3 , @xmath0 , @xmath4 ) = ( @xmath43 , @xmath44 , @xmath45 ) km s@xmath2 ( see table7 ) .
it is a pleasure to thank dr .
j. binney for a critical reading of the manuscript and constructive comments .
this work is supported by national key basic research program of china 2014cb845700 .
the guoshoujing telescope ( the large sky area multi - object fiber spectroscopic telescope , lamost ) is a national major scientific project built by the chinese academy of sciences .
funding for the project has been provided by the national development and reform commission .
lamost is operated and managed by the national astronomical observatories , chinese academy of sciences .
liu x. -w . , yuan h. -b . , huo z. -y .
et al . , 2014 , in feltzing s. , zhao g. , walton n. , whitelock p. , eds , proc .
298 , setting the scene for gaia and lamost , cambridge university press , pp .
310 - 321 , preprint ( arxiv : 1306.5376 ) | we re - estimate the peculiar velocity of the sun with respect to the local standard of rest ( lsr ) using a sample of local stars within 600pc of the sun , selected from the large sky area multi - object fiber spectroscopic telescope ( lamost ; also named the guoshoujing telescope ) spectroscopic survey of the galactic anti - centre ( lss - gac ) .
the sample consists of 94,332 fgk main - sequence stars with well - determined radial velocities and atmospheric parameters . to derive the lsr ,
two independent analyses are applied to the data .
firstly , we determine the solar motion by comparing the observed velocity distribution to that generated with the analytic formulism of schnrich & binney that has been demonstrated to show excellent agreement with rigorous torus - based dynamics modelling by binney & mcmillan .
secondly , we propose that cold populations of thin disc stars , selected by applying an orbital eccentricity cut , can be directly used to determine the lsr without the need of asymmetric drift corrections .
both approaches yield consistent results of solar motion in the direction of galactic rotation , @xmath0 , that are much higher than the standard value adopted hitherto , derived from strmgren s equation .
the newly deduced values of @xmath1 are 12 kms@xmath2 smaller than the more recent estimates derived from the geneva - copenhagen survey ( gcs ) sample of stars in the solar neighbourhood ( within 100pc ) .
we attribute the small difference to the presence of several well - known moving groups in the gcs sample that , fortunately , hardly affect the lss - gac sample .
the newly derived radial ( @xmath3 ) and vertical ( @xmath4 ) components of the solar motion agree well with the previous studies .
in addition , for all components of the solar motion , the values yielded by stars of different spectral types in the lss - gac sample are consistent with each other , suggesting that the local disk is well relaxed and that the lsr reported in the current work is robust .
our final recommended lsr is , @xmath5 .
galaxy : kinematics and dynamics galaxy : solar neighbourhood galaxy : stellar content galaxy : fundamental parameters . |
99 joachim c , gimzewski j k , schlittler r r and chavy c 1995 _ phys .
lett . _ * 74 * 2102 one should be cautious here because when the fermi level is moved the charge on the molecule changes and therefore the hamiltonian and by extension the transmission coefficents are modified .
however , if @xmath2 remains inside the hl gap the overall charge changes are negligible and the mean - field hamiltonian remains the same . | we study the electronic and transport properties of two novel molecular wires made of atomic chains of carbon atoms ( polyynes ) capped with either , benzene - thiols or pyridines . while both molecules are structurally similar , the electrical conductance of benzene - thiol - capped chains attached to gold electrodes is found to be much higher than that of pyridine - capped chains .
we predict that the conductance is almost independent of molecular length , which suggests that these molecules could be ideal molecular wires for sub-10 nm circuitry . both systems exhibit negative differential resistance ( ndr ) but its origin and characteristics depend on the type of molecule . we find a novel type of ndr mechanism produced by the movement of the lumo resonance with bias .
we also show that by gating the pyridine - capped molecules it is possible to make the ndr disappear and dramatically modify the @xmath0-@xmath1 characteristics and the length dependence .
molecular wires are currently receiving increasing attention due to recent experimental @xcite and theoretical @xcite advances .
these systems could be candidates for substituting silicon components in nanoscale circuits once the atomic limit is approached .
however , progress in this field has been hindered by the extremely low conductance found in the vast majority of molecules .
this limiting property is a consequence of the location of the injection energy , defined by the fermi level @xmath2 of the leads , which typically lies in the homo - lumo ( hl ) gap .
such a level misalignment moves the electronic transport to the tunneling regime , where the current decreases exponentially as the length of the molecule increases , and makes it very difficult to use molecular wires as interconnects .
for this reason the search for molecular wires whose resonances are broad enough to reach the fermi energy or whose level misalignment is small for any molecular length is of paramount importance for this field .
recently polyyne - based molecular wires , made with carbon chains with an even number of atoms , contacted directly to gold leads by sulphur atoms @xcite were predicted to have a high length - independent conductance and ohmic behavior for a large range of biases .
unfortunately these molecules can not be made in large quantities and contacted to leads because they are extremely reactive @xcite .
one possible way of overcoming this crucial deficiency is by capping the carbon chains with aromatic rings , which makes them much more stable and easier to manipulate @xcite .
the result is a new type of molecules made of a very high conducting backbone , the polyyne , connected to two less conducting rings . in this article
we study the electronic and transport properties of the above mentioned polyynes capped with benzene - thiol ( b-@xmath3-b ) or pyridine groups ( p-@xmath3-p ) for @xmath3 between 1 and 6 ( between 2 and 12 atoms in the carbon chain ) .
an example can be seen in fig .
( [ fig1 ] ) for @xmath4 .
we predict that this new family of molecular wires not only possesses an almost length - independent conductance , but also exhibits negative differential resistance ( ndr ) , which results in a reduction of the current as the bias voltage increases .
ndr is very important in the field of electronic technology and also in basic research . since its discovery @xcite lots of applications in the area of semiconductor physics have been found , which include amplification @xcite , digital applications @xcite , and oscillators @xcite . in the context of molecular electronics
there are various types of mechanisms that can lead to the phenomenon of ndr .
one is due to the movement of resonances to the bulk silicon gap as the bias is applied @xcite , which allows the possibility of tuning the ndr peak by varying the coupling between the stm tip and the molecule .
others include chemical changes @xcite , the destruction of conductance resonances as a consequence of the misalignment of localized or interface states @xcite and local orbital symmetry matching @xcite .
here we show the presence of two types of ndr , which depend on the type of end - group that cap the polyyne .
the first one , which is found in the b-@xmath3-b molecules , is due to the destruction of the homo resonance by the bias - induced asymmetry ; the second one , which is found in the p-@xmath3-p molecules , is a novel class of ndr produced by the bias - induced movement of the lumo resonance . to obtain these predictions we used density functional theory ( dft ) @xcite as implemented in the siesta code @xcite , which employs norm - conserving pseudopotentials to get rid of the core electrons and linear combinations of atomic orbitals to expand the valence states .
the basis set included double zetas and polarization orbitals ( dzp ) both in the simulations to perform the structural relaxations and the transport calculations .
we used the local density approximation ( lda ) to compute the exchange and correlation energy , which works rather well for light elements and systems where electrons are delocalized .
the hamiltonian , overlaps and electronic densities were evaluated in a real space grid defined with a plane wave cutoff of 200 ry .
the dimensions of the unit cell were long enough along @xmath5 and @xmath6 to avoid overlaps and strong electrostatic interactions with molecular images .
the molecular coordinates were relaxed until all forces were smaller than 0.05 ev / . to obtain the transport properties we used the smeagol code @xcite ,
which is interfaced to siesta and uses the non - equilibrium green s functions formalism ( negf ) to calculate self - consistently the charge density , the transmission coefficients and the @xmath0-@xmath1 characteristics .
the system was divided in three parts : a left lead , a right lead and an extended molecule ( molecule + part of the leads modified by the presence of the molecule and the surfaces ) and a separate calculation was performed for each of them @xcite . in the bulk calculation
the unit cell was made of three slices of fcc gold grown along ( 111 ) , with 9 atoms per slice and periodic boundary conditions along @xmath5 , @xmath6 and @xmath7 , but only @xmath8-points along @xmath7 , which we chose as the transport direction . in the transport calculation we had to include one of the previous unit cells on each side plus two slices of the same material on the left and right parts of the extended molecule , respectively .
each transport calculation involved around 120 atoms .
polyyne - based molecular wires .
( a ) polyyne between benzene - thiol groups , b-@xmath3-b , and ( b ) polyyne between pyridine groups , p-@xmath3-p , for @xmath4 . ] the transmission coefficients as a function of the bias voltage produced by the molecule with thiol connections for @xmath4 carbon pairs ( i.e. 6 atoms ) are shown in fig .
( [ fig2 ] ) ( a ) .
the molecule is connected to the three - atoms hollow site , which gives a strong bond to the surface and therefore produces broad resonances and high conductances at the fermi level . at zero bias
the transmission coefficients show the typical two - peak structure associated with the homo and lumo levels .
however , compared to the case of pure polyynes capped by sulphur atoms @xcite these peaks have a maximum value of only unity .
this difference is due to the fact that only a single channel at the fermi level is sustained by the aromatic rings compared with the carbon chain which has two .
these channels come from the @xmath9 orbitals , which are weakly bonded states and therefore have an energy much closer to the fermi level than the strongly bonded @xmath10 states .
the number of available @xmath9 states is two on the polyyne because on each carbon atom the hybridization is @xmath11 and only two sigma bonds couple directly to other carbon atoms , leaving two weakly - bonded @xmath9 orbitals . in the aromatic rings , however , the hybridization is @xmath12 and the third sigma bond couples to the hydrogen atom , leaving only one @xmath9 state on the ring .
the highest electrical resistance is therefore associated with the rings . when a bias is applied the most noticeable effect
is the different behavior of the homo and lumo resonances .
while the homo resonance is destroyed as the absolute value of the bias increases , the lumo resonance remains the same .
this different behavior can be understood by looking at the spatial distribution of the associated levels .
the state related to the homo resonance is largely localized on the polyyne chain and the sulphur atoms , while it has little weight on the rings .
since these three regions are only weakly coupled the outer parts move differently under an applied bias , i.e. one to lower energies and the other to higher energies , and their electronic occupations change .
this can be clearly seen if an electric field is applied to the isolated molecule with hydrogens attached to the sulphurs .
the spatial projection of the homo shows that the weight on one of the sulphurs increases while the weight on the other sulphur decreases .
the state becomes therefore asymmetric and , in accordance with the breit - wigner model of resonances @xcite , the transmission peak is reduced , as seen in fig .
( [ fig2 ] ) ( a ) @xcite .
the case of the lumo is however rather different because this state is mainly localized on the polyyne chain and is not modified by the bias voltage .
therefore the shape of the lumo resonance does not change under bias and the tunneling probability through it remains the same , as can be seen in fig .
( [ fig2 ] ) ( a ) .
the shift of the homo and lumo resonances with increasing bias is produced by charge transfer to the leads , which moves the states to lower energies @xcite .
this charge transfer is due to asymmetric partial occupation of states within the bias window near the homo or lumo resonances .
if the fermi level is close to the homo , like in this case , the weight inside the bias window is larger for states that were before completely occupied than for states that were before empty
. therefore the partial occupations do not sum up to the previous value and the molecule looses charge .
if the fermi level is however close to the lumo the effect is the opposite and the molecule wins charge , as we will see later .
fig .
( [ fig4 ] ) ( a ) shows that the zero bias conductance decreases by less than 2 as @xmath3 increases from 1 to 6 .
this trend , which is similar to what was found in pure polyynes @xcite and agrees with recent preliminary experiments @xcite , is a significant result because it proves it is possible to have molecular wires that can connect different parts of a circuit with almost the same conductance .
transport properties of b-@xmath3-b molecules .
( a ) transmission coefficients as a function of the bias potential for @xmath4 .
( b ) @xmath0-@xmath1 characteristics as a function of the number of carbon pairs ( @xmath3 ) in the chain . ] the presence of ndr in these systems , which is shown in fig .
( [ fig2 ] ) ( b ) , arises from the sensitivity of the homo resonance to the applied bias .
since the current is calculated by integrating the transmission coefficient over the bias window , @xmath13 , the suppression of the homo resonance can cause the current to decrease with bias because the reduction of the transmission compensates the increase of the bias window . for higher voltages the window reaches again regions with larger conductances and
the current grows again .
the shape of the ndr and the location of its maximum are very similar in all cases , which is a consequence of the analogous behavior of all these molecules under bias .
transport properties of of p-@xmath3-p molecules .
( a ) transmission coefficients as a function of the bias potential .
( b ) @xmath0-@xmath1 characteristics as a function of the number of carbon pairs ( @xmath3 ) in the chain . ]
when the benzene - thiol rings are substituted by pyridines , the presence of nitrogen on the rings produces a series of effects which greatly modify the behavior of the whole system . on the one hand ,
the higher electronegativity of nitrogen compared to that of sulphur and carbon moves all molecular states downwards , positioning the fermi level closer to the lumo .
on the other hand , the most stable bonding configuration to the surface moves from the hollow to the top position , which dramatically reduces the coupling to the electrodes .
this reduction can be clearly seen in the transmission coefficients of fig .
( [ fig3 ] ) ( a ) , where the resonances are much sharper than in the b-@xmath3-b case .
the length dependence is even more interesting than in the b-@xmath3-b case . as can be seen in fig .
( [ fig4 ] ) ( b ) , the zero bias conductance increases now with @xmath3 , which is a surprising result .
this behavior , which is due to the movement of the lumo to lower energies as the length increases , is not seen in recent experiments @xcite , where a smooth decrease in conductance with @xmath3 was observed .
the predicted length dependence is however sensitive to the relative position of @xmath2 , whose calculated value may be incorrectly predicted by dft @xcite and experimentally can be affected by the ambient environment @xcite .
fig .
( [ fig4 ] ) ( c ) shows that decreasing @xmath2 by 0.4 ev @xcite , causes the conductance to decrease , rather than increase with @xmath3 .
this striking change of tendency can be explained by taking into account two competing effects . on the one hand
, the increasing distance between the leads reduces exponentially the transmission in the hl gap , which dramatically reduces the conductance . on the other hand ,
the increasing electronic delocalization in longer molecules decreases the level repulsion and moves the states downwards , which increases the conductance near the lumo . as a consequence ,
depending on wether the fermi level is closer or not to the lumo one effect dominates the other and the conductance tendency can change .
this result proves that the length dependence of these molecular wires can be substantially modified and even reversed by a gate voltage . as the bias increases , the lumo resonance in fig .
( [ fig3 ] ) ( a ) moves to higher energies , in contrast to fig .
( [ fig2 ] ) ( a ) .
this trend can again be explained by charge transfer under bias , which in this case moves back to the molecule .
the reason now is due to the location of the fermi level near the lumo . the weight inside the bias window of states that were previously unoccupied is bigger than the weight of states that were completely occupied and the molecule wins charge , which moves the states to higher energies
this movement is the main responsible of the ndr effect .
as the bias window increases it encounters first the lumo resonance but at some bias this state rises faster than the bias window and the current decreases .
notice also the ndr is much more pronounced in this case compared to the b-@xmath3-b system and the positions of the maxima also depend on voltage .
the position of the minimum and subsequent rise in the @xmath0-@xmath1 characteristics moves to lower voltages with increasing @xmath3 , reflecting the decrease in the hl gap with increasing length .
the ndr however disappears on the same range of bias voltages if the fermi level is moved to lower energies , as can be seen in fig .
( [ fig4 ] ) where the current is calculated by integrating the bias - dependent transmission coefficients on a bias window centered at @xmath14 ev . the continuous increase in the current
is produced now by the proximity of the homo resonance which moves to higher energies as the bias increases . by gating these molecules
it is therefore possible to alter significantly their @xmath0-@xmath1 characteristics and tune the appearance of ndr .
zero - bias conductance of ( a ) b-@xmath3-b and ( b ) p-@xmath3-p molecules calculated with the fermi level provided by the calculation .
( c ) zero - bias conductance and ( d ) @xmath0-@xmath1 characteristics for p-@xmath3-p molecules calculated at @xmath14 ev . ] in conclusion , we have studied the electronic and transport characteristics of two novel types of molecules , b-@xmath3-b and p-@xmath3-p , using a combination of density functional theory and non - equilibrium green s functions formalism .
we found very interesting properties which make these systems rather promising from a technological and fundamental point of view .
the most important result is an almost constant length dependence of the conductance , which is very smooth and under some circumstances can even increase with the number of carbon pairs in the chain .
this allowed us to predict that it is possible to alter dramatically the length dependence of molecular wires by using a gate voltage .
we also found the presence of two types of negative differential resistance . in the b-@xmath3-b molecule
the ndr is due to the destruction of the homo resonance under bias , while in the p-@xmath3-p the movement of the lumo is the main responsible behind the decrease of the current .
we also found that the shape of the ndr was almost constant in the first molecule while in the second it depended on the molecular length .
finally , we showed that by gating the p-@xmath3-p molecules it is possible to alter the shape of the @xmath0-@xmath1 characteristics quite significantly and make the ndr disappear for the same range of bias voltages .
we thank martin bryce , richard nichols , wolfgang haiss , simon higgins and santiago martn for useful discussions .
this work was supported by the european comission , qinetiq , and the british epsrc , department of trade and industry , royal society and northwest regional development agency . |
two mobile agents , starting from different nodes of a network , have to meet at the same node at the same time .
this distributed task is known as _ rendezvous _ and has received a lot of attention in the literature .
agents can be any mobile autonomous entities .
they might represent human - made objects , such as software agents in computer networks or mobile robots navigating in a network of corridors in a building or a mine .
they might also be natural , such as animals seeking a mate , or people who want to meet in an unknown city whose streets form a network .
the purpose of meeting in the case of software agents or mobile robots might be the exchange of data previously collected by the agents or samples collected by the robots .
it may also be the coordination of future network maintenance tasks , for example checking functionality of websites or of sensors forming a network , or decontaminating corridors of a mine .
the network is modeled as an undirected connected graph with @xmath0 unlabeled nodes .
we seek deterministic rendezvous algorithms that do not rely on the agents perceiving node identifiers , and therefore can work in anonymous graphs as well ( cf .
the reason for designing such algorithms is that , even when nodes of the network have distinct identifiers , agents may be unable to perceive them because of limited sensory capabilities ( e.g. , a mobile robot may be unable to read signs at corridor crossings ) , or nodes may be unwilling to reveal their identifiers to software agents , e.g. , due to security or privacy reasons . from a methodological point of view , if nodes had distinct identifiers visible to the agents , the agents could explore the graph and meet at the node with the smallest identifier .
hence , in this case , rendezvous reduces to graph exploration . on the other hand , we assume that , at each node @xmath7 , each edge incident to @xmath7 has a distinct _ port number _ from the set @xmath8 , where @xmath9 is the degree of @xmath7 .
these port numbers are fixed and visible to the agents .
port numbering is _ local _ to each node , i.e. , we do not assume any relation between port numbers at the two endpoints of an edge .
note that in the absence of port numbers , edges incident to a node would be undistinguishable for agents and thus rendezvous would be often impossible , as an agent may always miss some particular edge incident to the current node , and this edge could be a bridge to the part of the graph where the other agent started . the previously mentioned security and privacy reasons for not revealing node identifiers to software agents are irrelevant in the case of port numbers . if the graph models a system of corridors of a mine or a building , port numbers can be made implicit , e.g. , by marking one edge at each intersection ( using a simple mark legible even by a mobile robot with very limited vision ) , considering it as corresponding to port 0 , and all other port numbers increasing clockwise .
agents are initially located at different nodes of the graph and traverse its edges in synchronous rounds .
they can not mark visited nodes or traversed edges in any way , and they can not communicate before meeting .
the adversary wakes up each of the agents , possibly in different rounds .
each agent starts executing the algorithm in the round of its wake - up .
it has a clock that ticks at each round and starts at the wake - up round of the agent . in each round
, each agent either remains at the current node , or chooses a port in order to move to one of the adjacent nodes .
when an agent enters a node , it learns the node s degree and the port number by which it enters the node .
when agents cross each other on an edge while traversing it simultaneously in different directions , they do not notice this fact .
each agent has a distinct integer label from a fixed _ label space _ @xmath1 , which it can use as a parameter in the same deterministic algorithm that both agents execute .
it does not know the label nor the starting round of the other agent .
since we study deterministic rendezvous , the absence of distinct labels would preclude the possibility of meeting in highly symmetric graphs , such as rings or tori , for which there exist non - trivial port - preserving automorphisms .
indeed , in such graphs , identical agents starting simultaneously and executing the same deterministic algorithm can never meet , since they will keep the same positive distance in every round .
hence , assigning different labels to agents is the only way to break symmetry , as is needed to meet in every graph using a deterministic algorithm . on the other hand ,
if agents knew each other s identities , then the smaller - labelled agent could stay inert , while the other agent would try to find it . in this case rendezvous
reduces to graph exploration . assuming such knowledge , however , is not realistic , as agents are often created independently in different parts of the graph , and they know nothing about each other prior to meeting .
the rendezvous is defined as both agents being at the same node in the same round .
the main efficiency measure of a rendezvous algorithm s performance is its _ time _
, i.e. , the number of rounds from the start of the later agent until the meeting . if @xmath2 is the distance between the initial positions of the agents , then @xmath3 is an obvious lower bound on the time of rendezvous .
however , if the agents have no additional knowledge , time @xmath4 is usually impossible to achieve .
this is due to two reasons .
first , without any knowledge about the graph , even the easier task of _ treasure hunt _ @xcite , in which a single agent must find a target ( treasure ) hidden at an unknown node of the graph , takes asymptotically larger time in the worst case .
treasure hunt is equivalent to a special case of rendezvous where one of the agents is inert . in the worst case , this takes as much time as graph exploration , i.e. , having a single agent visit all nodes .
second , even when the graph is so simple that navigation of the agents is not a problem , breaking symmetry between the agents , which is often necessary to achieve a meeting , may take time larger than @xmath2 . indeed , even in the two - node graph , where @xmath10 , rendezvous requires time @xmath11 @xcite .
we study the amount of information that has to be given _ a priori _ to the agents to achieve rendezvous in optimal time @xmath5 . following the paradigm of _ algorithms with advice _
@xcite , this information is provided to the agents at the start , by an oracle knowing the entire instance of the problem , i.e. , the graph , the starting positions of the agents , their wake - up rounds , and both of their labels .
the oracle helps the agents by providing them with the _ same _ binary string called _ advice _ , which can be used by each agent , together with its own label , during the execution of the algorithm .
the length of this string is called the _ size of advice_. our goal is to find the smallest size of advice ( up to constant factors ) which enables the agents to meet in time @xmath5 . in other words
we want to answer the question : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ what is the minimum information that permits the fastest possible rendezvous ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ where both `` minimum '' and `` fastest '' are meant up to multiplicative constants .
notice that , since the advice given to both agents is identical , it could not help break symmetry if agents did not have distinct labels . hence , even with large advice , the absence of distinct labels would preclude rendezvous in highly symmetric networks , as argued above . using the framework of advice permits us to quantify the amount of information needed for an efficient solution of a given network problem ( in our case , rendezvous ) , regardless of the type of information that is provided . for agents with labels from the set @xmath1 ,
we show that , in order to meet in optimal time @xmath5 in @xmath0-node networks , the minimum size of advice that has to be provided to the agents is @xmath6 .
the upper bound is proved by constructing an advice string of this size , and providing a natural rendezvous algorithm using this advice that works in time @xmath5 for all networks .
the matching lower bound , which is the main contribution of this paper , is proved by exhibiting classes of networks for which it is impossible to achieve rendezvous in time @xmath5 with smaller advice .
our algorithm works for arbitrary starting times of the agents , and our lower bound is valid even for simultaneous start . as far
as the memory of the agents is concerned , our algorithm has very modest requirements : an agent must only be able to store the advice and its own label .
hence memory of size @xmath12 is sufficient . on the other hand , our lower bound on
the size of advice holds even for agents with unlimited memory .
the problem of rendezvous has been studied both under randomized and deterministic scenarios .
an extensive survey of randomized rendezvous in various models can be found in @xcite , cf .
also @xcite . deterministic rendezvous in networks has been surveyed in @xcite .
several authors considered geometric scenarios ( rendezvous in an interval of the real line , e.g. , @xcite , or in the plane , e.g. , @xcite ) .
gathering more than two agents was studied , e.g. , in @xcite .
for the deterministic setting , many authors studied the feasibility and time complexity of rendezvous .
for instance , deterministic rendezvous of agents that are equipped with tokens used to mark nodes was considered , e.g. , in @xcite .
most relevant to our work are the results about deterministic rendezvous in arbitrary graphs , when the two agents can not mark nodes , but have unique labels @xcite . in @xcite ,
the authors present a rendezvous algorithm whose running time is polynomial in the size of the graph , in the length of the shorter label and in the delay between the starting times of the agents . in @xcite ,
rendezvous time is polynomial in the first two of these parameters and independent of the delay .
memory required by the agents to achieve deterministic rendezvous was studied in @xcite for trees and in @xcite for general graphs .
memory needed for randomized rendezvous in the ring is discussed , e.g. , in @xcite . apart from the synchronous model used in this paper , several authors investigated asynchronous rendezvous in the plane @xcite and in network environments @xcite . in the latter scenario ,
the agent chooses the edge to traverse , but the adversary controls the speed of the agent . under this assumption ,
rendezvous at a node can not be guaranteed even in very simple graphs .
hence the rendezvous requirement is relaxed to permit the agents to meet inside an edge .
providing nodes or agents with arbitrary kinds of information that can be used to perform network tasks more efficiently has been proposed in @xcite .
this approach was referred to as _ algorithms with advice_. advice is given either to nodes of the network or to mobile agents performing some network task .
in the first case , instead of advice , the term _ informative labeling schemes _ is sometimes used .
several authors studied the minimum size of advice required to solve network problems in an efficient way . in @xcite , given a distributed representation of a solution for a problem , the authors investigated the number of bits of communication needed to verify the legality of the represented solution . in @xcite
the authors compared the minimum size of advice required to solve two information dissemination problems using a linear number of messages . in @xcite
it was shown that advice of constant size given to the nodes enables the distributed construction of a minimum spanning tree in logarithmic time . in @xcite
the advice paradigm was used for online problems . in @xcite
the authors established lower bounds on the size of advice needed to beat time @xmath13 for 3-coloring cycles and to achieve time @xmath13 for 3-coloring unoriented trees . in the case of @xcite
the issue was not efficiency but feasibility : it was shown that @xmath14 is the minimum size of advice required to perform monotone connected graph clearing . in @xcite
the authors studied radio networks for which it is possible to perform centralized broadcasting in constant time .
they proved that constant time is achievable with @xmath15 bits of advice in such networks , while @xmath16 bits are not enough . in @xcite
the authors studied the problem of topology recognition with advice given to nodes . in @xcite
the task of drawing an isomorphic map was executed by an agent in a graph and the problem was to determine the minimum advice that has to be given to the agent for the task to be feasible . among the papers using the paradigm of advice ,
@xcite are closest to the present work , as they both concern the task of graph exploration by an agent . in @xcite
the authors investigated the minimum size of advice that has to be given to unlabeled nodes ( and not to the agent ) to permit graph exploration by an agent modeled as a @xmath17-state automaton . in @xcite
the authors established the size of advice that has to be given to an agent in order to explore trees while obtaining competitive ratio better than 2 . to the best of our knowledge ,
rendezvous with advice has never been studied before .
consider any @xmath0 node graph , and suppose that the distance between the initial positions of the agents is @xmath2 . in this section ,
we construct an advice string of length @xmath18 and a rendezvous algorithm which achieves time @xmath2 using this advice .
we first describe the advice string .
let @xmath19 be the underlying graph and let @xmath20 and @xmath21 be the distinct labels of the agents , both belonging to the label space @xmath1 .
call the agent with label @xmath20 the first agent and the agent with label @xmath21 the second agent .
let @xmath22 be the smallest index such that the binary representations of @xmath20 and @xmath21 differ on the @xmath22th bit .
without loss of generality assume that the @xmath22th bit is 0 in the binary representation of @xmath20 and 1 in the binary representation of @xmath21 .
let @xmath23 be a fixed shortest path in @xmath19 between the initial positions @xmath24 and @xmath7 of the agents .
the path @xmath23 induces two sequences of ports of length @xmath2 : the sequence @xmath25 of consecutive ports to be taken at each node of path @xmath23 to get from @xmath24 to @xmath7 , and the sequence @xmath26 of consecutive ports to be taken at each node of path @xmath23 to get from @xmath7 to @xmath24 .
let @xmath27 be the sequence corresponding to the direction from the initial position of the second agent to the initial position of the first agent .
denote @xmath28 .
let @xmath29 , for @xmath30 , be the binary representation of the integer @xmath31 .
additionally , let @xmath32 be the binary representation of the integer @xmath22 .
the binary strings @xmath33 will be called substrings .
the sequence of substrings @xmath33 is encoded into a single advice string to pass to the algorithm .
more specifically , the sequence is encoded by doubling each digit in each substring and putting 01 between substrings .
this permits the agent to unambiguously decode the original sequence .
denote by @xmath34 this encoding and let @xmath35 be the inverse ( decoding ) function , i.e. , @xmath36 . as an example , @xmath37 .
note that the encoding increases the total number of advice bits by a constant factor .
the advice string given to the agents is @xmath38 .
the idea of the algorithm fast rendezvous using the advice string @xmath39 is the following .
each agent decodes the sequence @xmath33 from the string @xmath39
. then each agent looks at the @xmath22th bit of its label , where @xmath22 is the integer represented by @xmath32 .
if this bit is 0 , the agent stays inert at its initial position , otherwise it takes the consecutive ports @xmath40 , where @xmath31 , for @xmath30 , is the integer with binary representation @xmath29 .
after these @xmath2 moves , the agent meets the other agent at the latter s initial position . below is the pseudocode of the algorithm .
[ ub ] let @xmath19 be an @xmath0-node graph with two agents initially situated at distance @xmath2 from one another .
algorithm fast rendezvous achieves rendezvous in time @xmath2 , using advice of size @xmath18 .
the correctness and time of the algorithm are straightforward .
it remains to prove that the length of the advice string @xmath39 is @xmath18 .
to do this , it is enough to show that the sum of the lengths @xmath41 of substrings @xmath29 , for @xmath42 , is @xmath18 .
note that @xmath43 .
also , note that @xmath44 , where @xmath45 are consecutive nodes on some shortest path between the initial positions of the agents .
it is well - known ( cf .
e.g. , @xcite ) that the sum of degrees on a shortest path between any two nodes in an @xmath0-node graph is bounded above by @xmath46 .
hence @xmath47 .
we have @xmath48 the product of @xmath2 positive numbers whose sum is at most @xmath46 is maximized when they are all equal to @xmath49 .
hence @xmath50 .
this concludes the proof .
in this section , we prove a lower bound on the size of advice permitting rendezvous in optimal time @xmath4 , where @xmath2 is the initial distance between the agents .
this lower bound will match the upper bound established in theorem [ ub ] , which says that , for an arbitrary @xmath0-node graph , rendezvous can be achieved in time @xmath4 using advice of size @xmath18 . in order to prove that this size of advice can not be improved in general
, we present two classes of graphs : one that requires advice @xmath51 and another that requires advice @xmath52 to achieve optimal time of rendezvous . to make the lower bound even stronger ,
we show that it holds even in the scenario where agents start simultaneously .
the @xmath51 lower bound will be first proved for the simpler problem of treasure hunt .
recall that in this task , a single agent must find a stationary target ( treasure ) hidden at an unknown node of the graph at distance @xmath2 from the initial position of the agent .
we then show how to derive the same lower bound on the size of advice for the rendezvous problem .
the following technical lemma gives a construction of a graph which will provide the core of our argument for the @xmath51 lower bound .
[ lowerboundgraph ] let @xmath0 and @xmath2 be positive integers such that @xmath53 .
consider any treasure - hunting algorithm @xmath54 that takes @xmath55 bits of advice .
for any fixed even integer @xmath56 and every integer @xmath57 , there exists a graph of size @xmath58 , an initial position of the agent in this graph , and a location of the treasure at distance @xmath59 from this initial position , for which algorithm @xmath54 uses @xmath60 rounds .
we define a class of graphs @xmath61 such that each graph in @xmath61 has @xmath62 nodes .
we will prove that there is a non - empty subset @xmath63 of @xmath61 such that , on each graph in @xmath63 , algorithm @xmath54 uses @xmath60 rounds to complete treasure hunt , for some initial position of the agent and a location of the treasure at distance @xmath59 from this location .
each graph @xmath19 in the class consists of @xmath64 copies of a @xmath17-clique @xmath65 ( with a port numbering to be described shortly ) , which are chained together in a special way .
we will refer to these cliques as @xmath66 .
let @xmath67 denote the nodes of @xmath65 .
it should be stressed that names of nodes in cliques are for the convenience of the description only , and they are not visible to the agent .
we choose an arbitrary edge - colouring of @xmath65 using the colours @xmath68 , which is always possible for cliques of even size , cf .
e.g. , @xcite , theorem 7.6 . for an arbitrary edge @xmath69 in @xmath65 ,
let @xmath70 denote the colour assigned to @xmath69 .
the port numbers of @xmath65 are simply the edge colours , i.e. , for any edge @xmath71 , the port numbers corresponding to this edge at @xmath24 and @xmath7 are both equal to @xmath72 .
each graph @xmath73 is obtained by chaining together the copies @xmath66 of the clique @xmath65 in the following way .
we will call node @xmath74 in clique @xmath75 the _ gate _
@xmath76 of @xmath75 .
the initial position of the agent is @xmath77 .
each gate @xmath76 , for @xmath78 , is placed on ( i.e. , subdivides ) one of the edges of clique @xmath79 not incident to @xmath80 .
we denote this edge by @xmath81 .
finally , an additional _ treasure node _ @xmath82 is placed on ( i.e. , subdivides ) one of the edges of clique @xmath83 not incident to @xmath84 , and this edge is denoted by @xmath85 .
hence @xmath77 has degree @xmath86 , each @xmath76 , for @xmath87 , has degree @xmath88 , and @xmath89 has degree 2 , cf .
figure 1 ( a ) .
note that , since @xmath76 , for @xmath78 , subdivides an edge that is not incident to @xmath80 , we have @xmath90 .
port numbering of graph @xmath19 is the following .
port numbers in each clique @xmath91 are unchanged , the new port numbers at each node @xmath92 , for @xmath93 , are @xmath86 and @xmath17 , with @xmath86 corresponding to the edge whose other endpoint has smaller index , and the new port numbers at node @xmath89 are @xmath94 and @xmath95 , with @xmath94 corresponding to the edge whose other endpoint has smaller index , cf . figure 1 ( b ) .
all graphs in the class @xmath61 are isomorphic and differ only by port numbering .
note that each graph in @xmath61 is uniquely identified by the sequence of edges @xmath96 .
therefore , the number of graphs in @xmath61 is @xmath97 .
( b ) port numbering of each clique @xmath75 , for @xmath98 , with gate @xmath99 inserted ] notice that an agent navigating in a graph @xmath73 always knows when it arrives at a gate @xmath76 , for @xmath87 , because these are the only nodes of degree @xmath88 .
call a walk of an agent _ normal _ , if the agent visits each gate @xmath76 , for @xmath87 , exactly once ( i.e. , never exits a gate by port @xmath86 or @xmath17 ) .
it is enough to prove our lower bound on the time of treasure hunt only for algorithms where the agent always performs a normal walk .
indeed , for any walk , there exists a normal walk using at most the same time . from now on we restrict attention to such algorithms .
we prove our lower bound on the class of graphs @xmath61 .
the idea is that , in order to find the treasure node , the agent must visit each of the nodes @xmath100 . to get from @xmath76 to @xmath99 ,
the agent must find the edge @xmath101 of @xmath75 that the node @xmath99 subdivides . with little advice ,
this amounts to searching many edges of the clique @xmath75 , and hence increases time . for any graph @xmath19
, the agent is given some advice string @xmath102 and executes its treasure - hunting algorithm @xmath54 . with @xmath55 bits of advice
, there exists a set @xmath63 of at least @xmath103 graphs for which the agent is given the same advice string .
next , we provide an upper bound on the number of graphs in @xmath63 . by comparing this upper bound with @xmath103
, we will get the desired lower bound on the number of rounds needed to find the treasure .
let @xmath104 be the maximum running time of algorithm @xmath54 on graphs of class @xmath61 .
let @xmath105 be the function that maps each graph from @xmath106 to an @xmath64-tuple @xmath107 , where , for each @xmath108 , @xmath109 is the number of edge traversals performed by the agent in clique @xmath75 .
this function is well - defined since we consider only deterministic algorithms .
we now prove that this function is injective .
[ injective ] for any two graphs @xmath110 in the set @xmath63 , we have @xmath111 .
let @xmath19 and @xmath112 be represented by the distinct sequences of edges @xmath96 and @xmath113 , respectively .
let @xmath114 be the first index for which @xmath115 .
since the advice string for graphs @xmath19 and @xmath112 is the same , the sequence of ports taken by the agent in these graphs is the same until the agent reaches nodes of different degree in @xmath19 and @xmath112 .
let @xmath116 and @xmath117 .
hence @xmath118 for @xmath119 .
without loss of generality assume that @xmath120 . for each round @xmath121
, the agent takes the same port number in @xmath19 and @xmath112 .
in round @xmath122 the agent reaches the gate @xmath123 in @xmath19 . since @xmath115 , the agent does not reach the gate @xmath123 in @xmath112 in this round .
since the walk of the agent is normal , we conclude that @xmath124 . by claim [ injective ] ,
the number of graphs in @xmath63 is bounded above by the size of the range of @xmath105 .
consider an arbitrary @xmath64-tuple @xmath107 in the range of @xmath105 . by the definition of @xmath61 ,
for each @xmath108 , the agent must traverse at least two edges to get from @xmath76 to @xmath99 .
further , @xmath104 is an upper bound on the number of edge traversals performed in any execution of the algorithm .
therefore , the size of the range of @xmath105 is bounded above by the number of integer - valued @xmath64-tuples with positive terms whose sum is at most @xmath104 .
clearly , this is bounded above by the number of real - valued @xmath64-tuples with non - negative terms whose sum is at most @xmath104 , i.e. , by the size of the simplex @xmath125 . from @xcite ,
the volume of @xmath126 is equal to @xmath127 .
thus , we have shown that the size of @xmath63 is bounded above by @xmath127 . comparing this to our lower bound @xmath103 on the size of @xmath63
, we get @xmath128{\ell ! \frac{n}{2^{dz } } } \geq \sqrt[\ell]{\ell ! \frac{((k-1)(k-2)/2)^{\ell}}{2^{dz } } } \geq \sqrt[\ell]{\ell ! } \frac{(k-2)^{2}/2}{2^{dz/\ell}}\\
= \sqrt[\ell]{\ell !
} \frac{(k-2)^{2}/2}{2^{2z}}.\ ] ] by stirling s formula we have @xmath129 , for sufficiently large @xmath64 .
hence @xmath130{\ell ! } \geq
\ell^{1/(2\ell ) } \cdot ( \ell / e)$ ] .
since the first factor converges to 1 as @xmath64 grows , we have @xmath130{\ell ! } \in \omega(\ell)$ ] .
hence the above bound on @xmath104 implies @xmath131 .
[ lb1 ] let @xmath0 and @xmath2 be positive integers such that @xmath53 .
if an algorithm @xmath54 solves treasure hunting in @xmath4 rounds whenever the treasure is at distance @xmath2 from the initial position of the agent , then there exists an @xmath0-node graph @xmath19 with treasure at this distance such that @xmath54 requires @xmath51 bits of advice .
we suppose that only @xmath132 bits of advice are provided , and show that there is an @xmath0-node graph for which @xmath54 completes treasure hunt in @xmath133 rounds .
* @xmath134 . +
the amount of advice provided is @xmath55 , where @xmath135 . by lemma [ lowerboundgraph ] with @xmath136 or @xmath137 ,
whichever number is even , and @xmath138 , there exists a graph @xmath139 such that , for some positive constant @xmath140 , algorithm @xmath54 uses at least @xmath141 rounds to reach the treasure located at some node @xmath22 at distance @xmath59 from the initial position of the agent . * case 2 .
* @xmath142 .
+ since @xmath143 is constant and @xmath132 bits of advice are provided , it follows that the amount of advice provided is @xmath55 , where @xmath144 for some integer function @xmath145 such that @xmath146 . by lemma [ lowerboundgraph ] with @xmath147 or @xmath148 ,
whichever number is even , and @xmath149 , there exists a graph @xmath139 such that , for some positive constant @xmath140 , algorithm @xmath54 uses at least @xmath150 rounds to reach the treasure located at some node @xmath22 at distance @xmath59 from the initial position of the agent . in both cases ,
graph @xmath112 has @xmath62 nodes and the treasure is located at distance @xmath59 from the initial position of the agent .
we obtain a graph @xmath19 with @xmath0 nodes by attaching a path of @xmath151 nodes to node @xmath22 in @xmath112 . in this graph
@xmath19 the initial position of the agent is unchanged with respect to @xmath112 , and the treasure is located on the attached path at distance @xmath2 from this initial position .
we now show how to deduce a lower bound on the size of advice for rendezvous ( even with simultaneous start ) from the lower bound for treasure hunt .
[ cor ] let @xmath152 be positive integers .
there exist @xmath153 and @xmath154 such that if an algorithm @xmath54 solves rendezvous in time @xmath4 in @xmath0-node graphs whenever the initial distance between the agents is @xmath2 , then there exists an @xmath0-node graph for which @xmath54 requires @xmath51 bits of advice .
let @xmath19 be a @xmath155-node graph with a treasure located at distance @xmath156 from the initial position of the agent such that an @xmath157-time treasure hunting algorithm requires @xmath158 bits of advice .
such a graph exists by theorem [ lb1 ] .
let @xmath7 be the initial position of the agent and let @xmath159 be the location of the treasure in graph @xmath19 .
let @xmath160 and @xmath161 . hence @xmath153 and @xmath154 .
construct the graph @xmath162 which consists of two disjoint copies of @xmath19 with the treasure locations joined by an edge .
locate two agents in @xmath162 , each at the node @xmath7 in different copies of @xmath19 .
the graph @xmath162 has @xmath0 nodes and the initial distance between the agents is @xmath2 . in order to accomplish rendezvous in time @xmath4 ,
at least one of the agents has to traverse the edge joining the copies , hence it must find the node @xmath159 in its copy in time @xmath157 , which requires advice of size @xmath163 . the second part of our lower bound on the size of advice , i.e. , the lower bound @xmath52 , will be proved on the class of oriented rings .
a ring is _ oriented _ if every edge has port labels 0 and 1 at the two end - points .
such a port labeling induces orientation of the ring : at each node , we will say that taking port 0 is going clockwise and taking port 1 is going counterclockwise .
we assume that agents operate in an oriented ring of size @xmath0 . in order to make the lower bound as strong as possible
, we prove that it holds even for simultaneous start of the agents .
[ lb2 ] let @xmath152 be positive integers .
consider any algorithm @xmath54 that solves rendezvous for agents with labels from the set @xmath164 .
there exist @xmath153 and @xmath154 such that if @xmath54 uses time @xmath4 in the @xmath0-node oriented ring whenever the initial distance between the agents is @xmath2 , then the required size of advice is @xmath52 .
assume that @xmath102 is the advice string given to the agents .
consider an agent with label @xmath165 executing algorithm @xmath54 using advice @xmath102 .
the actions of the agent in consecutive rounds until rendezvous are specified by a _
behaviour vector _ @xmath166 .
in particular , @xmath166 is a sequence with terms from @xmath167 that specifies , for each round @xmath168 , whether agent @xmath22 moves clockwise ( denoted by @xmath169 ) , remains idle ( denoted by @xmath94 ) , or moves counter - clockwise ( denoted by @xmath95 ) . note that an agent s behaviour vector is independent of its starting position , since all nodes of the ring look the same to the agent .
this behaviour vector depends exclusively on the label of the agent and on the advice string @xmath102 .
let @xmath170 , @xmath171 and @xmath172 . hence @xmath153 , @xmath154 , @xmath156 divides @xmath0 , and @xmath173 . as the initial positions of the agents , fix any nodes @xmath7 and @xmath159 of the @xmath0-node oriented ring , where @xmath159 is at clockwise distance @xmath2 from @xmath7 . since @xmath173 , agents are at distance @xmath2 in the ring .
partition the nodes of the ring into @xmath174 consecutive blocks @xmath175 of size @xmath156 , starting clockwise from node @xmath7 .
hence the initial positions @xmath7 and @xmath159 of the agents are the clockwise - first nodes of block @xmath176 and block @xmath177 , respectively . since agents start simultaneously , we have the notion of global round numbers counted since their start .
partition all rounds @xmath178 into consecutive _ time segments _ of length @xmath156 .
hence , during any time segment , an agent can be located in at most two ( neighbouring ) blocks .
fix a behaviour vector @xmath166 of an agent with label @xmath22 .
we define its _ meta - behaviour vector _ as a sequence @xmath179 with terms from @xmath167 as follows .
suppose that the agent is in block @xmath180 in the first round of the @xmath168-th segment .
the @xmath168-th term of @xmath179 is @xmath181 , if , in the first round of the @xmath182-th time segment , the agent is in the block @xmath183 , where index addition is modulo @xmath174 . since the initial position of an agent is the clockwise - first node of a block , for a fixed behaviour vector of an agent its meta - behaviour vector
is well defined .
suppose that algorithm @xmath54 takes at most @xmath184 rounds , for some constant @xmath140 .
this corresponds to @xmath9 time segments for some constant @xmath185 .
hence , all meta - behaviour vectors describing the actions of agents before the meeting are of length @xmath9 ( shorter meta - behaviour vectors can be padded by zeroes at the end . ) let @xmath186 be the set of sequences of length @xmath9 with terms from @xmath167 .
sequences from @xmath186 represent possible meta - behaviour vectors of the agents .
the size of @xmath186 is @xmath187 .
since the initial positions of the agents are in blocks that are separated by two other blocks , agents with the same meta - behaviour vectors must be in different blocks in every round , and hence they can never meet . indeed , in the first round of every time segment they must be in blocks separated by two other blocks , and during any time segment , an agent can either stay in the same block or get to an adjacent block .
suppose that the number of bits of advice is at most @xmath188 .
it follows that the set @xmath39 of advice strings is of size at most @xmath189 . for any label @xmath165 ,
let @xmath190 be the function from @xmath39 to @xmath186 , whose value on an advice string @xmath191 is the meta - behaviour vector of the agent with label @xmath22 when given the advice string @xmath102 .
functions @xmath190 are well - defined , as the meta - behaviour vector of an agent whose initial position is the clockwise - first node of a block depends only on its behaviour vector , which in turn depends only on the label of the agent and on the advice string . if the set @xmath192 of all functions from @xmath193 to @xmath194 had fewer elements than @xmath195 , then there would exist two distinct labels @xmath196 and @xmath197 of agents such that , for any advice string @xmath102 , these agents would have an identical meta - behaviour vector . as observed above
, these agents could never meet .
this implies @xmath198 .
hence @xmath199 , which contradicts the fact that @xmath9 is a constant .
this shows that the number of bits of advice must be larger than @xmath200 .
corollary [ cor ] and theorem [ lb2 ] imply : [ lb ] let @xmath152 be positive integers . consider any algorithm @xmath54 that solves rendezvous for agents with labels from the set @xmath164 .
there exist @xmath153 and @xmath154 such that , if @xmath54 takes time @xmath4 in all @xmath0-node graphs whenever the initial distance between the agents is @xmath2 , then the required size of advice is @xmath201 .
theorems [ ub ] and [ lb ] imply the following corollary which is the main result of this paper . the minimum number of bits of advice sufficient to accomplish rendezvous of agents with labels from the set @xmath202 in all @xmath0-node graphs in time @xmath4 , whenever the initial distance between the agents is @xmath2 , is @xmath203 .
we established that @xmath6 is the minimum amount of information ( advice ) that agents must have in order to meet in optimal time @xmath5 , where @xmath2 is the initial distance between them .
it should be noted that the two summands in this optimal size of advice have very different roles .
on one hand , @xmath204 bits of advice are necessary and sufficient to accomplish , in @xmath4 time , the easier task of treasure hunt in @xmath0-node networks , where a single agent must find a target ( treasure ) hidden at an unknown node of the network at distance @xmath2 from its initial position .
this task is equivalent to a special case of rendezvous where one of the agents is inert . on the other hand , for agents whose
labels are drawn from a label space of size @xmath195 , @xmath205 bits of advice are needed to break symmetry quickly enough in order to solve rendezvous in time @xmath4 , and hence , are necessary to meet in optimal time @xmath5 , even in constant - size networks . it should be stressed that the first summand in @xmath18 is usually larger than the second . indeed , only when @xmath195 is very large with respect to @xmath0 and @xmath2 does the second summand dominate .
this means that `` in most cases '' the easier task of solving treasure hunt in optimal time is as demanding , in terms of advice , as the harder task of solving rendezvous in optimal time . in this paper
, we assumed that the advice given to both agents is identical .
how does the result change when each agent can get different advice ?
it is clear that giving only one bit of advice , 0 to one agent and 1 to the other , breaks symmetry between them , e.g. , the algorithm can make the agent that received bit 0 stay inert .
thus , if advice can be different , one bit of advice reduces rendezvous to treasure hunt .
the opposite reduction is straightforward .
hence it follows from our results that @xmath204 bits of advice are necessary and sufficient to accomplish rendezvous in optimal time @xmath5 in @xmath0-node networks , if advice can be different .
this holds regardless of the label space and is , in fact , also true for anonymous ( identical ) agents .
e. bampas , j. czyzowicz , l. gasieniec , d. ilcinkas , a. labourel , almost optimal asynchronous rendezvous in infinite multidimensional grids , proc .
24th international symposium on distributed computing ( disc 2010 ) , 297 - 311 .
p. fraigniaud , d. ilcinkas , a. pelc , communication algorithms with advice , journal of computer and system sciences 76 ( 2010 ) , 222232 .
p. fraigniaud , d. ilcinkas , a. pelc , tree exploration with advice , information and computation 206 ( 2008 ) , 12761287 .
p. fraigniaud , a. korman , e. lebhar , local mst computation with short advice , theory of computing systems 47 ( 2010 ) , 920933 . | two mobile agents ( robots ) , starting from different nodes of an @xmath0-node network at possibly different times , have to meet at the same node .
this problem is known as _
rendezvous_. agents move in synchronous rounds using a deterministic algorithm . in each round
, an agent decides to either remain idle or to move to one of the adjacent nodes .
each agent has a distinct integer label from the set @xmath1 , which it can use in the execution of the algorithm , but it does not know the label of the other agent .
the main efficiency measure of a rendezvous algorithm s performance is its _ time _ , i.e. , the number of rounds from the start of the later agent until the meeting .
if @xmath2 is the distance between the initial positions of the agents , then @xmath3 is an obvious lower bound on the time of rendezvous .
however , if each agent has no initial knowledge other than its label , time @xmath4 is usually impossible to achieve .
we study the minimum amount of information that has to be available _ a priori _ to the agents to achieve rendezvous in optimal time @xmath5 . following the standard paradigm of _ algorithms with advice _ , this information is provided to the agents at the start by an oracle knowing the entire instance of the problem , i.e. , the network , the starting positions of the agents , their wake - up rounds , and both of their labels .
the oracle helps the agents by providing them with the _ same _ binary string called _ advice _ , which can be used by the agents during their navigation .
the length of this string is called the _ size of advice_. our goal is to find the smallest size of advice which enables the agents to meet in time @xmath5 .
we show that this optimal size of advice is @xmath6 .
the upper bound is proved by constructing an advice string of this size , and providing a natural rendezvous algorithm using this advice that works in time @xmath5 for all networks .
the matching lower bound , which is the main contribution of this paper , is proved by exhibiting classes of networks for which it is impossible to achieve rendezvous in time @xmath5 with smaller advice .
* keywords : * rendezvous , advice , deterministic distributed algorithm , mobile agent , time . |
in the early stages of the formation of a star ( see , e.g. , the review paper of * ? ? ?
* ) , a cloud of gas and dust collapses and forms a protostar with a disk surrounding it .
later on , while the accretion from the cloud continues , the star / disk system also begins to eject matter into outflows whose strength varies in time . finally , accretion and outflow cease and over the next few million years the star loses its disk and evolves onto the main sequence .
a major refinement of this paradigm has been an exploration of the parameter space for which multiple objects form from a single collapse ( see e.g. * ? ? ?
* ; * ? ? ?
the dynamical evolution of multiple systems , once they are formed , remains insufficiently well understood however .
observations @xcite have shown that in fact more stars in low mass star forming regions ( e.g. the taurus star forming region ) are found in binary or higher order multiple systems than is the case for older systems like the field stars in the solar neighborhood .
further , nearly all of these stars show evidence for circumstellar disks .
the existence of such multiple stars in forming systems suggests that the evolution of circumstellar material in orbit around or infalling onto those stars will be significantly altered from the evolution in a single star@xmath0disk system .
several observed binary star@xmath0disk systems do in fact show evidence of mutual interactions of various strengths ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in the l1551
irs 5 system for example , in which the observed angular separation on the sky corresponds to a @xmath550 au physical separation , disks around each component of the binary are truncated at about 10 au , a size which is far smaller than is typically observed in single systems or in binaries with larger orbital separations ( e.g. that observed in the hk tau system ; @xcite ) .
in addition to disk truncation , tidal interactions due to the companion star can significantly alter the evolution of the circumstellar disks by generating large scale shocks and other internal heating .
these factors act to accelerate mass transport through the disk , significantly reducing its lifetime unless that mass is replenished from a circumbinary disk on a similar timescale . such heating and transport may negatively influence the formation of planets in binary systems , as described in @xcite .
the situation becomes significantly more complex when the binary system is also surrounded by a circumbinary torus and/or disk .
this additional player may exchange mass and angular momentum with each star and its disk , affecting both the binary orbit and the mass evolution of the circumstellar disks .
for example , their dispersal due to accretion onto the star , enhanced by the binary perturbations , may be slowed by continued replenishment from circumbinary material .
if this reservoir is massive enough , the feeding mechanism may extend the lifetime of the inner disks so that their dissipation timescale is not significantly different from to single stars ( @xcite , see also @xcite for a review ) .
evidence for such replenishment has recently been observed with using high resolution interferometry on alma @xcite .
the total number of observed circumbinary disks is not large . of those ,
most are found in binaries with a small separations , such as gw ori , dq tau , ak sco , v4046 sgr @xcite .
circumbinary disks in wider binaries have been imaged in only a few cases , such as l1551 irs 5 , sr 24 n , uy aur and .
is particularly interesting because of the wealth of observational data available for it , which can be compared to theoretical models .
was first determined to be a multiple system in the early study of @xcite .
later observations of @xcite , using speckle interferometry , confirmed this observation and set the projected separation between the components to be @xmath10 .
the same observations also resolved both the primary and secondary components into a binaries .
imaging by @xcite , refined the distance measurements , setting projected separations of 0.25 and 1.48 between the components of the a and b binaries respectively , and 10.1 between the hierarchical binary pairs . at an assumed 140 pc distance of taurus , these separations correspond to distances of @xmath11 au and @xmath12 au , for the binaries , and a distance of @xmath13 au , for the hierarchical pairs .
the much more massive ` primary ' component of the system , , has received extensive attention since its discovery , and a progressively more detailed picture of the system has emerged .
measurements have been made of the stellar masses , luminosities and spectral types , their age , accretion rates , their relative motions and orbit characteristics , the mass and mass distribution of the circumbinary ring and disk and its rotation curve , and estimates of the circumstellar disk configurations and substructure ( an incomplete list of references includes * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . among the most detailed models of the various components of the system
is that described in ( * ? ? ?
* hereafter gds99 ) , which we summarize here . in their model , the circumbinary material
is composed of two components , consisting of a ring or torus and a disk with a combined mass of @xmath14 , of which some 70% of the total is contained in the torus .
both the torus inner and outer edges are quite sharp , with edges @xmath15 au in width , and are located at distances of 180 and 260 au from the apparent system center , respectively . assuming that the torus is circular , its inclination is determined from the observed projected image to be 37 degrees .
the disk component is fit to a rather steep power law in density with exponent @xmath16 , and a similarly steep power law in temperature with exponent @xmath17 .
the disk s outer edge is not as well resolved , but appears to extend to approximately 800 au , a distance consistent with tidal truncation due to the b components of the combined quadruple system .
kinematic measurements of the circumbinary material constrain the total mass of the a components to be [email protected] .
unresolved emission originating from the circumstellar environment sets a lower limit on the mass of such material to be @xmath19 .
the wealth of observations of is not without inconsistencies , based both in the data themselves , and in attempts to fit theoretical models to it .
for example , while the combined mass of the ` a ' components has been determined ( as noted above ) to be @xmath20 , the masses of the two stars have been estimated from their spectral signatures to be @xmath21 and @xmath22for the primary and secondary , respectively , based on spectroscopic measurements and fits to evolutionary tracks @xcite .
much of the question regarding the system masses may be resolved with the recent discovery of @xcite , that the secondary , gg tau ab , is itself a close binary with separation of @xmath23 au and masses estimated to be 0.38 @xmath24 amd 0.250.35 @xmath24 , respectively .
similarly , the orbital parameters of the stars have been established from fits to their relative motions , and are estimated to define an orbit with semi - major axis @xmath25 au and eccentricity @xmath26 @xcite .
these parameters appear to be inconsistent with the expectations of theoretical models which predict a semi - major axis @xmath27 au @xcite based on the positions of various orbital resonances and the @xmath5180 au inner diameter of the circumbinary disk .
@xcite explored this discrepancy in an analysis of their results , noting that with a slightly increased error bar on the system s observed inclination , an alternate fit to the data yield a semimajor axis of 62 au and an eccentricity of 0.35 .
this orbit is consistent with the dynamical constraints implied by the inner disk edge location and with the observed projected separation of @xmath28 au ( see , e.g. * ? ? ?
our goals in this work are to understand the dynamical influences that the disks and stars in a forming multiple star system have on each other , what their observational consequences are and to make inferences regarding the physical characteristics and behavior of the system based on the outcomes of numerical simulations .
to those ends , we will present a series of two dimensional numerical simulations configured to model systems like , using the numerical code vine .
the system is interesting because of the depth and quality of the observations made of it , which permit it to serve as representative of a larger class of objects in which there are significant dynamical interactions between the various components
. our strategy will be largely qualitative in nature and will focus on comparisons between the observations and simulations modeling a range of plausible physical conditions for the system .
comparisons of features seen in the simulations with those in the observations will then permit us to rule out or confirm various physical properties of the system , by virtue of being consistent ( or not ) with those observations , and to assign their origin to specific physical phenomena active in the system .
the presence and characteristics of the binary will have influences on the circumstellar and circumbinary materials at all scales .
as noted above however , the orbital parameters of the binary are not yet uniquely defined by observations , so the characteristics of the influence are likewise not uniquely determined .
therefore , a logical starting point for our explorations will be to study the spectrum of outcomes that different orbital parameters may produce in the behavior of the gaseous material in the system . with this in mind , our strategy will be to run a suite of simulations in which each member consists of the same circumbinary initial condition and physical model , but stars with different orbit parameters .
each member of the suite will then define a unique numerical experiment to display the consequences that a particular set of binary orbit parameters have for the system behavior , and which characteristics are common to all members of the suite . given the simulation results ,
we can make assessments in terms of comparisons of the appearance of the various gaseous components to the observations .
in particular , we will address three broad categories of questions appropriate for the evolution at different spatial scales .
first , what are the influences of the binary on the large scale structure of the circumbinary material and what structures develop from independently of the binary ?
second , what are the influences of the binary on the circumstellar material and the material ` in transition ' between the circumbinary and circumstellar regions ?
third , what are the likely orbit parameters of the binary ?
we begin in section [ sec : binphys ] with specifications of the initial conditions , the physical models and the numerical methods used to simulate their evolution .
then , in section [ sec : evo - circumbin ] , we introduce the suite of models that form the basis of our study and describe the evolution of each of the systems at large scales . in section [ sec : evo - circumstar ] , we revisit many of these same simulations , but turn our focus towards the smaller scale characteristics of the circumstellar environment .
we will find that many of the results described in these two sections are relevant to more than one of the questions just noted .
we therefore limit our initial discussion to the various characteristics of the simulations themselves , with limited reference to particular observations .
then , in section [ sec : compare ] , we compare our results to the observed system .
there , we focus first on assessing their consistency with the observations in order to rule out less likely configurations , and then on assessing the results for their physical significance , i.e. , what the models may reveal about the physical state and the evolution of the system . finally , in [ sec :
discussion ] , we summarize our results , comparing them to previous work and discussing a number of questions which may be of interest to address in the future . as a guide to the reader , we briefly introduce the various numerical experiments and the goals we set for them .
we introduce the main suite of simulations and begin our characterization of the evolution at large spatial scales in sections [ sec : eccentric ] and [ sec : circular ] . a similar introduction to many of the same simulations appears in section [ sec : circstar - conf ] , where we concentrate on the smaller scale , circumstellar evolution .
these sections will focus on exploring the morphology of large and small scale structures developing under the tidal force of the central stellar system . at large scales ,
the presence of a binary affects the evolution of the torus by continuously stirring the material , generating morphological features in response , and by heating it .
also , in the specific example of the torus , where the observed temperature and mass imply that it is susceptible to gravitational instabilities , we expect that features due to such instabilities to develop as well .
as noted above , is known to harbor assymmetries and other features within its torus . given these features , our first analyses will be qualitative : are any similar features identifiable in our simulations ?
we address this question in section [ sec : torus - features ] . after finding qualitative correspondence between features in the observed and simulated systems
, we turn to investigations to couple the physical processes that are active in the system with the features they are responsible for producing .
the most striking feature of the circumbinary material is that it is composed of two very distinct components : a torus and a disk , with a sharp boundary distinguishing them from each other . in section [ sec :
transport ] , we test the possibility that these two components actually have a common origin as a single primordial torus , out of which the disk forms as an ` excretion ' feature generated by the outward mass transport driven by the propagation of spiral arms .
we run simulations initially configured to include only a torus , in order to compare the behavior of such systems to the torus@xmath0disk systems already studied , focusing specifically on the relative efficiency of outward mass transport and the resulting circumbinary mass distribution .
we continue with a study to determine the relative importance of various sources of heating in the torus .
this question is important because of its bearing on the susceptibility of the torus to fragment . depending on its characteristics ,
if indeed any fragmenation occurs , such fragmentation may lead to the formation of additional planetary , brown dwarf or stellar components of the system . in section [ sec : heatoff ] , we describe the behavior of two simulations in which we suppress , respectively , the radiative heating from the stars and the internal heating due to the dissipation of shocks generated both by the tidal action of the binary and self gravitating instabilities .
we will expect changes in the behavior of the evolution if in fact either of these heating sources are important .
the physical significance of our results depends critically on having both an accurate inventory of the important physical processes and an accurate numerical realization of the evolution .
when fragmentation driven by gravitational instabilities plays a role , accurate numerical simulations require that certain minimum resolution conditions be observed over the lifetime of the simulation .
we assess the physical and numerical significance of the fragmentation behavior seen in our simulations in [ sec : fragmentation ] .
other properties of our simulations may also be affected by numerical issues , with one notable case being that they provide only very coarse resolution of the circumstellar disks .
an important consequence may be that numerical dissipation there is unphysically high , leading to unphysically high accretion rates onto the stars .
as a component of our investigation of the accretion in section [ sec : masstrans ] , we describe simulations run with two very different particle counts , thereby resolving the circumstellar accretion disks to different extents , in order to test whether or not this numerical issue affects our results .
our suite of simulations cover a parameter space within which we expect the actual system to reside , though its exact state is currently uncertain . in order to set more stringent constraints , in section [ sec : quantitative ]
we will make a number of quantitative comparisons between our simulations and the observations .
in particular , we fit ellipses to the torus to determine its size and shape over time and thereby infer other properties of the system , such as the binary orbit parameters and the system s inclination .
other system constraints can be derived by considering the influence that the binary orbit has on the characteristics of both the circumstellar disks and material in the gap between the circumstellar disks and the torus . in section
[ sec : masstrans ] , we calculate the amount of mass transferred from the circumbinary torus onto the circumstellar disks in our simulations , to test how much this mass flux depends on the orbit parameters of the binary .
we also evaluate the rates at which mass accretes from the circumstellar disks onto the stars in different system configurations .
the net rate of accretion into and out of the circumstellar disks determines their lifetimes .
given long enough lifetimes , planet formation may be enhanced .
we discuss this possibility in section [ sec : plan - form ] .
the initial conditions for the simulations performed in this work are based on the observed configuration of the system , as discussed above , except that we have not incorporated the newly discovered binary nature of b into our models . in order to investigate various aspects of the evolution , we vary the initial conditions over the plausible parameter space outlined by both observational data and theoretical analyses , exploring uncertainties in that data .
the details of the physical models and the methods used to set up the initial configurations are similar to those used in ( * ? ? ?
* ; * ? ? ?
* hereafter and , respectively ) and we refer the reader to those works for a more complete exposition . in this section , we shall summarize the models described previously , and describe the details of the set up specific to the present study here .
we use the smoothed particle hydrodynamics ( sph ) code ` vine ' @xcite , to evolve our simulations forward in time .
sph is a particle based , lagrangian method for solving the hydrodynamic equations , and so is very well suited to the complex geometry present in the multi - component systems we study .
particles naturally follow the flow , concentrating resolution specifically in regions where mass itself concentrates , and removing it and the accompanying computational cost , from uninteresting regions where little material exists .
this property permits the evolution of both the circumbinary torus / disk and the circumstellar disks to be modeling at the same time in the same simulation , despite their very large difference in spatial scales .
it also permits detailed studies of the the mass and momentum exchange among the different components of the system , such as between the circumbinary material and the circumstellar disks .
we consider a simplified model of the system in which the ` b ' component of the system is not included .
this approximation will be reasonably accurate unless the orbit of b has a pericenter smaller than @xmath5800 au .
even when such conditions exist , we expect it to affect only the outer portions of the circumbinary disk @xcite , at distances well beyond the regions of interest for our investigations .
we further simplify our simulations by neglecting the recent observation of @xcite that the gg tau ab component is also a multiple .
instead , we model the system as two stars , each harboring a circumstellar disk , and a distribution of circumbinary material , as described by gds99 as a combined torus / disk configuration in orbit around the binary . following this description ,
we configure the initial conditions for each of the three components of the system independently , then combine them together to create a synthesized , complete initial condition for each simulation .
we discuss the set up of each component in turn .
we model the mass distribution of the circumbinary material as a torus@xmath0disk combination .
we set the dimensions of the combination such that its inner and outer edges each define a circle 180 au and 800 au in diameter , respectively .
the outer boundary of the torus is set to 260 au .
the center of mass of the circumbinary material is set to coincide with the system barycenter .
the initial conditions of the circumbinary material are summarized graphically in figure [ fig : circumbin - ic ] .
sph particles are initially laid out on a series of concentric rings centered on the origin and extending from the inner torus / disk edge at a radius of 180 au outwards to a radius of 800 au .
the particle distribution follows a surface density power law : @xmath29^{-{p\over{2}}},\ ] ] where the values of @xmath30 and @xmath31 are adjusted to give the desired mass ratio between the torus and disk masses .
gds99 fit a power law to the volume density in the disk with a proportionality @xmath32 .
assuming a constant ratio of scale height to radius ( @xmath33 ) , this translates to a surface density power law with exponent @xmath34 .
this exponent is broadly consistent with that expected for most circumstellar disk configurations and , for the sake of simplicity , we choose the exponent @xmath35 , used in our previous work for both the circumbinary disk and for the circumstellar disks , discussed below .
the core radius @xmath36 for the power laws is set to @xmath36=1au , a value insignificant for the circumbinary material , but important for the circumstellar disks described below . in the suite of simulations described in this work ,
we explore both torus@xmath0disk configurations and torus - only configurations . for the torus@xmath0disk models
, we define the torus mass such that it contains 70% of the total , consistent with gds99 .
the two factors , @xmath37 and @xmath38 , are linearly decreasing functions near the torus or disk boundary , respectively .
they are each defined by : @xmath39 where @xmath40 plays the role of the defined torus or disk edge , for @xmath37 and @xmath38 , respectively . with this definition
, the outer torus and disk edges are smoothed in the region within a distance @xmath41 inward and outward of their nominally defined radii .
we smooth both the outer edge of the disk and the torus / disk interface .
this smoothing both improves consistency with the observed width of the interface , where data are available , and side - steps the unphysically large pressure gradients that would otherwise perturb the initial condition if the interface were perfectly sharp .
consistent with the observed width as discussed in gds99 , we have chosen the width of the smoothed torus / disk interface to be @xmath42 au .
no observational constraints are readily available for the disk s outer edge , and we choose a wider smoothing parameter of @xmath43 au for the outer edge of the disk . as a practical matter
, the inner edge of the circumbinary material is unsmoothed in order to permit a more evenly distributed initial particle configuration throughout the torus . in consequence ,
some small shear instabilities develop over the course of the first few hundred years of the simulations .
these features are quickly overwhelmed by the larger scale evolution of the system , and so are not significant on any longer time scale .
we specify the initial thermodynamic state of the disk assuming that the specific internal energy , or equivalently , temperature , in the torus and disk follows a single power law profile , given according to : @xmath44^{-{q\over{2}}}.\ ] ] the exponent , is set to @xmath17 , a value chosen to reproduce the fitted value obtained by gds99 and the value @xmath45 is chosen to reproduce , in combination with the equation of state , the observed temperature of @xmath46 k at a distance of 300 au from the origin .
as for the surface density law , the core radius , @xmath36 , is set to @xmath36=1au , which will not significantly alter the profile for the circumbinary material , but will influence the inner edge of the circumstellar disks defined below . at the outer edge of the disk , we modify the temperature profile following the same smoothing procedure as used for the surface density , such that it falls to a minimum of 3 k at the outermost extent .
matter is set up on initially circular orbits around the origin assuming rotational equilibrium .
radial velocities are set to zero . for purposes of generating an initial state only
, we replace the two binary components with a single point mass at the origin .
it s mass , of 1.3 , is the sum of that assumed for the two stars used in the actual simulations .
gravitational and pressure forces are balanced by centrifugal forces by setting @xmath47 where @xmath48 is the gravitational potential of the disk and the other symbols have their usual meanings .
the magnitudes of the pressure and gravitational forces are small compared to the stellar term , therefore the disk is nearly keplerian in character .
the circumbinary material , at a mass of one tenth the combined stellar masses , contains sufficient mass to be marginally self - gravitating . the extent to which this is true is quantified by the well known , toomre @xmath49 parameter defined as : @xmath50 where @xmath51 is the local epicyclic frequency and @xmath52 is the sound speed .
together , the conditions specified above combine to determine the radial profile for @xmath49 shown in figure [ fig : circumbin - ic ] . within the torus values
range from @xmath53 , and we may therefore expect significant signatures of self gravitating flows . further out , in the disk , values increase to @xmath54 , and we expect this region will be largely insensitive to self gravitating disturbances
. note that the @xmath49 profile shown includes small amplitude ` ringing ' oscillations , particularly in the disk , which are artifacts of the numerical derivatives ( of @xmath55 with respect to radius ) that are used to determine the epicyclic frequency .
they are of little significance and may be disregarded .
the two stellar components are each independently defined as point masses , and each is orbited by a circumstellar disk .
the stars have masses of @xmath56 or @xmath57 for the primary and secondary , respectively .
these values are consistent with the total system mass determined from the rotation curve of the circumbinary material ( gds99 ) and from orbital motions @xcite .
it lies slightly below the ` best ' values derived from their luminosities and fits to evolutionary tracks ( see , e.g. , * ? ? ?
* where values of 0.78 and 0.68 are fit for the primary and secondary , respectively ) , but remain within the error bars imposed by those models . as discussed in section [ sec : energy - gen ] , below , we incorporate a simple model of the radiative heating on the disks , supplied by the stars .
following @xcite , we specify their luminosities to be @xmath58 and @xmath59 , for primary and secondary , respectively .
gravitational forces due to the stars are softened , with a softening radius of @xmath60 au and , as the systems evolve , mass may be accreted onto the star if its trajectory takes it inside this radius . the mass and momenta of accreted particles
are added to that of the star .
we assume that the mass of the disk around each star is @xmath61 , consistent with the observed lower limit on the mass of @xmath62 imposed by millimeter observations of the circumstellar emission ( gds99 ) .
this assumption is also consistent with the masses inferred from the more recent observations of @xcite , of @xmath63 . as for the circumbinary material ,
we assume the mass distribution follows a surface density power law using equation [ eq : bindenslaw ] , with an exponent of @xmath35 . for the circumstellar disks , we omit the contribution in equation [ eq : bindenslaw ] due to the torus .
we set the spatial dimensions of the disks to be approximately consistent with the maximum radius dynamically permitted by the orbital parameters we assume for a given simulation .
specifically , for simulations with @xmath1 au , we assume the initial outer radius of each disk to be 10 au , and for simulations with @xmath2 au , we set the outer radius to be 4 au .
in each case , we smooth the outer edge , over a range @xmath64 au of the nominal edge location , following the same prescription used above for the circumbinary material
. no observations of sufficient detail exist , which may be used to determine an initial thermal state for the circumstellar disks with any precision .
we are therefore free to choose conditions that provide a simple and convenient starting point , from which the disks may evolve as the various components of the system interact .
therefore , as for the circumbinary disk , we set the initial specific internal energy distribution to be a power law , identical in form to equation [ eq : circumbin - temp ] . here
again , we choose an exponent of @xmath17 for the power law , and a radial scale of @xmath65 au .
we set the initial scale of the power law , @xmath45 , to be such that the disks are assumed to have initial minimum toomre stability = 1.5 . in practice , and as one of us found in previous work on binary systems @xcite , we will find that the initial thermodynamic properties of the disks matters very little , since they readjust both their mass and temperature distributions within a few orbits around their primary , as the system evolves . similarly ,
the mass distribution also changes in response to the evolution , though on a slower ( @xmath66 yr ) timescale .
once specified as isolated systems , the two star / disk systems defined in the last section are combined to form a binary system , assuming a keplerian orbit around the system barycenter , defined to be at the origin . for systems with an initial eccentricity ,
the initial positions of the stars are such that the system is at apoapse , so that the interactions between the components are at their weakest levels .
we then embed this combined system into the circumbinary environment defined in section [ sec : circumbin ] , replacing the single point mass originally used to as a place - holder to configure the circumbinary system in isolation .
once configured , the combined system , including the orbital elements of the binary , is free to evolve in response to gravitational forces between all of the components .
the thermodynamic properties of the system are very similar to those presented in , and interested readers should consult that work for a detailed description . briefly summarized , we assume that the disk matter can be approximated by an ideal gas and that radiative cooling from the disk surfaces is the only thermal energy sink .
thermal energy is generated by internal sources ( shocks and viscous turbulence ) , implemented by means of an artificial viscosity term in the numerical method .
in addition , for this work , we also employ a radiative heating model , with the two stars as the sources of radiant heating , as described below .
the hydrodynamic equations are solved assuming a vertically integrated gas pressure and a single component , ideal gas equation of state given by : @xmath67 where @xmath68 is the ratio of specific heats , @xmath69 is the vertically integrated pressure and @xmath70 is the specific internal energy of the gas .
the value of @xmath68 assumed for these calculations is the same as that in , of @xmath71 , for the hydrodynamical evolution and @xmath72 , for the three dimensional calculation of the disk vertical structure .
as in , thermal energy in the disks is permitted to escape from the system using a simple model of the disk s vertical structure to determine a ` photosphere temperature ' based on local conditions at each time step .
the temperature is then used as an input into a blackbody cooling rate .
the blackbody temperature is derived for each particle and at each time , using the surface density , specific internal energy and a distance as input parameters for a calculation of the temperature and density structure as a function of altitude , @xmath73 , above the midplane .
this calculation assumes the disk is instantaneously vertically adiabatic , plane parallel and self gravitating . from the newly derived @xmath74 structure ,
we derive the photosphere temperature by integrating the optical depth , @xmath75 , downwards from @xmath76 to the altitude at which @xmath77 in our previous work , the opacities were derived from a combination of the tabulations of @xcite and @xcite . for this work
, we have updated our model to use the rosseland mean opacities from the work of @xcite . in optically thin regions ,
for which @xmath78 at the midplane , we assume the photosphere temperature is that of the midplane . as in , we multiply the opacity by a factor @xmath79 in regions where the midplane temperature is above that of the grain destruction temperature , in order to crudely model the effect of grain destruction and reformation on the opacity .
the present configuration is considerably more complex than that described originally in , where the disk s vertical structure was defined assuming only the gravitational influence of a single star contributed to the vertical structure . in this work , of course ,
two stars contribute , each with a different relative magnitude depending on the distance of a given location from those stars . to simplify the specification of the vertical structure ,
we define three distinct regions , in which one or another limiting case holds and only one point mass can be assumed to contribute the only significant external contribution to the disk s vertical structure . for each of these limiting cases ,
we create separate tabulations of the vertical structure .
the regions are , respectively , locations where we may assume that the gravitational forces of either the primary or secondary alone dominate the vertical structure , and locations where the combined effect of both stars contribute to the vertical structure .
we assume arbitrarily that all material at distances greater than 150 au from the system center of mass falls into the latter case . for purposes of defining the disk s vertical structure in this region ,
the mass used to define the gravitational influence of the two stars is set to the sum of the masses of both stars , and its location is set artificially to the system center of mass . for material inside this 150 au limit , we determine the distance to each star and define the material as ` belonging ' to the closer of the two .
given the blackbody temperature , each sph particle loses thermal energy at a rate defined by @xmath80 where @xmath81 is the stefan - boltzmann constant , @xmath82 and @xmath83 are the specific internal energy and surface density of particle @xmath84 and @xmath85 is its photosphere temperature .
the surface density @xmath83 serves the dual purposes of first , relating the blackbody flux ( quantified as an energy flux ` per unit area ' ) to the specific internal energy ( quantified as an energy ` per unit mass ' ) used in our simulations and , second , of disambiguating the ` surface area ' of a single sph particle , which would otherwise suffer from confusion because of the overlap of sph particles with each other .
the factor of two accounts for the two surfaces of the disk . in regions where @xmath86 we multiply the derived flux by the optical depth , @xmath75 , in order to more correctly model the cooling of optically thin regions . thermal energy in the disk is generated from bulk mechanical energy via both reversible and irreversible processes during the course of the system s evolution .
thermal energy is generated reversibly as spiral arms and similar dynamical structures become compressed and @xmath87 work is done on the gas .
thermal energy is generated irreversibly in viscous processes and shocks .
both of the latter processes are modeled through the use of an artificial viscosity , as is common in many implementations of hydrodynamic codes .
two time and space dependent artificial viscosity terms are included . as discussed in ,
the first is proportional to @xmath88 ( the ` bulk viscosity ' ) and corresponds to unresolved viscous turbulence in the system .
the second is proportional to @xmath89 ( the ` von neumann - richtmyer ' viscosity ) and corresponds to shock dissipation .
because of the time and space dependence , an exact equivalance between the artificial viscosity and @xcite ` @xmath90 ' models does not exist .
however , a correspondence can be drawn between the bulk viscous term in the code and the local value of the @xcite @xmath90 @xcite , which we can use to calibrate the magnitude of internal disspation in our simulations in comparison to the model . in 2d
, the correspondence can be expressed for particle @xmath91 as @xmath92 where @xmath93 is the coefficient to reduce numerical shear viscosity for sph simulations defined by @xcite , @xmath94 is the artificial bulk viscosity coefficient , @xmath95 is the orbit frequency , @xmath96 the sound speed and @xmath97 the smoothing length of particle @xmath91 . averaged over many particles in some suitable radial range , equation [ eq : balsara ] provides a useful estimate of the magnitude of the artificial viscosity . the calculations of showed that the artificial viscosity produced a dissipation comparable in magnitude to an @xmath90 equal to a few @xmath98 , but that the dissipation was resolution dependent so that the physical dissipation of the real system was probably lower than this value .
the circumbinary torus / disk configuration used in this work is both different than the circumstellar disks in and is resolved with far more particles , so we do not expect dissipation at comparable levels to that work .
indeed , using the relationship quoted in @xcite , we find energy dissipation rates in typical regions of the torus to be an order of magnitude or more smaller , comparable to models with @xmath99 or less . in the case of the circumstellar disks , which are configured similarly to @xcite
, we estimate the energy generation rates to correspond to those in @xmath90 models where @xmath100 .
it is important to note in these estimates that while the magnitude of the dissipation in our simulations does correspond to that in a @xcite model with a given parameter , the physics as realized in the simulation does not corresponds directly to the physics underlying such models .
specifically , the turbulent eddies out of which the dissipation is assumed to arise are absent in our simulations , because we model only two dimensions .
rather , the presence of such dissipation is effectively an assumption that such turbulent eddies are present and are the only source of internal energy generation in the fluid .
as discussed in , this includes the assumption that the flow is reasonably smooth , and that strong shocks are not present .
we shall find this assumption is violated in the work presented here , as it was in our earlier work on binary systems @xcite .
strong interactions between the binary components produce large amplitude spiral structures in the circumstellar disks , including strong shocks that inevitably accompany such structures .
the circumbinary material also produces large amplitude spiral structures and shocks generated by the stirring motion of the binary itself . in both cases , additional heating of the disk matter occurs . in comparison to the bulk viscosity , these heating events generate internal energy at several times the smooth ` background ' rate in the torus , though only in a much smaller fraction of the total volume , where the spiral structures themselves are present . in early simulations ,
performed only with internal heating sources , we found that temperatures in the circumbinary torus and disk fell to values well below those observed for the system and , in the outer regions of the disk , even to the lower limit of 3 k we have implemented in the code itself .
the rate of thermal energy generation due to internal processes is less than is required to balance the radiative cooling in the circumbinary material .
we conclude that a correct model of the system requires that additional , external heating sources be included in the inventory of physical processes describing the system .
although insufficient by themselves to heat the disk and balance radiative cooling losses , we will show below ( e.g. section [ sec : heatoff ] ) that heating from dynamical processes does still provide a significant source of energy input to the system .
therefore a passive radiative heating model , such as described by @xcite , will also not be an appropriate ingredient to an accurate physical model of the system .
we have therefore developed a simple model for the radiative heating of the disks by the stars , using a prescription which is similar in spirit to the radiative cooling model described above .
specifically , we assume that each star radiates isotropically at a constant luminosity , and that all of the stellar radiation impinging on the two disk surfaces ( i.e. both above and below the disk midplane ) is absorbed .
we estimate that energy absorption rate as follows . by definition ,
a star s luminosity is the energy flux that it emits per unit time , with its value per unit area of any spherical volume centered on that star being @xmath101 .
this flux falls on the two surfaces of the disk , each oriented at an angle to the radially outward directed stellar flux .
the flux absorbed per unit area by the disk s surface will be diluted from the stellar flux due to the mismatch between the normal directions of the stellar flux and the disk surface . in order to quantify the energy absorbed correctly , we must determine the size relationship between a surface element oriented radially and one oriented normal to the disk surface .
following gds99 , we assume that stellar flux impinges on the disk horizontally and that the disk surface s normal vector is tilted at an angle @xmath68 , as measured downwards from a vector pointed towards the disk s north or south ` pole ' .
then , simple geometric arguments provide the relationship we require : the flux per unit area of disk surface will be diluted from the flux per unit area of stellar flux by the dot product of the unit vectors defining their respective orientations . that dot product reduces to the cosine of the angle between them and ,
if we assume that the disk is not significantly flared , it s value will be defined at every location by it s radial distance from the star and the altitude of surface at which the disk material becomes optically thick to stellar radiation : @xmath102 then , the rate of energy absorption of the stellar flux at any given location of the disk s surface will be given by the relation @xmath103 the factor of two accounts for the fact that both the upper and lower disk surfaces will be heated by the stellar flux .
the surface density term plays the same role in this equation that it does in equation [ eq : dudt - radcool ] above .
namely , to recast the energy flux per unit area in terms of the specific internal energy ( i.e. as a change to the energy per unit mass ) , which is the conserved quantity employed in vine .
the remaining unknown in equation [ eq : dudt - radheat ] is the value of @xmath104 , through its presence in the cosine term in equation [ eq : cosgamma ] . in general , the value of @xmath104 will be different than the altitude of the disk s photosphere however , because the latter characterizes an optical thickness to its own internal radiation , rather than to stellar photons .
we therefore utilize a measure of the disk s scale height , derived from the same vertical structure calculation used to determine the cooling rates described in in section [ sec : energy - diss ] . as for the cooling calculations , and to simplify the model , we employ the same three limiting regions as were used in the radiative cooling model . at distances
more than 150 au from the system center of mass , we the calculate vertical structure assuming a single star at the origin whose mass is the sum of the two binary components . at closer distances ,
the scale height is determined assuming that each sph particle belongs to the nearer of the two stars in the binary .
the radiative heating and cooling models used in this work are deliberately simplified from the actual radiative transfer processes present in the real systems . here
we explore the implications that these simplifications make on our simulations , in order to provide insights into the conclusions we may draw from them . as noted above
, we multiply the derived flux by the optical depth , @xmath75 , in regions where @xmath86 , in order to more correctly model the cooling of optically thin regions . the circumstellar disks and the torus region of the circumbinary material
are optically thick to their own radiation throughout the calculation , so that in practice , this modification has little effect . in the circumbinary disk region however , the optical depth falls to near unity , making the disk only marginally optically thick and endangering the assumptions made in deriving the cooling rates there .
because the focus of our work is mainly on the interactions between the stars , their circumstellar material , and the torus , errors in the cooling rates at large distances , and their consequences for the dynamical behavior , will be comparatively small . for other regions , such as the gap between the circumstellar and circumbinary material ,
material densities become even more tenuous , with correspondingly smaller optical depths as well . in these regions ,
the coupling of the material to the radiation becomes increasingly tenous as well , and the radiative heating and cooling treatments used in our work fail to provide accurate treatments of the energy flow . since we expect that the radiative energy fluxes into and out of the material will be substantially diminished by the reduced coupling efficiency , we simply disable both the radiative heating and cooling entirely when such conditions are found in our simulations .
specifically , when the surface density of any given sph particle falls to a value less than .05 gm/@xmath105 , we assume that no radiative heating or cooling occurs for that particle . as for radiative cooling ,
the accuracy of the radiative heating suffers in the outer regions of the circumbinary disk .
because our model assumes that thermal energy is deposited in an entire vertical column of material , rather than only in a surface layer , we overestimate the actual amount of heating there , since in fact most of the column is shielded by material closer to the stars . because we are interested primarily in the dynamical behavior of the torus and circumstellar disks where the effect is smaller , the over estimates at larger radii will be of limited concern for interpreting our results .
the dimensions of the system are typical of a number of similar systems , as noted in section [ sec : ggtau ] , so that our models serve as prototypes for the behavior of that class . of course
, the exact dimensions of each system vary and , with them , so do the time scales important for their evolution . in order to provide a calibration for our readers , we define here several relevant time and distance scales for the initial conditions used in our simulations . together with the masses of the binary components , the semi - major axis assumed for the binary orbit
defines its orbital period .
the initial conditions defined for our simulations include setups with semi - major axes of @xmath2 au and @xmath1 au , which correspond to orbital periods of 158 yr and 426 yr , respectively .
the orbital period of material in the torus provides another interesting timescale .
given its initial inner boundary at 180 au and the initial condition specification above , we can estimate its orbit period to be @xmath106 yr .
these orbit radii , together with the inner torus edge , define dimensionless ratios of @xmath107 for the 32 au semi - major axis , and @xmath108 for the 62 au semi - major axis , where @xmath109 is the torus inner radius .
we defined the initial configuration of the circumstellar disks to have outer radii that are each a large fraction of the orbital separation between the components making up the binary , consistent with our expectation that such a configuration will be nearer to the ` asymptotic ' long term configuration .
as the simulations proceed , they will grow or shrink according to the amount of accretion into the circumstellar environment from the torus and the amount of accretion out of the disks onto the star around which they orbit . in this context ,
another relevant dimension for the system is the limiting distance for stable orbits around the star to exist .
@xcite have calculated such criteria for a variety of system configurations and , for the systems studied here , derive outermost stable radii of @xmath110 and @xmath111 au for the a@xmath11262 au systems with eccentricity of @xmath4 or @xmath3 , respectively , and @xmath113 au and @xmath114 au for the a@xmath11232 au systems with @xmath4 and @xmath3 , respectively .
these radii are slightly larger than the initial sizes of the disks in our study
. the difference does not imply any inconsistency in the model because we anticipate ( and as we shall observe below , correctly so ) that the disks will quickly reconfigure themselves according to the dynamical evolution of the system , and because the present calculations include hydrodynamical effects not addressed in @xcite .
using the initial conditions and physical models described above we have performed a number of simulations modeling systems similar to .
the simulations typically follow the evolution for 6500 yr , corresponding to approximately three orbits of the inner edge of the circumbinary disk .
these times also correspond to @xmath115 binary orbits for the 62 au series , and to @xmath116 orbits for the 32 au series .
complete specification of the initial parameters of the simulations are listed in table [ tab : bin - params ] . in the discussion
below , we will often refer to the two classes of simulations as the ` wide ' ( 62 au ) binary simulations and ` close ' binary ( 32 au ) simulations .
we examine the qualitative nature of our simulations first , then examine in detail the structures which form and their physical origin .
lcccccccc clos3elo & 2289 & 1955 & 421384 & 4 & 0.3 & 32 & 4400 & 2.09 + wide0elo & 2207 & 1892 & 421384 & 10 & 0.0 & 62 & 6500 & 3.09 + wide3elo & 2207 & 1892 & 421384 & 10 & 0.3 & 62 & 6500 & 3.09 + clos0ehi & 9995 & 8592 & 1861769 & 4 & 0.0 & 32 & 7000 & 3.33 + clos3ehi & 9995 & 8592 & 1861769 & 4 & 0.3 & 32 & 7500 & 3.57 + wide0ehi & 10134 & 8418 & 1861769 & 10 & 0.0 & 62 & 6500 & 3.09 + wide3ehi & 10134 & 8418 & 1861769 & 10 & 0.3 & 62 & 6500 & 3.09 + singlehi & - & - & 1861769 & - & - & - & 2100 & 1.00 + noheat3ehi & 10134 & 8418 & 1861769 & 10 & 0.3 & 62 & 2100 & 1.00 + clost3elo & 2289 & 1955 & 421081 & 4 & 0.3 & 32 & 4000 & 1.90 + clost0ehi & 9995 & 8592 & 1867987 & 4 & 0.0 & 32 & 6500 & 3.09 + clost3ehi & 9995 & 8592 & 1867987 & 4 & 0.3 & 32 & 6500 & 3.09 + widet0ehi & 10134 & 8418 & 1867987 & 10 & 0.0 & 62 & 6500 & 3.09 + widet3ehi & 10134 & 8418 & 1867987 & 10 & 0.3 & 62 & 6500 & 3.09 + the analyses of the proper motion data lead to two alternatives for the orbital motion of the stars in , with @xmath2 au and @xmath1 au . in both cases ,
the orbital eccentricity is calculated to be @xmath117 . in this section ,
we examine the results of two simulations , each with eccentricity @xmath4 , but with differing semi - major axis , referenced in table [ tab : bin - params ] as simulation _ wide3ehi _
( @xmath1 au ) and _ clos3ehi _ ( @xmath2 au ) . in figure
[ fig : wide - ecc - manyorb ] we show twelve images of the same simulation , _
wide3ehi _ , at the same orbital phase , spread over 16 orbits of a binary orbiting with semi - major axis @xmath1 au .
the mass distribution in the initial condition , seen in the first frame , is completely axisymmetric and clearly shows the various components in the idealized initial configuration we employ in all of our simulations . as the simulation time increases and the system begins to evolve ( frames 1 - 3 ) , the mass density begins first to exhibit many small scale disturbances at the interface between the torus and disk near @xmath118 au , which penetrate into the low density disk surrounding it .
larger scale disturbances are present in the mass distribution at the torus s inner boundary as well .
we attach little physical significance to these early time features and instead attribute them merely to the growth and relaxation of ` start up transients ' related to our idealized initial condition for the actual complex system .
after the first several orbits of the binary , such small scale instabilities have been largely overwhelmed by the growth of features driven by the strong stirring action of the stars on the circumbinary material .
initially , the torus condenses into a narrowly peaked structure some 10 - 20 au in radial extent , with several higher density lumps approximately equally spaced in azimuth .
later , after 6 - 8 binary orbits and as the results of interactions between the torus and binary begin to accumulate , the density inhomogeneities become more and more widespread . at this time , the lumps have largely dissipated , and the ring itself has broken up into several filamentary spiral arms , each of which contain still finer scale substructure . over time , the original sharp inner edge of the torus becomes progressively more ragged and diffuse as the binary interactions perturb it .
as it evolves , it both moves inwards towards the stars and its shape changes , becoming significantly non - circular and producing spiral streams that extend inwards towards the binary .
as pictured , the inner edge of the torus appears similar to an ellipse whose semi - major axis is approximately perpendicular to the axis joining the two stars .
because each of the images in the mosaic is generated at the same orbital phase , the relative phase of the binary orbit and the elliptical torus edge appears nearly constant .
we will explore the extent to which this is always true in section [ sec : quantitative ] , below . at larger
radii , and as they begin to overwhelm the start - up transients , spiral structures become ubiquitous features throughout the torus at all times in the simulation .
the spiral structures are generated by a combination of self gravitating instabilities and the stirring action of the stars in their orbit .
the finer scale substructures in the torus are largely the remnants of streams of material which propagate through the gap region and return to the torus .
the details of the size and shape of these streams vary from one orbit to the next . for some orbits , long streamers of material
are visible extending inwards towards one or the other star , while in others they are nearly absent
. we will discuss the behavior of these streams over the course of one binary orbit in detail in section [ sec : circstar - conf ] .
of importance for our discussion here however is the fact that comparatively little material is actually accreted into the circumstellar environment compared to the fraction that returns to the torus .
instead , most material falls through the gap , interacting with one or both stars only through a relatively distant encounter during which only its trajectory is changed .
then , as it continues on its new trajectory , material begins to propagate outward again , returning to the torus from which it originated . in figure
[ fig : close - ecc - manyorb ] , we show twelve images of the _ close3ehi _ simulation , each at the same orbital phase , spread over 47 orbits of a binary orbiting with semi - major axis @xmath2 au . as in the prototype wide orbit configuration in the last section ,
the mass distribution in the circumbinary disk first begins to exhibit small scale inhomogeneities which we attribute to start up transients as the simulation begins to evolve away from its idealized initial configuration .
after @xmath119 yr of evolution ( frame 3 ) , larger scale structures develop in the torus , which extend througout its radial extent . unlike the wide orbit case however , the structures are more filamentary in character , with no features comparable to the radially narrow , peaked structure noted in panel of figure [ fig : wide - ecc - manyorb ] developing , even at later times . instead ,
smaller amplitude spiral structures develop which , after @xmath120 yr ( panel 7 - 9 ) , begin to organize into larger scale condensations . these condensations continue to grow and coalesce over the next @xmath121 yr and , in the final configuration of the simulation , only one remains .
we expect that if the simulation were to continue , this fragment would continue to grow , perhaps to a size at which it would begin to disrupt the disk itself
. we will discuss what interpretation should be given to such fragments in section [ sec : fragmentation ] . over the lifetime of the simulation
, the inner portion of torus material migrates inwards towards the stars , with its inner edge eventually reaching distances from the system center of mass of only about half of its original distance . unlike the wide orbit case , large streamers of material
do not appear in the gap region and , to the extent that they are present at all , they do not change their character from one orbit to the next .
this behavior is consistent with the increased distance from the stars to the inner edge of the torus and the correspondingly smaller perturbing influence that they exert on it . even at late times ,
when the streamers appear most distinct , they do not extend inwards to the positions of the stars or their disks as much as is seen in the wide orbit case above .
instead , they appear as extensions of the spiral structures further out , and we associate their presence to the fact that precursors of the fragments noted above have begun to influence the torus morphology . the behavior of the cirumbinary environment surrounding a system with an eccentric orbit will of course differ from that of a circular orbit , due to the presence or absence of time variations in the strength of the gravitational forces exerted by the stars . we have investigated these differences with a set of simulations whose conditions are identical to those in sections [ sec : wide - morph ] and [ sec : close - morph ] , except that we assume that the orbit of the binary is defined by an eccentricity of @xmath3 .
we show the system configurations at the end of each simulation in figure [ fig : e0-end ] , for the same two semi - major axes ( @xmath1 au and @xmath2 au ) described in sections [ sec : wide - morph ] and [ sec : close - morph ] above . in both the wide and close cases ,
the inner edges of the tori have moved inwards over the time scale of the simulation to radii of 120 - 140 au from the system barycenter or less .
both models show evidence of streams of material present in the gap region . although less pronounced than in the @xmath4 cases , they extend further inwards towards the stars , likely because they are not disrupted to as great an extent by close encounters with the stars .
relative to each other the wide binary configuration shows more massive streams than the close binary configuration , as was the case in the @xmath4 models , in addition , the wide configuration also exhibits a small band of material joining the two components , which appear to originate from material pulled from the outermost extents of the circumstellar disks .
closer to the stars , the circumstellar disks exhibit only minimal spiral structure , with the highest densities found in the outer extents of the disks rather than close to the stars .
the spiral structures that develop in each torus are much more diffuse in character than in the corresponding @xmath4 simulations , and exhibit much less internal structure . in consequence ,
the outer edges of each torus are both more circular and less sharply defined than in the eccentric binary cases , where well defined edge features form as structures propagate through the tori to their outer limits . in spite of the less pronounced spiral structure , the circumbinary tori in both simulations
have produced large fragments , some 10 - 20 au in diameter . in each case , the fragments have grown to such an extent that they have begun to generate large amplitude spiral wakes in the torus , independently of the influence of the binary s stirring action .
we will discuss the physical significance of these fragments in section [ sec : fragmentation ] .
the observed , two - component torus / disk configuration of the circumbinary material provides a challenge for simple models of star formation to reproduce .
while one can easily imagine mechanisms by which the actions of the binary conspire to keep a sharply bounded inner ` gap ' region clear , similar arguments are more difficult to make at the outer edge of the torus . how did a sharply bounded torus form with a surrounding low density disk , and how is the structure maintained in the system over time ?
an important result of the sections above was to show that the stirring action of the binary on the surrounding material is very strong , with large scale spiral structures being continually generated .
such stirring serves as a foundation for the model described in ( e.g. * ? ? ?
* ) for example , and one consequence he describes is the formation of an ` excretion ' disk as a portion of the circumbinary material gains angular momentum from the binary and moves outwards in response .
is this mechanism the origin of the low density disk in the system ?
in this section , we explore this question first by simulating the evolution of a family of systems in which the initial circumbinary configuration includes _ only _ a torus with the same mass as the combined torus / disk configuration as above .
then , we study behavior of the torques generated by the the stars and torus on each other .
system configuration of four simulations which assume all material in the circumbinary environment is initially located in a torus .
each panel shows the configuration at the binary s apoapse for the time closest to @xmath122 yr of evolution , corresponding to @xmath123 torus orbit periods , such that the systems have evolved for a similar amount of time .
the top panels show the two simulations with the assumed binary eccentricity set to @xmath4 ( _ clost3ehi _ and _ widet3ehi _ ) , while the lower panels show the two configurations with assumed binary eccentricity set to @xmath124 ( _ clost0ehi _ and _ widet0ehi _ ) .
the right and left panels show the @xmath2 au and @xmath1 au semi - major axes , respectively . ]
figure [ fig : torus - morphs ] shows the counterparts of each of the four torus / disk configurations described above , each after evolving for @xmath122 yr . as in the torus / disk models discussed above , torus material migrates inwards , partially filling the gap region between the cirumstellar environment and the initial inner boundary of the torus .
the gap region in the close binaries is visibly smaller in size than in the wide , and more in the @xmath3 cases than in the eccentric orbit cases . in every case , and similarly to the torus / disk models above , well defined spiral structures form in the tori and , in several examples , they fragment into clumps . in all cases , the spiral structures are driven by the stirring action of the stars and propagate outwards , carrying material as they do so .
also as in the torus / disk models , the edges of the spiral arms are sharply defined , particularly in the two wide binary examples where the stars interact more strongly with the torus .
although sharply defined during their passage through the torus , they become less and less distinct at greater distances from the system center , eventually reaching a lower and smoother background density at distances beyond the original outer edge of the torus .
the presence of material at large separations is evidence for some level of outwards mass transport in the system .
figure [ fig : mass - outside ] shows the amount of mass found there as a function of time for each of the simulations .
as defined in the figure , the ` outer disk mass ' is defined as the amount of mass found at radii larger than an arbitrarily defined critical distance from the system barycenter .
guided by the results we will describe in section [ sec : quantitative ] below , we set a value of 350 au for this radius , which we choose as a compromise between capturing the true mass of the disk component over time , and masking out transient changes to it that are due to portions of the torus flexing such that parts of it extend beyond our critical radius . in every case , the mass increases over time , showing that the process itself is robust across a wide variety of configurations .
the magnitude and overall pattern of the mass transport varies however , both from simulation to simulation and from time to time within the same simulation .
after @xmath125 torus orbits , corresponding to 6500 yr run time of the simulations , the total mass transport corresponds to about 5% of the total initial mass of the circumbinary material in the torus@xmath0disk systems , and is @xmath126 times higher , at @xmath127% , in the torus only systems . in the torus@xmath0disk runs , the outer disk mass increases at a near constant rate , with small oscillations visible , superimposed on the long term trend .
in contrast , in the torus only runs , the mass transport is almost entirely limited to the first @xmath128 yr of the evolution , with the outer disk mass becoming essentially constant thereafter or even decreasing slightly . only in the last @xmath129
yr of the runs , does the outer disk mass begin to increase again .
the duration of our simulations are long in the context of the circumstellar environments and the number of binary orbit periods we simulate .
unfortunately , they are short in comparison to the orbital timescales of circumbinary material , where a time of 6500 yr corresponds to only @xmath126 orbits . therefore , due to the high cost of running the simulations , we have not followed the evolution for the long times required to determine whether or not the outer disk masses return to their high rate of growth , or if the decrease is simply a signature of the overall flexing action of the circumbinary structures as they respond to the binary s stirring action .
we are therefore unable to conclude that the outward material migration will continue at the same rate or decrease , or even if time variations within a given simulation continue indefinitely .
nevertheless , the fact that the outer disk mass increases in all of our models , is strong evidence that the excretion process will continue in the initially torus - only systems , such that they evolve to become robust torus@xmath0disk systems . given the rate of outward material movement , and the fact that the disk in the system is observed ( gds99 ) to contain a mass of @xmath130 , the entire disk could be generated from such transport activity within a few @xmath131 yr .
we therefore conclude that the disk material surrounding the torus most likely originates as torus material which was thrown outwards by the stirring action of the binary .
gravitational torques between the stars and circumbinary material play an important role in their mutual evolution . here
, we turn to a study of these torques , in order to understand their importance in forming multiple star systems like .
torques between the stars and the torus will affect both and , if large enough , cause measurable changes even over the comparatively short duration of our simulations .
figure [ fig : sumtorq ] shows the net torque exerted by the stars on the circumbinary material
. in all cases , the torque magnitudes vary both in amplitude and sign on time scales comparable to the orbit periods of the binaries . also ,
while oscillations are correlated with the binary orbit , consecutive peaks and troughs are neither evenly spaced in time nor are their amplitudes the same .
the amplitude of the variations are larger in the torus - only systems than in the torus@xmath0disk systems , consistent with the increased mass concentration in the former case . with few exceptions ,
the magnitude of the peaks in the wide orbit examples exceed the magnitude of adjacent troughs , such that the average over many orbits is positive .
as seen , e.g. , in section [ sec : wide - morph ] above , the torques from the binary drive strong spiral waves into the torus in wide configurations , but are much less effective in the close configurations shown later .
their absence leads to the question of what process remains to drive the excretion flow discussed above ?
we note that , while stellar torques are weak in such systems , torques due to self gravity remain , and we conclude that they are important contributors to the excretion flow . with a change in sign
, the total torque on the circumbinary material is the same as that on the binary . to
what extent do torques of these magnitudes alter its orbit over time ?
@xcite makes estimates using a diffusion approximation based analysis where the torques are mediated by viscosity . here ,
no such model is required , since we have a direct quantification of the torques from the simulations themselves .
we therefore estimate the time scales using time averaged torques determined directly from the curves shown in figure [ fig : sumtorq ] . while the torques in the wide orbit runs exhibit positive maxima of @xmath132 dyn cm and negative minima of @xmath133 dyn cm , their averages over time take on values near @xmath134 dyn cm .
the circumbinary material thus gains angular momentum at the expense of the binary , consistent with the analysis in @xcite .
the close orbit torques are each more than an order of magnitude smaller and are also more nearly symmetric over their positive and negative extrema , with time averages consistent with zero net torque . neglecting factors near unity ,
the net angular momentum of the stars around the system barycenter is given by @xcite as @xmath135 , with @xmath136 being the combined mass of the two stars .
the time scale for changes in the semi - major axis is @xmath137 . in terms of the parameters of the wide and close orbit systems
, the angular momentum translates to numerical values of @xmath138 or @xmath139 gm @xmath105/s , respectively .
thus , combined with the time averaged torques derived from figure [ fig : sumtorq ] , the timescale for significant changes to the semi - major axis becomes @xmath140 myr for the wide systems and @xmath141 myr for the close systems .
the long time scales in the close orbit models , in pringle s interpretation , correspond to a system in an initial phase where the torus has yet to ` find out ' about the presence of the binary .
in contrast , the shorter time scales in the wide orbit simulations correspond to the period in which the torus is gradually expelled by the torques applied by the binary .
they are also comparable to the pre main sequence lifetimes of stellar systems @xcite , indicating that the final system configuration has yet to be established .
before we may conclude that the action of the orbiting stars on the circumbinary material is significant for the overall evolution of the system , we must compare the results of our simulations to systems with analogous configurations where the binary is not present .
similarly , in order to conclude that the radiative heating from the stars is important , we must compare to simulations in which that heating is omitted . to those ends , we have configured and run a simulation in which the radiative heating term has been neglected , and another in which we have replaced the two components of the binary with a single star of the same mass , located at the origin . in order to provide a similar level of heating to the system , this star radiates with a luminosity equal to the sum of the luminosities of the binary components .
the surface density morphology of simulations _ singlehi _ ( top ) and _ noheat3ehi _ ( bottom ) , each after @xmath142 yr ( approximately one orbit of the torus ) of evolution.,title="fig : " ] the surface density morphology of simulations _ singlehi _ ( top ) and _ noheat3ehi _ ( bottom ) , each after @xmath142 yr ( approximately one orbit of the torus ) of evolution.,title="fig : " ] figure [ fig : heatoff - morph ] shows the results of these simulations after they have evolved for @xmath142 yr , corresponding to one orbit of the inner edge of the torus .
the image shows that in each case , approximately 38 - 39 separate clumps have formed from the initially smooth tori during this comparatively short timescale .
although not shown in the images themselves , these clumps initially appeared after only @xmath13 yr of evolution and continued to develop further as the simulations proceeded .
we monitored the status of the simulations for adherence to the numerical resolution requirements discussed in @xcite and found that these clumps are indeed sufficiently resolved according to the criteria described there .
evidence for self gravitating instabilities is present in the torus region of all of our other simulations , including several which exhibit one or more clearly developed fragments .
however , even the simulations in which fragmentation was observed do not produce fragments so quickly or in such large numbers . in this regard ,
the configurations that develop here are unique in our suite of simulations , and follow a clearly different evolutionary path than any of the simulations that include both the binary components and radiative heating . since such ubiquitous fragmentation requires cold torus material , and the action of the stellar irradiation and the tidal interactions of the binary will be to increase the internal energy of the material , we conclude that both the stellar irradiation and the stirring action of the binary are critical contributors to the overall thermodynamic state of the circumbinary environment of . in the sections above ,
we have discussed the overall morphology of the circumbinary material as it evolves using qualitative metrics of its behavior .
we now turn to more quantitative metrics to describe the evolution . in particular ,
we have noted above that the torus appears to flex both radially and azimuthally over time , with various components moving both inwards and outwards as spiral waves pass through them .
here , we quantify the changes in the shape of the torus , by fitting its shape to a set of three ellipses , chosen as approximations of its inner and outer edges and its midpoint . in the following sections , we show example fits at two snapshots in time in order to illustrate the fidelity of the fits to the underlying data , and then as functions of time , in order to show their variations as the system evolves . for each ellipse , we fit for three important quantities : its semi - major axis , its eccentricity and its orientation . we discuss the fits for the two simulations _
wide3ehi _ and _ clos0ehi_. we choose the first because both the binary s wide orbit ( semi - major axis ) and its eccentricity carry it closer to the torus than any other .
we choose the second for exactly the opposite reason : its tighter orbit and zero eccentricity mean that it stays most distant from the torus over the course of its orbit .
we expect that these system configurations will therefore bracket the parameters for which the effect of the binary on the torus is the largest and smallest of all the configurations in our study .
sph particles do not easily lend themselves to standard procedures for generating fits to the ellipses we use to model the system shape . therefore , in order to generate fits , we first map the surface density of the disk material onto an 1800@xmath1431800 cartesian grid , on which the fitting can be accomplished straightforwardly .
the grid has spatial dimensions of 900@xmath143900 au and is centered on the origin , such that each grid cell is 0.5 au in linear extent .
the mapping implements the same interpolation procedure as is used in the sph algorithm to determine the surface densities for each particle in the simulation itself .
unfortunately in the present context however , the resolution afforded by our simulations and the mapping retains very small scale features in the flow , including accretion streams and other spiral structures only peripherally associated with the torus .
this level of detail frequently causes the fitting procedure outlined below to produce poorer fits of the large scale torus morphology in favor of matching these peripheral structures . in order to more accurately capture only the large scale features of the torus , we therefore convolve the density mapping with a gaussian psf with a characteristic length scale ( i.e. , its half - width , half maximum ) of 10 au . with this modification of the density mapping ,
we compute an azimuth averaged surface density and extract a maximum over all radii .
this maximum is used , in turn , as a critical value to extract the set of all grid points lying on three distinct isodensity contours .
we define one such contour by the condition that the local surface density matches the maximum azimuth averaged value , and two others by the condition that the local density is 20% of that maximum value , and falls either radially inside of , or outside of , a circle centered on the origin and 220 au in radius .
these contours correspond roughly to the torus inner and outer boundaries and to its high density core .
finally , given the points defining the three contours , we use the numerical recipes routine ` mrqmin ' @xcite to extract the semi - major axis , the eccentricity and the axis of pericenter for the best fit ellipses to each isosurface .
in addition to the three fit parameters , we also permit the origin of each ellipse to be offset from the system barycenter in the @xmath144 and @xmath145 coordinates , taking the final ellipse parameters from the offset for which the @xmath146 value for the fit is minimized . in order to avoid contamination of our fits by circumstellar material , we mask out a circular region of the grid , centered on the origin . its radius is defined to be slightly larger than the largest extent to which the circumstellar material extends at any time during its orbit , which will occur when the binary is at apoapse and also includes the combined radii of the two circumstellar disks . specifically , we define radius of the circumstellar mask region as : @xmath147 with the identification of the semi - major axis and eccentricity of the binary using their usual identifiers , @xmath148 and @xmath149 , the first term defines half the apoapse separation of the binary components . in the second term , the quantities @xmath150 and @xmath151 are each defined using the expression derived in @xcite , to approximate the largest radius for which an orbit around the star is stable over a suitably long duration .
we multiply the combined radii by a scale factor , @xmath152 , to account for the fact that material may be present intermittently at larger distances from each star for shorter durations than are accomodated by the @xcite formalism .
figure [ fig : torus - fits ] shows isodensity contours and fits for simulation _
wide3ehi _ , at each of the three locations defined above , both as raw contours and as convolved with a 10 au psf .
the times at which the fits were obtained correspond to the same apoapse and periapse passages shown in figure [ fig : wide - ecc-1orb ] , below . in each panel ,
the fitted ellipses are shown along with the contours but are frequently obscured by the data , visually demonstrating the fit quality .
although the differences between the raw and the convolved contours are not large , they are nevertheless significant . for example , in the case of the maximum isodensity contours ( center panels ) , the accretion streams are eliminated in the convolution .
little change occurs in the fitted ellipses in these cases , but both the inner and outer contours exhibit much larger differences , particularly so in the case of the orientation .
for example , the difference is @xmath153 degrees in each of the two snapshots of the outer contour , a value comparable to the precession angle over the timespan of the two fit times .
the largest differences are seen in the fits for inner contour at apoapse ( upper left panel ) , where the difference is nearly 90 degrees .
the inner contours are also those for which both the difference between the two times is most significant and the fidelity of the fit to the underlying data is least accurate . while the apoapse snapshot ( top ) requires significant eccentricity for a good fit to both the raw and convolved data , the periapse snapshot ( bottom ) is best fit to a much more circular shape . in this case , although the contour is indeed intrinsically more circular , the fit remains compromised by the incompletely obscured presence of the accretion streams in the contours which contribute non - negligibly to the fit .
we conclude that while our fits to the middle and outer contours are minimally affected by such features , fits to the inner contour will be of lesser quality due to such contributions .
figure [ fig : torus - radii ] shows fits for each of the three semi - major axes defined above , as functions of time , along with the offset of each ellipse from the actual system barycenter . in each case the semi - major axis varies over time , with the inner ellipse featuring the most rapid changes and the outer ellipse the slowest .
the amplitudes of the oscillations extend some @xmath8 au in either direction from its long term averaged value .
the inner and outer ellipses also exhibit slow secular trends to smaller and larger extents , respectively , as the torus itself becomes distorted by the passage of spiral structures through it .
although the trends are visible over the entire duration of the simulation , it is unlikely that they will continue over much longer time spans . for the inner ellipse ,
the trend will be constrained by the fact that the action of the binary will tidally truncate the torus edge .
the outer edge trend will be limited by the fact that the spiral structures decrease in amplitude as they move outwards , merging with the smooth background disk . in each case
, the offset of the best fit ellipse s origins from the system barycenter fluctuates most rapidly of all , but with comparatively small amplitudes of @xmath154 au , and often with evident correlations between the set of offsets for the three ellipses at a given time . at late times in the simulation ( @xmath155 yr ) , the offsets each increase to between 10 and 20 au , in response to the larger inhomogeneities in the tori from which they are derived , as seen in section [ sec : evo - circumbin ] .
even these late time offsets remain substantially smaller than the @xmath156 au offsets derived from the system by gds99 and @xcite however . in part
, we attribute this difference to the fact that the fits to the observations assume a circular torus , while our ellipse fits include additional degrees of freedom , which permit the ellipse origin to relax toward the system s barycenter , while retaining a good fit to the shape .
more importantly however , the discrepancy serves to highlight an important difference between the characteristics of the observations and our simulations .
namely , that the fitted parameters for our simulations do not trace exactly the same physical features as fits of various torus parameters derived from observed systems do . here , they account only for density features in our 2d representation of the flow , while in observed systems such as , the various system characteristics include features that trace both the system s 3d geometry , such as the disk scale height , and its radiated output , both from scattered and internal sources .
such characteristics may be correlated with the density , but do not trace it directly .
therefore , although the specific radii will be similar to those observed for corresponding features in the torus , it is important to make the distinction clear , in order to avoid confusion .
the eccentricity of the three ellipses described in the text , defining the inner ( solid ) , center ( dotted ) and outer ( dashed ) limits of the torus , each as a function of time , for simulations _
clos0ehi _ and _ wide3ehi _ , as labeled . ]
figure [ fig : torus - ecc ] shows the eccentricities as functions of time , for the same fits shown in figure [ fig : torus - radii ] . in both simulations , and for all three fits , the tori quickly become significantly eccentric . the eccentricities for simulation _
clos0ehi _ however , typically near @xmath157 , fall well below those for _ wide3ehi _ , where they are found near @xmath158 .
the difference is consistent with the fact that the torus is more strongly perturbed in the wide binary configuration .
for all three fits , the eccentricities also rise to values typical of their longer term evolution within only 1 - 2 orbits of the binary , but continue to vary widely over the full duration of the simulation .
this rise time is far shorter than a single orbit of the torus material defining the edge itself , and demonstrates the importance of the stars on the torus morphology .
consistent with its proximity to the perturbing effects of the stars , the torus inner edge suffers the largest variations from its initial circular form . in the case of the wide orbit configuration ,
the eccentricity undergoes repeated excursions to values greater than @xmath159 and less than @xmath160 , corresponding to a nearly circular shape .
also , the variations appear to be nearly periodic , perhaps arising as consequences of the binary s influence on the inner torus . to the extent that they are quantifiable however , the ocscillations do not correspond to the binary orbit period through any simple relationship , arguing against any direct connection .
similar variability is present in the fits derived from central torus region , but the maxima do not extend to values as large as at the inner edge .
also , in both simulations , the eccentricities of the central ellipses remain near zero for long time spans , alternating with periods in which their values are much higher .
in contrast , both the inner and outer ellipses typically retain their large eccentricities during these periods .
direct examination of the simulation morphology reveals that the systems have developed several , more or less equally spaced , spiral structures at these times . in consequence , though the non - axisymmetric character of the torus remains , the best fit ellipse derived from it becomes more circular , as the various features driving the fit to higher eccentricity cancel each other out .
later , as the spiral structures evolve into more asymmetric shapes , the eccentricity increases again .
the trend towards slower and smaller amplitude variations continues in the eccentricity for the outer edge . for both configurations ,
the outer ellipses require more time to increase to similar values as are seen for the other two .
in contrast to the inner and central ellipses , the outer ellipses exhibit only comparatively small excursions away from their long term averages and the variations occur most slowly with time .
they also exhibit none of the periods of near circular shapes ( low eccentricities ) seen for the other two . even at the end of the simulations ,
the long term trend values continue to increase , meaning that the outer edge has yet to evolve to a configuration for which all artifacts in the results due to the initial conditions have been fully erased . the axis of pericenter of the three ellipses described in the text , defining the inner , middle , and outer limits of the torus as labeled .
wide3ehi _ is shown with the solid curves , while simulation _
clos0ehi _ is shown with the dotted curves . ]
figure [ fig : torus - omega ] shows the axis of pericenter angles fitted for the tori over time . by definition ,
this parameter specifies the angle along which the long axis of the ellipse points , with a zero angle being defined here as the long axis of the stellar orbit ellipse , for the eccentric binary case , with the same zero point used for the for the circular binary simulation as well . in the case of the eccentric binary configuration
, we might expect the torus to remain aligned along that same axis , since the binary s orbital motion implies a strong forcing function to facilitate such alignment .
this expectation is clearly not met in our simulations . instead , the tori around both the eccentric and circular orbit binary appear to precess over time , as indicated by their continuing increases in angle and their periodic returns to zero as the motion sweeps around the compass . the tori precess at different rates depending on distance from the system barycenter , with both simulations producing similar rates for the fits at the middle and outer edge of the torus , as measured by the similar slopes of the curves and the time required to return to a given orientation .
in contrast , the precession behaviors for the inner ellipses in the two simulations are not similar to each other .
instead , the precession in the eccentric binary run corresponds approximately to a periodicity of twice the binary orbit period ( i.e. @xmath161 yr ) , over the course of many binary orbits , while for the circular binary run , it is much longer .
we speculate that the precession in the first case is being driven by the forcing action of the binary . in the circular binary case , and for the center and outer tori ,
the influence of the stars on the tori is much weaker . in these regions ,
the precession periods correspond loosely to the orbital periods of the tori at each location , with the exception discussed below for the center torus .
although the correspondence itself is apparent , we do not believe a direct , causal connection between the precession rates and the orbital motion of the torus material can be established however because , in general , the precession is clearly only approximately periodic .
the axis of pericenter angle neither changes at a uniform rate , nor does it increase monotonically with time . instead , occasional reversals occur , as the fit adapts to details of the local flow , and sudden discontinuities in the angle are also common , as the fit adjusts to the growth or decay of some structure in the torus .
such events are visible at @xmath162 yr in fit for the outer torus edge in the _ clos0ehi _ simulation , for example , and much more frequently in the center and inner torus fits .
jumps are particularly common in the central torus , and occur at the same times for which we noted above that the fitted eccentricity hovers near zero . given this correspondence , we attribute the jumps there to the same cause : i.e. that the torus has evolved into a set of spiral structures of near equal spacing and appearance , causing the fitted axis of pericenter to become highly sensitive to small scale features of the torus as a result . of some interest
is that the phase of the precession in one simulation is frequently similar to that in the other the pericenter angles are nearly the same at the same times for the outer and middle tori .
we believe that the similarity is not driven by any physical condition common to the two configurations , but is rather a consequence of the identical initial condition specified for the torus , which has not yet been erased by its subsequent evolution .
the inner torus edge follows the same pattern as well , both for the correlation in angle and in periodicity , but the correlation remains close for less than a single orbit period there , before being lost .
observations of have conclusively demonstrated that a significant quantity of material is present in very close proximity to the stars themselves .
recent observations have even detected material in locations where its dynamical lifetime can not be exceed a few hundred years @xcite before being accreted into the circumstellar disks or otherwise cleared from the system . in this work , we have assumed that all circumstellar material is initially contained in circumstellar disks orbiting each star , and in this section , we discuss the evolution of that material as it evolves .
we will investigate the processes by which material is transfered into the circumstellar environment from the circumbinary environment , and what configurations the cirumstellar disks evolve towards over time as material moves through them into and out of them .
the morphology of a system with a wide , eccentric orbit ( simulation _ wide3ehi _ ) is shown in figure [ fig : wide - ecc-1orb ] for various phases of its evolution through one binary orbit period .
details of the orbital evolution vary between orbits , and we have chosen to show the eighth orbit after the simulation s beginning ( i.e. beginning at @xmath163 yr ) because it exhibits the largest magnitude episode of mass accretion of any single orbit during the simulation ( see section [ sec : acc - onto - disks ] below ) . it therefore provides the most easily visible illustrations of similar events that occur at other times , but which are of smaller magnitude . in addition , rather than
show frames equi - distant from each other in time , we have chosen frames selected to highlight various aspects of the flow as the binary travels through a complete orbit , referencing the time since last apoapse for various time intervals in the orbit in the text . in the first frame of the mosaic , where the components are at apoapse , the disks show little evidence of internal structure such as spiral density waves .
consistent with its higher mass ( see section [ sec : acc - onto - disks ] below ) , the mass densities in the secondary disk ( to the left in the image ) , are higher than in the primary .
the shape of the cavity between the stellar components and the circumbinary material is significantly non-`round ' and includes fragmentary spiral structures that are remnants of previous interactions between the components .
these fragments are traveling outwards towards the circumbinary torus , which they will soon rejoin .
as the binary components begin another orbit , falling towards each other , large streams of material are pulled away from the torus and follow the stars on similar inward trajectories . due to its orbital motion , the inbound material forms large scale spiral structures
as it travels through the gap region and does not directly impact either circumstellar disk . by the time the stars have reached their 1/4 phase ( frames 2 - 4 , some 80 - 140 yr after apoapse ) ,
the leading edges of the spiral structures have overtaken the stellar components and wrapped around them , with the close tidal interactions causing them to begin to spread out again as they move away from the star .
a portion of this material may impact the other component of the binary , while the remainder forms a temporary ` bar ' extending between the two binary components .
material arriving slightly later ( frames 4 - 7 , 140 - 220 yr after apoapse ) , does directly impact portions of the outer circumstellar disks and the flow pattern becomes quite complex .
the time interval shortly before and during periapse is , unsurprisingly , the interval over which the most complex interactions take place . the various interactions between the infalling streams , the ` bar ' and disks cause substantial disruption in each . as a consequence , the spatial extent of the circumstellar material increases , with mass from all three sources contributing to the extended disks for each star .
the amount and distribution of material in the accretion streams is near its maximum at this time as well .
it is noteworthy however , that only a small fraction of the material in the streams actually accretes into the circumstellar environment , with the remainder returning to the torus .
illustrating this fact , figure [ fig : velocity ] shows the velocity magnitude at the same periapse passage shown in the sixth panel of the mosaic in figure [ fig : wide - ecc-1orb ] , along with velocity vectors on a small set of representative particles as well .
the latter show the motion of the streams extending well around and away from both the stars and the surrounding circumstellar material .
near the circumstellar disks , velocity magnitudes in the streams increase to as much as 5 - 8 km / s , comparable to the keplerian motion of the material within the circumstellar disks themselves , and well in excess of the 2 - 3 km / s orbital velocities of the components themselves .
the high velocities generated as the streams move through the gap are also important for conditions well within the circumstellar environment itself .
figure [ fig : disk - temps ] shows the midplane temperatures of the material , again at periapse . while the accretion streams themselves contain the coldest of all material in the circumstellar environment , at temperatures below @xmath550 k , temperatures in small regions near the impact points of the streams increase to 3 - 500 k as the kinetic energy of the infall is converted to heat .
these temperatures are well in excess of the @xmath164 k temperatures found in immediately adjacent regions of the disks , not affected by the streams impact . these @xmath164 k temperatures at the outer disk edges are the lowest found within the disks . at other locations ,
temperatures increase to as high as 1000 k ( extending to values beyond the color bar range shown in the figure ) within 1 au of each star .
non - azimuth symmetric structure is also visible , particularly in the disk around the secondary in which temperatures increase to as high as 300 k inside the spiral structures that develop .
we will discuss implications of these temperatures in section [ sec : plan - form ] , below .
the midplane temperature distribution in the 120@xmath143120 au region around the system barycenter at the time of the eighth periapse after simulation _
wide3ehi _ begins .
the secondary is on the right .
midplane temperature is determined for each sph particle and at each time using the procedure described in section [ sec : energy - diss ] . ] as the stars pass through periapse and begin to recede from each other again ( frames 7 - 9 of figure [ fig : wide - ecc-1orb ] , 200 - 240 yr after apoapse ) , the extended disk begins to shrink as the material it contains begins to reintegrate into the main volume of the circumstellar disks . at the same time , the disk structures closer to the stars become more inhomogeneous , due to their interactions with the highly eccentric flows of the infalling material that disrupt their keplerian motion .
the disk structures generated by the interactions near periapse decay to near axisymmetry over the next 1 - 200 yr , as the stars recede from each other and again approach apoapse ( frames 8 - 12 , 250 - 420 yr after apoapse ) . during these later phases of the orbit , and at larger distances from the stars , material in the tidal streams which was able to pass by both circumstellar regions without impacting either , continues on trajectories which returns it to the circumbinary torus from which it came .
as they travel through the gap , the spiral structures into which material is organized begin to lose the sharp definition that characterizes their appearance in the earlier phases of the binary orbit .
such loss of coherence is most noticable in the innermost portions of the streams , for which tidal interactions with the binary were largest .
nevertheless , such structures remain coherent well after their return to the main volume of the torus and , as discussed in section [ sec : evo - circumbin ] above , their existence in and influence on the torus are of major importance for the overall density and temperature structure well beyond its inner edge .
a markedly different pattern of behavior occurs in the case of a @xmath2 au orbit for the binary .
figure [ fig : clos - ecc-1orb ] shows the evolution of the _ clos3ehi _ simulation over the course of one binary orbit . in comparison to the wide orbit case , a much larger fraction of torus material has moved inwards into what was originally the outer gap region . despite this , and in contrast to the behavior seen with the wide orbit model , the remaining gap region continues to be clear of material over the entire binary orbit .
accretion streams do develop , but are quite weak and remain distinct for only a fraction of the distance inward to the stars .
no bar develops between the stars at periapse because the weaker accretion streams are unable to feed the circumstellar region with enough material to produce one .
also , due to the closer orbital separation , the circumstellar disks are spatially more compact than in the wide orbit case and assymetries that develop within them are also correspondingly less distinct . with small variations ,
the same behaviors appear during each binary orbit in both the wide and close cases .
we have not shown the @xmath3 models corresponding to the eccentric orbit models described here , however we note that their behaviors largely mimic those of their eccentric cousins .
an important exception to this trend is the fact that only the wide , eccentric model in figure [ fig : wide - ecc-1orb ] shows the strong accretion streams and episodic accretion .
we have seen that as the system evolves through time , the action of the binary generates large amplitude density structures throughout the circumbinary torus , extending outwards into the low density disk that surrounds it .
likewise , similar structures propagate inwards from the torus and , at various orbital phases , contact the circumstellar disks .
some of the mass in these streams becomes a permanent part of the circumstellar environment , while the remainder propagates outwards to rejoin the torus again .
in addition to mass accretion from the circumbinary environment , mass contained in the circumstellar disks is subject to the action of the same strong gravitational torques that perturb the circumbinary material .
it may therefore migrate radially inward towards the star and , when it gets close enough , be accreted by the star . in this section
, we quantify the rates and timing of the accretion into and out of the cirumstellar environment , and the masses of the disks that result . as one aspect of our discussion
, we will describe simulations of the same physical configuration at two different resolutions , differing by a factor of four in particle count . through the relationship in eq .
[ eq : balsara ] , this difference will be sufficient to change the numerical viscosity by a factor of @xmath165 .
a difference of this magnitude will in turn be sufficient to change the accretion by a similar factor , if indeed resolution plays a significant role in the accretion processes we model .
we will discuss such issues in section [ sec : stellar - env ] . .
figures [ fig : masstransw ] and [ fig : masstransc ] show the mass accreted by each of the circumstellar disks as a function of time for the various simulations in our study . for purposes of defining the `` circumstellar disks '' for the primary and secondary stars , we define an outer boundary to that environment to be 50% larger than the radius defined as the last stable orbit by @xcite . for each of the wide binary simulations ,
the total mass transfered into each of the circumstellar disks over the 6500 yr duration of the runs is comparable in magnitude to the assumed initial disk masses themselves .
accretion occurs preferentially onto the disk around the secondary star for both the wide and close systems , and for both eccentric and circular orbits .
the torus - only models generate higher accretion rates than do the torus / disk models , presumably because the higher mass concentration in the inner circumbinary region is more readily available for transfer inward . for the close binaries , mass transfer occurs at rates some @xmath8 times lower than occurs for corresponding simulations with wide orbits , so that only a small fraction of the initial disk masses are replaced over the lifetimes of the simulations . at early times in these simulations
, mass actually appears to be transfered out of the circumstellar disks .
this is , however , merely a consequence of the initial spreading of the disks away from their initial condition and the artificial limit we imposed to define the disk edge .
the accretion rates in these configurations are low enough to show small , apparent oscillations in the calculated disk masses for the eccentric orbit case ( fig [ fig : masstransc ] ) .
in fact , these oscillations are more likely due to the changes in the disk s spatial dimensions at different phases of its orbit .
such ` flexing ' also causes a small fraction of the disk mass to migrate outwards temporarily , beyond the limiting radius we have defined for the disk edge , before moving again into closer proximity to its host star .
eccentric orbital motion enhances the time averaged accretion rates over those observed for circular orbit systems , with rates a factor of @xmath166 higher in an eccentric orbit system than in an otherwise identical circular orbit system .
in addition , for the eccentric wide binaries , accretion is highly episodic , with substantial mass transfer occuring in bursts some 30 - 40 yr in duration and relatively little transfer occuring at other times .
these bursts reflect the periodic gravitational perturbations of the torus by the stars as they travel closer and further from its inner edge and excite spiral structures there . as described in sections [
sec : wide - morph ] and [ sec : circstar - conf ] , spiral structures generated in the circumbinary disk by the action of the binary fall through the otherwise cleared gap region outside the binary orbit . as they travel through the circumstellar environment , a portion of their length may encounter one or the other circumstellar disks and by accreted by it . while similar features form in the simulations with @xmath2 au ( albeit much weaker ) , the trajectories of the spiral structures through the gap region do not intersect with the circumstellar disks due both to the smaller spatial extent of the disks and to their greater distance from the torus itself . due to the evident variability in the magnitude of the accretion events from orbit to orbit , we have not attempted a detailed analysis of their exact timing relative to the binary s orbital phase .
we note however that the bursts do vary from orbit to orbit , as does the instantaneous accretion rate and total mass accreted during one event . typically , the events begin at orbital phases prior to orbital periapse and reach their maximum intensity near periapse , declining again as the stars recede from each other
. the timing does not appear to be strongly related to the binary motion itself , but rather to the time required for material originating outside the circumstellar environments to travel inwards far enough to be accreted into one or the other circumstellar disk .
in addition to mass transfer into the circumstellar environment from the torus , mass leaves the circumstellar disks and accretes onto the stars .
figures [ fig : massaccw ] and [ fig : massaccc ] show the total mass accreted onto the stars from their disks as functions of time , for the wide and close series of simulations , respectively .
in contrast to the episodic accretion patterns of mass entering the circumstellar disks , accretion onto the stars exhibits no periodicities or outburts , apart from an initial transient period during which the disk material loses its memory of its initial condition and moves towards a more physically appropriate distribution . for the wide binary simulations , accretion rates after the initial transient fall between @xmath167/yr for the @xmath4 series , and approximately half those rates for the @xmath3 series .
the eccentric orbit models each exhibit a slight increase in rates at late times , while the @xmath3 models remain steady at near constant rates . given the larger accretion rates _ into _
the circumstellar disks in the eccentric orbit simulations , increases in the accretion rates _ out of _ those same disks is a natural consequence as the larger material volume evolves along the same evolutionary path .
exactly the opposite effect is true of the close orbit models . in those cases ,
the initially rapid mass accretion transient drains a substantial fraction of the mass from the disks . because this material is not replaced as rapidly ( see section [ sec : acc - onto - disks ] , above ) , the disks become depleted of material ( see section [ sec : diskmasses ] below ) and the accretion onto the stars continues to slow down for the entire duration of the simulations . in all of our simulations , the resolution of the circumstellar disks is extremely coarse .
of the nearly @xmath168 million particles employed in our high resolution simulations , only about 20000 are initially allocated to the circumstellar disks and , as we shall see in section [ sec : diskmasses ] below , that number does not increase substantially over the life of the runs .
because of this low resolution , the viscous dissipation originating from numerical sources is quite large . in order to ensure that our conclusions regarding the accretion rates through the disk are valid , we must therefore demonstrate that those accretion rates are not sensitive to the numerical viscosity inherent in the simulations themselves rather than to physical processes present in the actual system we are modeling . to this end
, we ran identical physical models at two different resolutions , and have included both the low and high resolution results in figures [ fig : massaccw ] and [ fig : massaccc ] . in every case , apart from a larger rate of accretion during the initial transient , the cumulative mass accreted onto the stars for the low resolution variant follows a curve with an essentially identical slope to its high resolution counterpart .
we conclude that our results are not sensitive to the high numerical dissipation caused by the limited resolution of the circumstellar environment .
our conclusions based on these results are therefore also insensitive to the resolution .
although quantifying the mass accretion into and out of the circumstellar disks as functions of time are important measures of the system , by themselves they are incomplete because they do not specify the mass contained in the disk at any given time .
figures [ fig : massdiskw ] and [ fig : massdiskc ] show the mass of the disks orbiting the primary and secondary stars as a function of time , for the full set of wide and close simulations in our study . in each case , the most significant changes in the disk masses over the first @xmath169 yr are due to the transient redistribution of disk material as it loses memory of its initial condition .
thereafter , and as the circumbinary material begins to exhibit activity , disks in the wide , eccentric orbit models exhibit periodic accretion events , which correspond to the orbital motion of the binary .
in contrast , the disk masses in simulations where the orbital eccentricity is zero settle into steady states with little variation over time . for the close orbit models ,
the masses of the circumstellar disks continues to decrease for both the @xmath4 and @xmath3 series of simulations over the entire duration of the runs . after the 6500 yr lifetime of the simulations , the masses of both circumstellar disks
has fallen to values an order of magnitude or more lower than their initial masses .
we anticipate that the disks would have been entirely accreted within a comparatively short time of additional evolution .
further evolution was not possible however , because the number of particles remaining in the circumstellar disks decreased to such an extent that instabliities developed in the time step controls for the simulations ( due to the poor resolution of the disks ) .
little forward progress of the simulations was possible under such conditions , and we therefore terminated them . because of the larger mass accretion during the initial transient , our lower resolution runs could only be run for about two orbits of the circumbinary torus , much earlier than was required for the higher resolution realizations .
so far , we have described the behavior of simulations with initial conditions specified to be very similar to , but we have not yet discussed in any depth , direct comparisons of our results to it .
other systems similar to also exist , to which the results of our simulations may be suitably generalized as well . of these , one of the most similar is uy aur , with the recent discussions of @xcite providing the most comprehensive description of the system to date .
the similarities between it and the system are striking , both in the data themselves , and in the difficulties experienced in trying to reconcile them with a single self - consistent model of the system .
although the uy aur binary has a wider separation and its disks much lower masses , both systems exhibit a remarkably similar inventory of features otherwise .
such features include structures in the circumbinary and circumstellar environments , streamers and stellar orbit planes which may be misaligned with the circumbinary disk if , as @xcite propose , the system is actually a higher order multiple rather than a binary .
here , we discuss comparisons of our results with observations of and make a number of conclusions about the system and its evolution , with an implicit understanding that the results will be applicable to a larger class of hierarchical multiple systems of similar character .
the observed circumbinary torus of contains a number of features @xcite , including spokes , gaps , variations in brightness and torus width at different position angles along the torus .
other work ( e.g. gds99 ) , using high resolution interferometry , has shown that the edges of the torus are no more than @xmath10 au in radial extent . despite these observations ,
detailed characterization of the structures and attribution to specific causes is difficult because the features themselves approach size scales comparable to the resolution limits of the telescopes .
while similar constraints apply to our simulations , the spatial scale at which we are constrained is far smaller , thereby permitting both comparisons to be made , as well as , to some extent , predictions of types of features that may be revealed by still higher resolution observations .
figure [ fig : wide - ecc - closeup ] shows a blowup of one panel of the mosaic in figure [ fig : wide - ecc - manyorb ] , along with a snapshot of the system at the preceding periapse passage of the binary .
two streamers are clearly visible in the gap region between the circumstellar material and torus , particularly in the periapse image , which extend outwards to the torus and join with it .
the streamers remain distinguishable as coherent structures for well over a full revolution around the system and , as we will discuss in more detail in section [ sec : evo - circumstar ] , they do not persist as stable structures over the course of the binary s or torus orbit . instead , they travel through the gap region , with a small portion impacting the circumstellar disks , but with the largest fraction of material contained in the streamer simply traveling outwards again , where it rejoins the torus .
just such behavior has occured in the time leading up to the apoapse snapshot , where the streamers no longer extend all the way inwards to the stars , but instead appear effectively as the torus inner boundary . at this time
, portions of each streamer also exhibit small scale structure , particularly at the lower left and upper right portion of the gap , due to the fact that as they traveled these components have been stretched to the largest extent . even after reentering the torus
, the structures themselves do not lose coherence .
instead , they continue to propagate as distinct entities , interacting with other streamers already present , which were generated during previous orbits of the binary .
as is clear from these images , as well as similar panels in figure [ fig : wide - ecc - manyorb ] , the resulting torus is far from a smooth , azimuth invariant entity .
the combined density structures continue to stretch and deform as they evolve further . in some portions of the torus ,
the remmants of multiple streamers converge to form closely connected filaments . in others , few such filamants are found .
each of these features is defined by very distinct edges , which appear both as its inner and outer boundaries and also as fine structure entirely within the main body of the torus .
eventually , the features reach the outer edges of the torus , where they effectively define its outer edge .
the torus internal structure as seen in our simulations invites comparison with the various structure observed in the system .
most easily compared are the streamers which appear in the gap region itself .
although many observations are at the limits of what can be observed , so that their interpretations are not without ambiguities , some , more recent observations , such as those of @xcite and @xcite have begun to remove those uncertainties .
for example , @xcite interpret their observations of emissions from hot gas as evidence for streamers of material infalling onto circumstellar material , which has been shock heated as it impacted the circumstellar disks .
long wavelength observations of @xcite reinforce this interpretation , showing dust emission extending inwards from the torus to the circumstellar environments where the hot gas resides .
correspondence between the streamers in our simulations and these features in the observed system are easily made , particularly for the images shown near periapse , for which the streamers extend inwards all the way to the circumstellar disks .
some portions are also visible as a small ` bar ' in the region extending between the binary components themselves .
the short time scales over which the features in our simulations exist before dissipating back into the torus , also correspond well with the locations where the hot material is observed , in locations where its presence can not be explained except as a dynamically transient phenomenon .
finally , the presence of high temperature material in the outer portions of the disks ( see figure [ fig : disk - temps ] , above ) , shows that the accretion streams do , indeed , generate exactly the observed behavior pattern . within the torus ,
the distribution of spiral structure as a function of azimuth is highly inhomogeneous .
for example , in the periapse image , while some comparatively low density structures are visible in the image both to the left and right of the stars , far higher density structures ( yellow and red in the images ) are visible above and below them .
as discussed in sections [ sec : wide - morph ] and [ sec : close - morph ] , such structures are visible over the entire duration of the simulations , and change their appearance as system evolves .
although the orbital timescales in the torus are thousands of years , the time scales for changes are far shorter , as evidenced by the marked differences between the two images in figure [ fig : wide - ecc - closeup ] , which are separated in time by @xmath170 yr . from the earlier image to the later , the large amplitude spiral arms have both traveled forwards in their orbits , and changed in shape .
for example , the large spiral arm near the top of each image is longer and more dense in the top image than in the bottom .
similarly , the spiral arm appearing in the bottom left quadrant of each image is of higher density and angular extent in the top image than in the bottom .
the features in our simulations are qualitatively similar to the descriptions of depressions , filaments and gaps in the hst images discussed in @xcite and @xcite .
unfortunately , although the features are quite clear in our simulations , the interpretation of the features in the observations themselves less so , due simply to the fact that they are near the detection limits of the telescope , where they are difficult to distinguish from various sources of noise .
the features in our simulations are natural consequences of the dynamical activity generated by the motion of binary and here , we make tentative identifications of these features with those reported in the system .
further observations would substantially bolster the case for such identifications , but would require substantially more finely resolved images that are presently available .
we anticipate that any such improved observations , repeated over the course of multi - decade timescales , may also begin to resolve variability of the systems appearance in time , though such an observing program would be an extremely challenging endeavor . finally ,
the streamers , and their coherence over long periods of time , provide a natural explanation for the observed sharply defined edges of the torus . in figure
[ fig : dens - lineouts ] , we show the radial profiles of the surface density in the torus along three rays extending outwards from the system barycenter , each derived from the surface density mappings described in section [ sec : quantitative ] , below . the high density regions in the profiles correspond to slices through the various spiral structures visible in figure [ fig : wide - ecc - closeup ] , and their sharply defined edges to their inner and outer extents along the ray .
for example , the green curves , corresponding to two rays extending upwards in the two images each intersect a spiral arm whose outer edge is found just inside 250 au in the top panel , and just inside 260 au in the bottom . at those locations , the surface density decreases abruptly from @xmath171 gm/@xmath105 to @xmath154 gm/@xmath105 over a spatial scale of @xmath172 au .
similar features are present at various other locations in the profiles , with those in the radially inner portions corresponding to streamers located within the gap intersecting the rays , while those at larger distances corresponding to larger scale spiral disturbances within the torus itself . in some cases , the inner edge appears more sharply defined , as is the case for the feature at @xmath173 au along the 45 ray in the upper panel . in others , the outer edge is sharper , as discussed above for the 90 rays .
features with lower surface densities and lower contrasts are also present in the profiles , extending to distances as far as @xmath174 au from the system barycenter .
each of the profiles include density contrasts of up to factors of @xmath175 , over spatial scales of @xmath172 au in extent along the ray .
these scales are comparable to or smaller than the @xmath10 au widths derived for the inner and outer ring edge width by @xcite and @xcite for the system .
each of these edge width derivations are based on observations that approached the resolution limits of the telescopes themselves so that , in principle at least , the widths might be expected to be significantly narrower . on the other hand , the smoothing lengths of the sph particles located in these regions is @xmath176 au , and we conclude that both the spiral structures and their edges are well resolved in our simulations .
therefore , given the results of our simulations , we expect that future observations will fully resolve the edge structure at size scales of a few au .
in section [ sec : circumbin ] we described the process by which we defined initial conditions for the torus and disk and noted that determination of self consistent initial conditions ( i.e. rotation velocities consistent with the implied pressure and gravitational forces ) were difficult to specify self consistently at the interface between the two components .
the sharp density gradients produced correspondingly large pressure gradients that generated difficulties in defining self consistent orbit velocities near the interface . to sidestep the inconsistencies ,
we employed the ` workaround ' of widening the interface in our initial condition over a radial region some @xmath177 the observed width of the interface . given the the sharp interfaces described this section , a much more palatable physical model for interpreting the data is that the sharpness of the inner and outer edges of the torus is not a feature of a cylindrically symmetric distribution of material in rotational equilibrium , but is more likely simply to be due to the spiral structures generated by the actions of the binary on the torus and of self gravitating disturbances .
in the discussion above , we noted that in many of our simulations the circumbinary torus fragmented into one or more massive objects .
we did not however , discuss the significance of those objects , either in the context of their significance from a numerical standpoint , or in the context of what they might mean for the system .
we now consider such questions . of primary importance in evaluating our results is to answer the question of whether or not the fragments are simply artifacts of our numerical methods or of deficiencies in the physical assumptions used to model the system .
our first task , then , is to determine whether or not the simulations obey known requirements for resolving fragmentation , such as those described in ( * ? ? ?
* hereafter n06 ) .
the most pertinent of these criteria is the requirement that the fragments be resolved by a sufficiently large number of particles , both as they form and afterwards .
a quantification of ` sufficiently large ' is that the locally value of the ` toomre mass ' , as defined in n06 , be resolved by no fewer than @xmath178 the number of sph neighbor particles , which corresponds to about 200 particles in 2d .
we find that the conditions within the fragments produced in our simulations are such that the toomre mass reaches its smallest values of @xmath10 in the highest density portions of a fragment .
given the torus mass and the particle counts in our simulations , the mass of a single sph particle is @xmath179 .
the ratio of these two masses is approximately 10@xmath180far higher than the required number of sph neighbors , and we conclude that the fragmentation is well resolved by our simulations .
additional considerations for checking the sensitivity of our simulations to various numerical and physical parameters are found in a series of papers by meru & bate . in @xcite , they find that , in addition to the result of @xcite that disks fragment if their cooling timescale is shorter than @xmath181 orbit periods ( depending on the adiabatic exponent , @xmath68 ) , fragmentation is also sensitive to a variety of other physical parameters .
in addition to the cooling timescale , they examine the sensitivity of fragmentation in disk simulations to the opacity in models with radiative cooling , and to the mass and temperature profiles in the disk .
they determine an empirical relationship between the cooling timescale required for fragmentation and the mass distribution and also note that disks heated by external radiation , from e.g. stellar irradiation ( as are those in our models ) , tend to be less likely to fragment due to the additional heating .
later however , in @xcite , they realized that their previous results were sensitive not only to changes in the various physical models , but also to the number of sph particles used to resolve the disks in their simulations .
they concluded that in many cases , their simulations were not converged numerically , and questioned whether any numerical simulations that included radiative transport could possibly converge to a result which accurately modeled fragmentation in circumstellar disks .
after a further investigation @xcite , they revise their initial conclusion to note that fragmentation can be resolved , but depends strongly on the magnitude of the artificial viscosity employed . are these conditions important for the simulations described in our work ?
since our models also include an approximate radiative cooling model , they too may be affected by a similar deficiency , and we therefore explore that possibility here .
we note first , that meru & bate did not discuss the application of any of the n06 resolution criteria to their results .
given this fact , we begin with such a comparison in order to understand whether the n06 conditions are more or less restrictive than the convergence criteria they describe .
the condition most directly available for comparison in the various meru & bate works is the n06 requirement that the vertical structure in 3d disks be resolved by at least @xmath182 sph smoothing lengths per disk scale height , at the disk midplane .
figure 10 of @xcite shows this quantity , and we observe that the azimuth averaged values of the ratio @xmath183 ( smoothing length over scale height the inverse of the quoted n06 criterion ) , are larger than the required 1/4 for all but the outer portions of the two million particle run shown . based on this observation , we conclude that their simulations with two million particles are at best marginally resolved in their outer regions , while the inner regions and those runs with fewer particles are insufficiently resolved .
the direct consequence , as shown in n06 , is that the mass densities in the disk midplane in those runs will under - shoot their physically correct ( i.e. fully resolved ) values by a factor of as much as 30% when @xmath184
. this condition will be true for all of the simulations in @xcite and many in their later works @xcite as well .
the effect such errors will induce in other physics based quantities or the overall outcome of a given simulation is difficult to assess directly . it is noteworthy however that the sensitivities they describe are much less pronounced above the 2 million particle threshhold @xcite at which the vertical structure criterion of n06 is marginally satisfied in the outer parts of their disks .
we conclude that the trends in their simulations are likely to be manifestations of numerical inaccuracies similar to those underlying the n06 vertical structure resolution requirement .
this conclusion , and the arguments on which it is based , is also shared by @xcite . in addition , those authors argue that one possible origin for the resolution sensitivity may lie in resolution dependent changes in the thermal energy input supplied by the bulk artificial viscosity , even when that contribution is small overall .
they emphasize however that the case for such a mechanism is ambiguous , because some simulations , at high resolution and for which their model predicts fragmentation , actually do not fragment . that absence
is then taken as evidence for physical mechanisms to suppress fragmentation .
we have discussed the importance of two such mechanisms in our work ( radiative heating and stirring generated by the binary interactions ) , and so discount the resolution criterian in meru & bate , in favor of satisfying the n06 toomre mass criterion directly , as the best indicator that numerically induced fragmentation is absent in our simulations . given that our simulations do not contain numerical deficiencies which generate fragmentation
, we now discuss the significance of the relevant physical models and their contribution to fragmentation .
fragmentation occurs when an imbalance betweeen gravitational and pressure forces develops which favors gravitational forces in some volume so that it begins to collapse , typically on timescales of a few orbits @xcite .
the imbalance develops when either insufficient internal energy is absorbed by the disk via radiative heating , or is generated in it by dissipation of kinetic energy , or when cooling permits internal energy to escape more quickly than it can be replaced .
each of these processes decrease the gas pressure through the equation of state , and each are strongly time dependent in character , varying as the disk evolves .
all of our simulations exhibit strong spiral structures generated by the combination of self gravitating instabilities and the tidal interactions with the binary .
some also produce fragments on short timescales compared to the system lifetime .
we conclude that our simulations model systems lie very near their fragmentation boundary , with some configurations being slightly more susceptible to fragmentation and others slightly less so .
in contrast , and although marked structure exists , the system is not known to contain any well defined fragments within its torus , even though the observed torus conditions imply a strongly self gravitating structure .
this means that the balance between heating and cooling must be very finely held , because the instabilities are neither fully suppressed , nor do they grow to sufficient amplitude to generate fragements .
further , this balance has been sustained over a long time , in spite of the sensitive time dependence noted in @xcite , since the system lifetime is long compared to the @xmath5few orbit timescale required for fragmentation to occur . given
that our simulations model configurations defined to be very similar to , such that similar physical behaviors are expected in both , we conclude that the balance between heating and cooling is much more finely held in reality than is possible for our models to reproduce . nevertheless , two important , additional conclusions about the system are possible based on our results .
first , we recall our conclusions above that large scale structures seen in all of our simulations owe their origin not only to the tidal action of the binary but also to self gravitating instabilities in the torus .
although no clear pattern emerges from our simulations that would indicate which configurations are more likely to generate fragments in the system , their presence in systems with many different binary orbital parameters indicate that the conditions are generally favorable for fragmentation .
we can therefore conclude that the system itself is very near its fragmentation boundary .
given our results however , we can not predict whether or not such an outcome actually will occur .
second , we conclude that the combined energy input from the radiative heating from the stars and the stirring action of the binary are responsible for a large fraction of the thermal energy input into the torus , and thereby play an important role in its long term stability .
this is demonstrated by comparing the simulations described in section [ sec : heatoff ] . in the first
, we model a single star of the same mass and luminosity as the combined masses and luminosities of the binary stars in our other simulations , but with the same torus . in the second , we model a binary while omitting the stellar radiative heating source . the torus is marginally stable with an embedded binary that includes heating , producing none or at most one or two fragments within the several orbit time frame of the simulations . in contrast , removing either the radiative heating or the tidal stirring causes the torus to become violently unstable , producing nearly 40 fragments in less than a single torus orbit .
such an outcome clearly demonstrates both the unphysical nature of the evolution where those processes are absent and , at the same time , their importance where they are present .
gds99 report an inclination of 37@xmath181 degrees for the torus relative to the line of sight .
this value is derived from the observations by deprojecting the observed eccentric appearance of the circumbinary torus into a circular shape . given this inclination ,
attempts to fit the available astrometric data for the binary s orbit presently remain somewhat ambiguous . if an orbit coplanar with a circular torus is assumed , then the best fit orbital parameters are @xmath185 au and @xmath186 @xcite .
unfortunately , these parameters prove to be somewhat unsatisfactory from the point of view that the semi - major axis is smaller than is expected based on dynamical arguments and the position of the inner edge of the torus . recognizing this shortcoming , the same authors and others fit for the orbital parameters under the relaxed assumption that the planes of the stellar orbit and the torus may be misaligned . in this case
, @xcite determined best fit orbit parameters of @xmath9 au , @xmath187 and a mutual inclination of @xmath188 degrees .
they also report a second family of fits with mutual inclinations of nearly 90 degrees , but find that the disk in such systems is either disrupted entirely or becomes very thick .
following similar arguments regarding alignment , the later analysis of @xcite found the ` most plausible ' orbit parameters to be @xmath189 au and @xmath190 , with a relative inclination between the two planes of @xmath191 degrees . under both sets of assumptions however , the fitted orbital parameters still remain unsatisfactory based on the dynamical argument that a substantial misalignment between the stars and torus would quickly generate a warp in the torus , while no such feature is observed .
given the analysis of section [ sec : quantitative ] , we believe that the data and the analyses of them may be fully reconciled if we relax the assumption that the torus is circular , rather than the assumption that the torus and binary orbit planes are inclined with respect to each other .
under this alternate assumption , deprojection of the torus actual eccentric shape into a circular shape results in an erroneous determination of its inclination .
then , the perceived misalignment is simply a consequence of the incorrect determination of the torus inclination , rather than a property of the system itself
. this alternate assumption will be plausible if the mutual inclination determined from the fit is comparable to the inclination derived from misinterpreting an elliptical torus as a circular torus inclined at an angle , @xmath84 , with respect to the line of sight .
for a given ellipse which is interpreted as a circle , the inferred inclination is related to the true eccentricity , @xmath149 , via the equation @xmath192 therefore , given a torus with eccentricity @xmath4 , its inclination would be perceived to be @xmath193 degrees .
similarly , a torus with @xmath160 or @xmath159 would be perceived to have an inclination of @xmath194 or @xmath195 degrees , respectively .
the same error is made if some other plane defines the torus absolute orientation , rather than the plane of the sky . in the case of the system , this plane can be specified as the plane defining the orbit of the two stars in the binary .
the mutual inclinations derived in each of the analyses above fall well within the range of ` false ' inclinations which could be attributed to misinterpreting the eccentric shape of the torus itself . on this basis
, we conclude that the ambiguity in orbit parameters can be resolved in favor of the larger semi - major axis .
this conclusion restores consistency between the observations of the torus dimensions specifically the location of its inner edge and the dynamical results of @xcite , which predict a wide binary orbit based on the location of lindblad resonances that drive tidal truncation .
it is further reinforced by the fact that only the circumstellar disks in our @xmath1 au simulations remain massive enough to remain consistent with the masses derived from observations .
the disk masses in all of our simulations with @xmath2 au decrease quickly to much smaller values due to the combination of losses through accretion onto the stars and the lack of replenishing material accreting onto them from the circumbinary environment .
observations indicating the presence of accretion streams reinforces our conclusion further , for two reasons .
first , such streams are essentially non - existent in our @xmath2 au simulations , but ubiquitous in our @xmath1 simulations , particularly in the @xmath4 run .
second , assuming an @xmath1 au semi - major axis , the system must be relatively close to its periapse passage , given its apparent separation .
this is important because our simulations indicate that accretion streams extending inward to the circumstellar disks occur only around this time , becoming weaker or entirely absent at other orbital phases .
thus , the simulations and observations suggest the same consistent model for the system itself .
we note that an important consequence of adapting this assumption is that deprojection of the apparent shape of the torus becomes a much less accurate method of determining the system s actual inclination .
the difficulty is compounded by the fact that the eccentricity of the torus varies widely and its orientation is not correlated with the orientation of that of the binary s orbit plane .
other constraints will be required in order to disambiguate the torus actual inclination and its actual eccentricity .
observations of the system show conclusive evidence both for circumstellar material and for accretion into the circumstellar environment .
both hot and cold material are observed @xcite in a region extending some tens of au from the stars , in both circumstellar disks and in accretion streams .
unfortunately , the observations extend over only a very short time window in comparison to a full orbit of the binary , or even a full periapse passage .
they therefore define only a snapshot of the system rather than a full evolutionary picture of its behavior over time . given the similarities we have already noted between our wide orbit simulations and those observed in , it will be useful to apply the comparisons first to make interpretations of the current state of the system in the context of an evolutionary model , and then to make predictions of its behavior as it evolves in the future .
above , we concluded that the degeneracy in the orbital motion data for the stars in the system can be resolved in favor of an orbit with @xmath9 au .
this conclusion is further reinforced by the presence of significant amounts of material in the circumstellar environment of because mass accretes into the circumstellar environments at rates sufficient to sustain and even grow the disks _ only _ in our wide orbit simulations .
little accretion occurs in the close orbit runs .
instead , mass transfer into the circumstellar disks is very limited in those models and the disks initially assumed to be present quickly accrete onto the stars , almost entirely emptying them of material .
these conditions are inconsistent with the observations of features extending tens of au from the stars in the system .
the presence and sizes of the circumstellar disks are also consistent with the interactions by each component of the system on the other , which lead to tidal truncation of the circumstellar disks at sizes comparable to the dynamical stability limit described by @xcite and shown numerically by @xcite .
the episodic accretion seen in our eccentric binary simulations is quite reminiscent of the same feature in the work of @xcite . as in their work , we find that accretion occurs preferentially onto the secondary component and that , due to this preference , the mass ratio between the components will tend towards equality over long periods . the same conclusion was drawn in @xcite , for the long term , time averaged accretion rates from massless , circumbinary disks with isothermal evolution assuming temperatures both warmer and colder than in our study . given that the system is in a wide orbit , the fits provided in @xcite show that the stars are approaching their periapse passage , rather than receding from it .
examination of our simulations ( e.g. panels 6 - 8 of figure [ fig : wide - ecc-1orb ] ) show that at this time the circumstellar environment should include a significant amount of material originating in the accretion streams and wrapping around the disks and extending outwards to the torus .
a similarly important feature of the system is that the circumstellar accretion disks should exhibit spiral structures and other asymmetries , driven by the tidal interactions of the binary components .
both classes of features are consistent with the observations .
to these interpretations of the system s short term state , we add the longer term prediction that , over the duration of the periapse passage , as much as a few tenths of a jupiter mass will accrete onto the circumstellar disks of the system , preferentially onto the secondary component , as shown in figure [ fig : masstransw ] . an important feature of the system that our simulations do not reproduce is that the mass determined for the disk around the secondary component is substantially less than that around the primary component ( as discussed , e.g. , by * ? ? ?
* ) . in our simulations ,
the reverse is true
mass accretes preferentially onto the disk around the secondary , and its mass increases to @xmath165 times that of the disk around the primary ( figure [ fig : massdiskw ] ) .
should we interpret this inconsistency as an indication that our models should be ruled out as inconsistent with the observations ?
fortunately for our models , itself provides the means to reconcile simulation and observation : it simply is not the binary we have modeled .
instead , recent interferometric observations of @xcite have shown that the gg tau ab component is itself a binary consisting of two m stars separated by @xmath23 au . unlike the system we modeled , the circumstellar disks in this system will be tidally truncated to radii no larger than about 11.5 au each .
the shorter dynamical times inherent such small disks will lead to correspondingly reduced disk lifetimes and masses , consistent with the observations , even in the presence of rapid replenishment from the accretion streams .
given this confounding factor , the inconsistency between our simulations and the data is to be expected , rather than being evidence for their invalidity .
after its accretion into one or the other circumstellar disk , mass continues to migrate through the disk , ultimately accreting onto the star around which it orbited .
accretion rates onto the stars in have been determined by @xcite ( who derive a rate of @xmath196/yr ) and @xcite ( who derive a rate of @xmath197/yr ) and are in very good agreement with the rates inferred from our simulations , of a few @xmath198/yr .
each of these works consider only the accretion onto both stars as a sum , rather than onto the individual components .
our results , summed over both stars , fall roughly half way between the two observational determinations , supporting the conclusion that our simulations accurately model the system .
our rates , in turn , are about an order of magnitude higher than those computed by @xcite . while we have not investigated the origin of the differences between our results and theirs
, we note that our initial setup was significantly different from theirs , as was our numerical method .
the accretion rates correspond , in numerical terms , to rates of 1 - 2 particles per year , or several hundred per binary orbit . given this level of discreteness
, such values suggest that we should consider the accretion in our simulations to be , at least in part , a phenomenon governed by numerical processes rather than physics .
perhaps surprisingly however , the rates do not appear to be strongly correlated to resolution .
accretion driven primarily by numerical viscosity for example , should be proportional to the particle smoothing length , through equation [ eq : balsara ] , and in turn to the square root ( in 2d ) of the number of particles . instead , both the high and low resolution variants of the same physical configuration return similar patterns of accretion over time .
we conclude , due to this relative insensitivity , that the accretion is not driven primarily by viscous processes derived from numerial effects , but rather by other physical processes , such as gravitational torques .
mass transfer into the circumstellar disks occurs at rates similar to those for stellar accretion for wide systems but at much lower rates in the close systems .
thus , concerns similar to those noted for the stellar accretion must be addressed for these cases as well .
@xcite studied the sensitivity of the mass transfer rates into circumstellar environments to the resolution employed in 2d sph simulations .
they found that rates were numerically converged in simulations where the ratio of particle smoothing length to disk scale height fell in the range @xmath199 near roche radii of the stars .
this occured at accretion rates of a few hundred particles per binary orbit . in our high resolution runs
, we find that @xmath200 in this region .
given the similarity in these parameters between their work and ours , we conclude that the mass transfer rates in our simulations are also converged .
in contrast , the streams in the close orbit systems are far less massive and are correspondingly more poorly resolved
. the comparatively poorer resolution of the mass transfer will not be fatal to our conclusion that the rates are much lower than the wide orbit runs , because the simulations are based on identical initial configurations of the circumbinary material .
thus , the weaker streams are simply a consequence of the intrinsic differences between the physical configuration of the stellar and circumstellar material rather than any shortcoming in the numerical realization .
the results we present may be extended to considerations regarding the likelihood for planet formation in or other multiple systems . from an observational perspective for example , @xcite detect the presence of both cold and hot material in the circumstellar environment of and interpret the presence of the cold material as evidence for possible planet formation in the system . from a theoretical perspective
however , such a conclusion requires not only the presence of substantial amounts of material in circumstellar disks , but also a mechanism for accumulating that matter into massive , condensed objects on short timescales .
theory suggests that the most likely candidates responsible for forming jovian mass planets and low mass brown dwarfs are either gravitational fragmentation of large scale spiral structure @xcite or coagulation of small solid grains followed by later accretion of additional gas ( i.e. core accretionsee * ? ? ?
* ) in the disks of forming stellar systems . in n00
, one of us concluded that planet formation in the circumstellar disks of equal mass binary systems with a separation similar to that in the system was unlikely .
the conclusion was based on the fact that temperatures in the disks were too high either to support gravitational instabilities or to permit solid grain growth since grains would be vaporized in large portions of the disks .
it was based on models of a system similar to l1551 irs 5 , a system much younger than and for which the disks are still relatively massive . to what extent
does a similar conclusion hold for systems like , which has evolved to a much later evolutionary epoch , and for which the circumstellar disks are far less massive ? in answer to this question , we note that the temperatures seen in figure [ fig : disk - temps ] ( section [ sec : circstar - conf ] , above ) are similar to those reported in n00 , while the disk masses are more than an order of magnitude lower .
the effect of similar temperatures and smaller disk masses will make the disks even more stable against the growth of self gravitating structures , as defined by the value of the toomre @xmath49 parameter , than reported in n00 .
planet formation by way of gravitational fragmentation is therefore suppressed to a correspondingly greater degree .
similarly , high disk temperatures are significant for the core accretion model , because the budget of solids entrained in the gas decreases where temperatures rise above the vaporization temperatures of those solids . in the present instance , the ` ice line ' , at which planet formation via the core accretion model is enhanced by the presence of additional solid material , will lie near to or outside the actual boundaries of the disks .
water ices will therefore not be present within the circumstellar disks of , except perhaps at high altitudes above the midplanes , and planet formation via core accretion will be inhibited .
in addition , while the accretion of new material into the circumstellar environment might be expected to enhance the growth of dust and small solid bodies , the fact that the overall accretion rate through the disk is high will negate this benefit . any accumulations of dust grains and small planetesimals
will quickly be swept along with the overall mass flow through the disk and be lost onto the star .
therefore , based on the same arguments presented in n00 , we can extend the conclusion in n00 to the later stages of evolution typical of .
in extending the n00 conclusion , we are mindful of apparently contradictory data embodied in the detections of planets in mature binary systems ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in many cases , these systems are characterized by semi - major axes similar to that found in making the contradiction more acute .
it can be resolved however , by noting that the companion masses derived for these objects ( or implied for them through their effective temperatures ) are substantially lower than those reported for the primaries and that few planets in binaries with near equal mass ratios were found .
such systems are therefore distinct from the cases considered here and our conclusion does not apply .
further contradictions emerge however , when observations of young systems are accounted for , such as is seen in @xcite , who show that @xmath201 of young , close binary systems have lost their circumstellar disks by the time they are 1 - 2myr in age , without regard to stellar mass ratios .
they conclude that planet formation will be less likely in binary s compared to single stars .
such contradictions suggest that a comparative study of the two classes of systems ( of equal and unequal mass components ) would be of great interest , to explore the physical reasons for the dichotomy between the presence of planets in unequal mass binary systems and their paucity in equal mass systems .
given these dynamical constraints , we conclude that the cold material observed by @xcite has yet to interact substantially with the circumstellar environment or , for some portion , will soon return to the circumbinary torus as parts of the accretion stream moves out of the circumstellar environment .
we can not therefore confirm their conclusion that its presence in the circumstellar environment is evidence for planet formation .
nevertheless , the presence of both cold and hot material will remain of considerable interest for evolutionary models because together they define a reservior of nearly pristine material which has either recently undergone , or will soon undergo , its first significant heating event .
it may therefore retain signatures of the processes important during that event . in a comparison study of single and moderate separation binary systems ,
@xcite conclude that the first stages of planet formation
grain growth will not be affected by the presence or absence of a binary companion .
they base their conclusions on the lack of observed differences between the 10@xmath202 m silicate features seen in the spectra of disks around stars in single and multiple systems .
while the presence of silicates and other solids in disks is certainly a requirement for the formation of planets , in and of itself , simple presence is insufficient to differentiate between a system where planet formation is more likely and a similar system where planet formation is less likely .
these features may instead trace material which is unlikely to become important for planet formation because of its short lifetime in the disks .
we believe that such is the case in the system .
having completed our study of systems configured to be similar to and discussed their physical significance in relation it , we now summararize our results and suggest future studies that may profitably build on ours . we have modeled the evolution of a forming binary star system , configured to resemble the system , using the sph code vine .
the system consists of two stars , each surrounded by a circumstellar disk and in orbit around each other , with a circumbinary torus@xmath0disk structure in orbit around the combined system .
our simulations include the self - gravity of the disk material , external heating of the gas due to radiation from two stars and radiative cooling from the disks .
we follow the evolution of a suite of systems for @xmath56500 - 7500 yr , corresponding to about three orbits of the torus around the center of mass .
we find that strong spiral structures develop in the torus due to the stirring action of the binary and to its own self gravity .
the spiral structures are well defined in space , are characterized by edges of @xmath203 au in width and frequently extend more than 360 around the center of mass .
we interpret the observations of sharply defined structures in the torus as evidence for such spiral structures .
the spiral structures propagate outwards over time , generating a net outward mass flux extending into the low density circumbinary disk surrounding the torus . at large distances , they lose coherence , and we interpret the presence of the disk component of the circumbinary material as an excretion feature produced as the spiral structures transport mass to that region . in our torus - only configurations ,
about 15% of the mass of the torus moves outwards over the @xmath204 yr duration of our simulations about half of the observed mass of the present disk ( 30% of the torus mass ) .
the rates are much lower in the torus@xmath0disk initial configuration , and we expect that on a longer timescales , as the disk component gains mass , a nearly stationary state is reached between the outwards mass transport into the disk , and the mass escaping from it , through its outer edge or other mechanisms . structures created at different times frequently interact with one another and , late in the evolution of several simulations , fragment into one or two clumps with initial masses comparable to brown dwarfs . in a model where we replaced the binary with a single star of the same mass , some 38 clumps formed in less than a single orbit of the torus .
we interpret this result as an indication of the importance of the binary s stirring action as a contributor to the thermal energy input of the torus , and as a strong indication of the importance of self gravity in the torus . the torus changes shape and orientation as the spiral structures propagate through it .
we fit its shape to an ellipse whose eccentricity varies between @xmath205 and an orientation which precesses on timescales comparable to the torus orbit period .
we point out a degeneracy in the interpretation of the data in the calculation of the torus inclination and its eccentricity , if the torus is actually eccentric rather than circular in shape . given this degeneracy , we resolve the question of the supposed misalignment between the torus and the binary s orbit , by interpreting the data as a coplanar system in which the torus is eccentric , rather than circular .
we simulate systems configured with semi - major axes of either @xmath1 au ( ` wide ' ) or @xmath2 au ( ` close ' ) , and with assumed orbital eccentricity of either @xmath3 or @xmath4 .
although torus structure appears in all configurations , only wide configurations generate substantial streams of material extending inward to the circumstellar disks . a small fraction of the material in these streams accretes into the circumstellar environment , with the remainder returning to the torus . in wide binary configurations , and over the time span of our simulations ,
mass comparable to the disks themselves is accreted onto both primary and secondary components .
mass accretes preferentially onto the secondary at steady rates in @xmath3 configurations , but is episodic in @xmath4 configurations with highest rates occuring near binary periapse .
in contrast to the wide configurations , the streams in close binary configurations contain much less material and the rate of accretion into the circumstellar environment is two orders of magnitude smaller .
accretion onto the stars is steady in all cases , at rates sufficient to drain the disks almost entirely of material within the time span of our simulations , if they are not otherwise replenished .
the net mass transport into and out of the disks in the wide binary configurations , produces disk masses which remain steady near their initial values in @xmath3 cases , but double over the same time span in @xmath4 cases . in close binary configurations however , the net mass inflow and outflow is insufficient to replenish the disks before they accrete onto the stars .
replenishment of disk material from the torus in wide orbit configurations can therefore significantly extend the lifetime of the circumstellar disks .
mass replenishment is a necessary , but not sufficient , condition for planet formation in the circumstellar disks .
we find that other conditions , such as material temperatures , are not compatible with planet formation in the circumstellar disks , and conclude that planet formation there is unlikely .
given the results of our simulations , we interpret the observations of the system in favor of the wide , eccentric binary orbit solutions of @xcite and @xcite , for which parameters of @xmath9 au and @xmath186 are derived . in each case , the fits require that the torus and binary orbit be inclined with respect to each other . as noted above , we resolve this inconsistency by interpreting the supposed mutual inclination as an eccentricity in the torus .
we also favor these parameters over the fits otherwise prefered by each of the previous works , namely those with @xmath206 au and @xmath186 , because the circumstellar disks in such systems would be accreted onto their respective stars within a few thousand years , and because material streams in such systems would be far weaker than are observed .
the results presented in this work answer a number of questions regarding the character of the system , while raising others , and leaving others untouched .
first among the questions raised by our results are the questions of whether the detailed morphology of features in the circumbinary torus as seen in our simulations are actually present in the system .
the current best resolution observations of the torus provide tantalizing hints that such features do exist , but a definitive statement that such features are present must await higher resolution observations of quality similar to those described by @xcite , for the hl tau circumstellar disk . another important question concerns planet formation in multiple systems . whether or not any theoretical mechanism presently exists to explain the formation of giant planets in binary systems
, the fact remains that at least a few binary systems , such as @xmath68 cephei @xcite , gi86 @xcite and hd 41004 @xcite , do harbor planets .
therefore some formation mechanism does in fact exist . while we find that accretion into the circumstellar disks occurs rapidly enough so that they can survive for the comparatively long time scales needed to form planets , other conditions , such as temperatures , remain quite unfavorable .
what are the mechanisms still missing from our models that permit such objects to form ? finally , our simulations model the evolution of the ` a ' component of the full system over a time span extending over only a tiny fraction of its formation time scale and in only two spatial dimensions .
we neglected the full dynamical effects expected to be present in the system , insofar as our results include neither the distant binary ` b ' component of the system , nor the newly discovered tight binary nature of the b component .
even so , we find large scale morphological changes even over this short time span and restricted dimensionality and physical system .
given the vigor of the activity over such a short time scale , we would expect activity of similar scale over to occur over longer time spans as well , with correspondingly large consequences on the system morphology .
to what extent will 3d effects also play a role in the evolution ? what will be the end state configuration of the system as a whole ? will the components eventually break apart ?
future investigations extending the work presented here will be required in order to answer these questions .
we wish to thank the anonymous referee for comments which improved our manuscript .
portions of this work were carried out under the auspices of the national nuclear security administration of the u.s .
department of energy at los alamos national laboratory under contract no .
de - ac52 - 06na25396 , for which this is publication la - ur-16 - 23283 . | we present 2-dimensional hydrodynamic simulations using the smoothed particle hydrodynamic ( sph ) code , vine , to model a self - gravitating binary system .
we model configurations in which a circumbinary torus@xmath0disk surrounds a pair of stars in orbit around each other and a circumstellar disk surrounds each star , similar to that observed for the system .
we assume that the disks cool as blackbodies , using rates determined independently at each location in the disk by the time dependent temperature of the photosphere there .
we assume heating due to hydrodynamical processes and due to radiation from the two stars , using rates approximated from a measure of the radiation intercepted by the disk at its photosphere .
we simulate a suite of systems configured with semi - major axes of either @xmath1 au ( ` wide ' ) or @xmath2 au ( ` close ' ) , and with assumed orbital eccentricity of either @xmath3 or @xmath4 .
each simulation follows the evolution for @xmath56500 - 7500 yr , corresponding to about three orbits of the torus around the center of mass .
our simulations show that strong , sharply defined spiral structures are generated from the stirring action of the binary and that , in some cases , these structures fragment into 1 - 2 massive clumps .
the torus quickly fragments into several dozen such fragments in configurations in which either the binary is replaced by a single star of equal mass , or radiative heating is neglected .
the spiral structures extend inwards to the circumstellar environment as large scale material streams for which most material is found on trajectories which return it to the torus on time scale of 1 - 200 yr , with only a small fraction accreting into the circumstellar environment .
the spiral structures also propagate outwards through the torus , generating net outwards mass flow and eventually losing coherence at large distances from the stars .
the torus becomes significantly eccentric in shape over most of its evolution . in all configurations ,
accretion onto the stars occurs at a steady rate of a few @xmath6/yr , with the net result that , without replenishment , the disk lifetimes would be shorter than @xmath7 yr .
our simulations show that only wide orbit configurations are able to retain circumstellar disks , by virtue of accretion driven from the robust material streams generated in wide configurations , which are very weak in close configurations . in wide , eccentric orbit configurations ,
accretion is episodic and occurs preferentially onto the secondary , with rates strongly peaked near binary periapse . based on our results
, we conclude that the torus is strongly self gravitating and that a major contribution to its thermal energy input is the shock dissipation associated with spiral structures generated both by self gravitating disturbances and by the stirring action of the binary .
we interpret the sharply defined features observed in the torus as manifestations of such spiral structures .
we interpret the low density disk surrounding it as an excretion disk created by the outward mass flux generated by the spiral arms as they propagate outwards .
typical eccentricities calculated for the shape of the tori modeled in our simulations are large enough to account for the supposed @xmath8 mutual inclination between the stellar orbit plane of and its surrounding torus , through a degeneracy between the interpretation of inclination of the torus and its eccentricity .
we therefore interpret the observations in favor of a coplanar system with an eccentric torus . because accretion onto the disks occurs at rates sufficient to sustain them only in wide orbit configurations ,
we conclude that the gas currently resident in the circumstellar disks of the system has been accreted from the torus within the past few thousand years .
although circumstellar disks will persist over time spans long enough to permit planet formation , the overall environment remains unfavorable due to high temperatures and other conditions .
given , the presence of circumstellar disks , robust accretion streams , and our interpretation of the stellar orbit plane as coplanar with the torus surrounding it , we conclude that the system is in an eccentric , @xmath9 au orbit , resolving questions in the literature regarding its orbit parameters . |
the zone method approach to the fermion sign problem is based on the observation that in many finite temperature simulations fermion permutations are short ranged @xcite .
this can be true even for systems with massless modes and long distance correlations , as we demonstrate with an explicit example . in this proceedings article
we derive an estimate for the finite temperature fermion wandering length and discuss the main features of the zone method .
we begin with a brief review of the worldline formalism @xcite .
we introduce ideas in one spatial dimension before moving on to higher dimensions .
let us consider a system with one species of fermion on a periodic chain with @xmath0 sites , where @xmath0 is even . aside from an additive constant ,
the general hamiltonian can be written as @xmath1 + \sum_{i}c_{i}a_{i}^{\dagger } a_{i}.\ ] ] following @xcite we break the hamiltonian into two parts , @xmath2 and @xmath3 , @xmath4 .
\nonumber\ ] ] we note that @xmath5 . for large @xmath6 , we can write @xmath7 where @xmath8 inserting a complete set of states at each step , we can write @xmath9 $ ] as @xmath10 the worldline trajectory of each of the fermions can now be traced from euclidean time @xmath11 to time @xmath12 .
since we are computing a thermal trace , the worldlines from @xmath11 to @xmath12 define a permutation of identical fermions .
even permutations carry a fermion sign of @xmath13 while odd permutations carry sign @xmath14 .
the generalization to higher dimensions is straightforward . in two dimensions , for example
, @xmath15 takes the form @xmath16
\right ] ^{n } \\
\qquad \qquad \qquad \approx ( s_{\text{o}}^{y}s_{\text{e}}^{y}s_{\text{o}% } ^{x}s_{\text{e}}^{x})^{n}. \end{array } \right . \label{2d}\ ] ]
in one spatial dimension , the pauli exclusion principle inhibits fermion permutations except in cases where the fermions wrap around the lattice boundary . for the remainder of our discussion , therefore , we consider systems with two or more dimensions . the first question we address is how far fermion worldlines can wander from start time @xmath11 to end time @xmath17 .
we can put an upper bound on this wandering distance by considering the special case with no on - site potential and only nearest neighbor hopping .
let us consider motion in the @xmath18-direction .
for each factor of @xmath19 in ( [ 2d ] ) a given fermion may remain at the same @xmath18 value , move one lattice space to the left , or move one lattice space to the right . if @xmath20 is the hopping parameter , then for large @xmath6 the relative weights for these possibilities are approximately @xmath21 for remaining at the same @xmath18 value , @xmath22 for one move to the left , and @xmath23 for one move to the right . in ( [ 2d ] )
we see that there are @xmath6 factors of @xmath24 . therefore for a typical worldline configuration at low filling fraction , @xmath25
we expect @xmath26 hops to the left and @xmath26 hops to the right . for
non - negligible @xmath27 some of the hops are forbidden by the exclusion principle .
assuming random filling we expect @xmath28 hops to the left and @xmath29 hops to the right .
the net displacement is equivalent to a random walk with @xmath30 steps .
the expected wandering length , @xmath31 is therefore given by @xmath32 in cases with on - site potentials , fermion hopping is dampened by differences in potential energy .
hence the estimate ( @xmath33 ) serves as an upper bound for the general case .
we have checked the upper bound numerically using simulation data generated by many different lattice hamiltonians with and without on - site potentials .
let @xmath34 be the logarithm of the partition function , @xmath35 \right\ } .\ ] ] let us partition the spatial lattice , @xmath36 , into zones @xmath37 such that the spatial dimensions of each zone are much greater than @xmath38 . for notational convenience
we define @xmath39 . for any @xmath40 ,
let @xmath41 be the logarithm of a restricted partition function that includes only worldline configurations such that any worldline starting at @xmath11 outside of @xmath42 returns to the same point at @xmath12 .
in other words there are no permutations for worldlines starting outside of @xmath42 .
we note that @xmath43 and @xmath44 is the logarithm of the restricted partition function with no worldline permutations at all . since the zones are much larger than the length scale @xmath38 , the worldline permutations in one zone has little or no effect on the worldline permutations in another zones . therefore @xmath45 using a telescoping series , we obtain @xmath46 @xmath47 for translationally invariant systems tiled with congruent zones we find @xmath48 where @xmath49 , the number of zones
. for general zone shapes one can imagine partitioning the zones themselves into smaller congruent tiles
. therefore the result ( [ est ] ) should hold for large arbitrarily shaped zones . for this case
we take @xmath50 to be the number of nearest neighbor bonds in the entire lattice and @xmath51 to be the number of nearest neighbor bonds in the zone .
we will refer to @xmath51 as the zone size of @xmath52 .
this is just one choice for zone extrapolation .
a more precise and complicated scheme could be devised which takes into account the circumscribed volume , number of included lattice points , and other geometric quantities . as an example of the permutation zone method
, we compute the average energy @xmath53 for a free fermion hamiltonian with only hopping on an @xmath54 lattice .
we set the number of time steps @xmath55 and consider values @xmath56 , @xmath57 , and @xmath58 .
the corresponding values for @xmath38 are @xmath59 , @xmath60 , and @xmath61 respectively .
the monte carlo updates were performed using our own version of the single - cluster loop algorithm @xcite . in figure 1
we show data for rectangular zones with side dimensions @xmath62 , @xmath63 , @xmath64 , @xmath65 , @xmath66 , @xmath67 , ... , @xmath68 .
we also show a least - squares fit ( not including the smallest zones @xmath62 and @xmath63 ) assuming linear dependence on zone size as predicted in ( [ est ] ) .
we find agreement at the @xmath69 level or better when compared with the exact answers shown on the far right , which were computed using momentum - space decomposition . while the physics of the free hopping hamiltonian is trivial , the computational problems are in fact maximally difficult .
the severity of the sign problem can be measured in terms of the average sign , @xmath70sign@xmath71 , for contributions to the partition function . for @xmath56 , @xmath70sign@xmath72 for @xmath73 , @xmath70sign@xmath71 @xmath74 and for @xmath75 , sign@xmath76 .
direct calculation using position - space monte carlo is impossible by several orders of magnitude for @xmath77 .
we have reviewed the zone method approach to the fermion sign problem .
we have demonstrated that the exchange of identical fermions is short - ranged and has a maximum range of @xmath78 lattice sites , where @xmath79 is the inverse temperature , @xmath20 is the hopping parameter , and @xmath80 is the filling fraction . we have introduced the notion of permutation zones , special regions of the lattice where identical fermions may interchange and outside of which they may not . using successively larger permutation zones , one can extrapolate to obtain thermodynamic observables .
applications of the zone method to chiral symmetry breaking in ( 2 + 1 ) dimensional qed is discussed in another article in these proceedings @xcite .
h. g. evertz , to appear in _ numerical methods of lattice quantum many - body problems _ , edited by d. j. scalapino , perseus , berlin , 2001 ; h. g. evertz , g. lana , m. marcu , phys .
lett . * 70 * , 875 ( 1993 ) ; h. g. evertz , m. marcu in _
quantum monte carlo methods in condensed matter physics _ , p. 65 , edited by m. suzuki , world scientific , 1994 . | we review a recently proposed approach to the problem of alternating signs for fermionic many body monte carlo simulations in finite temperature simulations .
we derive an estimate for fermion wandering lengths and introduce the notion of permutation zones , special regions of the lattice where identical fermions may interchange and outside of which they may not . using successively larger permutation zones , one can extrapolate to obtain thermodynamic observables in regimes where direct simulation is impossible . |
the recent discovery @xcite of the higgs boson at the cern large hadron collider ( lhc ) has certainly crowned in the most spectacular way an almost half - century long history of successes of the standard theory of the electroweak interactions , the so called standard model ( sm ) .
albeit the tremendous efforts put in by the experimental collaborations working in the lhc experiments the hunt for new physics ( supersymmetry , compositeness , extra dimensions etc .. ) has been so far unsuccessful .
impressive and increasingly higher new bounds on the scale of several beyond the standard model ( bsm ) scenarios are continually being updated .
however the cms collaboration has recently reported an excess over the sm background expectations in the @xmath0 and @xmath7 final states where @xmath8 is the missing transverse energy . the analysis in @xcite for a search of right - handed gauge boson , @xmath9 , based on 19.7 fb@xmath10 of integrated luminosity collected at a center of mass energy of 8
tev reports a 2.8@xmath4 excess in the @xmath11 invariant mass distribution in the interval @xmath12 .
a cms search @xcite for first generation lepto - quarks at a center of mass energy of 8 tev and 19.6 fb@xmath10 of integrated luminosity reported an excess of respectively 2.4@xmath4 and 2.6@xmath4 in the @xmath0 and @xmath13 channels respectively .
several attempts have been proposed in the literature to explain the above cms excesses in the context of various standard model extensions .
for instance in @xcite the authors propose an explanation of the excesses in the context of @xmath9 decay by embedding the conventional lrsm @xmath14 in the @xmath15 gauge group .
studies of the @xmath0 excess in the context of @xmath9 and @xmath16 gauge boson and heavy neutrinos ( n ) coupling mainly to electrons production and decay appear in @xcite .
similarly @xcite discusses a model with pseudo - dirac heavy neutrinos providing a fit to all excesses in a generic lrsm with arbitrary @xmath17 , @xmath18-@xmath9 boson mixing , heavy neutrino mass n and the @xmath19-n mixing .
in addition , the authors point we point out the consequences of the excesses for neutrino - less double beta decay @xmath20 decay , and find for example that @xmath20 actually provides a pretty severe limit on the @xmath19-n mixing assuming the excesses are real .
other interpretations have been proposed within the context of models with vector - like leptons as in @xcite showing that resonant pair production of such vector - like leptons decaying to an electron and two jets leads to kinematic distributions consistent with the observed cms data .
the @xmath0 excess has been shown to arise as well in @xmath21-parity violating models through spleton resonant production @xcite .
an alternative scenario based on lepto - quarks is proposed in @xcite , discussing also possible connections to dark matter , which fits the data of the excess seen by cms . in @xcite
the observed cms excesses are explained within superstring inspired @xmath22 models which can also accomodate for the baryon asymmetry of the universe via lepto - genesis .
other studies have emphasized that the observed differences between the @xmath0 , @xmath23 , same sign ( ss ) and opposite sign ( os ) channels and could be addressed including mixing and cp phases of the heavy neutrinos @xcite . on the other hand
it is well known that the like - sign di - lepton and di - jet ( @xmath0 ) , @xmath24 violating final state ( keung - senjanovic process ) , is the golden signature to look for heavy majorana neutrinos at high energy hadron collisions @xcite . studies of heavy ( pseudo - dirac ) neutrino production at the lhc within the inverse see - saw mechanism have been performed @xcite , also considering the quark - gloun fusion mechanism @xcite . in this paper , assuming the existence of a heavy majorana neutrino arising from the well known scenario of compositeness of quarks and leptons , complemented with contact interactions , we propose to study the @xmath0 signature and show that the model in its simplest version can reproduce the essential features of the observed excess in the @xmath0 invariant mass distribution .
we discuss how , with some refinement , it has the potential to address also other aspects of the excess , such as the absence of a peak in the @xmath25 invarant mass distribution , the charge asymmetry of the excess and the fact that the same excess is not observed in the @xmath23 channel @xcite . in this scenario the heavy excited states ( @xmath26 ) couple , through gauge interactions , with the ordinary sm fermions via magnetic type couplings .
current bounds on excited lepton masses ( generically indicated by @xmath1 ) have been recently strengthened by the lhc run i analyses @xcite of the @xmath27 signature arising from @xmath28 production ( @xmath29 ) , via four fermion contact interactions with a compositeness scale @xmath2 , followed by the decay @xmath30 .
in particular in @xcite the atlas collaboration reporting an analysis at @xmath31 tev with an integrated luminosity of 13 fb@xmath10 give a lower bound on the mass of the excited leptons @xmath32 tev ( derived within the hypothesis @xmath33 ) . in @xcite the cms collaboration reported the results of data collected with 19.7 fb@xmath10 at @xmath31 tev and ( always assuming @xmath33 ) excluded excited electron ( muon ) masses up to 2.45 ( 2.48 ) tev .
preliminary studies within the compositeness scenario of the like - sign di - lepton and di - jet signature were performed long ago @xcite , assuming the excited neutrino @xmath34 to be a majorana particle . here
our aim is to complement the composite majorana neutrino model of ref .
@xcite with contact interactions which are again a generic expectation of a composite fermion scenario @xcite . based on previous studies related to the production at lhc of exotic doubly charged leptons @xcite we expect these contact interactions to be the dominant mechanism for the resonant production of the heavy majorana neutral particles n in the process
. this expectation is indeed verified by our numerical simulations performed with a custom version of calchep @xcite where our model has been implemented .
the heavy majorana neutrino is produced resonantly in association with a lepton ( @xmath36 ) and then given the relatively important branching ratio for the decay @xmath37 we perform a detailed kinematic study of the like - sign di - lepton and di - jet final state : @xmath38 including the relevant standard model backgrounds .
our study shows clearly that a full fledged analysis of the upcoming data from the run ii of lhc at @xmath39 tev has the potential of observing the signature or alternatively excluding larger portions of the model parameter space compared to those already excluded from analyses of run i @xcite .
the rest of the paper is organized as follows : in sec .
[ sec_model ] we review the theoretical composite model ; in sec . [ sec_production ] we discuss the heavy neutrino production cross sections and decay rates ; in sec .
[ sec_sig - bkg ] we discuss the same - sign di - lepton and di - jet signature and the main associated sm backgrounds ; in sec .
[ sec_fast+simulation ] we present the results of the fast simulation obtained through the delphes @xcite software ; finally sec . [ sec_conclusions ] gives the conclusions with outlooks .
in this section we review the composite model of excited fermions investigated in @xcite within the hypothesis of a heavy majorana neutrino .
compositeness of ordinary fermions is one possible scenario beyond the standard model . in this approach
quarks and leptons are assumed to have an internal substructure which should become manifest at some sufficiently high energy scale , the compositeness scale @xmath2 .
ordinary fermions are then thought to be bound states of some as yet unobserved fundamental constituents generically referred to as _
preons_. while building a fully consistent composite theory has proven to be quite difficult some important and model independent features of the compositeness scenario can be phenomenologically addressed .
quite natural properties of this picture are @xcite : ( @xmath40 ) the existence of excited states of such low lying bound states of preons @xmath41 with masses @xmath42 ; and ( @xmath43 ) contact interactions between ordinary fermions and also between ordinary and excited fermions .
let us consider here the various possible composite models with respect to the idea of introducing lepton number violation ( lnv ) via a composite majorana neutrino .
+ ( a ) . + the homo - doublet model @xcite contains a left handed excited doublet along with a right handed excited doublet : @xmath44 typically the left and right handed doublet are assumed to have the same mass .
it is known that two left and right majorana fields with the same mass combine to give a dirac field ( with a dirac mass ) @xcite . the homo - doublet model ,
as laid out , can not therefore accomodate majorana excited neutrinos , and hence lepton number violation ( lnv ) .
this becomes possible if one is willing to introduce a mass difference between the left and right doublet ( @xmath45 mixing ) or , in other words , a breaking of the l - r symmetry .
such a possibility has been discussed for instance in ref .
@xcite where the @xmath46 is possibly a linear combination ( with mixing coefficients ) of majorana mass eigenstates . on the other end ,
if we do not want to introduce a mass splitting ( or mixing ) between the left and right components in the homo - doublet model , we can account for lnv advocating different models within the compositeness scenario which naturally can accomodate a majorana neutrino @xcite .
these are the following : + ( b ) .
+ the sequential model contains excited states whose left handed components are accomodated in su(2 ) doublets while the right handed components are su(2 ) singlets : @xmath47\,;\ ] ] and the notation [ @xmath48 means that @xmath49 is necessary if the excited neutrino is a dirac particle while it could be absent for a majorana excited neutrino .
the magnetic type interactions in this case can be constructed by coupling the left - handed excited doublet to the sm fermion singlets via the higgs doublet @xcite .
this results in interaction couplings suppressed by a factor @xmath50 @xcite where @xmath51 gev is @xmath52 times the vacuum expectation value of the higgs field .
+ it is assumed to contain a right handed doublet and left handed singlets : @xmath53\ , ; \qquad l^*_r=\left ( \begin{array}{c } \nu^*_r \\ e^*_r \end{array}\right)\,,\ ] ] where we may assume that there is no left handed excited neutrino ( @xmath54 ) so that we can associate to @xmath49 a majorana mass term and @xmath46 is a majorana particle .
this model is described by a magnetic type coupling between the left - handed sm doublet and the right - handed excited doublet via the @xmath55 gauge fields @xcite : @xmath56 where @xmath57 is the ordinary @xmath58 lepton doublet , @xmath59 and @xmath60 are the @xmath58 and @xmath61 gauge couplings and @xmath62 , @xmath63 are the field strength for the @xmath58 and @xmath61 gauge fields ; @xmath64 and @xmath65 are dimensionless couplings which are typically assumed to be of order unity .
the relevant charged current ( gauge ) interaction of the excited majorana neutrino @xmath66 is then : @xmath67 the above mirror type model is therefore the model to which we will refer our detailed simulation of the like sign di - lepton signature at the run ii of the lhc .
incidentally we note that sm extensions involving mirror fermions have been recently considered @xcite with respect to the phenomenology of the production of mirror quarks at the lhc .
at last we may add that one could also consider extended isospin composite models @xcite where the excited states are grouped in triplets ( @xmath68 ) or quadruplets ( @xmath69 ) instead of doublets ( @xmath70 ) as considered above .
such extensions of the composite scenario contain exotic charge states like doubly charged leptons and quarks of charge @xmath71 .
some phenomenology of these extensions relative to the doubly charged leptons has been addressed recently @xcite .
such extended weak isospin composite models could also be considered with the additional hypothesis that the excited neutrino is a majorana particle .
the model with @xmath68 can only couple @xcite the triplet @xmath72 with the right - handed lepton singlet @xmath73 .
therefore we may assume a sequential type structure with a left - handed triplet and right - handed singlets .
if the @xmath74 is missing we may assume for the @xmath75 a majorana mass term and so the excited neutral @xmath76 of the triplet is a majorana neutrino ( @xmath77 ) .
the magnetic type interaction reads : @xmath78 where @xmath79 is an unknown dimensionless coupling in principle different from @xmath64 appearing in eqs .
( [ mirror]&[laggi ] ) . the relevant charged current interaction of the neutral component of the triplet @xmath76 is in this case : @xmath80 which differs form the one in eq . in the chirality of the projection operator .
the @xmath69 quadruplet @xmath81 couples instead @xcite with the left - handed sm doublet , so that assuming a mirror type scenario and that there is no @xmath75 we can assign to @xmath74 a majorana mass term so that the @xmath76 neutral of the quadruplet can be a majorana neutrino ( @xmath77 ) .
the magnetic type interaction is @xcite : @xmath82 where @xmath83 is an unknown dimensionless coupling in principle different from @xmath84 and @xmath85 are clebsch - gordan coefficients . in particular in the case of @xmath69
the relevant neutrino charged current interaction turns out to have the same structure as in eq . :
@xmath86 therefore the interaction in eq . effectively describes the charge current interaction of a heavy majorana neutrino both in the ( @xmath70 ) mirror type model or in a composite model with extended weak isospin ( @xmath69 ) , always of the mirror type , provided that we make the correspondence @xmath87 .
contact interactions between ordinary fermions may arise by constituent exchange , if the fermions have common constituents , and/or by exchange of the binding quanta of the new unknown interaction whenever such binding quanta couple to the constituents of both particles @xcite .
the dominant effect is expected to be given by the dimension 6 four fermion interactions which scale with the inverse square of the compositeness scale @xmath2 : [ contact ] @xmath88 where @xmath89 and the @xmath90 factors are usually set equal to unity . in this work
the right - handed currents will be neglected for simplicity .
the single production @xmath91 proceeds through flavour conserving but non - diagonal terms , in particular with currents like the third term in eq .
[ jcontact ] which couple excited states with ordinary fermions : @xmath92 which were not considered in @xcite while are now fully implemented in our simulations . the gauge interactions in eq . and the contact interactions in eq .
have been implemented in the calchep @xcite generator .
the feynman rules corresponding to the lagrangians in eqs .
( [ laggi],[lcontact2 ] ) have been derived with feynrules @xcite a mathematica @xcite package which allows to derive the feynman rules of a given quantum field theory model once the lagrangian is given . in this study
we have added the gauge interactions given in eq . and
contact interactions in eq . taking into account the majorana nature of the excited heavy neutrino @xmath77 .
we conclude this section with one final remark regarding the assumption that in this work the dimensionless couplings @xmath93 are @xmath94 .
the production cross sections and the simulations presented in the following are obtained assuming @xmath95 .
this should be recalled when quoting the resulting bounds on the other parameters of the model , namely ( @xmath96 ) . at this
regard we point out that our results can not not be easily rescaled if @xmath97 due to the presence of the contact interaction mechanism which does not depend explicitly on these constants .
although the production mechanism is dominated by contact interactions the gauge interactions do affect the decay of the heavy neutrinos .
a direct comparison with other studies @xcite which derived bounds on the mixing parameters for the electron falvour of the heavy neutrinos is therefore not possible at the moment .
indeed our production cross sections can not be simply rescaled because of the presence of the contact interaction production mechanism .
we would need to implement in the generator a model with the additional parameters @xmath93 .
heavy majorana neutrinos @xmath77 can be singly produced in association with a lepton @xmath98 in @xmath99 collisions .
the process @xmath100 can occur via both gauge ( fig .
[ interactions ] , first diagram in the right - hand side ) and contact interactions ( fig . [ interactions ] , second diagram in the right - hand side ) .
we now present here the production cross section for the heavy majorana neutrino @xmath77 in @xmath99 collisions expected at the cern lhc collider stemming from the partonic collisions .
owing to the qcd factorization theorem , the hadronic cross section are given in terms of convolution of the partonic cross sections @xmath101 , evaluated at the partons center of mass energy @xmath102 , and the universal parton distribution functions @xmath103 which depend on the parton longitudinal momentum fractions , @xmath104 , and on the factorization scale @xmath105 : @xmath106 for the calculation of the production cross section in proton - proton collisions at lhc , we have used cteq6 m parton distribution functions @xcite .
the factorization and renormalization scale has been set to @xmath107 . in fig .
[ sig_prod ] ( left ) we present the cross section against the heavy neutrino mass for @xmath108 tev for the lhc center of mass energy @xmath109 tev .
it is evident that the contact interaction dominates the production of the heavy composite majorana neutrino by a factor that ranges between two and three orders of magnitude , varying the heavy neutrino mass between 1 and 5 tev , and for the given choice of the compositeness scale ( @xmath110 tev ) . in fig .
[ sig_prod ] ( right ) we compare the cross sections of @xmath111 with the one of @xmath112 .
the cross section for the production of positive di - lepton is larger than that for the production of negative di - leptons as expected in proton - proton collisions due to the larger luminosity of a @xmath113 pair ( nedded to produce @xmath114 ) compared to that of a @xmath115 ( nedded to produce @xmath114 ) .
the heavy majorana neutrino @xmath77 can decay again through both gauge and contact interactions .
the decay amplitudes are related , via appropriate crossing symmetry exchanges , to those describing the single production and depicted in fig .
[ interactions ] .
the possible decays are : @xmath116 in the first we can have a positive lepton , a down - type quark and an up - type antiquark or a negative lepton an up - type quark and a down - type antiquark ; in the second owing to the majorana character of @xmath77 we can have either a neutrino or an antineutrino of the same flavor of the heavy neutrino @xmath77 and accordingly two opposite sign leptons belonging to a family that can be the same or different from the other one , or in alternative a positive ( negative ) lepton of the same family of the heavy neutrino and a negative ( positive ) lepton and a antineutrino ( neutrino ) belonging to a family that can be the same or differnt from the other one ; in the third we can have a neutrino or an anti neutrino and a quark and an antiquark both of up - type or both of down - type . in fig .
[ decay ] we present the width @xmath117 and the branching ratio @xmath118 for @xmath119 , the decay that gives the final signature under examination of two like - sign leptons and di - jet , @xmath120 which is a signature well known to be rather clean ( due to the low expected sm background ) .
relevant yields are ensured by the rather large @xmath118 .
this peculiar final state being a lepton number violating process ( @xmath121 ) is only possible if the heavy neutrino is of majorana type . in this work we chose to focus on the specific signature with two positive leptons due to its larger cross section as shown in fig .
[ p1vsp2](right ) .
comparison between the parton - level cross sections of the process with resonant production of heavy majorana neutrino ( solid line ) and that with a virtual heavy majorana neutrino ( dashed line ) .
both gauge and contact interactions are considered in each case.,width=302 ] ( color online ) invariant mass distribution of the @xmath0 system at @xmath109 tev .
the signal distribution gives clearly an excess relative to the standard model expectation in the region of large invariant masses . , width=302 ] it is important to remark that the like - sign di - lepton plus di - jet signature can be realized by two distinct classes of feynman diagrams which are shown in fig . [ processes ] . in fig .
[ processes](a ) there is a @xmath122-channel exchange of a virtual heavy majorana neutrino while in fig .
[ processes](b ) the heavy neutrino is resonantly produced ( @xmath123-channel ) and its subsequent decay ; in fig . [ processes ] , each dark ( grey ) blob includes both a gauge or a contact interaction term whose feynman diagrams are those shown in fig .
[ interactions ] . the process in fig . [ processes](a ) is the collider - analogous of the neutrinoless double-@xmath124 decay ( @xmath20 ) , the well known lepton number violating ( @xmath125 ) nuclear rare decay @xcite which
, if detected , would unambiguosly verify the majorana nature of neutrinos .
the half life of the @xmath20 is currently bounded as @xmath126 yr @xcite at 90% confidence level , from the data of the @xmath127xe exo ( enriched xenon observatory)-200 experiment .
previous searches with @xmath128ge ( the gerda experiment ) @xcite and with @xmath127xe ( the kamland - zen experiment ) @xcite had established a half - life longer than @xmath129 years . in an high energy collider
a heavy majorana neutrino can be produced in resonance , fig .
[ processes](b ) , if the mass of the neutrinos is kinematically accessible @xmath130 , where @xmath131 is the energy in the parton center of mass frame . in this case
the cross section for the signature @xmath132 is approximated by @xmath133 .
the resonant production rate is dominant relative to the virtual neutrino exchange contribution .
this was demonstrated in @xcite for the gauge - only case and it is still true in the current model including also the contact interactions . this has been explicitly verified and is shown in fig .
[ final_states](b ) .
( color online ) current bounds on the plane of parameters ( @xmath2 , @xmath1 ) from cms and atlas searches of @xmath134 ( run i ) versus the expected exclusion curves at run ii from the heavy majorana neutrino signature ( @xmath135 ) . see text for further details . ]
as it is well known in the standard model the lepton number @xmath136 is strictly conserved and thus processes like those in eq . with @xmath137 are not allowed .
however within the sm there are several processes that can produce two same sign leptons in association with jets .
the following processes are considered as main backgrounds @xcite [ bgs ] @xmath138 we discuss here the main kinematic differences between the signal and the background to choose suitable cuts for optimizing the signal / background ratio . from the point of view of the leptons transverse momentum distributions in fig .
[ kinematics ] ( top - left and center - left ) , signal and background are very well separated , for the given values of the parameters ( @xmath139 gev and @xmath110 tev ) and we can reduce drastically the background applying a cut on the transverse momentum of the leading positron at 200 gev and a cut in the transverse momentum of the second - leading positron at 100 gev : [ cuts ] @xmath140 on the contrary we can see from fig . [ kinematics ] ( top - right and center - right ) that the angular distributions of the leading and second - leading leptons are quite similar between signal and background . from fig .
[ kinematics ] ( bottom - left ) regarding the signal we can see that a large fraction of the events have the two jets with a very small separation in the @xmath141 plane , @xmath142 , ( @xmath90 is the pseudorapidity and @xmath143 the azimuthal angle in the transverse plane ) .
the corresponding @xmath144 distribution is peaked at @xmath145 .
therefore , in the reconstruction process , we are likely to have merging , i.e. the two jets can be often reconstructed as a single jet .
let us also remark the fact that information about the mass of the heavy majorana neutrino can be obtained from the invariant mass distribution of the second - leading lepton and the two jets . indeed in fig .
[ kinematics ] ( bottom - right ) we show that this distribution has a very sharp peak in correspondence of the heavy majorana neutrino mass .
this is indeed expected since in the resonant production the heavy majorana neutrino @xmath77 is decaying to @xmath146 and the lepton from @xmath77 is expected to be the second - leading , while the leading lepton is the one produced in association with @xmath77 , in @xmath147 .
finally we show explicitly that the di - lepton plus di - jet signature from a heavy composite majorana neutrino can easily explain the excess observed by the cms collaboration @xcite in the @xmath0 invariant mass distribution in the interval 1.8 tev @xmath148 2.2 tev . fig .
[ fig : eejjinvmass ] shows , at generator level and for a particular point of the parameter space ( @xmath110 tev , @xmath139 gev ) , that the @xmath0 invariant mass distribution can easily accomodate an excess in the interval where it has been claimed by the cms collaboration . the fact that the excess is observed in the electron channel but not in the muon channel could be explained in our model simply by invoking a rather natural mass splitting between the excited electron ( @xmath149 ) and muon ( @xmath150 ) instead of assuming that @xmath151 .
in order to take into account the detector effects , such as efficiency and resolution in reconstructing kinematic variables , we interface the lhe output of calchep with the software delphes that simulates the response of a generic detector according to predefined configurations .
we use a cms - like parametrization . for the signal
we consider a scan of the parameter space ( @xmath2 , @xmath1 ) within the ranges @xmath152 $ ] tev with step of red @xmath153 tev and @xmath154 $ ] gev with step of @xmath155 gev . for each signal point and each background we generate @xmath156 events in order to have enough statistics to evaluate the reconstruction efficiencies ( @xmath157 , @xmath158 ) of the detector and of the cuts previously fixed ( see eq .
[ cut1 ] , [ cut2 ] ) .
then we select the events with two positive electrons , while , although our final state has two jets , we do not require the presence of two jets because , as shown in the previous section , we have merging in a great fraction of the events .
once we have the number of the selected events we evaluate the reconstruction efficiencies , then for a given luminosity @xmath136 it is possible to estimate the expected number of events for the signal ( @xmath159 ) and for the background ( @xmath160 ) and finally the statistical significance ( @xmath161 ) : @xmath162 in fig . [ significance ] top - left , top - right and bottom - left we show the contour plots of @xmath163 and @xmath164 in the parameter space ( @xmath2 , @xmath1 ) for three different values of integrated luminosity @xmath165 fb@xmath166 .
the regions below the curves are excluded .
the colored filled bands are an estimate of the statistical error . in fig .
[ significance ] bottom - right we compare the 5-sigma curves at the three integrated luminosity values . finally in fig .
[ fig : comparison ] we compare our 3-@xmath4 contour plots ( @xmath163 ) for the three different values of integrated luminosity @xmath165 fb@xmath166 of fig .
[ significance ] with the 95% confidence level exclusion bounds from two run i analyses at @xmath167 tev : atlas with 13 fb@xmath10 @xcite and cms with 19.7fb@xmath10 @xcite .
the shaded regions below the solid , dashed and dot - dashed lines are the current cms exclusion at @xmath31 tev with 19.7 fb@xmath10 of integrated luminosity ( blue ) @xcite and the atlas exclusion at @xmath31 tev with 13 fb@xmath10 ( yellow ) and the region of the parameters where the model is not applicable ( grey ) i.e. @xmath168 . such experimental exclusion regions from run i are compared with the contour plots expected from run ii , considering the process studied in this work .
the solid ( magenta ) , dashed ( red ) and dot - dashed ( green ) without shading are the projected contour maps for @xmath163 ( 3-@xmath4 ) in the parameter space ( @xmath2 , @xmath1 ) of the statistical significance for @xmath109 tev and for the follwing three values of the integrated luminosity @xmath169 fb@xmath10 .
the great potential of discovery or improving the current model limits of the @xmath0 signature from a heavy composite majorana neutrino as function of the increasing integrated luminosity is clearly evident and is the authors opinion that a dedicated experimental analysis of this channel should be undertaken within runii of the lhc .
we have performed a phenomenological study of the production of heavy composite majorana neutrinos at lhc in view of the recent observation by the cms collaboration of a 2.5-@xmath4 excess in the @xmath0 channel .
we take up the well known composite model scenario with heavy excited fermions ( quarks and leptons ) .
these states have been searched for in a number of dedicated direct searches . in this study
we reconsider the hypothesis that the excited neutrino is of majorana type .
this hypothesis had been considered in @xcite where a model based only on gauge interactions was used to describe the production rates and decay of the composite majorana neutrino .
however a composite model scenario quite generically predicts , in addition to the presence of excited states , also four fermion contact interactions .
we have included the contribution of contact interactions in the phenomenology of the excited majorana neutrino , hitherto not considered in the literature .
the model is implemented in calchep which allows quite extensive simulations at the generator level .
the contact interaction mechanism turns out to be dominant in the resonant production of the heavy majorana neutrino .
we have performed a kinematic study of the same - signed di - lepton plus di - jet signature analyzing in detail both signal and background in order to optimize the signal over background ratio .
we find that the invariant mass distribution of the system made up by the second - leading lepton and the two jets is highly correlated to the heavy majorana neutrino mass , see fig .
[ kinematics ] ( bottom - right ) .
a fast simulation of the detector effects and efficiencies in the reconstruction process is performed using the delphes @xcite package based on a cms - like configuration .
we scanned the two dimensional parameter space @xmath170 for some benchmark values and computed the statistical significance .
we provide the contour plots of the statistical significance @xmath161 at 3- and 5-sigma ( see fig .
[ significance ] ) .
we find for instance that with @xmath171 tev the lhc can reach a 3-sigma sensitivity for masses up to @xmath172 gev respectively for an integrated luminosity of @xmath169 fb@xmath10 .
finally we compared ( see fig .
[ fig : comparison ] ) such expected exclusion bounds in the parameter space ( @xmath96 ) with the experimental data of run i c.f .
such analyses have investigated signatures of excited electrons and muons ( @xmath173 ) being produced by contact interactions ( @xmath174 ) and decaying via @xmath175 . strictly speaking such analyses access the parameters spaces ( @xmath176 ) and ( @xmath177 ) that are in principle different from the one presented here ( @xmath178 ) .
however all excited states masses can be assumed to be approximately degenerate , at least at a first order approximation . under this hypothesis
@xmath179 the @xmath0 signature discussed in this work provides contour maps that can be considered on the same parameter space of the other analyses based on @xmath180 @xcite .
this comparison shows that the @xmath0 signature from a heavy composite majorana neutrino has the potential to improve sensibly the current constraints on the composite scenario . before concluding we would like to comment briefly about another anomaly reported by the atlas collabortion and on recent results from atlas and cms collaborations from runii , and how our model could interpret them . in a search @xcite for high - mass di - boson resonances with boson - tagged jets at @xmath181
tev the atlas collaboration has reported an excess at around 2 tev with a global significance of 2.5 standard deviations ( note however that the same search performed by the cms collaboration did not observe a similar excess @xcite ) .
our model contains fermion resonances ( excited quarks and leptons ) which do not couple directly to a pair of gauge bosons .
indeed a fermion can not decay to a pair of gauge bosons by angular momentum conservation . on general grounds our fermion resonances could produce final states with a pair of gauge bosons but
these would always be accompained by other objects such as leptons and jets ( sm fermions ) . as an example one might think to pair produce the excited neutrinos @xmath182 with the excited neutrinos decaying leptonically @xmath183 .
one obtains a final signature of @xmath184 which is different from the one considered in the atlas search for high mass di - boson resonances consisting of only gauge boson pairs ( @xmath185 or @xmath186 ) .
however one might imagine to pair produce the charged excited fermions , for instance @xmath149 and/or @xmath187 , almost at threshold ( if they are very massive ) .
such pair of heavy fermions could in principle form a 1s bound state ( via the known coulomb and/or color interaction ) which in turn could decay to a pair of intermediate vector boson given the high mass of the hypothetical heavy fermions @xcite
. therefore our model has _ in principle _ the potential to reproduce an excess in the di - boson signal .
the estimate of such effects is certainly very interesting and it would surely be worth further investigation ( one needs for instance to understand whether such bound states could form at all in the first place ) .
however a quantitative analysis goes beyond the purview of the present work and we postpone it to a future study .
very recently the cms and atlas collaborations have released the first results of run ii of the lhc at @xmath188 tev , with respectively 42 pb@xmath10 and 80 pb@xmath10 of integrated luminosity @xcite , reporting about a search for hadronic resonances in the di - jet channel and showing @xmath189 1-sigma excess(es ) at an invariant mass of about 5 tev in the measured di - jet invariant mass distribution .
such excess(es ) , if confirmed by further data and statistics , could in principle signal the first hadronic ( excited quark ) resonance level in a composite model scenario beyond the 2 tev @xmath0 anomaly . indeed
the analysis in @xcite excludes excited quarks masses from around 3 tev at 95% c.l ( if @xmath33 ) while in @xcite excited lepton masses are excluded at 95% c.l . from @xmath190 tev
( again assuming @xmath33 ) . these experimental bounds would seem to preclude the possibility of an excited fermion bound state with regard to the explanation of the di - boson anomaly at 2 tev ( see above ) since a @xmath187 ( @xmath191 bound state would need to have to a mass of at least 6 tev ( 5 tev ) .
however the quoted bounds for the excited fermions are for @xmath33 which is the limit of validity of the effective composite model . for values of @xmath2
higher than @xmath1 the actual bounds on @xmath1 are lower ( see for example fig . [
fig : comparison ] ) . in such regions of the parameter space
the 2 tev di - boson anomaly could still be explained ( in principle ) by a @xmath192 ( or @xmath193 ) bound state with a mass @xmath194 tev . as a last remark concerning the @xmath0 anomaly we would like to comment on the fact that ( @xmath40 ) the same excess is not observed in the @xmath23 channel and ( @xmath43 ) the observed charge asymmetry of the like sign di - leptons @xcite .
the absence of the excess in the @xmath23 channel could be explained by our model simply assuming that the excited muon state ( @xmath150 ) is somewhat heavier than the @xmath149 and
so it would be observable only at higher energies .
the observed @xmath0 excess consists indeed of 14 events of which 13 are opposite sign ( os ) and only one is same sign ( ss ) .
it must be said that our mirror type composite model with one majorana neutrino will produce the same yield of os and ss events .
such feature could be explained within our composite model assuming the existence of an additional majorana @xmath46 state with a slightly different mass .
indeed it has been shown , albeit within a different ( seesaw ) model @xcite , that the interference between the contributions of two different majorana states could depress the ss yield relative to the os .
the interference effect could also explain the absence of a peak in the observed invariant mass distribution of the second leading electron and the two jets ( see our fig .
[ kinematics ] bottom right ) . in view of this
it could be worthwhile either to upgrade the calchep implementation of our mirror model to include other majorana states or alternatively to reconsider the homo - doublet model with @xmath54-@xmath49 mixing . in order to address quantitatively this issue
we would need to build a new model ( with more than one majorana neutrino state ) in the calchep generator .
this goes beyond the scope of the present work and will be addressed in a future study .
in summary the results presented in this work are quite encouraging and certainly endorse the interest and feasibility of a full fledged analysis of the experimental data of the upcoming lhc run ii for a search of the heavy composite majorana neutrino , within a mirror type model , in proton proton collisions at @xmath109 tev .
66ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty \doibase http://dx.doi.org/10.1016/j.physletb.2012.02.044 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.physletb.2012.02.064 [ * * , ( ) ] link:\doibase 10.1140/epjc / s10052 - 014 - 3149-z [ * * , ( ) ] , @noop _ _ , ( , , ) @noop ( ) , link:\doibase 10.1140/epjc / s10052 - 014 - 3107 - 9 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.91.015018 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.90.115010 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.115.181803 [ * * , ( ) ] @noop ( ) , @noop ( ) , link:\doibase 10.1103/physrevd.91.035019 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.91.011702 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.91.015011 [ * * , ( ) ] , link:\doibase 10.1007/jhep02(2015)142 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.91.035006 [ * * , ( ) ] , @noop ( ) , @noop ( ) , \doibase http://dx.doi.org/10.1016/j.physletb.2015.06.077 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.50.1427 [ * * , ( ) ] link:\doibase 10.1103/physrevd.47.961 [ * * , ( ) ] link:\doibase 10.1103/physrevd.50.3195 [ * * , ( ) ] link:\doibase 10.1103/physrevd.62.015013 [ * * , ( ) ] link:\doibase 10.1103/physrevd.62.075004 [ * * , ( ) ] link:\doibase 10.1103/physrevd.65.035005 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.171804 [ * * , ( ) ] http://stacks.iop.org/1126-6708/2009/i=05/a=030 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.081801 [ * * , ( ) ] http://stacks.iop.org/1367-2630/17/i=7/a=075019 [ * * , ( ) ] link:\doibase 10.1103/physrevd.88.113001 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.physletb.2014.06.058 [ * * , ( ) ] @noop ( ) , http://stacks.iop.org/1367-2630/15/i=9/a=093011 [ * * , ( ) ] @noop ( ) @noop `` , '' ( ) link:\doibase 10.1103/physrevd.90.035001 [ * * , ( ) ] @noop \doibase http://dx.doi.org/10.1016/j.cpc.2013.01.014 [ * * , ( ) ] link:\doibase 10.1007/jhep02(2014)057 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.15.480 [ * * , ( ) ] link:\doibase 10.1103/physrevd.22.184 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.50.811 [ * * , ( ) ] link:\doibase 10.1016/0370 - 2693(84)91850 - 1 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , , ) pp
. link:\doibase 10.1143/ptp.94.1097 [ * * , ( ) ] , link:\doibase 10.1088/1674 - 1137/38/9/090001 [ * * , ( ) ] @noop ( ) , link:\doibase 10.1016/0370 - 2693(84)90649-x [ * * , ( ) ] link:\doibase 10.1103/physrevd.85.095018 [ * * , ( ) ] , link:\doibase 10.1016/j.cpc.2009.02.018 [ * * , ( ) ] , @noop `` , '' , link:\doibase 10.1007/jhep10(2014)094 [ * * , ( ) , 10.1007/jhep10(2014)094 ] http://stacks.iop.org/0295-5075/105/i=1/a=11001 [ * * , ( ) ] @noop * * , ( ) , link:\doibase 10.1143/ptps.83.1 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.80.481 [ * * , ( ) ] link:\doibase 10.1038/nature13432 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.111.122503 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.062502 [ * * , ( ) ] @noop ( ) , link:\doibase 10.1007/jhep08(2014)173 [ * * , ( ) , 10.1007/jhep08(2014)173 ] link:\doibase 10.1103/physrevd.72.015005 [ * * , ( ) ] link:\doibase 10.1103/physrevd.81.115001 [ * * , ( ) ] https://cds.cern.ch/record/2048099[__ ] , ( , , ) http://cds.cern.ch/record/2048113[__ ] , ( , , ) | motivated by the recent observation of an excess in the @xmath0 channel by the cms collaboration we investigate the search for heavy majorana neutrinos stemming from a composite model scenario at the upcoming lhc run ii at a center of mass energy of 13 tev .
while previous studies of the composite majorana neutrino were focussed on gauge interactions via magnetic type transition coupling between ordinary and heavy fermions ( with mass @xmath1 ) here we complement the composite model with contact interactions at the energy scale @xmath2 and we find that the production cross sections are dominated by such contact interactions by roughly two / three orders of magnitude .
this mechanism provides therefore very interesting rates at the prospected luminosities .
we study the same sign di - lepton and di - jet signature ( @xmath3 ) and we discuss how it can account for the excess in the @xmath0 invariant mass distribution , and perform a fast detector simulation based on delphes .
we compute 3@xmath4 and 5@xmath4 contour plots of the statistical significance in the parameter space ( @xmath5 ) .
we find that the potentially excluded regions at @xmath6 tev are quite larger than those excluded so far at run i considering searches with other signatures . |
consider a smooth 4-manifold @xmath2 and a surface @xmath3 .
let @xmath4 be a ( possibly achiral ) smooth lefschetz fibration with singular values set @xmath5 and regular fiber @xmath6 , the compact connected orientable surface of genus @xmath7 with @xmath8 boundary components .
let @xmath9 be another surface .
we assume that @xmath2 , @xmath3 and @xmath9 are compact , connected and oriented with ( possibly empty ) boundary .
we say that a smooth map @xmath10 with regular values set @xmath11 is @xmath12-regular if @xmath13 and @xmath14 .
in other words , @xmath15 is @xmath12-regular if and only if @xmath15 and @xmath16 are transverse to @xmath12 .
if @xmath15 is @xmath12-regular then @xmath17 is a 4-manifold and the map @xmath18 given by @xmath19 is a lefschetz fibration .
moreover @xmath20 , @xmath21 , is a fiber preserving map which sends each fiber of @xmath22 diffeomorphically onto a fiber of @xmath12 , so the regular fiber of @xmath22 can be identified with @xmath23 .
we get the following commutative diagram we say that @xmath24 is the _ pullback _ of @xmath12 by @xmath15 and we make use of the notation @xmath25 . two lefschetz fibrations @xmath26 and @xmath27 are said _ equivalent _ if there are orientation - preserving diffeomorphisms @xmath28 and @xmath29 such that @xmath30 .
the equivalence class of @xmath12 will be indicated by @xmath31 $ ] .
we say that a lefschetz fibration @xmath12 is _ allowable _ if the vanishing cycles of @xmath12 with respect to a ( and hence to any ) hurwitz system for @xmath32 are homologically essential in @xmath23 .
we consider only allowable lefschetz fibrations , if not differently stated .
given @xmath12 we define the set @xmath33\}$ ] where @xmath15 runs over the @xmath12-regular maps @xmath10 and @xmath9 runs over the compact , connected , oriented surfaces .
analogously , we define the set @xmath34\ } \subset l(f)$ ] where @xmath15 runs over the @xmath12-regular orientation - preserving immersions @xmath10 , with @xmath9 as above .
[ univ / def ] a lefschetz fibration @xmath35 with regular fiber @xmath23 is said universal ( resp .
strongly universal ) if every lefschetz fibration over a surface with non - empty boundary and with regular fiber diffeomorphic to @xmath23 belongs to a class of @xmath36 ( resp .
@xmath37 . in other words
@xmath38 is universal ( resp . strongly universal ) if and only if any lefschetz fibration over a bounded surface and with the same fiber can be obtained as the pullback of @xmath38 by a @xmath38-regular map ( resp .
immersion ) . note that this notion of universality is analogous to that in the theory of vector bundles @xcite .
we denote by @xmath39 the mapping class group of @xmath40 whose elements are the isotopy classes of orientation - preserving diffeomorphisms of @xmath40 which keep the boundary fixed pointwise ( assuming isotopy through such diffeomorphisms ) .
it is well - known that for a lefschetz fibration @xmath4 with regular fiber @xmath40 the monodromy of a meridian is an element of @xmath41 which can be represented by the oriented boundary of an embedded disk in @xmath3 which intersects @xmath42 in a single interior point , cf . definition [ immers - merid / def ] .
] of @xmath32 is a dehn twist @xmath43 .
if @xmath3 is not simply connected , the monodromy of an element of @xmath44 which is not a product of meridians is not necessarily the identity on @xmath45 , and so it induces a permutation on the set of boundary components of @xmath40 .
we will denote by @xmath46 the permutation group of this set .
these considerations allow us to define three kind of monodromies .
let @xmath47 be the smallest normal subgroup of @xmath44 which contains all the meridians of @xmath32 . the _ lefschetz monodromy _ of @xmath12 is the group homomorphism @xmath48 which sends a meridian of @xmath32 to the corresponding dehn twist in the standard way @xcite .
let @xmath49 be the extended mapping class group of @xmath40 , namely the group of all isotopy classes of orientation - preserving self - diffeomorphisms of @xmath40 .
the _ bundle monodromy _
@xmath50 is the monodromy of the locally trivial bundle @xmath51 .
we consider also the natural homomorphism @xmath52 which sends an isotopy class to the permutation induced on the set of boundary components .
the composition @xmath53 passes to the quotient @xmath54 and gives a homomorphism @xmath55 which we call the _ permutation monodromy _ of @xmath12
. let @xmath56 be the set of equivalence classes of homologically essential simple closed curves in @xmath57 up to orientation - preserving homeomorphisms of @xmath40 .
it is well - known that @xmath56 is finite .
moreover @xmath58 for @xmath59 and @xmath60 ( * ? ? ?
* chapter 12 ) .
now we state the main results of the paper . in the following proposition
we characterize the universal and strongly universal lefschetz fibrations .
[ main / thm ] a lefschetz fibration @xmath35 with regular fiber @xmath40 is universal ( resp . strongly universal )
if and only if the following two conditions @xmath61 and @xmath62 ( resp . @xmath61 and @xmath63 are satisfied : * @xmath64 and @xmath65 are surjective ; * any class of @xmath56 can be represented by a vanishing cycle of @xmath38 ; * any class of @xmath56 contains at least two vanishing cycles of @xmath38 which correspond to singular points of opposite signs . in particular ,
if @xmath60 and @xmath64 is surjective then @xmath38 is universal .
if in addition @xmath38 admits a pair of opposite singular points , then @xmath38 is strongly universal . as a remarkable simple consequence we have that ( strongly ) universal lefschetz fibrations actually exist for any regular fiber @xmath40 .
moreover , the surjectivity of @xmath66 implies @xmath67the minimum number of generators of @xmath68 , and this inequality is sharp .
so we can assume that the base surface @xmath3 of a universal lefschetz fibration is the disk for @xmath69 , the annulus for @xmath70 , and such that @xmath71 for @xmath72 .
consider a knot @xmath73 and let @xmath74 be the oriented 4-manifold obtained from @xmath75 by the addition of a 2-handle along @xmath76 with framing @xmath77 . in the following theorem
we construct very special strongly universal lefschetz fibrations when the fiber is the torus or @xmath78 with @xmath79 .
[ strong - univ / thm ] there is a strongly universal lefschetz fibration @xmath80 with fiber @xmath40 and with @xmath81 , + @xmath82 , and + @xmath83 , where @xmath84 and @xmath85 denote respectively the trivial and the figure eight knots in @xmath86 .
[ paral / cor ] let @xmath87 be a lefschetz fibration with fiber of genus one .
suppose that no vanishing cycle of @xmath12 disconnects the regular fiber with respect to some ( and hence to any ) hurwitz system
. then @xmath2 immerses in @xmath88 and so is parallelizable . by means of theorem [ strong - univ / thm ] we are able to give a new elementary proof of the following corollary .
this was known since the work of phillips @xcite about submersions of open manifolds because there exists a bundle monomorphism @xmath89 for any oriented 4-manifold @xmath2 which is homotopy equivalent to a cw - complex of dimension two ( obtained by means of the classifying map to the complex universal vector bundle @xcite ) .
[ immers / cor ] any compact oriented 4-dimensional 2-handlebody - dimensional @xmath90-handlebody is a smooth @xmath77-manifold built up with handles of index @xmath91 . ]
admits an orientation - preserving immersion in @xmath92 .
universal maps in the context of lefschetz fibrations over closed surfaces can be constructed in a different way .
this generalization will be done in a forthcoming paper .
the paper consists of three other sections . in the next one
we review some basic material on lefschetz fibrations needed in the paper .
section [ proof / sec ] is dedicated to the proofs of our results , and in section [ remarks / sec ] we give some final remarks and a comment on positive lefschetz fibrations on stein compact domains of dimension four . throughout the paper we assume manifolds ( with boundary ) to be compact , oriented and connected if not differently stated .
we will work in the @xmath93 category .
let @xmath2 be a 4-manifold ( possibly with boundary and corners ) and let @xmath3 be a surface .
[ lf / def ] a lefschetz fibration @xmath4 is a smooth map which satisfies the following three conditions : 1 .
the singular set @xmath94 is finite and is mapped injectively onto the singular values set @xmath95 ; 2 .
the restriction @xmath51 is a locally trivial bundle with fiber a surface @xmath23 ( the regular fiber of @xmath12 ) ; 3 . around each singular point @xmath96
, @xmath12 can be locally expressed as the complex map @xmath97 for suitable chosen smooth local complex coordinates . if such local coordinates are orientation - preserving ( resp . reversing ) , then @xmath98 is said a _ positive ( resp .
negative ) singular point _ , and @xmath99 is said a _ positive ( resp .
negative ) singular value_. observe that this positivity ( resp .
negativity ) notion does not depend on the orientation of @xmath3 .
obviously , around a negative singular point there are orientation - preserving local complex coordinates such that @xmath100 .
most authors add the adjective ` achiral ' in presence of negative singular points .
we prefer to simplify the terminology and so we do not follow this convention .
the orientations of @xmath2 and of @xmath3 induce an orientation on @xmath23 such that the locally trivial bundle associated to @xmath12 is oriented .
we will always consider @xmath23 with this canonical orientation .
let @xmath40 be the orientable surface of genus @xmath101 with @xmath102 boundary components .
a lefschetz fibration @xmath4 with regular fiber @xmath103 is characterized by the _ lefschetz monodromy homomorphism _ @xmath104 , which sends meridians of @xmath32 to dehn twists , and by the _ bundle monodromy homomorphism _ @xmath105 which is the monodromy of the bundle associated to @xmath12 .
sometimes we use the term ` monodromy ' by leaving the precise meaning of it to the context .
it is well - known that the monodromy of a counterclockwise meridian of a positive ( resp .
negative ) singular value is a right - handed ( resp .
left - handed ) dehn twist around a curve in @xmath23 @xcite .
such a curve is said to be a _ vanishing cycle _ for @xmath12 .
we recall the following definition .
[ hurwitz / def ] a hurwitz system for a cardinality @xmath77 subset @xmath106 is a sequence @xmath107 of meridians of @xmath42 which freely generate @xmath108 and such that the product @xmath109 is the homotopy class of the oriented boundary of @xmath110 , where @xmath111 is a disk such that @xmath112 and @xmath113 .
if a hurwitz system @xmath107 for @xmath114 is given , the set @xmath115 can be numbered accordingly so that @xmath116 is a meridian of @xmath117 .
it is determined a sequence of signed vanishing cycles @xmath118 ( the _ monodromy sequence _ of @xmath12 ) , where @xmath119 corresponds to the dehn twist @xmath120 and the sign of @xmath121 equals that of @xmath117 as a singular value of @xmath12 .
clearly , @xmath121 is defined up to isotopy for all @xmath122 .
sometimes the plus signs are understood .
the fact that the @xmath121 s are all homologically ( or homotopically ) essential in @xmath23 does not depend on the actual hurwitz system , and so this is a property of the lefschetz fibration @xmath12 .
the monodromy sequence of @xmath87 determines a handlebody decomposition of the total space as @xmath123 where @xmath124 is a trivialization of the bundle associated to @xmath12 over a subdisk contained in @xmath125 and the 2-handle @xmath126 is attached to @xmath124 along the vanishing cycle @xmath127 for a suitable subset @xmath128 cyclically ordered in the counterclockwise direction @xcite .
the framing of @xmath129 with respect to the fiber @xmath130 is @xmath131 where @xmath132 is the sign of the singular point @xmath117 .
note that @xmath124 can be decomposed as the union of a 0-handle , some 1-handles , and also a 2-handle in case @xmath133 starting from a handlebody decomposition of @xmath23 and making the product with the 2-dimensional 0-handle @xmath134 . in this paper
we consider only the so called _ relatively minimal _ lefschetz fibrations , namely those without homotopically trivial vanishing cycles .
then in our situation the monodromy sequence can be expressed also by a sequence of dehn twists @xmath135 , where @xmath136 is assumed to be right - handed .
let @xmath137 be the homomorphism such that @xmath138 ) = [ \phi]$ ] for all @xmath139 \in { \mathop{\cal m}\nolimits}_{g , b}$ ] .
we have the exact sequence @xmath140 where @xmath141 is the boundary permutation homomorphism defined in the introduction .
the monodromy homomorphisms @xmath142 and @xmath143 of a lefschetz fibration @xmath12 satisfy @xmath144 . for a finite subset @xmath106
we indicate by @xmath145 the normal subgroup generated by the meridians of @xmath42 .
given @xmath3 , @xmath42 and two homomorphisms @xmath146 and @xmath147 such that @xmath148 sends meridians to dehn twists and @xmath149 , there exists a lefschetz fibration @xmath4 with regular fiber @xmath40 such that @xmath150 , @xmath151 and @xmath152 . moreover , such @xmath12 is unique up to equivalence by our relative minimality assumption , unless @xmath3 is closed and the fiber is a sphere or a torus ( because in such cases the diffeomorphisms group of the fiber is not simply connected @xcite ) . in particular ,
if @xmath3 has boundary the 4-manifold @xmath2 is determined up to orientation - preserving diffeomorphisms . in @xcite
we give a very explicit construction of @xmath12 starting from the monodromy sequence . if @xmath10 is @xmath12-regular with respect to a lefschetz fibration @xmath4 then the pullback @xmath25 satisfies @xmath153 , @xmath154 and @xmath155 , where @xmath156 is the homomorphism induced by the restriction @xmath157 .
the base points are understood and are chosen so that @xmath158 with @xmath159 in the domain and @xmath160 in the codomain .
the @xmath12-regularity of @xmath15 implies that @xmath161 .
let @xmath162 and @xmath163 .
then @xmath15 is a local diffeomorphism around @xmath164 .
it is immediate that the sign of @xmath164 as a singular value of @xmath22 is given by that of @xmath165 multiplied by the local degree of @xmath15 at @xmath164 , in other words @xmath166 . in order to prove theorem [ strong - univ / thm ] we recall the definition of _ stabilizations _ ( the reader is referred to @xcite or @xcite for details ) . given a lefschetz fibration
@xmath87 whose regular fiber @xmath23 has non - empty boundary , we can construct a new lefschetz fibration @xmath167 by an operation called stabilization which is depicted in figure [ stabil / fig ] .
the new regular fiber @xmath168 is obtained by attaching an orientable 1-handle @xmath169 to @xmath23 , and the new monodromy sequence is given by the addition to the old one of a signed vanishing cycle @xmath170 which crosses @xmath169 geometrically once .
the inverse operation is called _ destabilization _ and can be applied if there is a properly embedded arc @xmath171 in the regular fiber @xmath172 of @xmath173 which meets a single vanishing cycle @xmath174 , and it does so geometrically once .
the arc @xmath171 is the cocore of a 1-handle of @xmath172 .
let @xmath23 be @xmath172 cut open along @xmath171 , and let the new monodromy sequence be that of @xmath173 with @xmath170 removed ( no matter whichever is the sign ) . in terms of handlebody decompositions , stabilizations ( resp .
destabilizations ) correspond to the addition ( resp .
deletion ) of a cancelling pair of 1 and 2-handles , hence @xmath175 .
we end this section with the following straightforward propositions , needed in the proof of our main results .
let @xmath4 be a lefschetz fibration and let @xmath9 be a surface .
if @xmath176 , @xmath177 $ ] , is a homotopy through @xmath12-regular maps then @xmath178 . if @xmath10 is an orientation - preserving immersion ( resp .
embedding ) then the fibered map @xmath20 associated to the pullback @xmath179 is also an orientation - preserving immersion ( resp . embedding )
we first prove the following lemma .
[ van - cic / lem ] suppose that @xmath35 satisfies conditions @xmath61 and @xmath62 ( resp . @xmath61 and @xmath63 of proposition [ main / thm ] .
then each class of @xmath56 can be represented by a vanishing cycle ( resp . a vanishing cycle of prescribed sign ) in any monodromy sequence of @xmath38 .
let @xmath107 be a hurwitz system for @xmath180 and let @xmath181 .
there is a meridian @xmath182 of @xmath183 such that @xmath184 is a dehn twist ( resp . a dehn twist of prescribed sign ) around a curve @xmath185 .
it is well - known that @xmath182 is conjugate to some @xmath116 , hence @xmath186 for some @xmath187 .
let @xmath121 be the vanishing cycle of @xmath188 .
put @xmath189 , @xmath190 and @xmath191 .
we get @xmath192 ( because the standard right to left composition rule of maps differs from that in the fundamental group ) .
there are two cases depending on whether @xmath188 is or not the identity .
if @xmath188 is the identity then @xmath193 is also the identity .
it follows that @xmath174 and @xmath121 are boundary parallel ( by the relatively minimal assumption they can not be homotopically trivial ) , hence @xmath194 . if @xmath188 is not the identity then @xmath195 @xcite .
we need also the following definition which gives a generalization of the notion of meridian .
[ immers - merid / def ] an immersed meridian for a finite subset @xmath106 is an element of @xmath196 which can be represented by the oriented boundary of an immersed disk @xmath197 such that @xmath198 .
the immersed meridians of @xmath42 are precisely the conjugates in @xmath199 of the meridians .
we consider first the case of universal lefschetz fibrations . _
` if ' part .
_ suppose that @xmath38 satisfies conditions @xmath61 and @xmath62 of the statement .
consider a lefschetz fibration @xmath200 with regular fiber @xmath40 , where @xmath9 is an oriented surface with non - empty boundary .
we are going to show that @xmath201 for some @xmath38-regular map @xmath10 .
without loss of generality we can assume @xmath202 .
there is a handlebody decomposition of @xmath9 with only one 0-handle @xmath203 and @xmath204 1-handles @xmath205 , so @xmath206 .
we can assume that @xmath207 .
fix base points @xmath208 and @xmath209 and let @xmath210 be a hurwitz system for @xmath211 .
fix also a set of free generators @xmath212 for @xmath213 with @xmath214 a hurwitz system for @xmath215 .
we assume that the @xmath216 s are represented by meridians contained in @xmath203 and that @xmath217 is represented by an embedded loop ( still denoted by @xmath217 ) which meets the 1-handle @xmath205 geometrically once and does not meet any other @xmath218 for @xmath219 as in figure [ surface / fig ] . in this figure
the 1-handles @xmath205 s are contained in the white lower box .
we assume also that @xmath220 for @xmath221 .
s and @xmath217 s . ]
there are disks @xmath222 as those depicted in dark grey in the same figure [ surface / fig ] such that @xmath223 represents @xmath216 as a loop and @xmath224 for all @xmath221 .
there is also a disk @xmath225 which is a neighborhood of @xmath159 such that @xmath226 and @xmath227 $ ] for all @xmath122
. then @xmath228 is diffeomorphic to @xmath134 ( up to smoothing the corners ) .
it follows that @xmath9 is diffeomorphic to the surface @xmath229 obtained from @xmath110 by the addition of orientable 1-handles @xmath230 where @xmath231 has attaching sphere the endpoints of the arc @xmath232 for any @xmath233 .
consider the dehn twists @xmath234 and @xmath235 around respectively the vanishing cycles @xmath121 ( for @xmath38 ) and @xmath236 ( for @xmath12 ) , where @xmath237 ( @xmath188 and @xmath238 are assumed to be right - handed ) . by lemma [ van - cic / lem
] @xmath236 is equivalent in @xmath56 to some @xmath239 .
it follows that @xmath240 for some @xmath241 which sends @xmath236 to @xmath239 @xcite ( note that this composition is well - defined in @xmath39 as the isotopy class of @xmath242 where @xmath243 and @xmath244 are representatives of @xmath245 and @xmath246 respectively ) .
since @xmath64 and @xmath66 are surjective , @xmath247 is also surjective and so there is @xmath248 such that @xmath249 .
it follows that @xmath250 where @xmath251 is an immersed meridian of @xmath252 .
so @xmath253 can be represented by the oriented boundary of an immersed disk @xmath254 which intersects @xmath183 only at @xmath252 . the @xmath255 s can be chosen so that for a suitable embedded disk @xmath256 which is a neighborhood of @xmath160 , we have @xmath257 and @xmath258 for all @xmath259 .
now take @xmath260 such that @xmath261 for @xmath262 .
we represent @xmath263 by a transversely immersed loop ( still denoted by @xmath263 ) in @xmath264 such that @xmath265 $ ] and @xmath266 for @xmath221 .
consider the moves @xmath267 , @xmath268 and @xmath269 of figure [ permutation / fig ] , where @xmath267 acts on pairs @xmath270 , @xmath268 on pairs @xmath271 , and @xmath269 on pairs @xmath272 .
these moves allow us to make the @xmath255 s and the loops @xmath263 s intersect @xmath273 in the same order as the @xmath274 s and the @xmath217 s do with @xmath275 .
take a map @xmath276 which sends @xmath275 to @xmath273 diffeomorphically , immerses @xmath274 onto @xmath255 by preserving the orientation , and immerses the 1-handle @xmath231 onto a regular neighborhood of @xmath277 for all @xmath122 .
assume also that @xmath278 and @xmath279 .
it follows that @xmath280 is a @xmath38-regular immersion and that the homomorphism @xmath281 induced by the restriction @xmath282 satisfies @xmath283 and @xmath284 for all @xmath285 .
now fix identifications @xmath286 \times [ -1 , 1]$ ] for @xmath285 , such that the ordinate is 1 along the arc @xmath287 and such that the singular value @xmath288 has coordinates @xmath289 .
let @xmath290 be defined by the identity outside @xmath291 and by the map of @xmath292 to itself given by @xmath293 up to the above identification .
so @xmath294 shrinks a proper arc of @xmath292 to a point , preserves the orientation above this arc and reverses the orientation below it , as depicted in figure [ reverse / fig ] .
moreover , @xmath295 and the homomorphism induced by the restriction @xmath296 satisfies @xmath297 , @xmath298 for @xmath299 , and @xmath300 for all @xmath301 . on @xmath110 . ]
let @xmath302 be the set of those @xmath122 such that @xmath303 and put @xmath304 ( with @xmath305 if @xmath306 is empty ) .
then @xmath15 is @xmath38-regular , @xmath307 , @xmath308 and @xmath309 where @xmath310 is induced by the restriction of @xmath15 .
it follows that @xmath201 . _
` only if ' part .
_ let @xmath35 be universal with regular fiber @xmath40 .
consider a lefschetz fibration @xmath200 with the same regular fiber and which satisfies the conditions @xmath61 and @xmath62 .
there is a @xmath38-regular map @xmath10 such that @xmath311 .
then @xmath22 satisfies the conditions @xmath61 and @xmath62 of the statement .
since @xmath312 and @xmath313 we obtain that @xmath64 and @xmath65 are surjective .
consider now a class @xmath181 .
we can find a meridian @xmath314 such that @xmath315 is a dehn twist around a curve @xmath185 .
since @xmath15 is @xmath38-regular , @xmath316 is an immersed meridian of @xmath183 , hence @xmath317 for a meridian @xmath182 of @xmath183 and for some @xmath318 .
it follows that @xmath319 satisfies @xmath320 and so @xmath184 is a dehn twist around @xmath321 .
then @xmath322 can be represented by a vanishing cycle of @xmath38 .
the case of strongly lefschetz fibrations can be handled similarly by tracing the same line of the previous proof .
we just give an idea of the ` if ' part : if conditions @xmath61 and @xmath323 are satisfied then the @xmath239 s in the proof of the first part can be chosen so that @xmath324 . then the set @xmath306 defined above is empty and so @xmath15 is an orientation - preserving @xmath38-regular immersion such that @xmath325 .
finally , the last part of the proposition follows since @xmath326 for @xmath60 .
we consider three cases , depending on the values of @xmath7 and of @xmath8 .
_ case i : @xmath327 and @xmath328 .
_ consider the lefschetz fibration @xmath329 with regular fiber @xmath330 and with monodromy sequence given by the @xmath331 signed vanishing cycles @xmath332 , @xmath333 , @xmath334 depicted in figure [ lf / fig ] , where a hurwitz system is understood . in this figure
the surface @xmath330 is embedded in @xmath335 as part of the boundary of a standard genus @xmath7 handlebody . for @xmath327 . ] by a theorem of wajnryb @xcite @xmath336 is generated by the @xmath331 dehn twists @xmath337 , @xmath253 and @xmath188 around the curves @xmath117 , @xmath338 and @xmath121 respectively .
it follows that @xmath339 is surjective and so @xmath340 is strongly universal by proposition [ main / thm ] .
now we analyze the 4-manifold @xmath341 . in figure [ lefschetz2/fig ]
we pass from @xmath342 to @xmath343 in a two step destabilization process ( first destabilize @xmath117 by the arc @xmath344 and then destabilize @xmath345 by the arc @xmath346 ) .
this operation can be done whenever @xmath347 , and so by induction we can assume that @xmath348 .
in other words @xmath349 for @xmath350 . for @xmath351 . ] in figure [ lefschetz3/fig ]
we give some more destabilizations ( first of @xmath352 along @xmath353 and then of @xmath354 along @xmath171 ) , and finally we get the lefschetz fibration depicted in the left lower part of the same figure .
this has fiber @xmath355 and three boundary parallel vanishing cycles ( two negative and one positive ) . . ]
so a kirby diagram for @xmath341 is that depicted in figure [ kirby / fig ] , which by a straightforward kirby calculus argument can be recognized to be @xmath356 ( slide the outermost 2-handle over that with framing @xmath357 so that the latter cancels and by another simple sliding and deletion we get the picture for @xmath356 in the right side of figure [ kirby / fig ] ) . for @xmath327 . ] _ case ii : @xmath358 . _
consider the lefschetz fibration @xmath359 with regular fiber @xmath360 and with monodromy sequence @xmath361 depicted in figure [ torus / fig ] . by @xcite @xmath362 is generated by the two dehn twists @xmath363 and @xmath364 around the curves @xmath165 and @xmath8 respectively and so proposition [ main / thm ] implies that @xmath365 is strongly universal . .
] by a double destabilization we get a lefschetz fibration with regular fiber @xmath134 and without singular values , hence @xmath366 is diffeomorphic to @xmath367 .
_ case iii : @xmath368 .
_ let @xmath369 be the lefschetz fibration with fiber @xmath370 and with monodromy sequence @xmath361 depicted in figure [ torus - mon / fig ] .
then @xmath371 is surjective and so @xmath372 is strongly universal by proposition [ main / thm ] . .
] consider @xmath373 as the fiber of @xmath365 over a point of @xmath374 .
so @xmath375 is a knot in @xmath86 .
moreover , by pushing off @xmath76 along @xmath360 we get the framing zero ( in terms of linking number ) , since @xmath360 is a seifert surface for @xmath76 . therefore the addition of a 2-handle to @xmath75 along @xmath76 with framing zero produces @xmath376 @xcite . in figure [ kirby - m / fig ]
is depicted a kirby diagram for @xmath376 .
this and the next three figures are referred to the blackboard framing , namely that given by a push off along the blackboard plus the extra full twists specified by the number near the knot
. note that in figure [ kirby - m / fig ] the blackboard framing coincides with that of the fiber @xmath360 . .
] now we apply the moves @xmath377 and @xmath378 of figures [ trick / fig ] and [ trick2/fig ] respectively , where the thick arcs with framing zero belong to the same knot ( which is assumed to be unlinked with the thin one ) with the orientations indicated in these figures .
such moves are proved by kirby calculus in the same figures . .
] . ]
we get figure [ torus - simpl / fig ] where the two kinks in the second stage are opposite and so do not affect the framing . the last stage , which gives the figure eight knot , is obtained by framed isotopy .
since the writhe of the figure eight knot is zero , the blackboard framing zero is the same as linking number zero , and this concludes the proof . .
] by a theorem of harer @xcite any 4-dimensional 2-handlebody @xmath2 admits a lefschetz fibration @xmath87 with regular fiber @xmath330 for some @xmath59 ( see also @xcite or @xcite for different proofs ) . up to stabilizations
we can assume @xmath327 .
theorem [ strong - univ / thm ] implies that @xmath379 for some orientation - preserving @xmath340-regular immersion @xmath380 .
then we get a fibered immersion @xmath381 .
it is well - known that @xmath382 is orientation - preserving diffeomorphic to a tubular neighborhood of a projective line in @xmath92 .
then we can consider @xmath383 , and this concludes the proof .
let @xmath384 be the regular fiber of @xmath12 .
if @xmath328 the corollary follows immediately from theorem [ strong - univ / thm ] since @xmath2 admits a fibered immersion in @xmath385 and hence is parallelizable . if @xmath386 we consider the 4-manifold @xmath387 obtained from @xmath2 by the addition of 2-handles along all but one boundary components of the regular fiber @xmath388 with framing zero with respect to @xmath384
. then @xmath389 .
moreover , @xmath12 extends to a lefschetz fibration @xmath167 with regular fiber @xmath360 whose monodromy is obtained from @xmath142 by composition with the homomorphism from @xmath390 to @xmath391 induced by capping off by disks all but one boundaries components of @xmath384 .
the non - separating assumption on the vanishing cycles of @xmath12 implies that @xmath173 is allowable , and so @xmath387 immerses in @xmath88 by the first case . if @xmath392 , @xmath2 fibered immerses in the manifold @xmath83 of theorem [ strong - univ / thm ] .
we conclude by observing that @xmath393 immerses in @xmath88 as a tubular neighborhood of @xmath394 where @xmath395 is a self - transverse immersed disk with boundary the knot @xmath85 .
it is not difficult to see that for @xmath396 @xmath397 which is a lower bound for the number of singular points of a universal lefschetz fibration with fiber @xmath40 @xmath398 denotes the integer part of @xmath399 . in order to include also the not allowable lefschetz fibrations it suffices to replace , in proposition [ main / thm ] , @xmath56 with the set of @xmath400-equivalence classes of homotopically essential curves .
the proof is very similar .
in @xcite loi and piergallini characterized compact stein domains of dimension four , up to orientation - preserving diffeomorphisms , as the total spaces of positive lefschetz fibrations ( meaning with only positive singular points ) over @xmath134 with bounded fiber .
we can express this theorem in terms of universal positive lefschetz fibrations .
following the notations of the proof of theorem [ strong - univ / thm ] let @xmath401 be the lefschetz fibration with fiber @xmath330 and monodromy sequence given by @xmath402 for @xmath403 and @xmath404 , @xmath333 , @xmath405 for @xmath327 as showed in figure [ univ - stein / fig ] .
then @xmath406 is universal ( but not strongly universal ) by proposition [ main / thm ] .
moreover @xmath407 and @xmath408 has the kirby diagram depicted in figure [ kirby - stein / fig ] for @xmath327 .
that @xmath409 follows by the same argument used in the proof of theorem [ strong - univ / thm ] .
it follows that compact 4-dimensional stein domains with strictly pseudoconvex boundary coincide , up to orientation - preserving diffeomorphisms , with the total spaces of lefschetz fibrations that belong in @xmath410 for some @xmath412 . | in analogy with the vector bundle theory we define universal and strongly universal lefschetz fibrations over bounded surfaces . after giving a characterization of these fibrations
we construct very special strongly universal lefschetz fibrations when the fiber is the torus or an orientable surface with connected boundary and the base surface is the disk . as a by - product
we also get some immersion results for @xmath0-dimensional @xmath1-handlebodies .
keywords : universal lefschetz fibration , dehn twist , 4-manifold .
# 1 |
in this section we recall some notation , notions , and results which will turn out to be useful in proving both main results of the paper , theorem [ 1 ] and theorem [ 3 ] .
since the problem of tilings by crosses comes originally from the area of error - correcting codes we will stick to some of its terminology .
let @xmath32 be a @xmath20-tiling of @xmath2 by crosses .
then the set @xmath48 of centers of crosses in @xmath32 uniquely determines @xmath32 .
sometimes we will abuse the language and call @xmath34 a tiling of @xmath2 by crosses .
the elements of @xmath49 will be called words while the words in @xmath34 will be called codewords .
we will also say that a codeword @xmath50 covers a word @xmath51 if @xmath52 as usual @xmath53 stands for the manhattan distance of @xmath54 and @xmath55 given by @xmath56 we set @xmath57 @xmath58 where @xmath59 the following simple observation will be used several times : [ 0 ] let @xmath32 be a tiling of @xmath2 by crosses . then permuting the order of coordinates of each codeword in @xmath32 and/or changing a sign of a coordinate for each codeword in @xmath32 and/or adding @xmath60a word @xmath61 to each codeword results in a set @xmath62 which constitutes a tiling of @xmath2 by crosses congruent to @xmath32 .
if @xmath34 @xmath63is a tiling of @xmath2 by crosses then for each word @xmath51 in @xmath49 there is a unique codeword @xmath50 in @xmath34 so that @xmath52 therefore @xmath34 can be also seen as a decomposition ( tiling ) of @xmath49 by lee spheres @xmath64 of radius @xmath65 centered at @xmath66 ; i.e. , @xmath67 and vice versa , each tiling of @xmath49 by translations of @xmath64 induces a tiling of @xmath2 by crosses . as usual , @xmath68 where the @xmath69-th coordinate equal to @xmath70 a tiling @xmath34 of @xmath2 ( @xmath49 ) by translations of @xmath51 can be described in the form @xmath71 .
then @xmath34 is a lattice tiling if @xmath32 is a lattice .
the following theorem stated in @xcite turns out to be useful when proving both main results of the paper . [ 14 ] let @xmath51 be a subset of @xmath72 then there is a lattice tiling of @xmath49 by translations of @xmath51 if and only if there is an abelian group @xmath73 of order @xmath74 and a homomorphism @xmath75 so that the restriction of @xmath76 to @xmath51 is a bijection .
in addition , if @xmath76 satisfies this condition , then the lattice tiling of @xmath49 by translations of @xmath51 is given by @xmath77 .
as an immediate consequence we get : [ 20 ] let @xmath78 , the cyclic group of order @xmath79 be a homomorphism so that , for all @xmath80 @xmath81 is not an inverse element to @xmath82 that is @xmath83 then @xmath84 is a lattice tiling of @xmath49 by @xmath85 let @xmath32 be a collection of crosses that tile @xmath86 we will always assume wlog that the cross @xmath87 centered at the origin belongs to @xmath32 .
then each cross @xmath88 can be seen as a translation of @xmath87 by a vector @xmath89 .
so @xmath90**@xmath91**@xmath92 for the sake of brevity we will use @xmath93 for a cross centered at a point @xmath94
in this section we provide a proof of theorem [ 3 ] . the following lemma will be the key ingredient of the proof .
we recall that by theorem [ m ] there is a unique lattice tiling of @xmath2 when @xmath1 is a prime .
[ ll]let @xmath1 be a prime , and let @xmath95 be a unique lattice tiling of @xmath2 by crosses . if @xmath96 is a cross in @xmath95 , the shifting @xmath96 along any axis will cause that all crosses of @xmath95 will be shifted as well . consider the homomorphism @xmath98 the cyclic group of order @xmath79 given by @xmath99 for all @xmath100 then , by corollary [ 20 ] , @xmath76 induces a lattice tiling @xmath101**@xmath91**@xmath102 of @xmath2 by crosses
. let @xmath103 , @xmath104 be fixed .
we will prove that shifting the cross @xmath87 along the @xmath103-th axis would shift all crosses in @xmath95 .
we start with describing vectors @xmath105 that form a basis of the lattice @xmath106 let @xmath107 be the element inverse to @xmath103 in the multiplicative abelian group @xmath108 for each @xmath109 @xmath110 we set @xmath111 and @xmath112 clearly , @xmath113 that is , @xmath114 indeed , for @xmath110 @xmath115 and @xmath116 let @xmath51 be the matrix whose rows are vectors @xmath117 it is easy to calculate @xmath118 as the rows and columns of @xmath51 can be permuted such that the resulting matrix is a lower triangular having @xmath119 as its diagonal entries .
therefore , @xmath120 which in turn implies that @xmath105 form a basis of the lattice @xmath106 assume that the cross @xmath121has been shifted along the @xmath103-th axis .
then this will cause that the cross @xmath122 , @xmath109 will be shifted as well . indeed , for @xmath110 the cross @xmath122 contains the unit cube @xmath123 centered at @xmath124 ( centered at @xmath125 for @xmath126 ) ;
that is , the center of @xmath123 lies on @xmath103-th axis .
further , the cross @xmath87 contains the cube @xmath127 centered at @xmath128 thus , when shifting @xmath87 along the @xmath103-th axis we shift the cube @xmath127 along this axis , and this will cause the cube @xmath123 to get shifted ; i.e. , the cross @xmath122 will be shifted along the @xmath103-th axis for all @xmath129 .
consider now a cross @xmath93 in @xmath130 as @xmath95 is a lattice tiling , the above proved statement is true for any cross @xmath131 hence : with this claim in hand it is easy to provide the closing argument of our proof .
let @xmath134 .
we will prove that shifting the cross @xmath87 along the @xmath103-th axis will cause that the cross @xmath93 will be shifted as well . since @xmath135 , it is @xmath136 and
because @xmath105 form a basis of @xmath137 @xmath89 can be written as a linear combination @xmath138 where @xmath139 for all @xmath140 so to finish the proof it is suffices to apply repeatedly claim a. * of theorem * [ 3 ] .
let @xmath32 @xmath141**@xmath91**@xmath142 be a non - regular tiling of @xmath2 .
then there is @xmath143 and a vector @xmath144 such that @xmath145 is not an integer .
let the fractional part of @xmath145 is equal to @xmath146 denote by @xmath147 the set of all vectors @xmath148 in @xmath149 such that @xmath150 it is known , see e.g. @xcite , that the collection of crosses @xmath151 forms a prism @xmath152 along the @xmath69-th axis ; i.e. , if a point @xmath153 then , for all @xmath154 also the point @xmath155 hence , shifting all crosses @xmath156 by any vector @xmath157 parallel to @xmath158 independently on other crosses in @xmath32 , results in a new tiling of @xmath2 by crosses .
moreover , if @xmath159 then the shift results in a tiling where all crosses @xmath156 are now centered at points with the @xmath69-th coordinate being an integer . repeatedly applying this procedure to other crosses that have a non - integer coordinate , we arrive at a @xmath20-tiling @xmath160 of @xmath2 by crosses .
since we have started with a non - regular tiling @xmath32 , there is a proper subset @xmath161of @xmath160 of crosses so that @xmath162 comprises a prism along one of the axis . by lemma [ ll ] ,
if the lattice tiling @xmath95 contains a prism along any axis , this prism constitutes all crosses in @xmath95 .
therefore the above tiling @xmath160 is not congruent to the tiling @xmath95 .
however , this contradicts our assumption that there is a unique @xmath20-tiling of @xmath2 by crosses .
the proof of theorem [ 3 ] is complete . for the rest of this paper @xmath163 @xmath49 will stand for a @xmath20-tiling of @xmath2 by crosses . since , from now on , we deal only with @xmath20-tilings by crosses we will drop @xmath20- and refer to @xmath34 simply as a tiling .
in this section we prove the second of the main results of the paper . since this proof is quite long , we are going to break it into phases . in a sense it is possible to say that it is a `` backward '' proof .
as mentioned in the introduction molnar @xcite proved that the number of non - congruent lattice tilings of @xmath2 by crosses equals the number of non - isomorphic abelian groups of order @xmath37 as @xmath1 is a prime for @xmath41 there is only one abelian group of order @xmath164 and thus there is a unique , up to congruence , lattice tiling of @xmath6 by crosses .
thus , to prove the main result it suffices to show : let @xmath50 be a codeword in @xmath34 .
then @xmath165 , the @xmath166-neighborhood of @xmath167 will be the set of codewords @xmath51 in @xmath34 at the distance at most @xmath166 from @xmath167 that is , @xmath168 @xmath169 in the case of @xmath170
we will write @xmath171 instead of @xmath172 we will say that two @xmath166-neighborhoods @xmath165 and @xmath173 are equal ( are congruent ) if the two sets @xmath174 and @xmath175 are equal ( are congruent ) . clearly , for each codeword @xmath50 , the neighborhoods @xmath176 and @xmath177 are empty sets .
[ 4]let @xmath34 be a tiling of @xmath6 by crosses .
then , for each codeword @xmath50 in @xmath34 , the neighborhood @xmath178 and @xmath179 are equal , and @xmath179 is symmetric ; that is , if @xmath180 then @xmath181 as well .
* of theorem [ 33]*. to show that @xmath34 is a lattice tiling it suffices to prove that , for all codewords @xmath182 , @xmath183 as well . as @xmath34 is a tiling by crosses , it is not difficult to see that , for each codeword @xmath184 , there is a sequence @xmath185 of codewords in @xmath34 such that @xmath186 then @xmath187 and because , by theorem [ 4 ] , the -neighborhoods @xmath188 and @xmath179 are equal , which in turn implies , again by theorem [ 4 ] , that @xmath189 as well for all @xmath190 repeatedly applying theorem [ 4 ] we get that @xmath191 @xmath192 @xmath193 is a codeword .
the proof of theorem [ 33 ] is complete . hence , to prove the main result it suffices to prove theorem 4 .
it turns out that in order to be able to do so one needs to look at `` wider '' neighbourhoods .
in fact , to be able to prove theorem [ 4 ] we will have to prove the same type of a theorem for @xmath194-neighbourhoods .
this recalls a situation when one wants to prove a statement @xmath195 by using mathematical induction , but to be able to prove the inductive step a statement stronger than @xmath195 has to be proved .
we will do it in four steps . to facilitate our discussion we introduce more notation and terminology . by a word of type @xmath198 $ ]
we mean a word having @xmath199 coordinates equal to @xmath200 ... , and @xmath201 coordinates equal to @xmath202 the other coordinates equal to zero .
e.g. , both words @xmath203 and @xmath204 are of type @xmath205 .$]there are three types of words @xmath51 with @xmath206 either @xmath51 is of type @xmath207 , $ ] or of type @xmath208 , $ ] or of type @xmath209 .$ ] let @xmath210 then @xmath20 will be called a codeword of a given type with respect to @xmath50 if @xmath211 is of the given type ; the number of codewords of type @xmath198 $ ] in @xmath165 will be denoted @xmath212\right\vert _ { w}$ ] .
if the codeword @xmath50 will be clear from the context , we will drop the subscript @xmath213 similarly , each word @xmath214 is either of type @xmath215 , $ ] or @xmath216 , $ ] or @xmath217 , $ ] or @xmath218 , $ ] or @xmath219 .$ ] \(a ) let @xmath34 @xmath60be a tiling of @xmath49 by crosses .
first we prove a quantitative statement , namely that , for each @xmath220 , the @xmath221-neighbourhood @xmath222 contains the same number of codewords of each type .
we prove it not only for @xmath223but for all @xmath224 we believe that this statement might turn to be very useful when proving conjecture [ 2 ] for other values of @xmath0 , where @xmath1 is a prime .
[ 5]let @xmath34 be a tiling of @xmath2 by crosses where @xmath224 then , for each @xmath225 , @xmath226\right\vert = 0,$ ] @xmath227\right\vert = 2n,$ ] and @xmath228\right\vert = \frac{2n(n-2)}{3}.$ ] further , @xmath229\right\vert = \left\vert [ 2^{2}]\right\vert = 0,\left\vert [ 3^{1},1^{1}]\right\vert = 2n,\left\vert [ 2^{1},1^{2}]\right\vert = 2n(n-2),$ ] and @xmath230\right\vert = \frac{n(n-2)(n-3)}{3}.$ ]
\(b ) we prove the analogue of theorem [ 5 ] for the @xmath194-neighbourhood .
however , we get the explicit values for the number of codewords of individual types only for @xmath41 while for @xmath231 we get those values only as a function of the number of codewords of type @xmath232 .$ ] we point out , that this is not because the methods used are not satisfactory but for some values @xmath233 say @xmath234 there are two ( lattice ) tilings of @xmath49 by crosses with different number of codewords of type @xmath232 .$ ] we stress that for @xmath234 the number @xmath235 is not a prime , hence it does not provide a counterexample to our conjecture .
\(d ) as the last step we show that for any two codewords in @xmath34 their @xmath194-neighborhoods are not only congruent but the two @xmath194-neighborhoods equal , and this joint neighborhood is symmetric , so we prove theorem [ 00 ] . for any codeword @xmath50 in @xmath34
there are @xmath236 words @xmath51 of type @xmath237 $ ] with respect to @xmath50 .
( we recall that this means that @xmath238 is of given type ) .
each of them is covered by a codeword of type @xmath207 , $ ] or by a codeword of type @xmath208 , $ ] with respect to @xmath213 on the other hand , each codeword of type @xmath239 $ ] and of type @xmath208 , $ ] with respect to @xmath167 covers exactly one word of type @xmath237 $ ] with respect to @xmath213 thus we get , for each codeword @xmath167 @xmath240\right\vert + \left\vert [ 2^{1},1^{1}]\right\vert = 2n \label{p}\ ] ] the above and the following equalities are valid for each codeword @xmath167 therefore in what follows we drop the index @xmath241 also we will not repeat any longer that all codewords of given type are meant with respect to @xmath50 .
in @xmath49 there are @xmath242 words @xmath51 of type @xmath243 $ ] .
each of them is covered by a codeword of type @xmath244 , $ ] or by a codeword of type @xmath208 .$ ] on the other hand , each codeword of type @xmath209 $ ] covers three of them while a codeword of type @xmath208 $ ] covers exactly one codeword of type @xmath245 .$ ] hence @xmath246\right\vert + 3\left\vert \left [ 1^{3}\right ] \right\vert = 4\binom{n}{2 } \label{pp}\ ] ] equation ( [ p ] ) and ( [ pp ] ) are `` global '' equations . to get their `` local '' form we need to introduce some more notation .
often we will need to express the number of ( code)words in a set @xmath247 having their @xmath69-th coordinate positive , or their @xmath69-th coordinate negative .
therefore , to simplify the language , we will introduce the notion of the @xmath248 @xmath249 in @xmath72 for the rest of the paper by the set of signed coordinates we will understand the set @xmath250 by @xmath251 we will denote the number of words in @xmath247 with the @xmath69-th coordinate different from @xmath252 that is , @xmath253 stands for the number of words in @xmath247 with the first coordinate being a positive number , while @xmath254 represents the number of words in @xmath247 with the third coordinate being a negative number .
if we need to stress that the non - zero value is @xmath255 we will use @xmath256for the number of codewords with the @xmath69-th coordinate equal @xmath257 thus , for each @xmath258 @xmath259 _ { i}\right\vert $ ] is the number of codewords of type @xmath260 $ ] with the non - zero @xmath69-th coordinate , while @xmath259 _ { i}^{(2)}\right\vert $ ] stands for the set of codewords of type @xmath261 $ ] with the @xmath69-th signed coordinate equal to @xmath262 finally , if @xmath51 is a word in @xmath49 , then by @xmath263 we mean that the @xmath264-th coordinate of @xmath51 equals @xmath255 if @xmath265 and equals @xmath266 if @xmath267 clearly , in the notation of signed coordinates , if @xmath268 then @xmath269 .
e.g. , if @xmath270 then @xmath271 and @xmath272 now we are ready to state the local form of ( [ p ] ) and ( [ pp ] ) . as for each @xmath273
there is in @xmath49 one word @xmath51 of type @xmath274 $ ] with @xmath275 and @xmath276 words @xmath277 of type @xmath278 $ ] with @xmath279 , we get : indeed , if @xmath280 is a codeword of type @xmath281 $ ] with @xmath282 ( and then @xmath283 for all @xmath284 then @xmath280 covers a word @xmath51 of type @xmath274 $ ] with @xmath285 however , a codeword @xmath286 of type @xmath287 $ ] covers @xmath51 only if @xmath288 but does not cover it if @xmath289 on the other hand , @xmath60a codeword @xmath286 with @xmath290 covers one word @xmath291 of type @xmath278 $ ] with @xmath292 regardless whether @xmath293 or @xmath289 clearly , a codeword @xmath162 of type @xmath294 $ ] with @xmath295 covers exactly two words @xmath291 of type @xmath278 $ ] with @xmath296 now we derive equalities analogous to ( [ p ] ) - ( [ pp1 ] ) for codewords of absolute value equal to @xmath297 as ( [ p ] ) - ( [ pp1 ] ) have been derived in great detail , and the same type of ideas are used to prove equalities ( [ a ] ) - ( [ c1 ] ) we will leave a part of the proofs to the reader .
in @xmath49 there are @xmath236 words of type @xmath281.$ ] each of them is covered by a codeword of type @xmath281 $ ] or @xmath298 $ ] or @xmath299,$ ] and each of those codewords covers exactly one word of type @xmath281.$ ] therefore,@xmath240\right\vert + \left\vert [ 4^{1}]\right\vert + \left\vert [ 3^{1},1^{1}]\right\vert = 2n , \label{a}\]]and , for each @xmath258 we have@xmath240_{i}\right\vert + \left\vert [ 4^{1}]_{i}\right\vert + \left\vert [ 3^{1},1^{1}]_{i}^{(3)}\right\vert = 1 .
\label{a1}\]]further , in @xmath49 there are @xmath300 words of type @xmath287.$ ] they are covered by codewords of type @xmath287,$ ] or @xmath299,$ ] or @xmath301 , $ ] or @xmath302 $ ] .
each codeword of type @xmath301 , $ ] or @xmath302 $ ] covers two such words , while each codeword of type @xmath287 $ ] , or @xmath299 $ ] covers one of them .
+ \left\vert [ 3^{1},1^{1}]\right\vert + 2\left\vert [ 2^{2}]\right\vert + 2\left\vert \left [ 2^{1},1^{2}\right ] \right\vert = 2^{3}\binom{n}{2 } \label{b}\]]the above identity has two local forms .
there are @xmath276 words @xmath277 of type @xmath287 $ ] with @xmath303 and @xmath276 words @xmath277 of type @xmath287 $ ] with @xmath304 for each @xmath273 we get@xmath305 _ { i}^{(2)}\right\vert + \left\vert [ 3^{1},1^{1}]_{i}^{(3)}\right\vert + \left\vert [ 2^{2}]_{i}\right\vert + 2\left\vert \left [ 2^{1},1^{2}\right ] _ { i}^{(2)}\right\vert = 2(n-1 ) , \label{b1}\]]and@xmath305 _ { i}^{(1)}\right\vert + \left\vert [ 3^{1},1^{1}]_{i}^{(1)}\right\vert + \left\vert [ 2^{2}]_{i}\right\vert + \left\vert \left [ 2^{1},1^{2}\right ] _ { i}^{(1)}\right\vert = 2(n-1 ) .
\label{b2}\]]further , in @xmath49 there are @xmath306 words of type @xmath294.$ ] they are covered by codewords of type @xmath307 , or @xmath302 $ ] , or @xmath308 $ ] .
each codeword of type @xmath308 $ ] covers four of them .
\right\vert + \left\vert \left [ 2^{1},1^{2}\right ] \right\vert + 4\left\vert \left [ 1^{4}\right ] \right\vert = 2^{3}\binom{n}{3 } \label{c}\]]the local form of ( [ c ] ) reads as follows:@xmath309 _ { i}\right\vert + \left\vert \left [ 2^{1},1^{2}\right ] _ { i}\right\vert + 3\left\vert \left [ 1^{4}\right ] _ { i}\right\vert = 2^{2}\binom{n-1}{2 } \label{c1}\]]as in @xmath49
there are @xmath310 words @xmath277 of type @xmath294 $ ] with @xmath311 and each codeword @xmath51 of type @xmath308 $ ] with @xmath312 covers three of them .
[ 6]let @xmath314 and @xmath34 be a tiling of @xmath2 by crosses .
then , for each @xmath225 , @xmath226\right\vert = 0,$ ] @xmath227\right\vert = 2n,$ ] and @xmath315 \right\vert = \frac{2n(n-2)}{3}.$ ] as to the local values , for each @xmath258 @xmath259 _ { i}^{(2)}\right\vert = \left\vert \left [ 2^{1},1^{1}\right ] _ { i}^{(1)}\right\vert = 1,$ ] that is , @xmath259 _ { i}\right\vert = 2,$ ] and @xmath315 _ { i}\right\vert = n-2.$ ]
let @xmath50 be a codeword in @xmath34 .
clearly , then also the set @xmath316 for some @xmath51 in @xmath317 is a tiling of @xmath49 by crosses .
therefore , wlog we assume @xmath318 . from ( [ p1 ] )
we have @xmath319 _ { i}^{(2)}\right\vert \leq1 , $ ] while from ( [ pp1 ] ) we get @xmath320 _ { i}\right\vert $ ] is even , hence @xmath320 _ { i}^{(2)}\right\vert \leq\left\vert \left [ 2^{1},1^{1}\right ] _ { i}\right\vert $ ] . on the other hand , there is no @xmath273 with @xmath319 _
{ i}^{(2)}\right\vert < \left\vert \left [ 2^{1},1^{1}\right ] _ { i}\right\vert $ ] as @xmath321 _ { i}^{(2)}\right\vert = { \displaystyle\sum\limits_{i\in i } } \left\vert \left [ 2^{1},1^{1}\right ] _ { i}^{(1)}\right\vert .$ ] thus we proved : * lemma a. * for each @xmath322 either @xmath320 _ { i}\right\vert = 0 $ ] or @xmath319 _ { i}\right\vert = 2.$ ] in the latter case @xmath320 _ { i}^{(2)}\right\vert = \left\vert \left [ 2^{1},1^{1}\right ] _ { i}^{(1)}\right\vert = 1.$ ] \(i ) @xmath323_{i}\right\vert = 1.$ ] then , by ( [ a1 ] ) , @xmath324_{i}^{(3)}\right\vert = 0,$ ] and by ( [ p1 ] ) , @xmath319 _ { i}^{(2)}\right\vert = 0,$ ] which implies , by lemma a , that @xmath319 _ { i}\right\vert = 0.$ ] this in turn implies , see ( [ pp1 ] ) , @xmath325 _ { i}\right\vert = n-1.$ ] substituting it into ( [ c1 ] ) gives @xmath326 _ { i}\right\vert + 3\left\vert \left [ 1^{4}\right ] _ { i}\right\vert = ( n-1)(2n-5).$ ] as we deal with the case @xmath327 then @xmath328 as well , and therefore @xmath326 _ { i}\right\vert = 2(\operatorname{mod}3).$ ] subtracting ( [ b2 ] ) from ( [ b1 ] ) , and using @xmath324_{i}^{(3)}\right\vert = \left\vert \left [ 2^{1},1^{1}\right ] _ { i}^{(2)}\right\vert = \left\vert \left [ 2^{1},1^{1}\right ] _ { i}^{(1)}\right\vert = 0,$ ] we get @xmath329 _ { i}^{(2)}\right\vert = \left\vert \left [ 2^{1},1^{2}\right ] _ { i}^{(1)}\right\vert + \left\vert [ 3^{1},1^{1}]_{i}^{(1)}\right\vert .$ ] as @xmath326 _ { i}\right\vert = \left\vert \left [ 2^{1},1^{2}\right ] _ { i}^{(2)}\right\vert + \left\vert \left [ 2^{1},1^{2}\right ] _ { i}^{(1)}\right\vert , $ ] adding @xmath326 _ { i}^{(2)}\right\vert $ ] to both sides yields @xmath330 _ { i}^{(2)}\right\vert = \left\vert \left [ 2^{1},1^{2}\right ] _ { i}\right\vert $ ] @xmath331_{i}^{(1)}\right\vert .$ ] we showed above that in this case of @xmath226_{i}\right\vert = 1 $ ] it is @xmath326 _
{ i}\right\vert = 2(\operatorname{mod}3).$ ] therefore @xmath324_{i}^{(1)}\right\vert > 0,$ ] that is , @xmath324_{i}^{(1)}\right\vert > \left\vert [ 3^{1},1^{1}]_{i}^{(3)}\right\vert .$ ] \(ii ) @xmath323_{i}\right\vert = 0.$ ] by ( [ p1 ] ) , we get @xmath319 _ { i}^{(2)}\right\vert = 1,$ ] which implies , by , that @xmath319 _ { i}\right\vert = 2.$ ] this in turn implies , see ( [ pp1 ] ) , @xmath325 _ { i}\right\vert = n-2.$ ] substituting it into ( [ c1 ] ) gives @xmath326 _ { i}\right\vert + 3\left\vert \left [ 1^{4}\right ] _ { i}\right\vert = ( n-2)(2n-3).$ ] as @xmath327 it is @xmath332 and therefore @xmath326 _ { i}\right\vert = 0(\operatorname{mod}3).$ ] subtracting ( [ b2 ] ) from ( [ b1 ] ) , and using @xmath319 _ { i}^{(2)}\right\vert = \left\vert \left [ 2^{1},1^{1}\right ] _ { i}^{(1)}\right\vert = 1,$ ] we get @xmath333 _ { i}^{(2)}\right\vert + \left\vert [ 3^{1},1^{1}]_{i}^{(3)}\right\vert -\left\vert [ 3^{1},1^{1}]_{i}^{(1)}\right\vert = \left\vert \left [ 2^{1},1^{2}\right ] _ { i}^{(1)}\right\vert \,,$ ] and adding @xmath326 _ { i}^{(2)}\right\vert $ ] to both sides gives @xmath334 _ { i}^{(2)}\right\vert + \left\vert [ 3^{1},1^{1}]_{i}^{(3)}\right\vert -\left\vert [ 3^{1},1^{1}]_{i}^{(1)}\right\vert = \left\vert \left [ 2^{1},1^{2}\right ] _ { i}\right\vert .$ ] as @xmath326 _ { i}\right\vert = 0(\operatorname{mod}3)$ ] in this case , we have @xmath324_{i}^{(3)}\right\vert -\left\vert [ 3^{1},1^{1}]_{i}^{(1)}\right\vert = 0(\operatorname{mod}3),$ ] which yields @xmath324_{i}^{(1)}\right\vert \geq\left\vert \lbrack3^{1},1^{1}]_{i}^{(3)}\right\vert $ ] as @xmath324_{i}^{(3)}\right\vert \leq 1 $ ] for all @xmath258 see ( [ a1 ] ) .
so , @xmath323_{i}\right\vert = 1 $ ] implies @xmath335_{i}^{(1)}\right\vert > \left\vert [ 3^{1},1^{1}]_{i}^{(3)}\right\vert , $ ] while @xmath323_{i}\right\vert = 0 $ ] gives @xmath324_{i}^{(1)}\right\vert \geq\left\vert \lbrack 3^{1},1^{1}]_{i}^{(3)}\right\vert .$ ] however , @xmath336_{i}^{(1)}\right\vert = { \displaystyle\sum\limits_{i\in i } } \left\vert [ 3^{1},1^{1}]_{i}^{(3)}\right\vert , $ ] therefore there is no @xmath337 with @xmath323_{i}\right\vert = 1,$ ] that is @xmath226\right\vert = 0,$ ] and , for all @xmath258 @xmath338_{i}^{(1)}\right\vert = \left\vert [ 3^{1},1^{1}]_{i}^{(3)}\right\vert , \text { and } 3\left\vert \left [ 2^{1},1^{2}\right ] _ { i}^{(2)}\right\vert = \left\vert \left [ 2^{1},1^{2}\right ] _ { i}\right\vert \label{s}\ ] ] since @xmath323\right\vert = 0,$ ] by ( [ p ] ) we get @xmath339\right\vert = 2n,$ ] which in turn implies , by ( [ pp ] ) , that @xmath340 \right\vert = \frac{2n(n-2)}{3}.$ ] further , from @xmath319 _ { i}\right\vert = 2,$ ] we get @xmath340 _ { i}\right\vert = n-2.$ ]
the proof is complete . [
7 ] let @xmath34 be a tiling of @xmath2 by crosses where @xmath224 then , for each @xmath220 , @xmath342 \right\vert = 2n(n-2),$ ] and @xmath341\right\vert=\frac{n(n-2)(n-3)}{3}.\,$ ] in addition , for all @xmath258 it is@xmath343 _ { i}\right\vert=3(n-2),$ ] @xmath326 _ { i}^{(2)}\right\vert = n-2,$ ] and @xmath341
_ { i}\right\vert=\frac{2(n-2)(n-3)}{3}.$ ] * proof of lemma * [ l2 ] .
assume by contradiction that there is @xmath346say @xmath347such that @xmath345 _ { 1}\right\vert\geq2.$ ] let , w.l.o.g , @xmath348 be two codewords of type @xmath349 $ ] with @xmath350 .
we proved that , for each @xmath258 it is @xmath351 _ { i}^{(2)}\right\vert=1.$ ]
so there is a codeword @xmath286 of type @xmath352 $ ] , with @xmath353 we may assume w.l.o.g .
that @xmath354 if @xmath355 then @xmath356 that is , the codeword @xmath357 is with respect to the codeword @xmath286 of type @xmath358,$ ] which is a contradiction as we proved that @xmath359 \right\vert=0.$ ] so let @xmath360 then @xmath361 and @xmath362 that is , with respect to the codeword @xmath363 we get @xmath364_{i}^{(1 ) } \right\vert=2,$ ] which contradicts that @xmath364_i^{(1 ) } \right\vert=1 $ ] for all @xmath365 therefore , @xmath366 _ { i } \right\vert\leq1 $ ] for all @xmath258 which in turn implies @xmath366 \right\vert=\frac{1}{2}{\displaystyle\sum\limits_{i\in i } } \left\vert \left [ 2 ^ 2\right ] _
i \right\vert\leq n.$ ] this proves lemma [ l2 ] .
with this in hand we find the values of @xmath342 \right\vert$ ] and @xmath342_{i } \right\vert.$ ] the equality ( [ b1 ] ) states that @xmath367_i^{(2 ) } \right\vert+\left\vert \left [ 3^{1 } , 1^{1 } \right]_i^{(3 ) } \right\vert+\left\vert \left [ 2^{2 } \right]_i \right\vert+2\left\vert \left [ 2^{1 } , 1^{2 } \right]_{i}^{(2 ) } \right\vert=2(n-1\dot{)}.$ ] in addition , by lemma [ 6 ] , it is @xmath364_i^{(2 ) } \right\vert=1,$ ] by ( [ a1 ] ) @xmath368_i^{(3 ) } \right\vert\leq1,$ ] and by the above lemma @xmath366_i \right\vert\leq1.$ ] as @xmath364_i^{(2 ) } \right\vert+\left\vert \left [ 3 ^ 1 , 1 ^ 1 \right]_i^{(3 ) } \right\vert+\left\vert \left [ 2 ^ 2 \right]_i \right\vert$ ] is an even number , we get @xmath369_i^{(2 ) } \right\vert+\left\vert \left [ 3 ^ 1 , 1 ^ 1 \right]_i^{(3 ) } \right\vert+\left\vert \left [ 2 ^ 2 \right]_i \right\vert=2.\label{ee}\ ] ] therefore @xmath342_{i}^{(2 ) } \right\vert = n-2,$ ] thus @xmath342 \right\vert=$ ] @xmath370_{i}^{(2 ) } \right\vert=2n(n-2).$ ] by ( [ s ] ) , @xmath342_{i } \right\vert=3\left\vert \left [ 2^{1 } , 1^{2 } \right]_{i}^{(2 ) } \right\vert=3(n-2).$ ] the values of @xmath371 \right\vert$ ] and @xmath371_{i } \right\vert$ ] are easily obtained from ( [ c ] ) and ( [ c1 ] ) , respectively .
the proof is complete . to be able to determine the values of @xmath372 \right\vert,\left\vert
\left [ 3^{1 } , 1^{1 } \right ] \right\vert,$ ] and @xmath373 \right\vert,$ ] we need to consider codewords from the @xmath194-neighbourhood .
each word @xmath374 is either of type @xmath375 , $ ] or @xmath376 , $ ] or @xmath377 , $ ] or @xmath378 , $ ] or @xmath379 , $ ] or @xmath380 , $ ] or @xmath381 .$ ] let @xmath50 be a codeword in @xmath34 .
then the number of codewords @xmath20 in the @xmath194-neighbourhood of @xmath50 of the given type @xmath375 $ ] will be denoted by @xmath382 \right\vert,$ ] of type @xmath383 $ ] by @xmath384 \right\vert,$ ] etc .
assume by contradiction that there are two codewords @xmath386 and @xmath387 of type @xmath388 $ ] with @xmath389 for @xmath390 by ( [ a1 ] ) we have @xmath391_i \right\vert+\left\vert \left [ 3 ^ 1 , 1 ^ 1 \right]_i^{(3 ) } \right\vert=1 $ ] as we know from theorem [ 6 ] that @xmath392_i \right\vert=0.$ ] so , assume first that @xmath372_{i } \right\vert=1.$ ] then there is a codeword @xmath291 , with @xmath393 say , w.l.o.g , @xmath394 and @xmath395 @xmath396 . as @xmath397 and @xmath398 we arrived at a contradiction since with respect to @xmath291 we have @xmath367_{i}^{(1 ) } \right\vert>1.$ ] suppose now that @xmath368_i^{(3 ) } \right\vert=1,$ ] i.e. , there is codeword @xmath399 of type @xmath299 $ ] so that @xmath400 say @xmath401 if @xmath402 then @xmath403a contradiction as @xmath404 \right\vert=0 $ ] with respect to all codewords .
so we may assume that @xmath405 and @xmath406 then @xmath407 and @xmath408 i.e. , with respect to @xmath167 @xmath367 _ { i}^{(1 ) } \right\vert>1,~$ ] a contradiction .
the proof of the first part follows .
now let @xmath286 be a codeword of type @xmath208 $ ] with @xmath409 further , let @xmath385_{i}^{(2 ) } \right\vert\geq3 $ ] , and @xmath410 and @xmath411 be codewords of type @xmath412 $ ] with @xmath413 @xmath414 we assume w.l.o.g . that @xmath415 and @xmath416 then there are at least two of the codewords @xmath417 say @xmath386 and @xmath387 having the second coordinate equal to @xmath418 as otherwise the two codewords would be at distance less than @xmath313 .
we assume w.l.o.g . @xmath419 and @xmath420 hence @xmath421 and @xmath422 i.e. , with respect to the codeword @xmath423 we get @xmath424 _ { -2}^{(1 ) } \right\vert>1.$ ] this contradicts ( [ s ] ) because @xmath368_i^{(3 ) } \right\vert\leq1 $ ] for all @xmath365 the proof is complete .
before we prove the next lemma we get equalities related to covering words of absolute value @xmath425 in @xmath49 there are @xmath236 words of type @xmath298.$ ] by theorem [ 6 ] , there is no codeword of type @xmath281.$ ] hence we have@xmath426 \right\vert+\left\vert \left [ 5^{1 } \right ] \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert=2n,\label{d}\ ] ] and , for all @xmath258 the local form reads as follows:@xmath426_{i } \right\vert+\left\vert \left [ 5^{1 } \right ] _ { i } \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert=1\label{d1}\ ] ] further , in @xmath49 there are @xmath300 words of type @xmath299.$ ] each of them is covered by a codeword of type @xmath427 , $ ] or @xmath208 , $ ] or @xmath428 , $ ] or @xmath429 , $ ] or @xmath430 , $ ] or @xmath431 .$ ] only codewords of type @xmath432 $ ] cover two words of type @xmath299.$ ] in addition we know that there is no codeword of type @xmath207 $ ] .
thus , @xmath433 \right\vert+\left\vert \left [ 3^{1 } , 1^{1 } \right ] \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right ] \right\vert+2\left\vert \left [ 3^{1 } , 1^{2 } \right ] \right\vert=8\binom{n}{2}\label{e}\ ] ] for each @xmath258 there are @xmath276 words @xmath51 of type @xmath299 $ ] with @xmath434 and , at the same time , @xmath276 words @xmath51 of type @xmath299 $ ] with @xmath435 thus the local forms of ( [ e ] ) read as follows:@xmath436_i^{(2 ) } \right\vert+\left\vert \left [ 3 ^ 1 , 1 ^ 1 \right]_i^{(3 ) } \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert+\\2\left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(3 ) } \right\vert=2(n-1),\label{e1}\end{gathered}\ ] ] and @xmath437 _ { i}^{(1 ) } \right\vert+\left\vert \left [ 3^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert+\\ \left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(1 ) } \right\vert=2(n-1).\label{e2}\end{gathered}\ ] ] in @xmath49 there are @xmath242 words of type @xmath349.$ ] each word of this type is covered by a codeword of type @xmath438 , $ ] or @xmath439 , $ ] or @xmath440 , $ ] or @xmath441 $ ] .
as each such codeword covers one word of type @xmath349 $ ] we have @xmath433 \right\vert+\left\vert \left [ 2^{2 } \right ] \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right ] \right\vert+\left\vert \left [ 2^{2 } , 1^{1 } \right ] \right\vert=2^{2}\binom{n}{2}\label{f}\ ] ] and , for each @xmath258 we get@xmath433 _ { i } \right\vert+\left\vert \left [ 2 ^ 2 \right]_i \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right ]
_ { i } \right\vert+\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert=2(n-1)\label{f1}\ ] ] in @xmath49 there are @xmath442 words of type @xmath443.$ ] each of these words is covered by a codeword of type @xmath208 , $ ] or @xmath294 , $ ] or @xmath444 , $ ] or @xmath431 , $ ] or @xmath445 , $ ] or @xmath441 .$ ] as each codeword of type @xmath446 $ ] covers @xmath447 of them , each codeword of type @xmath294 $ ] and of type @xmath448 $ ] covers three of them , and each codeword of type @xmath449 $ ] covers two of them , we get@xmath450 \right\vert+3\left\vert \left [ 1^{3 } \right ] \right\vert+\left\vert \left [ 2^{1 } , 1^{2 } \right ] \right\vert+\left\vert \left [ 3^{1 } , 1^{2 } \right ] \right\vert+\\ 3\left\vert \left [ 2^{1 } , 1^{3 } \right ] \right\vert+2\left\vert \left [ 2^{2 } , 1^{1 } \right ] \right\vert=24\binom{n}{3}\label{g}\end{gathered}\ ] ] for each @xmath258 in @xmath49 there are @xmath310 words @xmath51 of type @xmath443 $ ] with @xmath285 hence , a local form of ( [ g ] ) is@xmath451_i^{(2 ) } \right\vert+\left\vert \left [ 1^{3 } \right ] _
{ i } \right\vert+\left\vert \left [ 2^{1 } , 1^{2 } \right]_{i}^{(2 ) } \right\vert+\left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(3 ) } \right\vert+
\\ 3\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { i}^{(2 ) } \right\vert+\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert=2^{2}\binom{n-1}{2}.\label{g1}\end{gathered}\ ] ] in @xmath49 there are @xmath452 words @xmath51 of type @xmath453 $ ] with @xmath435 it is not difficult to see that:@xmath437 _ { i}^{(0 ) } \right\vert+2(n-2)\left\vert \left [ 2^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert+2\left\vert \left [ 1^{3 } \right ] _
{ i } \right\vert+\left\vert \left [ 2^{1 } , 1^{2 } \right ] _ { i}^{(1 ) } \right\vert+\\ \left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(1 ) } \right\vert+2\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { i}^{(1 ) } \right\vert+\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert+\\ 2\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert=8\binom{n-1}{2},\label{g2}\end{gathered}\ ] ] where @xmath367 _ { i}^{(0 ) } \right\vert$]stands for codeword @xmath51 of type @xmath261 $ ] with @xmath454 as @xmath367 _
{ i } \right\vert=\left\vert \left [ 2^{1 } , 1^{1 } \right ]
_ { -i } \right\vert=2,$ ] we have @xmath367 _ { i}^{(0 ) } \right\vert=2(n-2).$ ] finally , in @xmath49 there are @xmath455 words of type @xmath456 .$ ] each of them is covered by a codeword of type @xmath294 , $ ] or @xmath219 , $ ] or @xmath457 , $ ] or @xmath458 .$ ] clearly each codeword of type @xmath294 $ ] covers @xmath459 words of type @xmath460 , $ ] while each codeword of type @xmath381 $ ] covers @xmath194 words of type @xmath219 .$ ] therefore:@xmath461 \right\vert+\left\vert \left [ 1^{4 } \right ] \right\vert + \left\vert \left [ 2^{1 } , 1^{3 } \right ] \right\vert+5\left\vert \left [ 1^{5 } \right ] \right\vert=2^{4}\binom{n}{4}.\label{h}\ ] ] the local form of ( [ h ] ) reads as follows@xmath462 _ { i}^{(0 ) } \right\vert+2(n-3)\left\vert \left [ 1^{3 } \right ]
_ { i } \right\vert+\left\vert \left [ 1^{4 } \right ] _ { i } \right\vert+\left\vert \left [ 2^{1 } , 1^{3 } \right ] _
{ i } \right\vert+\\ 4\left\vert \left [ 1^{5 } \right ] _
{ i } \right\vert=2^{3}\binom{n-1}{3},\label{h1}\end{gathered}\ ] ] where @xmath463 _ { i}^{(0 ) } \right\vert$ ] stands for the number of codewords @xmath51 of type @xmath464 $ ] with @xmath454 now we are ready to prove a lemma crucial for the proof of the next theorem .
( [ d1 ] ) states that @xmath372_{i } \right\vert+\left\vert \left [ 5^{1 } \right ] _ { i } \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert=1.\,\ $ ] thus , the lemma covers all possible cases .
[ l5]let @xmath365 if @xmath372_{i } \right\vert=1,$ ] then @xmath366_i \right\vert=1,$ ] @xmath368_i^{(3 ) } \right\vert=0,$ ] and @xmath385_{i}^{(3 ) } \right\vert=1,$ ] while if @xmath465 _
{ i } \right\vert=1,$ ] then @xmath366_i \right\vert=0,\left\vert \left [ 3 ^ 1 , 1 ^ 1 \right]_i^{(3 ) } \right\vert=1,$ ] and @xmath385_{i}^{(3 ) } \right\vert=0,$ ] but if @xmath466 _ { i}^{(4 ) } \right\vert=1 $ ] then @xmath366_i \right\vert=0,\left\vert \left [ 3^{1 } , 1^{1 } \right ] _ { i}^{(3 ) } \right\vert=1,$ ] and @xmath385_{i}^{(3 ) } \right\vert=1.$ ] in particular , @xmath385 \right\vert=\left\vert \left [ 4^{1 } \right ] \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert.$ ] assume first that @xmath372_{i } \right\vert=1,$ ] and let @xmath291 be a codeword in @xmath34 of type @xmath215 $ ] with @xmath467 . as @xmath372_{i } \right\vert=1,$ ] then taking into account @xmath404 _ { i } \right\vert=0 $ ] and ( [ a1 ] ) , we get @xmath368_i^{(3 ) } \right\vert=0.$ ] this in turn implies , due to ( [ ee ] ) , that @xmath366_i \right\vert=1.$ ] consider the @xmath313-neighbourhood @xmath468 of @xmath291 . by theorem
[ 6 ] , we have for @xmath468 that @xmath367 _ { i}^{(1 ) } \right\vert=1 $ ] for all @xmath365 that is there has to be in @xmath34 a codeword @xmath286 of type @xmath446 $ ] with respect to @xmath469 with @xmath470 that is @xmath471 thus @xmath472 and there is a @xmath473 so that @xmath474 hence , @xmath286 is a codeword of type @xmath412 $ ] with respect to the origin @xmath475 and therefore @xmath385_{i}^{(3 ) } \right\vert\geq1,$ ] while by lemma [ l3 ] we get @xmath385_{i}^{(3 ) } \right\vert=1.$ ] the first part of the proof is complete .
let now @xmath465 _
{ i } \right\vert=1.$ ] then there is a codeword @xmath50 in @xmath34 of type @xmath476 $ ] with @xmath477 further , by ( [ d1 ] ) , @xmath372_{i } \right\vert=0,$ ] which in turn implies , see ( [ a1 ] ) , that @xmath368_i^{(3 ) } \right\vert=1,$ ] and by ( [ ee ] ) , @xmath366_i \right\vert=0.$ ] now we prove that in this case @xmath385_{i}^{(3 ) } \right\vert=0 $ ] as well .
let @xmath478be a codeword of type @xmath208 $ ] with @xmath479 w.l.o.g .
, let @xmath480 and @xmath481 assume that there is a codeword @xmath162 of type @xmath412 $ ] with @xmath482 then @xmath483 and @xmath484 that is , with respect to the codeword @xmath167 we have @xmath424 _ { -1}^{(3 ) } \right\vert=1 $ ] but also @xmath373 _ { -1 } \right\vert=1,$ ] which contradicts ( [ ee ] ) as @xmath364_i^{(2 ) } \right\vert=1 $ ] for all @xmath365 thus @xmath385_{i}^{(3 ) } \right\vert=0 $ ] in this case .
the proof of the second part of the statement is complete .
now , assume that @xmath466 _ { i}^{(4 ) } \right\vert=1.$ ] we need to show that in this case @xmath385_{i}^{(3 ) } \right\vert=1 $ ] as well .
however , by ( [ e1 ] ) , @xmath364_i^{(2 ) } \right\vert+\left\vert \left [ 3 ^ 1 , 1 ^ 1 \right]_i^{(3 ) } \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert$ ] is an even number , and in this case we have @xmath364_i^{(2 ) } \right\vert=\left\vert \left [ 3 ^ 1 , 1 ^ 1 \right]_i^{(3 ) } \right\vert=\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert=1 .
$ ] hence @xmath385 _ { i}^{(3 ) } \right\vert$ ] is odd , and , by lemma [ l3 ] , @xmath385_{i}^{(3 ) } \right\vert=1.$ ] finally , @xmath385
\right\vert={\displaystyle\sum\limits_{i\in i } } \left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert={\displaystyle\sum\limits_{\left\vert \left [ 4^{1 } \right]_{i } \right\vert=1 } } \left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert + { \displaystyle\sum\limits_{\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert=1 } } \left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert \\+{\displaystyle\sum\limits_{\left\vert \left [ 5^{1 } \right ] _
{ i } \right\vert=1 } } \left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert={\displaystyle\sum\limits_{\left\vert \left [ 4^{1 } \right]_{i } \right\vert=1 } } \left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert+{\displaystyle\sum\limits_{\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert=1 } } \left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert=\left\vert \left [ 4^{1 } \right ] \right\vert + \left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert.$ ]
the proof is complete .
[ 8]for each codeword @xmath50 in @xmath485 we have @xmath424 \right\vert=2n,$ ] @xmath60and @xmath372 \right\vert=\left\vert \left [ 2^{2 } \right ] \right\vert=0.$ ] in addition , for all @xmath258 @xmath368_i^{(3 ) } \right\vert=\left\vert \left [ 3^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert=1.$ ] as above , we assume w.l.o.g .
that @xmath344 we recall that , by ( [ s ] ) , @xmath368_i^{(3 ) } \right\vert=\left\vert \left [ 3^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert$]for all @xmath273 . since @xmath385 _
{ i } \right\vert=\left\vert \left [ 3^{1 } , 2^{1 } \right ] _ { i}^{(3 ) } \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert,$ ] we get from ( [ f1 ] ) @xmath437 _ { i } \right\vert+\left\vert \left [ 2^{2 } \right ]
_ { i } \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert+\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert+\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert\\ = 2(n-1)\label{rr}\end{gathered}\ ] ] combining lemma [ l5 ] with ( [ e1 ] ) , ( [ e2 ] ) , and ( [ rr ] ) yields : if @xmath372_{i } \right\vert=1,$ ] then @xmath486 _ { i}^{(3 ) } \right\vert = n-2,\\ \left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(1 ) } \right\vert=2n-3-\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert-\left\vert \left [ 3^{1 } , 2^{1 } \right ] _ { i}^{(2 ) } \right\vert,\\ \text { and } \left\vert \left [ 2^{2 } , 1^{1 } \right ]
_ { i}^{(2 ) } \right\vert=2n-6-\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert\label{r1}\end{gathered}\ ] ] for @xmath487 _ { 1 } \right\vert=1,$ ] it is @xmath486 _ { i}^{(3 ) } \right\vert = n-2,\\ \left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(1 ) } \right\vert=2n-4-\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert-\left\vert \left [ 3^{1 } , 2^{1 } \right ] _ { i}^{(2 ) } \right\vert=,\\ \text { and } \left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert=2n-4-\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert,\label{r3}\end{gathered}\ ] ] and for @xmath488 _ { i } \right\vert=1 $ ] we get@xmath486 _ { i}^{(3 ) } \right\vert = n-3 , \\
\left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(1 ) } \right\vert=2n-4-\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert-\left\vert \left [ 3^{1 } , 2^{1 } \right ] _ { i}^{(2 ) } \right\vert=,\\ \text { and } \left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert=2n-5-\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert.\label{r2}\end{gathered}\ ] ] @xmath489 \right\vert={\displaystyle\sum\limits_{i\in i } } \left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(3 ) } \right\vert={\displaystyle\sum\limits_{\left\vert \left [ 4^{1 } \right]_{i } \right\vert=1 } } \left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(3 ) } \right\vert+{\displaystyle\sum\limits_{\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert=1 } } \left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(3 ) } \right\vert+{\displaystyle\sum\limits_{\left\vert \left [ 5^{1 } \right ] _
{ i } \right\vert=1 } } \left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(3 ) } \right\vert=$ ] @xmath372 \right\vert(n-2)+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert(n-3)+\left\vert \left [ 5^{1 } \right ] \right\vert(n-2).$ ] hence , using @xmath372 \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert+\left\vert \left [ 5^{1 } \right ] \right\vert=2n,$ ] see ( [ d ] ) , we get @xmath490 \right\vert=2n(n-2)-\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert;\label{eee1}\ ] ] @xmath491 \right\vert={\displaystyle\sum\limits_{i\in i } } \left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert={\displaystyle\sum\limits_{\left\vert \left [ 4^{1 } \right]_{i } \right\vert=1 } } \left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert+{\displaystyle\sum\limits_{\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert=1 } } \left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert+{\displaystyle\sum\limits_{\left\vert \left [ 5^{1 } \right ] _ { i } \right\vert=1 } } \left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) }
\right\vert=$ ] @xmath372 \right\vert(2n-6)+\left\vert \left
[ 4^{1 } , 1^{1 } \right ] \right\vert(2n-5)+\left\vert \left [ 5^{1 } \right ] \right\vert(2n-4)-{\displaystyle\sum\limits_{i\in i } } \left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert=2n(2n-6)+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert+2\left\vert \left [ 5^{1 } \right ] \right\vert-\left\vert \left [ 3^{1},2^{1 } \right ] \right\vert$ ] as , see , @xmath385 \right\vert=\left\vert \left [ 4^{1 } \right ] \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert.$ ] thus @xmath492 \right\vert = n(2n-6)+\left\vert \left [ 5^{1 } \right ] \right\vert-\frac{\left\vert \left [ 4^{1 } \right ] \right\vert}{2}.\label{eee2}\ ] ] substituting into ( [ g ] ) for @xmath493 \right\vert,\left\vert \left [ 1^{3 } \right ] \right\vert,\left\vert \left [ 2^{1},1^{2 } \right ] \right\vert$ ] from theorem [ 7 ] and for @xmath489 \right\vert$ ] and @xmath491 \right\vert$ ] from above we get @xmath494 \right\vert+ \\
3\left\vert \left [ 2^{1 } , 1^{3 } \right ] \right\vert+2n(2n-7)+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert+3\left\vert \left [ 5^{1 } \right ] \right\vert = 24\binom{n}{3 } , \end{gathered}\ ] ] substituting for @xmath364_i^{(2 ) } \right\vert,\left\vert \left [ 1^{3 } \right ] _ { i } \right\vert,\left\vert \left [ 2^{1 } , 1^{2 } \right]_{i}^{(2 ) } \right\vert$ ] from theorem [ 7 ] , to ( [ g1 ] ) yields @xmath489 _ { i}^{(3 ) } \right\vert+3\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { i}^{(2 ) } \right\vert+\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert=4\binom{n-1}{2}-4(n-2)=2(n-2)(n-3).$ ] applying ( [ r1 ] ) , ( [ r2 ] ) , and ( [ r3 ] ) in turn implies , if @xmath372_{i } \right\vert=1 $ ] or @xmath466 _ { i}^{(4 ) } \right\vert=1 , $ ] then @xmath496 _ { i}^{(2 ) } \right\vert=2(n-2)(n-3)-3n+8+\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert,\label{lambda}\ ] ] and if @xmath465 _ { i } \right\vert=1,$ ] then @xmath496 _ { i}^{(2 ) } \right\vert=2(n-2)(n-3)-3n+6+\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert.\label{lambda1}\ ] ] as @xmath327 also @xmath497 thus , for @xmath372_{i } \right\vert=1 $ ] and @xmath493 _ { i}^{(4 ) } \right\vert=1,$ ] we have @xmath385_{i}^{(2 ) } \right\vert=1(\operatorname{mod}3),$ ] while for @xmath465 _ { i } \right\vert=1 $ ] we get @xmath385_{i}^{(2 ) } \right\vert=0(\operatorname{mod}3).$ ] by lemma [ l3 ] , @xmath385_{i}^{(2 ) } \right\vert\leq2 $ ] for all @xmath365 hence , for @xmath372_{i } \right\vert=1 $ ] and @xmath466 _ { i}^{(4 ) } \right\vert=1,$ ] we have @xmath385_{i}^{(2 ) } \right\vert=1,$ ] and for @xmath465 _ { i } \right\vert=1 $ ] we get @xmath385_{i}^{(2 ) } \right\vert=0.$ ] again , by lemma [ l3 ] , we have the same conclusion for @xmath385_{i}^{(3 ) } \right\vert.$ ] thus , @xmath498_{i}^{(3 ) } \right\vert=\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert\label{gama}\ ] ] for all @xmath365 substituting into ( [ g2 ] ) for @xmath367 _ { i}^{(1 ) } \right\vert,\left\vert \left [ 1^{3 } \right ] _
{ i } \right\vert,$ ] and @xmath342 _ { i}^{(1 ) } \right\vert$ ] from theorem [ 7 ] yields @xmath489 _ { i}^{(1 ) } \right\vert+2\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { i}^{(1 ) } \right\vert+\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert+2\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(1 ) }
\right\vert=4(n-1)(n-2)-8(n-2)=4(n-2)(n-3).$ ] if @xmath372_{i } \right\vert=1 $ ] or @xmath466 _ { i}^{(4 ) } \right\vert=1,$ ] then in this case @xmath385_{i}^{(2 ) } \right\vert=1 $ ] , and by ( [ r1 ] ) , ( [ r2 ] ) , we get @xmath499 _ { i}^{(1 ) } \right\vert+2\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert=4(n-2)(n-3)-4n+11+\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert.\ ] ] thus , @xmath488 _ { i}^{(1 ) } \right\vert$ ] is odd for @xmath372_{i } \right\vert=1 $ ] and for @xmath466 _ { i}^{(4 ) }
\right\vert=1.$ ] however , @xmath500 _ { i}^{(1 ) } \right\vert={\displaystyle\sum\limits_{i\in i } } \left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(4 ) } \right\vert$ ] and @xmath466 _ { i}^{(4 ) } \right\vert\leq1.$ ] therefore , for all @xmath258@xmath501 _ { i}^{(4 ) } \right\vert=\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert,\label{beta}\ ] ] which in turn implies @xmath372_{i } \right\vert=0 $ ] for all @xmath502 hence @xmath372 \right\vert=0 $ ] . combining @xmath404 \right\vert=0 $ ] with ( [ a ] ) yields
@xmath424 \right\vert=2n,$ ] which in turn implies @xmath373 \right\vert=0,$ ] see ( [ ee ] ) .
the proof is complete . in this subsection
we deal with the number of codewords of individual types of absolute value equal to 5 .
first we summarize results for all @xmath327 then we concentrate on the case @xmath503 . for @xmath327
all these values are expressed as a function of @xmath487 \right\vert$ ] , the number of codewords of type @xmath476.$ ] we point out that for some @xmath231 , there are two tilings of @xmath2 by crosses with different number of codewords of type @xmath476.$ ] hence , unlike with codewords of absolute value @xmath313 and @xmath504 the values of @xmath505\right\vert , \left\vert [ 4^{1},1^{1}]\right\vert , \left\vert [ 3^{1},2^{1}]\right\vert , \left\vert [ 3^{1},1^{2}]\right\vert , \left\vert [ 2^{2},1^{1}]\right\vert , \left\vert [ 2^{1},1^{3}]\right\vert , $ ] and @xmath506\right\vert $ ] do not depend only on the value of @xmath0 but also on a given tiling of @xmath507by crosses .
[ 9]let @xmath42 for each codeword @xmath50 in @xmath34 we have : @xmath508\right\vert = 2n-\left\vert [ 5^{1}]\right\vert , \left\vert [ 3^{1},2^{1}]\right\vert = 2n-\left\vert [ 5^{1}]\right\vert , \left\vert [ 3^{1},1^{2}]\right\vert = 2n(n-3)+\left\vert [ 5^{1}]\right\vert , $ ] @xmath509\right\vert = n(2n-6)+\left\vert [ 5^{1}]\right\vert , \left\vert [ 2^{1},1^{3}]\right\vert = \frac{1}{3}2n(n-3)(2n-7)-\left\vert [ 5^{1}]\right\vert , $ ] and @xmath510\right\vert = \frac{1}{5}(2^{4}\binom{n}{4}-n(n-3)(3n-8)+\left\vert [ 5^{1}]\right\vert ) .$ ] we have proved in the previous theorem that @xmath372 \right\vert=0.$ ] therefore , by ( [ d ] ) , we have @xmath487 \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert=2n.$ ] in addition , by lemma [ l5 ] , it is @xmath385 \right\vert=\left\vert \left [ 4^{1 } \right ] \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert,$ ] that is @xmath385 \right\vert=\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert=2n-\left\vert \left [ 5^{1 } \right ] \right\vert.$ ] it is @xmath489 \right\vert=2n(n-2)-\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert,$ ] see([eee1 ] ) , hence @xmath487 \right\vert+\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert=2n$ ] implies @xmath489 \right\vert=2n(n-3)+\left\vert \left [ 5^{1 } \right ] \right\vert,$ ] while ( [ eee2 ] ) implies @xmath511 \right\vert = n(2n-6)+\left\vert \left [ 5^{1 } \right ] \right\vert.$ ] the value of @xmath512 \right\vert$ ] is given by ( [ i ] ) .
finally , the value of @xmath513 \right\vert$ ] follows from ( [ h ] ) after substituting for @xmath514 \right\vert$ ] and @xmath515 \right\vert$ ] from theorem [ 5 ] .
the proof is complete .
[ 10]let @xmath42 then for each codeword in @xmath34 we have : if @xmath465 _ { i } \right\vert=1,$ ] then @xmath488 _ { i } \right\vert=0$],@xmath516 _ { i } \right\vert=0,$ ] @xmath489 _ { i}^{(1 ) } \right\vert=2\left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(3 ) } \right\vert=2(n-2),\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(2 ) } \right\vert=2\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert=2(n-2),\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { i}^{(1 ) } \right\vert=3\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { i}^{(2 ) } \right\vert=(n-2)(2n-9).$ ] if @xmath465 _ { i } \right\vert=0 $ ] , then @xmath466 _ { i}^{(4 ) } \right\vert=\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert=1,\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert=\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert=1,\left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(1 ) } \right\vert=2\left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { i}^{(3 ) } \right\vert=2(n-3),$ ] @xmath511 _ { i}^{(2 ) } \right\vert=2\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert=2(n-3),\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { i}^{(1 ) } \right\vert=3\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { i}^{(2 ) } \right\vert=(n-3)(2n-7\dot{)}.$ ] in both cases @xmath517 _ { i } \right\vert=$ ] @xmath518 _ { i } \right\vert-\frac{10}{3}(n-2)(n-3)).$ ] let @xmath465 _ { i } \right\vert=1.$ ] then by , lemma [ l5 ] , @xmath385_{i}^{(3 ) } \right\vert=0,$ ] and by ( [ gama ] ) @xmath385_{i}^{(2 ) } \right\vert=0;$ ] also @xmath466 _ { i}^{(4 ) } \right\vert=0,$ ] and ( [ gama ] ) implies @xmath466 _ { i}^{(i ) } \right\vert=0.$ ] hence @xmath385 \right\vert=\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert=0.$ ] using the same arguments in the case @xmath465 _ { i } \right\vert=0 $ ] yields @xmath466 _ { i}^{(4 ) } \right\vert=\left\vert \left [ 4^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert=1,\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(3 ) } \right\vert=\left\vert \left [ 3^{1 } , 2^{1 } \right]_{i}^{(2 ) } \right\vert=1.$ ] with this in hand , the values of @xmath519 _ { i } \right\vert$ ] and @xmath520 _ { i } \right\vert$ ] follow from ( [ r3 ] ) and ( [ r2 ] ) , while the values of @xmath512 _ { i } \right\vert$ ] are obtained from ( [ lambda ] ) and ( [ lambda1 ] ) . finally , to determine @xmath513 _ { i } \right\vert$ ] it suffices to substitute into ( [ h1 ] ) .
the proof is complete .
we showed above , that the number of codewords in @xmath34 of absolute value @xmath313 and @xmath221 does depends only on @xmath521 while the number of codewords of absolute value @xmath194 depends also on the tiling @xmath34 .
however , for @xmath503 also all these values are constant .
[ 11 ] if @xmath41 then @xmath404 \right\vert=\left\vert \left [ 4^{1 } \right ] \right\vert=\left\vert \left [ 2^{2 } \right ] \right\vert=\left\vert \left [ 5^{1 } \right ] \right\vert=0,\left\vert \left [ 2^{1 } , 1^{1 } \right ] \right\vert=\left\vert \left [ 1^{3 } \right ] \right\vert=\left\vert \left [ 3^{1 } , 1^{1 } \right ] \right\vert=\left\vert \left [ 1^{4 } \right ] \right\vert=\left\vert \left [ 4^{1 } , 1^{1 } \right ] \right\vert=\left\vert \left [ 3^{1 } , 2^{1 } \right ] \right\vert=10,$ ] @xmath489 \right\vert=\left\vert \left [ 2^{2 } , 1^{1 } \right ] \right\vert=\left\vert \left [ 2^{1 } , 1^{3 } \right ] \right\vert=20,$ ] while @xmath522 \right\vert=30,$ ] and @xmath517 \right\vert=2,$ ] and @xmath517 _
{ i } \right\vert=1 $ ] for all @xmath365 the values of @xmath523 \right\vert,\left\vert \left [ 2^{1 } , 1^{1 } \right ] \right\vert,\left\vert \left [ 1^{3 } \right ] \right\vert,\left\vert \left [ 4^{1 } \right ] \right\vert,\left\vert \left [ 3^{1 } , 1^{1 } \right ] \right\vert,\left\vert \left [ 2^{2 } \right ] \right\vert,\left\vert \left [ 2^{1 } , 1^{2 } \right ] \right\vert,$ ] and @xmath515 \right\vert$ ] are obtained from theorem [ 6 ] , [ 7 ] , and [ 8 ] by substituting @xmath503 .
the other values depend on the value of @xmath487 \right\vert,$ ] see theorem [ 9 ] .
we will prove that there is no codeword of type @xmath476,$ ] i.e. , that @xmath487 \right\vert=0.$ ] in order to do it we have to consider not only local equalities but also so - called double - local equalities for the individual type of codewords . to be able to introduce these
we need one more piece of notation .
let @xmath96 be a set of codewords .
then , for @xmath524 we denote by @xmath525 the number of codewords @xmath526 in @xmath96 so that @xmath527 for an ordered pair @xmath528 we denote by @xmath529 the number of codewords @xmath526 in @xmath96 with @xmath530 and @xmath531 substituting @xmath503 into theorem [ 9 ] and into theorem [ 10 ] yields @xmath517 \right\vert=2+\frac{\left\vert \left [ 5^{1 } \right ] \right\vert}{5},$ ] and @xmath513 _ { i } \right\vert>0 $ ] for all @xmath258 respectively . assume by contradiction that @xmath487 \right\vert>0.$ ] then we have @xmath517 \right\vert \geq3,\,$ ] and@xmath60at least two codewords of type @xmath532 $ ] have to coincide in at least two coordinates . thus , there have to be signed coordinates @xmath533 such that @xmath517_{ij } \right\vert\geq2.$ ] to reject the assumption of @xmath487 \right\vert>0 $ ] we prove that @xmath517 _ { ij } \right\vert\leq1 $ ] for all @xmath534 we will start by setting several double - local equalities . for each @xmath533 there is a unique word @xmath51 in @xmath535 of type @xmath278 $ ] with @xmath536 therefore , see also the explanation to ( [ pp ] ) , @xmath433 _ { ij } \right\vert+\left\vert \left [ 1^{3 } \right ] _ { ij } \right\vert=1\label{x0}\ ] ] for each @xmath533 there are six words @xmath51 in @xmath535 of type @xmath294 $ ] with @xmath536 therefore , see also ( [ c ] ) , @xmath537 _ { ij } \right\vert+\left\vert \left [ 2^{1 } , 1^{2 } \right ] _ { ij } \right\vert+2\left\vert \left [ 1^{4 } \right ] _ { ij } \right\vert=6\label{x1}\ ] ] clearly , it is not difficult to see that @xmath342 _ { ij } \right\vert=\left\vert \left [ 2^{1 } , 1^{2 } \right ] _ { ij}^{(2,1 ) } \right\vert+\left\vert \left [ 2^{1 } , 1^{2 } \right ] _ { ij}^{(1,2 ) } \right\vert \\+ \left\vert \left [ 2^{1 } , 1^{2 } \right ] _ { ij}^{(1,1 ) } \right\vert\leq3.$ ] indeed , @xmath342 _ { ij}^{(2,1 ) } \right\vert>1 $ ] ( @xmath342 _ { ij}^{(1,2 ) } \right\vert>1 $ ] ) would imply that there are in @xmath34 two codewords of type @xmath443 $ ] of distance @xmath538 a contradiction .
finally , @xmath342 _ { ij}^{(1,1 ) } \right\vert>1 $ ] would imply , that there are in @xmath34 two codewords of type @xmath443 $ ] coinciding in two coordinates where their value is @xmath539 say @xmath540 and @xmath541 or @xmath542 however , in the former case @xmath543 and in the latter @xmath544 , thus @xmath286 would be with respect to @xmath280 a codeword of type @xmath298,$ ] and of type @xmath349,$ ] respectively , which contradicts @xmath372 \right\vert=\left\vert \left [ 2^{2 } \right ] \right\vert=0,$ ] see theorem [ 8 ] . combining ( [ x0 ] ) with ( [ x1 ] )
we get : now we state a double - local equality for codewords of type @xmath308.$ ] for each @xmath524 there are twelve words @xmath51 in @xmath535 of type @xmath308 $ ] with @xmath545 using arguments similar to proving ( [ h ] ) yields:@xmath462 _ { ij}^{(0,1 ) } \right\vert+\left\vert \left [ 1^{3 } \right ] _ { ij}^{(1,0 ) } \right\vert+4\left\vert \left [ 1^{3 } \right ] _ { ij } \right\vert+\left\vert \left [ 1^{4 } \right ] _ { ij } \right\vert+\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { ij } \right\vert+\\ 3\left\vert \left [ 1^{5 } \right ] _ { ij } \right\vert=12,\label{x2}\end{gathered}\ ] ] where @xmath463 _ { ij}^{(0,1 ) } \right\vert$ ] is the number of codewords @xmath162 of type @xmath294 $ ] such that @xmath546 and @xmath547 for each @xmath524 we have @xmath463 _ { ij } \right\vert\leq1,$ ] otherwise there would be two codewords of type @xmath294 $ ] at distance less than @xmath297 therefore , for all @xmath524 there is at most one codeword @xmath51 of type @xmath294 $ ] with @xmath545 or @xmath548 or @xmath549 as @xmath514 _ { i } \right\vert=3,$ ] we get that both @xmath463 _ { ij}^{(0,1 ) } \right\vert$ ] @xmath550 and @xmath463 _
{ ij}^{(1,0 ) } \right\vert\geq2 $ ] for @xmath463 _ { ij } \right\vert=0,$ ] and both @xmath463 _ { ij}^{(0,1 ) } \right\vert$ ] @xmath551 and @xmath463 _ { ij}^{(1,0 ) } \right\vert\geq1 $ ] if @xmath463 _ { ij } \right\vert=1.$ ] in aggregate , if @xmath463 _ { ij } \right\vert=1,$ ]
we get @xmath463 _ { ij}^{(0,1 ) } \right\vert+\left\vert \left [ 1^{3 } \right ] _ { ij}^{(1,0 ) } \right\vert+4\left\vert \left [ 1^{3 } \right ] _ { ij } \right\vert\geq6,$ ] and @xmath463 _ { ij}^{(0,1 ) } \right\vert+\left\vert \left [ 1^{3 } \right ] _ { ij}^{(1,0 ) } \right\vert+4\left\vert \left [ 1^{3 } \right ] _ { ij } \right\vert\geq4 $ ] for @xmath463 _ { ij } \right\vert=0.$ ] @xmath552 _ { ij } \right\vert+3\left\vert \left [ 1^{5 } \right ] _ { ij } \right\vert\leq5\text { for } \left\vert \left [ 1^{3 } \right ] _ { ij } \right\vert=1 , \text { and } \\ \left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { ij } \right\vert+3\left\vert \left [ 1^{5 } \right ] _ { ij } \right\vert\leq6\text { for } \left\vert \left [ 1^{3 } \right ] _ { ij } \right\vert=0 .
\end{gathered}\ ] ] thus , @xmath517 _ { ij } \right\vert\leq1 $ ] for @xmath463 _ { ij } \right\vert=1.$ ] to prove that @xmath517 _ { ij } \right\vert\leq1 $ ] also when @xmath463 _ { ij } \right\vert=0,$ ] we will show that in this case @xmath553 _ { ij } \right\vert>0.$ ] if @xmath463
_ { ij } \right\vert=0,$ ] then @xmath367 _ { ij } \right\vert=1,$ ] see ( [ p1 ] ) .
let @xmath286 be the codeword of type @xmath208 $ ] having non - zero the @xmath69-th and the @xmath103-th sign coordinate@xmath554 then either @xmath293 and @xmath555 or @xmath556 and @xmath474 assume that the former is the case .
in @xmath535 there are six words @xmath51 of type @xmath443 $ ] with @xmath275 and @xmath557 each of these is covered either ( i ) by a codeword @xmath50 of type @xmath446 $ ] with @xmath558 or ( ii ) by a codeword @xmath50 of type @xmath446 $ ] with @xmath559 or ( iii ) by a codeword @xmath50 of type @xmath294 $ ] with @xmath560 or ( iv ) by a codeword @xmath50 of type @xmath561 $ ] with @xmath559 or ( v ) by a codeword @xmath50 of type @xmath562 $ ] with @xmath563 or ( vi ) by a codeword @xmath50 of type @xmath564 $ ] with @xmath565 or ( vii ) by a codeword @xmath50 of type @xmath564 $ ] with @xmath559 or ( viii ) by a codeword @xmath50 of type @xmath566 $ ] with @xmath567 using our notation we can write @xmath437 _ { ij}^{(2,0 ) } \right\vert+6\left\vert \left [ 2^{1 } , 1^{1 } \right ] _ { ij}^{(2,1 ) } \right\vert+\left\vert \left [ 1^{3 } \right ] _ { ij } \right\vert+\left\vert \left [ 2^{1 } , 1^{2 } \right ] _ { ij}^{(2,1 ) } \right\vert+\left\vert \left [ 3^{1 } , 1^{2 } \right ] _ { ij}^{(3,1 ) } \right\vert\\ + \left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { ij}^{(2,2 ) } \right\vert+\left\vert \left [ 2^{2 } , 1^{1 } \right ] _ { ij}^{(2,1 ) } \right\vert+\left\vert \left [ 2^{1 } , 1^{3 } \right ] _ { ij}^{(2,1 ) } \right\vert=6 . \end{gathered}\ ] ] we have chosen @xmath568 so that @xmath367 _ { ij}^{(2,1 ) } \right\vert=\left\vert \left [ 1^{3 } \right ] _ { ij } \right\vert=0.$ ] further , it is easy to see that in each of the cases ( i ) and ( iv)-(vi ) there is at most one codeword of each type , as otherwise we would have two codewords at distance less than @xmath313 . as to ( vii ) , we have @xmath511 _ { ij}^{(2,1 ) } \right\vert\leq1 $ ] as well , because if there were two codewords @xmath50 of type @xmath569 $ ] with @xmath570 and @xmath571 then their difference would be a codeword either of type @xmath298 $ ] or @xmath349,$ ] a contradiction .
so we get @xmath553 _ { ij}^{(2,1 ) } \right\vert\geq1 $ ] in this case , thus @xmath553 _ { ij } \right\vert\geq1,$ ] and in turn @xmath517 _ { ij } \right\vert\leq1 $ ] also in the case @xmath463 _ { ij } \right\vert=0.$ ] the proof is complete .
let @xmath51 be a word in @xmath572 then by @xmath573 we denote the collection of words comprising @xmath574 and the words obtained by cyclic shifts of coordinates of @xmath51 .
hence , e.g. , @xmath575@xmath576 @xmath577we note that @xmath573 contains five words except for the case @xmath51 @xmath578 finally , we set @xmath579 @xmath580 @xmath581 by the canonical @xmath194- neighborhood , or simply a canonical neighborhood , we mean the set of words@xmath582 @xmath583 @xmath584 a simple inspection shows that the number of words of individual types in the canonical neighborhood coincides with the values given by theorem [ 11 ] .
e.g. , @xmath585 is the set of ten words of type @xmath287,$ ] ( @xmath367 \right\vert=10 $ ] ) , while @xmath586 comprises ten words of type @xmath294 $ ] ( @xmath463 \right\vert=10 $ ] ) .
[ 12 ] let @xmath34 be a tiling of @xmath6 by crosses .
then , for each codeword @xmath50 in @xmath34 , @xmath60the @xmath194-neighbourhood of @xmath50 is congruent to the canonical one .
moreover , the @xmath194-neighborhood of @xmath50 is uniquely determined by the set of codewords of type @xmath446 $ ] . as in other proofs in this paper
we assume w.l.o.g . that @xmath344 two words @xmath587 will be called sign equivalent in the @xmath103-th
coordinate if @xmath588that is , they are sign equivalent if @xmath589 and the two non - zero values have the same sign .
it was proved in theorem [ 11 ] that @xmath517 \right\vert=2,$ ] and @xmath517 _
{ i } \right\vert=1 $ ] for each @xmath590 i.e. , the two codewords of type @xmath532 $ ] differ in each coordinate (= are not sign equivalent in any coordinate ) .
that is , if @xmath591 is a codeword of type @xmath532,$ ] then @xmath592 is the other codeword of type @xmath532.$ ] let @xmath593 be a set of codewords of type @xmath446,$ ] and @xmath594 be the set of codewords of type @xmath294 $ ] .
we know by theorem [ 6 ] that @xmath595 .
there are in total @xmath596 words of type @xmath278 $ ] that are sign equivalent in two coordinates with @xmath591 and another @xmath596 words of type @xmath278 $ ] that are sign equivalent in two coordinates with @xmath597 each word of type @xmath278 $ ] is covered by a codeword in @xmath598 , thus each of these @xmath599 words is covered by a codeword in @xmath598 .
if a codeword @xmath162 in @xmath594 covered two of these @xmath599 words , then @xmath162 would be sign equivalent in three coordinates with @xmath591 or @xmath600 and the distance of @xmath162 to one of @xmath591 or @xmath592 would be less than @xmath313 .
therefore , each codeword in @xmath594 covers at most one of these @xmath599 words of type @xmath278 $ ] .
hence , each codeword in @xmath593 has to cover one of these @xmath599 words , that is , each codeword of type @xmath446 $ ] is sign equivalent in both non - zero coordinates either with the codeword @xmath591 or with the codeword @xmath592 .
we know by theorem [ 6 ] that , for each @xmath258 @xmath364_i^{(2 ) } \right\vert=\left\vert \left [ 2^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert.$ ] thus , five codewords in @xmath593 are sign equivalent in two coordinates with @xmath601 the other five with @xmath592 .
it turns out that graph theory has a very suitable language to describe the structure of the set @xmath593 .
let @xmath73 be a graph with the vertex set @xmath602 the set of signed coordinates , and the edges of @xmath73 be all pairs of vertices in @xmath603 except for @xmath604 @xmath605 thus @xmath73 is a complete graph on @xmath596 vertices , @xmath606 with a @xmath65-factor ( = perfect matching ) removed .
we denote this one factor by @xmath28 so @xmath607 in @xmath535 there are forty words of type @xmath608 .$ ] in a natural way each word of type @xmath245 $ ] is associated with an edge of @xmath609 if @xmath51 is a word of type @xmath245 $ ] with @xmath545 ( and then @xmath610 for all @xmath611 ) then we assign to @xmath51 the edge @xmath612 of @xmath609 so there is a one - to - one correspondence between words of type @xmath245 $ ] and the edges of @xmath609 in addition , each codeword @xmath613 is associated with the edge ( word of type @xmath278)$ ] covered by @xmath213 the set of words of type @xmath614 $ ] covered by codewords in @xmath593 will be denoted by @xmath615 the condition @xmath364_i^{(2 ) } \right\vert=\left\vert \left [ 2^{1 } , 1^{1 } \right ] _ { i}^{(1 ) } \right\vert=1 $ ] implies that the words in @xmath616 form a @xmath617-factor , say @xmath618 it follows from the above discussion that @xmath619 consists of two cycles of length @xmath194 such that both cycles contain exactly one vertex of each edge in the matching @xmath591 .
let @xmath162 be the set of five words in @xmath616 constituting one of the two @xmath194-cycles in @xmath619 . clearly , by suitably permuting the order of coordinates of each codeword in @xmath34 and/or changing a sign of a coordinate for each codeword in @xmath34 maps @xmath162 onto @xmath620 and the codewords covering words in @xmath162 onto @xmath621 . @xmath60by claim [ 0 ] the above transformation is a congruence mapping .
therefore we will assume that @xmath621 are codewords in @xmath34 .
if one of the two @xmath194-cycles of @xmath616 is @xmath620 then the other one is formed by codewords having both non - zero coordinates negative .
there are four non - isomorphic ways how to choose it .
it is either the @xmath194-cycle comprising the edges corresponding to @xmath625 or to @xmath626 or@xmath627 @xmath628 , or@xmath629 @xmath630 the graph consisting of the edges of the @xmath617-factor @xmath619 and the matching @xmath591 is a cubic graph .
the four cubic graph corresponding to the cases described above are depicted in figure 1 .
the first one is the prism on @xmath596 vertices , the second is the petersen graph , the labels for the other two are taken from @xcite . .
each codeword of type @xmath294 $ ] covers three words of type @xmath278,$ ] so we associate with each codeword of type @xmath294 $ ] a triangle ( cycle of length @xmath313 ) in the graph @xmath73 . as mentioned many times , all words of type @xmath278 $ ]
are covered by codewords in @xmath631 therefore , the triangles corresponding to codewords in @xmath594 have to form an edge decomposition of the complement of the cubic graph consisting of the edges of the @xmath617-factor @xmath619 and the matching @xmath28 it is known , see @xcite , that the complement of the two cubic graphs @xmath632 and @xmath633 is not decomposable into triangles .
therefore , in the case of @xmath632 and @xmath633 it is impossible to choose the codewords of type @xmath294.$ ] thus , we are left with the prism and the petersen graph .
these two decomposition are isomorphic but we will we need to consider both of them , as the automorphism of the graph @xmath637 which maps one decomposition on the other , maps the set @xmath585 of codewords of type @xmath446 $ ] on the set @xmath638 let @xmath643 be the set of ten codewords in @xmath34 of type @xmath219 .$ ] clearly , @xmath644 where @xmath645 is constructed as follows : first let @xmath645 be the set of all eighty words in @xmath535 of type @xmath219 .$ ] as any codeword in @xmath643 has to have a distance at least @xmath313 from both codewords of type @xmath381 , $ ] and all ten codewords of type @xmath244 , $ ] we delete from @xmath645 all codewords that coincide with a codeword in all four non - zero coordinates , or with a codeword in @xmath594 in three non - zero coordinates .
after these two procedures there are exactly thirty words left in @xmath646 there are twenty words of type @xmath278 $ ] with both non - zero coordinates of the same sign .
ten of them are covered by codewords in @xmath593 , the other ten by the codewords in @xmath594 .
further , by claim b stated in the proof of theorem [ 11 ] , it is @xmath371 _ { ij } \right\vert\geq2 $ ] for @xmath647 such that @xmath367 _ { ij } \right\vert=1,$ ] and @xmath371 _ { ij } \right\vert\geq1 $ ] for @xmath463 _ { ij } \right\vert=1 $ ] .
hence , to each codeword @xmath50 in @xmath648 there are in @xmath643 at least two codewords sign equivalent with @xmath50 in two coordinates .
since no codeword in @xmath643 has all non - zero coordinates of the same sign ( otherwise its distance to a codeword of type @xmath532 $ ] would be less than @xmath313 ) , and the codewords in @xmath593 form two @xmath194-cycles , no codeword in @xmath643 can be sign equivalent in two coordinates with three codewords in @xmath593 .
this in turn implies , because @xmath649 that * claim c. * each codeword in @xmath643 has to be sign equivalent in two coordinates with two codewords of type @xmath446 $ ] and with one word of type @xmath278 $ ] with both non - zero coordinates of the same sign . in particular , each codeword in @xmath643 has three coordinates of the same sign . as in all cases
@xmath650 , @xmath643 has to contain a codeword @xmath651 where exactly one of @xmath652 equals @xmath653 and the other equals @xmath654 and all cyclic shifts of coordinates of @xmath213 ( ia ) first we describe the set @xmath646 at the beginning of the process@xmath655 @xmath656 we have to remove from @xmath645 words @xmath657 that have the distance from the codewords @xmath658 of type @xmath532 $ ] less than @xmath297 next , in this case , the set @xmath594 of codewords of type @xmath294 $ ] is @xmath659 therefore we need to remove from @xmath645 forty words at distance less than @xmath313 from any codeword in @xmath594 . these words are @xmath660 and @xmath661 thus , at the end of the process @xmath662 clearly , the only way how to choose a set of ten codewords satisfying claim c is to set @xmath663 ( iib ) in this case all words in @xmath645 that belong to @xmath674 are:@xmath675 @xmath676@xmath677 it is impossible to choose from this set three words of type @xmath678 and @xmath679 that would be pair - wise at distance at least @xmath297 therefore in this case it is impossible to choose a required set of codewords of type @xmath308,$ ] and we do not need to consider this case any longer .
in @xmath535 there are eighty words of type @xmath294.$ ] by ( [ c ] ) , ten of them are covered by codewords in @xmath594 of type @xmath294,$ ] forty of them by codewords in @xmath643 of type @xmath308,$ ] and the set @xmath680of the remaining thirty words covered by the set @xmath681 of codewords of type @xmath443 $ ] .
clearly , if a codeword @xmath50 of type @xmath443 $ ] covers a word @xmath51 in @xmath682 then we can see the codeword @xmath50 as obtained from @xmath51 by multiplying one of the non - zero coordinates of @xmath51 by two .
further , in @xmath535 there are eighty words of type @xmath308.$ ] by ( [ h ] ) , forty of them are covered by codewords in @xmath594 , ten of them by codewords in @xmath683 ten by codewords of type @xmath532 , $ ] and the remaining twenty , belonging to the set @xmath684 , by the set @xmath685 of codewords of type @xmath448 $ ] .
as above , if a codeword @xmath50 in @xmath685 covers a word @xmath51 in @xmath684 , then we can see @xmath50 as obtained from @xmath51 by multiplying one of the non - zero coordinates of @xmath51 by two .
( ia ) since @xmath686 and @xmath687 it is @xmath688@xmath689 and @xmath690@xmath665 we need to consider two possible choices of codewords of type @xmath446 $ ] covering the other @xmath194-cycle @xmath691 of the @xmath617-factor @xmath618 it is either @xmath692 or @xmath693 .
[ c]rrrrrr@xmath694 & @xmath695 & @xmath654 & @xmath418 & @xmath418 & @xmath696 + @xmath697 & @xmath418 & @xmath418 & @xmath538 & @xmath539 & @xmath696 + @xmath698 & @xmath539 & @xmath654 & @xmath539 & @xmath539 & @xmath696 + @xmath699 & @xmath654 & @xmath654 & @xmath539 & @xmath539 & @xmath696 + @xmath700 & @xmath539 & @xmath418 & @xmath539 & @xmath539 & @xmath696 + @xmath701 & @xmath653 & @xmath654 & @xmath539 & @xmath539 & @xmath696 + @xmath702 & @xmath539 & @xmath654 & @xmath539 & @xmath418 & @xmath696 clearly , @xmath703 are codewords of type @xmath446,$ ] while @xmath704 , and @xmath705 . to get a codeword @xmath706 covering the word @xmath707@xmath708 we need to multiply a non - zero coordinate of @xmath709 by @xmath617 .
as there are five words @xmath707 and all of them have the fifth coordinate equal to zero , two of them have to have the same non - zero coordinate multiplied by @xmath617 .
it is possible only for @xmath710 and @xmath711 if their forth coordinate is chosen , as otherwise the two resulting codewords would be at distance less than @xmath313 ( note that multiplying the first coordinate of @xmath710 by @xmath617 results in a codeword at distance less than @xmath313 from @xmath712 because of the distance to the codeword @xmath713 only the word @xmath714 can have its third coordinate multiplied by @xmath617 .
this implies that @xmath715 has to have its second coordinate multiplied by @xmath617 while for @xmath716the first one is the only choice .
the same type of argument can be applied to a set of words / codewords obtained by the above one by cyclically shifting coordinates of each word / codeword and/or multiplying all words / codewords by @xmath717 therefore setting@xmath718 and@xmath719 which is identical to@xmath720and @xmath721 is the unique choice so that all codewords in @xmath593 , @xmath681 , and @xmath685 are pair - wise at distance at least @xmath313 .
[ c]rrrrrr@xmath723 & @xmath418 & @xmath418 & @xmath654 & @xmath695 & @xmath696 + @xmath698 & @xmath654 & @xmath539 & @xmath654 & @xmath654 & @xmath696 + @xmath699 & @xmath654 & @xmath418 & @xmath654 & @xmath654 & @xmath696 + @xmath700 & @xmath654 & @xmath539 & @xmath654 & @xmath418 & @xmath696 + @xmath701 & @xmath418 & @xmath539 & @xmath654 & @xmath654 & @xmath696 it is easy to check that @xmath724 and @xmath725 @xmath726 while @xmath286 is a codeword . let @xmath706 be a codeword covering the word @xmath727 as mentioned above , @xmath706 can be viewed as obtained from @xmath728by multiplying one non - zero coordinate of @xmath709 by @xmath617 .
it is easy to see that , for @xmath729 if @xmath706 and @xmath730 were obtained by multiplying the same coordinate of @xmath709 and @xmath731 by @xmath617 than their distance would be less than @xmath313 . on the other hand ,
if , for some @xmath732 the fourth coordinate of @xmath706 equaled to @xmath733 , then the distance of @xmath706 to @xmath286 would be less than three .
thus in this case , codewords covering @xmath734 do not exist .
[ c]rrrrrr@xmath723 & @xmath538 & @xmath539 & @xmath418 & @xmath418 & @xmath696 + @xmath698 & @xmath539 & @xmath539 & @xmath654 & @xmath539 & @xmath696 + @xmath699 & @xmath539 & @xmath539 & @xmath418 & @xmath539 & @xmath696 + @xmath700 & @xmath539 & @xmath539 & @xmath654 & @xmath418 & @xmath696 + @xmath701 & @xmath418 & @xmath539 & @xmath654 & @xmath539 & @xmath696 then @xmath739 @xmath740 by the same type of an argument as in ( ia ) it is easy to see that the required codewords @xmath741 do not exist , as multiplying the first coordinate of @xmath709 by @xmath617 leads to a codeword at distance less than @xmath313 from @xmath286 .
moreover , our example shows that in this case ( ib ) it is impossible to choose the sets of codewords of type @xmath742,$ ] and @xmath448,$ ] regardless whether @xmath743 or @xmath744 is in @xmath593 .
( iia ) we show that also in this case it is impossible to choose the sets of codewords of type @xmath443,$ ] and @xmath448.$ ] there are two ways how to choose codewords in @xmath593 of type @xmath446 $ ] covering words @xmath745 either @xmath746 or @xmath747 there are four ways how to choose the set @xmath643 of codewords of type @xmath308.$ ] we treat here only one of them as the other three are nearly identical to this one .
we deal with the option @xmath748 then @xmath749 @xmath750 while@xmath751 of the two cases when
@xmath752 or @xmath753 we treat here the first one .
let @xmath754{rrrrrr}$]b_1=(@xmath7552,@xmath7551,@xmath7550,@xmath7550,@xmath7550)@xmath756b_2=(@xmath7550,@xmath7550,@xmath755 - 2,@xmath7550,@xmath755 - 1)@xmath756v_1=(@xmath7551,@xmath7551,@xmath755 - 1,@xmath7550,@xmath755 - 1)@xmath756v_2=(@xmath7550,@xmath7551,@xmath755 - 1,@xmath7550,@xmath755 - 1)@xmath756v_3=(@xmath7551,@xmath7551,@xmath7550,@xmath7550,@xmath755 - 1)@xmath757 , where @xmath758 and @xmath759 are codewords and @xmath760 , @xmath761@xmath762 , @xmath763 .
because of @xmath758 none of @xmath764s can have the first coordinate multiplied by @xmath538 while because of @xmath759 none of @xmath764s can have its third coordinate multiplied by @xmath262 thus , codewords covering @xmath765 do not exist .
in @xmath535 there are eighty words of type @xmath446.$ ] by ( [ b ] ) , ten of them are covered by the codewords of type @xmath446 $ ] , and sixty by the codewords of type @xmath443.$ ] the remaining ten words @xmath767 are to be covered by codewords of type @xmath299.$ ] thus , @xmath768 are in @xmath34 . in @xmath769there
are @xmath770 words of type @xmath443.$ ] by ( [ g ] ) sixty of them are covered by codewords of type @xmath446,$ ] thirty by codewords of type @xmath294,$ ] another thirty by codewords of type @xmath771 $ ] , and sixty by codewords of type @xmath448.$ ] the remaining sixty words of type @xmath443 $ ] are @xmath772@xmath773 then @xmath774 @xmath775 and @xmath776 @xmath777 as each codeword of type @xmath449 $ ] covers two words of type @xmath443\,$ ] there have to be in @xmath34 codewords @xmath778 and @xmath779 of type @xmath449,$ ] and the remaining twenty words of type @xmath742 $ ] are covered by codewords @xmath780 of type @xmath781 .
$ ] finally , the ten remaining words @xmath782 of type @xmath299 $ ] have to be covered by codewords @xmath783 of type @xmath784\,,$ ] and the ten remaining words@xmath785 of type @xmath349 $ ] by codewords@xmath786 of type @xmath412.$ ] [ 13 ] let @xmath787 be two words from the @xmath194-neighborhood of a codeword @xmath50 .
then @xmath788 is in this neighborhood , that is , the @xmath194-neighborhood is symmetric , and if @xmath789 then @xmath790 belongs to this @xmath194-neighborhood as well . in particular ,
if @xmath791 are from the @xmath194-neighborhood of the origin then @xmath792 and @xmath793 if @xmath794 belong to this neighborhood as well
. again it suffices to prove the statement for @xmath344 from the previous theorem we know , that the @xmath194-neighborhood of each codeword in congruent to the canonical one .
clearly , a congruence mapping retains the properties described in this theorem .
therefore , it suffices to prove the statement for the canonical neighborhood . to show that the canonical neighborhood satisfies these properties we prove that this neighborhood is a part of ( the unique ) lattice tiling of @xmath6 by crosses .
then the proof will follow from the fact that if words @xmath795 belong to a lattice @xmath32 then also @xmath792 and @xmath796 @xmath60are in @xmath32 . consider a homomorphism @xmath797 the cyclic group of order @xmath164 given by @xmath798 and @xmath799 then @xmath76 satisfies the assumptions of corollary [ 20 ] .
thus , @xmath76 induces a lattice tiling @xmath34 of @xmath6 by crosses , where the set @xmath800 is the set of the centers of crosses in @xmath34 .
it is easy , although time consuming , to check , that all codewords from the canonical neighborhood belong to @xmath801 that is , if @xmath802 belongs to the @xmath194-neighborhood then @xmath803 the proof is complete .
let @xmath34 be a tiling of @xmath6 by crosses .
we proved that the @xmath194-neighborhood of any codeword @xmath50 in @xmath34 is congruent to the @xmath194-neighborhood of the origin . by claim
[ 0 ] , we may assume that the @xmath194-neighborhood of the origin is the canonical one .
we have proved , see theorem [ 6 ] , that for each codeword @xmath50 the @xmath313-neighborhood of @xmath50 comprises twenty codewords ; i.e. , there are in @xmath34 twenty codewords at distance @xmath313 from @xmath50 .
we will call these codewords at distance @xmath313 from @xmath50 the codewords adjacent to @xmath213 ten of the adjacent codewords are of type @xmath446,$ ] and ten of them are of type @xmath294.$ ] the proof of the theorem is based on the following claim , which states that all codewords adjacent to @xmath50 have `` the same '' set of codewords of type @xmath446 $ ] as @xmath50 has .
* claim d. * let @xmath50 be a codeword in @xmath34 , and let @xmath277 be a codeword adjacent to @xmath213 further , let @xmath804 and @xmath805 be the set of codewords of type @xmath446 $ ] with respect to @xmath50 and @xmath806 respectively .
then @xmath807 where @xmath808 where @xmath809 theorem [ 12 ] states that the @xmath194-neighbourhood of each codeword @xmath50 is uniquely determined by the codewords of type @xmath446 $ ] .
thus , with claim d in hands , we know that any two adjacent codewords have the same @xmath194-neighborhood .
the rest of the proof of the theorem follows easily by induction because to each codeword @xmath277 there is a sequence of codewords @xmath810 such that the codeword @xmath811 is adjacent to the codeword @xmath812 for all @xmath813 w.l.o.g .
we prove claim d only for @xmath170 the origin .
consider a codeword @xmath277 that is adjacent to @xmath128 to prove claim d for @xmath277 we need to show that @xmath814 where @xmath815 is a codeword in @xmath816 .
to do so , it suffices either indeed , in the case ( a ) , we know that for each codeword @xmath50 its @xmath194-neighborhood is symmetric with respect @xmath167 hence if @xmath822 is a codeword then also @xmath823 is a codeword because @xmath824 in the case ( b ) consider a codeword @xmath825 . by theorem
[ 13 ] , if there are codewords @xmath826 so that @xmath827 and @xmath828 then @xmath829 is a codeword as well .
however , ( i ) guarantees that all assumptions of theorem [ 13 ] are satisfied , therefore ( ii ) guarantees that @xmath796 is in the @xmath194-neighborhood of the codeword @xmath277 .
first we choose the codeword adjacent to the origin to be of type @xmath446.$ ] let @xmath830 if @xmath831 then @xmath832 is a codeword , and hence by ( a ) @xmath833 is a codeword as well .
the following table provides a suitable choice for the other four codewords in @xmath834 [ c]|l|l|l|l|l|@xmath835 & @xmath836 & @xmath837 & @xmath838 & @xmath839 + @xmath840 & @xmath841 & @xmath842 & @xmath843 & @xmath844 + @xmath845 & @xmath846 & @xmath847 & @xmath843 & @xmath848 + @xmath849 & @xmath846 & @xmath847 & @xmath850 & @xmath851 + @xmath852 & @xmath853 & @xmath854 & @xmath855 & @xmath856 + theorem [ 13 ] guarantees that the @xmath194-th neighborhood of each codeword is symmetric .
therefore @xmath796 is a codeword also for all@xmath857 let @xmath858 @xmath859 then we apply the same cyclic shift to @xmath825,@xmath860 given in the table to obtain the required codewords .
the same applies to @xmath792 and its cyclic shifts .
finally , let @xmath277 be a codeword of type @xmath294 $ ] adjacent to the origin .
say @xmath861 the proper choice of @xmath818 for each @xmath862 is given in the table below : [ c]|l|l|l|l|l|@xmath863 & @xmath836 & @xmath864 & @xmath865 & @xmath839 + @xmath846 & @xmath841 & @xmath866 & @xmath843 & @xmath867 + @xmath840 & @xmath845 & @xmath868 & @xmath869 & @xmath870 + @xmath845 & @xmath845 & @xmath871 & @xmath872 & @xmath873 + @xmath849 & @xmath846 & @xmath874 & @xmath875 & @xmath876 + @xmath852 & @xmath846 & @xmath854 & @xmath877 & @xmath878 + while for @xmath879 , @xmath796 is a codeword as the @xmath194-neighbourhood is symmetric . as in the case of @xmath277 being an adjacent codeword of type @xmath446,$ ] a suitable choice of @xmath818 for other cases can be obtained by a cyclic shift .
the proof is complete . | an @xmath0-dimensional cross comprises @xmath1 unit cubes : the center cube and reflections in all its faces .
it is well known that there is a tiling of @xmath2 by crosses for all @xmath3 albdaiwi and the first author proved that if @xmath1 is not a prime then there are @xmath4 non - congruent regular (= face - to - face ) tilings of @xmath2 by crosses , while there is a unique tiling of @xmath2 by crosses for @xmath5 .
they conjectured that this is always the case if @xmath1 is a prime . to support the conjecture we prove in this paper that also for @xmath6 there is a unique regular , and no non - regular , tiling by crosses
so there is a unique tiling of @xmath7 by crosses , there are @xmath4 tilings of @xmath8 but for @xmath6 there is again only one tiling by crosses .
we guess that this result goes against our intuition that suggests `` the higher the dimension of the space , the more freedom we get '' .
tilings of @xmath2 by unit cubes go back to 1907 when minkowski conjectured @xcite that each lattice tiling of @xmath2 by unit cubes contains twins , a pair of cubes sharing a complete @xmath9 dimensional face .
this conjecture was proved by hajs @xcite in 1942 . in 1930 , when minkowski s conjecture was still open
, keller @xcite suggested that the lattice condition in the conjecture is redundant , that the nature of the problem is purely geometric , and not algebraic as assumed by minkowski .
thus he conjectured that each tiling of @xmath2 by unit cubes contains twins .
it is trivial to see that each tiling of @xmath10 by unit cubes contains twins , and it is also easy to verify it for @xmath11 however , a proof that each tiling of @xmath12 @xmath13 contains twins takes in aggregate 80 pages , see @xcite .
there was no progress on keller s conjecture for more than 50 years . only in 1992
lagarias and shor @xcite constructed a tiling of @xmath14 by unit cubes with no twins .
first they found such a tiling in @xmath15 which we consider a very surprising and remarkable result .
however , once one has such a tiling in hand , it is relatively easy to find it for @xmath16 as well .
the second part supports our belief that `` the higher the dimension of the space , the more freedom we get '' .
mackey @xcite proved that the keller s conjecture is false for @xmath17 as well .
as to the remaining value of @xmath18 there are only some partial results , see @xcite .
since late fifties tilings of @xmath2 by different clusters of unit cubes have been considered , see e.g. @xcite and @xcite , many of them related to perfect error - correction codes in lee metric .
the golomb - welch conjecture @xcite has been a main motivating power of the research in this area for the last forty years .
a perfect @xmath19-error correcting lee code over @xmath20 ( over @xmath21 of block size @xmath0 , denoted @xmath22 ( denoted @xmath23 ) , is a set @xmath24 ( @xmath25 ) of codewords so that each word @xmath26 ( @xmath27 ) is at lee distance at most @xmath19 from exactly one codeword in @xmath28 golomb - welch . for @xmath29 and @xmath30
there is no @xmath22 code .
clearly , the above conjecture , if true , implies that there is no @xmath23 code for @xmath29 , @xmath30 and @xmath31 for the state of the art on the conjecture we refer the reader to @xcite .
in this paper we focus on tilings by @xmath0-crosses .
an @xmath0-dimensional cross comprises @xmath1 unit cubes : the `` central '' one and reflections in all its faces .
a tiling @xmath32 of @xmath2 by crosses is called a @xmath20-tiling ( a lattice tiling ) if centers of all crosses in @xmath32 have integer coordinates ( form a lattice in @xmath33 .
a regular ( face - to - face ) tiling is a tiling that is congruent to a @xmath20-tiling ; otherwise the tiling is called non - regular .
we recall that two tilings @xmath34 and @xmath35 of @xmath2 are congruent if there exists a linear , distance preserving bijection of @xmath2 which maps @xmath34 on @xmath35 .
it seems that krteszi @xcite was the first to ask whether there exists a tiling of @xmath7 by crosses .
such a tiling was constructed by freller in 1970 ; korchmros about the same time treated the case @xmath36 .
golomb and welch showed the existence of these tilings in terms of error - correcting codes , see section 3.5 in @xcite . immediately after the existence question has been answered
, the enumeration of tilings has been studied . in @xcite molnr
proved : [ m]molnar .
the number of pair - wise non - congruent lattice @xmath20-tilings of @xmath2 by crosses equals the number of non - isomorphic commutative groups of order @xmath37 szabo @xcite constructed a non - regular lattice tiling of @xmath2 by crosses in the case when @xmath1 is not a prime . using refinements of this construction
it was proved in @xcite that in this case there are @xmath38non - congruent @xmath20-tilings of @xmath2 by crosses . in a strict contrast to this result
it was proved there that , for @xmath39 and @xmath40 , there is a unique , up to a congruence , tiling of @xmath2 by crosses .
it is conjectured in @xcite , see also @xcite : [ 2]if @xmath1 is a prime then there exists , up to a congruence , only one @xmath20-tiling of @xmath2 by crosses .
it seems to us that the above conjecture , if true , would totally go against our intuition that suggests : the higher the dimension of the space , the more freedom we get ; see also an above comment related to the lagarias - shor result on keller s conjecture . to provide supporting evidence for conjecture [ 2 ]
we prove in this paper : [ 1]there exists , up to a congruence , a unique @xmath20-tiling of @xmath6 by crosses .
although we proved conjecture [ 2 ] only for @xmath41 an essential part of the proof of theorem [ 1 ] holds for all @xmath42 we believe that this part will be helpful when proving this conjecture for some other values of @xmath3 clearly , if @xmath32 is a @xmath20-tiling of @xmath2 by crosses , then centers of crosses in @xmath32 form a @xmath43 code .
it is easy to check that the unique tiling of @xmath6 by crosses is @xmath44-periodic .
thus , as an immediate consequence we get : there is a @xmath45 code if and only if @xmath46 as to the non - regular tilings of @xmath2 by crosses , it was mentioned above that such a tiling exists if @xmath1 is not a prime .
a result of redei @xcite implies that , if @xmath1 is a prime , then there is no lattice non - regular tiling of @xmath2 by crosses .
it is easy to check that a non - regular tiling of @xmath10 by crosses does not exist .
the same result for @xmath40 has been proved in @xcite . as the other main result of this paper
we will show that : [ 3]let @xmath1 be a prime .
if there is a unique @xmath20-tiling of @xmath2 by crosses , then there is no non - regular tiling of @xmath2 by crosses .
combining theorem [ 1 ] with [ 3 ] we get : there is a unique , up to a congruence , tiling of @xmath6 by crosses , and this tiling is a @xmath20-tiling . thus , there is a unique tiling of @xmath7 by crosses , there are @xmath38 pair - wise non - congruent @xmath20-tilings of @xmath47by crosses , but for @xmath6 there is again a unique tiling by crosses . also , by means of theorem [ 3 ] , it is straightforward that conjecture [ 2 ] is equivalent to if @xmath1 is a prime then there exists , up to a congruence , a unique tiling of @xmath2 by crosses , and this tiling is a lattice @xmath20-tiling . in the next section
we introduce needed notation , definitions and state some auxiliary results .
theorem [ 1 ] will be proved in section 3 . |
surveys of molecular clouds indicate that 60% of young stars form in clusters @xcite .
it is therefore important to study star formation in clusters to understand the influence of the cluster environment in the process of star and planet formation . of fundamental importance
is the understanding of the spatial structure of clusters , and of the evolution of that structure .
observations with the _ spitzer _ and _ chandra _ space telescopes are of great importance in the study of cluster structure as they provide the means to identify young stellar objects from the protostellar to pre - main sequence phases , and to categorize these objects into the canonical evolutionary classes ( class 0/i , ii and iii @xcite ) .
the number and distribution of sources in each of these classes provides unique information on the distribution of star formation sites , the motion of the stars after they form , and the dynamical state of the cluster at large . from these studies
, we can better understand the physical processes that govern the fragmentation of molecular clouds , the ensuing trajectories that the resulting young stars make through the cluster , and the eventual fate of the embedded cluster .
clusters also serve as laboratories for the evolution of young stellar objects , and can be used to study the evolution of disks around young stars and the evolution of hot x - ray emitting plasma commonly found around such stars ( see for example : @xcite ) in this paper , we describe a detailed study of the population of young stellar objects in the serpens cluster identified by observations with _ spitzer _ and _ chandra_. the serpens region is an example of a very young , deeply embedded cluster , containing a number of protostars @xcite . the embedded cluster is heavily extinguished , with a peak extinction exceeding 40 magnitudes in the visual . at a distance of @xmath0260
pc , the serpens cloud core is one of the nearest regions of clustered star - formation to the sun ( see @xcite for a discussion of the distance to serpens ) .
this makes it an excellent candidate for study with _
spitzer _ as it is close enough to both resolve the individual members and to detect the lowest mass members to below the hydrogen - burning limit .
furthermore , the serpens cluster is rich in protostars .
sub - millimetre and millimetre observations of the region , at 450 @xmath1 , 850 @xmath1 and 3 mm , identify at least 14 objects in the serpens cloud core @xcite . in this paper , we extend the sample of known young stellar objects in serpens using the high sensitivity of _ spitzer _ and _ chandra_. we then discuss three topics : the mid - ir spectral energy distributions of protostellar objects detected in the sub - millimeter , the spatial distribution of the serpens cluster members , and finally , the x - ray properties of the young stellar objects as a function of their evolutionary class .
we have obtained _ spitzer _ images of the serpens cloud core in six wavelength bands : the 3.6 , 4.5 , 5.8 and 8.0 @xmath3 m bands of the infrared array camera ( irac ; @xcite ) and the 24 @xmath3 m and 70 @xmath3 m bands of the multi - band imaging photometer for _ spitzer _ ( mips ; @xcite ) .
the photometry extracted from these data was supplemented by @xmath9 , @xmath10 and @xmath11-band photometry from the 2mass point source catalog @xcite , resulting in data in nine photometric bands spanning 1 - 70 @xmath3 m .
the uncertainties for the point sources as a function of magnitude are shown in fig .
[ fig3 ] . in the following analysis ,
only data with uncertainties @xmath12 magnitudes were used .
below , we describe the observations , image reduction and photometry for the _ spitzer _ data .
the _ spitzer _ space telescope has observed the serpens cluster with the irac instrument at two different epochs .
the first epoch was on 1st april 2004 ; these observations were part of the guaranteed time observation program pid 6 , _ the structure and incidence of embedded clusters_. the second epoch was on 5th april 2004 , during which _ spitzer _ executed observations from pid 174 from the _ cores to disks _ ( c2d ) legacy program .
the mosaicked regions cover a @xmath13 field in all four of the irac wavelength bands ( fig.[fig1 ] ) .
the field is centered on the main serpens cluster in the northern cloud core .
the observations were taken in the high dynamic range mode with a 12 second frame time . in each epoch , two dithered 10.4 second frames and two dithered 0.4 second frames were obtained at each map position .
the analysis used the basic calibrated data ( bcd ) fits images from the s11.4 pipeline of the _ spitzer _ science center .
these data were combined into mosaics using the mopex program .
the 3.6 @xmath3 m and 4.5 @xmath3 m images were pre - processed by using the idl - based pulldown corrector by lexi moustakas , and then additional interactive processing steps were applied to locate and remove the remaining column pulldown and mux bleed artifacts in the 3.6 @xmath3 m and 4.5 @xmath3 m images , and banding effects in the 5.8 @xmath3 m and 8.0 @xmath3 m images @xcite .
the overlap correction module was used to minimize instrumental offsets between adjacent frames , and the mosaics were constructed on a pixel scale of 0.8627 @xmath14/pixel , or @xmath15 the irac pixel size .
the typical total integration time is 41.6 seconds per pixel , although in some regions near the edges of the mosaic , the integration time is 20.8 seconds .
the calibration uncertainty across the irac bands is estimated at @xmath16 mag @xcite .
in addition , there are @xmath05% position dependent variations in the calibration of point sources in the flat - fielded bcd data ; these have not been corrected for in our data .
furthermore , these systematic uncertainties have not been added into the random photometric uncertainties reported in this paper .
we used a custom idl program for point source finding ; this program was a heavily modified version of the daofind program in the idlphot package @xcite .
this program first creates a smoothed version of the mosaic by convolving with a gaussian with @xmath17 pixels , approximately double the fwhm of the point sources , and then subtracts the smoothed version from the mosaic to filter out the extended nebulosity .
the noise in a @xmath18 pixel region around the source was calculated to assess the combined instrumental noise , shot noise and noise from the spatially varying extended nebulosity .
point sources with peak values more than 5 @xmath19 above the background were considered candidate detections .
after the point sources were identified , aperture photometry then was obtained using the aper.pro program in the idlphot package .
an aperture radius of 2.8 pixels ( @xmath20 ) was used , and a sky annulus from 4.2 to 8.4 pixels ( @xmath21 to @xmath22 ) was used to measure the contribution from extended emission in the aperture .
the zero points for these apertures were , 17.8204 , 17.3025 , 16.7408 , and 15.9440 mag for the 3.6 , 4.5 , 5.8 , and 8.0 @xmath3 m bands , respectively , in the native bcd image units of mjy pixels sr@xmath23 .
there were 25853 , 22703 , 6595 , and 3946 detections in the 3.6 , 4.5 , 5.8 , and 8.0 @xmath1 respectively , with uncertainties less than 0.2 mag , with 2758 of these detected in all four irac bands .
the sensitivity of the 8.0 @xmath1 band is the limiting factor in four band detections : it suffers from lower photospheric fluxes , higher background emission , and the presence of bright , structured nebulosity in some parts of the image .
the detection threshold in the 3.6 @xmath1 band was 17 magnitudes - far below the hydrogen burning limit for a 1 myr old star at the distance of serpens ( @xmath012 mag at 3.6 @xmath3 m @xcite ) .
_ spitzer _ observed the serpens cluster twice with the multiband imaging photometer for _ spitzer _ ( mips ; rieke et al .
2004 ) . in both epochs ,
the medium scan rate was used . for the first epoch ,
part of the guaranteed time observation program pid 6 , taken on the 6th april 2004 , six 0.5 degree scan legs with full - array cross - scan offsets were used . for the second epoch observations ( pid 174 ) , 12th april 2005 , 12 scan legs of length 0.5
degrees and half - array cross - scan offsets were used . in both epochs ,
the total map size was approximately 0.5 x 1.5 degrees including the overscan region .
all three mips bands are taken simultaneously in this mode ; because the data at 160 micron are saturated , we do not consider it in our subsequent analysis .
the 2nd epoch map had full 70 @xmath3 m coverage , while the first epoch mapped only six @xmath24 wide bands , covering half of the map .
both epochs were combined to form the final maps .
the typical effective exposure time per pixel is about 80 seconds at 24 @xmath3 m and 40 seconds at 70 @xmath3 m .
the mips images were processed using the mips instrument team data analysis tool , which calibrates the data and applies a distortion correction to each individual exposure before combining into a final mosaic @xcite .
the data were further processed using various median filters to remove saturation effects and column - dependent background structure .
the resulting mosaics have a pixel size of @xmath25 at 24 @xmath3 m and @xmath26 at 70 @xmath3 m .
stars in the 24 @xmath3 m mosaic were then identified using photvis @xcite , which searched for all point sources with peaks 5 times the rms noise .
aperture photometry in a 5 pixel aperture was performed on these sources with photvis and the photometry was corrected to a 12 pixel aperture radius , with a sky annulus from 12 - 15 pixels .
adopting a conversion of 6.711 jy per @xmath27 , multiplying by @xmath28 to correct for the smaller mosaic pixels and by 1.146 to correct from a 12 mosaic pixel aperture to an infinite aperture , and using a zero magnitude flux of 7.3 jy , the aperture photometry was converted to magnitudes . due to the larger @xmath29 fwhm of the mips 24 @xmath3 m point spread function ( psf ) , we performed psf fitting photometry on all the detected sources using the idl version of daophot in the idlphot package .
the psf was generated from six bright ( 3 - 6 magnitude ) stars in the image ; these stars were chosen to be bright and uncontaminated by nebulosity .
psf fitting photometry was then performed on the point source detections using the nstar.pro routine in idlphot .
the nstar routine requires the input of aperture photometry .
sky values were taken from an annulus from 12 to 15 pixels .
these photometry are then adjusted by fitting the psf to the image data and scaling accordingly .
a total of 269 sources were extracted with uncertainties less than 0.2 mag .
three of the sources , numbers 11 , 35 , and 36 , were saturated at 24 @xmath3 m .
the fits of these data were adjusted by scaling the point spread function manually , subtracing the psf from the image , and visually inspecting the residual in the wings of the psf .
the tabulated magnitudes produced the lowest apparent residual .
since raising / lowering the magnitudes by 0.1 mag would produce a distinct over / undersubtraction in the residual image , we adopted an uncertainty of 0.1 magnitude for these values .
the 70 @xmath3 m photometry was extracted from the mosaic using a similar procedure to the 24 @xmath3 m photometry .
aperture photometry was performed on the sources in a 3.66 pixel aperature , with a sky annulus from 3.66 - 7.92 pixels .
the conversion used was 0.675 jy per @xmath27 , with a factor of 1.927 correcting from a 3.66 pixel to an infinite aperture , and a zero magnitude flux of 0.775 jy .
psf photometry was not performed on the point sources as there were not enough detections uncontaminated by nebulosity to generate the psf .
the fwhm of the psf at 70 @xmath3 m is 18@xmath14 .
the typical uncertainty on the 70 @xmath3 m flux is 15% .
the x - ray data were taken from the @xmath30 anchors ( an archive of chandra observations of regions of star formation ) archive , obsid 4479 .
the raw x - ray data are dominated by events of non - astrophysical origin . to remove these ,
the raw data were processed using the @xmath31 x - ray center s standard processing version 6.13.2 ( july 2005 ) . using acis_process_events on the level 1 events
file , a gain correction ( conversion from pulse heights to x - ray energy ) is applied from caldb 2.21 .
the cti correction for pulse heights distorted by charge transfer inefficiency and vfaint background cleaning to remove soft cosmic rays were also applied .
next an energy filter was applied to remove photons above 8 kev and below 300 ev which are typically not of stellar origin .
finally events with bad grades and bad status ( grades 1 , 5 and 7 , status @xmath320 indicative of x - ray signals ) and bad time intervals were filtered using the ciao tool _
dmcopy_. time - dependent gain corrections were applied and acis_process_events rerun .
while there were over 3@xmath33 level 1 events , our `` cleaned '' data file used for analysis contained 171,973 events . due to vignetting and small gaps between the chips of
the i array the effective collecting area on each part of the sky differs .
the effective area also depends on the energy of the photon .
the exposure map corrects for the changes in the effective area .
an exposure map was created using _
merge_all _ , accurately representing the average effective area for a 1.7 kev photon .
we chose this single energy for the exposure map since it is intermediate between the maximum of the effective area of the hrma / acis system and an estimated mean source energy of @xmath0 2.0 kev .
the exposure map is later applied automatically by ciao tools extracting count rates and spectra . to perform source detection ,
the data were split into two event lists , one to concentrate on cooler sources with limited noise , consisting of photons with energies between 0.5 and 2.0 kev .
the other list contained photons of higher energy between 2.0 and 7.5 kev .
wavdetect was used for source detection on our cleaned and separated events list to identify sources across the entire i array .
threshold significance was set to detect sources down to 3.5 sigma and the data were searched on scales of 0.5 to 16@xmath14 . with these settings , a false detection rate of @xmath34
is expected .
the source detection lists were combined to make a single source list of 88 sources . in comparison ,
@xcite construct their source list from the same data using different filtering criteria on the bottom 10% of the data .
there is @xmath095% agreement between the two source catalogues . from the location of the 88 sources ,
regions were calculated which would contain 95% of the x - ray energy of each sources .
the regions are based on the @xmath31 psf and chip position for each source . the routine _
mk_psf _ was used to obtain images of the psf at various off - axis angles @xmath35 ( arcmin ) , and rotation angles @xmath36 ( degrees ) , around the acis array . at each source location
an ellipse was generated to enclose 95% of the total x - ray energy following wolk et al .
( 2006 ) .
analysis of the x - ray spectrum of each source was performed to determine the bulk temperature of the corona and the intervening column of hydrogen . for each source with over 25 counts , source and background pulse height distributions in the total band ( 0.3 - 8.0 kev )
were constructed .
the final fits were done with ciao version 3.1.0.1 using the ciao script _
psextract _ to extract source spectra and to create an ancillary response function ( arf ) and redistribution matrix function ( rmf ) files which correct for the detector response at each location .
model fitting of spectra was performed using _ sherpa _ @xcite .
the data for each source were grouped into energy bins which required a minimum of 8 counts per bin and background subtracted .
the optimization method was set to levenberg - marquardt and @xmath37gehrels statistics were employed .
an absorbed one temperature `` raymond smith '' plasma @xcite , was fitted using a two step fitting procedure .
initial conditions were set so that @xmath38 and @xmath39 kev .
then an initial fit was made with an absorbed thermal blackbody model .
these fit results were then used as initial conditions for the absorbed raymond
smith plasma model .
to generate a catalog of the photometry from all detected sources in the serpens field , the photometry from the combined irac , mips and _ chandra _ data were merged .
in addition , the data were also merged with the 2mass point source catalog to provide @xmath9 , @xmath10 and @xmath11-band photometry for each detected _ spitzer _ source @xcite .
sources observed in different wavelength bands which were located within @xmath40 of each other were considered to be the same source ; if while comparing two bands , multiple sources in one of the bands satisfied this criteria , the closest source was chosen
. the field of view ( fov ) of each instrument did not cover the same region of the cluster , being most limited by the irac fov in the infrared , and overall by the _
the mips data covered the entire irac field , while the 2mass survey is not spatially limited .
furthermore , the irac detectors are split into two groups , channels 1 & 3 ( 3.6 & 5.8 @xmath1 ) and channels 2 & 4 ( 4.5 & 8.0 @xmath1 ) , whose fovs are offset from one another by @xmath41 .
the field studied in the remainder of this paper is the overlap region of all five _ spitzer _ bands , the ir - field ( fig.[fig2 ] ) .
the size of the overlap regions is 26@xmath42 @xmath43 28@xmath42 . in comparison ,
the _ chandra _ image of serpens covered a 17@xmath42 @xmath43 17@xmath42 field of view ( the irx - field ) , centered on the northwestern edge of the cluster , @xmath02@xmath42 from the center of the irac fov ( fig . [ fig2 ] ) .
x - ray data is not available for all the sources in our catalog .
the serpens field contains 19,181 sources with a detection in at least one band with photometric uncertainty @xmath44 in the ir - field ( 122 having detections in all five bands ) ; however , only a small fraction of these are bona fide ysos .
three methods were applied to these data to identify possible young stellar objects : selection of stars with ir excesses on ir color - color diagrams , identification of x - ray luminous ysos by comparison of x - ray sources with ir detections , and finally , a search for extremely red mid - ir sources among the detections .
young stellar objects can be identified by their excess emission at ir wavelengths .
this emission arises from reprocessed stellar radiation in the dusty material of their natal envelopes or circumstellar disks .
the infrared identification of ysos is carried out by identifying sources that possess colors indicative of ir excess and distinguishing them from reddened and/or cool stars @xcite .
the main limitation of this method is the contamination from extragalactic sources such as pah - rich star - forming galaxies and agn ; both of these have colors similar to young stellar objects .
four color - color diagrams were used to determine the color excess of the sources : an irac [ 3.6 ] - [ 4.5 ] vs. [ 5.8 ] - [ 8.0 ] diagram , an irac - mips [ 3.6 ] - [ 4.5 ] vs. [ 8.0 ] - [ 24 ] diagram , and two irac-2mass diagrams j - h vs. h - [ 4.5 ] and h - k vs. k - [ 4.5 ] , fig.[fig7 ] . in the following analysis we required all photometry to have uncertainties @xmath44 mags in all bands used for a _
particular _ color - color diagram .
the numbers of sources for each diagram which satisfy this criteria are 2417 , 122 , 3748 , and 4251 , respectively . to identify sources with true ir excesses , it is necessary to distinguish between reddened or cool stars and those with excess emission arising from heated dust grains .
a reddening law in the irac bands was determined by flaherty et al.(2006 ) , which shows the [ 5.8 ] - [ 8.0 ] color to be particularly insensitive to reddening and stellar temperature , and is thus a reliable measure of excess emission due to dust .
sources with a color @xmath45 beyond @xmath46 - [ 8.0 ] > 0.2 $ ] mag are likely to possess excesses and not to be reddened or cool stars ( fig.[fig4 ] ) .
extragalactic sources such as pah rich star - forming galaxies and agn will also satisfy this criteria @xcite .
sources with a color more than @xmath47 below @xmath48 - [ 4.5 ] < 0.1 $ ] were considered galaxies and were removed ; those with a 24 @xmath1 detection are reconsidered separately .
gutermuth et al.(submitted ) have recently developed a method for substantially reducing extragalactic contamination built on the bootes shallow survey data @xcite and the _ spitzer cores to disks _ legacy program methods @xcite .
the galaxies are eliminated either by their colors , which unlike ysos are often dominated by pah features in the 5.8 and 8.0 @xmath1 bands , or by their faintness .
agn are typically much fainter than the ysos found in nearby star - forming regions and are removed by this criterion .
it should be noted that very faint or embedded flat spectrum sources may be erroneously filtered by this method also .
fig.[fig5 ] shows the color - color and color - magnitude diagrams used to identify star - forming galaxies with strong pah emission and agn .
nebulosity may also result in the misidentification of sources .
there are 55 sources with excesses in the [ 5.8]-[8.0 ] color , but little evidence of an excess in the [ 3.6]-[4.5 ] color .
examination of the spectral energy distributions for these sources ( sec .
6 ) showed that these sources exhibit excess emission only in the 8.0 @xmath1 band .
subsequent analysis showed that seventeen of these also showed excesses in the 24 @xmath1 band , but 38 showed no 24 @xmath1 detection .
the seventeen 24 @xmath1 excess sources followed the distribution of the other bona fide ysos in the field , while the 38 sources with weak 8.0 @xmath1 and no 24 @xmath1 excess were found in regions of bright 8.0 @xmath1 nebulosity to the east of the cluster ( fig . 2 ) .
these sources appear to have their 8.0 @xmath1 photometry contaminated by the nebulosity and were removed from the sample . in general
, all sources with [ 5.8]-[8.0 ] excesses which do not show an excess at 24 @xmath1 excess or a @xmath48-[4.5 ] > 0.1 $ ] are removed . in total , 146 potential ysos were identified in the irac color - color diagram . of these , 62 are likely to be contaminants and have been removed from the source list . the shorter wavelength 3.6 and 4.5 @xmath1 irac bands are much more sensitive to stellar photospheres than the longer wavelength 5.8 and 8.0 @xmath1 bands ; where the shorter wavelength bands have 17,385 and 15,539 detections with @xmath49 , respectively , the longer wavelength bands having 4,670 and 2721 detections .
hence many sources can not be characterized with the irac color - color diagram .
for this reason , it is important to develop methods to identify sources with infrared excesses that rely only on the shorter wavelength bands . since we can not distinguish between reddening and infrared excess from circumstellar dust with only two bands , we combine the near - ir data from the 2mass point source catalog with the 3.6 and 4.5 @xmath1 band photometry @xcite .
this provides the ability to detect objects which are too faint for detection in the 5.8 and 8.0 @xmath1 bands .
in particular , we concentrate on the @xmath50 vs. @xmath51 $ ] diagram and the @xmath52 vs @xmath51 $ ] diagram . these diagrams take advantage of the excellent sensitivity of _ spitzer _ in the 4.5 @xmath3 m band and the stronger infrared excess emission at 4.5 @xmath3 m compared to that at shorter wavelength @xcite . for highly reddened sources which are not detected in the @xmath9-band
, the @xmath52 vs. @xmath53 $ ] diagram can be used .
it should be noted that the irac 3.6 and 4.5 @xmath1 data can detect sources too faint or reddened to have been detected by 2mass .
deeper near - ir imaging is needed to detect these sources in the @xmath9 , @xmath10 and @xmath11-bands .
we calibrated the combined irac-2mass color - color diagrams using a sample of stars which show no evidence for infrared excesses out to 8.0 @xmath1 , and which had uncertainties @xmath54 in all irac-2mass bands , c.f . fig .
they were identified primarily by their [ 5.8]-[8.0 ] color , which is not significantly affected by extinction ( flaherty et al .
the criteria used were the following : @xmath55 the [ 3.6]-[4.5 ] color was also limited to values less than one to eliminate contamination from protostars which can have [ 5.8]-[8.0 ] close to that of pure photospheres @xcite .
this sample of reddened photospheres was then plotted on the @xmath50 v @xmath56 $ ] and @xmath52 vs. @xmath53 $ ] color - color diagrams to define where the reddened photospheres fall in the diagrams ( fig.[fig4 ] ) . reddening vectors derived from flaherty et al .
( 2006 ) were placed on the diagrams so that all of the pure photospheres were blueward ( to the left ) of the vectors .
sources which were more than @xmath57 redward of the reddening vectors were selected as having an excess . using this analysis on the @xmath50 vs. @xmath51 $ ] diagram ,
the following criteria for infrared excess stars was established : @xmath58 where 0.912 is the slope of the reddening curve .
there were 72 sources detected using these criteria , 27 of which are not found with the irac color - color diagram .
of the 27 , 12 were found to be contaminants . in the @xmath52 vs. @xmath53 $ ] diagram
, the infrared excess stars were identified using the following criteria : @xmath59 where 0.873 is the slope of the reddening curve .
there were 50 sources detected via this plot , 8 not detected on the irac or @xmath50 vs. @xmath51 $ ] color - color diagrams .
three of the eight were found to be contaminants .
there were 24 sources identified as having an excess on the irac-2mass color - color diagrams that did not have photometry in all the irac bands and did not show a [ 3.6]-[4.5 ] color excess . in these cases ,
the observed infrared excess resulted from a discontinuity between the 2mass and irac photometry ; such a discontinuity may be the result of time variability or contaminated photometry .
these sources were not selected as infrared - excess sources . to minimize contamination from agn , sources which lacked 5.8 and 8.0 @xmath1 detections and were identified solely in the irac-2mass color - color diagrams were rejected if their 3.6 @xmath1 magnitude was fainter than 15 mag . at the distance of serpens ,
this magnitude is well below the cutoff of the hydrogen - burning limit for 1 - 3 myr stars ( 12 - 13 mag ; @xcite ) . from our analysis of the irac color - color diagram ( sec 5.1.1 ) ,
sources identified as agn dominate this region ( fig .
[ fig6 ] ) . for young stars with circumstellar disks , the [ 3.6 ] - [ 4.5 ] vs. [ 8.0 ] - [ 24 . ]
color - color diagram is very sensitive to the ir excess emission from dust at @xmath603 - 5 au @xcite .
it is particularly effective at identifying stars with little to no mid - ir excess shortward of 8.0 @xmath1 due to inner holes in their disks .
these transition disks appear to be in a state of evolution where the inner disk has been cleared of small dust grains by planets or grain growth @xcite .
we used the following selection criteria : @xmath61 a lower limit of 0.4 mag was used to conservatively eliminate pure photospheres even in the presence of photometric scatter due to nebulosity and source confusion .
this criteria also excludes likely extragalactic sources which have colors [ 8.0 ] - [ 24 . ]
a total of 75 young stellar objects were identified on these diagrams , 15 of which were not previously identified in the irac and irac-2mass color - color diagrams .
three of the fifteen were identified as extra - galactic contaminants .
estimates for the 90% completeness limit of our _ spitzer _ photometry were calculated via the method of inserting artifical stars into the mosaics and then employing our detection algorithms to identify them .
the 90% completeness estimates were 15.0 , 15.0 , 14.5 , 12.5 , 7.5 at 3.6 , 4.5 , 5.8 , 8.0 , and 24 @xmath1 , respectively , for the irac and mips photometry .
[ figcomp ] plots the histograms and completeness of sources at 3.6 @xmath1 .
the dashed black line plots the number of 3.6 @xmath1 detections by magnitude .
we have also plotted the number of sources which have sufficient signal to noise in the relevant bands that they can be placed on the irac , irac - mips and/or irac-2mass color - color diagrams ( upper grey line ) .
these are the sources which can be searched for ir - excesses .
the final yso catalogue sources are also plotted ( lower gray line ) , as are the ysos detected in x - rays ( lower light gray line ) . the ratio of the number of sources with @xmath63 mag . that can be placed on the color - color diagrams to the total number of sources with @xmath63 mag .
( and correcting this number for completeness by dividing by the fraction of artifical stars recovered in each magnitude bin ) is @xmath64 or 81% .
the same ratio for detections above 14.5 and 14 magnitudes , the percentages are 93% and 97% respectively .
these fractions represent lower limits on the completeness of the ysos , as the number of contaminating field stars is rising with increasing magnitude , while the number of ysos does not show an equivalent rise .
magnitudes of 14.5 and 15 for irac 3.6 @xmath1 , correspond to masses of @xmath00.02 @xmath65 and @xmath00.03 @xmath65 at 1 myrs , and @xmath00.03 @xmath65 and @xmath00.04 @xmath65 at 3 myrs @xcite .
this completeness limit is for pre - main sequence stars with ir - excess emission from circumstellar disks .
since the fraction of sources with ir - excesses varies with mass and age @xcite , the completeness with respect to all pre - main sequence can not be estimated with the available data .
deeply embedded protostars may only be observable at 8.0 or 24 @xmath1 , and thus not be detected by the previous methods @xcite . in order
that these objects are not overlooked , the sed of each of the 24 @xmath1 detections that was not already selected was visually examined .
three sources were identified at 24 @xmath1 only , and one source was identified at 8.0 and 24 @xmath1 .
one of sources detected at 24 @xmath1 lies at the position of a known sub - millimetre source , smm1 @xcite ; it is one of the brightest 24 @xmath1 sources on our image ( @xmath66 ) and is also seen at 70 @xmath1 .
another source coincides with smm8 @xcite ; this is detected in both 8.0 @xmath1 and 24 @xmath1 .
it is a fainter source , 11.5 mag at 8.0 @xmath1 and 6.3 mag at 24 @xmath1 .
one of th esources is near the ( sub)mm source smm3 , and is probably associated with that source .
however , given the lack of clear detections in the irac - bands and the lack of a clearly identifiable 70 @xmath1 counterpart in this crowded region , we consider this 24 @xmath1 source a tentative yso .
the remaining source , detected only at 24 @xmath1 , was detected in only one of the two epochs of mips observations , and is probably an asteroid .
young stellar objects may also possess elevated levels of x - ray emission ( @xmath67 ) which can be used to distinguish them from field stars @xcite .
we utilize this property to identify ysos that do not have emission from a dusty disk ( evolutionary class iii ) and would otherwise be indistinguishable from field stars
. protostars ( class 0/i ) and pre - main sequence stars with disks ( class ii ) with elevated x - ray emission may also be identified .
image of serpens covered a 17@xmath43 17 field of view centred on the cluster , and x - ray data is not available for all the sources in our catalog , c.f .
table [ tbl2 ] .
the _ chandra _ observations detected 88 sources in the region , a mix of genuine ysos and agn .
the coordinates of these sources were matched against the 19,000 ir sources , with 67 matches , or 76% .
the lower limit of the x - ray luminosity detectable in serpens with chandra can be estimated by comparison with the coup data , see @xcite , as @xmath68 @xmath69 @xmath70 , assuming a distance of 260 pc to serpens and an exposure time of 88 ks .
we match the x - ray sources with ir analogues to reduce the contamination from agn ; an x - ray source with no ir counterpart is assumed to be agn contamination . of the 67 matches ,
40 were ir excess sources previously identified , 3 sources had only been detected in some of the infrared bands and consequently could not be placed on the color - color diagrams used to identify infrared excess sources .
the remaining 24 sources did not exhibit an ir - excess .
of the 67 ir matches , 27 of the x - ray luminous stars are not in the list of infrared excess sources . of these 27 sources , five of the sources with _
chandra _ and _ spitzer _ detections were identified as probable agn using the criteria in sec 5.1.1 .
an additional two sources which did not show detections in all of the irac bands had 3.6 @xmath1 magnitudes fainter than 15 ; these seven sources were cosidered to be contaminating agn . in total , 60 ysos were detected in x - rays , 20 of which are new to the yso catalog .
the completeness of the x - ray data can be assessed from fig .
[ figcomp ] ; it should be noted that only 78% of the ysos from class 0/i to transistion disk lie in the irx - field , thus the completeness of the x - ray data is limited partially by the smaller field of view .
for @xmath71 we detect 57% of the ysos in x - rays , this percentage drops quickly for fainter magnitudes .
this corresponds to a mass of 0.18 @xmath65 for an age @xmath72 myr and an @xmath73 @xcite .
a total of 229 candidate objects were identified in the overlap ir - field through the methods outlined above .
a list of 200 candidate members was compiled from the color - color diagrams , 66 were detected in x - rays in the irx - field , and three as extremely red mid - ir objects .
after contaminant removal there remained a total of 138 , which we consider to be bona fide young stellar objects .
the membership of serpens is estimated to be more tahn 97% complete to 14 mag at 3.6 @xmath1 , or @xmath00.04
@xmath65 at 1 myrs @xcite , for sources with mid - ir excess emission .
as will be shown , @xmath050% of objects in each evolutionary class are detected in x - rays , thus we are likely missing @xmath021 class iii members in the irx - field . by scaling the @xcite results to our 0.2 @xmath74 field , we estimate that there is of order 1 agb contaminant in our catalogue .
we have removed 15 agn and 13 galaxies with strong pah - emission features , and estimate that @xmath01 such contaminating object remains in the final source list @xcite .
we estimate that 1 or less of the class iii sources may be a dme contaminant .
a list of the coordinates 137 of the sources identified as young stellar objects in the serpens cloud core is given in table [ tableids ] with associated identifiers .
the photometry of the sources is given in table [ tablephot ] .
these two tables do not list the photometry for two sources detected only by mips , smm1 and smm3 , whose fluxes are listed in table [ tablesmm ] . since smm3 is only clearly detected by spitzer at 24 @xmath1 , we have not included smm3 in the 138 sources .
the evolutionary state of a young stellar object can be inferred from the _ spitzer _ mid - infrared photometry @xcite .
the classification of the serpens sources was carried out using four different diagnostics : mid - ir colors @xcite , the slope of the spectral energy distribution ( sed ) @xcite , the dereddened sed slope , and the shape of the seds ( as ascertained by visual inspection ) .
each source was classified as either class 0/i , flat spectrum , class ii , transition disk , or class iii ( * ? ? ?
* see following subsections for the definition of these classes ) .
classification of the sources was carried out by first noting their locations on the irac , irac-2mass and irac - mips color - color diagrams and assigning each source a preliminary class ( @xcite , see fig .
[ fig8 ] ) .
typically , the irac color - color diagram was used for the initial assignment of the evolutionary class .
an exception was made in the case of the transition disk objects which do not always possess an 8.0@xmath1 excess and are most clearly identified by their [ 8.0 ] - [ 24 . ] color from the irac - mips diagram .
this was followed by the construction of seds for all sources .
examples of the seds for the different evolutionary classes are given in fig.[fig9 ] .
for each sed , a slope @xmath75 was calculated . the conversion from magnitudes to fluxes in w
@xmath76 s@xmath23 used the following zero fluxes for the @xmath9 , @xmath10 , @xmath77 , [ 3.6 ] , [ 4.5 ] , [ 5.8 ] , [ 8 ] and [ 24 ] bands respectively : @xmath78 .
the slope was calculated by a least - squares fit over the available irac bands , and where possible irac and mips .
the near - ir magnitudes were not included as they are most susceptible to extinction . the serpens cluster contains many deeply embedded members with extinctions reaching @xmath040 @xmath79 . for sources with detections in at least two of the near - ir bands
, we have measured the extinction using a method developed by gutermuth ( 2006 ) .
this method uses the extinction law of flaherty et al ( 2006 ) and the yso loci from meyer et al .
( 1997 ) and gutermuth ( 2006 ) to determine the extinction and deredden the sources .
the slope of the sed of the dereddened source was also measured . in the majority of cases the sed slope and color - color diagram methods agreed ( 118 , 86% ) . where they did not , further examination was undertaken to ascertain where the discrepancy arose : 18 members ( 14% ) were classed differently via the irac and the dereddened sed @xmath80 methods .
visual inspection of these objects was carried out to better distinguish their class . in 17/18 cases
the class derived from the dereddened data was the final choice .
the remaining two objects were detected at 8.0 and 24 @xmath1 , and 24 @xmath1 only , and are coincident with known ( sub)mm protostars .
class 0/i sources are protostellar objects surrounded by in - falling dusty envelopes .
they are characterized by rising seds in the infrared @xcite .
the standard criteria , that class 0 sources with @xmath81 @xcite , can not be established from the _ spitzer _ data . as some known class 0 sources have been detected with _ spitzer _
we refer to all sources showing rising seds as class 0/i objects @xcite .
these sources were identified in our data by their rising seds ( @xmath82 ; note we used the 3.6 - 24 @xmath1 slope when available , these may be different to the 3.6 - 8.0 @xmath1 slopes listed in table [ tablephot ] ) , and by satisfying one of the two following criteria : @xcite : @xmath83 there can be some overlap between highly reddened class ii sources ( stars with disks ) and class 0/i objects @xcite . to distinguish between these sources , we used the dereddened seds .
sources that showed dereddened seds that looked like other class ii sources were deemed class ii .
these objects are found in the class 0/i region of the irac color - color diagram in fig .
although all class 0/i objects might be thought of as class ii like objects reddened by their protostellar envelopes ; the envelope also results in the scattering of a significant component of light in the near and mid - ir @xcite .
in addition , there may be a contribution from thermal emission from the inner envelope .
for this reason , class i objects can not be simply dereddened , and the application of our dereddening algorithm results in seds which appear distinctly different than the class ii objects ( see fig .
9 ) . in total
, 22 class 0/i protostars were identified .
flat spectrum sources possess a flat sed ( @xmath84 ) @xcite , they do not exhibit the steeply rising sed of protostars ( class 0/i ) , but the have too much excess for it to arise simply from a circumstellar disk ( class ii ) .
flat spectrum sources are thought to be an intermediate phase between the class 0/i and class ii phase where the central star and disk are surrounded by a thin infalling envelope @xcite .
initially 26 objects were tentatively classified as flat spectrum sources , using the undereddened @xmath80 , many because they lacked the necessary near - ir bands to further constrain their classification .
this class is particularly sensitive to contamination from agn , as both have similar colors .
each flat spectrum object was individually examined , and where possible its dereddened slope was used , to distinguish between class 0/i ( @xmath85 ) , class ii ( @xmath86 ) , and flat spectrum , ( @xmath87 ) .
we used the 3.6 - 24 @xmath1 slope for sources with 24 @xmath1 detections ; in other cases we used the 3.6 - 8.0 @xmath1 slope ( @xmath88 ) given in table [ tablephot ] .
finally , the locations of all the flat spectrum members were plotted on the irac color - color diagram ; these sources delineated the boundary between the class 0/i and ii objects ( fig .
8) . in total
, 16 flat spectrum objects were identified .
class ii sources are identified by having an @xmath89 @xcite , and the following mid - ir colors @xcite : @xmath90 note that we have lowered the minimum slope from -1.6 to -2 in order to contain evolved or anaemic disks found in other star forming regions @xcite .
the class ii members were refined by addition of the reddened class ii objects initially classed as protostars : in total 8 sources ( 6% ) were reclassified from class 0/i to class ii .
class ii sources from the irac-2mass diagrams were distinguished from protostars by their slopes in the available irac bands . in total , 62 class ii stars were identified .
seventeen sources were plotted that had an excess at 8.0 and/or 24@xmath1 but none at shorter irac wavelengths . @xmath91
these sources are considered to be transition disk objects , stars with a cleared inner disk but retaining an outer disk starting at a few au @xcite .
it is thought that the inner disk might be cleared due to formation of planets or the agglomeration of the dust particles into larger mm - sized grains @xcite .
those with weaker @xmath92 excess ( @xmath93 ) may be debris disks @xcite .
identification of these objects by sed slope is unreliable , due to the jump in flux at the longer wavelengths ( fig .
. visual examination of the sed must be carried out to verify that they show excess only longward of 8.0 @xmath1 .
they tend to lie near the lower boundary of the class ii region on the irac color - color diagram , showing little or no excess on the [ 3.6 ] - [ 4.5 ] color - axis and varying degrees of excess along the [ 5.8 ] - [ 8.0 ] color - axis ( fig .
they are more reliably identified using the irac - mips color - color diagram ( fig .
in total we find 17 transition disk candidates .
an additional 3 sources were initially classed by their mid - ir color excesses as class ii sources , but were identified as transition disk objects from the lack of excess shortward of 24 @xmath1 in their _ dereddened _ seds .
these sources appear in the class ii region in the irac and irac - mips color - color diagrams .
there is a possible source of contamination in the transition disk sources : agb stars , which have similar colors @xcite , with @xmath94 .
this affects eight of the sources , spectroscopy will be needed to determine whether these sources are contamination or members .
class iii objects are pre - main sequence stars which do not have circumstellar disks detectable in the mid - ir .
they exhibit colors in the color - color diagrams consistent with reddened photospheres , an sed which approximately follow a rayleigh - jeans law @xcite , and exhibit no excess in the mid - ir . in total , 21 of the x - ray identified ysos were found to be class iii
. we can not distinguish between class iii stars without elevated x - ray emission and field stars in the line - of - sight .
considering both the smaller field of view of the _ chandra _ observations and the fact that only a fraction of young stars show elevated x - ray emission ( see the x - ray discussion in sec [ xrayc ] ) , a substantial number of unidentified class iii objects may exist in the cluster .
further spectroscopic information on stars in the field will be required to positively identify these class iii sources .
we have identified 138 young stellar objects : 22 class 0/i objects , 16 flat spectrum sources , 62 class ii objects , 17 transition disks , and 21 class iii objects . in the following discussion
we use this sample to address four topics .
first , we compare our results to previous studies of the serpens embedded cluster .
second , we identify _ spitzer _ analogs to sub - millimeter sources previously found in this region .
third , we study the spatial distribution of the sources as a function of their evolutionary class .
finally , we discuss the x - ray properties of the ysos detected in the _ chandra _ x - ray observavtions . eiroa & casali ( 1992 ) observed the serpens cluster in the j , h , k , and nbl bands and identified 50 young stellar objects in the region .
we match all 50 sources to detections in our overlap region , although an offset of up to 9@xmath14 was needed to identify their counterparts in our data . of the 50 sources ,
34 have been identified as ysos in our analysis .
the remaining sixteen sources were found to have irac colors of reddened stellar photospheres without infrared excesses or to have too few bands to classify reliably , and none of these sixteen sources showed x - ray emission .
these sources can not be established as ysos by our criteria . a previous study by @xcite using the isocam instrument on - board iso , identified 77 cluster members in a 0.13 @xmath74 region centred on the serpens core , at wavelengths of 6.7 @xmath1 and 14.3 @xmath1 . of these ,
70 matched with a _
detection in the ir - region , with the remaining 7 outside of the overlap region but detected in one or more _ spitzer _ bands . a maximum offset of 9@xmath14
was used to match the sources to our data . of the 70 sources ,
56 are identified as ysos in our final catalogue , while the remaining 14 were found to be reddened stellar photospheres without infrared excesses or x - ray detections .
these remaining 14 were not identified as ysos in our analysis . of the ysos in common with kaas , two of our class
ii objects are classified as class i by kaas and one of our class iii objects is classified as class ii by kaas .
a previous study of the serpens core was undertaken in x - rays by @xcite and @xcite with the xmm - newton satellite . in one observation of @xmath95 ks and two observations of @xmath96
ks each over and energy range of 0.15 - 5 kev , 47 sources were detected ; 4 class i , 2 flat spectrum , and 41 class ii young stars .
these observations are not as deep as the 88 ks _ chandra _ integration , and do not possess as high an angular resolution .
we detect a total of 60 x - ray sources in the _ chandra _ data , 13 more members than the preibisch study .
twelve of these newly discovered x - ray sources are class 0/i or flat spectrum sources , indicating that the _ chandra _ observations are detecting more deeply embedded x - ray sources .
@xcite describes vla 3.5 cm observations of the serpens core , identifying 22 sources , thirteen of which matched to infrared counterparts .
nine of these are class 0/i sources including smm1 and smm5 , one is a flat spectrum , two are class ii and one is class iii .
three further sources were matched to counterparts detected in one or two bands only .
the class iii object is detected in the x - ray , thus all thirteen sources have one of our two criteria for membership : an infrared excess and/or an infrared counterpart ( with or without an excess ) to an x - ray detection .
the remaining six sources without infrared counterparts are most likely background sources which are too faint and/or reddened to be detected in the 2mass or _ spitzer
_ infrared data .
the fluxes of the vla sources coincident with the protostars detected in the ( sub)mm are given in table [ tablesmm ] .
eight of the 13 vla sources with infrared counterparts have _ chandra _ x - ray counterparts as well ; hence , 60% of the vla sources are detected by chandra . in comparison , in the @xmath97
ophiuchus core ( 165 pc ) , 8/28 or 36% of the vla sources had both x - ray and infrared counterparts @xcite .
in the coronet cluster ( 140 pc ) , 9/15 or 60% of the radio sources have been detected in multi - epoch analysis of x - ray data @xcite .
the measurements are consistent within the statistical uncertanties of the ratios , even though the sensitivities vary substantially ; the x - ray detection rates in the coronet cluster were found in 20 ks observations , while the @xmath97 oph core observation was nearly 5 times longer .
@xcite have recently submitted the results of the _ chandra _ observations of the serpens cluster .
this work was done in collaboration with our team , and they have reported the evolutionary classes listed in this paper .
table [ tableids ] , listing the coordinates and identifiers of the ysos , provides the cross reference for the @xcite source numbers in the eighth column .
@xcite presents irac observations of the serpens cloud as part of the _ spitzer _ legacy project `` from molecular cores to planet - forming disks '' ( c2d ; @xcite ) .
the surveyed area covers a much larger field of @xmath98 deg@xmath99 than covered in this paper .
they identified 257 young stellar objects throughout the field , in two main groupings .
one , labeled a in @xcite , is the serpens main core cluster studied in this paper , the second region , cluster b , being about @xmath100 to the south of the main core .
an important difference between this work and that of @xcite is the selection of class iii objects .
the @xcite criterion for class iii sources required an @xmath101 ; their class iii objects are sources with weak infrared emission from a disk that we would classify as a weak class ii or transition disk object . in contrast , our class iii sources are defined as lacking in infrared emission yet exhibit detectable x - ray emission . bearing this in mind , the numbers of class i , flat spectrum , class ii , and class iii members identified in their larger field were 30 , 33 , 163 , and 31 , respectively .
the number of young stellar objects identified is 138 .
there are 38 protostars ( 22 class 0/i , 16 flat spectrum ) and 100 pre - main sequence stars ( 62 class ii , 17 transition disks , and 21 class iii ) .
protostellar class 0/i and flat spectrum sources account for 28% of the ysos detected .
the ratio of protostars to pre - main sequence stars with disks is 48% ; this high number indicates that serpens is unusually rich in protostars , and is in agreement with the 56% fraction found in @xcite . over the entire serpens cloud
, the ratio was found to be 38% @xcite . in comparison ,
the fraction of protostars to stars with disks detected with the _ spitzer _ cores to disks ( c2d ) survey was 14% in the ic 348 cluster and 36% in the ngc 1333 cluster @xcite , and @xmath050% in chameleon ii @xcite .
previous observations at sub - millimeter and millimeter wavelengths have identified a number of presumably protostellar sources in the serpens cluster .
studies by @xcite , @xcite , and @xcite identify fourteen objects detected at wavelengths of 3 mm or 850 @xmath3 m with flux densities greater than 3 mjy beam@xmath23 , and significance above 5 @xmath19 .
table [ tablesmm ] lists these objects with their identifiers , coordinates , and fluxes from the above mentioned papers combined with our _ spitzer _ photometry .
these are also shown in fig .
[ fig12 ] .
we now discuss these sources : * smm1 : * at the location of smm1 we find a bright 24 and 70 @xmath1 source .
outflow knots associated with smm1 are visible in our 4.5 @xmath1 irac mosaic , and diffuse emission is visible in the 8.0 @xmath1 ; however , we do not identify a compact source at this location .
this appears to be a umambiguous example of a class 0 object too deeply embedded to be detected shortward of 24 @xmath1 .
* smm2 : * two _ spitzer _ identified class 0/i sources are found within @xmath102 of the smm2 source ; these sources appear to be beyond the positional uncertainties in the 3 mm data ( @xmath103 ) and are not coincident with smm2 .
there is also 70 @xmath3 m emission in the region around smm2 , it is not clear which of the three sources ( smm2 and the two class 0/i ) contribute to 70 @xmath3 m emission .
interestingly , there is a faint slightly extended 4.5 @xmath3 m source toward the position of smm2 .
this suggests that the smm2 source may be a protostellar source with associated infrared emission , although the lack of 24 @xmath3 m emission is puzzling .
* smm3 : * there is a 24 @xmath3 m source within @xmath104 of the location of smm3 , which we tentatively assign to it .
while there is extended 70 @xmath3 m emission in the region of smm3 , no point source could be photometered due to flux contamination from nearby bright sources .
while there is faint emission towards this source in the irac bands , it is confused with image artifacts from neighboring bright sources .
* smm4 : * there are no sources identified in the ir coincident with smm4 , but two small areas of nebulosity can be seen in the 4.5@xmath1 image .
again , this suggests that a protostellar source may reside in smm4 , but if so , there is a lack of bright 24 @xmath3 m and 70 @xmath3 m emission from the source .
* smm5 , smm9 , s68nb , s68nc and s68nd : * these five sources are associated with the group of four class 0/i objects on the northwestern side of the cluster .
in addition , the outflow jet associated with object s68nc is visible in the irac bands .
this region has previously been considered devoid of more evolved pre - main sequence stars , however we have identified two class ii members in the vicinity .
* smm6 , smm8 , smm10 , ps2 : * there is a flat spectrum source at the position of smm6 and class 0/i objects at the positions of smm8 , smm10 and ps2 .
recent work by @xcite identified smm6 as the primary component of a double system .
we do not see the companion in our data , as it is contaminanted by the flux from smm6 itself . *
smm11 : * while there is no ir counterpart at the position of smm11 from @xcite , a class 0/i object @xmath105 north of its position is detected out to 70 @xmath3 m .
we also note that the peak of smm11 in the scuba 850 @xmath1 data is @xmath106 southeast of our class 0/i source and @xmath107 northwest of the testi source .
while it is possible that these are three separate objects , it seems unlikely ; the position of this source needs to be re - examined in subsequent submillimeter observations . of the fourteen ( sub)mm objects , seven
correspond directly to sources detected in the irac bands , three are detected only with mips , and four are not detected at @xmath108 m . with the exception of smm11 , all of the ( sub)millimeter sources are coincident with 70 @xmath3 m emission ; however , the low angular resolution of the 70 @xmath3 m map ( @xmath109 ) makes it difficult to extract the 70 @xmath3 m photometry for an individual source . in seven of these sources
, the 70 @xmath3 m flux can be extracted ; these fluxes are tabulated in table [ tablesmm ] and the resulting seds are shown in figure [ figsmmseds ] .
for the remaining sources , the 70 @xmath3 m emission is too confused to photometer accurately .
three of the fourteen sources have x - ray detections in our data : smm5 , smm6 , and s68nb .
these appear to be class i sources in that they show bright emission shortward of 24 @xmath3 m ; there is no detectable x - ray emission toward the sources with weak or no detected emission shortward of 24 @xmath3 m , i.e. the probable class 0 sources .
the spatial distribution of young stellar objects in a cluster gives insight into the fragmentation processes leading to the formation of protostellar cores and the subsequent dynamical evolution of the stars as they evolve from the protostellar to the pre - main sequence stage @xcite .
recent work has shown that in many clusters the sources trace the underlying molecular gas distribution @xcite .
@xcite made the first maps of the distribution of class ii and class i objects in serpens , demonstrating that the class i sources were significantly more clustered than the class ii sources .
[ fig13 ] shows the distribution of young stars as a function of their evolutionary class .
this is shown for the sample of ir - excess stars and the sample of x - ray luminous stars .
the most significant difference is between the class 0/i and flat spectrum sources , which are concentrated in a narrow filament in the center of the cluster , and the class ii sources .
although there is a peak in the class ii density in the center of the cluster , the majority of the class ii sources , as well as the transition disk and class iii sources , are found in an extended halo surrounding the protostars .
the protostars are coincident with the dense molecular ridge mapped by the 850 @xmath1 scuba map of @xcite .
the class 0/i and flat spectrum sources concentrated in two groups coincident with the two dominant molecular gas clumps ( fig .
[ fig12 ] ) , confirming the previous work of @xcite and @xcite .
these are coincident with the column density peaks of the 850 @xmath3 m map and have size 0.2 pc in diameter .
the northern group contains 9 class 0/i and 1 flat spectrum source , seven of which are found in a region only 0.1 pc in diameter , and 4 class ii sources .
the southern group contains a mixture of 12 class 0/i , 10 flat spectrum , 16 class ii , 1 transition disk , and 5 class iii sources in a @xmath110 pc diameter region . in the southwestern quadrant of this grouping
is a small wishbone shaped sub - grouping .
this sub - grouping of stars contains eight stars in an area 0.1 pc in diameter ; including three class 0/i objects , one flat spectrum source and four class ii sources
. the higher fraction of class ii sources in the southern group suggests that it is more evolved than the northern group .
finally , the x - ray data shows a grouping of four class iii objects in the southwestern quadrant of the cluster within a 0.1 pc diameter region .
this grouping is unusual in that it is the only apparent group of class iii objects ; all of the remaining class iii objects are scattered around the cluster .
if this is a bona fide group and not a chance alignment , it is of great interest why these objects have dispersed their circumstellar material .
the spatial distribution of the cluster sources was examined for each of the five evolutionary classes using a nearest neighbour technique .
the nearest neighbour distance is the projected distance to the nearest yso of the same evolutionary class , using the adopted distance of 260 pc .
figure [ fignn ] ( _ left _ ) shows the distribution of nearest neighbour distance between the members of each class . the class
0/i sources are by far the most densely clustered , with a median separation of 0.024 pc . in comparison , the median nearest neighbor distances of the remaining evolutionary classes are significantly greater : flat spectrum - 0.079 pc , class ii - 0.097 pc , transition disks - 0.132 pc and class iii - 0.131 pc .
the median distances for the flat spectrum and transition disk are biased to higher values by the small numbers ( and hence low densities ) of these sources .
the class iii sources mean distance may be biased to a lower values by the limited field of view of the _ chandra _ data .
these scales are smaller than the 0.12 and 0.25 pc separations for class i and ii previously derived from isocam observations @xcite , which were limited by the lower angular resolution of iso . also plotted in figure [ fignn ] are the cumulative distribution curves for the ysos by class , showing how densely the sources are distributed over the field . a kolmogorov - smirnov ( k - s ) test was performed to ascertain the probability that they are derived from the same parent distribution .
the class i sources are dissimilar to each of the other classes , reflecting their more highly clustered nature ( table [ tprob ] ) .
the distributions of the four remaining classes are statistically indistinguishable ( table [ tprob ] )
. for each evolutionary class , 10k random distributions of stars were generated , with the number of stars equal to the number of objects in the given evolutionary class .
for the class iii sources , the randomly distributed stars were constrained to fall within the irx - field , for all other classes , the random distribution covered the region of the ir - field that contained 90% of the sources .
the resulting nearest neighbor distributions of the observed ysos and the random distributions were compared for each class using the k - s test .
the probabilities that the random and observed distributions were drawn from the same parent distribution were calculated for all 10k distributions ; the mean values of the probablilities for each class are listed in table [ tprob ] .
our results indicate that the median projected spacing of protostars in serpens sources is only 5000 au , and the spacing is as close as 2000 au in certain regions such as the wishbone and the northern group .
the actual physical spacings are probably larger ; if we assume the sources are randomly distributed relative to each other , the median spatial separation between sources would be @xmath111 au .
if we assume that each protostar will form an 0.5 @xmath65 star , then for dense core densities of @xmath112 and @xmath113 @xmath114 @xcite , such a protostar would have to accrete from a volume 12,000 and 4000 au in diameter , respectively .
thus , assuming volume densities typical of dense molecular cores , the accretion volumes would be densely packed if not overlapping .
is the observed spacing consistant with jean s fragmentation ?
the jean s length is given by @xcite : @xmath115 .
if we set the jean s length to be @xmath116 pc , adopt temperatures of 12 and 17 k @xcite , and solve for the density , the resulting densities are @xmath113 cm@xmath23 and @xmath118 cm@xmath23 .
the observed gas densities range from @xmath112 @xmath114 to @xmath113 @xmath114 @xcite .
thus , the observed median separation is consistant ( i.e. within a factor of two ) with jeans fragmentation given the highest observed densities in serpens . at these highest observed densities , the spacing would also be approximately equal to the radius of a critical bonner - ebert sphere ( @xmath119 ) @xcite . in either case , the spacing of objects at the jeans spacing and bonner - ebert radius further indicate that the volumes of gas accreting onto the protostars are tightly packed or overlapping .
we note that unlike the case of @xcite , there is not a well defined peak in the distribution of nearest neighbor separations ; many objects are found at separations much smaller than the median separation .
this is in keeping with @xcite who resolved one of the teixeira et al .
sources into a multiple system .
the close spacing of the protostars raises the possibility that competitive accretion is taking place . in competitive accretions models , several protostars accrete from a common reservoir of gas .
these objects compete for gas from the reservoir ; this process can lead to a distribution of masses similar to the initial mass function @xcite . in models of competitive accretion , protostars accrete mass through a bondi - hoyle accretion at a rate of @xmath120 where @xmath97 is the volume density of the gas , @xmath121 is the instantaneous mass of the accreting protostar , @xmath122 is the sound speed of the gas and @xmath123 is the velocity of the star relative to the gas @xcite . @xcite
measured the velocities of 5 of the protostars in the northern group by using detections in the @xmath124 ( @xmath125 ) line conicident with the protostars .
the rms 1-d velocity dispersion of these protostars is 0.26 km s@xmath23 , implying a 3-d velocity dispersion of 0.45 km s@xmath23 . assuming a density of @xmath126 @xmath114 , a velocity of 0.45 km s@xmath23 , and a kinetic temperature of 17 k ( implying a sound speed of 0.22 km s@xmath23 ) , for a protostar with a mass of 0.1 m@xmath127 , the bondi - hoyle accretion rate is @xmath128 m@xmath127 per year , and for a protostar with a mass of 0.5 m@xmath127 , the accretion rate is @xmath129 m@xmath127 per year . given these accretion rates , adopting a protostellar lifetime of 300,000 @xcite , and considering that the rate will increase as the object grows in mass , the young protostars in the northern group could accrete a significant portion of their ultimate mass through bondi - hoyle accretion . consequently
, competitive accretion remains a viable process for the formation of stars in the serpens cluster . finally , the close spacing and velocities of the protostars in the northern group suggest that their envelopes could impinge on one another .
we note that the 3 mm continuum observations ( @xmath130 beamsize ) of @xcite did not resolve the the sizes of the protostars in the northern group of serpens .
this does not rule out our previous estimates that the accretion volumes extend 5000 to 10,000 au ; the bima measurements may only be sensivite to dense inner regions of the protostars .
however , for the present analysis , we use a size of 1000 au , assuming that the protostars are spheres with diameters just below the angular resolution of the bima observations .
we follow the analysis in @xcite , adopting a mass of 0.5 m@xmath127 , a velocity of 0.26 km s@xmath23 and a peak density for the northern group of seven protostars in a volume 0.1 pc in diameter .
the resulting time between collisions for a given object is 200,000 years , less than the protostellar lifetime .
consequently , if the objects in this group remain in bound orbits within this region for 300,000 years , the typical protostellar lifetime @xcite , all of the protostars could experience collisions between their envelopes and the envelopes of other protostars in the group . in summary , jean s length ,
bonner - ebert sphere and simple volumetric arguments all suggest that the accreting envelopes around the protostars in serpens are often tightly packed , if not overlapping .
we find that for the conditions present in the northern group , the protostars may accrete a significant amount of their mass through bondi - hoyle accretion from a common envelope , and competitive accretion may occur within the group .
furthemore , collisions between the sources are likely . in total , our analysis suggests that the interactions between protostars can play an important role in the formation of stars within the dense groups observed in serpens . using the models of @xcite we calculate the magnitude of the hydrogen burning limit at k - band to be 12.5 mag for an assumed 1 myr old cluster at a distance of 260 pc to serpens .
of the 138 sources in serpens , 34 have k - band detections fainter than 12.5 magnitudes , 2 of which are class 0/i and one flat spectrum . since we can not measure the dereddened photospheric luminosities of the class 0/i and flat spectrum objects , these sources are discounted , leaving 31 candidate ( 23 class ii and 8 class iii ) brown dwarfs members .
@xcite has also identified numerous substellar candidates in the serpens b cluster to the south of the core .
we can not rule out that these sources are older objects or objects that appear underluminous for other reasons @xcite .
searches for brown dwarfs in other regions have shown that infrared spectroscopy is essential to confirm that faint objects are protostellar in nature @xcite .
assuming that the 23 class ii objects were to be positively identified as substellar , this would lead to an upper limit of 23/138 or 17% of the total yso population being substellar .
this would be somewhat higher than the substellar fractions found in chameleon 1 ( 5% ) and ic348 ( 8% ) @xcite , though in agreement with @xcite , who estimate 20% of serpens sources are substellar .
interestingly , the spatial distribution of the @xmath131 12.5 class ii objects appears more extended that the brighter class ii ( fig .
[ figbdc ] ) .
in particular , although smaller in number , the most distant objects north , west , and east of the core of the cluster have @xmath131 12.5 .
this may be biased by the high reddening in the centre of the cluster which would preclude detections of faint objects in the near - ir .
using deeper near - ir photometry of the center of the cluster @xcite also found that sources with dereddened @xmath131 12.5 were non - centralised .
spectroscopy of the fainter objects is needed to confirm their substellar nature . in recent years
many studies have been undertaken to investigate the emission properties of pre - main sequence stars and protostars in the higher energy x - ray region of the spectrum @xcite .
in developed , hydrogen burning stars , x - ray activity arises from magnetic fields generated as a result of shear between the core radiative zone and the outer convective zone .
the process behind the generation of the highly increased levels of emission in young stellar objects remains uncertain ; suggested causes being magnetic disk - locking between the star and disk @xcite , accretion onto the star @xcite , and alternative dynamo models for coronal emission @xcite . in main sequence stars , there is a trend of decreasing x - ray activity with age ; this trend has not yet been conclusively shown for pre - main sequence stars .
_ chandra _ has recently examined the x - ray properties of the orion nebula cluster ( onc ) , in a project known as the _ chandra _ orion ultra - deep project ( coup ) .
although the serpens sample is much smaller than the coup survey , the analysis of serpens has the advantage of excellent 3.6 - 24 @xmath1 photometry ; observavtions longward of 4.5 @xmath1 are difficult in the center of the onc due to the bright infrared emission from the molecular cloud .
this photometry allows us to accurately ascertain the evolutionary class of each object , and study the dependence of the x - ray properties with evolutionary state .
only those detections with x - ray count in excess of 100 were considered when examining emission properties to ensure reliable estimates . of the 138 identified members , 60
were conclusively shown to exhibit x - ray emission .
nine of the x - ray detections are associated with class 0/i sources ( 43% of class 0/i s in irx - field ) , a further eight with flat spectrum members ( 56% ) .
twenty detections have class ii counterparts ( 45% ) , while two of the transition disks are detected in x - rays ( 25% ) .
the remaining 21 sources did not exhibit an ir - excess and were identified as a class iii yso solely on their x - ray emission ; hence , 21/138 or 15% of the sources were identified solely by chandra .
there appears to be no evidence that the detection rate depends on evolutionary class ; within 1 @xmath19 , the detection rates of class 0/i , flat spectrum , and class ii sources are identical
. the detection rate may be largely a function of the sensitivity of the _ chandra _ observations ; the majority of t tauri stars have @xmath132 @xmath133 @xmath134 @xcite while our sample is complete to @xmath132 @xmath133 @xmath135 .
since we can not identify class iii sources without x - ray emission , we can not establish the fraction of class iii sources detected in x - rays . as we will argue subsequently ,
the x - ray properties of the class ii and lii objects appear to be indistinguishable . assuming that this similarity holds for the fraction of detected sources , then the fraction of classs iii sources detected is the same as that for the class ii objects , and we estimate the total number of class iii objects to be @xmath136 . if , further , we take into account that the irx - field contains only 71% of the class ii population , and we assume that the class iii sources show a similar spatial distribution , then the total population of class iii objects may be as high as 65 .
the ratio of the number of x - ray detected pre - main sequence stars with disks ( class ii and transition disk ) to the total number of x - ray detected pre - main sequence stars ( class ii , class iii , and transition disk ) is found to be 22/(22 + 21 ) = 51% .
the number counts method of @xcite was not applicable here due to the high density of background stars .
these values are comparable to the disk fraction of a @xmath02 myr old cluster @xcite ; consistent with the suggestion of @xcite that the class ii population of serpens is 2 myr old .
however , this older age seems odd considering the large number of protostars in this region . in a future paper , we will examine the positions of these objects on the h - r diagram to better determine the age of the serpens cluster .
fig.[figx](a ) shows plasma temperature against the column density of hydrogen in the intervening gas between us and the hot plasma .
the class i and flat spectrum sources show column densities five times greater than the class ii and iii sources , with one exception .
this is not surprising since the class 0/i and flat spectrum sources are surrounded by infalling envelopes .
the one exception is a class ii object with an unusually high column density , this may be a source where an edge - on disk around the star is occulting the hot plasma .
there does appear to be a correlation of plasma temperature with column density : the values of these 26 sources show a spearman rank coefficient of 0.55 .
however , the mean values of kt for @xmath137 greater and less than @xmath138 @xmath139 are @xmath140 kev and @xmath141 kev , respectively , indicating that there is no strong dependence .
this indicates that the column density is not biasing our determination of the temperature to higher values by selectively absorbing the soft x - ray flux .
soft x - ray photons are more highly absorbed than hard x - ray photons , by up to 90% when @xmath142 @xcite .
further more , it shows that there is no significant trend of plasma temperature with evolutionary class .
fig.[figx](c ) shows the x - ray flux corrected for absorption and @xmath143 for the different classes .
although the class 0/i objects appear slightly more luminous , this may be a bias due to the fact that the class 0/i sources are much more highly absorbed and consequently the weaker class i objects may not have had the 100 counts we require to perform the analysis of the x - ray spectrum .
the 8 class ii and 9 class iii sources have statistically indistinguishable values ; the mean values of the class ii are @xmath143 @xmath144 kev and @xmath145 @xmath146 @xmath147 @xmath148 and the mean value for the class iii sources is @xmath143 @xmath149 kev and @xmath145 @xmath150 @xmath147 @xmath148 .
the mean values for the class i are @xmath143 @xmath151 kev and @xmath145 @xmath152 @xmath147 @xmath148 , for the flat spectrum are @xmath143 @xmath153 kev and @xmath145 @xmath154 @xmath147 @xmath148 , and for the transistion disks are @xmath143 @xmath155 kev and @xmath145 @xmath156 @xmath147 @xmath148 . using the combined class ii and class iii sample we find only some indication of a weak correlation of kt with x - ray flux , with a spearman rank coefficient of 0.39 . in comparison ,
@xcite find a trend of increasing kt with x - ray in a combined study of several young clusters .
finally , the mean plasma temperatures for the class 0/i and flat spectrum sources are not significantly different than those values for class ii and class iii sources .
thus , we find no significant dependence of plasma temperature on evolutionary class for the detected x - ray sources . the x - ray flux of pre - main sequence stars has been found to vary with bolometric luminosity @xcite .
although we can not measure the bolometric luminosities , we can examine the x - ray flux as a function of the dereddened j - band magnitude , which is the infrared band least contaminated by infrared excess emission and is thus a good proxy for photospheric emission ( fig.[figx](b ) ) .
we find again that the class ii and class iii sources appear indistinguishable .
in particular , if we limit our analysis to dereddened @xmath157 where we detect 90% of the class ii sources in x - rays , we find that the @xmath145 @xmath158 @xmath147 @xmath148 for class ii and @xmath145 @xmath159 @xmath147 @xmath148 for class iii . a similar lack of dependence is found in ngc 1333 and @xmath19 ori @xcite .
this suggests that the mechanism generating the x - ray flux is similar during the class ii and iii phases .
@xcite find a clear relationship between @xmath145 and j - band magnitude of slope @xmath160 in ngc1333 , while @xcite find a slope of 0.30 in the ophiuchi cloud .
our results give a slope of @xmath161 , which is consistent with these other young clusters , although there is a large uncecrtainty due to the scatter in our data . in conclusion
, we find that there is no clear observational motivation for invoking different mechanisms for generating the x - ray emission in the class i , ii and iii phases .
the combined x - ray and infrared results can be used to calibrate the relationship between gas column density and the extinction measured in the infrared .
we calculated @xmath6 for each star using the method of @xcite , which is based on the reddening loci from @xcite and the extinction law of @xcite .
these values can be compared to the column density of hydrogen atoms , @xmath7 , which is calculated from the inferred absorption of the x - ray emission .
previous measurements of this value lead to an approximately linear fit of @xmath162 @xcite for stellar sources , while the value for the diffuse ism ranges from @xmath163 @xcite .
fig.[figx](c ) plots the hydrogen column density against the calculated exctinction in the k - band .
as shown in this plot , the @xcite relationship agrees with our data for @xmath164 , but diverges from the points with @xmath8 .
a similar divergence has also been found in the more deeply embedded stars in the rcw 108 region @xcite .
interestingly , the class ii sources always show a higher value of @xmath7 for a given @xmath6 .
the colors of the class ii objects can be affected by both absorption and scattering of the light from the central source by a flared disk ; the extent of this effect would be determined , predominantly , by the inclination of the disk and the sizes of grains in the disk @xcite .
for example , the colors of an edge - on disk system , where the central star is highly obscured by the disk , would be dominated by scattered light .
in such a case , the disk absorbs both the infrared and the x - ray emission from the star , but the scattered infrared light reduces the observed reddening of the star and hence the inferred value of @xmath6 . to eliminate the uncertainties in the determination of @xmath6 due to circumstellar material , we fit only the class iii objects .
the resulting relationship is @xmath165 . assuming a standard conversion of @xmath166 @xcite , this slope is 2.5 times lower than that derived by previous authors . in comparison ,
a fit to the class ii objects results in @xmath167 . the poor fit and higher slope are further indication that the disks around these objects affect the measurement of @xmath6 , possibly leading to spuriously low values of @xmath6 .
we now focus on the slope derived from the class iii objects , the possible systematic uncertainties in that slope , and a possible explanation for the low value of the slope .
_ systematic underestimation of @xmath7 : _ if the measurements of @xmath7 were systematically underestimated by a factor of 2.5 , then this would lead to the shallower slope observed .
however , the determinations of the @xmath7 values appear to be robust .
we calculated @xmath7 only for those sources with relatively high s / n ( @xmath168 counts ) .
these values of @xmath7 were found to be consistent using two different models for the plasma .
a bias could occur in the determination of @xmath7 when the soft x - ray emission is totally absorbed ; however inspection shows that this is not the case here .
furthermore , values of @xmath7 calculated by our software for stars in the orion coup survey are within 5% of the values published in @xcite ; this is further validation of the accuracy of our analysis .
_ sytematic overestimation of @xmath6 : _ the value of @xmath6 is determined by the selective absorption in the near - infrared bands @xcite ; this selective absorption was converted to @xmath6 by using an assumed value of @xmath169 .
we use the @xcite reddening law where the adopted ratio of @xmath170 .
this is equivalent to a near - ir general to selective absorption of @xmath171 .
changes in the reddening law in the molecular cloud could change this ratio . in molecular clouds ,
@xcite report a flattening in the infrared reddening law for @xmath172 @xmath1 compared to the diffuse ism reddening law measured by @xcite ; however , they found no significant difference in the reddening law for @xmath173 @xmath1 between the molecular clouds and the diffuse ism . alternatively , @xcite report @xmath169 of 1.44 , lower than the value in @xcite , and they find evidence of variations in this value toward different lines of sight toward the galactic center .
if the slope found in our data were wholly due to an overestimation of @xmath6 , then the value of @xmath169 would have to decrease to 0.73 , and the value of @xmath174 must change to 2.37 .
this value of @xmath174 is higher than the values in the literature and would require a steepening of the reddening law between 1 & 2 @xmath1 .
_ growth and coagulation of dust grains : _ the ratio @xmath175 can be affected by increases in the sizes of dust grains ; this would simply increase the value of @xmath6 and hence reduce the ratio of @xmath175 .
the value of @xmath6 is given by : @xmath176 where @xmath177 is the column density of dust , @xmath178 is the radius of a dust grain , and @xmath179 is the extinction efficiency @xcite . in the limit of @xmath180 , @xmath179 , which would be the case for a standard grain size distribution @xcite of @xmath181 , @xmath182 where @xmath183 is the refractive index .
thus , there is a strong dependence on the size of the grain , and the value of @xmath6 can be changed simply by growing grains to larger sizes ; for example the scattering term in @xmath179 goes as @xmath184 , hence a 25% increase in grain size can increase the scattering efficiency by a factor of 2.5 .
grain growth in molecular clouds may occur through both coagulation and the accretion of grain mantles ; these processes would be accelerated in the dense molecular gas found in the serpens cloud core .
for example , @xcite explained changes in the @xmath185 ratio in ophiuchus through the coagulation of grains .
the collision time between grains is @xcite @xmath186 v_d [ km~s^{-1}])^{-1}\end{aligned}\ ] ] for a @xmath187 and a velocity of 0.1 km s@xmath23 @xcite , the collision timescale is 30,000 years .
thus it is likely that a dust grain in the serpens cloud would have had multiple collisions with other grains .
alternatively , the growth can be through the accretion of mantles of volatiles such as h@xmath188o , nh@xmath189 , o@xmath188 , n@xmath188 and co. the accretion timescale is given by @xmath190 where s is the specific density of the mantle , @xmath191 is the change in the size of the dust grain due to the mantle , @xmath192 is the sticking coefficient ( @xmath193 ) , @xmath194 is the temperature of the gas , @xmath195 is the density of the molecule forming the mantle , and @xmath183 is the mass of the molecule forming the mantle @xcite .
thus the rate of mantle accretion is proportional to the density of the gas ; again , the high density of the gas in the serpens cloud would accelerate the rate of mantle accretion .
a detailed calculation of the rate requires consideration of the gas phase abundances of the different possible volatiles , and is beyond the scope of this paper .
self consistent modeling of the extinction over multiple wavelength bands is needed to study the impact of grain growth on both the relationship between @xmath7 and @xmath6 and the shape of extinction law with which @xmath6 is derived .
further observational work is also needed to confirm our observed @xmath175 slope using a larger number of class iii objects from multiple star forming regions . however
, this study demonstrates the value of using combined chandra , near - ir , and spitzer observations to identify deeply embedded class iii objects and search for variations in the extinction law within molecular clouds .
we have undertaken an intensive study of the serpens cloud core in the mid - ir with _
spitzer _ and in the near - ir with 2mass .
we have merged these data with _
chandra _ x - ray observations and investigated various x - ray properties of protostars and pre - main sequence stars .
these data provide perhaps the most complete census to date of the young stellar objects in one of the nearest embedded clusters to the sun .
* we identify 138 young stellar objects , after removing contamination .
these include 38 protostars ( 22 class 0/i , 16 flat spectrum ) and 100 pre - main sequence stars ( 62 class ii , 17 transition disks , and 21 class iii ) . the ratio of protostellar objects ( class 0/i and flat spectrum ) to pre - main sequence stars ( class ii , transition disk , and class iii ) is 38:100 or 29% of the total number of identified members .
the ratio of protostars to pre - main sequence stars with disks is 48% .
these high numbers indicate that serpens is unusually rich in protostars .
we estimate that as many as 44 class iii sources may be missing , implying a total membership of 182 .
* we have searched for _ spitzer _ counterparts to 14 ( sub)mm sources in the serpens cluster .
seven are detected as point sources at 3.6 - 8.0 @xmath1 , a total of 10 have been detected at 24 @xmath1 , and an additional two show nebulosities at @xmath196 @xmath1 .
all but one are coincident with 70 @xmath1 emission , although for many of the protostars , the low angular resolution of the 70 @xmath1 data precludes accurate photometry of the sources . *
we find that the spatial distribution of the protostars is distinctly different than the pre - main sequence stars . while the class 0/i sources and most of the flat spectrum sources are concentrated in the central molecular gas filament , the class ii , transition disk and class iii sources are distributed over a much more extended region .
the median projected spacing of the class 0/i sources is only 5000 au ; these sources are highly clustered .
in contrast , the distribution of the remaining classes are statistically indistinguishable from a random distribution . *
the protostars are concentrated in two groups coincident with the two dominant molecular gas clumps , as has been previously reported .
the density of protostars in these groups suggest that the volumes of gas from which these protostars accrete may touch or overlap .
the spacing is also similar to both the jean s length and the radii of critical bonner ebert spheres .
based on published ambient gas densities and velocities of the protostars in the northern group , we find that competitive accretion could be occurring in this groups .
furthermore , collisions between protostellar envelopes may be common .
* of the 138 sources , 60 exhibit detectable x - ray emission .
nine of the detections are associated with class i sources ( 43% of members in class ) , a further eight with flat spectrum members ( 56% ) .
twenty - one detections have class ii counterparts ( 45% ) , while only two of the transition disks are detected in x - rays ( 25% ) . twenty class iii sources were identified solely by their x - ray emission .
there appears to be no evidence of a trend in the detection rate of sources by evolutionary class .
the fraction of x - ray detected pre - main sequence stars ( i.e. class ii , transition disks and class iii sources ) with disks is 51% .
* we find no evidence for a dependence of the temperature of the x - ray emitting plasma on the evolutionary class , and find no difference in the x - ray luminosity of detected class ii and classs iii objects .
consequently , there is no evidence that the generation mechanism of the x - ray emission depends on the presence of an infalling envelope or circumstellar disk .
* we find an observed relationship of @xmath165 , which is much shallower than found in other studies .
the most likely explanation for this is a change in the near - ir reddening law that would affect the shape and magnitude of the reddening law .
such changes could result from grain coagulation and/or grain growth from the accretion of mantles of volatiles ; both such processes may occur relatively quickly due to the high gas ( and presumably dust grain ) densities in the serpens cloud core .
this work is based on observations made with the _ spitzer _ space telescope ( pid 6 , pid 174 ) , which is operated by the jet propulsion laboratory , california institute of technology under nasa contract 1407 .
support for this work was provided by nasa through contract 1256790 issued by jpl / caltech .
support for the irac instrument was provided by nasa through contract 960541 issued by jpl .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
this research has made use of the nasa / ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
e. winston would like to thank the irish research council for science , engineering , and technology ( ircset ) for funding and support .
cccccc [ tprob ] random & @xmath197 & 0.05 & 0.13 & 0.67 & 0.21 + i & & @xmath198 & @xmath199 & @xmath200 & @xmath201 + flat & @xmath198 & & 0.45 & 0.28 & 0.73 + ii & @xmath199 & 0.45 & & 0.016 & 0.60 + trans . &
@xmath200 & 0.28 & 0.016 & & 0.35 + cccccccc 1 & 18:30:02.09 & 1:13:58.81 & ck7 & & & & + 2 & 18:29:57.83 & 1:12:51.62 & ec95 & kaas317/345 & & 72 & 53 + 3 & 18:29:59.57 & 1:11:58.80 & & kaas330 & & 25 & 60 + 4 & 18:29:58.76 & 1:14:26.47 & ec103 & kaas326 & & 23 & + 5 & 18:29:47.01 & 1:16:26.82 & ec27 & kaas237 & & & + 6 & 18:29:49.24 & 1:16:31.74 & & kaas249 & & & + 7 & 18:29:49.58 & 1:17:05.92 & ec76 & kaas254 & s68nb & 17 & 28 + 8 & 18:29:57.59 & 1:13:00.44 & svs4 - 10 & kaas312 & & 12 & 46 + 9 & 18:29:56.88 & 1:14:46.62 & ck3 & kaas307 & & 67 & 43 + 10 & 18:30:02.73 & 1:12:28.14 & ec129 & kaas347 & ps2 & & + 11 & 18:29:49.11 & 1:16:19.97 & & & sn68d & & + 12 & 18:29:57.64 & 1:13:04.54 & & & & & + 17 & 18:30:00.68 & 1:13:01.23 & & & & & + 18 & 18:29:59.92 & 1:13:11.58 & & kaas331 & & & + 19 & 18:29:59.61 & 1:14:12.33 & & & & 78 & 61 + 20 & 18:29:52.85 & 1:14:56.08 & & kaas276 & & & 33 + 21 & 18:29:52.20 & 1:15:47.59 & & kaas270 & smm10 & & + 22 & 18:29:52.20 & 1:15:59.02 & & & & 20 & 32 + 23 & 18:29:48.09 & 1:16:44.98 & & kaas241 & smm9/s68n & & + 24 & 18:29:51.14 & 1:16:40.49 & & kaas265 & smm5 & 18 & 30 + 27 & 18:29:58.18 & 1:15:21.91 & ck4 & kaas321 & & 74 & 55 + 28 & 18:30:00.20 & 1:14:03.56 & & & & 79 & 62 + 30 & 18:30:01.09 & 1:13:24.44 & ec121 & kaas341 & & 81 & 66 + 31 & 18:30:01.53 & 1:15:14.93 & ec123 & & smm8 & & + 32 & 18:29:18.85 & 1:14:16.94 & & & & & + 33 & 18:30:23.43 & 1:05:04.84 & & & & & + 34 & 18:29:53.84 & 1:12:47.40 & & & & & + 35 & 18:29:57.72 & 1:14:06.06 & svs20 & kaas314 & smm6 & 70 & 48 + 36 & 18:29:31.97 & 1:18:42.66 & & kaas159 & & 9 & 9 + 37 & 18:29:57.82 & 1:12:37.97 & ec94 & kaas318 & & 22 & 52 + 38 & 18:29:55.68 & 1:14:31.34 & ck9 & kaas298 & & 63 & 37 + 39 & 18:29:54.07 & 1:07:11.10 & & kaas287 & & & + 40 & 18:30:18.17 & 1:14:16.89 & & & & 44 & 83 + 44 & 18:29:59.38 & 1:10:40.92 & & & & & + 45 & 18:29:59.02 & 1:12:24.92 & & kaas327 & & & + 47 & 18:30:02.48 & 1:12:43.44 & & & & & + 48 & 18:29:49.70 & 1:14:56.71 & ec40 & kaas253 & & & + 53 & 18:29:14.32 & 1:07:30.14 & & & & & + 54 & 18:30:05.01 & 1:12:35.99 & & & & & + 55 & 18:30:22.83 & 1:16:21.34 & & & & & + 56 & 18:29:55.56 & 1:08:34.34 & & & & & + 57 & 18:29:21.08 & 1:24:06.66 & & & & & + 58 & 18:30:22.74 & 1:13:23.58 & & & & & + 59 & 18:29:59.22 & 1:14:07.82 & ck8 & kaas328 & & 77 & 59 + 60 & 18:30:04.09 & 1:06:55.86 & & & & & + 61 & 18:29:56.66 & 1:12:39.17 & ec80 & kaas306 & & 21 & 42 + 62 & 18:29:55.19 & 1:13:22.43 & ec73 & kaas294 & & & + 64 & 18:29:38.18 & 1:09:17.53 & & & & & + 65 & 18:29:43.92 & 1:07:20.99 & & kaas219 & & & + 66 & 18:29:51.43 & 1:09:32.94 & & & & & + 67 & 18:30:09.22 & 1:17:52.66 & & & & & + 68 & 18:29:47.19 & 1:07:16.02 & & & & & + 69 & 18:29:53.83 & 1:13:30.80 & ec68 & kaas285 & & & + 70 & 18:29:57.81 & 1:12:28.17 & ec91 & kaas320 & & & + 71 & 18:29:44.28 & 1:04:53.50 & & kaas221 & & & + 72 & 18:29:04.26 & 1:18:39.08 & & & & & + 73 & 18:30:07.70 & 1:12:04.49 & & kaas366 & & 86 & 75 + 74 & 18:30:13.28 & 1:02:48.93 & & & & & + 75 & 18:29:32.24 & 1:12:39.94 & & & & & + 76 & 18:30:16.97 & 1:13:07.47 & & & & & + 77 & 18:30:00.89 & 1:22:03.59 & & & & & + 78 & 18:30:03.41 & 1:16:19.45 & ec135 & kaas348 & & 84 & 69 + 79 & 18:29:46.28 & 1:10:25.33 & & & & & + 80 & 18:29:56.53 & 1:12:59.53 & svs4 - 2 & kaas304 & & 66 & 40 + 81 & 18:29:53.59 & 1:17:01.91 & ec67 & kaas283 & & 60 & 34 + 82 & 18:29:33.37 & 1:08:24.42 & & kaas173 & & & + 83 & 18:29:57.80 & 1:15:31.89 & ck13 & kaas319 & & 71 & 50 + 84 & 18:30:06.08 & 1:06:17.11 & & kaas357 & & 42 & 73 + 85 & 18:29:56.94 & 1:12:47.88 & ec84 & kaas309 & & 68 & 44 + 86 & 18:29:53.16 & 1:12:27.78 & & & & & + 87 & 18:29:54.50 & 1:14:47.11 & ec70 & kaas291 & & 61 & 35 + 88 & 18:30:41.16 & 1:13:15.94 & & & & & + 89 & 18:30:14.42 & 1:04:06.44 & & & & & + 90 & 18:30:44.52 & 1:23:30.19 & & & & & + 91 & 18:29:35.18 & 1:13:10.95 & & & & & + 92 & 18:30:01.66 & 1:04:42.96 & & & & & + 93 & 18:29:42.15 & 1:20:20.93 & & kaas216 & & 47 & + 94 & 18:29:39.87 & 1:17:56.02 & & kaas202 & & 57 & 18 + 95 & 18:29:51.18 & 1:13:19.71 & ec51 & kaas266 & & 49 & 31 + 96 & 18:29:53.09 & 1:06:17.85 & & & & & + 97 & 18:29:05.82 & 1:13:22.94 & & & & & + 98 & 18:30:04.89 & 1:14:39.48 & & & & 85 & 71 + 99 & 18:29:53.54 & 1:13:05.34 & & & & & + 100 & 18:29:44.53 & 1:13:11.59 & ec11 & kaas224 & & 59 & 23 + 101 & 18:30:11.09 & 1:12:37.93 & & kaas393 & & 87 & 79 + 102 & 18:30:27.67 & 1:22:35.02 & & & & & + 103 & 18:29:41.44 & 1:07:38.15 & & kaas207 & & & + 104 & 18:29:20.41 & 1:21:03.78 & & & & & + 105 & 18:29:57.89 & 1:12:46.42 & ec98 & kaas322 & & 73 & 54 + 106 & 18:29:48.77 & 1:13:42.12 & ec33 & kaas242 & & & + 109 & 18:30:02.31 & 1:08:08.82 & & & & & + 110 & 18:30:00.31 & 1:09:44.59 & & & & 13 & 63 + 112 & 18:29:56.28 & 1:13:21.14 & & & & & + 114 & 18:29:49.69 & 1:15:28.85 & & kaas258 & & & + 115 & 18:29:38.59 & 1:18:01.39 & & & & & + 118 & 18:30:01.79 & 1:18:59.84 & & & & & + 119 & 18:29:29.80 & 1:21:57.38 & & & & 5 & 6 + 122 & 18:29:13.01 & 1:07:00.64 & & & & & + 124 & 18:29:08.17 & 1:05:44.68 & & & & & + 127 & 18:29:37.67 & 1:11:30.01 & & & & & + 128 & 18:30:23.12 & 1:20:09.65 & & & & 46 & 85 + 134 & 18:29:37.58 & 1:11:17.55 & & & & & + 136 & 18:29:16.36 & 1:08:22.13 & & & & & + 140 & 18:30:37.33 & 1:16:09.02 & & & & & + 143 & 18:30:53.70 & 1:11:20.64 & & & & & + 156 & 18:30:00.07 & 1:03:05.81 & & & & & + 157 & 18:30:22.39 & 1:20:44.12 & & & & 45 & 84 + 158 & 18:30:00.60 & 1:15:20.21 & & & & & + 161 & 18:30:17.29 & 1:21:32.83 & & & & & + 165 & 18:29:58.43 & 1:12:50.35 & & & & 75 & + 166 & 18:30:13.96 & 1:08:51.47 & & kaas407 & & & + 170 & 18:30:09.43 & 1:02:47.06 & & kaas379 & & & + 174 & 18:30:03.58 & 1:16:41.04 & & & & & + 176 & 18:29:41.52 & 1:10:04.52 & & & & & + 180 & 18:29:47.64 & 1:04:22.28 & & & & & + 181 & 18:30:19.60 & 1:05:34.31 & & & & & + 187 & 18:30:16.80 & 1:19:17.13 & & & & & + 189 & 18:30:03.01 & 1:19:49.71 & & & & 83 & + 190 & 18:29:57.39 & 1:14:50.47 & & & & 88 & 45 + 192 & 18:29:57.65 & 1:10:53.56 & & & & 69 & 47 + 193 & 18:29:57.55 & 1:10:47.01 & & & & 50 & + 195 & 18:30:01.24 & 1:15:03.73 & & & & 82 & 67 + 196 & 18:29:59.11 & 1:11:20.09 & & & & 76 & 58 + 198 & 18:29:59.05 & 1:11:13.82 & & & & 24 & 57 + 199 & 18:29:47.21 & 1:22:34.69 & & & & 40 & 27 + 201 & 18:29:55.34 & 1:10:34.31 & & & & 62 & 36 + 203 & 18:30:05.50 & 1:14:24.87 & & & & 51 & 72 + 204 & 18:29:56.33 & 1:12:18.04 & & & & 65 & 39 + 205 & 18:29:56.17 & 1:10:57.66 & & & & 64 & 38 + 209 & 18:29:50.53 & 1:08:58.22 & & & & 41 & 29 + 215 & 18:29:42.32 & 1:12:01.99 & & & & 58 & 21 + 216 & 18:30:00.64 & 1:13:40.38 & serpens13 & kaas338 & & 80 & 65 + 218 & 18:29:49.47 & 1:12:04.29 & & & & 48 & + 220 & 18:30:24.44 & 1:19:50.66 & & & & 36 & + 221 & 18:29:22.70 & 1:10:33.09 & & & & 37 & 2 + 224 & 18:29:31.64 & 1:15:05.59 & & & & 54 & + 225 & 18:29:33.06 & 1:17:16.68 & & & & 56 & 11 + 1 & @xmath3216.044 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & -0.06 & + 2 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & 72 & 0.56 & + 3 & @xmath3217.913 & @xmath3217.371 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & 25 & 1.27 & + 4 & @xmath3215.185 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & 23 & 0.46 & + 5 & @xmath3217.731 & @xmath3215.836 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & -0.44 & + 6 & @xmath3218.248 & @xmath3217.326 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & 1.35 & + 7 & @xmath3216.889 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & 17 & 0.76 & + 8 & @xmath3216.454 & @xmath3214.104 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & 12 & 0.77 & + 9 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & 67 & 0.06 & + 10 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & 0.01 & + 11 & @xmath3218.253 & @xmath3217.301 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & 2.21 & + 12 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & & 0/1 & & 0.74 & + 17 & & & & [email protected] & [email protected] & [email protected] & [email protected] & & 0/1 & & 1.39 & + 18 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & 4.35 & + 19 & & & & [email protected] & [email protected] & [email protected] & & [email protected] & 0/1 & 78 & 1.03 & + 20 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & 2.16 & + 21 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & 1.12 & + 22 & & & & [email protected] & [email protected] & [email protected] & [email protected] & & 0/1 & 20 & 3.40 & + 23 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & & 1.54 & + 24 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 0/1 & 18 & 1.38 & + 27 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & 74 & -0.74 & + 28 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & 79 & -0.26 & + 30 & @xmath3217.925 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & 81 & -0.77 & + 31 & & & & & & & [email protected] & [email protected] & 0/1 & & & + 32 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & & 0.34 & + 33 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & & -0.91 & + 34 & @xmath3218.32 & @xmath3217.294 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & & -0.67 & + 35 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & 70 & 0.21 & + 36 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & 9 & 0.62 & + 37 & @xmath3217.197 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & fs & 22 & -0.45 & + 38 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & 63 & 0.06 & + 39 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & & -1.07 & + 40 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & 44 & -0.32 & + 44 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & & 1.59 & + 45 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & & 0.09 & + 47 & & & & [email protected] & [email protected] & [email protected] & [email protected] & & fs & & -0.06 & + 48 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & fs & & 0.99 & + 53 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.58 & 0.21 + 54 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -1.18 & 0.28 + 55 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & 2 & & -1.85 & 0.29 + 56 & @xmath3214.229 & [email protected] & [email protected] & [email protected] & [email protected] & & & & 2 & & -0.93 & + 57 & [email protected] & [email protected] & [email protected] & & [email protected] & & & & 2 & & & 0.09 + 58 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & 2 & & -0.67 & 0.29 + 59 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 77 & -0.64 & 0.57 + 60 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.43 & 0.45 + 61 & @xmath3218.127 & @xmath3216.539 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 21 & -0.39 & + 62 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -0.90 & 1.92 + 64 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & & 2 & & -2.86 & 0.58 + 65 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.53 & 0.68 + 66 & [email protected] & [email protected] & @xmath3213.202 & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -1.68 & + 67 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.27 & 0.53 + 68 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -1.03 & 0.52 + 69 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -0.94 & 1.44 + 70 & @xmath3218.165 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -0.77 & 3.51 + 71 & [email protected] & [email protected] & [email protected] & & [email protected] & [email protected] & & [email protected] & 2 & & 0.18 & 1.23 + 72 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & & 2 & 3 & -2.49 & + 73 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 86 & -1.01 & 0.70 + 74 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.69 & 0.29 + 75 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & 2 & & -2.13 & + 76 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.34 & 0.38 + 77 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & 2 & & -1.72 & + 78 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 84 & -0.65 & 0.51 + 79 & 13.210 @xmath2020.026 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.49 & 0.23 + 80 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 66 & -0.82 & 1.48 + 81 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 60 & -1.45 & 0.58 + 82 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -0.54 & + 83 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 71 & -1.26 & 2.76 + 84 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 42 & -1.06 & 0.38 + 85 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & 68 & -1.32 & 2.17 + 86 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -0.86 & 1.33 + 87 & [email protected] & [email protected] & 12.8 @xmath2020.035 & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 61 & -2.35 & 0.02 + 88 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.02 & + 89 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.07 & 0.29 + 90 & @xmath3214.481 & [email protected] & [email protected] & [email protected] & [email protected] & & & & 2 & & -1.41 & + 91 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -1.62 & 0.11 + 92 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -0.80 & 1.16 + 93 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 47 & -0.35 & + 94 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 57 & -1.06 & 0.95 + 95 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 49 & -1.38 & 1.44 + 96 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.48 & 0.77 + 97 & [email protected] & [email protected] & @xmath3214.814 & [email protected] & [email protected] & & & & 2 & & -2.47 & + 98 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 85 & -1.20 & 0.24 + 99 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -1.86 & 0.03 + 100 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 59 & -0.23 & 0.36 + 101 & @xmath3213.258 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 87 & -0.74 & 0.14 + 102 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & 2 & & -2.12 & + 103 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.18 & 0.42 + 104 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -1.82 & 0.83 + 105 & @xmath3216.553 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & 73 & -0.46 & 3.45 + 106 & @xmath3217.230 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & -0.89 & 1.84 + 109 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & & 0.06 & + 110 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2 & 13 & 0.86 & + 112 & & & & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -0.11 & + 114 & & & & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -0.21 & + 115 & & & & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -1.81 & + 118 & & & & [email protected] & [email protected] & [email protected] & [email protected] & & 2 & & -1.65 & + 119 & & & & & [email protected] & [email protected] & [email protected] & & 2 & 5 & -1.09 & + 122 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.42 & 0.89 + 124 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.63 & + 127 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.33 & 1.34 + 128 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & 46 & -2.86 & 0.06 + 134 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.38 & 1.35 + 136 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath3212.029 & [email protected] & 2 & & -1.75 & 0.29 + 140 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.58 & 1.00 + 143 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.64 & 0.92 + 156 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -0.98 & 0.90 + 157 & 13.1 @xmath2020.027 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & [email protected] & td & 45 & -2.51 & 0.24 + 158 & @xmath3217.663 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -1.71 & 6.26 + 161 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.35 & 1.01 + 165 & @xmath3218.127 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & td & 75 & -1.05 & 3.89 + 166 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -1.46 & 0.47 + 170 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.28 & 1.06 + 174 & @xmath3218.198 & @xmath3216.001 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -1.24 & + 176 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -2.36 & 1.32 + 180 & & & & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & td & & -1.52 & + 181 & & & & [email protected] & [email protected] & & [email protected] & [email protected] & 2 & & -1.73 & + 187 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & & 3 & 35 & -2.74 & + 189 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & 3 & 83 & -3.14 & + 190 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & 3 & 88 & -1.87 & 1.08 + 192 & [email protected] & 8.13 @xmath2020.053 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 3 & 69 & -2.91 & + 193 & [email protected] & [email protected] & [email protected] & & [email protected] & [email protected] & [email protected] & & 3 & 50 & -2.96 & + 195 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 82 & -2.31 & 1.81 + 196 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 76 & -2.58 & 1.10 + 198 & @xmath3216.917 & @xmath3216.191 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 24 & -1.57 & + 199 & [email protected] & 10.52 @xmath2020.031 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 40 & -2.70 & 0.71 + 201 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 62 & -2.83 & + 203 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 51 & -2.49 & 0.04 + 204 & [email protected] & [email protected] & 13.59 @xmath2020.044 & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 65 & -2.68 & 0.29 + 205 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 64 & -2.58 & 0.16 + 209 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 41 & -2.36 & 0.79 + 215 & @xmath3216.836 & 14.37 @xmath2020.059 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 58 & -2.45 & 2.35 + 216 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 80 & -2.50 & 1.68 + 218 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 48 & -2.78 & 0.80 + 220 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & & 3 & 36 & -2.77 & + 221 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 37 & -2.81 & 0.47 + 224 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 54 & -2.93 & 0.65 + 225 & [email protected] & 10.33 @xmath2020.031 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & & 3 & 56 & -2.72 & 0.47 + cccccccccccccccc smm1 & 18:29:49.69 & 1:15:20.55 & 18:29:49.66 & 1:15:21.56 & & & & 10.7 & 1538 & 28034 & 35700 & 6100 & 233 . & 7.54 & + smm2 & 18:30:00.33 & 1:12:55.33 & & & & & & & & & 3400 & 700 & & & + smm3 & 18:29:59.41 & 1:14:01.25 & 18:29:59.27 & 1:14:02.90 & & & & & 472 & & 7100 & 1500 & & & + smm4 & 18:29:56.73 & 1:13:17.07 & 18:29:56.69 & 1:13:20.71 & 0.29 & 2.53 & 2.25 & & & & 10800 & 2900 & & 0.12 & + smm5 & 18:29:51.16 & 1:16:40.65 & 18:29:51.14 & 1:16:40.47 & 27.8 & 59.4 & 125 & 183 & 1376 & 3705 & 3900 & 900 & 20.7 & 0.15 & 18 + smm6 & 18:29:57.71 & 1:14:08.12 & 18:29:57.73 & 1:14:06.03 & 2269 & 3163 & 4899 & 5897 & 9676 & 5452 & 1900 & 500 & & 0.17 & 70 + smm8 & 18:30:01.79 & 1:15:09.43 & 18:30:01.53 & 1:15:14.90 & & & & 1.55 & 21.6 & & 1300 & 400 & & & + smm9 & 18:29:48.16 & 1:16:44.43 & 18:29:48.10 & 1:16:44.97 & 1.82 & 6.41 & 11.1 & 15.1 & 289 & 5664 & 9100 & 1600 & 36.1 & 0.28 & + s68nb & 18:29:49.55 & 1:17:10.53 & 18:29:49.58 & 1:17:05.89 & 39.8 & 87.4 & 137 & 169 & 585 & & & & 17.7 & & 17 + s68nc & 18:29:48.71 & 1:16:57.00 & 18:29:48.88 & 1:16:55.23 & & & & 2.41 & 0.27 & & & 20.4 & & + s68nd & 18:29:48.96 & 1:16:23.01 & 18:29:49.12 & 1:16:19.95 & 10.8 & 61.9 & 200 & 543 & 7340 & 7489 & & & 32.3 & & + smm10 & 18:29:52.08 & 1:15:48.74 & 18:29:52.20 & 1:15:47.59 & 6.83 & 25.5 & 37.2 & 42.1 & 1487 & 4953 & 2700 & 600 & 25.3 & 0.14 & + smm11 & 18:29:59.67 & 1:11:32.27 & [ 18:29:59.57 ] & [ 1:11:58.77 ] & & & & & & & 2700 & 900 & & & [ 3 ] + ps2 & 18:30:03.15 & 1:12:27.53 & 18:30:02.73 & 1:12:28.11 & 230 & 345 & 439 & 534 & 2275 & 2319 & & & & & + | we present _ spitzer _ and _ chandra _ observations of the nearby ( @xmath0260 pc ) embedded stellar cluster in the serpens cloud core . we observed , using _
spitzer s _ irac and mips instruments , in six wavelength bands from 3 to 70 @xmath1 , to detect thermal emission from circumstellar disks and protostellar envelopes , and to classify stars using color - color diagrams and spectral energy distributions ( seds ) .
these data are combined with _
chandra _ observations to examine the effects of circumstellar disks on stellar x - ray properties .
young diskless stars were also identified from their increased x - ray emission .
we have identified 138 ysos in serpens : 22 class 0/i , 16 flat spectrum , 62 class ii , 17 transition disk , and 21 class iii stars ; 60 of which exhibit x - ray emission .
our primary results are the following : 1 . )
ten protostars detected previously in the sub - millimeter are detected at @xmath2 @xmath3 m , seven at @xmath4 @xmath3 m , 2 . )
the protostars are more closely grouped than more evolved ysos ( median separation : @xmath5 pc ) , and 3 . )
the luminosity and temperature of the x - ray emitting plasma around these ysos does not show any significant dependence on evolutionary class .
we combine the infrared derived values of @xmath6 and x - ray values of @xmath7 for 8 class iii objects and find that the column density of hydrogen gas per mag of extinctions is less than half the standard interstellar value , for @xmath8
. this may be the result of grain growth through coagulation and/or the accretion of volatiles in the serpens cloud core . |
the problem of _ variable - length _ lossy source coding is one of the fundamental research topics in shannon theory . for this problem ,
several studies have adopted the _ excess distortion probability _ as a distortion criterion ( e.g. , @xcite , @xcite , @xcite , @xcite ) .
the excess distortion probability is defined as the probability that the distortion between a source sequence and its reproduction is greater than a certain threshold . for the problem of variable - length lossy source coding under the excess distortion probability , there are mainly two criteria on codeword length : the _ mean codeword length _ and the _
overflow probability_. kostina et al .
@xcite have considered the mean codeword length and shown the non - asymptotic characterization on the optimal mean codeword length .
they also have performed the asymptotic analysis on the optimal mean codeword length for i.i.d .
sources . on the other hand , yagi and nomura @xcite and nomura and yagi @xcite have considered the overflow probability . in @xcite ,
they have treated the case where either the overflow probability or the excess distortion probability is less than or equal to a positive constant asymptotically for general sources . in @xcite , they have dealt with the case where the probability of union of events that the overflow occurs and the excess distortion occurs is less than or equal to a positive constant asymptotically for general sources .
this paper considers the excess distortion probability and the overflow probability as in @xcite and @xcite .
however , the primary differences are 1 ) we address the case where both the excess distortion probability and the overflow probability may be positive and 2 ) we analyze both non - asymptotic and asymptotic cases . the contribution of this paper is the non - asymptotic ( one - shot ) and asymptotic characterizations on the minimum threshold of the overflow probability by using a new smooth max entropy - based quantity . in the non - asymptotic regime ,
coding theorems are shown for both stochastic and deterministic encoders . to show the achievability results
, we give an explicit code construction instead of using the random coding argument .
it turns out that the constructed code satisfies the properties of the optimal code shown in @xcite .
further , using the results obtained in the non - asymptotic regime , we establish asymptotic coding theorem for general sources .
let @xmath0 be a source alphabet and @xmath1 be a reproduction alphabet , where both are finite sets .
let @xmath2 be a random variable taking a value in @xmath0 and @xmath3 be a realization of @xmath2 .
the probability distribution of @xmath2 is denoted as @xmath4 .
a distortion measure @xmath5 is defined as @xmath6 the pair of an encoder and a decoder @xmath7 is defined as follows .
an encoder @xmath8 is defined as @xmath9 where @xmath10 denotes the set of all binary strings and the empty string @xmath11 , i.e. , @xmath12 an encoder @xmath8 is possibly _
stochastic _ and produces a non - prefix code . for @xmath13 ,
the codeword length of @xmath14 is denoted as @xmath15 .
decoder @xmath16 is defined as @xmath17 variable - length source coding _ without _ the prefix condition is discussed as in , for example , @xcite and @xcite .
the performance criteria considered in this paper are the excess distortion and the overflow probabilities . given @xmath18 , the excess distortion probability for a code @xmath7
is defined as @xmath19 given @xmath20 , the overflow probability for a code @xmath7 is defined as @xmath21 using these criteria , we define a @xmath22 code .
given @xmath23 and @xmath24 , a code @xmath7 satisfying @xmath25 is called a @xmath22 code .
the fundamental limits are the minimum thresholds @xmath26 and @xmath27 for given @xmath28 , @xmath29 , and @xmath30 .
given @xmath31 and @xmath32 , we define @xmath33 consider the special case @xmath34 . from a @xmath35 code
, we can construct a fixed - length code achieving rate @xmath36 and the excess distortion probability @xmath37 .
thus , @xmath38 or @xmath39 represents the minimum rate in fixed - length lossy source coding under the excess distortion criterion .
the _ smooth max entropy _ , which is also called the smooth rnyi entropy of order zero ,
has first introduced by renner and wolf @xcite .
later , uyematsu @xcite has shown that the smooth max entropy can be defined in the following form .
given @xmath40 , the smooth max entropy @xmath41 is defined as @xmath42 where @xmath43 represents the cardinality of the set .
one of the useful properties of the smooth max entropy , which is used in the proof of the achievability result in our main theorem , is schur concavity . to state the definition of a schur concave function ,
we first review the notion of _
let @xmath44 be the set of non - negative real numbers and @xmath45 be the @xmath46-th cartesian product of @xmath44 , where @xmath46 is a positive integer .
suppose that @xmath47 @xmath48 and @xmath49 satisfy @xmath50 if @xmath51 and @xmath52 satisfy , for @xmath53 , @xmath54 then we say that @xmath55 _ majorizes _ @xmath56 ( it is denoted as @xmath57 ) .
schur concave functions are defined as follows .
we say that a function @xmath58 is a _ schur concave _ function if @xmath59 for any @xmath60 satisfying @xmath57 . from the definition of the smooth max entropy and schur concave functions , it is easy to see that the smooth max entropy is a schur concave function is a schur concave function for @xmath61 and a schur convex function for @xmath62 . ] .
next , using the smooth max entropy , we introduce a new quantity , which plays an important role in producing our main results . given @xmath18 and @xmath63 ,
@xmath64 is defined as @xmath65 where @xmath66 denotes a conditional probability distribution of @xmath67 given @xmath2 . for
a given @xmath18 and @xmath68 , suppose that @xmath69 then , there are no codes whose excess distortion probability is less than or equal to @xmath29 .
conversely , if such codes do not exist for given @xmath28 and @xmath29 , ( [ inf ] ) holds . in this case
, we define @xmath70 and @xmath71 .
further , if ( [ inf ] ) holds , we also define @xmath72 because there is no conditional probability distribution @xmath66 on @xmath1 satisfying @xmath73 .
the next lemma shows the achievability result on @xmath36 of a @xmath22 code .
assume that @xmath74 for any @xmath18 and @xmath63 , there exists a @xmath22 code such that @xmath75 see section iv - a . to prove the achievability result , we do not use the random coding argument but give an explicit code construction .
the constructed code satisfies the properties of the optimal code shown in @xcite .
the next lemma shows the converse bound on @xmath36 of a @xmath22 code . for any @xmath18 and @xmath76
, any @xmath22 code satisfies @xmath77 see section iv - b . combining lemmas 1 and 2
, we can immediately obtain the following result on @xmath26 . for any @xmath18 and @xmath63
, it holds that @xmath78 by theorem 1 , the minimum threshold @xmath79 can be specified within one bit in the interval not greater than @xmath64 , regardless of the values @xmath28 , @xmath29 , and @xmath30 .
this result is mainly due to an explicit construction of good codes , rather than the random coding argument , given in section iv - a .
the next lemma shows the achievability result on @xmath36 of a deterministic @xmath22 code .
assume that @xmath74 for any @xmath18 and @xmath63 , there exists a deterministic @xmath22 code such that @xmath80 see section iv - c . from lemma 3 and the fact that @xmath81 , the following result on @xmath27 is obtained .
for any @xmath18 and @xmath63 , it holds that @xmath82 by theorem 2 , @xmath27 can be specified in the interval within four bits , which is slightly weaker than the result for stochastic codes .
let @xmath83 and @xmath84 be the @xmath85-th cartesian product of @xmath0 and @xmath1 , respectively .
let @xmath86 be a random variable taking a value in @xmath83 and @xmath87 be a realization of @xmath86 .
the probability distribution of @xmath86 is denoted as @xmath88 . in this section ,
coding problem for general sources @xmath89 @xcite is considered . a distortion measure @xmath90
is defined as @xmath91 an encoder @xmath92 is possibly stochastic and produces a non - prefix code .
a decoder @xmath93 is deterministic .
we define an @xmath94 code as follows .
given @xmath23 and @xmath95 , a code @xmath96 satisfying @xmath97 for all @xmath98 with some @xmath99 is called an @xmath94 code . the fundamental limit is the following minimum threshold . given @xmath31 and @xmath95 , @xmath100 the next theorem characterizes @xmath101 by the smooth max entropy - based quantity @xmath102 . for any @xmath18 and @xmath63 , @xmath103 where @xmath102 is defined as @xmath104 see section iv - d .
as shown in theorem 3 , we characterize the minimum threshold on the overflow probability by the quantity related to the _ entropy _ , whereas previous studies such as @xcite and @xcite have characterized it by the quantity related to the _ mutual information_. in the non - asymptotic ( one - shot ) regime , the results on the minimum threshold are different for stochastic encoders and deterministic encoders as shown in theorems 1 and 2 . in the asymptotic regime , however , the restriction to only deterministic encoders does not change the minimum threshold . instead of ( [ cd1 ] ) and ( [ cd2 ] ) , as a generalization of the problem in @xcite , we can consider the conditions @xmath105 and define @xmath106 as the minimum threshold under the conditions ( [ cd1d ] ) and ( [ cd2d ] ) . then , by almost the same proof of theorem 3 , we have @xmath107 for any @xmath18 and @xmath63 .
first , some notations are defined before the construction of the encoder and decoder is described . * for @xmath108 and @xmath18 ,
@xmath109 is defined as @xmath110 * we define @xmath111 ( @xmath112 ) by the following procedure and @xmath1 are finite sets .
however , we can assume countably infinite @xmath0 and @xmath1 if this operation is admitted for countably infinite @xmath0 and @xmath1 . ] .
let @xmath113 be defined as @xmath114 and for @xmath115 , let @xmath111 be defined as @xmath116 * for @xmath117 , @xmath118 from the definition , we have @xmath119 * if @xmath120 , let @xmath121 be the integer satisfying @xmath122 if @xmath123 , we define @xmath124 .
* let @xmath125 be the integer satisfying @xmath126 from this definition , it holds that @xmath127 .
* let @xmath128 and @xmath129 be defined as @xmath130 and @xmath131 * let @xmath132 be the @xmath133-th binary string in @xmath10 in the increasing order of the length and ties are arbitrarily broken .
for example , @xmath134 etc .
we construct the following encoder @xmath135 and decoder @xmath17 * [ encoder ] * * for @xmath136 ( @xmath137 ) , set @xmath138 . * for @xmath139 , set from ( [ ks1 ] ) . ]
@xmath140 * for @xmath141 , set @xmath142 . *
[ decoder ] * set @xmath143 ( @xmath144 ) .
now , we evaluate the excess distortion probability .
we have @xmath145 for @xmath136 ( @xmath146 ) since @xmath147 .
furthermore , we have @xmath145 with probability @xmath148 for @xmath139 .
thus , @xmath149 therefore , we have @xmath150 next , we evaluate the overflow probability . from the construction of the encoder
, it is easily verified that @xmath151 ( @xmath152 ) .
hence , setting @xmath153 , we have @xmath154 where the last inequality is due to the definition of @xmath128 and ( [ ih1 ] ) and the last equality is due to the definition of @xmath129 . therefore , the code @xmath7 is a @xmath22 code with @xmath153 . to complete the proof of the theorem
, we shall show @xmath155 .
we define @xmath156 .
notice that @xmath157 where @xmath158 and @xmath159 follow from the definition of the encoder and decoder and @xmath160 is due to ( [ hc ] ) .
then , the equality in ( [ notice ] ) does not hold . however , @xmath161 is true since @xmath162 .
] , @xmath163 which implies that @xmath164 . hence
, if @xmath165 is shown , the desired equation @xmath155 is obtained .
to show ( [ hg ] ) , the following lemma is useful .
if @xmath166 which is induced by @xmath167 satisfying @xmath168 majorizes any @xmath169 which is induced by @xmath170 satisfying @xmath171 , then it holds that @xmath172 the lemma follows from the fact that the smooth max entropy is a schur concave function and the definition of @xmath64 . in view of lemma 4
, we shall show that @xmath173 majorizes any @xmath169 induced by @xmath170 satisfying @xmath171 .
to show this fact , suppose the following condition : ( @xmath174 ) _ there exists a @xmath169 satisfying @xmath171 but not being majorized by @xmath173 .
_ assuming ( @xmath174 ) , we shall show a contradiction .
let @xmath175 give the largest @xmath176 in @xmath1 , @xmath177 give the largest @xmath176 in @xmath178 , @xmath179 give the largest @xmath176 in @xmath180 , etc .
that is , @xmath181 and @xmath182 for all @xmath183 considering the fact that the support of @xmath173 is @xmath184 and the assumption ( @xmath174 ) , we can say that there exists a @xmath185 satisfying @xmath186 on the other hand , the excess distortion probability under @xmath187 is evaluated as @xmath188 where @xmath189 is the indicator function and the last inequality is due to @xmath190 @xmath191 for all @xmath13 . for the second term in ( [ ed ] )
, it holds that @xmath192 where @xmath158 follows from the definition of @xmath193 , @xmath159 follows from ( [ a1 ] ) and ( [ a2 ] ) , and @xmath160 follows from ( [ y11 ] ) and ( [ y12 ] ) . plugging ( [ st ] ) into ( [ ed ] ) gives @xmath194 where the last inequality is due to ( [ mu ] ) .
this is a contradiction to the fact that @xmath171 . for any @xmath22 code @xmath7 , set @xmath195
the definition of a @xmath22 code gives @xmath196 let @xmath197 then , the first inequality in ( [ npc ] ) is rewritten as @xmath198 hence , @xmath199 where the superscript `` @xmath200 '' represents the complement . from ( [ npc3 ] ) and the definition of the smooth max entropy , @xmath201 on the other hand , since @xmath202 for @xmath203 ,
@xmath204 combining ( [ npc4 ] ) and ( [ npc5 ] ) yields @xmath205 thus , from the second inequality in ( [ npc ] ) , we have @xmath206 first , some notations are defined .
* let @xmath125 be the integer satisfying ( [ ks1 ] ) and ( [ ks2 ] ) .
* define @xmath207 as @xmath208 .
then , it holds that @xmath209 .
* let @xmath210 be the integer satisfying @xmath211 we construct the following _ deterministic _ encoder @xmath135 and decoder @xmath17 * [ encoder ] * * for @xmath136 ( @xmath144 ) , set @xmath138 .
* for @xmath141 , set @xmath142 . *
[ decoder ] * set @xmath143 ( @xmath144 ) .
now , we evaluate the excess distortion probability . from the definition of the encoder and decoder , @xmath212 therefore
, we have @xmath213 next , we evaluate the overflow probability . from the definition of the encoder , we have @xmath214 setting @xmath215 , it holds that , then @xmath216 . ]
@xmath217 where the last inequality is due to ( [ js1 ] ) .
therefore , the code @xmath7 is a deterministic @xmath22 code with @xmath215 .
let @xmath218 be the integer satisfying ( [ ih1 ] ) and ( [ ih2 ] ) .
then , from the proof of lemma 1 , it holds that @xmath155 . since @xmath209
, it is easily verified that @xmath219 and @xmath220 , meaning that @xmath221 . if @xmath222 , @xmath223 obviously holds .
then , assuming that @xmath224 , we shall show that @xmath225 this leads to @xmath226 where the rightmost inequality is due to taylor s expansion , and we obtain ( [ th4 ] ) .
the first step to show @xmath223 is the following inequality : @xmath227 where @xmath158 follows from ( [ ih1 ] ) and ( [ js2 ] ) and @xmath159 follows from @xmath228 inequality ( [ jsks ] ) is equivalent to @xmath229 thus , we obtain @xmath230 , implying that @xmath231 .
( direct part ) from lemma 3 , there exists a code @xmath96 satisfying @xmath232 fix @xmath233 arbitrarily .
then , it holds that @xmath234 and @xmath235 for all @xmath98 with some @xmath99 .
then , from ( [ t31 ] ) , we have @xmath236 for all @xmath98 .
thus , from ( [ t30 ] ) and ( [ t33 ] ) , @xmath96 is an @xmath94 code with @xmath237 for all @xmath98 .
since @xmath233 is arbitrary , this indicates that @xmath238 ( converse part ) for any @xmath94 code , lemma 2 gives @xmath239 therefore , it holds that @xmath240 for any @xmath94 code .
hence , we have @xmath241
this work was supported in part by jsps kakenhi grant numbers 26289119 , 16k00195 , 16k00417 , and 16k06340 .
s. saito and t. matsushima , `` threshold of overflow probability using smooth max - entropy in lossless fixed - to - variable length source coding for general sources , '' _ ieice trans .
fundamentals _ , vol .
e99-a , no .
12 , pp . 22862290 , dec .
2016 .
t. weissman and n. merhav , `` tradeoffs between the excess - code - length exponent and the excess - distortion exponent in lossy source coding , '' _ ieee trans .
inf . theory _ ,
2 , pp . 396415 , feb . 2002 . | this paper investigates the problem of variable - length lossy source coding .
we deal with the case where both the excess distortion probability and the overflow probability of codeword length are less than or equal to positive constants .
the infimum of the thresholds on the overflow probability is characterized by a smooth max entropy - based quantity .
both non - asymptotic and asymptotic cases are analyzed . to show the achievability results
, we do not utilize the random coding argument but give an explicit code construction . |
in the concluding remarks of their pioneering paper on nonradial oscillations of neutron stars thorne and campollataro described it as `` _ just a modest introduction to a story which promises to be long , complicated and fascinating _ '' .
the story has undoubtedly proved to be intriguing , and many authors have contributed to our present understanding of the pulsations of neutron stars . of special interest
are the attempts to calculate the eigenfrequencies of the so - called quasinormal modes of a stellar system .
these are solutions to the perturbation equations which agree with the necessary boundary conditions at the centre and at the surface of the star , and at the same time behave as purely outgoing waves at spatial infinity . in general relativity pulsations of the stellar fluid will normally be damped due to the emission of gravitational radiation .
hence , the characteristic frequencies of the system will have complex values .
the real parts correspond to physical oscillation frequencies , while the imaginary parts describe the damping rate of each oscillation mode .
the first attempts to compute outgoing - wave modes for relativistic stars concerned modes that have an analogue in newtonian theory .
it was believed ( and verified by the early calculations for nonrotating stars ) that the newtonian modes of oscillation are shifted only slightly because of the coupling to gravitational radiation : each characteristic frequency adopts a very small imaginary part . in practice , it is not straightforward to determine these modes with great accuracy .
this difficulty became obvious with the first numerical calculations @xcite .
an approach based on a variational principle devised by detweiler and ipser also suffers from this problem @xcite .
the difficulty of determining small imaginary parts with some accuracy was not overcome until much later when lindblom and detweiler combined numerical integration of the equations for the interior with a numerically integrated solution for the exterior .
using this technique they computed the fundamental ( f ) mode frequency for many realistic neutron star equations of state .
however , they later realized that the fourth - order system of equations they had used for the stellar interior could become singular .
consequently , they suggested that a different choice of dependent variables be made @xcite .
also worth mentioning in this context is a paper by balbinski _
_ where results obtained by the lindblom - detweiler scheme were found to be in satisfactory agreement with the predictions of the standard quadrupole formula .
the last few years have seen several important contributions to this field . in an impressive series of papers , chandrasekhar and ferrari
have reformulated the problem . in their final system of equations
all fluid perturbations have been eliminated @xcite .
this has the advantage that one can view the problem of gravitational waves being scattered by a star in a way similar to that used for black holes @xcite . in the second paper of the series ,
chandrasekhar and ferrari show that odd - parity ( axial ) modes can be excited by the even - parity ( polar ) modes in the slow rotation limit @xcite .
the coupling is due to the lense - thirring effect . in the following papers ,
they demonstrate that their algorithm for computing eigenfrequencies can be trusted so long as the imaginary part is considerably smaller than the real part @xcite , that the odd - parity ( axial ) modes can be of importance for extremely compact stars @xcite and that the idea of a complex angular momentum can be useful in astrophysical situations @xcite .
a summary of their work has been written by ferrari .
another recent reformulation of the equations governing a perturbed stellar model was proposed by ipser and price .
they discuss the relation between the regge - wheeler gauge ( and the resulting equations ) and the diagonal gauge used by chandrasekhar and ferrari @xcite .
in fact , they reformulate the stellar perturbation equations in such a way that all fluid perturbations can be eliminated also in the regge - wheeler gauge . an interesting question is whether these new formulations of the pulsation problem will make computation of mode - frequencies any easier
. we will not approach that issue here , but feel that an investigation of that kind is of great importance and should be encouraged .
another chapter of the story begins with the study of a very simple toy model .
kokkotas and schutz suggested that a plausible model for a stellar system would be a finite string representing the fluid of the star coupled by means of a spring to a semi - infinite string the dynamical spacetime .
they found that such systems could accommodate a new family of oscillation modes .
these would be associated with the gravitational degrees of freedom , rather than the pulsations of the fluid , and would be strongly damped .
recently , a slightly more realistic toy model led baumgarte and schmidt to much the same conclusions .
yet another toy model , suggested by kokkotas , shows a lot of similarities with the axial ( odd parity ) modes discussed by chandrasekhar and ferrari and has recently led to the discovery of a branch of strongly damped axial modes @xcite .
a computation of highly damped modes for a realistic stellar model is not trivial , however .
the main difficulty involves numerically separating the ingoing and outgoing - wave solutions at spatial infinity : the ingoing solution dies exponentially as @xmath0 while the outgoing one grows .
this problem is well - known from studies of quasinormal modes for black holes ( see andersson _
for a detailed discussion ) .
it seems reasonable to assume that methods that have proved useful in studies of the black - hole problem can be adapted to the stellar situation . a series expansion approach based on a four - term recurrence relation
similar to that developed by leaver for black holes @xcite was used by kojima to verify that the strongly damped modes ( referred to as w - modes because of their connection to the gravitational aves ) indeed do exist for realistic stellar models .
kokkotas and schutz used a wkb approach ( essentially a geometrical optics assumption of no reflection of waves in the exterior vacuum ) in their calculations of w - mode spectra for several models .
their main results have recently been verified by leins , nollert and soffel . in their calculations , leins _
et al . _ employed two different approaches for the exterior : leaver s continued fraction approach @xcite and a wronskian technique that has proved extremely powerful for schwarzschild black holes @xcite .
although the results obtained by leins _ et al . _ generally agree with those of kokkotas and schutz there are some differences .
et al . _ found new modes with considerably smaller oscillation frequencies and also higher damping than those that had been found by the wkb - technique .
their calculations also suggest that a `` kink '' that was apparent in the spectra of kokkotas and schutz does not exist . at first sight
, one would be tempted to believe that the calculations of leins _ et al .
_ are the most reliable since the two approaches they used to deal with the exterior do not , in principle , involve any approximations .
this , however , does not explain why the wkb approach becomes less reliable as the oscillation frequency increases .
in fact , this is contrary to all expectations : the geometrical optics argument should be valid for high frequencies . in this paper
we attempt to settle these matters .
we will combine a numerical phase - amplitude approach ( similar to that used by andersson for black holes ) to the exterior problem with the lindblom - detweiler @xcite scheme for the inside .
the key idea is to separate ingoing and outgoing solutions by numerically calculating their analytic continuations to a place in the complex - coordinate plane where they have comparabe amplitudes .
this is a new approach that could prove to be of great importance , especially for problems involving rotating stars . in such problems the exterior spacetime
is not known analytically and previous methods ( such as the wkb method of kokkotas and schutz and the two methods employed by leins _
_ ) will consequently fail .
but it seems likely that a method based on complexifying the coordinates will work even if the spacetime itself can only be approximated .
in fact , the method proposed here may well prove to be the only one that remains useful for the strongly damped modes of rotating stars .
the present approach also has the advantage that , although based on numerical integration and in that sense it provides an arbitrarily high numerical precision
it solves for quantities that are directly comparable to those used in the wkb scheme of kokkotas and schutz .
hence , in this first application of the new method , we hope to understand the reason for the discrepancies between the results of kokkotas and schutz and leins _ et al . _ and thus contribute further to the understanding of the oscillation modes of nonrotating relativistic stars .
in regge - wheeler gauge the perturbed metric for a relativistic stellar model takes the form @xcite @xmath1 where @xmath2 , @xmath3 and @xmath4 are functions of @xmath5 only , since we have assumed a harmonic dependence on time .
we also have the standard definition of @xmath6 : @xmath7 which plays the role of an effective mass inside radius @xmath5 .
as is well - known , a stellar model in equilibrium is described by the tolman - oppenheimer - volkov equations [ see for example equations ( a3)-(a5 ) of lindblom and detweiler ] .
these are a system of three coupled first order differential equations that determine the mass @xmath8 , the metric quantity @xmath9 and the pressure @xmath10 .
their solution requires that an equation of state @xmath11 for the energy density is specified . in the present investigation
we will assume a simple polytropic equation of state @xmath12 where @xmath13 and we are using units in which @xmath14 throughout this paper .
this choice of equation of state may not be very realistic , but it simplifies the calculations somewhat .
it also enables us to compare our results directly to those of balbinski _ et al . _ ,
kojima , kokkotas and schutz and leins _ et al . _ . inside the star , the perturbed fluid
is described by a lagrangian displacement @xmath15 , where @xmath16 @xmath17 @xmath18 on general grounds one would think that perturbations of a spherical star have four degrees of freedom .
two of these are associated with the fluid and the remaining two correspond to the gravitational waves .
this means that the five functions @xmath2 , @xmath3 , @xmath4 , @xmath19 and @xmath20 should not be independent . as was shown by detweiler and lindblom a system of equations that is free from singularities can be obtained by defining @xmath21 where a prime denotes a derivative with respect to @xmath5 .
( a misprint in equation ( 6 ) of detweiler and lindblom has been corrected here . )
then it follows , as a consequence of einstein s equations , that @xmath22 h_1 + \nonumber \\ & & \left [ nr -\omega^2r^3e^{-\nu } - { e^\lambda \over r } q(2 m - r + q ) \right ] k \ , \end{aligned}\ ] ] where @xmath23 and @xmath24 the interior problem now reduces to a system of four first order differential equations for @xmath3 , @xmath4 , @xmath19 and @xmath25 [ equations ( 8)-(11 ) in detweiler & lindblom ] .
only two of the four linearly independent solutions to this system are well - behaved at the centre of the star ( at @xmath26 ) .
furthermore , the perturbed pressure must vanish at the surface ( @xmath27 , which implies that @xmath28 .
these conditions specify a single acceptable solution for each frequency @xmath29 .
physically , this solution describes the response of the star when gravitational waves of the given frequency are incident upon it . in the exterior of the star
the fluid perturbations vanish , and the two metric perturbations @xmath3 and @xmath4 can be combined in such a way [ see for example fackerell ] that one obtains a single second - order differential equation known as the zerilli equation .
this equation can be written @xmath30 z = 0 \ , \label{zerilli}\ ] ] with the effective potential @xmath31 given by @xmath32 the tortoise coordinate @xmath33 is defined by @xmath34 clearly , this equation allows two linearly independent solutions . far away from the star one of these
can be identified as an outgoing wave , whereas the other describes an ingoing wave . for a general frequency , the physically acceptable solution for the interior of the star leads to a mixture of out- and ingoing waves at spatial infinity .
the quasinormal modes of the system are those special frequencies for which no waves come in from infinity . in a sense , these are the frequencies at which the star can be expected to radiate `` spontaneously '' .
in any numerical approach it is imperative that the quantities under consideration are smooth and nicely behaving . if that is not the case , an integration scheme may require ridiculously small steps in order to achieve the desired accuracy .
one standard way to avoid this difficulty when one has to solve for ( say ) an oscillating function such as the wave - like solutions in the stellar problem is to appeal to general properties of the solutions and express them in terms of slowly varying functions . in the following
we adopt this approach to deal with the exterior problem for stars .
although the tortoise coordinate @xmath33 appears naturally [ cf .
( [ zerilli ] ) ] in a formal description of the perturbation equations it may obscure a numerical solution of the problem , especially as we plan to use complex coordinates . in order to avoid any difficulties we introduce a new dependent variable @xmath35 according to @xmath36 then we get a new differential equation @xcite @xmath37 \psi = 0 , \label{ode}\ ] ] where @xmath38 \ .\ ] ] one of the important properties of an equation such as ( [ ode ] ) is that the wronskian of any two linearly independent solutions must be a constant .
this means that the general solution can always be constructed as a combination of two basic solutions that has the form @xmath39 \ , \label{pamsols}\ ] ] where the function @xmath40 is a solution to the nonlinear equation @xmath41 at first sight , it may seem as if we have replaced the relatively simple second - order differential equation ( [ ode ] ) with a more complicated equation . in principle that may be true , but it should be quite obvious that the function @xmath42 as determined from ( [ qequation ] ) may be slowly varying even when the solution to ( [ ode ] ) oscillates wildly . moreover , it is easy to make a connection to the wkb approximation here .
whenever @xmath43 is a slowly varying function of @xmath5 one can neglect the two derivatives in ( [ qequation ] ) .
that is , one can use @xmath44 . since that is the case for large @xmath5 , we shall use this approximation to generate initial values for the integration of ( [ qequation ] ) far away from the star .
this is done in such a way that @xmath45 as @xmath46 , _
i.e. _ , the function @xmath47 represents an ingoing wave whereas @xmath48 is outgoing .
the purpose of this section is to study to what extent a wkb approximation of the zerilli function at the surface of the star can be trusted .
specifically , we want to investigate whether a geometrical optics argument that the waves undergo no reflection in the exterior spacetime [ as was assumed by kokkotas and schutz ] can be used in this problem .
if that is the case , one can simply express the exterior solution at the surface in terms of @xmath49 .
in general , one would expect this kind of argument to be valid for high frequencies . in the case of really low frequencies , however , the situation becomes more obscure
. then waves can possibly be backscattered by the curvature of spacetime .
these qualitative expectations agree with the results of leins _ et al . _ who found a few highly damped modes with a very small real part that kokkotas and schutz were unable to find . on the other hand , leins _
_ did not confirm the kink that occurred for relatively high frequencies in the wkb spectra .
they suggested that this indicates that the wkb approach becomes less reliable as the oscillation frequency increases .
this seems unreasonable : if the wkb argument is at all valid , the approximation should become more accurate as the oscillation frequency increases .
it is clear that the exact phase - amplitude approach outlined in the previous section provides us with the means to test the wkb assumption quantitatively . by comparing the numerically determined function @xmath42 to the approximation
@xmath50 we can assess the reliability of the method used by kokkotas and schutz .
( it should be noted that this approximation is not exactly that used by kokkotas and schutz where the zerilli equation was approached directly .
nevertheless , a study such as the present one will provide results relevant also for that approximation . ) in approaching the problem for the stellar exterior numerically we must first devise a scheme that avoids a problem that arises for rapidly damped modes . the desired outgoing - wave solution to ( [ ode ] ) will grow exponentially as @xmath51 .
this means that it is difficult to resolve the ingoing solution ( that is exponentially small ) given a finite numerical precision .
this problem is well - known from studies of black - hole normal modes , and can be avoided by letting @xmath5 assume complex values [ see andersson ] . since @xmath52 as @xmath51 it is clear that the exponential divergence can be suppressed along a path in the complex @xmath5-plane . for large @xmath53
the preferred path is a straight line with slope given by @xmath54 .
such paths are parallell to the so - called anti - stokes lines that play a crucial role in a complex wkb analysis @xcite .
the anti - stokes lines are curves along which @xmath55 is a real quantity . in order to deal with the present problem we will integrate ( [ qequation ] ) along a straight line ( with the prescribed slope ) from a point in the asymptotic regime towards the stellar surface .
one can show , either using an analytic continuation argument or a wkb analysis such as that of arajo _ et al . _
@xcite , that asymptotic conditions introduced for complex values of @xmath5 in the way described above are , in fact , identical to the desired outgoing - wave boundary condition on the real @xmath5-axis . in a complex - coordinate wkb approach to ( [ ode ] )
the so - called stokes lines also play an important role [ see the discussion by andersson _
et al . _ ] .
these correspond to @xmath55 being purely imaginary . from each of the ,
possibly complex , zeros of @xmath50 the transition points of the problem emanate three such contours . when an approximate solution to ( [ ode ] ) is continued across a stokes line the character of the solution changes .
this is called the stokes phenomenon .
specifically it means that , if a certain linear combination of @xmath47 and @xmath48 ( with @xmath42 replaced by @xmath50 ) represents the desired solution to ( [ ode ] ) on one side of the stokes line , another linear combination should probably be used on the opposite side .
moreover , it is straightforward to show [ see the appendix in andersson ] that this effect is important also for numerical integration . if the integration of ( [ qequation ] ) is continued across a stokes line the function @xmath42 may become oscillatory .
we thus have a simple diagnostic of the `` no - reflection '' assumption used by kokkotas and schutz : if a stokes line ( emanating from one of the complex zeros of @xmath50 ) crosses the real @xmath5-axis between @xmath56 and @xmath57 we should have reflection .
for the stellar models considered here we find that this does indeed happen for the new modes discovered by leins _
et al . _ . for small oscillation frequencies
a stokes line crosses the real @xmath5-axis in the relevant interval and the stokes phenomenon should be accounted for in an approximate analysis .
it is , of course , possible that this would give rise to a very small correction . to test whether this is the case ,
we have compared @xmath42 as numerically determined from ( [ qequation ] ) to @xmath50 for several frequencies .
a typical example for one of the new modes that were identified by leins _
_ is displayed in figure 1 . from this figure
it follows that the wkb method used by kokkotas and schutz can not be trusted for modes with a very small real part and a relatively large imaginary part . in order to illustrate the difficulties involved in numerical integration along the real @xmath5-axis and thus the advantage of using complex coordinates we also show results of a real-@xmath5 calculation in figure 1
this specific example corresponds to the worst possible situation .
the numerically computed function @xmath42 has a singularity close to the real @xmath5-axis . that this can be expected if one fails to account for the stokes phenomenon is clear from the equations given in the appendix of andersson .
note also that the magnitude of @xmath42 drops dramatically immediately after passing close to this singularity .
when the surface of the star is reached @xmath58 has decreased more than ten orders of magnitude .
our study shows that the wkb technique should be reliable for modes with large real parts .
this conclusion would mean that the `` kink '' in the spectrum of kokkotas and schutz can not yet be dismissed .
however , we can proceed one step further and use the numerically determined function @xmath42 in the necessary matching at the stellar surface .
this should provide an accurate way of dealing with the exterior problem , and it will hopefully enable us to conclude whether the `` kink '' is real or not .
in principle , it is a simple task to derive a normal - mode condition based on the phase - amplitude approach .
we want to match the numerical solution to the inside problem to the solution for the exterior at the surface of the star .
that is , we require that the zerilli function and its derivative be continuous at the surface @xmath59
. then we can use ( [ zpsi ] ) and @xmath60 which follows immediately from ( [ tortoise ] ) .
let us first assume that the physically acceptable solution for the interior of the star corresponds to a mixture of out- and ingoing waves at spatial infinity .
then we have @xmath61 outside the star , and we get from ( [ pamsols ] ) @xmath62 -a_{\rm out } \psi^- \left [ iq + { 1\over 2q } { dq \over dr } \right]\ .\ ] ] the next step is to solve ( [ qequation ] ) for a particular value of @xmath29 from a point far away from the star in the way described in the previous section .
this means that the above expressions for @xmath35 and its derivative in terms of @xmath63 will remain valid .
moreover , if we take the lower limit of integration in ( [ pamsols ] ) to be the surface @xmath59 of the star ( where we do the matching ) we get @xmath64 which means that we need only determine @xmath42 and its derivative .
it should be pointed out that the amplitudes @xmath65 and @xmath66 as defined by ( [ combi ] ) differ slightly from those that follow from the asymptotic behaviour of the zerilli function ; @xmath67 the difference is essentially a phase - factor and is of no importance for a search for mode - frequencies . with our definition the amplitudes of the functions that asymptotically represent out- and ingoing waves are equal at the surface of the star [ see ( [ ampli ] ) ] .
straightforward algebra leads to @xmath68 + z_s^\prime \right\ } \ , \nonumber \\\end{aligned}\ ] ] and @xmath69 + z_s^\prime \right\ } \
, \nonumber \\\end{aligned}\ ] ] where @xmath70 represents the interior solution at the surface of the star , and a prime denotes a derivative with respect to @xmath33 . however , since the solution is only determined up to a constant factor by the physical conditions it makes more sense to study @xmath71 - z_s^\prime \over z_s \left [ \left ( 1 - { 2 m \over r } \right ) \left(iq + { 1\over 2q } { dq \over dr } \right ) + { m \over r^2 } \right ] + z_s^\prime } \ . \label{ratio}\ ] ] a quasinormal mode of the stellar system corresponds to a singularity of this ratio as a function of @xmath29
. it may be worthwhile to clarify the meaning of @xmath72 in the above equations , since we are assuming that @xmath5 is complex in the exterior . in principle
, one could simply interpret ( letting @xmath73 represent the complex @xmath5 ) @xmath72 as @xmath74 but this may not be very practical .
as mentioned previously , we integrate for the exterior solution along a straight line in the complex @xmath73-plane .
this line is parametrized by the real distance @xmath75 ( from the surface of the star ) with constant phase angle @xmath76 .
then @xmath77 and @xmath78 is the derivative of @xmath42 with respect to the real parameter .
thus we get @xmath79 which is easily implemented numerically .
the main purpose of the present study is to understand the discrepancies between the results of kokkotas and schutz and those of leins _ et al . _ .
specifically , we want to find the `` new '' modes that seem to exist for very small oscillation frequencies and large damping @xcite .
we want to see if our ad hoc explanation for why those modes could not be found when a wkb approach to the exterior problem was used ( see section 4 ) is correct , _
i.e. , _ that one must account for stokes phenomenon for frequencies close to the imaginary @xmath29-axis .
we also feel that it is imperative that we check the possible existence of the `` kink '' that kokkotas and schutz found in their spectra , especially since this feature was not found by leins _
et al . _ .
we have performed detailed calculations for the four models studied by kokkotas and schutz . a sample of the numerical results for model 4 ( that has characteristics : @xmath80 g/@xmath81 , @xmath82 km and @xmath83 , _ i.e. , _
@xmath84 ) can be found in table 1 .
.a sample of characteristic frequencies for stellar modes associated with model 4 of kokkotas and schutz ( 1992 ) .
the highly damped ( w ) modes are compared to the slowly damped ( p ) modes .
the results for w - modes are in good agreement with those of leins _ et al . _
( 1993 ) the frequencies found for the f - mode and the first of the p - modes are in perfect agreement with those obtained by kojima ( 1988 ) .
the `` new '' w - modes found by leins _
are indicated by @xmath85 ( the imaginary part of the one in parenthesis is too large for our program to provide truly reliable results , but there does seem to be a mode there ) . the same data is shown in figure 2 .
all entries are given in units of @xmath86 .
the given damping rates for the slowly damped modes are accurate to a few parts in @xmath87 .
[ cols="^,^,<,^ , < , < " , ] we set out on this investigation to verify the existence of modes with large damping and small oscillation frequencies that had been found by leins _
et al . _ , and also to find whether there is a `` kink '' in the w - mode spectrum or not .
the first of these questions was readily answered as soon as we had combined the numerical integration approach for the exterior vacuum ( see sections 35 ) with the numerical code that kokkotas and schutz had used for the interior problem .
the modes found by leins _
et al . _ do , indeed , exist . in view of our understanding of the failure of the wkb approach for these frequencies
this makes sense .
as is clear from figure 1 , the wkb approach does break down when the real part of the frequency is small .
leins _ et al .
_ assumed that the `` kink '' in the spectra of kokkotas and schutz was evidence that the wkb method fails also as @xmath88 increases .
we have found that this is not the case .
our first calculations gave results in good agreement with those of kokkotas and schutz for high frequencies , with a pronounced kink in all spectra , and it was clear that the wkb method is reliable for high oscillation frequencies .
_ argue that their approach to the exterior problem did not involve approximations and should therefore be considered as more reliable than the wkb approximation .
still , we found that our new numerical scheme gave results that did not agree well with those of leins _
et al._. at the same time , both methods have proved to be reliable and providing results of high accuracy in the case of black holes .
hence , it seemed likely that they should both be trusted , and that any discrepancies , _ eg . _ , the `` kink '' , were due to differences in the approach to the interior problem .
unfortunately , this meant that we faced a rather difficult situation .
the approach of detweiler and lindblom has been used in all studies of highly damped oscillations of relativistic stars .
hence , any differences would be in the numerical codes and rather difficult to find . with this in mind we scrutinized our code for the interior problem , and found points where
it could be improved .
the amended code is much more reliable , and numerically accurate , than the original one used by kokkotas and schutz .
it also runs considerably faster .
it turns out that most of the improvements have a minor effect on the numerical results , however .
( the results quoted in table 1 were , obviously , computed with the final version . ) after testing the code we have a much clearer idea of its restrictions .
most importantly , we have found that the `` kink '' is intimately related to the choice of point close to the centre of the star where power series expansions are used to initiate the numerical integration .
when choosing a starting point one introduces an artificial length - scale in the problem , and this scale is reflected by the oscillation frequency at which the `` kink '' occurs .
this can easily be illustrated if one considers the eigenfunctions : as the oscillation frequency increases the number of nodes in each eigenfunction increases .
one will eventually reach a frequency where nodes should occur for smaller radii than the one where the numerical integration is initiated .
the failure of the power - series solutions to account for this feature of the physical solutions leads to the `` kink '' in the spectrum . by initiating the integration sufficiently close to the centre of the star one
can ensure that this `` kink '' does not occur in the frequency regime of interest . in the calculations discussed here we initiated the integration at a point roughly corresponding to @xmath89 of @xmath59 .
it seems likely that leins _
_ initiated their integration at a point closer to the centre than we did originally , and had they continued their calculation to higher oscillation frequencies they would have found a `` kink '' similar to that of kokkotas and schutz .
the `` kinks '' are numerical artefacts .
we also find that it is extremely difficult to avoid numerical noise in the eigenfunctions .
this is mainly due to the fact that the eigenfunction @xmath25 decreases rapidly towards the surface of the star ( where it should vanish ) .
when solving for the interior of the star one matches a solution from the centre ( that satisfies the physical condition of regularity ) to one from the surface of the star at some intermediate point . in order to get a reliable result it is imperative that this matching is not done at too large a radius .
we find that the interior matching should be performed for radii smaller than something like @xmath90 .
furthermore , the calculation gets more difficult as @xmath91 increases .
this is due to the fact that @xmath92 is then many magnitudes smaller than any of the other three eigenfunctions @xmath93 and @xmath19 .
thus it is difficult to compute @xmath25 with acceptable accuracy .
for this reason the calculation is more difficult for less relativistic stellar models : the damping of the w - modes generally increases as the model gets less relativistic .
interestingly , this explains why the results of kokkotas and schutz for model 1 were not as good as for the other three models . in view of this fact we decided to restrict our detailed study to models 24 of kokkotas and schutz , and also not to attempt a search for modes with @xmath94 .
we feel that the interior problem must be reformulated if a reliable search for modes with very large damping is to be made . in this context
a study based on the alternative formulations of chandrasekhar and ferrari , or ipser and price , may provide interesting results .
however , this does not mean that our results should not be taken seriously .
we believe that the present study is the most accurate one to date within the range of validity that we have indicated .
an interesting effect of the improvements that we made to our code for the interior problem is that it is now considerably more accurate for modes with very slow damping . hence
, it is not at all difficult to compute the f - mode for the various models .
furthermore , we are able to iterate for several of the first p - modes .
as far as we know noone has attempted to compute these modes before , and the results that we obtain are somewhat surprising .
the damping of the p - modes does not necessarily , as has been generally presumed , decrease rapidly as one proceeds to higher oscillation frequencies .
in fact , for the most compact of our models ( model 4 ) we find an intriguing behaviour .
there is a low - order mode that damps away much faster than expected ( see table 1 ) .
this more rapidly damped mode occurs only for the most compact of our stellar models .
this is an interesting result that could be of potential importance for many of our astrophysical expectations , such as the life - time of gravitational - wave sources .
furthermore , methods used by other authors to study f - modes could be used to test its correctness . for model 4
the ratio of real to imaginary part for the relevant mode is not considerably different from that for the f - mode .
therefore , it seems likely that the program used by detweiler and lindblom , or that of kojima , could be used to test our results . in fact , dr kojima and dr lindblom have both very kindly performed this calculation .
the results they obtain for the p - modes of model 4 are in good agreement with those listed in our table 1
. it may be worthwhile to comment on the difficulty of finding p - modes here .
it is clear that a numerical code must be very robust and accurate if iteration for these slowly damped modes is to be at all possible .
the imaginary part is typically at least six orders of magnitude smaller than the real part , and if one requires single - digit precision in the imaginary part one must therefore achieve at least six digits in the real part .
our amended code should be reliable for imaginary parts larger than @xmath95 . in cases when the imaginary part of the mode - frequency is so small that iteration is not possible , one can always attempt to infer the real part from a graphical approach . we know that the modes correspond to the zeros of the asymptotic amplitude @xmath65 ( or the singularities of the ratio @xmath96 ) .
if such a zero is situated close to the real @xmath29-axis it should , in principle , be easy to distinguish in a plot of @xmath97 .
this idea was recently exploited by , for example , ferrari and germano .
that it works nicely is illustrated in figure 3 . when generating this figure we normalised the perturbed quantities in such a way that the radial displacement @xmath19 is equal to 1 at the surface of the star for all frequencies .
the slowly damped p - modes give rise to very narrow singularities that are easy to distinguish although the data in the figure correspond to @xmath98 .
it is not very time consuming to create a figure with this kind of resolution , and it is certainly worthwhile . in fact , we can not only pin down all the p - modes given in table 1 , but also guess at which oscillation frequencies there will be a w - mode .
it seems as if a peak in @xmath97 is an indication of a w - mode higher up in the complex @xmath29-plane .
we did not expect to find this feature in this kind of graph and the exact origin of these peaks is something of a mystery .
interestingly the peaks do not show up in similar graphs generated by the numerical code of lindblom ( private communication ) .
it thus seems that the difference between the two approaches , _
i.e. _ between ( i ) matching the interior solution to the exterior one at the surface of the star and ( ii ) using the value of the zerilli function at the surface as initial data for integration towards infinity where the solution is matched to an asymptotic form , is crucial .
we believe that the information contained in this sort of graph is important to our understanding of the stellar pulsations and are presently investigating this issue further .
although we do not yet have the a clearcut explanation of the phenomenon it is worthwhile to make one final point .
it is easy to show that the peaks in figure 3 originate from the interior solution and not the exterior one .
remembering that the zerilli function is generated from the two spacetime perturbations @xmath4 and @xmath3 , this agrees well with the notion that the spacetime perturbations dominate the fluid ones ( which set the normalisation for the figure ) at the w - mode frequencies .
leins _ et al .
_ argued that they had found a new family of highly damped modes .
given the present evidence we have to assess whether that is the case or not .
we have shown why kokkotas and schutz could not find modes with high damping and small oscillation frequencies with the wkb scheme they used .
but surely , that in itself does not imply that any modes they could not find belongs to a new family ?
different families of modes must be distinguished for clear physical reasons .
the difference between the two families ( p- and g - modes ) of slowly damped modes is well established , and the w - modes are certainly distinct . but given the knowledge available at the moment it is not at all clear that a similar split of the highly damped modes into different classes makes sense .
leins _ et al .
_ put forward two arguments for why their new modes belong to a different family .
the first relates to the number of nodes of the eigenfunctions corresponding to the various modes .
they find that the number of nodes increase systematically as one progresses up the w - mode sequence , but that the situation is not that clear for the new modes . in doing this
they claim to be studying the `` amplitude '' of the eigenfunctions .
what they actually study is the number of nodes in the real part of each eigenfunction .
an argument solely based on the real part of a manifestly complex eigenfunction clearly does not make much sense . from , for example ,
( [ metric ] ) it is evident that both the real and the imaginary part will be relevant .
hence , the nodes in the imaginary part of the eigenfunction must also be studied .
when that is done the conclusion regarding the higher oscillation frequencies remains the same , but for the new modes the situation is somewhat clearer . the imaginary part of each eigenfunction may have a node even if the real part does not , and vice versa . however , although it may provide a useful diagnostic , the number of nodes in the eigenfunctions is not a very good measure for distinguishing different families of modes .
it is well known [ see for example chapter 17 in cox ] that different p - modes may have an identical number of nodes in the eigenfunctions , and there is no reason why that could not be the case also for highly damped modes .
however , leins _
_ argue that their new modes are different from the ones of kokkotas and schutz because of other features of the eigenfunctions . for the new modes the eigenfunctions die out towards the centre of the star , whereas the amplitude stay roughly constant for all values of @xmath5 for the old modes . this difference is similar to that between g- and p - modes @xcite . for the g - modes which correspond to oscillation frequencies smaller than that of the f - mode
the eigenfunctions are known to die out towards the centre of the star , whereas this is not the case for the p - mode eigenfunctions .
could it be that the perturbations generally die out towards the centre of the star for very small oscillation frequencies ?
if so , a distinction of different w - mode families based on this feature would not be satisfactory . at the present time
, it seems premature to divide the w - modes of relativistic stars into different `` families '' .
if such a classification is to make sense it must be based on clear physical principles , and our understanding of the highly damped stellar oscillations is still far from satisfactory .
the issue could probably be resolved by further detailed studies of different families of stellar models @xcite .
a vital piece of information that is still missing from the puzzle regards the possible existence of modes with very large imaginary parts .
as mentioned previously we have found that the present approach to the interior problem suffers from numerical difficulties when @xmath94 or so .
this is not to be taken as an indication that modes with larger imaginary parts do not exist . on the contrary , such modes may well exist and a mode survey that covers the entire complex frequency plane could produce interesting results .
in order to understand the discrepancies between results obtained by kokkotas and schutz and leins , nollert and soffel , we have computed highly damped modes for relativistic stars .
the stellar models considered are simple polytropes used in several previous investigations .
we have outlined a new numerical approach to the problem for the exterior of the star .
this new scheme is based on numerical integration for complex coordinates , and is similar to one that has proved extremely reliable for black - hole problems @xcite . with this new approach to the stellar exterior we are able to find modes that were not identified by kokkotas and schutz .
these modes , that are highly damped and situated close to the imaginary frequency - axis , agree perfectly with modes found by leins _
that the wkb method employed by kokkotas and schutz failed to distinguish these modes can be understood if the so - called stokes phenomenon ( familiar from wkb theory ) is accounted for .
we have managed to explain the occurence of a `` kink '' in the spectra of kokkotas and schutz : it is a numerical artefact due to the choice of point close to the centre of the star where power - series expansions are used to initiate the numerical integration .
this conclusion was drawn after we had scrutinized the way that we dealt with the stellar interior .
we also found other points where our original code could be improved .
although none of those changes affected the numerical results significantly , the program is now more robust and runs considerably faster . that it allows us to do accurate calculations is illustrated by the fact that we could iterate for several of the very slowly damped p - modes
this led to a rather surprising result : for the most compact of our stellar models ( corresponding to @xmath84 ) the damping of the p - modes does not decrease monotonically as one proceeds to higher oscillation frequencies .
there is a low - order mode that damps away at least ten times faster than anticipated .
this result may significantly affect our expectations regarding , for example , stellar stability and the lifetime of gravitational - wave sources .
hence , it is of some importance that it be studied in more detail .
the new complex - coordinate method that we have employed in the present study seems promising also for rotating stars , _ i.e. _ , when the exterior spacetime is only known approximately . since all previously suggested methods will surely fail to handle that difficulty this new approach could turn out to be of considerable importance .
the anomalous p - mode damping that we have discovered here would be especially interesting if it exists for rotating star f - modes , because these limit stability .
such issues are , however , beyond the scope of the present investigation and we will return to them in the future .
we would like to thank hans - peter nollert for providing us with numerical results that were not published in their paper .
discussions with him provided valuable information that helped us find the reason for the discrepancies between the previous studies of this problem .
we are also grateful to yasufumi kojima and lee lindblom who most kindly provided us with independent checks of our results for slowly damped modes . | the oscillation modes of a simple polytropic stellar model are studied . using a new numerical approach ( based on integration for complex coordinates ) to the problem for the stellar exterior
we have computed the eigenfrequencies of the highly damped w - modes .
the results obtained agree well with recent ones of leins , nollert and soffel [ _ phys .
rev .
d _ * 48 * 3467 ( 1993 ) ] . specifically , we are able to explain why several modes in this regime of the complex frequency plane could not be identified within the wkb approach of kokkotas and schutz [ _ mon . not .
r. astr .
soc . _ * 255 * 119 ( 1992 ) ] .
furthermore , we have established that the `` kink '' that was a prominent feature of the spectra of kokkotas and schutz , but did not appear in the results of leins _ et al .
_ , was a numerical artefact . using our new numerical code we are also able to compute , for the first time , several of the slowly damped ( p ) modes for the considered stellar models . for very compact stars
we find , somewhat surprisingly , that the damping of these modes does not decrease monotonically as one proceeds to higher oscillation frequencies .
the existence of low - order modes that damp away much faster than anticipated may have implications for questions regarding stellar stability and the lifetime of gravitational - wave sources .
the present results illustrate the accuracy and reliability of the complex - coordinate method and indicate that the method could prove to be of great use also in problems involving rotating stars .
there is no apparent reason why the complex - coordinate approach should not extend to rotating stars , whereas it is accepted that all previous methods will fail to do so .
stars : neutron stars - gravitational waves |
lacoo@xmath2 crystallizes in a rhombohedral symmetry ( r@xmath12c ) and has quite unusual physical properties . at low temperatures it features a nonmagnetic and insulating ground state . with increasing temperature
a paramagnetic moment develops which causes a maximum in the susceptibility around 100k @xcite .
this is usually explained in terms of a temperature - driven spin - state transition of the co@xmath8 ions which can occur in three different spin configurations : a low - spin ( ls ) ( @xmath13 , @xmath7 ) , an intermediate - spin ( is ) ( @xmath14 , @xmath15 ) , or a high - spin state ( hs ) ( @xmath16 , @xmath17 ) . at low temperatures a ls ground state is realized .
the nature of the spin state at higher temperatures is controversially discussed , e.g. @xcite .
the spin - state transition causes also a large additional schottky - like contribution to the thermal expansion of lacoo@xmath2 @xcite .
the substitution of la@xmath8 by an amount @xmath11 of divalent ca , sr , or ba ions influences this contribution and leads to a strong suppression of the spin - state transition .
furthermore all doping series show long - range ferromagnetic order and metallic behavior at higher doping levels @xmath18 . in the case of ca doping ,
ferromagnetic behavior is observed for the insulating compounds @xmath19 , too @xcite .
ca@xmath1coo@xmath2 for ( a ) @xmath20 , ( b ) @xmath21 , ( c ) @xmath22 and ( d ) @xmath23 .
the curve for @xmath24 is shown in each panel .
the black circles denote the critical temperatures @xmath25 of the onset of ferromagnetic order which have been taken from the magnetization data given in ref .
@xcite . ] in fig .
[ tad ] the temperature dependence of the linear thermal - expansion coefficient @xmath26 of la@xmath0ca@xmath1coo@xmath2 is shown .
[ tad](a ) clearly reveals that the pronounced maximum in the thermal - expansion data of the undoped compound ( @xmath4 ) is strongly suppressed with increasing doping concentration @xmath11 . above @xmath27 this indication of the spin - state transition
has completely vanished .
instead several anomalies develop ( figs .
[ tad](b ) ( d ) ) : the curves for @xmath28 exhibit two distinct minima and for @xmath29 additional kinks occur .
the low - temperature minimum ( ltm ) shifts with increasing @xmath11 to lower whereas the high - temperature minimum ( htm ) moves to higher temperatures . for @xmath30 we could observe only the onset of the decrease in @xmath5 , because the anomaly has shifted out of the available temperature interval , i.e. @xmath31k , for higher doping concentrations . obtained from x - ray diffraction data of la@xmath32ca@xmath33coo@xmath2 :
the arrows indicate whether the data was taken with increasing or decreasing temperature .
the lines are to guide the eyes . ] at room temperature a doping induced structural phase transition from rhombohedral ( r@xmath12c , @xmath34 ) to orthorhombic symmetry ( pbnm , @xmath35 ) is observed @xcite .
this transition shifts to lower temperatures for smaller @xmath11 .
[ roentg ] shows the temperature dependence of the structural phase fractions for @xmath36 which was obtained from x - ray diffraction data .
it reveals a first - order phase transition .
the htm for @xmath36 occurs at @xmath37k which coincides with the crossover from r@xmath12c to pbnm in the x - ray data .
therefore the occurrence of the htm in @xmath5 can be attributed to a doping - induced structural phase transition . in fig .
[ tad ] , the circles ( @xmath38 ) mark the onset of ferromagnetic order .
the critical temperatures @xmath25 of the magnetic transitions have been taken from the @xmath39 data given in ref .
they match the occurrence of the kinks in @xmath5 for @xmath24 and @xmath40 ( figs .
[ tad](c ) , ( d ) ) .
therefore we conclude that the kink anomalies in the thermal - expansion data result from the onset of long - range ferromagnetic order . of la@xmath0ca@xmath1coo@xmath2
: @xmath41 and @xmath42 mark the occurrence of the ltm and htm in @xmath5 , @xmath38 signals the ferromagnetic phase transition .
panel ( b ) shows an enlarged view of the intermediate doping region @xmath43 . for clarity the different data sets
are shifted by 0.1j / molk@xmath44 with respect to each other . ] in fig .
[ cpt ] specific - heat data @xmath45 of la@xmath0ca@xmath1coo@xmath2 is presented .
one can identify several anomalies , too , although they are much less pronounced than in @xmath5 .
the temperatures of the minima in @xmath5 ( ltm : @xmath41 , htm : @xmath42 ) and the onset of ferromagnetic order ( @xmath38 ) are denoted .
[ cpt](b ) gives an expanded view for @xmath43 .
please note , that for clarity the different data sets for @xmath46 are shifted by 0.1j / molk@xmath44 with respect to each other . for @xmath47
all compounds feature one or more weak anomalies in @xmath45 , which coincide with the anomalies in @xmath5 . for @xmath36 and 0.17
the ltm and the magnetic phase transition occur at similar temperatures . in the @xmath5 data of the higher doped specimens
the onset of ferromagnetism is indicated by kinks .
these kinks might be hidden by the large ltm anomalies for @xmath36 and 0.17
. therefore one can not distinguish if the anomalies in @xmath45 for these two compounds are caused by the magnetic or a different transition .
for @xmath48 the specific - heat data features only a weak shoulder around the temperature of the corresponding ltm .
the anomalies visible in the data of the higher doped compounds @xmath35 are clearly related to the ferromagnetic phase transition .
the structural phase transition shifts with increasing @xmath11 to higher temperatures and causes in both quantities clear anomalies .
the appearance of the htm in @xmath5 ( @xmath42 ) and of the maxima at higher temperatures in @xmath45 take place at similar temperatures . for @xmath49
one can see a weak maximum in the specific - heat data above room temperature which agrees with the onset of a minimum in the thermal - expansion data .
the origin of the ltm which is visible in @xmath5 is not clear yet .
this anomaly vanishes for @xmath50 and possibly does not affect the specific heat .
the investigation of its origin is still under progress and will be reported elsewhere . in summary ,
we have studied the influence of hole doping on the series la@xmath0ca@xmath1coo@xmath2 by thermal - expansion and specific - heat measurements .
depending on the concentration @xmath11 we observe up to three distinct anomalies .
one of them is related to the ferromagnetic ordering , while a second one arises from a first - order structural phase transition .
the origin of the third anomaly remains to be clarified .
this work was supported by the deutsche forschungsgemeinschaft through sfb 608 . | we report thermal - expansion and specific - heat data of the series la@xmath0ca@xmath1coo@xmath2 for @xmath3 . for @xmath4
the thermal - expansion coefficient @xmath5 features a pronounced maximum around @xmath6k caused by a temperature - dependent spin - state transition from a low - spin state ( @xmath7 ) at low temperatures towards a higher spin state of the co@xmath8 ions .
the partial substitution of the la@xmath8 ions by divalent ca@xmath9 ions causes drastic changes in the macroscopic properties of lacoo@xmath2 . around @xmath10
the large maximum in @xmath5 has completely vanished . with further increasing @xmath11 three different anomalies develop
. 75.40,65.70 , specific heat , thermal expansion , spin - state transition , metal - insulator transition , double - exchange mechanism |
we start by proving the first lemma stated in the letter .
_ lemma 1._consider @xmath33 as dichotomic observables ( with possible outcomes @xmath2 and @xmath34 ) , @xmath35 as trichotomic observables ( with possible outcomes @xmath36 , @xmath37 , and @xmath38 ) , with @xmath39 and labels @xmath31 and @xmath32 corresponding to the respective party .
then , for any realistic theory , the following inequality holds : _ proof._each probability in can be written as a sum of a set of probabilities for complete hidden variable states . in our scenario there are four dichotomic observables , @xmath43 , @xmath45 , @xmath46 , and @xmath47 , and four trichotomic observables , @xmath48 , @xmath49 , @xmath44 , and @xmath50 .
the proof is a bookkeeping exercise showing that each term that adds to the positive term @xmath51 also appears ( with the right multiplicity ) at some of the negative terms , implying @xmath52 .
we will need some notation .
a complete hidden variable state will be given by @xmath79 any summation without defined index denotes summation over all the unspecified symbols . with this convention ,
the positive contributions for @xmath80 are given by @xmath81 namely , @xmath82 terms , where @xmath83 appear with multiplicity two .
let us begin by chasing all terms @xmath84 .
the first negative term will cancel all terms like @xmath85 and @xmath86 , so one eliminates all terms with @xmath87 or @xmath88 .
it remains to be canceled all terms such that the pair @xmath89 be given by @xmath90 , @xmath91 , @xmath92 , and @xmath93 .
the term @xmath94 cancels terms where @xmath95 or @xmath96 .
but we still have to look for the @xmath97 terms @xmath98 .
the term @xmath99 , with @xmath100 , cancels all remaining terms with @xmath101 , except those with @xmath102 .
some of these terms are included in @xmath103 , and we are now concerned only with the terms @xmath104 and @xmath105 . on the same way , but with @xmath106 , @xmath107 with @xmath108 goes away , leaving @xmath109 , @xmath110 , and @xmath111 .
the first of these terms goes with @xmath112 , the last goes with @xmath113 , while the remaining reduces to @xmath114 , due to @xmath115 , which is finally canceled by @xmath112 .
now we can follow a similar strategy for the terms @xmath116 .
starting from @xmath117 , which eliminates all terms with @xmath101 or @xmath107 , and @xmath118 , which eliminates @xmath108 and @xmath119 , we are left with @xmath120 .
the term @xmath121 , with @xmath100 , reduces the task to @xmath122 and @xmath123 .
the same term , with @xmath106 , reduces even more to @xmath124 , @xmath125 , @xmath126 , and @xmath127 .
the two first terms are eliminated by @xmath128 , while @xmath129 reduces the order two to @xmath130 , and @xmath131 .
the term @xmath115 with @xmath100 eliminates the first and reduces the second to @xmath132 , which is eliminated by the same term with @xmath106 . as already said ,
@xmath133 terms appear with twice each in the positive part .
we close the demonstration by proving that they also appear ( at least ) twice in the negative contributions .
they are @xmath134 .
let us start by the terms @xmath135 .
they appear once at @xmath117 , for @xmath100 , and a second time at @xmath136 , @xmath137 , if @xmath138 , at @xmath112 , @xmath100 , if @xmath108 , and at @xmath103 , in case @xmath139 .
analogously , for the terms @xmath140 , they appear once at @xmath94 , and again at @xmath121 , @xmath100 , if @xmath138 , at @xmath141 , @xmath137 , if @xmath108 , and at @xmath128 , if @xmath142 .
finally , for the states @xmath143 , they appear once at @xmath144 , @xmath145 , if @xmath146 , at @xmath141 , @xmath145 , if @xmath119 , and at @xmath115 , @xmath145 , if @xmath102 .
the second appearance can be noted at @xmath129 , if @xmath138 , at @xmath113 if @xmath108 , while for @xmath142 we split again into three terms : @xmath147 , @xmath106 for @xmath107 , @xmath94 , @xmath106 , for @xmath96 , and @xmath115 , @xmath106 , for @xmath148 .
as stated , only the following @xmath54 out of the @xmath55 realistic states are noncontextual : @xmath149 when considering two parties , the cross product must be applied , resulting in @xmath150 noncontextual states .
_ proof._let us prove inequality in detail .
the others are analogous .
note that only three one - party states allow for @xmath152 : @xmath153 , @xmath154 , and @xmath155 . in the same way
, only three states of party @xmath32 will allow for @xmath156 : @xmath157 , @xmath158 , and @xmath159 .
so we have @xmath160 noncontextual states contributing for @xmath161 . but
note that six of these states have @xmath138 , and the other three have @xmath146 , so once again all terms which add for the positive part also appear in the negative one .
as probabilities are non - negative , the resulting inequality applies .
to see that inequality is tight , just use state @xmath162 , while @xmath163 is a realistic contextual state ( in the sense that it is an assignment of values which does not obey the assumption of noncontextuality ) which violates it . _
proof._it is again necessary to cancel out the same nine terms which contribute to @xmath161 .
four of them have @xmath95 and @xmath138 .
also four of them have @xmath96 and @xmath146 , but one was taken twice .
the remaining two terms are canceled by the second summation term , since one has @xmath165 and the other @xmath166 .
_ quantum violation._now let us discuss in more details how the quantum violation of the presented inequalities is directly related to the first example discussed in the letter .
for this , we will focus on the inequality @xmath169 since every term is conditioned to @xmath77 , we can consider a projective measurement on part @xmath32 with the result corresponding to @xmath170 appearing .
since the conditional state of @xmath171 given that @xmath172 was obtained is @xmath173 , one is faced with the quantum probabilities | it has been argued that any test of quantum contextuality is nullified by the fact that perfect orthogonality and perfect compatibility can not be achieved in finite precision experiments .
we introduce experimentally testable two - qutrit violations of inequalities for noncontextual theories in which compatibility is guaranteed by the fact that measurements are performed on separated qutrits .
the inequalities are inspired by the basic building block of the kochen - specker proof of quantum contextuality for a qutrit , despite the fact that their proof is completely independent of it . _
introduction_contextuality is a fascinating property of nature : the result of an experiment may depend on other compatible experiments that may be performed .
this is surprising because the experiment gives the same result when repeated after any number of compatible experiments .
it is also surprising because the probability of obtaining any particular result does not depend on which compatible experiments might be performed ( probabilities are noncontextual ) .
nature s
contextuality was pointed out by the discovery that some predictions of quantum mechanics ( qm ) can not be reproduced by any noncontextual theory , and that this conflict occurs for any state of any system with three or more distinguishable states @xcite .
quantum contextuality has far - reaching consequences , such as the impossibility of describing nature by classical theories with bounded speed of information @xcite or bounded density of memory @xcite , or the possibility of device - independent eternally secure communications @xcite .
the original demonstration by kochen and specker ( ks ) @xcite is usually considered a _ tour de force _ because it is based on a @xmath0-vertex graph . however , this graph has an @xmath1-vertex building block , which we refer to as the basic diagram ( fig .
[ ksblock ] ) . in a ks graph , each vertex is seen as a yes or no proposition and linked vertices are compatible propositions where no more than one of them can be assigned the yes answer ( a property sometimes called exclusiveness ) .
the original @xmath0-vertex graph was built to give a state - independent proof of the impossibility of assigning noncontextual answers reproducing the predictions of quantum mechanics for a three - level system ( like a spin @xmath2 ) .
accordingly , a triangle represents a set of three compatible propositions , one of them receiving the yes answer ( given quantum mechanically by a projective complete measurement , like the @xmath3 component of a spin 1 particle ) .
the basic building block is also referred to as a yes - no diagram , in the sense that if the answer yes is noncontextually assigned to vertex @xmath4 , any valid assignment must give to the vertex @xmath5 the value no .
this letter was motivated by this basic diagram , as we will show below . in 1999 , a debate on the physical impact and experimental testability of the kochen - specker theorem started in the pages of physical review letters .
meyer @xcite and kent @xcite pointed out that finite precision measurement nullifies the physical content of the kochen - specker theorem .
more than 20 papers have been published since then supporting ( for instance , @xcite ) or criticizing ( for instance , @xcite ) this conclusion .
more recently , a series of experimental tests with ions @xcite , neutrons @xcite , photons @xcite , and nuclear magnetic resonance systems @xcite have been questioned due to a variant of the objection of meyer and kent called the compatibility loophole @xcite . in this letter we present a proposal for a type of experiment that ( we hope ) will definitely close the debate on the physical relevance and experimental testability of the kochen - specker theorem and stimulate a new generation of experiments .
since it is impossible to have contextuality for a two - level system , subjected to projective measurements , the three - state quantum system , or qutrit , is considered the simplest example of contextuality .
we give here a state - dependent quantum contextuality demonstration @xcite with a twofold objective : to provide a clear example for the nonspecialized reader and to fix notation to be used in the two - qutrit case .
suppose the qutrit is initially in the state @xmath6 where @xmath7 , @xmath8 , and @xmath9 are three orthogonal states .
then the measurement represented by the projector @xmath10 , with @xmath11 gives the result @xmath2 ( representing `` yes '' ) , with ( quantum mechanical ) probability @xmath12 this already establishes the _ yes - yes _ possibility and we need only to prove the existence of the six other vertices obeying all compatibility and exclusiveness properties of the ks basic building block . for this , consider the observable @xmath13 where @xmath14 , @xmath15 , and @xmath16 are the following orthogonal states : [ h0 ] @xmath17 then , according to qm , the result @xmath18 can never be obtained since @xmath19 . in addition , in any noncontextual theory in agreement with qm , the result of @xmath20 can not be @xmath21 since @xmath22 .
therefore , in any noncontextual theory in agreement with qm , the initial state @xmath23 together with the positive probability imply @xmath24 , whenever the yes answer should be found for @xmath25 . consider also @xmath26 where [ h1 ] @xmath27 the same reasoning leads to the conclusion that , in any noncontextual theory in agreement with qm , @xmath28 whenever @xmath5 is to happen .
however , @xmath24 and @xmath28 can not happen simultaneously since @xmath29 .
therefore , there is no noncontextual assignment of results to the six propositions @xmath30 in agreement with the predictions of qm for a qutrit prepared in the state @xmath23 and postselected in the state @xmath25 . indeed , the orthogonality relations between these eight states constitutes the building block of the kochen - specker proof @xcite .
these eight quantum states obey all orthogonality conditions represented on the building block diagram ( see fig .
[ ksblock ] ) , but the nonorthogonality of @xmath23 and @xmath25 allows for a yes - yes answer , ruled out by noncontextual assignments . building block of the kochen - specker proof of quantum contextuality ( left ) .
vertices represent propositions ; two of them are joined when the propositions are compatible and both can not be true .
kochen drawing the block in zrich in 2009 ( right).,width=283 ] what if perfect orthogonality can not be experimentally achieved ?
then the previous argument vanishes .
for instance , the block of fig .
[ ksblock ] can not be constructed if the unit vectors are restricted to vectors with rational components @xcite or to a set where each vector participates only of one context @xcite , although in both cases these sets are dense in the set of unit vectors and therefore are undistinguishable from the set of all unit vectors by experiments with finite precision @xcite .
a related problem affecting some recent experimental violations @xcite of noncontextual inequalities @xcite involving sequential measurements on the same physical system is the fact that these inequalities are based on the assumption that the sequential measurements are perfectly compatible , something which does not occur in actual experiments , so to conclude contextuality extra assumptions are needed @xcite . here
we avoid these extra assumptions by measuring one of the observables in one qutrit and the second observable on a distant qutrit .
spatial separation provides a physical basis to the assumption that both measurements are not only approximately but perfectly compatible . the combination of state - independent kochen - specker proofs and correlated systems is a standard way to show contextuality and nonlocality for composite systems @xcite . here
, instead , we do not require a state - independent proof .
our central result will be to adapt the previous one - qutrit example to find two - qutrit quantum violations for noncontextual inequalities derived under the same inspiration . for this purpose
we start by deriving an inequality valid for any realistic theory ( i.e. , those assigning predefined answers for all possible questions ) based on the observables of the block of kochen and specker and involving correlations between measurements on two different parties @xmath31 and @xmath32 .
_ lemma 1._consider @xmath33 as dichotomic observables ( with possible outcomes @xmath2 and @xmath34 ) , @xmath35 as trichotomic observables ( with possible outcomes @xmath36 , @xmath37 , and @xmath38 ) , with @xmath39 and labels @xmath31 and @xmath32 corresponding to the respective party .
then , for any realistic theory , the following inequality holds : [ bhinequality ] @xmath40 where @xmath41 where @xmath42 denotes the joint probability of obtaining the results @xmath2 and @xmath18 for @xmath43 and @xmath44 , respectively .
_ proof._each probability in can be written as a sum of a set of probabilities for complete hidden - variable states . in our scenario
there are four dichotomic observables , @xmath43 , @xmath45 , @xmath46 , and @xmath47 , and four trichotomic observables , @xmath48 , @xmath49 , @xmath44 , and @xmath50 .
the proof is a bookkeeping exercise showing that each term that adds to the positive term @xmath51 also appears ( with the right multiplicity ) at some of the negative terms , implying @xmath52 .
a detailed calculation is given in the supplemental material [ suppmat ] .
notice that , despite the fact that the inequality is inspired by the ks diagram applied to a pair of three - level systems in a singlet state , as will be made clear by eqs . , it is independent of it in the sense that the proof is not based on a diagram , and even a computer programmed to search for realistic inequalities could have found it .
it is also important to stress that inequality also holds for any local theory .
this means that , when the choice of measurement in one of the particles is spacelike separated from the result of the measurement on the other particle , then is a ks inspired bell inequality . in translating the ks basic diagram into four realistic questions , exclusiveness of each triangle
was used , but the noncontextual assumption deserves more . for each part , we also have the following logical implications : [ compat ] @xmath53 a noncontextual assignment is given by answers to the four questions following the conditions . only @xmath54 out of the @xmath55 realistic states are noncontextual . with fewer allowed states , more restrictions can be obtained .
one can now prove the following . _ lemma 2._consider @xmath33 and @xmath35 as in lemma 1 .
moreover , consider that these questions obey , for each party , conditions .
the following inequality holds : @xmath56 where @xmath57 is used to remind that only noncontextual hidden - variable states are taken into account . _
proof._again , it is simple bookkeeping showing that all terms which contribute to the positive term also appear at ( at least one ) negative ones .
details and other inequalities are provided in the supplemental material @xcite .
inequalities are tight , with respect to noncontextual states , and can be violated by realistic states which do not obey some of the relations . in this sense ,
even with an experiment prepared with spacelike separations between @xmath31 and @xmath32 , they should not be considered as bell inequalities .
their violation rules out a different set of hidden - variable models : noncontextual ones . from the technical point of view , that is a great advantage , since an experimentalist need not bother with guaranteeing spatial - like separations . with the same kind of proof one can deduce the following .
_ lemma 3._with the same conditions of lemma 2 , @xmath58 and also other similar expressions .
again , is tight with respect to noncontextual states , and can be violated with realistic contextual ones . _
quantum violation._the same quantum system will violate all inequalities shown here .
it is just a two particle generalization of the previous described state - dependent proof of impossibility of noncontextual models for quantum mechanics .
consider a two - qutrit system prepared in the state @xmath59 and the observables [ qmoperators ] @xmath60 with @xmath23 , @xmath25 , @xmath61 , @xmath62 , and @xmath63 as defined in eqs , , , and .
then the predictions of qm are @xmath64 where @xmath65 and @xmath66 , and the remaining probabilities in are null .
therefore , according to qm , @xmath67 violating inequality , and inequalities and read @xmath68 _ discussion._inequality was deduced using a translation of the basic ks diagram into a set of four propositions . after this translation
is done , its validity rests only on hidden - variable hypothesis of a preexisting reality which is reveled by measurements .
inequalities and are based on these four questions and on the constraints given by noncontextual hypothesis . because the lemmas were written and proved
, there is no necessity of going back to the diagrams , despite the fact that their essence is captured by those relations . in all these inequalities ,
each term involves pairs of compatible questions , asked for different parts of the composed system .
moreover , being a quantitative statement , the argument is not sensible to finite precision problems , given that precision is enough to make violations meaningful . in this sense
those are proposals for quantum violations of noncontextuality free from compatibility and finite precision loopholes . testing the violation of inequality is very demanding experimentally .
it requires testing @xmath69 joint probabilities and , assuming that one obtains @xmath70 and @xmath71 for the other @xmath72 probabilities , one must have @xmath73 in order to observe a violation of inequality .
the importance of this result is that it explicitly shows that there is no fundamental obstacle observing kochen - specker contextuality on ( pairs of ) qutrits .
no additional assumptions are needed to deal with the fact that measurements have a finite precision and compatibility between sequential measurements is not perfect .
inequalities are less demanding experimentally . indeed
, they can be translated as conditional probabilities giving inequalities like @xmath74 which gives larger violations , since qm predicts @xmath75 , allowing for @xmath76 .
another interesting point is that inequality only involves one question and one answer in part @xmath32 : @xmath77 , and an experiment would only involve three possible measurements in part @xmath31 , with coincidence countings of each of these three with one and the same condition in @xmath32 .
it is also stressed in the supplementary material @xcite how much this situation resembles the first example discussed in this letter . in this letter
we have established experimentally testable contextual inequalities , based on the building block of kochen - specker demonstration of the impossibility of assigning noncontextual answers to all possible predictions of quantum mechanics .
we hope that this proposal will definitely close the debate on the physical relevance and experimental testability of the kochen - specker theorem , and induce a new generation of loophole - free experiments .
this work was stimulated by a discussion with a. zeilinger .
the final work benefited from a discussion with d. cavalcanti .
a.c .
acknowledges support from capes and icex ( belo horizonte , brazil ) , the spanish micinn project no . fis2008 - 05596 , and the wenner - gren foundation . m.t.c . acknowledges support from brazilian cnpq , fapemig , and inct - iq . 99 s. kochen and e. p. specker , j. math . mech .
* 17 * , 59 ( 1967 ) .
j. s. bell , physics ( long island city , n.y . ) * 1 * , 195 ( 1964 ) .
m. kleinmann _
et al . _ , .
k. horodecki _
et al . _ , .
d. a. meyer , phys .
rev .
lett . * 83 * , 3751 ( 1999 ) .
a. kent , phys .
rev .
lett . * 83 * , 3755 ( 1999 ) .
r. clifton and a. kent , proc .
r. soc .
london , ser .
a * 456 * , 2101 ( 2000 ) .
j. barrett and a. kent , stud .
hist .
philos .
mod .
phys . * 35 * , 151 ( 2004 ) .
n. d. mermin , .
a. cabello , phys .
rev .
a * 65 * , 052101 ( 2002 ) .
g. kirchmair _
et al . _ , nature ( london ) * 460 * , 494 ( 2009 ) .
h. bartosik _
et al . _ ,
phys .
rev .
lett . *
103 * , 040403 ( 2009 ) .
e. amselem _
et al . _ ,
phys .
rev .
lett .
* 103 * , 160405 ( 2009 )
. b. h. liu _
et al . _ ,
phys .
rev .
a * 80 * , 044101 ( 2009 ) .
o. moussa _
et al . _ ,
phys .
rev .
lett . * 104 * , 160501 ( 2010 ) .
o. ghne _
et al . _ ,
phys .
rev .
a * 81 * , 022121 ( 2010 ) .
a. stairs , phil .
sci . * 50 * , 578 ( 1983 ) .
r. clifton , am .
j. phys .
* 61 * , 443 ( 1993 ) ; h. bechmann johansen , am .
j. phys .
* 62 * , 471 ( 1994 ) ; p. e. vermaas , am . j. phys .
* 62 * , 658 ( 1994 ) .
s. kochen and e. p. specker , in _
symposium on the theory of models _ , edited by j. w. addison _ et al .
_ , ( north - holland , amsterdam , holland , 1965 ) , p. 177 . a. cabello , phys . rev
. lett .
* 101 * , 210401 ( 2008 ) .
p. badziag _
et al . _ ,
phys .
rev .
lett . * 103 * , 050401 ( 2009 ) .
s. kochen , to a. shimony ( private communication ) , reported in @xcite , p. 581 , and in @xcite , p. 481 .
p. heywood and m. l. g. redhead , found .
phys .
* 13 * , 481 ( 1983 ) .
a. cabello , phys .
rev .
lett . * 104 * , 220401 ( 2010 ) .
l. aolita , r. gallego , a. acn , a. chiuri , g. vallone , p. mataloni , and a. cabello ( to be published ) .
see supplemental material below for the proofs of the lemmas , some generalizations , and technical discussion of the results . |
the problem of construction of such multidimensional nonlinear partial differential equations ( pdes ) which are either completely integrable or , at least , possess a big manifold of particular solutions is very attractive problem of integrability theory .
first of all , one should emphasize several remarkable classical works regarding the completely integrable models such as @xcite ( where the korteweg - de vries equation ( kdv ) has been first time studied ) , @xcite ( where the dressing method for a big class of soliton and instanton equations has been formulated ) , @xcite ( where the so - called sato approach to the integrability is applied to kadomtsev - petviashvili equation ( kp ) ) .
a big class of the first order systems of quasilinear pdes is integrated in @xcite using the generalized hodograph method .
however , most of the above integrable nonlinear pdes are ( 2 + 1)- and/or ( 1 + 1)-dimensional , excepting the selfdual - type pdes @xcite and the equations assotiated with commuting vector fields @xcite . in this paper
we use modification of the algorithm represented in @xcite allowing one to construct new multidimensional partially integrable nonlinear pdes .
it is shown in @xcite that frobenious reduction of the matrix fields of the nonlinear pdes integrable either by the hoph - cole substitution @xcite ( @xmath0-integrable pde @xcite ) or by the method of characteristics @xcite ( @xmath1-integrable pde ) leads to one of two big classes of the nonlinear pdes integrable by the inverse spectral transform method ( istm ) @xcite ( @xmath2-integrable pdes @xcite ) .
these classes are ( @xmath3 ) soliton equations , such as kdv @xcite , the nonlinear schrdinger equation ( nls ) @xcite , the kadomtsev - petviashvili equation ( kp ) @xcite , the deavi - stewartson equation ( ds ) @xcite , and ( @xmath4 ) instanton equations , such as the self - dual yang - mills equation ( sdym ) @xcite .
the natural question is whether the frobenius reduction ( or its modification ) can be used for construction new types of integrable ( or at least partially integrable ) systems starting with any known integrable system or this method works only for derivation of soliton and instanton pdes from @xmath0- and @xmath1-integrable ones ? at first glance the answer is negative .
in fact , one can verify that frobenius reduction applied to such matrix @xmath2-integrable pde as the gl(@xmath5 ) sdym , the @xmath5-wave equation and the kp produces the same @xmath2-integrable pde .
however , there is a method to generate new higher dimensional partially integrable systems of nonlinear pdes using frobenius type reduction after the appropriate differential reduction imposed on the matrix fields of the above @xmath2-integrable nonlinear pdes .
such combination of reductions has been already used in @xcite .
it is shown there that gl(@xmath5 ) sdym supplemented by the pair of reductions , namely , differential reduction relating certain blocks of the matrix field and frobenius reduction of these blocks , produces a new five - dimensional system of matrix nonlinear pdes ( with three - dimensional solution space ) , whose scalar reduction results in the nonlinear pde assotiated with commuting vector fields @xcite . following the strategy of ref.@xcite
, we consider an algorithm for construction new partially integrable pdes starting with the matrix kp ( although this algorithm may be applied to any @xmath2-integrable model ) .
we will derive two representative of matrix systems whose scalar reductions yield the following four - dimensional equations : @xmath6 another reduction , @xmath7 reduces these pdes into the following three - dimensional ones @xmath8 here we take @xmath9 and @xmath10 as evolutionary parameters ( times ) while @xmath11 , @xmath12 and @xmath13 are taken as space parameters . since the linear parts of eqs.([nl_ex2 ] ) and ( [ ex_3dim2 ] ) coincide with the linear parts of the kp and the kdv respectively , eqs.([nl_ex2 ] ) and ( [ ex_3dim2 ] ) may be treated as new variants of kp- and kdv - type equations respectively .
the feature of these equations is that the derivative with respect to @xmath12 appears only in the nonlinear parts .
for this reason , these equations may be referred to as dispersionless ones .
remember , that the kp originates from the ( 1 + 1)-dimensional @xmath0-integrable brgers hierarchy due to the frobenius reduction @xcite .
thus the complete set of transformations leading to eqs.([nl_ex1 ] ) and ( [ nl_ex2 ] ) is following ( see also fig.1 ) .
we start with the @xmath0-integrable brgers hierarchy of nonlinear pdes with independent variables @xmath11 and @xmath14 , @xmath15 . frobenius reduction of this hierarchy @xcite yields the proper hierarchy of discrete chains of nonlinear pdes , which is equivalent to the chains obtained in the sato approach to the integrability of ( 2 + 1)-dimensional kp @xcite .
the later may be derived as an intermediate result of our algorithm after eliminating all extra fields using combination of the first and the second representatives of the constructed discrete hierarchy .
next , apply the differential reduction introducing one more independent variable @xmath12 .
this step yields a three - dimensional system of nonlinear pdes , i.e. dimensionality coincides with that of kp .
finally , the frobenius type reduction applied to the matrix fields of the latter nonlinear pdes results in a four - dimensional system of matrix pdes whose scalar versions yield eqs.([nl_ex1],[nl_ex2 ] ) .
it will be shown that the derived four - dimensional nonlinear pdes may not be completely integrated by our method because the available solution spaces to them are restricted .
fig.1 the chain of transformations from the ( 1 + 1)-dimensional @xmath0-integrable brgers hierarchy to the four - dimensional pdes ( [ nl_ex1 ] ) and ( [ nl_ex2 ] ) . here
@xmath16 , @xmath17 are @xmath18 matrix fields , @xmath19 , @xmath20 , @xmath21 and @xmath22 are @xmath23 blocks , @xmath24 and @xmath25 are @xmath26- and @xmath27-dimensional identity matrices respectively . blocks @xmath21 and @xmath22 are defined by eqs.([qp_i ] ) . here
@xmath28 , @xmath29 and @xmath27 are arbitrary positive integer parameters .
this paper is organized as follows . in sec.[section : kp ] we briefly recall some results of @xcite and derive the discrete chains of nonlinear pdes produced by the ( 1 + 1)-dimensional brgers hierarchy ( with independent variables @xmath11 and
@xmath14 , @xmath30 ) supplemented by the frobenius reduction . using a few equations of theses chains we eliminate all extra fields and derive the matrix kp . in sec.[section : dr ] we suggest the special type differential reduction imposed on the blocks of the @xmath31 matrix fields of the above chains .
this reduction introduces a new independent variables @xmath12 and @xmath9 and allows one to generate @xmath32 three - dimensional matrix pdes .
frobenius type reduction of the above pdes results in the four - dimensional systems of @xmath33 matrix pdes , sec.[section : ftr ] .
scalar versions of these pdes result in eqs.([nl_ex1 ] ) and ( [ nl_ex2 ] ) which , in particular , may be reduced to eqs.([ex_3dim1 ] ) and ( [ ex_3dim2 ] ) respectively .
solution spaces to the nonlinear pdes derived in secs.[section : dr ] and [ section : ftr ] will be studied in sec.[section : solutions ] .
the obstacles to the complete integrability of eqs.([nl_ex1 ] ) , ( [ nl_ex2 ] ) , ( [ ex_3dim1 ] ) and ( [ ex_3dim2 ] ) as well as the obstacles to their integrability by the istm are briefly discussed in sec.[section : obstacle ] .
conclusions are given in sec.[section : conclusions ] .
here we use the brgers hierarchy as a simplest example of @xmath0-integrable hierarchies linearizable by the matrix hopf - cole substitution .
namely , let @xmath34 be solution of the following linear algebraic matrix equation : @xmath35 here @xmath36 and @xmath34 are @xmath37 matrix functions , @xmath38 are @xmath37 commuting constant matrices .
to anticipate , the matrices @xmath38 are introduced in order to establish the reductions eliminating derivatives with respect to @xmath14 from the nonlinear pdes , see , for example , reduction ( [ red ] ) .
the compatibility conditions of eqs .
( [ lin_chi ] ) and ( [ lin_t ] ) yield the linearizable brgers hierarchy .
we will need only the first and the second representatives of this hierarchy below , i.e. @xmath30 in eqs.([lin_t ] ) : @xmath39 introduce the frobenius reduction @xcite : @xmath40 where @xmath41 and @xmath42 are @xmath43 identity and zero matrices respectively , @xmath17 are @xmath44 matrix functions . substituting eq.([ww ] ) into eq.([chiw ] ) we obtain the following block structure of @xmath45 : @xmath46 where @xmath47 are @xmath44 matrix functions .
thus , eqs.([lin_chi ] ) and ( [ lin_t ] ) may be written as follows : @xmath48 where @xmath49 are some @xmath18 commuting constant matrices .
the meaning of the integer parameters @xmath29 and @xmath27 is clarified in fig.1 .
the compatibility conditions of eqs.([lin_kp ] ) and ( [ t_kp ] ) ( with @xmath30 ) yield the following discrete chains : @xmath50 alternatively , these chains may be derived substituting eq.([ww ] ) into eqs.([lin_t2 ] ) and ( [ lin_t3 ] ) .
the matrix kp is represented by the system of three equations involving fields @xmath17 , @xmath51 : eq.([burgers_ch ] ) with @xmath52 and eq.([burgers-3_ch ] ) with @xmath53 . eliminating @xmath54 and @xmath55 from this system one
gets the following nonlinear pde for the field @xmath56 : @xmath57-\frac{3}{4 } w^{(1)}_{t_2t_2}=0 , \end{aligned}\ ] ] where square parenthesis mean matrix commutator . in the scalar case this equation reduces to the following one , @xmath58 : @xmath59 which is the scalar potential kp
[ section : dr ] let matrices @xmath60 have the following block structure : @xmath61 where @xmath62 are @xmath32 matrix functions . we introduce @xmath12-dependence of @xmath62 by the following second order pde : @xmath63 here @xmath3 , @xmath64 and @xmath65 are @xmath32 diagonal constant matrices .
the block structure of @xmath60 ( [ dif_red ] ) suggests us the relevant block structure of @xmath66 : @xmath67 where @xmath68 , @xmath69 , @xmath70 and @xmath71 are @xmath72 matrix functions
. now matrix @xmath73 equations ( [ lin_kp ] ) may be written as two @xmath74 equations : @xmath75 where @xmath76 the compatibility condition of eqs.([b1 ] ) and ( [ b2 ] ) , @xmath77 yields the expressions for @xmath71 and @xmath70 in terms of @xmath69 and @xmath68 : @xmath78 thus only two blocks of @xmath66 are independent , i.e. @xmath68 and @xmath69 .
matrix equation ( [ g_r1 ] ) may be considered as the uniquely solvable system of @xmath79 linear @xmath80 matrix algebraic equations for the matrix functions @xmath19 and @xmath20 , @xmath81 , while eq.([b2 ] ) is the consequence of eq.([g_r1 ] ) .
eqs.([t_kp ] ) yield : @xmath82 where @xmath83 are @xmath32 commuting constant matrices .
eq.([x ] ) allows us to introduce one more set of parameters @xmath84 as follows : @xmath85 before proceeding further we select the following three equations out of system ( [ t],[tau ] ) : @xmath86 we also assume that @xmath3 , @xmath64 and @xmath65 are scalars , i.e. @xmath87 in order to derive the nonlinear pdes for @xmath19 and @xmath20 , we must consider the conditions providing the compatibility of eq.([b1 ] ) and eqs.([t2]-[t3 ] ) : @xmath88 these equations generate the following chains of nonlinear pdes for @xmath19 and @xmath20 , @xmath81 : @xmath89 @xmath90 @xmath91 in addition , we must to take into account the compatibility condition of eqs.([g_r2 ] ) and ( [ g_r1 ] ) , @xmath92 which gives us the following system of non - evolutionary nonlinear chains : @xmath93 one can show that the derived chains ( [ nl_ch_1]-[nl_ch_3],[nl4],[nl4_1 ] ) generate three compatible three - dimensional systems of nonlinear pdes as follows : @xmath94 we do not represent these equations explicitely since they are intermediate equations in our algorithm .
three systems ( [ dif_1]-[dif_3 ] ) represent three commuting flows with times @xmath13 , @xmath9 and @xmath10 respectively . in this section
we follow the strategy of ref.@xcite and show that hierarchy of nonlinear chains ( [ nl_ch_1]-[nl_ch_31],[nl4],[nl4_1 ] ) supplemented by the frobenius type reduction of the matrix fields @xmath19 and @xmath20 generates the hierarchy of four - dimensional systems of nonlinear pdes , see eqs.([nl_ch_1_fr_1_newvar]-[nl4_fr_1_newvar ] ) and text thereafter .
this is possible due to the remarkable property of chains of nonlinear pdes ( [ nl_ch_1]-[nl_ch_31],[nl4],[nl4_1 ] ) .
namely , these chains admit the following frobenious type reduction : @xmath95 where @xmath29 is an arbitrary positive integer parameter , @xmath96 and @xmath97 are arbitrary positive integer functions of positive integer argument , @xmath98 and @xmath99 are @xmath33 matrix fields . in particular , if @xmath100 , then this reduction becomes frobenius one @xcite , which is shown in fig.1 .
eqs.([g_f ] ) require the following block - structures of the functions @xmath101 and of the constant matrices @xmath102 : @xmath103 here @xmath104 are @xmath33 matrix functions and @xmath105 are @xmath33 commuting constant matrices .
in turn , eq.([b1 ] ) reduces to the following one : @xmath106,\;\;l= 1,\dots,2n_0,\;\;n , m=1,\dots , n_0,\end{aligned}\ ] ] while eqs.([t2]-[t3],[x ] ) yield @xmath107 then the chains of nonlinear pdes ( [ nl_ch_1]-[nl_ch_31],[nl4],[nl4_1 ] )
get the following block structures : @xmath108 in particular , nonlinear chains ( [ nl_ch_1],[nl_ch_11 ] ) and ( [ nl4],[nl4_1 ] ) with frobenious type reduction ( [ g_f ] ) generate the following chains with two discrete indexes : @xmath109 @xmath110 where @xmath111 , @xmath112 .
now our goal is to write the complete system of pdes for some fields @xmath113 and/or @xmath114 which would be independent on both parameters @xmath28 and @xmath29 ( in the spirit of the sato theory @xcite ) .
this may be done if , along with system ( [ nl_ch_1_fr]-[nl4_1_fr ] ) , one involves the discrete chains of pdes generated by either eqs.([nl_ch_2],[nl_ch_21 ] ) or eqs.([nl_ch_3],[nl_ch_31 ] ) together with the frobenious type reduction ( [ g_f ] ) .
for instance , system ( [ nl_ch_2],[nl_ch_21 ] ) yields @xmath115 the discrete chains generated by eqs .
( [ nl_ch_3 ] ) and ( [ nl_ch_31 ] ) are very cumbersome .
however , it will be shown that we need only equation ( [ nl_ch_3 ] ) with @xmath116 for derivation of the complete system of nonlinear pdes .
the later equation may be simplified after elimination @xmath117 , @xmath118 and @xmath119 using eq.([nl_ch_1 ] ) with @xmath120 and eq.([nl_ch_11 ] ) with @xmath116 .
this simplification is reasonable because , in any case , we are going to use chain ( [ nl_ch_1_fr ] ) ( produced by eqs.([nl_ch_1 ] ) ) in order to derive the complete system of nonlinear pdes .
one gets in result : @xmath121 after the frobenius reduction , this equation generates the following discrete chain : @xmath122 as a consequence , systems of nonlinear pdes ( [ dif_1]-[dif_3 ] ) generate the following compatible chains of nonlinear pdes , @xmath112 : @xmath123 having chains of equations ( [ nl_ch_1_fr]-[nl_ch_22_fr],[nl_ch_3_mod_fr ] ) we would like to exhibit the complete system of nonlinear pdes independent on both @xmath28 and @xmath29 taking a few equations out of these chains .
after some examination of these chains we conclude that it is enough to take the equations @xmath124 , @xmath125 , @xmath126 and @xmath127 . introducing the new fields @xmath128 we write these equations as follows : @xmath129 any two equations out of the system ( [ nl_ch_1_fr_1_newvar]-[nl_ch_3_mod_fr_1_newvar ] ) supplemented by eq.([nl4_fr_1_newvar ] ) represent the complete system of matrix nonlinear pdes for the fields @xmath130 , @xmath131 and @xmath132 .
two examples of the scalar nonlinear pdes are given below in sec.[section : examples ] .
[ [ example-1 . ] ] example 1 .
+ + + + + + + + + + consider the system ( [ nl_ch_1_fr_1_newvar],[nl_ch_2_fr_1_newvar],[nl4_fr_1_newvar ] ) .
in the scalar case this system reduces to the following single scalar nonlinear pde for the field @xmath130 : @xmath133 [ [ example-2 . ] ] example 2 .
+ + + + + + + + + + consider the system ( [ nl_ch_1_fr_1_newvar],[nl_ch_3_mod_fr_1_newvar],[nl4_fr_1_newvar ] ) . in the scalar case
this system reduces to the single equation for the field @xmath130 as follows : @xmath134
solutions to systems of matrix nonlinear pdes ( [ dif_1]-[dif_3 ] ) or , more general , to discrete chains ( [ nl_ch_1]-[nl_ch_31],[nl4],[nl4_1 ] ) are obtainable in terms of the functions @xmath62 @xmath135 taken as solutions of linear system ( [ x],[t2]-[t3 ] ) : @xmath136 where @xmath137 are the arbitrary scalar functions of single scalar variable , @xmath138 is the list of all independent variables of the nonlinear pdes , @xmath139 , @xmath120 , are roots of the characteristics equation assotiated with eq.([g_r2 ] ) where @xmath3 , @xmath64 and @xmath65 are given by eqs.([a_nu_mu ] ) : @xmath140 now , we can use eq.([b1 ] ) in order to find @xmath19 and @xmath20 .
in particular , there is a big class of solutions in the form of rational functions of exponents , such as solitary wave solutions .
we do not represent their explicite form since eqs.([dif_1]-[dif_3 ] ) are intermediate equations in our method of solving eqs.([nl_ex1],[nl_ex2 ] ) .
note nevertheless that the simplest nontrivial solution to eqs.([dif_1]-[dif_3 ] ) corresponds to @xmath141 .
then eq.([b1 ] ) reduces to the following four matrix equations : @xmath142 these equations , in general , are uniquely solvable for the matrix fields @xmath19 and @xmath20 , @xmath120 .
[ [ the - reductions - to - the - lower - dimensional - pdes . ] ] the reductions to the lower dimensional pdes .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the described solution space admits the following reductions @xmath143 where @xmath144 is either 2 or 3 .
in fact , these reductions mean @xmath145 in eq.([psi_sol0 ] ) .
for instance , let @xmath146 . then we must substitute the following expression into eq.([psi_sol0 ] ) : @xmath147 where @xmath148 , @xmath125 , are arbitrary constants . in this case
the fields @xmath113 and @xmath149 are solutions to linear algebraic system ( [ psi_ex ] ) .
more precisely , only part of system ( [ psi_ex ] ) is needed to define @xmath113 and @xmath149 .
in fact , system ( [ psi_ex ] ) may be viewed as two subsystems .
the first one corresponds to @xmath53 : @xmath150,\\\nonumber & & l=1,\dots,2 n_0\;\;,\;\;m=1,\dots , n_0.\end{aligned}\ ] ] this is the system of @xmath151 linear algebraic @xmath33 matrix equations for the same number of the matrix fields @xmath98 and @xmath99 , @xmath81 , @xmath152 .
namely eqs.([psi1 ] ) yield the functions @xmath153 , @xmath154 , @xmath155 , @xmath156 , @xmath157 , @xmath158 , @xmath159 defining the solutions @xmath130 , @xmath131 , @xmath132 to system ( [ nl_ch_1_fr_1_newvar]-[nl4_fr_1_newvar ] ) in accordance with formulae ( [ newvar ] ) .
the second subsystem corresponds to @xmath160 in ( [ psi_ex ] ) : @xmath161,\\\nonumber & & l=1,\dots,2 n_0,\;\;n , m=1,\dots , n_0,\;\;\ ; \psi^{(l;ij)}=0,\ ; { \mbox{if}},\;i\le 0.\end{aligned}\ ] ] this subsystem expresses recursively the functions @xmath104 , @xmath160 , in terms of the functions @xmath162 .
the simplest case corresponds to @xmath163 , @xmath164 ( the frobenius reduction ) .
the functions @xmath104 are solutions to system ( [ t2_ex]-[x_ex ] ) and may be written as follows : @xmath165 here @xmath166 are arbitrary scalar functions of one scalar argument . in particular , the derived formulae describe the big class of solutions having the form of rational functions of exponents , such as solitary wave solutions .
the simplest examples of them will be given below , see eqs.([solution]-[kk],[simplest ] ) .
in the scalar case ( corresponding to eqs.([nl_ex1 ] ) and ( [ nl_ex2 ] ) ) one has @xmath167 so that eq.([psi_sol0_fr ] ) reads : @xmath168 with the same expressions for @xmath139 given by eqs.([k ] ) .
[ [ the - reductions - to - the - lower - dimensional - pdes.-1 ] ] the reductions to the lower dimensional pdes .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + note that the described solution space admits reduction ( [ red ] ) embedded into the following set of reductions @xmath169 where @xmath144 is either 2 or 3 .
in fact , similar to reduction ( [ red_sol_1 ] ) , these reductions mean @xmath170 in eq.([psi_sol0_fr ] ) .
for instance , let @xmath146 which corresponds to reduction ( [ red ] ) .
then we must substitute the following expression into eq.([psi_sol0_fr ] ) : @xmath171 where @xmath172 ( @xmath125 ) are arbitrary constants . in the scalar case ( corresponding to nonlinear equations ( [ ex_3dim1 ] ) and ( [ ex_3dim2 ] ) ) eq.([psi_red_fr_scalar ] )
reads @xmath173 which must be substituted into eq.([psi_sol0_fr_scalar ] ) .
[ [ the-simplest-solution-to-eqs.nl_ex1-and-nl_ex2 . ] ] the simplest solution to eqs.([nl_ex1 ] ) and ( [ nl_ex2 ] ) .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the simplest nontrivial solution to eqs.([nl_ex1 ] ) and ( [ nl_ex2 ] ) corresponds to @xmath174 and @xmath175 , @xmath30
. then eq.([psi_sol0_fr_scalar ] ) reads @xmath176 eqs.([psi1 ] ) become the system of two following equations : @xmath177 their solution reads : @xmath178 formulae ( [ solution],[solution_b ] ) have four arbitrary functions of single variable @xmath179 , @xmath180 .
let us write the explicite formulae for the particular case @xmath181 , where both @xmath182 and @xmath183 are arbitrary constants : @xmath184 the solution @xmath130 has no singularities if @xmath185 , which is true if , for instance , the following relations are valid : @xmath186 for example , let @xmath187 then @xmath188 which is the kink .
[ [ the-simplest-solution-to-eqs.ex_3dim1-and-ex_3dim2 . ] ] the simplest solution to eqs.([ex_3dim1 ] ) and ( [ ex_3dim2 ] ) .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we take eq.([psi_sol0_fr_scalar ] ) with @xmath174 and @xmath189 : @xmath190 note that eqs.([ex_3dim1 ] ) and ( [ ex_3dim2 ] ) require reduction ( [ red ] ) , which means that function @xmath179 in eq.([psi_sol0_scalar_simple_red ] ) is given by eq.([psi_red_scalar ] ) . for instance , let @xmath191 then one has expressions @xmath192 instead of expressions ( [ expl_delta]-[expl_q ] ) , while relations ( [ kk ] ) remain the same .
the conditions similar to ( [ nonsing ] ) with replacement @xmath193 guarantee that the solution @xmath130 is non - singular in this case as well .
let , for example , @xmath194 , @xmath195 , @xmath196 and @xmath197 .
then one has @xmath198 which is the kink .
it is important to note that our algorithm does not describe the full solution spaces to nonlinear pdes ( [ nl_ex1_sc ] ) , ( [ nl_ex2 ] ) , ( [ ex_3dim1 ] ) and ( [ ex_3dim2 ] ) . to justify this statement we consider eqs.([nl_ex1_sc ] ) and
( [ nl_ex2 ] ) , while eqs.([ex_3dim1 ] ) and ( [ ex_3dim2 ] ) may be treated in a similar way .
the simplest argument is following . by construction ,
eqs.([nl_ex1_sc ] ) and ( [ nl_ex2 ] ) must be commuting flows , i.e. the expression @xmath199_{t_3 } -[{\mbox{eq.(\ref{nl_ex2})}}]_{y\tau}\end{aligned}\ ] ] must be identical to zero .
however , the direct calculation shows that this expression yields the nonlocal three - dimensional pde having rather complicated form ( we do not represent it here ) .
this means that the constructed solution space to the equations ( [ nl_ex1_sc ] ) and ( [ nl_ex2 ] ) is two - dimensional ( in other words , it may depend on the arbitrary functions of two independent variables ) while the full solution space to four - dimensional pdes ( [ nl_ex1_sc ] ) and ( [ nl_ex2 ] ) must be three dimensional .
the dimensionality of the solution space is confirmed in sec.[section : solutions ] by formulae ( [ psi_sol0_fr_scalar ] ) , which shows us that the solution space depends on @xmath200 arbitrary functions of single variable @xmath201 , @xmath202 , @xmath203 , @xmath120 .
since both @xmath28 and @xmath29 are arbitrary positive integers and may go to infinity , these functions may approximate arbitrary functions of two variables in the solution space .
thus , the possibility to have an arbitrary function of three independent variables in the solution space ( which would provide the full integrability ) remains an open problem for further study . without details ,
we remark that solution space to four - dimensional eq.([nl_ex1 ] ) ( available due to our algorithm ) is two - dimensional as well , while solution spaces to three - dimensional equations ( [ ex_3dim1 ] ) and ( [ ex_3dim2 ] ) are one - dimensional , which may be obtained owing to the formulae ( [ psi_sol0_fr_scalar ] ) and ( [ psi_red_scalar ] ) . since eqs.([nl_ex1_sc ] ) and ( [ nl_ex2 ] ) have been derived from the matrix kp and the later is well known ( 2 + 1)-dimensional pde integrable by the istm , one can expect the same type of integrability of eqs.([nl_ex1_sc ] ) and ( [ nl_ex2 ] ) .
in particular , one can expect that these equations are compatibility conditions of some overdetermined linear spectral problem derivable from the overdetermined linear spectral problem for the matrix kp .
however , this is not true . in the next subsection
we consider eq.([nl_ex2 ] ) as an example and show that the linear spectral problem for this equation is not well defined .
thus , at the moment , the istm is not a suitable tool for solving eq.([nl_ex2 ] ) .
the same conclusion is valid for eq.([nl_ex1_sc ] ) , ( [ ex_3dim1 ] ) and ( [ ex_3dim2 ] ) as well .
although the algorithm described above allows one to find a big manifold of solutions to eqs.([nl_ex1_sc ] ) , ( [ nl_ex2 ] ) , ( [ ex_3dim1 ] ) and ( [ ex_3dim2 ] ) it does not gives us an algorithm to derive the linear spectral problem for any of these equations , i.e. , at the moment , we are not able to obtain such overdetermined system of linear equations for some spectral function whose compatibility condition results in eq.([nl_ex1_sc ] ) , or eq.([nl_ex2 ] ) , or eq.([ex_3dim1 ] ) , or eq.([ex_3dim2 ] ) without additional requirements to the coefficients of linear system . as an example explaining this problem , we consider eq.([nl_ex2 ] ) as nonlinear equation derivable from the matrix kp ( [ matrixkp ] ) after differential reduction ( introduced in sec.[section : diff_reduction ] ) followed by the frobenious type reduction ( introduced in sec.[section : ftr ] ) in this subsection . the obstacle to derive the linear spectral problem for eq.([nl_ex2 ] ) is assotiated with the spectral representation of differential reduction ( [ ww_i],[qp_i ] ) which is a reduction of the potentials of the linear spectral problem for the matrix kp . in other words , we do not know which reduction must be imposed on the spectral function to generate the above differential reduction for the potentials of the spectral problem .
thus the problem of spectral representation of the eq.([nl_ex2 ] ) remains open as well as the problem of its complete integrability .
nevertheless , we derive some linear spectral problem whose compatibility condition leads to eq.([nl_ex2 ] ) imposing the above differential reduction `` by hand '' . the spectral problem for matrix
kp ( [ matrixkp ] ) reads @xmath204 where @xmath205 is a spectral function and @xmath206 is a spectral parameter .
differential reduction ( [ ww_i],[qp_i ] ) requires the following block structure of @xmath205 : @xmath207 where @xmath208 are @xmath209 matrices .
so that equations ( [ sp_p_kp ] ) acquire the following block structure : @xmath210 where @xmath211 are @xmath212 matrix equations . the spectral problem is now represented by the first - row blocks of eq.([f_red ] ) : @xmath213 while the second - row blocks of eq.([f_red ] ) are equivalent to the system ( [ lin_dif_1]-[lin_dif_4 ] ) up to the replacement of the spectral functions @xmath214 , @xmath125 .
emphasize that the differential reduction is already imposed on the coefficients of the above linear system due to the special form of the functions @xmath215 and @xmath216 given by eq.([qp_i ] ) .
it may be shown by the direct calculations that the compatibility conditions of eqs.([lin_dif_1]-[lin_dif_4 ] ) yield the system of nonlinear pdes which is equivalent to system ( [ dif_1 ] ) , ( [ dif_3 ] ) , i.e. we derive two complete compatible systems of nonlinear pdes so that eqs.([dif_1 ] ) describe the @xmath13-evolution while eqs.([dif_3 ] ) describe the @xmath10-evolution of fields . however , since differential reduction has not spectral representation , these nonlinear systems may not be integrated by the istm .
of course , this means that the eq.([nl_ex2 ] ) ( which has been derived from the above system of nonlinear pdes after the frobenious type reduction ( [ g_f ] ) ) may not be integrated by the istm as well .
however , for the sake of completeness , we consider the frobenious type reduction ( [ g_f ] ) imposed on the linear system ( [ lin_dif_1]-[lin_dif_4 ] ) .
first of all , note that this reduction requires the following structure of the functions @xmath217 and @xmath218 : @xmath219 where @xmath220 are @xmath33 functions .
eqs.([lin_dif_1]-[lin_dif_4 ] ) acquire the following block structure : @xmath221 where @xmath222 are @xmath33 matrix equations . similarly to eq.([f_red ] ) , only the first - row blocks of these equations represent the system of independent spectral equations : @xmath223 where @xmath224 and @xmath225 here we take into account eq.([a_nu_mu ] ) and take @xmath226 without loss of generality . the system of linear pdes ( [ lin_dif_fr_1]-[lin_dif_fr_4 ] ) , @xmath52 represents the overdetermined system whose compatibility condition results in the complete system of nonlinear pdes which includes equations ( [ nl_ch_1_fr_1_newvar ] ) , ( [ nl_ch_3_mod_fr_1_newvar ] ) and ( [ nl4_fr_1_newvar ] ) as a complete subsystem , whose scalar version ( @xmath167 ) results in eq.([nl_ex2 ] ) .
we have constructed two examples of the four - dimensional nonlinear pdes starting with the dressing method for the ( 1 + 1)-dimensional @xmath0-integrable brgers hierarchy and using combination of the frobenius type and the differential reductions .
one of these examples , eq.([nl_ex2 ] ) , has the same dispersion relation as the kp does and may be referred to as the kp type equation . as a consequence , its lower dimensional reduction , eq.([ex_3dim2 ] ) , has the same dispersion relation as the kdv does and may be referred to as the kdv type equation .
although the derived four - dimensional pdes are not completely integrable by our technique ( see sec.[section : obstacle ] ) , we are able to supply a big solution manifold to them with solitary wave solutions as most simple examples . the new feature of this algorithm in comparison with one represented in @xcite is the differential reduction which introduces the new independent variable @xmath12 into the nonlinear pdes . in turn
, this variable allows one to introduce the set of new @xmath9-variables by formula ( [ tau ] ) , which is a new method of increasing the dimensionality of solvable nonlinear pdes . an important problem is to overcome the obstacle to the complete integrability of the derived four - dimensional nonlinear pdes ( [ nl_ex1],[nl_ex2 ] ) and their three dimensional reductions ( [ ex_3dim1],[ex_3dim2 ] ) .
it is also important to find the physical application of the derived nonlinear pdes .
regarding this problem we must note that the additional independent variable @xmath12 appears only in the nonlinear terms of eqs.([nl_ex2 ] ) and ( [ ex_3dim2 ] ) , which usually means the existence of solutions with break of wave profiles .
so that one can expect that these equations describe the break of wave profiles in the physical systems where the kp and the kdv appear . | we represent an algorithm allowing one to construct new classes of partially integrable multidimensional nonlinear partial differential equations ( pdes ) starting with the special type of solutions to the ( 1 + 1)-dimensional hierarchy of nonlinear pdes linearizable by the matrix hopf - cole substitution ( the brgers hierarchy ) .
we derive examples of four - dimensional nonlinear matrix pdes together with they scalar and three - dimensional reductions .
variants of the kadomtsev - petviashvili type and korteweg - de vries type equations are represented among them .
our algorithm is based on the combination of two frobenius type reductions and special differential reduction imposed on the matrix fields of integrable pdes .
it is shown that the derived four - dimensional nonlinear pdes admit arbitrary functions of two variables in their solution spaces which clarifies the integrability degree of these pdes .
* differential reductions of the kadomtsev - petviashvili equation and associated higher dimensional nonlinear pdes . *
a. i. zenchuk _ institute of chemical physics ras , acad .
semenov av .
, 1 chernogolovka , moscow region 142432 , russia _ e - mail : [email protected] |
improvements in sensitivity of sub - mm interferometers over the past decade have enabled studies of the molecular gas content at galactic scales to reach very high redshifts . while the first measurements typically targeted highly luminous starbursts ( sbs ) or quasars ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , we are now able to build samples of normal , main sequence ( ms ) galaxies with _ integrated _ measurements of their co content out to redshift @xmath02.5 @xcite .
such studies have demonstrated that ms galaxies at z@xmath42 exhibit large molecular gas fractions ( @xmath550% of the baryonic mass , * ? ? ?
* ; * ? ? ?
* ) , which can drive disk fragmentation and generate the giant star - forming clumps @xcite that characterise the complex morphologies often observed at these redshifts @xcite .
_ resolved _ studies of the molecular gas at high redshift are still in their infancy and , although there have been few attempts to probe normal galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ) , current works have typically focused on the starburst tail of the galaxy population due to observational challenges @xcite .
molecular gas observations of clumpy , ms galaxies at scales comparable to the giant star - forming clumps ( stellar masses of @xmath6m@xmath7 and sizes of 0.1 - 1kpc ) , in particular , could offer important clues on the physical processes governing galaxy evolution .
migration of the clumps toward the galaxy centre and the torques they generate could play an important role in regulating bulge formation , black hole feeding and quenching of star formation in high redshift galaxies @xcite .
clump lifetimes and consequently the impact of clump - driven dynamical processes are however determined by the interplay of stellar feedback , gas accretion from the surrounding disks and efficiency of star formation @xcite .
strong radiative feedback and stellar winds could disrupt the clumps before they can reach the galaxy centre and significantly contribute to black - hole or bulge growth ( e.g. * ? ? ?
* ; * ? ? ?
powerful outflows are observed in high - z clumps @xcite , but theoretical works have shown that clumps can survive these outflows if they are able to re - create gas from the surrounding disk @xcite .
our observational knowledge of the molecular gas reservoirs in high - z clumps and how these compare with their host galaxies is , however , still limited .
in particular , it is not known whether the star formation efficiency in these giant clumps is similar to that seen in normal galaxy falling on the schmidt - kennicutt relation @xcite , or instead they display shorter time scales for molecular gas consumption .
indirect evidence presented in @xcite for a @xmath810myr old clump suggests a high star - formation efficiency , similar to that observed in starburst galaxies / starburst nuclei , during the early stages of clump formation .
other studies find depletion time scales of the order of @xmath9gyr for unresolved ensembles of clumps @xcite , i.e. , similar efficiencies as those observed in normal local galaxies and galactic regions @xcite .
more observations are hence necessary to discriminate among different scenarios . among the tracers of the different phases of molecular hydrogen ( @xmath10 ) ,
the co(5 - 4 ) transition is particularly advantageous for studying star formation occurring in a dense , clumpy medium .
excitation of the high-@xmath11 co transitions requires @xmath10 number densities of order 10@xmath12@xmath13 @xcite , which are much more similar to the densities reached in star - forming cores of molecular clouds than the critical densities of lower @xmath11 transitions @xcite . a single
, linear correlation between line luminosity , @xmath3 , and total infrared ( 8 - 1000@xmath14 m ) luminosity , @xmath15 , holds for normal and starburst galaxies both locally and at high redshift ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this has been interpreted as evidence that the co(5 - 4 ) line is a good tracer of the gas actually undergoing star - formation , irrespectively of the host galaxy properties . moreover
, the numerical simulations of @xcite demonstrated that giant star - forming clumps could be responsible for highly excited co spectral line energy distributions ( sleds ) as observed in ms , high - redshift galaxies , suggesting that this transition is indeed a good tracer of dense gas within the giant clumps .
further evidence for this was presented by @xcite who attempted a decomposition of the co sled of a high redshift galaxy into the emission from a diffuse component , and emission originating from a clump - dominated region and found the latter to display higher excitation , with a sled plausibly peaking at the co(5 - 4 ) transition .
finally , the co(5 - 4 ) line is much brighter and hence more easily observable at high redshift than other tracers that produce a linear correlation with star formation rate , e.g. , hcn @xcite . in this paper
we present @xmath00.5@xmath1 atacama large millimeter / submillimeter array ( alma ) observations of the co(5 - 4 ) transition in a [email protected] ms galaxy in the hubble ultra deep field ( hudf ) area @xcite .
the target galaxy , udf6462 @xcite , is a clumpy disk galaxy , enabling us to study the molecular content during this crucial phase for galaxy formation . with more than 4h integration time
, our observations reach a continuum sensitivity that is roughly a factor of 2 - 3 times deeper than other studies of the hudf covering a similar wavelength range ( but extend over a much larger area , e.g. , * ? ? ?
* ; * ? ? ?
the paper is organised as follows : in section [ sec : ancillarydata ] we describe the properties and available datatsets for the clumpy disk galaxy udf6462 .
clump identification and properties are discussed in section [ sec : clumpdescription ] .
our new alma observations are presented in section [ sec : almadata ] and the results obtained from these data discussed in section [ sec : results ] .
we summarise our findings in section [ sec : summary ] . throughout this paper
we use magnitudes in the ab system @xcite and a chabrier initial mass function ( imf , * ? ? ?
when necessary , measurements for comparison samples are converted accordingly . our definition of stellar mass correspond to the actual mass in stars , i.e.
, it is corrected for the fraction that is returned to the ism by evolved stars and differs from the integral of the star formation history , sometimes used in the literature .
we adopt a planck 2015 cosmology ( @xmath16=0.31 , @xmath17+@xmath16=1 and @xmath18=67.7kms@xmath19mpc@xmath19 , @xcite ) .
for our analyses we use a number of ancillary multi wavelength datasets available for udf6462 , including hubble space telescope ( hst ) imaging at @xmath0kpc resolution , infrared ( ir ) observations from the spitzer and herschel satellites and radio maps obtained with the karl g. jansky very large array ( vla ) .
we describe in detail these data in the following section .
the main physical properties for udf6462 are provided in table [ tab : udf6462prop ]
. m6 cm r r.a .
( deg , j2000 ) & 53.18172 + & @xmath2053.18176 + dec .
( deg , j2000 ) & -27.78298 + & @xmath20 - 27.78308 + @xmath21 & 1.57 + @xmath22 & [email protected] + l@xmath24(@xmath25 ) & 33@xmath239 + sfr@xmath24(@xmath26/yr ) & 33@xmath239 + sfr@xmath27(@xmath26/yr ) & 43@xmath237 + sfr@xmath28(@xmath26/yr ) & [email protected] + @xmath29(jykms@xmath19 ) & [email protected]
+ @xmath3(@xmath30 kkms@xmath19pc@xmath31 ) & [email protected] + @xmath32(@xmath14jy ) & [email protected] + m@xmath33(@xmath34 ) & 2.0@xmath35 + m@xmath36(@xmath34 ) & [email protected] + [ tab : udf6462prop ] in selecting our target we aimed to observe a typical , clumpy high redshift galaxy while maximising the amount of ancillary data supporting the interpretation of the alma observations .
udf6462 is located within the hudf area and hence has been observed by several spectroscopic and photometric surveys .
moreover , its morphological and stellar population properties have also been investigated by a number of other studies in the the literature @xcite .
spectroscopic redshifts for udf6462 were measured from the 3d - hst grism data @xcite , from the fors2 follow - up of the great observatories origins deep survey south ( goods - s ) field described in @xcite and from the sinfoni observations of @xcite .
all measurements are consistent with a @xmath21=1.570 . extremely deep hst observations in the @xmath37 , @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 and @xmath47 filters are available from the ultraviolet hubble ultra deep field ( uvudf ) , hudf and cosmic assembly near - infrared deep extragalactic legacy survey ( candels ) surveys and other follow - up campaigns @xcite .
in this paper , we specifically use @xmath39 to @xmath47 data extracted from the extreme deep field ( xdf ) obtained combining all of the available observations and reaching a death of @xmath030ab mag @xcite .
the @xmath37 and @xmath38 images are extracted from the hubble deep uv ( hduv ) data release ( oesch et al . , submitted ) , which includes the v2 release of the uvudf database @xcite .
the hst stamp images and colour maps of udf6462 are shown in figure [ fig : udf6462stamps ] .
near- to far - infrared ( fir ) coverage at wavelengths between 3.6@xmath14 m and 500@xmath14 m is also available from the spitzer , goods - herschel and herschel multi - tiered extragalactic survey ( hermes ) surveys @xcite . at these long wavelengths source confusion
becomes severe and flux deblending is necessary .
our mid - ir and fir photometry of udf6462 is based on new , improved deblending algorithms developed for goods - s by liu et al .
2017 ( in prep . , for fluxes at @xmath48@xmath49250@xmath14 m ) and wang et al .
2017 ( in prep .
, at @xmath48@xmath50250@xmath14 m ) . at even longer wavelengths ,
0.3@[email protected]@xmath1 resolution observations of the radio synchrotron emission were obtained from the vla c - band ( 6ghz , @xmath05 cm ) survey described in @xcite .
finally , maps of @xmath52 , [ nii ] and [ sii ] emission and the velocity field in udf6462 were derived from the sinfoni integral - field spectroscopic observations presented in @xcite .
together with providing a kinematical classification of udf6462 , these data also enabled the derivation of gas - phase metallicities in different regions of the galaxy and thus provide important constraints on the co - to-@xmath53 conversion factor ( @xmath54 ) in udf6462 ( see section [ sec : mgas_sfe ] ) . from left to right and top to bottom : composite @xmath55-@xmath43-@xmath47 image from xdf without ( left ) or with ( right ) highlighting the position and size of the clumps identified in section [ sec : clumpdescription ] ( white circles with red numbers ) ; ( @xmath55-@xmath43 ) colour map and ( @xmath43-@xmath47 ) colour map .
two other galaxies at a different redshift are also indicated in the top left panel .
axis coordinates are offsets relative to the phase centre of the alma observations .
, scaledwidth=49.0% ] for the integrated stellar mass of udf6462 we adopt the value of log(m@xmath56/@xmath26)=10.4 published by the 3d - hst team and derived from the multiwavelength photometric catalog ( uv to nir ) of @xcite .
independent spectral energy distribution ( sed ) fits that we performed using the @xcite candels catalog and different sets of stellar population models yielded consistent results within 0.15 dex @xcite . in the following
we will take this value to represent the typical uncertainty on the stellar mass of udf6462 .
thanks to the exquisite multiwavelength data sets available , we can calculate the integrated star formation rate ( sfr ) using different approaches .
our fiducial estimate of the dust - obscured sfr is obtained by fitting the ir ( 3.6@xmath14 m to 350@xmath14 m ) and radio photometry of udf6462 with the @xcite models , following the approach described in @xcite . in performing this fit
, we included the continuum flux measurement at 1300@xmath57 derived from the alma data presented in section [ sec : continuum ] , while we did not use the 500@xmath14 m datapoint due to uncertain deblending for this galaxy even in our improved photometric catalogs .
the best fit ir - model is shown in figure [ fig : irsed ] .
it returns a total ( 8 - 1000@xmath14 m ) ir luminosity of @xmath2=(33@xmath239)@xmath58 , corresponding to an integrated sfr equal to sfr@xmath24=(33@xmath599)@xmath26yr@xmath19 when using the @xcite l@xmath24 to sfr conversion factor for a chabrier imf .
very similar results ( within @xmath00.1 dex ) are obtained when fitting the sed using other commonly used codes ( e.g. , lephare or magphys , * ? ? ?
* ; * ? ? ?
* ) and different sets of ir templates @xcite
. observed ir / radio sed of udf6462 ( red circles with error bars ) obtained combing the spitzer / herschel photometric data points at @xmath60 m with the flux density measured from the alma continuum image at @xmath61 m ( section [ sec : continuum ] , shown as green point ) and the @xmath05 cm radio flux from the vla observations of @xcite .
the thick black line shows the best fit @xcite model derived in section [ sec : sfrmassint ] , with the stellar contribution included ( green dash - dotted line ) .
the blue and orange lines are the contributions from the diffuse ism component and the photodissociation regions , respectively ( see @xcite for a discussion on the fitting technique and components).,scaledwidth=48.0% ] from the hst data we can furthermore derive the unobscured component of the sfr . using the @xmath38 and @xmath39 band observations , we calculate a dust - uncorrected , rest frame uv luminosity of @xmath62=([email protected])@xmath5110@xmath63ergss@xmath19hz@xmath19 and a sfr@xmath28=([email protected])@xmath26yr@xmath19 using the relation between uv luminosity and sfr in @xcite , converted to a chabrier imf
. combining the ir and uv estimates , we obtain a total star formation rate of sfr@xmath64=sfr@xmath24+sfr@xmath28=(37@xmath239)@xmath26yr@xmath19 .
finally , we can also derive an extinction - free sfr from the vla c - band observations of @xcite .
the radio map of udf6462 is presented in the top right panel of figure [ fig : udf6462resmaps ] , showing clear radio emission from the galaxy central region .
the 6ghz flux density of udf6462 was measured following the gaussian fitting approach with the pybdsm software described in @xcite . to recover extended emission ,
the fitting was performed on several tapering scales and the one maximising the s / n was chosen as the optimal one for source extraction . with this method we obtained a total flux density of @xmath65=([email protected])@xmath14jy with a s / n=6.6 and extended over a gaussian with beam - deconvolved full width at half maximum ( fwhm ) of (
this flux density corresponds to a rest - frame luminosity at 1.4ghz of @xmath66 when assuming a radio power law @xmath67@xmath68@xmath69 with @xmath70=0.7 @xcite .
we then applied the radio luminosity to sfr conversion in @xcite , which for a chabrier imf implies a sfr@xmath27=3.18@xmath71l@xmath72=(43@xmath237)@xmath26yr@xmath19 .
+ the c - band flux corresponds to a logarithmic ir to radio flux ratio @xmath73=@xmath74@[email protected] , i.e. , udf6264 lies within the scatter of the observed local ir - radio correlation ( @[email protected]@xmath230.3 , e.g. , * ? ? ?
* ; * ? ? ?
to summarise , sfr values derived from ir+uv and radio data are consistent within their uncertainties . for the reminder of the paper
, we will thus use the average of the two measurements , @xmath8sfr@xmath50=(40@xmath235)@xmath26yr@xmath19 , as our fiducial value for the sfr of udf6462 . with a stellar mass of @xmath22=10.4 ,
udf6462 is hence slightly below ( 0.3dex ) the characteristic mass @xmath76 of the mass function of star - forming galaxies @xcite and lies on the @xmath77 main - sequence ( @xmath78=sfr / sfr@xmath79=1.2 , see also figure [ fig : msclumps ] ) . to understand whether udf6462 hosts an active galactic nucleus ( agn ) and
if hence its ir / radio and co line emission could be significantly affected by it , we investigate several agn diagnostics , which we present in figure [ fig : agncont ] .
as mentioned above , we estimated the rest - frame 1.4ghz luminosity of udf6462 to be 1.36@xmath80whz@xmath19 .
we note that this galaxy is not detected in the vla 1.4ghz maps of goods - south published in @xcite , placing a @xmath81 upper limit on the 1.4ghz luminosity(flux ) of @xmath82whz@xmath19 ( 19.5@xmath14jy ) .
both estimates are below the threshold of 10@xmath83whz@xmath19 , above which powerful radio - loud agns typically dominate the radio luminosity function ( e.g. * ? ? ?
* ; * ? ? ?
* ) .
the infrared properties of udf6462 are also consistent with little contamination from an agn component .
this is shown in the upper two panels of figure [ fig : agncont ] . in panel
( a ) in particular , the irac [ 3.6]-[5.8 ] and [ 4.5]-[8.0 ] colours of udf6462 place it in the region dominated by star - forming galaxies according to the criteria of @xcite which was optimised for high - redshift galaxies .
udf6264 would be classified as an agn according to the original cuts proposed by @xcite , however it was shown in @xcite that , especially at high redshift , this selection can suffer from significant contamination from star - forming galaxies .
the above results are also confirmed by the decomposition of the ir sed of our galaxy into an agn and a star - forming component , shown in panel ( b ) of figure [ fig : agncont ] . to perform this decomposition , we used the agn sed template of @xcite together with the ir best - fit template we obtained in section [ sec : sfrmassint ] and the public idl code for the fitting . from this analysis
, we obtain an agn contribution of @xmath015% to the total ir luminosity .
we note , however , that from the pure star - forming fit described in section [ sec : sfrmassint ] there is no evidence for the need of a second component .
the percentage quoted above hence likely represents an upper limit of the agn contribution to the ir .
x - ray observations of udf6462 are available from the 4ms and 7ms chandra deep field south data of @xcite and @xcite . these data provide x - ray luminosities equal to @xmath84(0.5 - 8kev)=1.4@xmath85ergs@xmath19 and @xmath84(0.5 - 7kev)=1.1@xmath85ergs@xmath19 . although these place udf6462 slightly above the locus of purely star - forming galaxies in the @xmath84-sfr plane ( see panel ( c ) of figure [ fig : agncont ] ) the x - ray luminosity is well below the threshold of 3@xmath85ergs@xmath19 used to identify luminous agn and udf6462 is classified as a normal galaxy in both catalogues
. we can finally also use the stellar mass of this galaxy and the [ oiii]/h@xmath86 ratio measured from the 3d - hst grism spectrum @xcite to find the location of udf6462 in the mass - excitation plane of @xcite ( mex ) . as discussed in @xcite , at high redshift
the dividing line between agn and normal star - forming galaxies shifts toward higher stellar masses with respect to the local relation as a consequence of selection effects and evolution of the overall galaxy population .
when calculating the agn probability for udf6264 using the mex diagnostic , we hence imposed a minimum line flux of 2.1@xmath87ergs@xmath19cm@xmath31 corresponding to the detection limit of 3d - hst @xcite and allow for the evolution of the characteristic luminosity , @xmath89 .
we find that udf6462 lies below the limiting line dividing star - forming galaxies from agn with a 64% probability that the galaxy could be a purely sf system . from the evidence above
, we conclude that udf6472 does not host a strong agn and its radio and ir properties can largely be ascribed to pure star formation .
the morphology udf6264 is characterised by several giant clumps , as shown in figure [ fig : udf6462stamps ] . as extensively discussed in @xcite
, udf6462 has a velocity gradient that is consistent with a rotationally supported disk ( albeit disturbed by the presence of some kinematically distinct clumps at small scales ) . the complex morphology seen in the optical / nir images of this galaxy is thus unlikely to be the outcome of a merger and can instead be explained with instability - driven formation of giant clumps , as observed in many z@xmath501 galaxies . using the available hst photometry spanning the wavelength range @xmath90@xmath41000 to 6000@xmath91
, we derived resolved colour maps and maps of stellar mass , star - formation rate and dust extinction for udf6462 following the approach in @xcite .
the details of the generation of these maps can be found in @xcite .
briefly , we performed pixel - by - pixel sed fitting to all available images , after homogenisation of the s / n and convolution to match the @xmath47-band resolution . in this work the wfc3 @xmath37 and @xmath38 pass - bands
were also included , providing better constraints on dust extinction and sfr .
in @xcite we used a set of @xcite templates with delayed , exponentially declining star formation histories ( delayed @xmath92-models ) , but for the scientific purposes of this paper , we preferred to limit our library to templates with a constant star formation ( csf ) .
it is known that sed fits in which the sfh and dust extinction are allowed to vary can result in degeneracies between highly obscured models and models dominated by an old stellar population . while this has a limited impact for stellar masses , it can lead to step - like spacial variations in the sfr maps , especially when individual pixels are fit independently from the others , as is the case in our analysis . to minimise such degeneracies
, we hence decided to fix the sfh and consider the maps obtained with csf templates as our fiducial estimate .
specifically , we used csf models with ages between 100myr and 2gyr and a solar metallicity ( in agreement with the estimate in @xcite ) and corrected for internal dust extinction assuming a @xcite law with @xmath93 ranging between 0 and 0.9 mag .
we note that neither from the observed @xmath94 & @xmath95 colours nor applying the @xmath96 colour criterion @xcite on the best - fit seds is there evidence for a region with little current star formation in udf6462 , justifying our choice of a star - forming template . for details on the different methods .
error bars correspond to the formal uncertainties in the derived parameters .
we show with empty grey circles the average sfr and masses derived in @xcite for the four clumps that we have in common with their sample .
note that , as discussed in section [ sec : clumpsextraction ] , although we visually confirmed that these clumps are spatially coincident , it was not possible to verify if they also extend over exactly the same areas and hence whether their estimates are directly comparable with those presented here.,scaledwidth=46.0% ] we also generated a less model dependent version of the sfr map by applying the @xmath97-sfr relation of @xcite , where the pixel - based uv luminosities where calculated by interpolating the @xmath38 and @xmath39 photometry and corrected for dust extinction using the @xmath93 vs. @xmath94 relation described in @xcite and a @xcite extinction curve . for illustration , we show in figure [ fig : udf6462resmaps ] the maps of stellar mass , sfr and specific sfr ( ssfr ) derived from the csf sed fit , the other versions of these maps can be found in figure [ fig : udf6462resmaps_appendix ] of appendix [ app : clumpmeasurements ] .
when summing up the individual pixels in the resolved maps calculated via csf sed fitting , we obtain a total stellar mass of 10@xmath98 and a sfr of 40@xmath26 yr@xmath19 , which are in good agreement with the integrated properties discussed above .
the sfr map derived from the corrected @xmath97 results in a sfr=26@xmath26 yr@xmath19 , suggesting that we are underestimating the dust extinction in this case .
we further discuss dust extinction in the resolved maps in section [ sec : nuclearregion ] and appendix [ app : clumpmeasurements ] .
to characterise the location of the clumps , we followed an approach similar to that outlined in @xcite , who studied clump properties for a sample of galaxies in the candels field , also including udf6462 .
specifically , we identified as clumps those structures that are brighter than the underlying galactic diffuse component at the 5@xmath99 level .
we characterised the diffuse component via both a non - parametric approach , i.e. , by smoothing the galaxy images with a 4-pixel gaussian kernel , and by modelling the galaxy surface brightness with galfit @xcite . in both cases we subtracted the diffuse component of the galaxy and
run the source extractor code ( sextractor , @xcite ) on the residual image ; sources of at least 5 pixels with a residual flux above the 5@xmath99 level were identified as clumps .
we performed the detection in the @xmath55 , @xmath43 and @xmath47 bands independently .
we obtained consistent results for the three bands ( see figure [ fig : appresiduals ] in appendix [ app : clumpsprops ] ) and using the two smoothing methods . for easier comparison with previous works
, we use the @xmath43 band as the reference , but our findings do not depend on the exact clump detection technique that we employ .
we identified 6 main clumps in our galaxy which we show in figures [ fig : udf6462stamps ] and [ fig : udf6462resmaps ] as white ellipses with extents given by the sextractor petrosian radii .
the four northern clumps ( clump i d 1 to 4 ) coincide with those described in the work of @xcite .
table [ tab : udf6462prop_clump ] summarises the clump main physical properties that we derived from the optical images and resolved maps . observed sizes for the clumps were obtained from the sextractor petrosian radius calculated on the residual @xmath100-band image , after subtraction of the diffuse component .
these radii enclose most of the clump fluxes in the optical images , as shown by the white circles in figure [ fig : udf6462stamps ] .
we also derived intrinsic clumps sizes by fitting with galfit the residual @xmath100-band image with multiple gaussian sources having centres fixed at the clump position .
given its petrosian radius ( see table [ tab : udf6462prop_clump ] ) , we estimate that clump # 3 is unresolved at the @xmath47-band resolution and hence treat it as a point source for the rest of the paper . to estimate the stellar mass and star - formation rates of the clumps we followed a number of approaches , an in - depth discussion of these methods is given in appendix [ app : clumpmeasurements ] .
briefly , the techniques we explored differ in terms of how the clump sed fitting is performed ( integrated clump photometry vs. pixel - by - pixel from resolved maps ) , how star formation rate is modelled / traced ( uv luminosity vs. csf templates ) and whether or not a contribution for a galactic diffuse component is subtracted . as evident in figure [ fig : clumpssfr ] , the derivation of the clump physical parameters is affected by systematics that depend on all these assumptions .
while these systematics are less important for stellar masses , which typically agree within a factor of 2 - 3 for each individual clump , sfr values can differ by a factor @xmath010 in some of the clumps .
we adopt the average of the different measurements for each clump as our preferred estimate for its mass and sfr but , to stress that these can be subject to large uncertainties , we will show the full range of possible values as error bars in our plots .
the 6 clumps constitute on average @xmath025% and @xmath010% of the total star formation and stellar mass in the galaxy , respectively . depending on the method applied , however
, the contribution to the sfr varies between 10% and 40% and between 6% and 16% for the stellar mass .
the position of the clumps in the sfr vs. mass and sfr density vs. mass density planes is shown in figure [ fig : msclumps ] .
we show for reference the ms of star - forming galaxies as derived from galaxy integrated properties ( using the parametrisation of the ms from * ? ? ?
* ; * ? ? ?
* ) . if this relation holds at the scale of clumps , our clump sample mostly consists of ms " clumps , with just one clump showing somewhat enhanced activity ( # 2 ) . ) .
, scaledwidth=50.0% ] in terms of ssfr , clumps in udf6462 have average values of the order of ssfr@xmath4@xmath101yr@xmath19 , i.e. , a stellar mass doubling timescale of 1/ssfr@xmath4@xmath102yr . as mentioned above
, these ssfr are comparable those measured in ms star - forming galaxies at the redshift of udf6462 , however are lower than the ssfr that can be attained in the most massive giant molecular clouds , where a few percent of their gas masses ( @xmath0@xmath103 ) is locked into stars with sfr of the order of @xmath104yr@xmath19 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , resulting in an approximate timescale of 1/ssfr@xmath4@xmath105yr .
as visible in figure [ fig : udf6462resmaps ] , the clumps also have typically low to intermediate dust extinction .
ccccccccccc i d & r.a . & dec & sma & @xmath106 & m@xmath56 & sfr & e(b - v ) & @xmath94 & @xmath95 & @xmath29 + & ( deg , j2000 ) & ( deg , j2000 ) & ( @xmath1 ) & ( @xmath1 ) & @xmath107 & ( @xmath26/yr ) & ( mag ) & ( mag ) & ( mag ) & ( jykms@xmath19 ) + + 1 & 53.181717 & -27.783123 & 0.19 & 0.06 & 8.56@xmath108 & [email protected] & 0.26/0.20 & [email protected] & [email protected] & @xmath80.020 + & @xmath2053.181732 & -27.783199 + 2 & 53.181583 & -27.783060 & 0.30 & 0.14 & 8.40@xmath109 & [email protected] & 0.29/0.28 & [email protected] & [email protected] & @xmath80.021 + & @xmath2053.181599 & -27.783136 + 3 & 53.181812 & -27.782932 & 0.16 & pt.s .
& 8.69@xmath110 & [email protected] & 0.26/0.25 & [email protected] & [email protected] & @xmath80.019 + & @xmath2053.181828 & -27.783009 + 4 & 53.181728 & -27.782829 & 0.29 & 0.18 & 8.89@xmath111 & [email protected] & 0.18/0.19 & [email protected] & [email protected] & @xmath80.021 + & @xmath2053.181744 & -27.782906 + 5 & 53.181370 & -27.783344 & 0.23 & 0.25 & 8.44@xmath110 & [email protected] & 0.25/0.19 & [email protected] & [email protected] & @xmath80.021 + & @xmath2053.181385 & -27.783421 + 6 & 53.181408 & -27.783510 & 0.29 & 0.31 & 8.73@xmath112 & [email protected] & 0.11/0.11 & [email protected] & [email protected] & @xmath80.025 + & @xmath2053.181423 & -27.783587 + at the centre of udf6462 there is a red nucleus ( orange symbol in figure [ fig : msclumps ] ) , which displays a high star formation rate and very high dust extinction in the resolved maps .
high attenuation toward the centre of galaxies is commonly found in high redshift galaxies with masses comparable to that of udf6462 ( e.g. * ? ? ?
* ) . the significant radio emission over this area in the vla radio map ( see top right panel in figure [ fig : udf6462resmaps ] ) , as well as the results from the alma observations we describe in the next section , are further evidence for this region being truly star - forming .
the red nucleus is also coincident with the peak of the mass distribution and is metal enriched with respect to the rest of the galaxy @xcite .
moreover , the galfit model of the galaxy @xmath47-band surface brightness distribution yields a fairly concentrated profile with a srsic index of [email protected] . a galfit fit to the mass maps results in a similarly high index of [email protected] .
the above properties are suggestive of a forming bulge , still hosting a non - negligible amount of ( obscured ) star formation ( see also discussion in @xcite ) . cccccccc date & configuration & n. of antennae & max .
baseline & time on source & angular resolution & pwv & max .
scale + & & & ( km ) & ( hours ) & ( arcsec ) & ( mm ) & ( arcsec ) + + 2014 - 07 - 18 & c34 - 4/5 & 32 & 0.65 & 0.56 & 0.65 & 0.58 & 8.4 + 2014 - 07 - 22 & c34 - 4/5 & 33 & 0.78 & 0.77 & 0.42 & 1.64 & 8.4 + 2015 - 06 - 06 & c34 - 5 & 37 & 0.74 & 0.77 & 0.61 & 0.56 & 7.8 + 2015 - 06 - 16 & c34 - 5 & 36 & 0.78 & 0.77 & 0.46 & 0.58 & 6.3 + 2015 - 08 - 28 & c34 - 7/(6 ) & 40 & 1.6 & 0.77 & 0.26 & 1.75 & 4.0 + 2015 - 08 - 28 & c34 - 7/(6 ) & 40 & 1.6 & 0.77 & 0.20 & 1.70 & 4.0 + we note that in the work of @xcite a clump has been identified in this region ( number 1 , in their figure 1 ) .
we also find some evidence for a clumpy structure ( see e.g. , figure [ fig : appresiduals ] ) that is slightly shifted eastward with respect to the centre of mass and the reddest peak in the ( @xmath55-@xmath43 ) and ( @xmath43-@xmath47 ) colour maps , although it is not as prominent at the other clumps .
it is possible that toward the centre of the galaxy a young clump in the foreground overlaps with the central mass concentration .
however , given the complexity of the galaxy in this area , we do not attempt a separation of these two components and instead we will consider the region as a whole and refer to it as the nuclear region " from now on . due to the high dust extinction , sed - based sfr estimates for this region vary strongly depending on the assumed star formation history and are consequently subject to large uncertainties .
however , the high - resolution vla observations , which are not affected by dust obscuration , enable us to derive more robust sfrs and also to quantify systematic biases in the sfr map from the sed fitting .
a comparison of the sfr surface density radial profiles derived from the radio observations and the maps from the sed fits over the nuclear region is presented in appendix [ app : clumpmeasurements ] .
we find that the sed - based maps result in shallower sfr profiles and underestimate the sfr in the nuclear region by roughly factors of 1.5 to 2.5 depending on the templates used .
for this reason , when considering the red nucleus of udf6462 , we use the radio image as our fiducial sfr tracer .
as we discuss in more detail in the next section , the bulk of the co(5 - 4 ) emission in our alma observations originates from this nuclear region . to be able to directly link its sfr and stellar mass to the molecular gas content
, we use the best - fit gaussian parameters of the co map ( section [ sec : co5 - 4emission ] ) to define the extent of the nuclear region , instead of performing the extraction on the @xmath43 image as was done for the clumps .
the stellar mass of the nuclear region is then obtained by summing up pixels in the mass map within an ellipse of major axis equal to the fwhm of the co(5 - 4 ) gaussian .
the sfr is derived by integrating the radio best - fit gaussian over the width of the co(5 - 4 ) gaussian .
we obtained @[email protected] and sfr=(29@xmath239)@xmath26 yr@xmath19 for the nuclear region .
alma band 6 observations for udf6462 were carried out during cycle 2 with angular resolution ranging from 0.2@xmath1 to 0.6@xmath1 ( project i d : 2013.1.01271.s , pi : cibinel ) .
a log of these observations is provided in table [ tab : obslog ] .
we configured the first alma baseband which contains the redshifted co(5 - 4 ) line observed and rest frequencies equal to @xmath113=224.23ghz and @xmath114=576.26ghz with a spectral resolution of 0.976mhz .
the other three baseband were used in time division mode ( tdm ) mode for continuum observations . considering line - free channels in all four bands ,
our continuum observations cover 7.4 ghz and the rest - frame wavelength range @xmath90@xmath4480 - 520@xmath14 m ( average observed wavelength @xmath115@xmath41300@xmath14 m ) .
we integrated for a total of 4.4h on source , reaching sensitivities of @xmath116=7.6@xmath14jy / beam ( natural weighting ) for the continuum and @xmath117=0.066mjy / beam over a 100km / s channel for the co line .
according to the co(5 - 4)-ir relation in @xcite , the @xmath81 line sensitivity corresponds to a sfr@xmath03@xmath26/yr@xmath19 in an individual resolution element .
calibration of the alma raw data was carried out using common astronomy software applications ( casa , * ? ? ?
* ) and the standard alma pipeline .
imaging of visibilities and the scientific analysis of the data cubes were instead mostly performed with the gildas software package @xcite . to maximise sensitivity ,
a natural weighting was used to generate the maps , resulting in a synthesised beam of fwhm @xmath118 .
the maximum recoverable scale of our observations is @xmath04@xmath119 .
as we discuss in more detail in appendix [ app : astrometryoff ] , there is an astrometric offset between the alma observations and hst data products on the goods - s field that has to be taken into account when interpreting the data ( see also @xcite ) . in this paper
we only refer to hst images which already incorporate the astrometric correction .
however , to allow an easy identification of the clumps / structures in the optical images and the comparison with previous studies , we also provide the uncorrected hst celestial coordinates in tables [ tab : udf6462prop ] and [ tab : udf6462prop_clump ] . for reference and comparison of our data with previous studies , we also provide in appendix
[ app : astrometryoff ] the @xmath01300@xmath14 m integrated fluxes for four other galaxies detected in our continuum image .
three of these galaxies are in common with the sample of @xcite and @xcite .
the velocity - integrated intensity map of udf6462 is shown in figure [ fig : colineint ] .
the line emission is concentrated on the nuclear region and detected at a 10@xmath99 significance level ( ratio of the peak line flux to the noise in the intensity map ) .
we extracted the total co(5 - 4 ) flux associated with udf6462 through @xmath120-plane fitting with an elliptical gaussian model , using the gildas task .
we assumed as initial guesses for the model centre and size the coordinates of the peak in the optical images and the alma synthesised beam parameters , respectively , and let the parameters vary freely .
the fitting was performed over a velocity range @xmath121__v__@xmath41680kms@xmath19 ( corresponding to a conservative velocity interval equal to @xmath04.5 times the observed line width derived below ) .
we obtained a velocity - integrated line intensity of @xmath122=@xmath123__v__=@xmath124mjykms@xmath19 , corresponding to a line luminosity of : @xmath125 where @xmath113=224.23ghz , @xmath126 and @xmath127 is the luminosity distance in mpc @xcite .
the estimates above have been corrected for the continuum contribution discussed in section [ sec : continuum ] . from the best - fit parameters ,
we also find a beam - deconvolved fwhm of the co(5 - 4 ) emission equal to @xmath128 , i.e. , roughly 2kpc in radius .
this is in good agreement with the extent of the nuclear region in the optical images .
we describe here the simulated alma observations utilised to test the robustness of the co(5 - 4 ) size measurements .
we started by generating an input model for the co(5 - 4 ) emission from the hst sfr map .
we show here simulations based on the sfr map obtained from the dust - corrected @xmath97 , but similar results are obtained if using the sfr from the sed fitting with csf models . as we discussed in section [ sec : cosfrdist ] , a direct conversion of the sfr into co(5 - 4 ) luminosity results in too low fluxes , likely as a consequence of underestimation of nuclear dust obscuration . for this reason
, we renormalised the input sfr map such to match the observed co(5 - 4 ) line flux within the @xmath129 gaussian region over which we detect the line emission , accounting for hst psf blurring .
( note that this effectively translates into a simple division of the sfr map by the flux within that area , hence the pixel relative fluxes and radial distribution are conserved . )
we then employed the task within casa to simulate alma observations .
the relevant task parameters were set to reproduce exactly the date & time of observation , antennae configuration , the total integration time and atmospheric conditions of the real alma data . during the simulation , we employed a channel width of 0.33ghz , i.e. , corresponding to the @xmath0430km / s velocity range over which we detect significant line emission in figure [ fig : colineint ] ( grey shaded area in right panel ) .
we show in figure [ fig : app : fakeobs ] the thus obtained artificial alma maps .
globally , the simulated observations reproduce the real data in terms of s / n and spatial distribution .
we modelled the simulated @xmath120-plane with a gaussian source , using the task as done for the real observation and measured a size of fwhm=([email protected]@[email protected] ) arcsec , hence recovering the intrinsic model size .
tapered versions of both the real and simulated images on a scale of 1@xmath1 are shown in the right panels of figure [ fig : app : fakeobs ] .
while tapering on the simulated data reveals the extended emission in the north - south direction and increases s / n , in the real alma observations this is not the case and the s / n is not significantly improved .
the findings above hence suggest that emission distributed on scales of 1@xmath1 , if present , should be recoverable at the s / n of our observations and that the small co size compared to the sfr / radio map is likely intrinsic . | we present deep alma co(5 - 4 ) observations of a main sequence , clumpy galaxy at z=1.5 in the hudf .
thanks to the @xmath00.5@xmath1 resolution of the alma data , we can link stellar population properties to the co(5 - 4 ) emission on scales of a few kpc . we detect strong co(5 - 4 ) emission from the nuclear region of the galaxy , consistent with the observed @xmath2@xmath3 correlation and indicating on - going nuclear star formation .
the co(5 - 4 ) gas component appears more concentrated than other star formation tracers or the dust distribution in this galaxy .
we discuss possible implications of this difference in terms of star formation efficiency and mass build - up at the galaxy centre .
conversely , we do not detect any co(5 - 4 ) emission from the uv - bright clumps .
this might imply that clumps have a high star formation efficiency ( although they do not display unusually high specific star formation rates ) and are not entirely gas dominated , with gas fractions no larger than that of their host galaxy ( @xmath050% ) .
stellar feedback and disk instability torques funnelling gas towards the galaxy centre could contribute to the relatively low gas content .
alternatively , clumps could fall in a more standard star formation efficiency regime if their actual star - formation rates are lower than generally assumed .
we find that clump star - formation rates derived with several different , plausible methods can vary by up to an order of magnitude .
the lowest estimates would be compatible with a co(5 - 4 ) non - detection even for main - sequence like values of star formation efficiency and gas content .
[ firstpage ] galaxies : evolution galaxies : ism galaxies : star formation galaxies : high - redshift submillimeter : galaxies |
let @xmath0 be a simple finite - dimensional lie superalgebra with a non - degenerate supersymmetric even invariant bilinear form @xmath10 , @xmath1 be a nilpotent element in the even part of @xmath0 , @xmath2 be a good grading of @xmath0 for @xmath1 , denoted by @xmath11 and @xmath3 be the ( affine ) @xmath4-algebra defined as the brst cohomology associated with @xmath12 and @xmath2 ( @xcite ) .
the @xmath4-algebra is a @xmath13-graded vertex superalgebra and it is conformal if its level is not critical i.e. @xmath14 , where @xmath15 is the dual coxeter number of @xmath0 . for fixed @xmath0 , @xmath1 and @xmath16 , the vertex superalgebra structure of the @xmath4-algebra @xmath3 does not depend on the choice of the good grading @xmath2 ,
although the conformal grading does ; see @xcite . in this paper , we present an arbitrary @xmath4-algebra as the intersection of kernels of the screening operators . in the special case that the good grading @xmath2 can be chosen such that @xmath17 is a cartan subalgebra
, our presentation gives a free field realization in terms of the screening operators , which recovers a result of feigin and frenkel @xcite in the case that @xmath0 is a simple lie algebra and @xmath1 is a regular nilpotent element . as applications , we prove that the @xmath4-algebra associated with a regular nilpotent element in @xmath6 is isomorphic to the @xmath7-algebra introduced by fateev and lukyanov , and that the @xmath4-algebra associated with a subregular nilpotent element in @xmath8 is isomorphic to the @xmath9-algebra introduced by feigin and semikhatov .
these resluts were conjectured in @xcite and @xcite , respectively . to clarify our results ,
let @xmath18 be a set of roots of @xmath0 and @xmath19 be a set of simple roots compatible with @xmath2 .
for @xmath20 , we denote by @xmath21 the set of roots whose root vector lies in @xmath22 and set @xmath23 .
then , there is a decomposition @xmath24 ; see e.g. @xcite .
denote by @xmath25 the restricted root system associated with @xmath2 ( @xcite ) and by @xmath26 the base of @xmath27 .
then , @xmath28 , where @xmath29 .
let @xmath30 be a root lattice of @xmath17 .
the equivalence relation on @xmath31 defined by @xmath32 for @xmath33 gives the quotient set @xmath34 $ ] of @xmath31 .
let @xmath35 $ ] be the equivalent class of @xmath36 in @xmath34 $ ] .
since the equivalence relation on @xmath31 is homogeneous with respect to the grading , the decomposition of @xmath31 induces that of @xmath34 $ ] : @xmath37=[{\pi^{\gamma}}_{\frac{1}{2}}]\sqcup[{\pi^{\gamma}}_{1}].\end{aligned}\ ] ] let @xmath38 be an affine vertex superalgebra associated with the subalgebra @xmath17 of @xmath0 with degree @xmath39 and its invariant bilinear form @xmath40 ( see for the definition ) , and @xmath41 be a neutral free superfermion vertex superalgebra associated with the subspace @xmath42 of @xmath0 with degree @xmath43 ( @xcite ) . for generic @xmath16 ,
the @xmath4-algebra @xmath3 is isomorphic to the vertex subalgebra of @xmath44 , which is the intersection of kernels of the screening operators @xmath45}$ ] for all @xmath35\in[{\pi^{\gamma}}]$ ] : @xmath46\in[{\pi^{\gamma}}]}{\mathrm{ker}}\ q_{[\beta]}.\end{aligned}\ ] ] see and for the definitions of the screening operators @xmath45}$ ] .
we note that theorem a describes the image of the miura map defined in @xcite , see section [ miura sec ] . in the special case
that @xmath17 is a cartan subalgebra @xmath47 , the vertex superalgebra @xmath38 is isomorphic to the heisenberg vertex algebra @xmath48 associated with @xmath47 .
moreover , since the equivalence relation on @xmath31 is trivial and @xmath49 is empty , the set @xmath34 $ ] is equal to @xmath19 .
let @xmath50 be a linear function on @xmath0 defined by @xmath51 for all @xmath52 , @xmath53 be the vertex operator on @xmath48 associated with @xmath54 , where we identify the dual space @xmath55 with @xmath47 via the fixed bilinear form , and @xmath56 be the vertex operator on @xmath41 associated with @xmath57 .
then we obtain a more explicit presentation of the @xmath4-algebra .
assume that @xmath2 can be chosen so that @xmath17 is a cartan subalgebra @xmath47 of @xmath0 .
then , for generic @xmath16 , the @xmath4-algebra @xmath3 is isomorphic to the vertex subalgebra of @xmath58 , which is the intersection of kernels of the screening operators : @xmath59 where @xmath60 .
theorem b for a simple lie algebra and @xmath1 a regular nilpotent element was known by feigin and frenkel @xcite , and for a simple lie superalgebra and @xmath1 a superprincipal with @xmath61=f$ ] ( @xmath1 being regular nilpotent ) and these two vectors form part of @xmath62 . ]
nilpotent element by heluani and dias @xcite ( see remark 3.4 of @xcite ) . as applications of theorem
a and b , we have the following results that were conjectured in @xcite and @xcite
. + 1 .
the @xmath4-algebra associated with a regular nilpotent even element @xmath63 in @xmath6 at level @xmath64 is isomorphic to the fateev - lukyanov @xmath7-algebra ( @xcite ) : @xmath65 2 .
the @xmath4-algebra associated with a subregular nilpotent element @xmath66 in @xmath8 at level @xmath67 is isomorphic to the feigin - semikhatov @xmath9-algebra ( @xcite ) : @xmath68 for @xmath69 , theorem c ( 1 ) was shown in @xcite as the @xmath70-algebra is the neveu - schwarz vertex superalgebra ( or the super virasoro vertex superalgebra ) ( @xcite ) .
we apply theorem b to @xmath6 and @xmath63 to prove theorem c ( 1 ) for generic @xmath16 , and use a result related to the miura map to extend the proof to non - critical level @xmath16 , see the proof of theorem [ wbn thm ] for the details . for @xmath71
, theorem c ( 2 ) follows from results of @xcite , for the @xmath72-algebra is the bershadsky - polyakov vertex algebra ( @xcite ) .
we apply theorem a to @xmath73 and @xmath74 , and prove theorem c ( 2 ) using the wakimoto construction of the affine vertex algebra of @xmath75 for generic @xmath16 and a result related to the miura map for non - critical level @xmath16 , see the proof of theorem [ w2n thm ] .
the paper is organized as follows . in section [ w - def subsec ] , we review the definition of the @xmath4-algebra @xmath3 . in section [ kw subsec ] , we recall the definition of the complex @xmath76 that is used to compute @xmath3 . in section [ classical subsec ] ,
we consider the classical limit of @xmath76 , which is used to prove theorem a. in section [ screening sec ] , we set up some notations to state our main results . in particular , we define the screening operators , which consist of the vertex operators on @xmath41 and certain operators @xmath77 . in section [ main thm sec ]
, we deduce theorem b from theorem a. section [ proof sec ] is devoted to the proof of theorem a. we define the weight filtration on @xmath76 in section [ weight subsec ] and study the first spectral sequence @xmath78 in section [ e1 subsec ] by using the classical limit . the key fact that we use to analyze @xmath78 is lemma [ cinf lemma ] , which is proved in section [ cinf subsec ] . in section
[ vastr sec ] , we give another construction of @xmath77 and derive some results from it . in section [ proof subsec ] , we prove theorem a. section [ miura sec ] is devoted to the study of the miura map . in section
[ application sec ] , we prove theorem c. _ acknowledgments_this paper is the master thesis of the author and he wishes to express his gratitude to his supervisor tomoyuki arakawa for suggesting the problems and lots of advice to improve this paper .
he thanks hiroshi yamauchi and kazuya kawasetsu for useful comments and discussions .
he is deeply grateful to toshiro kuwabara for explaining him to the geometric aspects of the @xmath9-algebra and giving him the crucial insight to prove theorem c ( 2 ) .
in this section we review the definition of the ( affine ) @xmath4-algebras via the generalized drinfeld - sokolov reduction given by kac , roan and wakimoto @xcite .
let @xmath0 be a simple lie superalgebra with a non - degenerate even supersymmetric invariant bilinear form @xmath10 , @xmath1 be a nilpotent element of the even part of @xmath0 , @xmath16 be a complex number and @xmath2 be a good grading of @xmath0 with respect to @xmath1 , where a @xmath79-grading @xmath11 is called good if @xmath80 and @xmath81 is injective for @xmath82 and surjective for @xmath83 .
let @xmath84 be a semisimple element in @xmath0 such that @xmath85=j u\}$ ] .
choose a cartan subalgebra @xmath47 containing @xmath84 , a basis @xmath86 of @xmath47 , where @xmath87 , the root system @xmath18 of @xmath88 and a set of simple roots @xmath19 such that @xmath89 for all positive roots @xmath90 , where @xmath91 is a non - zero root vector .
we normalize the invariant bilinear form @xmath10 on @xmath0 by @xmath92 , where @xmath93 is a highest root of the even part of @xmath0 .
let @xmath94 and set @xmath23 for all @xmath95 , @xmath50 be a linear function on @xmath0 defined by @xmath51 for @xmath52 , @xmath96 be the parity of @xmath97 for all @xmath98 and @xmath99 be a structure constant for all @xmath100 such that @xmath101=\sum_{\gamma\in i\cup\delta}c_{\alpha,\beta}^{\gamma}{e_{\gamma}}$ ] .
denote by @xmath102 the killing form of @xmath0 and by @xmath103 the supertrace on @xmath0 . given a vertex superalgebra @xmath104 ( see @xcite for the definitions )
, we denote by @xmath105 the vertex operator for @xmath106 and by @xmath107=\sum_{n=0}^{\infty}(a_{(n)}b)\frac{{\lambda}}{n!}$ ] the @xmath108-bracket of @xmath109 .
consider the vertex superalgebra @xmath110 where @xmath111 is the universal affine vertex superalgebra associated with @xmath0 , @xmath112 is the charged free superfermion vertex superalgebra associated with @xmath113 with reversed parities and @xmath41 is the neutral free superfermion vertex superalgebra associated with @xmath42 ( see @xcite for definitions ) , where @xmath114 denote by @xmath115 the vertex operator on @xmath111 associated with @xmath52 , by @xmath116 the vertex operators on @xmath112 associated with @xmath117 for @xmath118 , by @xmath119 the vertex operator on @xmath41 associated with @xmath120 . note that @xmath121=[u , v]+k(u|v){\lambda},\quad [ { \varphi_{\alpha}}_{{\lambda}}\varphi^{\alpha'}]=\delta_{\alpha,\alpha'},\\ & [ { \varphi_{\alpha}}_{{\lambda}}\varphi_{\alpha'}]=[{\varphi^{\alpha}}_{{\lambda}}\varphi^{\alpha'}]=0,\quad [ { \phi_{\beta}}_{{\lambda}}\phi_{\beta'}]=(f|[e_{\beta},e_{\beta'}])\end{aligned}\ ] ] for all @xmath122 , @xmath123 and @xmath124 .
using the charges in @xmath112 defined by @xmath125 and @xmath126 , one has the induced charge decomposition of @xmath127 .
let @xmath128 where @xmath129 then @xmath130 and @xmath131 is a differential compatible with the charge decomposition of @xmath127 .
thus , @xmath132 is a cohomology complex .
we denote the cohomology of this complex by @xmath133 which inherits a vertex superalgebra structure from @xmath127 . since the @xmath134-th comohology of is zero if @xmath135 @xcite , @xmath3 is the @xmath39-th cohomology of .
this vertex superalgebra is called the _ @xmath4-algebra associated with @xmath0 , @xmath1 , @xmath16 , and @xmath2_. we note that the vertex superalgebra structure of the @xmath4-algebra does not depend on the choice of the good grading @xmath2 @xcite ( but the conformal grading of the @xmath4-algebra depends on @xmath2 ) .
we have @xcite the decomposition of the complex @xmath136 such that @xmath137 and @xmath76 is a vertex superalgebra generated by @xmath138 for all @xmath139 , @xmath140 for all @xmath118 and @xmath119 for all @xmath120 , where @xmath141 for @xmath139 .
note that @xmath142=j^{[u , v]}+\tau_{k}(u|v){\lambda},\quad[{\varphi^{\alpha}}_{{\lambda}}j^{u}]=\sum_{\beta\in{\delta_{>0}}}c_{u,\beta}^{\alpha}\varphi^{\beta},\quad[{\phi_{\gamma}}_{{\lambda}}j^{u}]=0\end{aligned}\ ] ] for all @xmath143 , @xmath118 , @xmath144 , where @xmath145 the vertex superalgebra @xmath76 is a subcomplex of @xmath127 and we have @xmath146=-\sum_{\begin{subarray}{c}\beta\in{\delta_{>0}}\\ \alpha\in i\cup\delta_{\leq 0}\end{subarray}}(-1)^{{p(\alpha)}}c_{u,\beta}^{\alpha}:j^{{e_{\alpha}}}\varphi^{\beta}:+\sum_{\beta\in{\delta_{>0}}}(\mathit{a}_{k}(u|{e_{\beta}}){\partial}+\mathit{b}_{k}(u|{e_{\beta}}){\lambda})\varphi^{\beta},\\ & [ { { d_{\mathrm{ne}}}}_{{\lambda}}j^{u}]=\sum_{\begin{subarray}{c}\beta\in{\delta_{>0}}\\ \alpha\in{\delta_{\frac{1}{2}}}\end{subarray}}c_{u,\beta}^{\alpha}:\phi_{\alpha}\varphi^{\beta}:,\quad [ { { d_{\chi}}}_{{\lambda}}j^{u}]=\sum_{\beta\in{\delta_{>0}}}\chi([u,{e_{\beta}}])\varphi^{\beta},\\ & [ { { d_{\mathrm{st}}}}_{{\lambda}}\varphi^{\alpha}]=-\frac{1}{2}\sum_{\beta,\gamma\in{\delta_{>0}}}(-1)^{{p(\alpha)}{p(\beta)}}c_{\beta,\gamma}^{\alpha}:\varphi^{\beta}\varphi^{\gamma}:,\quad [ { { d_{\mathrm{ne}}}}_{{\lambda}}\phi_{\alpha}]=\sum_{\beta\in{\delta_{\frac{1}{2}}}}\chi([{e_{\beta}},{e_{\alpha}}])\varphi^{\beta},\\ & [ { { d_{\mathrm{ne}}}}_{{\lambda}}\varphi^{\alpha}]=[{{d_{\chi}}}_{{\lambda}}\varphi^{\alpha}]=[{{d_{\mathrm{st}}}}_{{\lambda}}\phi_{\alpha}]=[{{d_{\chi}}}_{{\lambda}}\phi_{\alpha}]=0,\end{aligned}\ ] ] where @xmath147 for @xmath148 and @xmath149 is a projection map . since @xmath137 , @xmath150 .
moreover , @xmath151 for all @xmath152 , see @xcite .
therefore @xmath153 let @xmath154 be a vertex subalgebra of @xmath76 generated by @xmath138 for all @xmath139 and @xmath140 for all @xmath118 .
since @xmath155 and by the above formulas , the vertex superalgebra @xmath154 is a complex with the differential @xmath156 .
we have @xmath157 let @xmath158 be a vertex subalgebra of @xmath154 generated by @xmath138 for all @xmath139 and @xmath38 be a vertex subalgebra of @xmath158 generated by @xmath138 for all @xmath159
. then @xmath160 provided that @xmath161 , define @xmath162 for @xmath139 and @xmath120 .
[ qcal lemma ] suppose @xmath161 and replace @xmath1 by @xmath163 . then @xmath164={{\epsilon}}({\bar{j}}^{[u , v]}+\bar{\tau}_{{{\epsilon}}}(u|v){\lambda}),\quad [ { \varphi^{\alpha}}_{{\lambda}}{\bar{j}}^{u}]={{\epsilon}}\sum_{\beta\in{\delta_{>0}}}c_{u,\beta}^{\alpha}\varphi^{\beta},\\ & [ { { \bar{\phi}}}_{\alpha\hspace{0.1mm}{\lambda}}{\bar{\phi}}_{\beta}]={{\epsilon}}\cdot\chi([e_{\alpha},e_{\beta}]),\hspace{10 mm } [ { { \bar{j}}^{u}}_{{\lambda}}{{\bar{\phi}}}_{\alpha}]=[{\varphi^{\alpha}}_{{\lambda}}\varphi^{\beta}]=[{\varphi^{\alpha}}_{{\lambda}}{\bar{\phi}}_{\beta}]=0,\end{aligned}\ ] ] @xmath165=-\sum_{\begin{subarray}{c}\beta\in{\delta_{>0}}\\ \alpha\in i\cup\delta_{\leq 0}\end{subarray}}(-1)^{{p(\alpha)}}c_{u,\beta}^{\alpha}:{\bar{j}}^{{e_{\alpha}}}\varphi^{\beta}:+\sum_{\beta\in{\delta_{>0}}}(\bar{\mathit{a}}_{{{\epsilon}}}(u|{e_{\beta}}){\partial}+\bar{\mathit{b}}_{{{\epsilon}}}(u|{e_{\beta}}){\lambda})\varphi^{\beta},\end{aligned}\ ] ] @xmath166=\sum_{\begin{subarray}{c}\beta\in{\delta_{>0}}\\ \alpha\in{\delta_{\frac{1}{2}}}\end{subarray}}c_{u,\beta}^{\alpha}:{\bar{\phi}}_{\alpha}\varphi^{\beta}:,\hspace{29 mm } & [ { { d_{\chi}}}_{{\lambda}}{\bar{j}}^{u}]=\sum_{\beta\in{\delta_{>0}}}\chi([u,{e_{\beta}}])\varphi^{\beta},\\ [ { { d_{\mathrm{st}}}}_{{\lambda}}\varphi^{\alpha}]=-\frac{1}{2}\sum_{\beta,\gamma\in{\delta_{>0}}}(-1)^{{p(\alpha)}{p(\beta)}}c_{\beta,\gamma}^{\alpha}:\varphi^{\beta}\varphi^{\gamma}:,\quad & [ { { d_{\mathrm{ne}}}}_{{\lambda}}{\bar{\phi}}_{\alpha}]=\sum_{\beta\in{\delta_{\frac{1}{2}}}}\chi([{e_{\beta}},{e_{\alpha}}])\varphi^{\beta},\\ [ { { d_{\mathrm{ne}}}}_{{\lambda}}\varphi^{\alpha}]=[{{d_{\chi}}}_{{\lambda}}\varphi^{\alpha}]=[{{d_{\mathrm{st}}}}_{{\lambda}}{\bar{\phi}}_{\alpha}]=[{{d_{\chi}}}_{{\lambda}}{\bar{\phi}}_{\alpha}]=0,\hspace{5mm}\end{aligned}\ ] ] where @xmath167 replace @xmath168 in the @xmath108-brackets in lemma [ qcal lemma ] by an independent variable @xmath169 and denote by @xmath170 the vertex superalgebra over @xmath171 $ ] generated by @xmath172 for all @xmath139 , @xmath140 for all @xmath118 and @xmath173 for all @xmath120 .
the vertex superalgebra @xmath170 over @xmath171 $ ] is a cochain complex over @xmath171 $ ] with the differential @xmath131 by lemma [ qcal lemma ] .
define a vertex superalgebra @xmath174 by @xmath175}{\otimes}{\mathbb{c}}_{{{\epsilon}}}\end{aligned}\ ] ] for @xmath176 , where @xmath177 is a one - dimensional @xmath171$]-module defined by @xmath178 in @xmath177 .
the vertex superalgebra @xmath174 is also a complex with the differential @xmath131 . by construction , @xmath179 if @xmath180 .
let @xmath181 be a vertex subalgebra over @xmath171 $ ] in @xmath170 generated by @xmath172 for all @xmath139 and @xmath140 for all @xmath118 .
the vertex superalgebra @xmath181 over @xmath171 $ ] is a chain complex over @xmath171 $ ] with the differential @xmath156 by lemma [ qcal lemma ] and the formula @xmath155 .
denote @xmath182}{\mathbb{c}}_{{{\epsilon}}}$ ] .
thus , @xmath183 and @xmath184 is also a complex with the differential @xmath156 .
let @xmath185 then @xmath186 is a complex with the differential @xmath131 .
it has a supercommutative vertex superalgebra structure by lemma [ qcal lemma ] in the case of @xmath187 and analogously for @xmath188 and @xmath156 .
the vertex superalgebra @xmath186 is a poisson vertex superalgebra through the poisson @xmath108-bracket @xmath189_{{y}=0},\end{aligned}\ ] ] for @xmath190 , where the @xmath108-bracket is that of @xmath170 .
then @xmath186 is called the _ classical limit _ of @xmath76 and @xmath191 is called the _ classical @xmath4-algebra associated with @xmath0 , @xmath1 , @xmath16 and @xmath2_. the cohomology @xmath191 inherits the poisson vertex superalgebra structure from @xmath186 .
in this section we fix some notations that will be used in the sequel .
let @xmath192 which is a restricted root system ( @xcite ) , @xmath193 be the base of @xmath27 such that the positive part of @xmath27 coincides with a set of positive roots in @xmath27 and @xmath30 be the root lattice of @xmath17 i.e. @xmath194 .
set @xmath29 .
we define the equivalence relation in @xmath31 by @xmath195 for @xmath33 .
denote by @xmath34 $ ] the quotient set of @xmath31 and by @xmath196 $ ] the equivalent class of @xmath197 in @xmath34 $ ] .
[ pp dec lemma ] @xmath28 .
if there exists @xmath198 for @xmath199 , the positive root @xmath90 is not simple because @xmath200 as is shown in @xcite .
thus , @xmath90 is decomposed to two positive roots in @xmath201 , which contradicts that @xmath197 .
lemma [ pp dec lemma ] induces the decomposition @xmath37=[{\pi^{\gamma}}_{\frac{1}{2}}]\sqcup[{\pi^{\gamma}}_{1}].\end{aligned}\ ] ] let @xmath202}=\bigoplus_{\alpha\in[\beta]}{\mathbb{c}}x_{\alpha}\end{aligned}\ ] ] be the @xmath17-module defined by @xmath203}c_{\gamma , u}^{\alpha}x_{\gamma}\end{aligned}\ ] ] for @xmath159 , where the complex number @xmath204 is defined by @xmath205=\sum_{\alpha\in[\beta]}c_{\gamma , u}^{\alpha}e_{\alpha}$ ] .
then @xmath206}$ ] is well - defined .
we show that @xmath207\cdot x_{\alpha}\end{aligned}\ ] ] for all @xmath208 and @xmath209 $ ] .
thus , it is enough to show that @xmath210}(c_{\gamma , u}^{\rho}c_{\rho , v}^{\alpha}-(-1)^{p(u)p(v)}c_{\gamma , v}^{\rho}c_{\rho , u}^{\alpha})=c_{\gamma,[u , v]}^{\alpha}\end{aligned}\ ] ] for all @xmath211 $ ] .
by jacobi identity , @xmath212=[[e_{\gamma},u],v]-(-1)^{p(u)p(v)}[[e_{\gamma},v],u].\end{aligned}\ ] ] this implies . therefore the assertion follows .
let @xmath213 be the lie superalgebra cohomology with the coefficients in the trivial @xmath214-module @xmath215 .
then @xmath216 is naturally a @xmath17-module induced by the adjoint action of @xmath17 on @xmath214 , and @xmath217\in[{\pi^{\gamma}}]}{\mathbb{c}}^{[\beta]}\end{aligned}\ ] ] is the irreducilble @xmath17-module decomposition described in @xcite .
denote by @xmath218\oplus{\mathbb{c}}k$ ] the central extension of the loop superalgebra @xmath219 $ ] via the even supersymmetric invariant bilinear form @xmath40 defined by .
we extend the @xmath17-module @xmath206}$ ] to a @xmath220\oplus{\mathbb{c}}k$]-module by @xmath220t=0 $ ] and @xmath221 on @xmath206}$ ] .
define a @xmath222-module @xmath223}$ ] by @xmath224}&=\mathrm{ind}^{\hat{{\mathfrak{g}}}_{0}}_{{\mathfrak{g}}_{0}[t]\oplus{\mathbb{c}}k}{\mathbb{c}}^{[\beta ] } = u(\hat{{\mathfrak{g}}}_{0})\underset{u({\mathfrak{g}}_{0}[t]\oplus{\mathbb{c}}k)}{\otimes}{\mathbb{c}}^{[\beta]}\\ & \simeq{v^{\tau_{k}}({\mathfrak{g}}_{0})}\otimes\bigoplus_{\alpha\in[\beta]}{\mathbb{c}}x_{\alpha}.\end{aligned}\ ] ] then @xmath223}$ ] is naturally a @xmath38-module . [ lz lemma ] suppose that @xmath161 .
let @xmath225 be the field on @xmath38 defined by @xmath226 where @xmath227 is a basis of @xmath17 and @xmath228 is the dual basis of @xmath17 such that @xmath229 .
then @xmath225 is a virasoro field on @xmath38 .
consider the decomposition @xmath230 where @xmath231 is the center of @xmath17 and @xmath232 is a simple lie superalgebras for @xmath233 .
let @xmath234 for @xmath235 be the complex number defined by @xmath236 for @xmath237 .
set @xmath238 .
then , by definition , @xmath239 for @xmath237 and @xmath240 .
let @xmath241 be a basis of @xmath232 and @xmath242 be the dual basis of @xmath232 so that @xmath243 . by the sugawara construction
, we have a virasoro field on @xmath38 defined by @xmath244 therefore the assertion follows .
[ vtau rem ] there exists the canonical isomorphism @xmath245 where we follow the notation in the proof of lemma [ lz lemma ] .
[ lxalpha lemma ] @xmath246\\ \gamma\in i\cup\delta_{0}\end{subarray}}(-1)^{{p(\beta)}{p(\gamma)}}c_{\beta,-\alpha}^{\gamma}j^{e_{\gamma}}_{(-1)}x_{\beta}.\end{aligned}\ ] ] direct calculations .
we introduce a formal power series @xmath247 for @xmath197 , where its coefficients are the operators @xmath248}\end{aligned}\ ] ] defined by @xmath249 for all @xmath250 , where @xmath251 is the vertex operator of @xmath252 on @xmath253}$ ] .
the formal power series @xmath77 satisfies the following formulas for generic @xmath16 ( see definition [ generic def ] ) . [ screening prop ] if @xmath16 is generic , @xmath254=\sum_{\beta\in[\alpha]}c_{\beta , u}^{\alpha}s^{\beta}_{m+n},\\ & { \partial}s^{\alpha}(z)=-\frac{(k+h^{\vee})^{-1}}{(e_{\alpha}|e_{-\alpha})}\sum_{\begin{subarray}{c}\beta\in[\alpha]\\ \gamma\in i\cup\delta_{0}\end{subarray}}(-1)^{{p(\beta)}{p(\gamma)}}c_{\beta,-\alpha}^{\gamma}:j^{e_{\gamma}}(z)s^{\beta}(z):\end{aligned}\ ] ] for all @xmath159 and @xmath197 .
proposition [ screening prop ] will be proved in section [ vastr sec ] .
we define the _ screening operators _
@xmath255}:{v^{\tau_{k}}({\mathfrak{g}}_{0})}\otimes{f({\mathfrak{g}}_{\frac{1}{2}})}\rightarrow m_{[\beta]}\otimes{f({\mathfrak{g}}_{\frac{1}{2}})}\end{aligned}\ ] ] for @xmath35\in[{\pi^{\gamma}}]$ ] by @xmath256}=\sum_{\alpha\in[\beta]}\int : s^{\alpha}(z)\phi_{\alpha}(z):dz\quad\mathrm{if}\ [ \beta]\in[{\pi^{\gamma}}_{\frac{1}{2}}],\\ \label{screening eq2}&q_{[\beta]}=\sum_{\alpha\in[\beta]}\chi(e_{\alpha})\int s^{\alpha}(z)dz\hspace{8.5mm}\mathrm{if}\ [ \beta]\in[{\pi^{\gamma}}_{1}],\end{aligned}\ ] ] where we have set @xmath257 for a formal power series @xmath258 .
[ main1 thm ] if @xmath16 is generic , @xmath3 may be described as a vertex subalgebra in @xmath44 , which is the intersection of kernels of the screening operators : @xmath259\in[{\pi^{\gamma}}]}{\mathrm{ker}}\ q_{[\beta]}.\end{aligned}\ ] ] here the screening operators @xmath255}:{v^{\tau_{k}}({\mathfrak{g}}_{0})}\otimes{f({\mathfrak{g}}_{\frac{1}{2}})}\rightarrow m_{[\beta]}\otimes{f({\mathfrak{g}}_{\frac{1}{2}})}\end{aligned}\ ] ] for @xmath35\in[{\pi^{\gamma}}]$ ] are defined by and .
the proof of theorem [ main1 thm ] is given in section [ proof sec ] .
suppose that @xmath260 .
let @xmath48 be the heisenberg vertex algebra associated with @xmath55 and its non - degenerate bilinear form @xmath10 induced by @xmath47 , @xmath261 be the @xmath48-module with the highest weight @xmath262 and @xmath263 be the formal power series whose coefficients are operators from @xmath48 to @xmath261 defined by @xmath264 where @xmath265 is the shift operator determined by the following formulas @xmath266=0\quad{\mathrm{for}\ \mathrm{all}}\ n\neq0\end{aligned}\ ] ] for @xmath267 , where @xmath268 is the highest weight vector of @xmath261 .
[ main2 thm ] if @xmath16 is generic and @xmath260 , then @xmath269 may be described as a vertex subalgebra in @xmath58 as follows : @xmath270 where @xmath60 .
if @xmath16 is generic and @xmath260 , then @xmath271 and @xmath272 .
since @xmath30 is empty , the equivalence relation in @xmath31 defined by is trivial .
therefore @xmath273 by theorem [ main1 thm ] . the vertex algebra @xmath274 is generated by @xmath138 for all @xmath275
satisfying @xmath142=(k+h^{\vee})(u|v){\lambda}\end{aligned}\ ] ] for @xmath276 since @xmath277 we identify @xmath47 with @xmath55 via the non - degenerate bilinear form @xmath10 , that is , @xmath278 such that @xmath279 for all @xmath280 .
denote @xmath281 for @xmath262 , where @xmath60 . then @xmath282=(\beta|\gamma){\lambda}\end{aligned}\ ] ] for @xmath283 .
hence @xmath284 is the heisenberg vertex algebra @xmath48 associated with @xmath55 . we show @xmath285 for @xmath54 .
we have @xmath286=({e_{\alpha}}|e_{-\alpha})t_{\alpha}\quad\mathrm{for}\ \mathrm{all}\ \alpha\in\delta_{+}.\end{aligned}\ ] ] by the invariant properties of @xmath10 . using this formula and proposition [ screening prop ]
, we obtain the following relations : @xmath287=-\frac{1}{{\nu}}(\beta|\alpha)s^{\alpha}_{m+n},\\ & { \partial}s^{\alpha}(z)=-\frac{1}{{\nu}}:\alpha(z)s^{\alpha}(z):\end{aligned}\ ] ] for @xmath54 and @xmath262 .
this implies . therefore the assertion follows . for @xmath73 ,
there is a one - to - one correspondence between nilpotent orbits in @xmath8 and partitions of @xmath288 .
if @xmath1 is a nilpotent element corresponding to the partition @xmath289 with @xmath290 , there exists a good grading for @xmath1 such that @xmath291 ( see @xcite ) .
we define a degree map @xmath292 by @xmath293 let @xmath294 and @xmath295 . since @xmath296 for all @xmath297 , @xmath298 is a filtration of a complex @xmath76 .
thus , the filtration @xmath298 gives a spectral sequence @xmath299 .
we call @xmath298 a _ weight filtration _ on @xmath76
. define the conformal weight of @xmath76 inherited from @xmath127 , that is , @xmath300 where we denote by @xmath301 the conformal weight of @xmath302 .
let @xmath303 be the homogeneous conformal weight space with the conformal weight @xmath288 in @xmath76 .
we have @xmath304 since @xmath131 preserves each homogeneous conformal weight spaces , @xmath303 is a subcomplex of @xmath76 .
hence the filtration @xmath298 induces a filtration and a spectral sequence on the subcomplex @xmath303 .
since @xmath305 , the spectral sequence on @xmath76(n ) converges for all @xmath306 so that the total spectral sequence @xmath299 also converges .
denote by @xmath307 the differential of @xmath308 and @xmath309 by construction , @xmath310 and @xmath311 for @xmath312 .
[ e1 lemma ] @xmath313 .
the assertion follows from the fact that the differential @xmath156 acts only on the subcomplex @xmath154 and .
suppose that @xmath104 is a super space .
then we denote by @xmath314 the supersymmetric algebra of @xmath104 , which is the tensor algebra of @xmath104 quotient by the two - sided ideal generated by @xmath315 for all @xmath316 .
let @xmath317t^{-1})}$ ] be the supersymmetric algebra of @xmath318t^{-1}$ ] and @xmath213 be the lie superalgebra cohomology with the coefficients in the trivial @xmath214-module @xmath215 .
the chevalley complex of @xmath213 is the superexterior algebra @xmath319 .
we denote by @xmath320 the element of @xmath317t^{-1})}$ ] corresponding to @xmath321t^{-1}$ ] and by @xmath322 the element of @xmath319 corresponding to @xmath323 with reversed parity for @xmath118 .
we can regard @xmath317t^{-1})}$ ] and @xmath319 as subalgebras of the supercommutative algebra @xmath188 by @xmath324t^{-1})}\ni\ j^{u_{1}}(-n_{1})\cdots j^{u_{s}}(-n_{s})\ & \longmapsto\ { \bar{j}}^{u_{1}}_{(-n_{1})}\cdots{\bar{j}}^{u_{s}}_{(-n_{s})}\ket{0}\ \in{c'}_{\infty},\\ \lambda({\mathfrak{g}}_{+})\ni\hspace{8mm}\varphi^{\beta_{1}}(0)\cdots\varphi^{\beta_{s}}(0)\quad&\longmapsto\quad\varphi^{\beta_{1}}_{0}\cdots\varphi^{\beta_{s}}_{0}\ket{0}\hspace{7.5mm}\in{c'}_{\infty}\end{aligned}\ ] ] for @xmath325 and @xmath326 . [ cinf lemma ] @xmath327t^{-1})}\otimes h({\mathfrak{g}}_{+},{\mathbb{c}})$ ]
. the proof of lemma [ cinf lemma ] will be given in section [ cinf subsec ] .
[ cx cor ] @xmath328 $ ] .
let @xmath329 for all @xmath297 .
since @xmath330 for all @xmath297 , @xmath331 is a filtration of the complex @xmath181 over @xmath171 $ ] and gives a spectral sequence @xmath332 .
we extend the conformal weight on @xmath154 to @xmath181 by @xmath333 and get the homogeneous conformal weight decomposition of @xmath181 . since each of the homogeneous spaces is a free @xmath171$]-module of finite rank , the induced spectral sequence on it converges so that the total spectral sequence @xmath332 also converges .
denote by @xmath334 the differential of @xmath335 for @xmath336 .
it is straightforward that @xmath337.\end{aligned}\ ] ] by lemma [ cinf lemma ] , @xmath338t^{-1})}\otimes h({\mathfrak{g}}_{+},{\mathbb{c}})\end{aligned}\ ] ] and this implies @xmath339}{\otimes}h({\mathfrak{g}}_{+},{\mathbb{c}}[{y}]),\end{aligned}\ ] ] where @xmath340 is a vertex subalgebra over @xmath171 $ ] in @xmath181 generated by @xmath172 for all @xmath159 and @xmath341)$ ] is the lie superalgebra cohomology with the coefficients in the trivial @xmath214-module @xmath171 $ ] whose complex is identified with a subspace in @xmath181 spanned by all the elements @xmath342 with @xmath326 and @xmath343 over @xmath171 $ ] . since @xmath344)=0,\end{aligned}\ ] ] @xmath345 for all @xmath346 and then the spectral sequence collapses at @xmath347 .
therefore the assertion follows from .
the set @xmath348 is zariski dense in @xmath215 . by corollary [ cx cor ] , @xmath349}{\otimes}{\mathbb{c}}_{{{\epsilon}}}\simeq h({c'}_{\infty},{{{d_{\mathrm{st}}}}_{(0)}}).\end{aligned}\ ] ] suppose that @xmath350 .
if @xmath351}{\otimes}{\mathbb{c}}_{{{\epsilon}}},{{{d_{\mathrm{st}}}}_{(0)}})=h({\widehat{c}'}_{{y}},{{{d_{\mathrm{st}}}}_{(0)}})\underset{{\mathbb{c}}[{y}]}{\otimes}{\mathbb{c}}_{{{\epsilon}}}\end{aligned}\ ] ] holds for @xmath180 , it follows that @xmath352 because @xmath353}{\otimes}{\mathbb{c}}_{{{\epsilon}}}.\end{aligned}\ ] ] furthermore holds for @xmath187 because @xmath354}{\otimes}{\mathbb{c}}_{0}.\end{aligned}\ ] ] therefore @xmath355}{\otimes}{\mathbb{c}}_{{{\epsilon}}},{{{d_{\mathrm{st}}}}_{(0)}})=h({\widehat{c}'}_{{y}}(n),{{{d_{\mathrm{st}}}}_{(0)}})\underset{{\mathbb{c}}[{y}]}{\otimes}{\mathbb{c}}_{{{\epsilon}}}\end{aligned}\ ] ] holds for @xmath187 and all @xmath356 , where @xmath357 is the homogeneous conformal weight space of conformal weight @xmath288 in @xmath181 . since @xmath357 is a free @xmath171$]-module of finite rank , the upper semi - continuous theorem ( cf .
@xcite ) in a neighborhood of @xmath358 implies that holds for some neighborhood @xmath359 of @xmath187 in @xmath215 .
the complement @xmath360 is a finite set because @xmath359 is a non - empty zariski open set in @xmath215 .
hence @xmath361 is a countable set .
therefore a set @xmath362 is dense in @xmath215 .
the assertion follows from the fact @xmath363 because holds for @xmath364 .
[ generic def ] @xmath365 is called generic if @xmath366 .
if @xmath16 is generic , lemma [ cinf lemma ] implies that @xmath367 where the complex @xmath319 of @xmath213 is identified with the subspace of @xmath154 spanned by @xmath342 for all @xmath326 and @xmath343 .
[ e1 prop ] if @xmath16 is generic , @xmath368 by lemma [ e1 lemma ] , @xmath369 the assertion immediately follows from .
recall that @xmath31 is the set of all indecomposable roots in @xmath201 defined in .
[ dstpp lemma ] let @xmath197 .
then @xmath370\\ \gamma\in i\cup\delta_{0}\end{subarray}}(-1)^{{p(\gamma)}}c_{-\alpha,\beta}^{\gamma}:j^{{e_{\gamma}}}\varphi^{\beta}:+(e_{-\alpha}|e_{\alpha})(k+h^{\vee}){\partial}\varphi^{\alpha}\end{aligned}\ ] ] let @xmath197 .
we have @xmath371 where @xmath372
the indecomposablility of @xmath90 implies that @xmath373 $ ] if @xmath374 for @xmath375 and @xmath376 .
it is enough to show that @xmath377 first , @xmath378 where @xmath379 is the projection map .
next , we have @xmath380 by the indecomposablility of @xmath90 and the invariant property of @xmath10 .
thus , @xmath381 therefore the formula holds .
this completes the proof .
notice that the cohomology @xmath382 inherits a vertex superalgebra structure from @xmath154 .
thus , @xmath383 is a vertex superalgebra .
assume that @xmath16 is generic .
by , @xmath384 recall that we denote by @xmath322 the element of the complex @xmath319 of @xmath213 corresponding to @xmath385 with reversed parity for @xmath118 .
[ trivial lemma ] @xmath386 where @xmath387 is the non - zero cohomology class of @xmath322 in @xmath216 .
denote by @xmath388 be the differential of the complex @xmath319 of @xmath213 . since @xmath389 we obtain the first formula .
moreover @xmath390 and @xmath391 where @xmath392 is any total order in @xmath201 .
therefore @xmath393 if and only if @xmath90 is indecomposable in @xmath201 .
this implies the second formula . here ,
@xmath394 by lemma [ trivial lemma ] .
the first formula is also a vertex superalgebra isomorphism .
let @xmath395 be the vertex operator of @xmath387 on @xmath382 for @xmath197 .
[ salpha lemma ] the following formulas hold : @xmath396=\sum_{\beta\in[\alpha]}c_{\beta , u}^{\alpha}{\widetilde{s}}^{\beta},\\ \label{salpha eq3}&{\partial}{\widetilde{s}}^{\alpha}(z)=-\frac{(k+h^{\vee})^{-1}}{(e_{\alpha}|e_{-\alpha})}\sum_{\begin{subarray}{c}\beta\in[\alpha]\\ \gamma\in i\cup\delta_{0}\end{subarray}}(-1)^{{p(\beta)}{p(\gamma)}}c_{\beta,-\alpha}^{\gamma}:j^{e_{\gamma}}(z){\widetilde{s}}^{\beta}(z):\end{aligned}\ ] ] for @xmath197 and @xmath159 .
the following formulas hold in @xmath154 : @xmath397=\sum_{\beta\in[\alpha]}c_{\beta , u}^{\alpha}\varphi^{\beta}\end{aligned}\ ] ] for @xmath197 and @xmath159 .
these formulas imply and .
moreover @xmath398 for all @xmath197 in @xmath382 . since @xmath370\\ \gamma\in i\cup\delta_{0}\end{subarray}}(-1)^{{p(\gamma)}}c_{-\alpha,\beta}^{\gamma}:j^{{e_{\gamma}}}\varphi^{\beta}:+(e_{-\alpha}|{e_{\alpha}})(k+h^{\vee}){\partial}\varphi^{\alpha}\end{aligned}\ ] ] for all @xmath197 by lemma [ dstpp lemma ] , the formula @xmath399 holds if and only if @xmath400\\ \gamma\in i\cup\delta_{0}\end{subarray}}(-1)^{{p(\beta)}{p(\gamma)}}c_{\beta,-\alpha}^{\gamma}:j^{e_{\gamma}}(z)\varphi^{\beta}(z):\end{aligned}\ ] ] this implies and completes the proof . by lemma [ salpha lemma ]
, @xmath401 is a @xmath38-module and the operator @xmath402 restricts to the maps @xmath403}{\mathbb{c}}\psi_{\beta}\end{aligned}\ ] ] for all @xmath197 and @xmath404 . recall that @xmath253}$ ] is a @xmath38-module defined by .
[ malpha lemma ] let @xmath405\in[{\pi^{\gamma}}]}m_{[\alpha]}\end{aligned}\ ] ] be the vector space isomorphism defined by @xmath406 for all @xmath250 and @xmath197 .
then this is also a @xmath38-module isomorphism . by , @xmath407=\sum_{\beta\in[\alpha]}c_{\beta , u}^{\alpha}{\widetilde{s}}^{\beta}_{n+m}\end{aligned}\ ] ] for @xmath159 , @xmath197 , @xmath408
. therefore @xmath409}c_{\beta , u}^{\alpha}\psi_{\beta},\quad j^{u}_{(n)}\psi_{\alpha}=0\quad\mathrm{for}\ n\geq1\end{aligned}\ ] ] by .
this implies that is also a @xmath38-module isomorphism . by lemma [ malpha lemma ]
, we can regard the operators @xmath402 as the maps @xmath410}.\end{aligned}\ ] ] recall that @xmath77 is the formal power series whose coefficients are the operators from @xmath38 to @xmath253}$ ] defined in .
[ ts prop ] @xmath411 as the operator from @xmath38 to @xmath253}((z))$ ] for all @xmath197 . by the skew - symmetry in @xmath382 , @xmath412 for all @xmath250 , where @xmath251 is the vertex operator on @xmath382 and @xmath413 is the translation operator .
all we need to show is that the actions of @xmath414 and @xmath413 on @xmath253}$ ] are the same via the isomorphism .
however , @xmath415 for all @xmath250 and @xmath416 because @xmath225 is a virasoro field on @xmath38 by lemma [ lz lemma ] . therefore the first terms of and are the same .
thus , it is enough to show that @xmath417 via the isomorphism . by , @xmath418\\ \gamma\in i\cup\delta_{0}\end{subarray}}(-1)^{{p(\beta)}{p(\gamma)}}c_{\beta,-\alpha}^{\gamma}j^{e_{\gamma}}_{(-1)}\psi_{\beta}.\end{aligned}\ ] ] this coincides with @xmath419 by lemma [ lxalpha lemma ] .
so the proof is completed . by proposition
[ ts prop ] , @xmath420 hence the assertion follows from lemma [ salpha lemma ] .
assume that @xmath16 is generic .
by proposition [ e1 prop ] , @xmath421 since @xmath422 is the vertex operator on @xmath78 induced by @xmath423 , @xmath424 let @xmath425 be a vertex operator on @xmath78 defined by @xmath426 which is the vertex operator induced by @xmath427 .
[ e1q lem ] assume that @xmath16 is generic .
let @xmath428 be a vertex operator on @xmath78 defined by @xmath429 then @xmath430 and @xmath431 is a complex with respect to the charge degrees .
moreover , @xmath432 recall that @xmath433 . since @xmath434=[{{{d_{\chi}}}_{(0)}},{{{d_{\chi}}}_{(0)}}]=0 $ ] , @xmath435=\frac{1}{2}[d_{1}+{\widetilde{d}}_{2},d_{1}+{\widetilde{d}}_{2}]=\frac{1}{2}[d_{1},d_{1}]=d_{1}^{2}=0.\end{aligned}\ ] ] therefore @xmath431 is a complex with respect to the charge degrees .
the filtration on @xmath78 induced by the weight filtration on @xmath76 gives a convergent spectral sequence @xmath436 such that @xmath437 for all @xmath438 .
therefore @xmath439 this completes the proof .
recall @xmath440 by lemma [ e1q lem ] , @xmath441 notice that @xmath442\in[{\pi^{\gamma}}]}q_{[\beta]}\end{aligned}\ ] ] by definition in and , and that @xmath255}\cdot({v^{\tau_{k}}({\mathfrak{g}}_{0})}\otimes{f({\mathfrak{g}}_{\frac{1}{2}})})\subset m_{[\beta]}\otimes{f({\mathfrak{g}}_{\frac{1}{2}})}\end{aligned}\ ] ] for @xmath35\in[{\pi^{\gamma}}]$ ] .
this implies the decomposition of the kernel of the operator @xmath428 : @xmath443\in[{\pi^{\gamma}}]}{\mathrm{ker}}\ q_{[\beta]}.\end{aligned}\ ] ] the isomorphism is the vertex superalgebra isomorphism ( lemma [ miura lem ] below ) .
therefore this completes the proof .
define the @xmath444 $ ] action on the affine superspace @xmath445 $ ] by @xmath446-{\partial}a(t)\end{aligned}\ ] ] for @xmath447 $ ] and @xmath448 $ ] .
this action is induced by the derivation of the gauge action of the loop supergroup associated with @xmath444 $ ] .
this @xmath444 $ ] action on @xmath449 gives rise to the @xmath444$]-module structure on the space of the functions of @xmath449 .
identify the space of the functions of @xmath449 with the supersymmetric algebra of @xmath450t^{-1}$ ] , denoted by @xmath451t^{-1})}$ ] , via the _ residue dual _ , that is , a non - degenerate bilinear form @xmath452 defined by the following formula @xmath453 for @xmath454 , @xmath455 , @xmath456 $ ] and @xmath457t^{-1}$ ] .
then @xmath451t^{-1})}$ ] is a @xmath444$]-module .
set @xmath458t^{-1})}\end{aligned}\ ] ] for @xmath139 , @xmath459 and @xmath460\end{aligned}\ ] ] for @xmath461 , @xmath462 . here ,
@xmath444 $ ] acts on @xmath451t^{-1})}$ ] by @xmath463 for @xmath118 , @xmath139 , @xmath464 and @xmath465 , and @xmath466 acts on @xmath451t^{-1})}$ ] as a super derivation with a parity @xmath467 .
consider the semi - infinite cohomology ( cf .
@xcite ) of the @xmath444$]-module @xmath451t^{-1})}$ ] , denoted by @xmath468,{s({\mathfrak{g}}_{\leq0}[t^{-1}]t^{-1})}).\end{aligned}\ ] ] the complex of is @xmath469t^{-1})}\otimes\lambda(({\mathfrak{g}}_{+}[t])^{*}),\end{aligned}\ ] ] where @xmath470)^{*})$ ] is the superexterior algebra of @xmath471)^{*}$ ] .
let @xmath472 be a generator of @xmath470)^{*})$ ] associated with the element @xmath473)^{*}$ ] for @xmath118 and @xmath474 .
@xmath475t^{-1})})=h({c'}_{\infty},{{{d_{\mathrm{st}}}}_{(0)}})$ ] .
the map @xmath476t^{-1})}\otimes\lambda(({\mathfrak{g}}_{+}[t])^{*})$ ] defined by @xmath477 for @xmath139 , @xmath118 , @xmath438 , @xmath464 is a superalgebra isomorphism by construction .
the derivation on @xmath188 coincides with the derivation on @xmath451t^{-1})}\otimes\lambda(({\mathfrak{g}}_{+}[t])^{*})$ ] via this map by direct calculations .
therefore the assertion follows .
denote by @xmath478t^{-1})}\otimes\lambda(({\mathfrak{g}}_{+}[t])^{*})\end{aligned}\ ] ] the complex of and by @xmath479 the derivation of @xmath480 .
let @xmath481 be a subspace of @xmath480 spanned by all the elements @xmath482 with @xmath483 , @xmath326 , @xmath484 , @xmath485 and @xmath486 . since @xmath487 @xmath481 is a filtration of the complex @xmath480 and gives the spectral sequence @xmath488 on @xmath480 .
the conformal weight on @xmath480 is defined in the same way as @xmath154 . since
each of homogeneous conformal weight spaces is finite dimensional , the spectral sequence restricted to each of homogeneous conformal weight spaces converges so that @xmath488 also converges . set @xmath489 for @xmath306 .
we have @xmath490 for @xmath118 , @xmath159 , @xmath465 and @xmath464 . for
@xmath118 and @xmath438 , denote by @xmath491 a subalgebra of @xmath480 generated by @xmath492 and @xmath472 .
let @xmath493 be a subalgebra generated by @xmath322 for all @xmath118 .
then @xmath494t^{-1})}\otimes{\mathcal{c}}_{0}\otimes\bigotimes_{\begin{subarray}{c}\alpha\in{\delta_{>0}}\\ n\geq1\end{subarray}}{\mathcal{c}}_{\alpha , n}\end{aligned}\ ] ] is a decomposition of the complex @xmath480 by .
therefore @xmath495t^{-1})}\otimes{\mathcal{c}}_{0}\otimes\bigotimes_{\begin{subarray}{c}\alpha\in{\delta_{>0}}\\ n\geq1\end{subarray}}h({\mathcal{c}}_{\alpha , n},d_{{\mathcal{c}},0})\end{aligned}\ ] ] since @xmath496t^{-1})}=d_{{\mathcal{c}},0}\cdot{\mathcal{c}}_{0}=0 $ ] .
we show that it is sufficient to prove that @xmath497 for all @xmath118 and @xmath438 .
then @xmath498t^{-1})}\otimes{\mathcal{c}}_{0}.\end{aligned}\ ] ] thus , @xmath499t^{-1})}\otimes h({\mathcal{c}}_{0},d_{{\mathcal{c}},1})\end{aligned}\ ] ] since @xmath500t^{-1})}=0 $ ] .
we have @xmath501 for @xmath118 .
this implies @xmath502 since @xmath503 for @xmath346 , the assertion follows .
therefore we only need to show for all @xmath118 and @xmath438 .
if @xmath467 is even , @xmath492 is even and @xmath472 is odd .
therefore @xmath491 is spanned by all the elements @xmath504 with @xmath336 .
we have @xmath505 by and . hence follows .
if @xmath467 is odd , @xmath492 is odd and @xmath472 is even .
therefore @xmath491 is spanned by all the elements @xmath506 with @xmath336 .
we have @xmath507 by and . hence follows .
therefore this completes the proof .
in this section we consider the @xmath4-algebra @xmath3 at non - generic level .
recall that , following @xcite , there exists an ascending vertex superalgebra filtration on @xmath76 such that the induced filtration on @xmath508 gives an isomorphism @xmath509 where @xmath510 is the centralizer of @xmath1 in @xmath0 . hence if @xmath511 and @xmath512 is a basis of @xmath513 such that @xmath514 for all @xmath515 , where all @xmath516 s are some non - negative half - integers , then the @xmath4-algebra @xmath3 is strongly generated by @xmath517 for all @xmath515 with conformal weights @xmath518 since the differential @xmath131 preserves the conformal weights on @xmath76 .
thus , @xmath3 is realized as a vertex subalgebra of @xmath519 generated by @xmath520 for all @xmath515 .
let @xmath521 be the vertex superalgebra homomorphism defined by @xmath522 for all @xmath523 and @xmath524 , and @xmath525 on @xmath44 .
this map @xmath526 induces the vertex superalgebra homomorphism @xmath527 by restriction , where we regard the @xmath4-algebra @xmath3 as a vertex subalgebra of @xmath528 .
this map @xmath529 is called _ the miura map_. [ miura lem ] the image of the isomorphism in theorem [ main1 thm ] coincides with that of the miura map @xmath529 .
assume that @xmath16 is generic .
by lemma [ e1q lem ] , @xmath530 which is the isomorphism .
this is the vector isomorphism induced by the weight filtration on @xmath76 . by proposition [ e1 prop ] , @xmath531 this implies that the leading terms of the cohomology classes in @xmath532 concentrate on the homogeneous space with degree @xmath39 .
therefore the isomorphism is the projection of the homogeneous terms with degree @xmath39 in the cohomology classes in @xmath532 .
hence the image of coincides with that of the miura map by definition .
this completes the proof .
[ miura cor ] if @xmath16 is generic , the miura map @xmath529 is injective and there exist strongly generating fields @xmath533 of the vertex subalgebra @xmath534 of @xmath44 such that @xmath535 for all @xmath515 .
though the following assertion is proved in @xcite for regular nilpotent elements @xmath1 , the same proof applies for arbitrary @xmath1 ( see section 5.9 of @xcite for the details ) .
[ miura1 lem ] the miura map @xmath529 is injective for all @xmath16 if @xmath0 is a simple lie algebra .
we note that the analogue of lemma [ miura1 lem ] for the classical @xmath4-algebras is proved in @xcite and this can be used to give an yet another proof of lemma [ miura1 lem ] .
recall that @xmath170 is a vertex superalgebra over @xmath171 $ ] generated by @xmath536 for all @xmath537 and also a chain complex over @xmath171 $ ] with the differential @xmath131 ( see section [ classical subsec ] for definitions ) .
let @xmath538 be the vertex subalgebras of @xmath170 over @xmath171 $ ] generated by @xmath539 for all @xmath540 . by kac and wakimoto arguments in the case of @xmath76 ,
one finds that the cohomology @xmath541 is a free @xmath171$]-module vertex superalgebra generated by @xmath542 for all @xmath515 in @xmath543 and we call it _ the @xmath4-algebra over @xmath171 $ ] _ denoted by @xmath544 by construction , if @xmath161 , @xmath545}{\otimes}{\mathbb{c}}_{{{\epsilon}}},\end{aligned}\ ] ] where @xmath177 is a one - dimensional @xmath171$]-module defined by @xmath178 and @xmath180
. we can also define the miura map @xmath546 in the same way and we have @xmath547}{\otimes}\mathrm{ev}_{{y}={{\epsilon}}}\end{aligned}\ ] ] for all @xmath14 , where @xmath548 is the evaluation map defined by @xmath549 .
[ miura2 lem ] the miura map @xmath550 is injective .
we extend the weight filtration on @xmath76 to @xmath170 by @xmath551 .
this filtration gives the convergent spectral sequence @xmath552 on @xmath170 . by construction , @xmath553 by corollary [ cx cor ] , @xmath554).\end{aligned}\ ] ] therefore , in the same way as theorem [ main1 thm ] , one finds that the @xmath4-algebra @xmath555 over @xmath171 $ ] is isomorphic to the vertex subalgebra over @xmath171 $ ] of @xmath556 , which is the intersection of kernels of the screening operators : @xmath557\in[{\pi^{\gamma}}]}{\mathrm{ker}}q_{[\beta],{y}},\end{aligned}\ ] ] where @xmath558,{y}}=\sum_{\alpha\in[\beta]}\int : s^{\alpha}_{{y}}(z){\bar{\phi}}_{\alpha}(z):dz\quad\mathrm{if}\ [ \beta]\in[{\pi^{\gamma}}_{\frac{1}{2}}],\\ & q_{[\beta],{y}}=\sum_{\alpha\in[\beta]}\chi(e_{\alpha})\int s^{\alpha}_{{y}}(z)dz\hspace{8.5mm}\mathrm{if}\ [ \beta]\in[{\pi^{\gamma}}_{1}],\end{aligned}\ ] ] and @xmath559 is defined in proposition [ screening prop ] by replacing @xmath180 by @xmath169 .
this implies that the miura map @xmath550 is injective .
the following lemma is useful in applications .
[ miura3 lem ] let @xmath560 for all @xmath515 .
then @xmath561 for all @xmath14 .
compare the leading term of @xmath562 with that of @xmath520 .
the assertion follows from the definition @xmath563 .
let @xmath564 and @xmath47 be a cartan subalgebra of a simple lie algebra of type @xmath565 with the non - degenerate symmetric bilinear form @xmath10 such that @xmath92 . denote by @xmath566 a set of simple roots of type @xmath565 such that the simple root @xmath567 is short .
let @xmath48 be the heisenberg vertex algebra associated with @xmath55 i.e. the vertex algebra @xmath48 is generated by the even fields @xmath568 for all @xmath569 satisfying @xmath570=(\alpha_{i}|\alpha_{j}){\lambda},\end{aligned}\ ] ] and @xmath571 be the vertex superalgebra generated by the odd field @xmath572 satisfying @xmath573=1.\end{aligned}\ ] ] choose @xmath574 .
let @xmath575 be an odd field on @xmath576 defined by @xmath577 where @xmath578 for @xmath569 , and @xmath579 is an even field on @xmath576 for @xmath580 defined by @xmath581=w_{0}+\sum_{i=1}^{n-1}\gamma_{i}(w_{2i-1}\frac{{\lambda}^{2i-1}}{(2i-1)!}+w_{2i}\frac{{\lambda}^{2i}}{(2i)!})+\gamma_{n}\frac{{\lambda}^{2n}}{(2n)!},\end{aligned}\ ] ] where @xmath582 [ wb0 lem ] let @xmath583 be the subspace of @xmath576 spanned by @xmath584 for all @xmath585
. then @xmath581\ \equiv\ w_{0}+\sum_{i=1}^{n-1}\gamma_{i}\frac{w_{2i}{\lambda}^{2i}}{(2i)!}+\gamma_{n}\frac{{\lambda}^{2n}}{(2n)!}\quad\pmod{c_{2}({\mathcal{h}}\otimes{\mathcal{f}})}\end{aligned}\ ] ] and @xmath586 for all @xmath587 .
notice that @xmath588=\delta_{i , j}{\lambda}\end{aligned}\ ] ] for all @xmath589 . if @xmath69 , @xmath581 & = \ : b_{1}^{2}:+\gamma{\partial}b_{1}+:({\partial}\psi)\psi:+(1 - 2\gamma^{2})\frac{{\lambda}^{2}}{2}\\ & \equiv\ : b_{1}^{2}:+(1 - 2\gamma^{2})\frac{{\lambda}^{2}}{2}\quad\pmod{c_{2}({\mathcal{h}}\otimes{\mathcal{f}})}.\end{aligned}\ ] ] let @xmath590 and @xmath591 then @xmath592 by inductions on @xmath288 , we can assume that @xmath593 satisfies our assertions . hence
@xmath594\ \equiv\ w'_{0}+\sum_{i=1}^{n-2}\gamma_{i}\frac{w'_{2i}{\lambda}^{2i}}{(2i)!}+\gamma_{n-1}\frac{{\lambda}^{2n-2}}{(2n-2)!}\quad\pmod{c_{2}({\mathcal{h}}\otimes{\mathcal{f}})}\end{aligned}\ ] ] and @xmath595 therefore @xmath596\equiv\ -\gamma^{2}{\lambda}^{2}[g'_{{\lambda}}g']-\gamma{\lambda}[g'_{{\lambda}}:b_{1}g':]+\gamma{\lambda}[g'_{{\lambda}}:b_{1}g':]+[:b_{1}g':_{{\lambda}}:b_{1}g':]\\ & \hspace{95mm}\pmod{c_{2}({\mathcal{h}}\otimes{\mathcal{f}})}\\ & \equiv\ : b_{1}^{2}w'_{0}:+\sum_{i=1}^{n-2}\gamma_{i}\frac{(:b_{1}^{2}w'_{2i}:+w'_{2i-2}){\lambda}^{2i}}{(2i)!}+\gamma_{n-1}\frac{(:b_{1}^{2}:+w'_{2n-4}){\lambda}^{2n-2}}{(2n-2)!}+\gamma_{n}\frac{{\lambda}^{2n}}{(2n)!}\\ & \hspace{95mm}\pmod{c_{2}({\mathcal{h}}\otimes{\mathcal{f}})}.\end{aligned}\ ] ] furthermore @xmath597 for @xmath598 by our assumptions .
this completes the proof .
the vertex subalgebra of @xmath576 generated by the odd field @xmath575 and the even fields @xmath599 for @xmath587 is _ the @xmath7-algebra for @xmath574 _ introduced by fateev and lukyanov @xcite .
let @xmath600 such that @xmath601 and @xmath602 .
define the operators @xmath603 for @xmath569 by @xmath604 where @xmath263 for @xmath262 is defined in section [ main thm sec ] . according to @xcite , @xmath605 for generic @xmath606 .
it is enough to show that @xmath610 for all @xmath515 .
let @xmath611 for @xmath569 and @xmath612 .
then , for @xmath613 , @xmath614_{{\lambda}=0 } = \ : ( \gamma{\partial}+b_{1})\cdots(\gamma_{+}+b_{i-1})[{s^{\alpha_{i}}}_{{\lambda}}:(\gamma{\partial}+b_{i})(\gamma{\partial}+b_{i+1})g_{i+2}:]_{{\lambda}=0}:\end{aligned}\ ] ] and @xmath615_{{\lambda}=0}\\ & = \gamma\gamma_{+}:{\partial}s^{\alpha_{i}}g_{i+2}:-\gamma_{+}:s^{\alpha_{i}}b_{i+1}g_{i+2}:+\gamma_{+}:b_{i}s^{\alpha_{i}}g_{i+2}:\\ & = \gamma_{+}:(\gamma{\partial}s^{\alpha_{i}}-\gamma_{+}{\partial}s^{\alpha_{i}}-b_{i+1}s^{\alpha_{i}}+b_{i}s^{\alpha_{i}})g_{i+2}:\\ & = \ : ( -{\partial}s^{\alpha_{i}}+\gamma_{+}\alpha_{i}s^{\alpha_{i}})g_{i+2}:\\ & = 0.\end{aligned}\ ] ] hence @xmath616 moreover @xmath617=\ : ( \gamma{\partial}+b_{1})\cdots(\gamma_{+}+b_{n-1})[:s^{\alpha_{n}}\psi:_{{\lambda}}:(\gamma{\partial}+b_{n})\psi:]_{{\lambda}=0}:\end{aligned}\ ] ] and @xmath618_{{\lambda}=0 } & = \gamma{\partial}s^{\alpha_{n}}-\gamma_{+}{\partial}s^{\alpha_{n}}+:b_{n}s^{\alpha_{n}}:\ = 0.\end{aligned}\ ] ] therefore @xmath619 this completes the proof .
let @xmath622 and @xmath623 be a set of simple roots of @xmath0 such that @xmath624 is an odd root and @xmath625 is a set of simple roots of the even part of @xmath0 .
note that the even part of @xmath0 is the symplectic lie algebra @xmath626 of rank @xmath288 .
let @xmath63 be a regular nilpotent element of the even part of @xmath0 defined by @xmath627 and @xmath628 be a @xmath75-triple .
we set @xmath629 and denote by @xmath630 the normalized even supersymmetric non - degenerate invariant bilinear form on @xmath0 .
notice that the dual coxeter number @xmath15 of @xmath0 is equal to @xmath631 .
a semisimple element @xmath84 defines the dynkin grading @xmath2 on @xmath0 with respect to @xmath63 denoted by @xmath632 where @xmath22 is the eigenspace of @xmath633 with an eigenvalue @xmath134 .
then @xmath634 where @xmath635 is a cartan subalgebra of @xmath0 containing @xmath84 .
the subspace @xmath42 of @xmath0 is the one - dimensional odd space spanned by @xmath636 .
the neutral free superfermion vertex superalgebra associated with @xmath42 is isomorphic to @xmath571 since @xmath637)\neq0 $ ] .
let @xmath638 be the heisenberg vertex algebra associated with @xmath635 .
if @xmath16 is generic , then by theorem [ main2 thm ] , the vertex superalgebra @xmath639 is isomorphic to the vertex subalgebra of @xmath640 , which is the intersection of kernels of the screening operators : @xmath641 where @xmath642 .
let @xmath643 then @xmath644 for all @xmath589 and the vertex algebra @xmath638 is isomorphic to @xmath48 .
moreover @xmath645 for all @xmath515 , where @xmath646 . therefore our assertion follows .
[ wbn thm ] if @xmath647 , @xmath648 where @xmath649 , @xmath650 , @xmath63 is a regular nilpotent element of the even part of @xmath6 and @xmath2 is the dynkin grading of @xmath63 .
moreover the fields @xmath651 are strongly generating fields of @xmath652 with conformal weights @xmath653 and @xmath654 .
let @xmath655 if @xmath16 is generic , by lemma [ wb2 lem ] , @xmath656 where @xmath621 , which describes the image of the miura map @xmath529 ( see section [ miura sec ] for the definition ) .
the conformal weights of @xmath576 induced by this isomorphism is determined by @xmath657 by lemma [ wb1 lem ] , the odd vector @xmath607 and @xmath608 for @xmath587 belong to @xmath658 and @xmath659 by definition , the @xmath7-algebra is generated by @xmath575 and @xmath599 for all @xmath515 as a vertex subalgebra in @xmath576 . since @xmath599 has a leading term @xmath660 for @xmath587 by lemma [ wb0 lem ] , there exists a set of strongly generating fields of the @xmath7-algebras such that @xmath575 and @xmath599 are included in this set .
let @xmath661 be the even part of @xmath0 and @xmath662 be the odd part of @xmath0 .
denote by @xmath513 the centralizer of @xmath1 in @xmath0 .
then @xmath663 since @xmath664 and @xmath1 is a regular nilpotent element of @xmath661 , @xmath665 and there exists a basis @xmath666 of @xmath667 such that @xmath668 for @xmath569 , where @xmath516 is called the @xmath515-th exponent of @xmath626 .
a subspace @xmath669 of @xmath513 is a one - dimensional odd space spanned by @xmath670 where we follow the notation in the proof of lemma [ wb2 lem ] .
therefore , by corollary [ miura cor ] , the vertex superalgebra @xmath3 is generated by an odd field @xmath593 and even fields @xmath671 with conformal weights @xmath672 for @xmath569 .
therefore , for generic @xmath16 , we can choose @xmath671 and @xmath593 such that @xmath673 hence the isomorphism implies that the miura map @xmath674 is an isomorphism for generic @xmath16 .
since the equations hold not only for generic @xmath16 but also for @xmath64 by lemma [ miura3 lem ] , the miura map is also an isomorphism for @xmath64 .
this completes the proof . for @xmath675 , the @xmath4-algebra @xmath652 is isomorphic to the vertex subalgebra of @xmath576 generated by @xmath575
. therefore the feigin - frenkel duality @xmath676 holds for this case , where @xmath677 .
let @xmath678 , @xmath365 such that @xmath679 and @xmath104 be a vector space spanned by @xmath680 with non - degenerate @xmath215-bilinear form @xmath681 defined by the following gram matrix : @xmath682 let @xmath48 be the heisenberg vertex algebra associated with @xmath104 and @xmath683 , and @xmath684 be the @xmath48-module with the highest weight @xmath685 for @xmath686 .
denote by @xmath687 the highest weight vector of @xmath684 .
then @xmath688 has a vertex algebra structure such that the vertex operator of @xmath687 is defined by @xmath689 for all @xmath686 .
since @xmath690 , the parity of @xmath691 is even for all @xmath686 and the following formula holds : @xmath692=0\end{aligned}\ ] ] for all @xmath693 .
the @xmath9-algebra is defined as a vertex subalgebra of @xmath694 generated by @xmath695 where @xmath696 we remark that we should substitute @xmath697 for @xmath571 after expanding @xmath697 formally .
for example , if @xmath698 , @xmath699 let @xmath700 by a result in @xcite , @xmath701 for generic @xmath16 . for @xmath698 ,
the vertex algebra @xmath702 is isomorphic to the universal affine vertex algebra associated with @xmath75 at level @xmath16 and for @xmath71 , the vertex algebra @xmath72 is isomorphic to the bershadsky - polyakov algebra ( @xcite ) .
assume @xmath709 .
let @xmath710 be the standard basis of @xmath711 and @xmath712 for @xmath613
. denote by @xmath713 the special linear lie algebra of rank @xmath288 .
then @xmath714 is a basis of the cartan subalgebra @xmath47 .
let @xmath715 be the dual element of @xmath716 for @xmath569 and @xmath717 for @xmath718 .
then @xmath719 is a root system of @xmath0 and @xmath720 is a set of simple roots of @xmath18 .
choose a root vector @xmath721 .
define a subregular nilpotent element @xmath74 of @xmath0 by @xmath722 and a semisimple element @xmath84 of @xmath0 by @xmath723 then @xmath633 defines a good grading @xmath2 with respect to @xmath1 .
the corresponding weighted dynkin diagram is the following : @xmath724 let @xmath22 be the homogeneous subspace of @xmath0 with degree @xmath134
. then @xmath725 denote by @xmath27 , @xmath31 , @xmath34 $ ] the restricted root system associated with @xmath2 , a base of @xmath27 , a quotient set of @xmath31 respectively , as defined in section [ screening sec ] .
we have @xmath726=\{[\alpha_{2}],\ldots,[\alpha_{n-1}]\},\end{aligned}\ ] ] where @xmath727=\{\alpha_{1}+\alpha_{2},\ \alpha_{2}\},\quad [ \alpha_{i}]=\{\alpha_{i}\}\quad\mathrm{for}\ i>2.\end{aligned}\ ] ] let @xmath10 be the non - degenerate symmetric invariant bilinear form on @xmath0 defined by @xmath728 for @xmath122 . by theorem [ main1 thm ] , the vertex algebra @xmath729 is isomorphic to the vertex subalgebra of @xmath38 , which is the intersection of kernels of the screening operators : @xmath730 for generic @xmath16
. let @xmath731 and @xmath732 be th fields on @xmath38 defined by @xmath733 it is enough to show that @xmath735_{{\lambda}=0}=[{s^{\alpha_{i}}}_{{\lambda}}f]_{{\lambda}=0}=0\end{aligned}\ ] ] for all @xmath736 . by proposition [ screening prop ] , @xmath737=\alpha(h)s^{\alpha}_{m+n}\quad\mathrm{for}\ \mathrm{all}\ \alpha\in{\pi^{\gamma}},\ h\in{\mathfrak{h}},\\ & [ s^{\alpha_{1}+\alpha_{2}}_{m } , j^{e_{\alpha_{1}}}_{(n)}]=\ s^{\alpha_{2}}_{m+n},\quad [ s^{\alpha}_{m } , j^{e_{\alpha_{1}}}_{(n)}]=0\quad\mathrm{for}\ \mathrm{all}\ \alpha\neq\alpha_{1}+\alpha_{2},\\ & [ s^{\alpha_{2}}_{m } , j^{e_{-\alpha_{1}}}_{(n)}]=\ s^{\alpha_{1}+\alpha_{2}}_{m+n},\quad [ s^{\alpha}_{m } , j^{e_{-\alpha_{1}}}_{(n)}]=0\quad\mathrm{for}\ \mathrm{all}\ \alpha\neq\alpha_{2}\end{aligned}\ ] ] and @xmath738 therefore @xmath735_{{\lambda}=0}=0\end{aligned}\ ] ] for all @xmath515 .
next , we show @xmath739_{{\lambda}=0}=0 $ ] for all @xmath515 .
let @xmath740 and @xmath741 for @xmath742 , where @xmath743 for all @xmath134 .
first , @xmath744_{{\lambda}=0 } = \ : ( ( k+n-1){\partial}+j^{b_{n-1}}(z))\cdots((k+n-1){\partial}+j^{b_{i+1}}(z))[{s^{\alpha_{i}}}_{{\lambda}}f_{i}]_{{\lambda}=0}:.\end{aligned}\ ] ] therefore it is enough to show that @xmath745_{{\lambda}=0}=0\end{aligned}\ ] ] for all @xmath515 . for @xmath746 , we have @xmath747_{{\lambda}=0}=&\ ( k+n-1){\partial}s^{\alpha_{1}+\alpha_{2}}+:j^{b_{2}}s^{\alpha_{1}+\alpha_{2}}:+:s^{\alpha_{2}}j^{e_{-\alpha_{2}}}:\\ = & \
( k+n){\partial}s^{\alpha_{1}+\alpha_{2}}+:(j^{b_{2}}s^{\alpha_{1}+\alpha_{2}}+j^{e_{-\alpha_{2}}}s^{\alpha_{2}}):\\ = & \ 0.\end{aligned}\ ] ] for @xmath748 , since @xmath749_{{\lambda}=0}=0 $ ] , we obtain that @xmath745_{{\lambda}=0}=&\ [ { s^{\alpha_{i}}}_{{\lambda}}:((k+n-1){\partial}+j^{b_{i}})((k+n-1){\partial}+j^{b_{i-1}})f_{i-2}:]_{{\lambda}=0}\\ = & -(k+n-1):{\partial}s^{\alpha_{i}}f_{i-2}:-:j^{b_{i}}s^{\alpha_{i}}f_{i-2}:+:s^{\alpha_{i}}j^{b_{i-1}}f_{i-2}:\\ = & \ -:((k+n){\partial}s^{\alpha_{i}}+j^{h_{i}}s^{\alpha_{i}})f_{i-2}:\\ = & \ 0.\end{aligned}\ ] ] this completes the proof .
define a vertex algebra homomorphism @xmath751 by @xmath752 then @xmath753 consider the lie algebra decomposition @xmath754 where @xmath75 is the subalgebra of @xmath17 spanned by @xmath755 and @xmath756 is the center of @xmath17 . if we restrict @xmath757 to the vertex subalgebra @xmath758 of @xmath38 , the map @xmath757 coincides with the vertex algebra homomorphism of the wakimoto construction of @xmath758 , which is injective ( @xcite ) . since @xmath759 ( see remark [ vtau rem ] ) and @xmath760 is the heisenberg vertex algebra associated with @xmath756 , @xmath757 is injective and we can regard @xmath38 as a vertex subalgebra of @xmath694 via @xmath757 .
then @xmath761 acts on @xmath38 as @xmath762 therefore , for generic @xmath16 , @xmath763 because @xmath764 for all @xmath736 .
moreover @xmath765 and these are generating fields of @xmath9 for all @xmath679 . hence @xmath766 induces a vertex algebra isomorphism @xmath767 for generic @xmath16 . by lemma [ miura1 lem ] , this isomorphism holds for all @xmath679 . | let @xmath0 be a simple finite - dimensional lie superalgebra with a non - degenerate supersymmetric even invariant bilinear form , @xmath1 a nilpotent element in the even part of @xmath0 , @xmath2 a good grading of @xmath0 for @xmath1 and @xmath3 the ( affine ) @xmath4-algebra associated with @xmath5 defined by the generalized drinfeld - sokolov reduction . in this paper , we present each @xmath4-algebra as the intersection of kernels of the screening operators , acting on the tensor vertex superalgebra of an affine vertex superalgebra and a neutral free superfermion vertex superalgebra . as applications ,
we prove that the @xmath4-algebra associated with a regular nilpotent element in @xmath6 is isomorphic to the @xmath7-algebra introduced by fateev and lukyanov , and that the @xmath4-algebra associated with a subregular nilpotent element in @xmath8 is isomorphic to the @xmath9-algebra introduced by feigin and semikhatov . |
radiation condition onboard leo spacecrafts is determined , particularly , by the charged particles penetrating from outside of the magnetosphere .
the measure of such a process in any given near - earth space location is a local geomagnetic cutoff rigidity value . in this work we assume , that `` cutoff rigidity '' phrase denotes an effective vertical cutoff rigidity ( @xmath0 ) , which value is gained from calculated discrete penumbra structure by ordinary technique using white spectra ( for example , @xcite ) .
it is well known , that magnitude of @xmath0 and penetration boundaries position depend on local time and different magnetospheric condition - related parameters @xcite .
cutoff rigidity value s variations can be measured in experiments or obtained by some type of computation using earth s magnetosphere model . however , usual method for @xmath0 calculation is based on resource consuming trajectory computation technique @xcite . in this work
we offer the method to calculate @xmath0 with accounting for local time and main empirical parameters , that characterizes magnetospheric condition , in any point of near - earth s space . for the applications where @xmath0 calculation is often needed ( e.g. transmissions for leos ) our method provides a huge speedup in comparison with direct trajectory computations .
the only things needed for the method are the igrf rigidity in exploring point and rigidity attenuation @xcite quotient s dependence on empirical parameters .
the first one can be obtained using interpolation in @xmath1 tabulated data ( see , for example , @xcite ) , which is regularly updated for coming epochs . and the second can be obtained only once by using particle trajectory computations in corresponding numerical magnetospheric model . in our case
the dynamic paraboloid model @xcite of earth s magnetosphere was chosen for solution of the problem .
we have adopted in our work the cutoff rigidity attenuation formalism @xcite , which has already successfully applied @xcite for transmission calculations using tsyganenko-89 @xcite model .
the rigidity attenuation quotient @xmath2 in our case is calculated as @xmath3 where @xmath4 value corresponds the set of model parameters for non - disturbed magnetosphere condition , and @xmath5 is correspond to some set of varied parameters regarding to quiet ones . as it was wordlessly postulated in @xcite , the @xmath2 dependence on point s geographical position
might not be taken into account because it is relatively small .
generally , it seems to be true at least for @xmath6 gv with good enough accuracy ( error s order is about percents even for lowest @xmath7 in this range ) .
hence , the approach of calculating a world - wide @xmath0 grid can be essentially simplified .
the algorithm we propose for obtaining @xmath8 in given point contains 3 stages : * calculating @xmath1 for the given point ; for practical needs , the interpolation in pre - calculated grid for some altitude @xmath9 is often used , it is especially easy when large amount of results is needed quickly .
the formula @xmath10 can be applied to transform @xmath1 value to new altitude @xmath11 . * calculating a model - dependent cutoff rigidity attenuation quotient + @xmath12 by interpolation in pre - calculated @xmath13 database ( or by extrapolation if @xmath1 value is out of applied basic @xmath1 array , see appendix ) * computing the resulting value @xmath14 table of @xmath1 values for some given altitude @xmath9 is needed to be renewed every 5 years with new igrf epoch coming .
values @xmath4 were obtained by trajectory computations using dynamic paraboloid model with parameter set @xmath15 $ ] ( sequence of variables is according to table [ t1 ] below ) . to compute @xmath13 database which is intended to provide dependence of @xmath16 on parameters and local time @xmath17 , we used the varying of all model parameters in wide enough ranges , comprising extreme , quiet and `` anti - extreme '' parameter sets .
the multidimensional grid , applied for this calculation , is presented in table [ t1 ] .
.the applied parameter grid . [ cols="^,^,^,^,^ " , ] values at the edges of this grid were obtained ( approximately ) from @xcite .
all calculations were performed for local time values @xmath18 and @xmath19 hours .
technical details of calculations are summarized in appendix .
let us now demonstrate some results of cutoff rigidity dependence on local time and model parameters , that have been computed using presented technique .
fig.[f1 ] gives an opportunity to weigh deposit of every model parameter in resulting @xmath0 value for all points of noted above set .
such case of partial dependencies is obtained by fixating all parameters as quiet and excepting the given one .
note , that the role of @xmath20 as @xmath0 depression factor wanes with decreasing of @xmath0 , but the @xmath21 one rises rapidly .
the weight of @xmath22 and @xmath23 seems to be not important for @xmath0 values , at any hand , in presented case of quiet basic values , because they result in @xmath0 changes that are comparable with the value of applied rigidity step size ( see appendix ) .
contrary , decrease of @xmath24 effects in huge @xmath0 grow ( see also fig.[f2 ] ) .
the effect of @xmath25 and @xmath26 here is clear but of moderate magnitude .
fig.[f2 ] shows family of @xmath0 daily variations for three points having normal cutoff values of different order . for the first picture
there are some hours where @xmath0 is negligibly small , so such a point can be accessible for particles of any rigidity .
the middle picture demonstrates the effect of @xmath24 `` anti - extreme '' setting , leading not only to very high @xmath0 mean values ( for this case ) , but also to the translocation of curve s maximum for time axis s positive direction .
the upper curve on the most right part of fig.[f2 ] exhibits the insufficiency of applied step size for penumbra calculation ( see appendix ) , leading to the underestimation of @xmath0 value for @xmath27 , which causes an ambiguous interpolation results .
however , this effect is rare and too weak to affect resulting transmission , for which calculating presented method was created .
a `` gnarlyness '' of some presented curves is due to effect of scarce grid , currently imperfect interpolation procedure and scale .
it is does not sufficiently affect the results of applications where much of @xmath0 values to be calculated , because interpolation uses many @xmath13 values and their errors ( which are of order of @xmath7 step size ) are to be evened .
extrapolating parameter values to out of grid ( but not very far ) is also possible and seems to be correct enough to get reliable result .
magnetospheric transmission functions are directly connected with such tasks as leo radiation conditions evaluating and the interpretation of orbital experiment results @xcite .
because of their importance , it is necessary to be able to account for more empiric factors that affecting transmission . here
we ll give some examples of calculated magnetospheric transmission functions for several ( quiet , disturbed and very extreme ) model parameters sets for iss - like orbit , obtained using presented technique . the essential note
is that , unlike tsyganenko models , the paraboloid one is primordially intended to describe even very extreme conditions , as it is not internally limited by experimental data arrays .
fig.[f3 ] presents some modeled transmissions for @xmath28 km @xmath29 circular orbit under different conditions for long mission duration , where all localities are averaged and transmission curve became smoothed . in comparison with additionally figured transmission , obtained by tsyganenko-89 ( t89 ) model with @xmath30 , it is obviously seen that the paraboloid model gives unapproachable for t89 results in modeling most extreme conditions .
the method we have developing is intended for applications using intensive geomagnetic cutoff calculations , first of all for express magnetospheric transmission calculation for given leo missions , even under changing magnetospheric conditions during flight .
test @xmath0 and transmission calculation , based on the presented technique , is available online by url http://dec1.sinp.msu.ru/vovka/riho ( a simplified version , where transmission calculation is allowed for only static conditions and circular orbits ) .
current version of the method uses the igrf rigidity values table as basic ( in step 1 of scheme ) , although it seems that the best choice would be to apply some other world basic grid , for example , one directly calculated @xmath4 table with using paraboloid model .
nevertheless , it does not understate the attenuation quotient methodology itself , because its drawback in our case is only correspond to relatively small systematic inaccuracies between conditional `` quiet '' @xmath13 values and igrf ones .
author thanks dr.nymmik for fruitful discussions on the topic of paper , and dr.kalegaev for the provided data and cluster computing facilities .
here we summarize technical moments that are related to the performed @xmath13 trajectory calculations with using paraboloid model .
vertically directed protons were test particles for reverse trajectory calculations of penumbra .
fourth order integration scheme was used for it , with accounting for time changing ( hence , magnetic field ) during particle s modeled flight .
the geomagnetic field there was superposition of igrf epoch 2005 field with dynamic paraboloid model ( version 2004 @xcite ) , with uniform field vector @xmath31 outside of the magnetopause , which position is natively given by the code realizing the paraboloid model .
maximal flying distance was equal to @xmath32 , the particles walked it over during motion were considered as allowed for penetration .
the earth was represented by wgs-82 ellipsoid with atmosphere layer at @xmath33 km above its surface , the particles which fell below it was considered as reentrant . rigidity step size for penumbra calculation was equal to @xmath34 gv . all calculations were performed for altitude @xmath35 km above mean earth s radii @xmath36 km for six selected points in northern geographic hemisphere with basic igrf rigidities @xmath37 gv . shea m.a . , smart d.f . and k.g.mccraken .
jgr , 70:4117 , 1965 stone e.c .
jgr , 69(17):3577 , 1964 smart d.f . , shea m.a . and r.gull .
jgr , 74(19):4731 , 1969 danilova o.a . and tyasto m.i
24th icrc , 5:1066 - 1069 , 1995 t.a .
ivanova et al .
, 25(1):7 - 12 , 1985 e.o .
flckiger et al .
jgr , 91(a7):7925 - 7930 , 1986 smart d.f .
magnetospheric models and trajectory computations .
space sci .
, 93:305 - 333 , 2000 smart d.f .
asr , 37:1206 - 1217 , 2006 smart d.f .
et al . proc .
26th icrc , 7 ( sh 3.6.28 ) , 1999a smart d.f . et al .
26th icrc , 7 ( sh 3.6.29 ) , 1999b smart d.f . and shea m.a
25th icrc , 2:397 - 400 , 1997 nymmik r.a . ,
diurnal variations of geomagnetic cutoff boundaries and the penetration function .
cosmic research , 29(3):491 - 493 , 1991 r.a .
nymmik et al .
30th icrc , 1:701 - 704 , 2007 n.a .
planet space sci .
37(1):5 - 20 , 1989 a.j .
tylka et al , creme96 : a revision of the cosmic ray effects on micro - electronics code .
ieee transactions on nuclear science , 44 , 2150 - 2160 ( 1997 ) ; and references therein .
alexeev et al .
magnetic storms and magnetotail currents .
jgr , 101(4):7737 - 7748 , 1996 i.i .
alexeev et al .
the model description of magnetospheric magnetic field in the course of magnetic storm on january 9 - 12 , 1997 .
jgr , 106(a11):25683 - 25694 , 2001 i.i .
alexeev et al .
modelling of the electromagnetic field in the interplanetary space and in the earth s magnetosphere .
space sci .
, 107 , n1/2:7 - 26 , 2003 v.v .
paraboloid model input for two large events in 2003 - 2004 .
private communication , 2006 http://www.magnetosphere.ru/iso r.a .
. doctoral dissertation .
sinp msu , moscow , 1998 | using dynamic paraboloid model of earth s magnetosphere , a large set of particles trajectory computations was performed . based on its result ,
the numerical algorithm for calculating effective cutoff rigidity dependence on empiric parameters has been developed for further use in magnetospheric transmission calculations .
_ keywords : cutoff rigidity , dynamic paraboloid model , magnetospheric transmission , radiation condition on leo _ |
uv observations of euv - detected stars have revealed the presence of about 15 hot white dwarf companions to bright stars in non - interacting binary systems ( e.g. , burleigh , barstow , & fleming 1997 ) . at optical wavelengths ,
these wds are hidden because of their close proximity to much more luminous companions which are main sequence ( spectral type k or earlier ) or evolved stars .
a fascinating variety of objects are known or proposed to contain white dwarf ( wd ) stars in interacting binary systems .
a partial list includes novae , cataclysmic variables , symbiotic stars , ba and ch giants , feige 24-type systems and dwarf carbon stars ( green & margon 1994 ) .
these systems offer great insights to evolution and dynamical processes in binaries .
a number of interacting binary systems where the white dwarf is the primary ( i.e. , optically brightest ) star have also been found among euv - detected systems ( e.g. , 6 close , interacting white dwarf / red dwarf binaries by vennes & thorstensen 1994 ) .
optical or ultraviolet spectral observations are most commonly used to detect companions to wd primaries , by searching for ( 1 ) the presence of narrow balmer line emission overlying the broad smooth balmer absorption of the wd , ( 2 ) a composite white dwarf + main sequence spectrum , or ( 3 ) radial velocity ( rv ) variations .
however , only wds with very close , or intrinsically active companions will be found by method ( 1 ) . for hot white dwarf systems , composite spectra ( 2 )
are only expected to be visible if the companion s spectral type is early enough .
rv variations ( 3 ) require multiple observations at high spectral resolution , and detection strongly favors close and/or massive companions .
all the discoveries mentioned above have been strongly dominated by these selection effects , with companions biased to earlier types than predicted by the simulations of dekool & ritter ( 1993 ) and others . scaling from dekool & ritter s results , vennes & thorstensen ( 1994 ) estimate that `` at least twice as many close binary systems remain to be identified from euv surveys , most of them with a low mass secondary . ''
the resulting sample of binaries known to date therefore must diverge strongly from the intrinsic distribution , in overall normalization , as well as in mass and spectral type of the main sequence companions . the current study , conceived as a complement to optical studies , began as a near - ir photometric survey for low mass companions to hot white dwarfs ( wds ) . by investigating only euv - detected wds
, we obtain a very reasonably - sized but complete sample of young wds , next to which very late - type dwarf companions can be detected in the near - infrared by searching for a @xmath1 excess over that expected from the white dwarf .
many hot white dwarfs ( @xmath2k ; finley et al . 1993 ) have been detected in the recent euv all - sky surveys .
euv detection of these hot wds depends primarily on their temperature , distance , and the intervening galactic ism .
our sample of euv wds ( whose selection we define below ) offers excellent flux contrast in the ir relative to optical ; cool companions will almost always be brighter in the @xmath1 band than the hot wds . to know what @xmath1 mag to expect for the wds , we benefit from constraints on log@xmath3 , radius , and @xmath4 derivable from optical spectra for the wds in our sample , using nlte model atmosphere fits ( napiwotzki et al . 1993 ,
napiwotzki 1997 ) .
the resulting predictions for @xmath1 magnitudes allow a direct search for any ir excess from a cool companion . in some cases , ir colors will also provide a preliminary spectral type .
results from the ir survey will be presented in an upcoming paper .
an additional motivation is the study of the white dwarf mass distribution . since the pioneering work of koester , schulz & weidemann ( 1979 ; ksw ) it is well established that the masses of white dwarfs cluster in a narrow range around @xmath5 , remarkable given that white dwarfs stem from progenitors with masses ranging from below @xmath6 up to @xmath7@xmath8 .
precise knowledge of the white dwarf mass distribution puts constraints on the theory of stellar evolution , especially the poorly understood mass loss process during the final stages of stellar evolution . with two recent exceptions ( beauchamp et al .
1996 and dreizler & werner 1996 , who analyzed samples of helium - rich db and do white dwarfs , respectively ) the mass distribution has only been determined for hydrogen - rich da white dwarfs .
however , this is not a severe limitation , because this spectral class comprises about 80% of all known white dwarfs .
the analysis of ksw , along with other follow - up investigations in the early eighties , used photometric data of which the strmgren and greenstein multichannel colors were the most suitable .
both systems provide temperature- and gravity - sensitive indices .
alternatively , ksw and others used trigonometric parallax measurements to directly calculate the stellar radius .
however , the latter method is practical only for a small sample and suffers from considerable measurement uncertainty .
unfortunately , the photometric indices have their highest sensitivity near 10000k . at this temperature
da white dwarfs have a convective atmosphere and the results depend critically on the adopted parameters of mixing length theory .
the situation improved at the beginning of the nineties , when the development of modern , highly efficient detectors made it possible to obtain high - quality spectra of large numbers of white dwarfs and determine the stellar parameters from a fit to the detailed profiles of the balmer lines .
this method yields sufficient accuracy for white dwarfs hot enough to have a radiative envelope .
the first comprehensive sample of white dwarfs analyzed by this method was presented by bergeron , saffer & liebert ( 1992 ; hereafter bsl ) .
as attributed to higher precision of spectroscopic methods , this investigation yielded a white dwarf mass distribution even narrower than found by ksw and other previous studies .
at the same time , kidder ( 1991 ) analyzed a sample of hot da white dwarfs discovered through positional coincidences of catalogued hot da white dwarfs in existing soft x - ray databases .
three soft x - ray sources corresponding to white dwarfs were found , having relatively low effective temperatures , @xmath925,000k , determined independently using complementary optical and uv spectroscopy .
kidder et al .
( 1992 ) analyzed an expanded sample to derive photospheric he abundances for the hotter objects and establish an effective observational low - temperature threshold for the detection of pure hydrogen da white dwarfs at soft x - ray wavelengths . in 1997
three groups ( marsh et al .
1997 , m97 ; vennes et al .
1997 , v97 ; finley , koester & basri 1997 , fkb ) published results on the mass distribution of extreme ultraviolet ( euv ) selected white dwarfs . due to the selection criterion
, these samples contain the hottest white dwarfs ( @xmath10@xmath11k ) , as cooler white dwarfs do not emit significant euv radiation .
the derived mass distributions in the euv - selected samples are similar to that of bsl , but show some interesting deviations in detail .
the frequency of very high mass white dwarfs is much larger , and that of very low mass white dwarfs much smaller , than in bsl .
these findings can at least partly be explained by selection effects ( see the discussion in fkb ) .
more serious is a trend of the peak mass with temperature .
v97 found that their mass distribution peaks at @xmath12 , while the bsl distribution peaks at @xmath13 , with masses computed using wood s ( 1995 ) mass - radius relation with `` thick '' layers ( @xmath14 , @xmath15 ) .
this discrepancy diminishes slightly , if the `` very thin layer '' ( @xmath16 , no hydrogen layer ) mass - radius relations are used ( peak masses of @xmath17 and @xmath18 for the v97 and bsl sample , respectively ) .
v97 interpreted this as evidence for a very thin hydrogen layer of the da white dwarfs .
however , the effects are small so this result depends strongly on the accuracy of the derived stellar parameters .
fkb estimated the internal accuracy of different analysis methods from monte carlo simulations .
the precision reachable by balmer line fitting is very compelling : @xmath19 @xmath10/ @xmath10@xmath20 for @xmath10@xmath21k .
however , for spectra with very high signal - to - noise ratios ( s / n ) , errors introduced by details of the observation and reduction techniques ( e.g. , extraction , flat fielding , flux and wavelength calibration ) might be more important , but are very difficult to determine .
additionally , one has to take into account differences in the model atmosphere calculations and fitting procedure . together with the results presented in this paper
, we now have four samples of hot white dwarfs , analyzed in a similar way , and with significant overlap .
this offers the opportunity to determine the real accuracy of the spectral analysis of hot white dwarfs , including many possible systematic effects .
we present the selection criteria of our sample in sect . 2 and the observations and data reduction procedures in sect .
details on our model atmospheres are given in sect .
the results and a detailed comparison with the previous analyses of euv selected white dwarfs are presented in sect .
we finish with a discussion of our results and an outlook .
we chose to limit our uniform sample to das , for which models provide the best temperature and mass constraints .
we start with 73 known da white dwarfs in the euve bright source list ( malina et al .
excluding sources at low galactic latitudes ( @xmath22 ) and southerly declinations ( @xmath23 ) yields a list of 28 das . a similar procedure for non - overlapping das listed in the _ rosat _ wide field camera survey bright source catalogue ( pounds et al . 1993 ) yields 29 objects .
we have removed from our uniform sample 2 well - known stars with published sensitive optical spectrophotometry and ir photometry ( feige 24 and hz 43 ) .
two das with broad line profiles due to magnetic splitting were also excluded ( pg1658 + 441 and pg0136 + 251 ) .
eight known binaries are also excluded from the uniform sample : v471 tau ( vennes , christian , & thorstensen 1998 ) , pg0824@xmath24289 , ( heber et al . 1993 ) hd74389b ( liebert , bergeron , & saffer 1990 ) , rej1016@xmath25052 ( v97 ) , pg1033 + 464 ( gd123 ) , ( green , schmidt , & liebert 1986 ) , rej1426@xmath24500 ( v97 ) , rej1629 + 780 ( catalan et al . 1995 ) , and ik peg ( wonnacott et al . 1993 ) , leaving 47 objects . in this paper , a handful of objects which fell outside the uniform sample definitions just outlined were included for observation .
these include the known binaries pg1033 + 464 and rej1629 + 780 and the magnetic white dwarf pg1658 + 441 , as well as mct0455 - 2812 which was outside the sample declination limits .
due to observing constraints ( a combination of weather , poor seeing and faint objects , or celestial placement of objects ) no spectra were obtained for sample objects rej0443 - 034 , rej0916 - 194 , or pg1040 + 451 .
pg1234 + 482 was originally classified as a sdb star and thus excluded ; it has since been reclassified as a da ( jordan et al .
the final sample we analyze here thus includes 46 da white dwarf stars for which we present new model nlte fits to optical spectra .
we note that since the sample selection was performed , several relevant discoveries pertaining to sample objects have been made .
rej0134 - 160 ( gd984 ) has central balmer emission components produced by a dme companion .
rej1440 + 750 turns out to be a magnetic da ( dreizler et al .
1994 ) . the uniform sample will be discussed in a followup paper treating the ir photometry and binary fraction .
dates are listed for all observations in table [ t : results ] . on the nights of 04 - 06 january 1996 we obtained spectra at steward observatory s kitt peak station using the bok 2.3-m reflector equipped with the boller &
chivens cassegrain spectrograph and uv - flooded loral 800@xmath261200 ccd .
most spectra were dispersed with a 600 @xmath27/mm first - order grating used behind a 4.5@xmath284@xmath29 long slit . the instrumentation provided wavelength coverage @xmath3034005600 at a spectral resolution of @xmath315fwhm . on the last night of the observing run we employed a new 400 l / mm grating providing wavelength coverage @xmath3035006790 at a spectral resolution of @xmath317 fwhm .
we also obtained spectra at the mmt on mt hopkins , 09 - 11 may 1996 , using the 300 @xmath27/mm grating in first order on the loral 3k@xmath321k ccd of the blue channel spectrograph .
this yields coverage from about 3350 to 8800 , and the 2@xmath33 slit width we used resulted in a spectral resolution of about 4 fwhm .
several objects were kindly obtained for us using the identical instrumental configuration at the mmt by perry berlind on 08 april 1997 .
exposures at both telescopes ranged from one to thirty minutes for program stars , and for all observations the long slit was rotated to the parallactic angle according to the calculations of filippenko ( 1982 ) .
the airmasses were held below 1.5 in almost all cases .
all spectra were extracted from the two - dimensional images and reduced to linear wavelength and intensity scales using standard reduction packages in the image reduction and analysis facility ( iraf ) .
these operations included bias subtraction , flat field division by images obtained by exposing on dome or internal quartz lamps , centroiding and summation of the stellar traces on the two - dimensional images , sky subtraction , wavelength calibration using spectra of he / ar arc lamps , and absolute flux calibration using the flux standards of massey et al .
details of these reduction procedures are given by bsl .
we present in figs .
[ f : steward ] and [ f : mmt ] our collection of white dwarf spectra .
the spectra obtained at the steward observatory 2.3 m telescope are found in fig .
[ f : steward ] , and the mmt spectra in fig .
[ f : mmt ] .
some stars observed repeatedly appear in both figures .
the signal - to - noise ratio of our spectra ranges from 35 to 200 with an average of 90 .
ninety percent of our spectra have a s / n of at least 45 .
we calculate hydrogen model atmospheres with the nlte code developed by werner ( 1986 ) .
basic assumptions are those of static , plane - parallel atmospheres in hydrostatic and radiative equilibrium . in contrast to the atmospheres
commonly used to analyze da white dwarfs , we relax the assumption of local thermal equilibrium ( lte ) and solve the detailed statistical equilibrium instead . as described in werner ( 1986 ) , the accelerated lambda iteration ( ali ) method is used to solve the set of non - linear equations . the impact of nlte on white dwarf atmospheres is discussed in detail in napiwotzki ( 1997 ) .
levels and lines are included in nlte up to @xmath34 .
line blanketing by the stark broadened hydrogen lines is taken into account consistently .
as the hydrogen atmospheres of da white dwarfs are stable for @xmath10@xmath35k , convection is not included in our atmospheric models .
pressure dissolution of the higher levels is described by the hummer & mihalas ( 1988 ) occupation probability formalism following the nlte implementation by hubeny et al .
the synthetic spectra are computed with the extended vcs broadening tables ( vidal et al .
1970 ) provided by schning & butler ( priv .
comm . ) and lemke ( 1997 ) .
we followed the prescription of bergeron ( 1993 ) and increased the critical ionizing field adopted to calculate the occupation probability by a factor two .
the motivation is not a flaw in the hummer & mihalas ( 1988 ) formalism , but a compensation for the inadequacy of the standard stark broadening theory when line wings overlap .
our nlte model grid covers the temperature range 17000k @xmath36 @xmath10@xmath37100000k ( stepsize increasing with @xmath10 from 2000k to 10000k ) and gravity range @xmath38 ( stepsize 0.25 ) .
although deviations from lte are small for most da white dwarfs they become significant for the hottest stars in our sample ( cf .
napiwotzki 1997 ) .
since we intend to compare our results with three other samples analyzed by means of lte atmospheres , we have produced a map with lte correction vectors .
for this purpose we calculated a set of lte model atmospheres using the technique described in napiwotzki ( 1997 ) of drastically enhancing collisional rates between the atomic levels in the nlte code .
this forces the occupation numbers to be in lte and guarantees consistency with the nlte atmospheres .
the synthetic lte spectra were transformed into `` observed '' spectra by convolving them with a gaussian of 5 fwhm , rebinning them to 2 and adding poisson noise corresponding to a continuum s / n of 100 .
these simulated spectra were analyzed with the nlte grid following the procedure outlined in sect .
[ s : analysis ] .
1000 simulations were run for every parameter set to eliminate the effect of random errors .
the resulting offsets are displayed in fig .
[ f : ltevectors ] .
the orientation of the vectors corresponds to the correction , which must be applied to transform lte results to the nlte scale . as expected from the results of napiwotzki ( 1997 ) the differences are negligible for da white dwarfs cooler than @xmath940000k , but can be significant for hotter stars .
the lte vectors show the trend that nlte effects increase with increasing temperature and decreasing gravity ( see e.g. fig . 3 and 4 in napiwotzki , 1997 , for the case of da white dwarfs ) .
the small corrections found for the models with the highest temperature and lowest gravity seems to contradict this behavior .
a look at the line profiles reveals that the nlte deviations are larger for , say , the @xmath10@xmath39k , @xmath40 model than the 70000k , 8.0 model .
however , the correction vectors are not a simple function of nlte deviations measured e.g. as equivalent width difference , but depend also on the way line profiles vary with temperature and gravity .
the small corrections found for the high temperature / low gravity models are produced by the cancellation of these effects .
we expect a reliable transformation between the lte and nlte temperature scales only if the deviations are not too large .
a conservative upper limit is @xmath970000k .
a thorough comparison with the previous lte analyses is presented in sect .
[ s : comparison ] .
we checked our models by a comparison of our lte spectra with some da model spectra kindly provided by d. koester .
model parameters were @xmath41k to 70000k in 10000k steps and @xmath42 for pure hydrogen models .
input physics are very similar .
in particular both model calculations adopt twice the critical ionizing field for the calculation of line profiles .
we treated koester s model spectra the same way we treated the lte spectra above and fitted them with our lte grid .
the result was quite satisfactory : the temperature differences were always below 1.5% and the gravity differences never exceeded 0.03dex .
metals were ignored in our calculations , but they can modify the hydrogen line profiles by their effect on the atmospheric structure .
lanz et al .
( 1996 ) analyzed the balmer lines of the hot da g191b2b with pure hydrogen lte and nlte atmospheres and an nlte model with full metal line blanketing .
they concluded that the effect of metal line blanketing on the balmer lines was relatively small , and the difference between lte and nlte was found to be the most important effect . a recent study by barstow et al . ( 1998 ) , which investigated several hot white dwarfs in the temperature range around 60000k derived larger metal line blanketing effects of the order of the nlte effects . since the lte analyses of m97 , v97 , and fkb are based on pure hydrogen models , our results should be consistent with theirs in any case
atmospheric parameters of our da white dwarfs are obtained by simultaneously fitting line profiles of the observed balmer lines with the nlte model spectra described above .
we use the least - square algorithm described in bsl . the observed and theoretical balmer line profiles are normalized to a linear continuum ( both spectra are @xmath43 ) in a consistent manner .
wavelength shifts are determined with a cross - correlation method and applied consistently to each complete spectrum .
the synthetic spectra are convolved to the observational resolution with a gaussian and interpolated to the actual parameters with bicubic splines , and interpolated to the observed wavelength scale .
the atmospheric parameters @xmath10 and @xmath44 are then determined by minimizing the @xmath45 value by means of a levenberg - marquardt steepest descent algorithm ( press et al .
several tests revealed that our interpolation routine is rather robust concerning spacing of our model grid and yields reliable results even at the edge of the model grid .
finally , an estimate of the internal errors can be derived from the covariance matrix .
in contrast to bsl , we estimate the noise of the spectra ( @xmath46 ) used for the @xmath45 fit from the neighboring continuum of each line .
the s / n is adopted to be constant throughout the line .
rlr@@xmath47lr@@xmath47lrl 0007 + 331 & gd2 & 46493&514 & 7.83&0.05 & 0.602 & 6 jan 1996 + 0134@xmath25160&gd984 , phl1043&44866&667 & 7.77&0.05 & 0.572 & 5 jan 1996 + 0237@xmath25122 & phl1400 & 32077&177 & 8.45&0.04 & 0.890 & 5 jan 1996 + 0348@xmath25005 & gd50 & 39508&464 & 9.07&0.06 & 1.215 & 4 jan 1996 + 0427 + 740 & & 48587&1044 & 7.93&0.08 & 0.646 & 5 jan 1996 + 0457@xmath25280 & mct0455@xmath252812&51199&786 & 7.72&0.05 & 0.570 & 4 jan 1996 + 0512@xmath25004 & & 31733&139 & 7.40&0.03 & 0.458 & 5 jan 1996 + 0521@xmath25102 & & 33186&301 & 8.60&0.06 & 0.980 & 5 jan 1996 + 0841 + 032 & & 38293&252 & 7.75&0.03 & 0.544 & 4 jan 1996 + 0902@xmath25040 & & 23218&160 & 7.84&0.02 & 0.544 & 4 jan 1996 + 0907 + 505 & pg0904 + 511 & 32167&338 & 8.11&0.07 & 0.695 & 4 jan 1996 + 0940 + 502 & pg0937 + 506 & 36034&283 & 7.69&0.04 & 0.519 & 9 apr 1997 + 0957 + 852 & & 51311&1348 & 8.37&0.10 & 0.866 & 5 jan 1996 + 1019@xmath25140 & & 31524&102 & 7.92&0.02 & 0.606 & 9 apr 1997 + 1029 + 450 & pg1026 + 454 & 35518&247 & 7.70&0.04 & 0.520 & 9 apr 1997 + 1032 + 532 & & 43587&506 & 7.95&0.05 & 0.644 & 4 jan 1996 + 1033@xmath25114 & g162@xmath2566 , ltt3870 & 24685&252&7.85&0.03 & 0.553 & 4 jan 1996 + 1036 + 460 & gd123 & 29361&251 & 8.02&0.05 & 0.648 & 4 jan 1996 + 1043 + 490 & & 41132&1178 & 7.94&0.13 & 0.635 & 5 jan 1996 + 1044 + 574 & pg1041 + 580 & 30338&153 & 7.81&0.03 & 0.550 & 4 jan 1996 + 1100 + 713 & pg1057 + 719 & 41104&814 & 7.84&0.09 & 0.593 & 4 jan 1996 + 1112 + 240 & ton61 & 39824&636 & 7.78&0.07 & 0.563 & 5 jan 1996 + 1122 + 434 & pg1120 + 439 & 26996&151 & 8.31&0.02 & 0.803 & 9 apr 1997 + 1126 + 183&pg1123 + 189 & 54334&1983 & 7.76&0.13 & 0.594 & 4 jan 1996 + 1128@xmath25025 & pg1125@xmath25026 & 30699&380 & 8.24&0.08 & 0.767 & 4 jan 1996 + 1148 + 183 & pg1145 + 188 & 25758&299 & 7.91&0.04 & 0.587 & 4 jan 1996 + 1235 + 233&pg1232 + 238 & 46569&523&7.83&0.05 & 0.602 & 11 + 12 may 1996 + 1257 + 220&gd153 & 38926&142 & 7.78&0.02 & 0.560 & 4 jan 1996 , 12 may 1996 + 1336 + 694 & pg1335 + 701 & 29607&87 & 8.34&0.02 & 0.824 & 10 may 1996 + 1431 + 370 & gd336 & 34404&115 & 7.91&0.02 & 0.608 & 11 may 1996 + 1446 + 632 & & 37947&254 & 7.79&0.04 & 0.562 & 11 may 1996 + 1629 + 780 & & 41043&338 & 7.92&0.04 & 0.627 & 11 may 1996 + 1638 + 350 & pg1636 + 351 & 35404&142 & 7.98&0.02 & 0.642 & 10 may 1996 + 1643 + 411&pg1642 + 414 & 28815&81&8.22&0.02 & 0.753 & 9 may 1996 + 1650 + 403 & & 38144&211 & 7.97&0.03 & 0.643 & 9 may 1996 + 1711 + 664 & & 48989&757&8.89&0.06 & 1.141 & 12 may 1996 , 9 apr 1997 + 1726 + 583&pg1725 + 586 & 53561&542&8.23&0.04 & 0.795 & 9 may 1996 + 1800 + 683 & kuv18004 + 6836&44723&424 & 7.80&0.04 & 0.585 & 10 may 1996 + 1820 + 580 & & 44099&264 & 7.78&0.03 & 0.574 & 9 may 1996 + 1845 + 682 & kuv18453 + 6819&36120&189 & 8.23&0.03 & 0.770 & 10 may 1996 + 2116 + 735&kuv21168 + 7338 & 50812&354&7.72&0.03 & 0.569 & 5 jan 1996 , 9 may 1996 + 2207 + 252 & & 26964&174 & 8.27&0.03 & 0.779 & 5 jan 1996 + 2312 + 104 & gd246 & 53088&968 & 7.85&0.07 & 0.624 & 4 jan 1996 + the results are given in table [ t : results ] , with illustrative examples shown in fig .
[ f : sampleres ] .
we adopt , for the moment , the usual practice and indicate in table [ t : results ] the internal errors estimated from the quality of the @xmath45 fit .
however , one should keep in mind that these errors can only serve as lower limits
. we will show below ( sect .
[ s : comparison ] ) that these formal errors derived from the @xmath45 fit significantly underestimate the real errors .
external errors can be estimated from multiple observations and analysis of the same star .
we obtained repeat observations for a subsample of 6 stars , for which results are given in table [ t : repeat ] .
the gravity values of all six stars agree within the estimated internal errors .
the same is true for four temperature comparisons , too .
however , the differences found for rej1650 + 403 and rej2116 + 735 are significantly larger .
this is in line with the external errors we estimate from a comparison with the studies of m97 , v97 , and fkb ( see sect . [
s : comparison ] ) .
rr@@xmath47lr@@xmath47lr@@xmath47lr@@xmath47l 1235 + 233 & 46308&837 & 7.85&0.08 & 46737&670 & 7.82&0.06 + 1257 + 220 & 39349&317 & 7.76&0.04 & 38820&159 & 7.78&0.02 + 1643 + 411 & 28813&120 & 8.23&0.02 & 28818&111 & 8.21&0.02 + 1650 + 403 & 37798&245 & 7.94&0.04 & 39126&413 & 8.01&0.05 + 1726 + 583 & 52712&927 & 8.27&0.06 & 54003&669 & 8.20&0.05 + 2116 + 735 & 50131&384 & 7.71&0.03 & 54604&906 & 7.76&0.06 + five stars in our sample , rej0134 - 160 , rej1036 + 460 , rej1043 + 490 , rej1126 + 183 , and rej1629 + 780 , show clear signs of binarity in the red part of their spectra .
two more stars , rej1711 + 664 and rej2207 + 502 , are members of visual binaries .
[ [ rej1629780 ] ] rej1629 + 780 : + + + + + + + + + + + + the red part of the spectrum of rej1629 + 780 is heavily contaminated by a m type main sequence companion . the composite spectrum and the spectrum of the m star after subtracting the white dwarf component
are shown in fig .
[ f : compspek ] .
the characteristic bands of tio are easily recognizable .
cataln et al .
( 1995 ) determined a spectral type dm4 .
the h@xmath48 and h@xmath49 lines are seen in emission , which indicates a chromospherically active me star .
sion et al.(1995 ) detected a flare - like increase of the balmer line emission .
the comparison in fig .
[ f : compspek ] demonstrates that the blue part ( @xmath50 ) of the white dwarf spectrum is not disturbed by the m dwarf .
thus , we excluded h@xmath49 and the core of h@xmath51 from the fit and derived the parameters @xmath10@xmath52k and @xmath53 from h@xmath51 to h@xmath54 .
these parameters are in reasonable agreement with the results of cataln et al .
( @xmath10@xmath55k ; @xmath42 ) , who however included h@xmath49 , and kidder ( 1991 ; @xmath10@xmath56k , @xmath57 ) , who fitted lyman-@xmath48 and the balmer lines h@xmath49 and h@xmath51 .
[ [ rej1036460-rej1043490-rej1126183 ] ] rej1036 + 460 , rej1043 + 490 , rej1126 + 183 : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + three more white dwarfs of our observed sample are known binaries ( rej1036 + 460 , rej1126 + 183 : green , schmidt , & liebert 1986 , rej1043 + 490 : schwartz et al .
1995 ) and show red excesses in our spectra : since the h@xmath49 lines of the white dwarfs are contaminated they were excluded from the fit .
h@xmath51 and higher balmer lines are virtually uncontaminated . a discussion of these stars and newly discovered binaries will be given in a forthcoming paper .
[ [ rej0134 - 160-gd984 ] ] rej0134 - 160 ( gd984 ) : + + + + + + + + + + + + + + + + + + + + although our spectrum ends at only 5600 a red excess is obvious .
subtraction of the theoretical white dwarf flux leaves an m star spectrum .
the m dwarf contribution is much smaller than in rej1629 + 780 and we excluded only the cores of h@xmath49 and h@xmath51 , filled in by the balmer line emission , from the fit .
bues & aslan ( 1995 ) suspected a hot third component in rej0134 - 160 .
however , outside of the h@xmath49 and h@xmath51 cores , the balmer lines are well reproduced by our best fit without any indication of a third component . indeed , a subdwarf component as suggested by bues & aslan ( 1995 ) is almost certainly not present , since its flux would dominate in the blue .
[ [ rej1711664 ] ] rej1711 + 664 : + + + + + + + + + + + + since a late type star is separated from the white dwarf by only @xmath58 , we took care to get an uncontaminated white dwarf spectrum .
we obtained a spectrum under good seeing conditions .
no excess is present up to the red limit at 8500 .
[ [ rej2207252 ] ] rej2207 + 252 : + + + + + + + + + + + + this white dwarf has a red companion @xmath59 away .
schwartz et al .
( 1995 ) estimated spectral type k4v and distance 65pc from its colors . with the parameters from table [ t :
results ] a white dwarf distance of 62pc results .
thus it is likely that both stars form a physical pair .
the angular separation corresponds to @xmath9500au .
two white dwarfs of our sample show zeeman splitting of the balmer line cores , indicative of a magnetic field .
rej1659 + 440 ( pg1658 + 441 ) is a well known star , already analyzed by schmidt et al .
the discovery of the magnetic nature of rej1440 + 750 ( hs1440 + 7518 ) was announced by dreizler et al .
( 1994 ; note the naming confusion corrected in dreizler et al .
1995 : hs1412 + 6115 should have been hs1440 + 7518 ) .
although rej1440 + 750 was analyzed by v97 , they did not remark on its magnetic nature .
this is likely due to the lack of coverage of the h@xmath48 line , which displays the most pronounced zeeman effect .
flux calibrated spectra of the magnetic white dwarfs are displayed in fig .
[ f : magwd ] .
the pg1658 + 441 analysis of schmidt et al.(1992 ) resulted in @xmath10@xmath60k and @xmath61 .
the magnetic splitting was best reproduced by a 3.5megagauss ( mg ) dipole inclined @xmath62 to the line of sight ( producing a mean surface field strength @xmath63 mg ) . from the linear zeeman effect we estimated a mean magnetic strength of 8 mg for rej1440 + 750 , consistent with the estimate given in dreizler et al .
( 1994 ) .
the temperature and gravity of rej1440 + 750 were derived from a fit of the higher balmer lines h@xmath51 to h@xmath54 , which are less affected by the magnetic splitting .
results are given in table [ t : magwd ] supplemented by the schmidt et al .
( 1992 ) fit of pg1658 + 441 .
our own fit gave similar results albeit with lower accuracy .
our fits of rej1440 + 750 can only provide a rough estimate of the stellar parameters .
accurate results can only be expected from a detailed treatment of the magnetic effects .
rlr@@xmath47lr@@xmath47lrrl 1440 + 750 & hs1440 + 7518 & 36154&275 & 8.87&0.05 & 1.128 & 7.7 & 10 may 1996 + 1659 + 440&pg1658 + 441 & 30510&200&9.36&0.07 & 1.311 & 2.3 & 10 may 1996 + we have now presented the results of a homogeneous analysis of a sample of 46 hot , euv selected white dwarfs based on balmer line fitting .
three other large samples analyzed with the same method were recently published by m97 , v97 , and fkb .
since considerable overlap exists between all four samples , this allows a direct check for systematic errors and the individual scatter on a star by star basis for white dwarfs hotter than 25000k .
since in contrast to previous works , our analysis is based on nlte model atmospheres , we applied the correction vectors given in fig .
[ f : ltevectors ] to correct for the lte assumption . since for
the hottest white dwarfs these corrections become large , while the accuracy of temperature and gravity estimates decreases for both lte and nlte analyses , one should exclude comparison of stars with @xmath10@xmath64k .
this does not affect our sample , which has a maximum temperature closer to 54000k .
differences ( after correction to nlte ) in @xmath10and @xmath44 between studies for stars in common with m97 , v97 , and fkb are displayed in fig .
[ f : diffteff ] as function of @xmath10 .
the magnetic white dwarfs and the binaries , which show significant contamination of the white dwarf spectrum by the companion , are excluded .
one can now focus on systematic differences between pairs of studies , e.g. our results ( ngs ) versus fkb ( fkb@xmath25ngs ) , v97@xmath25ngs , m97@xmath25ngs , fkb@xmath25m97 , m97@xmath25v97 , and so on .
however , we chose another approach and performed an optimization that simultaneously took into account all values from all stars in common between any samples .
in other words , the values given for the systematic differences between our study and the m97 , v97 , and fkb samples form a system for direct transformation between , say , fkb and the three other samples .
the running averages computed this way are plotted in fig .
[ f : diffteff ] as a solid line .
the actual value was computed in 1000k steps for all white dwarfs in the respective temperature intervals @xmath10@xmath65 .
although this curve does not represent the best fit to the data plotted in fig .
[ f : diffteff ] alone , it is a fairly good representation of the differences computed directly between our measurements and those of m97 , v97 , and fkb .
since the distribution is highly non - gaussian with many outliers , as evident in fig .
[ f : diffteff ] , we decided to adopt an underlying lorentzian ( or cauchy ) distribution for the optimization .
the tails of the cauchy distribution are much larger than that of the corresponding gaussian , yielding a much lower weight for deviant points ( see discussion in press et al .
the dotted lines represent the @xmath66 confidence interval _ of the mean _ , computed conservatively from the rms deviations .
first , we notice a considerable scatter , larger than expected from the internal error estimates ( see the discussion below ) . if one ignores the hot end , the agreement between the fkb and our temperature scale is good ; differences are below 1% , smaller than the maximum model differences to the koester models ( cf . sect . 4 ) , which were used by fkb .
the same atmospheres are used in m97 , and it is therefore surprising that significant differences with m97 are present .
these trends are most likely caused by different reduction and analysis techniques .
offsets of the same order are found in our comparison with v97 , where a different lte model atmosphere code is used .
although basically the same input physics is included , this might at least partly explain those shifts in @xmath10and @xmath44 .
lr@@xmath47lr@@xmath47lr@@xmath47l @xmath10@xmath67k & @xmath250.023&0.006 & 0.032&0.005 & 0.007&0.005 + @xmath68k & @xmath250.013&0.003 & 0.017&0.003 & 0.004&0.003 + @xmath69k & 0.036&0.006 & 0.050&0.006 & 0.046&0.007 + all with @xmath70k&@xmath250.006&0.003&0.024&0.003 & 0.012&0.003 + lr@@xmath47lr@@xmath47lr@@xmath47l @xmath71k & @xmath250.112&0.017 & @xmath250.001&0.016 & @xmath250.064&0.018 + @xmath68k & @xmath250.086&0.017 & 0.026&0.016 & @xmath250.034&0.016 + @xmath69k & @xmath250.066&0.030 & @xmath250.067&0.026 & @xmath250.062&0.027 + all with @xmath70k&@xmath250.084&0.013&@xmath250.002&0.012 & @xmath250.047&0.012 + our results are quantified in tables [ t : dteff ] and [ t : dlogg ] .
we divided the white dwarfs into three groups according to their temperature : a cool group with @xmath72k , a hotter one with @xmath73k , and the hottest considered group with @xmath74k .
nlte effects are still negligible in the range of effective temperatures of the two coolest groups .
mean shifts and the confidence range of the mean ( computed as described above ) are provided for these groups and the complete sample ( except stars with @xmath75k ) .
the shifts discussed above and shown in fig [ f : diffteff ] are statistically significant and reach values of @xmath95% in @xmath76 in the hottest bin . in @xmath44
the difference between our results and m97 reach @xmath90.1dex for the coolest bin . however , we emphasize that all four analyses are based on state - of - the - art model atmospheres and @xmath45 fitting techniques .
since there are no strong arguments to favor or discard one analysis , it seems these are the systematic shifts characteristic of modern analyses of hot white dwarfs .
lrr @xmath71k & 0.023 & 0.075 + @xmath68k & 0.023 & 0.101 + @xmath69k & 0.034 & 0.128 + all with @xmath70k & 0.026 & 0.107 + if we take the systematic shifts into account , we can use the samples to derive estimates of the observational scatter , which can be compared with the internal error estimates . since we have approximately the same scatter for different combinations of the samples , we compute a mean scatter @xmath77 for all possible combinations ( weighted by the number of stars in common ) . for this purpose we correct for the systematic shifts calculated for each of the three @xmath76 bins , whose temperature intervals are given in table [ t : scatter ] .
the individual measurement errors @xmath78 add quadratically , and if we assume that the inherent scatter is the same for all analyses ( and we found no reason to discard this assumption ) , the individual measurement errors can be estimated as @xmath79 . not surprisingly , the smallest scatter is found for the `` cool '' group ( @xmath72k ) with @xmath80% and @xmath81dex .
it increases to @xmath82% and @xmath83dex for the hottest bin .
this trend is expected from the statistical analysis presented in fkb ( their fig . 1 ) .
however , the values are larger by a factor of three or more than the internal parameter errors for a well exposed spectrum ( see e.g. table [ t : results ] ) .
therefore , we conclude that the accuracy is not limited by the noise for good spectra , and we suggest that other effects , such as details of the extraction or fluxing and normalization procedures , contribute more . considering these systematic uncertainties , the 0.3dex difference between the gravity determinations of fkb and napiwotzki et al .
( 1993 ) for hz43a is only a @xmath84 deviation and therefore not as serious as considered by fkb .
once the temperature and gravity of the white dwarfs are known the mass can be determined from theoretical mass - radius relations .
the recent investigations of m97 , v97 , and fkb based their interpretation on the model sequences of wood ( 1995 ) .
the applied models have h- and he - layer masses of @xmath85 and @xmath86 , respectively ( `` thick layers '' ) and a carbon core .
these model sequences are based on a set of pre - white dwarf models with end masses of @xmath87 and @xmath88 computed by kawaler ( see wood 1994 ) . starting models for other masses were constructed by homology transformations .
the mass - radius relations we use are based on the evolutionary calculations of blcker ( 1995 ) .
these models are calculated _ ab initio _ from the main sequence .
the resulting white dwarfs show important differences compared to wood s models .
in contrast with the carbon core models widely used for mass determination , the degenerate core contains a mass - dependent mixture of carbon and oxygen ( and trace elements ) .
however , note that wood ( 1995 ) also presented some white dwarf sequences with c / o cores .
the h- and he - layer masses resulting from evolutionary calculations depend on the stellar mass ( blcker 1995 ; blcker et al .
while the canonical thick hydrogen layer mass of @xmath89 is met for a @xmath90 white dwarf , it decreases from several @xmath91 for @xmath92 down to @xmath93 for @xmath94 .
this results in a mass - radius relation which is steeper than a relation based on a constant layer mass .
consequently , the derived mass distribution will be narrower , if evolutionary layer masses are used .
the lower mass limit for a c / o core white dwarf is @xmath95 , the limiting mass for central helium burning in low mass stars ( sweigart et al .
thus , white dwarfs with a lower mass possess a helium core . at the current age of the universe
he white dwarfs have not been produced by single star evolution , but are the result of binary evolution where the hydrogen - rich envelope was stripped away along the ( first ) red giant branch ( kippenhahn 1967 ; iben & tutukov 1986 ) .
we use the recent evolutionary models of he white dwarfs calculated by driebe et al .
( 1998 ) based on this scenario . the combined tracks of blcker ( 1995 ) and driebe et al .
( 1998 ) cover the mass range @xmath96 to @xmath97 .
we supplemented this set with the 1.0 , 1.1 , and @xmath98 carbon core sequences of wood ( 1995 ) with `` thin '' layers .
since the hydrogen envelope mass decreases with increasing white dwarf mass in blcker s models , their highest mass models effectively correspond to `` thin layer '' models .
since high mass white dwarfs are already close to the zero - temperature configuration , the remaining departures from consistent evolutionary calculations can be ignored for practical purposes . the position of the analyzed white dwarfs in the temperature / gravity plane is shown in fig .
[ f : tracks ] along with the tracks used for the mass determination .
the individual white dwarf masses are given in table [ t : results ] , and the resulting mass distribution is shown in fig .
[ f : mass ] .
we have also redetermined masses for the bsl sample with the blcker / driebe mass - radius relations and supplement fig .
[ f : mass ] with the resulting distribution .
our mass distribution possesses the same basic features as found in the euv selected samples of m97 , v97 , and fkb .
the sharp peak centered at @xmath99 is in principal agreement with the earlier investigations of ksw , weidemann & koester ( 1984 ) , and bsl , with a sharp falloff towards lower masses , and a less steep decline towards higher masses with a tail of high mass white dwarfs . considering the relatively large observational scatter we found in sect .
[ s : comparison ] , the _ underlying _ distribution may be extremely sharp .
the mean mass is @xmath100 .
however , this value is strongly biased by the few very high mass white dwarfs .
we decided to follow the recipe of fkb and fitted the mass peak with a gaussian .
although a gaussian is not a good representation of the white dwarf mass distribution , it gives a robust estimate of the peak mass .
the fit result is , in principle , dependent on the adopted binning of the mass intervals .
however , in some test calculations we found effects exceeding a few @xmath101 only if the bin width was larger than @xmath102 .
the values given in this paper were derived by fitting gaussians to virtually unbinned mass distributions ( formal bin width @xmath101 ) .
we were concerned by the possibility that the different fraction of low and high mass white dwarfs found in different samples ( see discussion below ) may skew the gaussian fit of the main peak to higher or lower masses .
we tested this by fitting the bsl , and fkb and our mass distributions with multi - component gaussians , which fitted the secondary high and low mass peaks separately .
we found no deviation higher than @xmath103 in any case and concluded that the single gaussian fit of the main peak yields a rather stable estimate . with this method
, we derived a peak mass of our sample of @xmath104 .
we reanalysed the fkb , m97 , and v97 samples with our mass - radius relations .
we applied the corrections to nlte and excluded the white dwarfs with temperatures in excess of 70000k .
the peaks masses are @xmath105 ( fkb ) , @xmath106 ( m97 ) , and 0.582 ( v97 ) .
these differences reflect the systematic differences in @xmath76 and @xmath44 determinations discussed above and probably different selection criteria in the case of fkb . the probable he core white dwarf rej0512 - 004 is the only object in our sample with a mass below @xmath107 , but four white dwarfs have masses in excess of @xmath108 .
the frequency of low and high mass white dwarfs found in our and other euv selected samples is qualitatively different from that found in optically selected samples such as bsl .
bsl detected in their sample of 129 white dwarfs 16 objects with masses below the limiting mass for c / o core white dwarfs and a total of 28 white dwarfs with @xmath109 .
these numbers are reduced to 10 and 16 respectively , if we redetermine the masses with the mass - radius relations of blcker ( 1995 ) and driebe et al .
( 1998 ) . however , that is still a much higher fraction than we find in our euv selected sample . why do optical and euv selected samples contain different fractions of high and low mass white dwarfs ?
all helium white dwarfs from the bsl sample have temperatures below 30000k ; the only helium white dwarf candidate in our sample has @xmath11032000k .
this implies that the detection probability of low mass white dwarfs in euv selected samples is much lower , because the fraction of low mass white dwarfs at higher @xmath76 is much lower than it is for @xmath71k .
this viewpoint is supported by the helium white dwarf sequences calculated by driebe et al .
we have plotted the variation of temperatures with time for white dwarfs with helium and c / o cores in fig .
[ f : tevol ] .
the detection probability in a given evolutionary stage goes as the inverse of the rate of temperature decrease at that stage .
new he white dwarfs cool down rapidly to a certain ( mass dependent ) temperature , whereafter the cooling rate drops drastically and the star remains on a temperature plateau for a long time .
this behavior is most pronounced for the lowest masses .
hydrogen shell burning plays an important role for this plateau phase .
although the energy production by hydrogen burning drops down dramatically after the star enters the white dwarf sequence , it still produces a significant fraction of the total luminosity , enough to bring the cooling to a standstill .
we can conclude from fig .
[ f : tevol ] that it is extremely unlikely to detect a helium white dwarf during its first rapid cooling phase .
it is easier to find them during their plateau phase or afterwards .
this means there is a strong bias towards detecting helium white dwarfs in samples of relatively `` cool '' da stars ( like bsl ) instead of the overall much hotter euv selected samples .
we can explain in a similar fashion why high mass white dwarfs are found preferentially in the @xmath111k range .
due to the drop in neutrino cooling , the evolutionary speeds of massive white dwarfs are relatively low in this range , as demonstrated by the @xmath112 track , the most massive stellar remnant calculated by blcker ( 1995 ) .
this increases the detection probability of more massive white dwarfs in euv selected samples .
even if we restrict our comparisons to hot subsamples of white dwarfs , the fraction of massive white dwarfs is much higher in the euv selected samples than it is in optically selected ones .
an extensive discussion of this problem is carried out by fkb .
fkb argue that the number of high mass white dwarfs discovered per sky area is similar for the whole sky euv surveys and the optical palomar - green ( pg ) survey ( green et al .
1986 ) , which covered roughly 25% of the sky .
in other words , a whole sky version of the pg survey should have discovered all but one of the euv detected white dwarfs with @xmath113 . at a given temperature ,
white dwarfs with lower masses are larger , consequently more luminous , and detectable to larger distances .
for instance at 30000k , a white dwarf of @xmath107 is detectable at distances @xmath114 larger than a @xmath115 white dwarf . at the maximum distance sampled by ,
e.g. , the pg survey , the probability is very high that interstellar matter effectively absorbs all euv radiation .
thus the dominant effect in euv surveys is not a selection for massive white dwarfs as suggested by v97 , but a selection _ against _ low mass white dwarfs , due to a sample volume strongly affected by interstellar absorption .
fkb noted a moderate apparent trend of the white dwarf peak mass with temperature .
since their sample also included optically selected white dwarfs , they analyzed a considerably higher fraction of stars with @xmath116k and therefore have a larger temperature baseline than pure euv samples .
analogous to this finding , v97 found an @xmath117 offset between the peak of the mass distribution in their euv selected sample and the ( cooler ) white dwarfs analyzed by bsl . in both cases this temperature dependence decreases if wood s models with `` very thin '' layers ( @xmath16 , no hydrogen layer ) instead of the canonical thick layer models ( @xmath118 ) are used .
however , part of the trend noted in v97 may be related to the systematic differences between the analyses discussed in sect 5.3 .
since v97 compared results from two different analyses , it is not possible to judge if the offset is real or just an artifact .
the case of fkb is stronger , because they based their case on a homogeneously analyzed sample .
fkb warned that the temperature dependence of the mass distribution peak may be caused by inadequacies remaining in the model atmospheres . a more detailed discussion is given in the next section .
some evidence that part of the temperature trend stems from the analyses comes from our intercomparisons presented in sect 5.3 .
we showed that systematic differences between the four investigated samples exist that vary with effective temperature .
this might mimic a temperature dependence of the sample peak mass .
a gravity offset of 0.1dex transforms into mass offsets of 0.050@xmath119 , @xmath120 , and @xmath121 for a @xmath5 white dwarf with 25000k , 40000k , and 60000k , respectively .
even a relatively small systematic @xmath44 difference of 0.05dex ( cf.table [ t : dlogg ] ) corresponds to a @xmath122 offset .
the scatter @xmath46 of _ individual _ gravity determinations reported in table [ t : scatter ] corresponds to @xmath123 , nearly independent of @xmath76 .
another part of the @xmath76 dependence of the mass peak may be caused by the use of wood s cooling tracks by fkb and v97 .
these tracks are not completely self - consistent since not all model sequences were not calculated _ ab initio _ from the main sequence .
this is not important for cooler white dwarfs but may cause deviations from _ ab initio _ tracks especially for the hottest white dwarfs , where structure depends sensitively on the evolutionary history ( blcker & schnberner 1990 ) .
we redetermined the masses of the bsl white dwarfs with the blcker / driebe mass - radius relations and derived a peak mass of @xmath124 .
that is @xmath117 lower than our peak mass , but having in mind the range of mass determination derived for the euve selected samples , we can not consider this a significant difference .
fkb divided their sample into cool ( @xmath125k ) and a hot ( 35000k @xmath12675000k ) subsamples and derived a @xmath127 higher peak mass for the hot sample .
our reanalysis of the fkb lte results with the blcker / driebe mass - radius relation yields virtually the same offset : @xmath128 .
however , the difference is brought down to @xmath129 ( @xmath130 vs. @xmath131 ) if we apply corrections to nlte .
one can imagine that a difference of this order ( if significant at all ) can easily be produced by our neglect of metallicity effects ( cf .
lanz et al .
1996 , barstow et al .
1998 ) .
therefore , given the combination of sample selection and systematic effects in analyses to date , our results do not confirm the presence of intrinsic , systematic mass differences between hot and cool white dwarfs .
we have obtained temperatures , gravities , and masses for a sample of 46 extreme ultraviolet selected da white dwarfs .
these data complement a near infrared survey for low mass companions .
the stellar parameters were determined by fitting the hydrogen balmer line profiles with nlte model spectra .
a map of lte correction vectors was constructed , which allows the transformation of lte results to the nlte scale .
these shifts become important for temperatures above 50000k ( see also napiwotzki 1997 ) .
three recent analyses of da white dwarfs have a considerable overlap with our sample and allow a direct check for systematic errors and individual measurement scatter .
although all four analyses applied similar model atmospheres and fitting techniques , we recognized systematic shifts up to 5% in temperature and 0.1dex in gravity determinations .
these systematic errors have to be considered when the results are interpreted and compared to other studies .
the individual measurement errors increase with temperature : @xmath132 from 2.3% to 3.3% and @xmath133 from 0.07dex to 0.13dex . this is larger than the typical errors computed with @xmath45 fit procedures for a well - exposed spectrum and
indicates that accuracy is limited by noise in combination with other effects such as details of the extraction , fluxing or normalization procedures .
it is unclear if these @xmath46 values indicate lower limits , or scale ( at least partially ) with the s / n ratio of the spectra .
it will be a challenge to resolve this question .
repeated observations performed by one observer , at one telescope , reduced the same way and analysed with one method as performed by bsl , fkb , and us ( table 2 ) do not reveal all effects .
independent observations analysed independently are necessary .
masses have been inferred from theoretical mass - radius relations based on the evolutionary calculations of blcker ( 1995 ) for c / o white dwarfs and driebe et al .
( 1998 ) for he white dwarfs
. an important feature of these calculations are the hydrogen and helium layer masses which depend on stellar mass .
we find a sharp peak centered at @xmath9@xmath0 in agreement with the previous investigations of ksw and bsl .
we redetermined masses of the bsl white dwarfs with the blcker / driebe mass - radius relation and derived a peak mass of @xmath134 . at first glance
this seems to confirm the systematic offset reported by v97 .
however , this offset could be explained by systematic differences of the model atmosphere analyses as well .
our reanalysis of the homogeneous fkb sample showed that the temperature dependence of the mass peak nearly vanishes , when corrected for their lte assumption .
we conclude that the observational data presented here and in similar studies are well explained by canonical stellar evolution theory , i.e. white dwarfs with thick envelopes .
we find only one object ( rej0512 - 004 ) with a mass below @xmath107 , which is a possible helium core white dwarf , but four white dwarfs with masses in excess of @xmath108 .
this ratio of high- and low - mass white dwarfs is quite different from that found by bsl .
this can partly be understood as the result of variable evolutionary time scales of high and low mass white dwarfs .
we agree with fkb that another part can be explained by a selection _ against _ low mass white dwarfs in euv selected samples .
the principal aim of our project is the search for and study of cool main sequence companions of our white dwarf sample .
five white dwarfs already show red contamination in our optical spectra . a comprehensive discussion of these and other binaries will be given in a forthcoming paper presenting ir photometry for the white dwarf sample . in this article we provided the basic white dwarf data necessary to interpret our binary sample .
we thank t. driebe , f. herwig , and t. blcker for providing us with their tracks and computing the white dwarf masses and d.koester , who made some model spectra available for our model comparison .
we are grateful to steward observatory for an unexpected , generous award of six nights of unclaimed 2.3-m time , which made possible simultaneous optical and infrared observations of our sample of stars .
we thank perry berlind for obtaining several white dwarf spectra .
pjg acknowledges support through nasa contract nas8 - 39073 ( asc ) .
barstow m. a. , hubeny i. , holberg j. b. 1998 , mnras , 299 , 520 beauchamp a. , wesemael f. , bergeron p. , liebert j. , saffer r. a. 1996 , in : hydrogen - deficient stars , eds .
c. s. jeffery & u. heber , asp conf .
96 , p. 295
bergeron p. 1993 , in : white dwarfs : advances in observation and theory , ed .
m. a. barstow , kluwer , dordrecht , p. 267
bergeron p. , saffer r. a. , & liebert j. 1992 , 394 , 228 ( bsl ) blcker t. 1995 , 299 , 755 blcker t. & schnberner d. 1990 , , 240 , l11 blcker t , herwig f , driebe t. , bramkamp h. , & schnberner d. 1997 , in : white dwarfs , eds .
i. isern et al . , kluwer , dordrecht , p. 57
bues i. , aslan t. 1995 , in : white dwarfs , eds .
d. koester & k. werner , springer , berlin , p. 259 burleigh , m. r. , barstow , m. a. , & fleming , t. a. 1997 , mnras , 287 , 381 cataln m. s. , sarna m. j. , jomaron c. m. , & smith r. c. 1995 , mnras , 275 , 153 dekool , m. & ritter , h. 1993 , a&a , 267 , 397 dreizler s. , heber u. , jordan s. , & engels d. 1994 , in : hot stars in the galactic halo , eds .
s. j. adelman , a. r. upgren , c. j. adelman , cambridge university press , p. 228
dreizler s. & werner k. 1996 , 314 , 217 dreizler s. , werner k. , & heber u. 1995 , in : white dwarfs , eds .
d. koester & k. werner , springer , berlin , p. 160 driebe t. , schnberner d. , blcker t. , herwig f. 1998 , , 339 , 123 filippenko a. v. 1982 , , 94 , 715 finley , d. s. , et al .
1993 , apj , 417 , 259 finley d. s. , koester d. , & basri g. 1997 apj 488 , 375 ( fkb ) green , p. j. & margon , b. 1994 , apj , 423 , 723 green r. f. , schmidt m. , & liebert j. 1986 apjs 61 , 305 hubeny i. , hummer d. g. , & lanz t. 1994 , , 282 , 151 hummer d. g. & mihalas d. 1988 , 331 , 794 iben , i. jr . & tutukov a. v. 1986 , 311 , 742 jordan s. , heber u. , weidemann v. 1991 , in : white dwarfs , ed .
g. vauclair & e. m. sion , kluwer , dordrecht , p. 121 kidder , k. m. 1991 , phd thesis , university of arizona kidder , k. m. , holberg , j. b. , barstow , m. a. , tweedy , r. w. , & wesemael , f. 1992 , apj , 394 , 288 kippenhahn r. , kohl k. , & weigert a. 1967 , zeitschrift f. astrophys . 66 , 58 koester d. , schulz h. , & weidemann v. 1979 , , 76 , 262 ( ksw ) lanz t. , barstow m. a. , hubeny i. , & holberg j. b. 1996 , , 473 , 1089 lemke m. 1997 122 , 285 liebert j. , bergeron p. , & saffer r. a. 1990 pasp , 102 , 1126 malina , r. et al .
1994 , aj , 107 , 751 marsh m. c. , barstow , m. a. , buckley d. a. et al .
1997 , mnras , 286 , 369 ( m97 ) massey p. , strobel s. , barnes j. v. , & anderson e. 1988 , , 328 , 315 napiwotzki r. 1997 , 322 , 256 napiwotzki r. , barstow m. a. , fleming t. , holweger h. , jordan s. , & werner k. 1993 , 278 , 478 pounds , k. et al .
1993 , mnras , 260 , 77 press w. h. , flannery b. p. teukolsky s. a. , & vetterling w. t. 1986 , numerical recipes , cambridge university press schmidt g. d. , bergeron p. , liebert j. , & saffer r. a. 1992 , , 394 , 603 schwartz r. d. , dawkins d. , findley d. , & chen d. 1995 , 107 , 667 sion e. m. , holberg j. b. , barstow m. a. , & kidder , k. m. 1995 , , 107 , 232 sweigart a. v. , greggio l. , & renzini a. 1990 , 364 , 527 vennes , s. & thorstensen , j. r. 1994 , apj , 433 , l29 vennes s. , thejll p. a. , galvan g. g. , & dupuis j. 1997 , , 480 , 714 ( v97 ) vennes s. , christian d. j. & thorstensen j. r. 1998 apj , 502 , 763 vidal c. r. , cooper j. , & smith e. w. 1970 jqsrt 10 , 1011 weidemann v. & koester d. 1984 , , 132 , 195 werner k. 1986 , 161 , 177 wonnacott d. , kellett b. j. , & stickland d. j. 1993 , mnras , 262 , 277 wood , m. 1994 , in iau coll . 147 : the equation of state in astrophysics , eds .
g. chabrier and e. schatzmann ( cambridge u. press : cambridge ) p. 612 .
wood m. 1995 , in : white dwarfs , eds .
d. koester & k. werner , springer , berlin , p. 41 | we present new determinations of effective temperature , surface gravity , and masses for a sample of 46 hot da white dwarfs selected from the euve and rosat wide field camera bright source lists in the course of a near - ir survey for low mass companions .
our analysis , based on hydrogen nlte model atmospheres , provides a map of lte correction vectors , which allow a thorough comparison with previous lte studies .
we find previous studies underestimate both the systematic errors and the observational scatter in the determination of white dwarf parameters obtained via fits to model atmospheres .
the structure of very hot or low mass white dwarfs depends sensitively on their history .
to compute white dwarf masses , we thus use theoretical mass - radius relations that take into account the complete evolution from the main sequence .
we find a peak mass of our white dwarf sample of @xmath0 , in agreement with the results of previous analyses .
however , we do not confirm a trend of peak mass with temperature reported in two previous analyses .
analogous to other euv selected samples , we note a lack of low mass white dwarfs , and a large fraction of massive white dwarfs .
only one white dwarf is likely to have a helium core .
while the lack of helium white dwarfs in our sample can be easily understood from their high cooling rate and therefore low detection probability in our temperature range , this is not enough to explain the large fraction of massive white dwarfs .
this feature very likely results from a decreased relative sample volume for low mass white dwarfs caused by interstellar absorption in euv selected samples . |
in 2003 babar collaboration discovered a positive - parity scalar charm strange meson @xmath1 with a very narrow width @xcite , which was confirmed by cleo @xcite later . in the same experiment cleo also observed the @xmath7 partner state at @xmath8 @xcite .
since these two states lie below the @xmath0 and @xmath9 threshold , respectively , the potentially dominant s - wave decay modes @xmath10 etc . , are kinematically forbidden .
thus the radiative decays and isospin - violating strong decays become the dominant decay modes .
therefore both of them are very narrow .
the discovery of these two states has triggered heated discussion on their nature in literature .
the key point is to understand their low masses .
the mass of @xmath1 is significantly lower than the expected values in the range of @xmath11 within quark models @xcite .
the model using the heavy - quark mass expansion of the relativistic bethe - salpeter equation predicted a lower value @xmath12 @xcite , which is still higher than the experimental data by about @xmath13 . from the experience with @xmath14 , van beveren and
rupp @xcite argued that the low mass of @xmath1 could arise from the mixing between the @xmath15 @xmath16 state and the @xmath0 continuum . in this way
the lowest @xmath15 state could be pushed much lower than that expected from the quark models .
the mass of @xmath2 state from the lattice qcd calculation is also significantly larger than the experimentally observed mass of @xmath1 @xcite .
it is also pointed out in ref .
@xcite that @xmath1 might receive a large component of @xmath0 continuum , which makes the lattice simulation very difficult .
the difficulty with the @xmath17 interpretation leads many authors to speculate that @xmath1 is a @xmath18 four quark state @xcite , or a strong @xmath19 atom @xcite .
however , calculations based on the quark model show that the mass of the four quark state is much larger than that of the @xmath15 @xmath16 state @xcite .
the radiative decay of @xmath1 also favors that it is a @xmath17 state @xcite .
furthermore , there are two @xmath15 states in the four quark system and one in the two - quark system .
only one @xmath15 state has been found below the @xmath20 resonance in the experimental search by babar @xcite , consistent with the @xmath17 interpretation .
this problem has been treated with qcd sum rules in the heavy quark effective theory in ref .
the resulting @xmath2 mass is consistent with the experimental data within large theoretical uncertainties .
however , the central value is still larger than the data by @xmath21 .
even larger result for the @xmath2 mass was obtained in the earlier work with the sum rule in full qcd @xcite .
it has been pointed out in ref .
@xcite that , in the formalism of qcd sum rules , the physics of mixing with @xmath0 continuum resides in the contribution of @xmath0 continuum in the sum rule , and including this characteristic contribution should render the mass of @xmath2 lower .
recently , there have been two investigations on this problem using sum rules in full qcd including the @xmath22 corrections . in ref .
@xcite the value of the charm quark pole mass @xmath23 is used , and the mass of @xmath15 @xmath16 state is found to be @xmath24 higher than the experimental data . on the other hand , in ref . @xcite the current quark mass @xmath25 ( corresponding to @xmath26 to @xmath22 ) is used , and the central value of the resulting @xmath15 @xmath16 mass is in agreement with the data . however , a low value of @xmath27 is used and the same value of the continuum threshold ( denoted by @xmath28 below ) is used for @xmath15 @xmath16 and @xmath29 @xmath16 . on the other hand , the perturbative three loop , order @xmath30 correction to the two - point correlation function with one heavy and one massless quark has been calculated @xcite .
it turns out that in the pole mass scheme used by many previous analyses including ref .
@xcite , the perturbative expansion is far from converging .
however , taking the quark mass in the modified minimal subtraction ( @xmath31 ) scheme @xcite , better convergence of the higher order corrections is obtained , and thus a more reliable determination of physical quantities of the lowest lying resonances becomes feasible @xcite . usually the contribution of two - particle continuum is neglected within the qcd sum rule formalism . however , because of the large s - wave coupling of @xmath32 @xcite and the adjacency of the @xmath2 mass to the @xmath0 threshold , this contribution may not be neglected in the considered case . in the present work , we shall therefore calculate this contribution and include it in the qcd sum rule . in the meantime
we take into account the perturbative three loop order @xmath30 correction and work in the @xmath31 scheme .
we find that the @xmath0 continuum contribution indeed renders both the mass and the decay constant of @xmath2 significantly lower . in section [ sec2 ] , we give a short overview of the traditional qcd sum rule for the scalar charm strange meson .
then we derive the @xmath0 continuum contribution and write down the full sum rule in section [ sec3 ] .
finally , the numerical results and our conclusions are presented in section [ sec4 ] . some relevant formulas and expressions used in this paper
are collected in appendices .
we consider the scalar correlation function @xmath33 where the renormalization invariant operator @xmath34 is defined as @xmath35 with @xmath27 and @xmath36 being the charm and strange quark current mass , respectively . up to a subtraction polynomial in @xmath37 , the correlation function @xmath38 satisfies the following dispersion relation @xmath39 at the quark gluon level , the spectra function @xmath40 is calculable using the renormalization group improved perturbation theory in the framework of the operator product expansion ( ope ) .
following jamin and lange @xcite , in this paper we shall adopt the @xmath31 running quark mass scheme rather than the pole mass one , and take into account both the @xmath41 terms in the perturbation theory obtained in ref .
@xcite and the corrections from the light quark mass up to order @xmath42 .
in addition we have included the contribution from the four quark condensation which affects the final result for @xmath2 mass only by a few @xmath43 . for convenience ,
all the relevant expressions for @xmath40 at the quark - gluon level , denoted by @xmath44 , are summarized in appendix a. on the other hand , @xmath40 can be phenomenologically written in terms of contributions from intermediate hadronic states .
generally , the spectral density at the hadronic level , denoted by @xmath45 , is taken to be the pole term of the lowest lying hadronic state plus the continuum starting from some threshold , with the latter identified with the qcd continuum @xmath46 where @xmath47 is the vector current decay constant of @xmath15 @xmath48 particle , analogous to @xmath49 .
@xmath50 is the mass of this particle , and @xmath28 is the continuum threshold above which the hadronic spectral density is modeled by that at the quark gluon level .
the recent works @xcite also use the above ansatz . after making the borel transformation to suppress the contribution of higher excited states and
invoking the quark - hadron duality , one arrives at the sum rule @xmath51 = \int \limits_{m_c^2}^{\infty } dt\ , \frac{\rho_{\rm qcd}(t)}{\pi}\ , \exp[-\frac{t}{m_b^2}]\,.\ ] ] following ref .
@xcite , the lower limit of the integration in the above equation is taken to be the charm quark pole mass @xmath52 , which can be expressed in terms of the running mass @xmath53 through the perturbative three - loop relation as defined in appendix b.
the contribution of two - particle continuum to the spectral density can safely be neglected in many cases , as usually done in the traditional qcd sum rule analysis .
one typical example is the @xmath54 meson sum rule , where the two pion continuum is of p - wave nature .
its contribution to the spectral density is tiny and the @xmath54 pole contribution dominates .
however , there may be an exception when the @xmath15 particle couples strongly to the two - particle continuum via s - wave .
in such case , there is no threshold suppression and the two - particle continuum contribution may be more significant . the strong coupling of the @xmath15 particle with the two - particle state and the adjacency of the @xmath15 mass to the @xmath0 continuum threshold result in large coupling channel effect , which corresponds to the configuration of mixing in the formalism of quark model . in the problem under consideration , the mass of @xmath1 is only about @xmath55 below the @xmath0 threshold , and the s - wave coupling of @xmath32 is found to be very large @xcite .
therefore , one may have to take into account the @xmath0 continuum contribution carefully .
the importance of @xmath56 continuum contribution in the sum rule for @xmath57 meson was first emphasized in ref .
@xcite , based on the duality consideration in the case where the @xmath57 mass is higher than the @xmath56 threshold .
based on the soft pion theorem , two of us also made a crude analysis of the @xmath58 continuum contribution in the case where the @xmath15 particle mass is higher than the threshold @xcite . in this work
, we calculate the continuum contribution more carefully in the case where the @xmath15 particle mass is lower and very close to the two - particle continuum threshold .
let @xmath59 be the form - factor defined by @xmath60 from the large s - wave coupling of @xmath32 and the adjacency of the @xmath2 mass to the @xmath0 threshold , one expects that in the low energy region , @xmath59 is dominated by the product of a factor of the @xmath2 pole and a factor from the final state interactions . in the low energy region with @xmath61
needed in our sum rule , the effect of inelastic @xmath0 scattering is suppressed by the phase space .
therefore , we take the approximation to consider only the @xmath0 scattering with only elastic intermediate states
. it can be described by the @xmath32 interaction and the @xmath62 chiral interaction in the low energy effective lagrangian , which can be represented by a series of bubble diagrams shown in fig .
[ fig1 ] .
the s - wave born amplitude of @xmath0 scattering represented by the black circles in fig . [ fig1 ] contains three terms .
the first one is the @xmath63 channel pole term @xmath64 , with @xmath65 being the @xmath32 coupling constant and @xmath66 being the mass parameter normalized at the scale @xmath67 in the effective lagrangian .
the second term corresponds to the direct @xmath62 interaction in the effective lagrangian .
let @xmath68 and @xmath69 be the four momentum of @xmath70 mesons in the initial and final state respectively , and @xmath71 . in the chiral effective lagrangian ,
the amplitudes for the processes @xmath72 , @xmath73 , and @xmath74 in the low energy @xmath75 region of @xmath76 meson needed in the qcd sum rule are all equal to @xmath77\,,\ ] ] in the center of mass system . here
@xmath78 and @xmath79 should be separately included in the integrals for two adjacent loops .
the third term of the s - wave born amplitude arises from the crossing @xmath80 channel pole term @xmath81 where @xmath82 for simplicity , we put the on - shell values of @xmath75 , @xmath83 , @xmath84 , @xmath85 into @xmath86 and @xmath87 in the above equations .
the effect of this approximation on our final results is expected to be small , since the contribution of the @xmath80 channel pole term is relatively small . as a result
, one finds that the @xmath80 channel pole term is an analytic function of @xmath63 with only a short cut , the length of which is only @xmath88 for experimental values of the corresponding masses .
therefore , it can be well approximated by a pole form @xmath89 , where @xmath90 } \,,\end{aligned}\ ] ] @xmath91 with the above results for the three terms of the s - wave born amplitude , we can now evaluate the sum of the series of the bubble diagrams shown in fig .
let @xmath92 be the partial sum of the series of loop diagrams in fig .
[ fig1 ] , with the loop number less or equal to @xmath93 .
it can then be written in the form @xmath94\,f_{n0 } + 2\,(\sqrt{t}-k'_0)\,f_{n1 } -f_{n2}\bigg\}\,,\ ] ] where @xmath83 and @xmath79 is the energy and momentum of final - state @xmath76 meson , and the three unknown functions @xmath95 ( @xmath96 ) correspond to the diagrams with a factor @xmath97 , @xmath75 , and @xmath98 respectively at the last vertex , which contributes to the integration over the last loop of each diagram .
let @xmath99 be integrals defined by eqs.(c1)-(c6 ) which appear as the loop integrals of the individual loop diagrams shown in fig .
they can be evaluated using dimensional regularization @xcite , with the corresponding analytic results given in ( c1)-(c6 ) .
a recurrence relations can be written between @xmath100 and @xmath101 , and hence between @xmath102 and @xmath103 , the coefficients of which are linear combinations of the loop - integral functions @xmath99 . taking the limit @xmath104 , @xmath105 and separating out terms with the factor @xmath97 , @xmath83 , and @xmath106
, we can obtain a system of three linear equations for the three unknown functions @xmath107 @xmath108 \,{\sqrt{t } } + 4\,\left ( f_1\,\sigma_1 + f_0\,\sigma_2 \right)\ , t \nonumber \\ & & + 4\,f_k^4\,g_0 ^ 2\,\big [ g_0\,\sigma_0 - f_0\,\left ( m_0 ^ 2 + g_0 ^ 2\,\sigma_0 - t \right ) \big]\ , \big [ \frac{c}{{\left ( t - t_0 \right ) } ^2 } + \frac{1}{\left ( t - m_0 ^ 2 \right ) \,\left ( t - t_0 \right ) } \big]\ , \nonumber \\ & & + 2\,f_k^2\,\frac{g_0\,\left ( 1 - 2\,f_0\,g_0 \right ) \,\left ( 2\,\sigma_1\,{\sqrt{t } } - \sigma_3 \right ) + 2\,f_1\,\big [ g_0 ^ 2\,\sigma_1 - \left ( m_0 ^ 2 + g_0 ^ 2\,\sigma_0 \right ) \,{\sqrt{t } } + t^{\frac{3}{2 } } \big]}{t - m_0 ^ 2}\,\nonumber \\ & & + f_2\,\bigg[\sigma_3 - 2\,\sigma_1\,{\sqrt{t } } + 2\ , f_k^2\ , g_0 ^ 2\,\sigma_0\,\left ( \frac{1}{t - m_0 ^ 2 } + \frac{c}{t - t_0 } \right ) - 2\ , f_k^2 \bigg]\ , \nonumber \\ & & + 2\,f_k^2\ , \frac{2\,c\,g_0 ^ 2\ , \big [ f_1\,\sigma_1 + f_0\,\sigma_3 - \left ( f_1\,\sigma_0 + 2\,f_0\,\sigma_1 \right ) \,{\sqrt{t } } \big]\,}{t - t_0}\ , \nonumber \\ & & + 4\,c\,f_0\,f_k^4\,g_0 ^ 4\,\sigma_0\ , \big [ \frac{c}{{\left ( t - t_0 \right ) } ^2 } + \frac{1}{\left ( t - m_0 ^ 2 \right ) \,\left ( t - t_0 \right ) } \big]=0 \,,\end{aligned}\ ] ] @xmath109 \,{\sqrt{t } } + 2\,f_1\,\left ( -f_k^2 + \sigma_2 - 2\,\sigma_1\,{\sqrt{t } } + \sigma_0\,t \right)\ , \nonumber \\ & & + 2\,f_k^2\,\frac{g_0\,\left ( 1 - f_0\,g_0 \right ) \,\sigma_1 + \big [ f_0\,m_0 ^ 2 - g_0\,\left ( 1 - f_0\,g_0 \right ) \,\sigma_0 \big ] \,{\sqrt{t } } - f_0\,t^{\frac{3}{2}}}{m_0 ^ 2 - t } + 2\,f_0\,\sigma_1\,t \nonumber \\ & & + 2\,f_k^2\ , \frac{c\,f_0\,g_0 ^ 2\,\left(\sigma_1 - \sigma_0\,{\sqrt{t } } \right ) } { t - t_0}=0\,,\end{aligned}\ ] ] @xmath110\ , \nonumber \\ & & + \sigma_0\,\left ( f_2 - \frac{2\,f_k^2\,g_0}{t - m_0 ^ 2 } - 2\,f_1\,{\sqrt{t } } \right)=0 \,,\end{aligned}\ ] ] from which the analytic forms for @xmath107 can then be deduced . the results are shown in eqs .
( c9)-(c12 ) . finally , with the explicit expressions for @xmath107 and @xmath99 given in appendix c , and putting the on - shell value of @xmath79 to eq .
( [ fnt ] ) , we obtain @xmath111\,f_{0 } + \frac{m_d^2-m_k^2}{\sqrt{t}}\,f_{1 } -f_{2}\bigg\ } \nonumber \\ & = & { { \lambda } \over { t - m_0 ^ 2-\delta(t)}}\,,\end{aligned}\ ] ] where @xmath112 is given by eq .
( [ deltat ] ) .
similarly , with the same series of @xmath0 loops included , the full propagator of @xmath2 meson is related to the function @xmath113 through @xmath114\,.\ ] ] using the solution for @xmath113 obtained above and given by eq .
( [ fit ] ) , it can be further rewritten as @xmath115 with @xmath116 given by eq .
( [ delta1 t ] ) .
we have chosen the scale @xmath117 so that the mass parameter @xmath66 is the physical mass of @xmath2 in our approximation , i.e. , @xmath118 .
the bare coupling constant @xmath65 is related to the physical coupling constant @xmath119 by @xmath120 , where @xmath121_{t = m_0 ^ 2}\,,\end{aligned}\ ] ] is the on - shell wave function renormalization constant of @xmath2 meson . in order to fix the unknown constant @xmath122 in @xmath59 given by eq .
( [ ft ] ) , we apply the soft - pion theorem @xmath123 to the extrapolated value of the matrix element @xmath124 at @xmath125 , from which we can deduce the constant @xmath126\,.\end{aligned}\ ] ] with all the above equipments , the @xmath0 continuum contribution to the hadronic spectral function can then be written as @xmath127 in the above calculations we have neglected the contribution of the @xmath128 channel .
the threshold of this channel is at @xmath129 .
our formula for the contribution of the two - particle term is proportional to @xmath130 .
the lower part of the integration in @xmath63 is more important . at the thresholds of the two channels
the factor @xmath130 for the @xmath0 channel is about 18 times larger than that for the @xmath128 channel .
therefore , the effect of the latter is expected to be small .
in our numerical analysis , we use the recent result for c quark current mass @xmath3 @xcite .
other input parameters are the following ( assuming the isospin symmetry ) : @xmath131 @xcite , @xmath132 @xcite , @xmath133 @xcite , @xmath134 @xcite , @xmath135 @xcite , @xmath136 @xcite , in the four quark condensation term eq .
( [ psibarpsi0 ] ) @xmath137 , @xmath138 @xcite , @xmath139 @xcite , @xmath140 @xcite , @xmath141 @xcite .
the renormalized coupling constant @xmath119 was determined to be in the interval @xmath142 in refs .
inclusion of the contribution of @xmath0 continuum in the sum rule analysis of the scalar current channel will lower the @xmath119 value . since the uncertainty is large , we have not calculated this correction and simply allow the renormalized @xmath119 to vary in the region @xmath143 .
a resonance of the @xmath144 system with the natural parity has has been observed experimentally at @xmath145 @xcite .
if it is an excited state of @xmath2 , we should confine us to @xmath28 smaller than and close to @xmath146 .
we shall first consider this case and then discuss the case that this resonance is not a @xmath15 state .
the convergence of the ope series and dominance of the sum by the pole and the @xmath0 continuum terms over the qcd continuum beyond @xmath28 constrain the borel mass @xmath147 in a region depending on the parameters @xmath148 and @xmath28 .
taking @xmath149 , @xmath150~{\rm gev}^2 $ ] for @xmath151 , and @xmath152~{\rm gev}^2 $ ] for @xmath153 . as mentioned in refs .
@xcite , the convergence of the perturbative expansion of the two - point correlation function , when written in terms of the pole quark mass , is rather poor , the order @xmath154 and @xmath30 loop contributions being of similar size with , or even larger than , the leading term , while the expansion in terms of the @xmath31 running mass converges much faster .
however , it should be noted that , even in the @xmath31 running mass scheme , the convergence of the asymptotic series in the @xmath144 meson system is worse than the one found in the @xmath155 meson system .
for @xmath2 the first order correction amounts to about @xmath156 and the second order to about @xmath157 of the leading term using the values of our input parameters and @xmath151 .
the same observation has also be made in ref .
@xcite .
we first move the @xmath0 continuum contribution to the right hand side of the sum rule .
then we obtain the curve of @xmath50 with respect to @xmath147 by taking the derivative of the logarithm of both sides of the sum rule as usually done .
since this curve depends on the unknown parameters @xmath66 , we have to do it self - consistently by requiring that the @xmath50 value determined by the sum rule for the input trial " value of @xmath66 both lies in the middle of the stability window and equals roughly to @xmath66 . for reliability of the results we also require that the ratio of @xmath0 contribution to the whole sum rule is not larger than about @xmath158 . with the input @xmath3 , we present the variation of @xmath50 with @xmath159 for @xmath151 and @xmath153 in figs .
[ fig2 ] and [ fig3 ] , respectively . for comparison
, we also show the case without @xmath0 continuum contribution with the same set of input parameters .
it can be seen clearly from the two figures that the inclusion of the @xmath0 continuum contribution can lower the @xmath50 value by @xmath160 .
the @xmath0 continuum contributes around @xmath161 to @xmath158 of the right hand side of the final sum rule for @xmath162 as shown in fig .
another interesting point is about the vector current decay constant @xmath163 of @xmath2 meson .
we find that the inclusion of @xmath0 continuum contribution lowers the decay constant @xmath163 from about @xmath164 to @xmath165 for the same @xmath28 value as can be seen from fig .
[ fig5 ] . for @xmath3 and @xmath5 , we found @xmath166 , being in agreement with the experimental data @xmath167 @xcite .
for the same value of input parameters , we found @xmath168 , which is , however , significantly lower than the ones obtained in previous literature .
here we have not included the errors due to uncertainties in the qcd sum rule except those from the variation of the results in the stability window and the @xmath28 interval , since our main interest is the central value of the results .
the previous results already shew that the @xmath2 mass lies in the large uncertainty interval of the qcd sum rule @xcite .
now we consider the case that the new resonance found in @xcite is not a @xmath15 state .
in this case the @xmath28 value can only be determined by stability analysis .
the results for the mass @xmath50 found for @xmath169 for @xmath170 and @xmath171 are shown in fig . 6 and fig .
7 respectively .
the working region for @xmath172 and @xmath173 are @xmath174~{\rm gev}^2 $ ] and @xmath175~{\rm gev}^2 $ ] respectively .
the region of the @xmath28 value can be chosen by requiring the least sensitivity of the results for the mass to the value of @xmath28 .
it is clear from these figures that this is the region between @xmath153 and @xmath151 which is just the region chosen above for the case of a @xmath15 resonance at @xmath176 .
also the best stability with respect to @xmath147 is achieved at @xmath153
. therefore , the results obtained above are essentially unchanged .
the above results show that the contribution of @xmath0 continuum , which contains the physics of the coupled channel effect in the formalism of qcd sum rule , is significant and is partly the reason for the unexpected low mass of @xmath15 @xmath177 state .
our analysis also explains partly why the extracted mass of the @xmath15 @xmath177 state from the quenched lattice qcd simulation is higher than the experimental value where the dk continuum contribution was not included .
* acknowledgements * we thank prof .
cheng for beneficial discussion and dr chun liu and dr .
xiang liu for their kind help .
s.l.z . was supported by the national natural science foundation of china under grants 10625521 and 10721063 and ministry of education of china .
in this appendix , all the relevant expressions for the spectral function @xmath40 are given .
for further details , we refer the readers to ref .
@xcite and references therein .
in perturbation theory , the spectral function @xmath44 has an expansion in powers of the strong coupling constant @xmath178 with @xmath179 . the leading order term @xmath180 results from a calculation of the bare quark - antiquark loop and
is given by @xmath181 and , up to order @xmath42 , the corrections in small mass @xmath36 can be found in ref .
@xcite @xmath182 where @xmath183 , and the appearing quark masses correspond to the running masses in the @xmath31 scheme with @xmath53 and @xmath184 evaluated at the scale @xmath185 the order @xmath154 correction @xmath186 can be written as @xmath187 + ( 1 - 2x)(3-x)\ln x + 3(1 - 3x)\ln \frac{\mu_m^2}{m_c^2 } + \frac{1}{2}(17 - 33x)\,\biggr\ } \,,\end{aligned}\ ] ] where @xmath188 is the dilogarithmic function .
the order @xmath154 mass corrections to the spectral function can be obtained by expanding the results given by @xcite up to order @xmath42 , after the higher dimensional operators have been expressed in terms of non - normal ordered condensates @xmath189 + 2(3 - 5x+x^2)\ln x + 3(2 - 3x)\ln \frac{\mu_m^2}{m_c^2 } + 2(7 - 9x)\,\biggr\ } \ , , \\
\vbox{\vskip 12 mm } \label{rho1m2 } \rho^{(1)}_{m^2}(s ) & = & -\,\frac{n_c}{8\pi^2}\,c_f\ , ( m_c+m_s)^2\,m_s^2\ , \biggl\{(1-x)\big[\,4l_2(x)+2\ln x\ln(1-x)\,\big ] { \nonumber}\\ { \vbox{\vskip 8mm}}&&\hspace{-0 mm } -\,(2+x)(4-x)\ln(1-x ) + ( 6 + 2x - x^2)\ln x + 6\ln \frac{\mu_m^2}{m_c^2 } + ( 8 - 3x)\,\biggr\ } \ , , \\
\vbox{\vskip 12 mm } \label{rho1m3 }
\rho^{(1)}_{m^3}(s ) & = & -\,\frac{n_c}{8\pi^2}\,c_f\ , ( m_c+m_s)^2\,\frac{m_c m_s^3 } { s}\,\biggl\{\,4l_2(x)+2\ln x\ln(1-x ) + \frac{(9 + 8x-9x^2)}{(1-x)^2 } { \nonumber}\\ { \vbox{\vskip 8mm}}&&\hspace{-0 mm } -\,2\frac{(7 + 7x-2x^2)}{(1-x)}\ln(1-x ) + 2\frac{(6 + 7x-2x^2 ) } { ( 1-x)}\ln x + 6\frac{(2-x^2)}{(1-x)^2}\ln\frac{\mu_m^2 } { m_c^2}\,\biggr\ } \ , , \\
\vbox{\vskip 12 mm } \label{rho1m4 }
\rho^{(1)}_{m^4}(s ) & = & \frac{n_c}{8\pi^2}\,c_f\ , ( m_c+m_s)^2\ , \frac{m_s^4}{s}\,\biggl\{\,2l_2(x)+\ln x\ln(1-x ) { \nonumber}\\ { \vbox{\vskip 8mm}}&&\hspace{-0 mm } -\,\frac{(13 - 24x-27x^2 + 2x^3)}{2(1-x)^2}\ln(1-x ) + \frac{(12 - 22x-27x^2 + 2x^3)}{2(1-x)^2}\ln x { \nonumber}\\ { \vbox{\vskip 8mm}}&&\hspace{-0 mm } + \,3\frac{(4 - 12x+x^2 + 3x^3)}{2(1-x)^3}\ln\frac{\mu_m^2}{m_c^2 } + \frac{(6 - 64x+15x^2 + 11x^3)}{4(1-x)^3}\,\biggr\ } \,.\end{aligned}\ ] ] the three - loop , order @xmath30 correction @xmath190 has been calculated by chetyrkin and steinhauser @xcite for the case of one heavy and one massless quark . in the present analysis , we shall make use of the program _ rvs.m _ , which contains the required expressions for @xmath190 @xcite . however , since the spectral function has been calculated only in the pole mass scheme , following jamin and lange @xcite , in the @xmath31 scheme we still have to add to @xmath190 the contributions resulting from rewriting the pole mass in terms of the @xmath31 mass
the two contributions @xmath191 and @xmath192 which arise from the leading and first order contributions , respectively , are given by @xmath193 \
\vbox{\vskip 9 mm } \label{del2rho2 } \delta_2\rho^{(2)}(s ) & \!=\ ! & -\,\frac{n_c}{8\pi^2}\,c_f\ , ( m_c+m_s)^2\,s\ , r_m^{(1)}\biggl\{(1-x)(1 - 3x)\big[\,4l_2(x)+2\ln x\ln(1-x)\,\big ] { \nonumber}\\ \vbox{\vskip 6 mm } & & \hspace{-0 mm } -\,(1-x)(7 - 21x+8x^2)\ln(1-x ) + ( 3 - 22x+29x^2 - 8x^3)\ln x { \nonumber}\\ \vbox{\vskip 9 mm } & & \hspace{-0 mm } + \,\frac{1}{2}(1-x)(15 - 31x)\,\biggr\ } \,,\end{aligned}\ ] ] where explicit expressions for the coefficients @xmath194 and @xmath195 can be found in appendix b. in the following , we summarize the contributions to the spectral function @xmath44 coming from higher dimensional operators , which arise in the framework of the ope and parameterize the appearance of non - perturbative physics .
since the spectral functions corresponding to the condensates contain @xmath196-distribution contributions , we shall present directly the borel transformed integrated quantity @xmath197 below , where @xmath198 with @xmath147 being the borel mass . the leading order expression for the dimension - three quark condensate is well known with the explicit form given by @xmath199 \,,\ ] ] where the expansion up to order @xmath200 has been included @xcite .
the first order correction to the quark condensate can be deduced based on the fact that the mass logarithms must cancel once the quark condensate is expressed in terms of the non - normal ordered condensate @xcite with @xmath201\ , e^{-m_c^2/u}\,\biggr\ } \,,\ ] ] where @xmath202 is the incomplete @xmath203-function . the next contribution in the ope
is the dimension - four gluon condensate with the corresponding expression given by @xmath204 the dimension - five mixed quark gluon condensate should also be included , since it is enhanced by the heavy quark mass and hence still has some influence on the sum rule . again
the result is well known with @xmath205 the last condensate contribution considered in this paper is the four - quark condensate @xmath206 where @xmath207 is the factor representing the deviation from vacuum saturation .
the contributions of all the other higher dimensional operators are extremely small and thus have been neglected .
the relationship between pole and running @xmath31 quark mass is given by @xcite @xmath208 \,,\ ] ] where @xmath209\ln\frac{\mu_m}{m(\mu_m ) } + \frac{\gamma_1}{2}\,(\gamma_1- \beta_1)\ln^2\frac{\mu_m}{m(\mu_m ) } { \nonumber}\\ { \vbox{\vskip 8 mm } } & & -\,\biggl[\,\gamma_1+\beta_1\ln\frac{\mu_m}{\mu_a}\,\biggr ] r_m^{(1 ) } \ , .
\label{eq : b.6}\end{aligned}\ ] ] the coefficients of the logarithms can be calculated from the renormalisation group @xcite , and the constant coefficients @xmath210 and @xmath211 are found to be @xcite @xmath212+c_ftn_f\biggl[\frac{71}{96}+\frac{1}{2}\zeta(2 ) \biggr ] { \nonumber}\\ { \vbox{\vskip 8 mm } } & & \hspace{-0mm}+\,c_ac_f\biggl[-\frac{1111}{384}+\frac{1}{2}\zeta(2)+ \frac{3}{8}\zeta(3)-\frac{3}{2}\zeta(2)\ln 2\biggr]+c_ft\biggl[\frac{3}{4}- \frac{3}{2}\zeta(2)\biggr ] \ , .
\label{eq : b.8}\end{aligned}\ ] ] with @xmath213 \ , , \qquad \beta_2 \ ; = \ ; \frac{1}{12}\,\big[\,17c_a^2 - 10c_atn_f-6c_ftn_f\,\big ] \,,\ ] ] and @xmath214 \,.\ ] ]
for convenience , in this appendix we collect some relevant expressions used in sec .
[ sec3 ] when discussing the @xmath0 continuum contribution .
firstly , we define the following loop integral functions @xmath99 ( with @xmath215 ) @xmath216 } \nonumber \\ & = & -\frac{1}{8\pi^2}\,b_0(t , m_d^2,m_k^2)\,,\end{aligned}\ ] ] @xmath217 } \nonumber \\ & = & -\frac{1}{8\pi^2}\,\bigg\{\,\frac{t+m_k^2-m_d^2}{2\sqrt{t}}\ , b_0(t , m_d^2,m_k^2 ) + \frac{1}{2\sqrt{t}}\,\big[a_0(m_d^2)-a_0(m_k^2)\big]\,\bigg\}\,,\end{aligned}\ ] ] @xmath218 } \nonumber \\ & = & -\frac{1}{8\pi^2}\,\bigg\{\,\frac{(t+m_k^2-m_d^2)^2}{4t}\ , b_0(t , m_d^2,m_k^2 ) + \frac{t+m_k^2-m_d^2}{4t}\,\big[a_0(m_d^2)-a_0(m_k^2)\big ] \nonumber \\ & & \hspace{5 mm } + \frac{1}{2}\,a_0(m_d^2)\,\bigg\}\,,\end{aligned}\ ] ] @xmath219 } \nonumber \\ & = & -\frac{1}{8\pi^2}\,\big[m_k^2\ , b_0(t , m_d^2,m_k^2 ) + a_0(m_d^2)\,\big]\,,\end{aligned}\ ] ] @xmath220 } \nonumber \\ & = & -\frac{1}{8\pi^2}\,\bigg\{\,\frac{t+m_k^2-m_d^2}{2\sqrt{t}}\ , m_k^2\,b_0(t , m_d^2,m_k^2 ) + \frac{m_k^2}{2\sqrt{t}}\ , \big[a_0(m_d^2)-a_0(m_k^2)\big ] \nonumber \\ & & \hspace{5 mm } + \sqrt{t}\,a_0(m_d^2)\,\bigg\}\,,\end{aligned}\ ] ] @xmath221 } \nonumber \\ & = & -\frac{1}{8\pi^2}\,\big[m_k^4\,b_0(t , m_d^2,m_k^2 ) + ( t+m_k^2+m_d^2)\,a_0(m_d^2)\,\big]\,.\end{aligned}\ ] ] here we have taken into account two intermediate states with different charges in eqs .
( [ sigma0])([sigma5 ] ) . @xmath222 and @xmath223 is the usual one - loop scalar one- and two - point function , respectively @xcite @xmath224\,,\end{aligned}\ ] ] @xmath225 } \nonumber \\ & = & ( \frac{2}{\epsilon}-\gamma_e+\ln{4\pi})+\ln{\mu^2 } - f_0(t , m_1 ^ 2,m_2 ^ 2)\,,\end{aligned}\ ] ] where @xmath226 in @xmath227-dimensional space time , @xmath117 is the introduced renormalization scale in dimensional regularization , and the explicit form of the function @xmath228 could be found in ref .
@xcite . from the three linear equations for the three unknown functions @xmath107 given by eqs .
( [ eq1])([eq3 ] ) , we can deduce the explicit expressions for the three unknown functions @xmath107 @xmath229 ( t - t_0)\ , , \\
f_1(t ) & = & \frac{1}{y(t)}\ , 2 f_k^4 g_0 \left[2 \sigma_1 f_k^2-\sigma_1 \sigma_3+\sigma_0 \sigma_4 + 2 \left(\sigma_1 ^ 2 - \sigma_0 \sigma_2\right ) \sqrt{t}\right ] ( t - t_0)\ , , \\
f_2(t ) & = & \frac{1}{y(t)}\ , 2 f_k^2 g_0 \left\{2 \sigma_3 f_k^4+\big[2 \sigma_1 \sigma_4-\sigma_3 ( 2 \sigma_2+\sigma_3)+\sigma_0 \sigma_5 + 2 ( \sigma_1 \sigma_3-\sigma_0 \sigma_4 ) \sqrt{t}\,\big ] f_k^2\ , \right . \nonumber \\ & & \left . \hspace*{1 cm } + \sigma_0 \sigma_4 ^ 2 - 2 \sigma_1 \sigma_3 \sigma_4+\sigma_1 ^ 2 \sigma_5+\sigma_2\left(\sigma_3 ^ 2-\sigma_0 \sigma_5\right)\right\ } ( t - t_0)\,,\end{aligned}\ ] ] with @xmath230 f_k^4\ , \nonumber \\ & & \hspace{-5 mm } + \left(m_0 ^ 2-t\right ) \left[4 t \sigma_1 ^ 2 - 4
\sigma_4 \sigma_1+\sigma_3 ^ 2-\sigma_0 \sigma_5 + 4 \sigma_2 ( \sigma_3-\sigma_0 t)-4 ( \sigma_1 \sigma_3-\sigma_0 \sigma_4 ) \sqrt{t}\right ] f_k^2\ , \nonumber \\ & & \hspace{-5 mm } + \big[2 \sigma_3 \sigma_4 \sigma_1-\sigma_5 \sigma_1 ^ 2-\sigma_0 \sigma_4 ^ 2-\sigma_2 \left(\sigma_3 ^ 2-\sigma_0 \sigma_5\right)\big ] \left(m_0 ^ 2-t\right)\biggr\ } ( t - t_0)\ , \nonumber \\ & & \hspace{-5 mm } -4 c f_k^4 g_0 ^ 2 \left(\sigma_0 f_k^2+\sigma_1 ^ 2-\sigma_0 \sigma_2\right ) \left(m_0 ^ 2-t\right)\,.\end{aligned}\ ] ] with the above results , the functions @xmath112 and @xmath116 can be , respectively , written as @xmath231 @xmath232 @xmath233 - 64f_k^4\,g_0 ^ 2\,\pi^2\,(t - t_0)\,a_0(m_k^2)\,a_0(m_d^2 ) { \nonumber}\\ & & \hspace{-3 mm } -32f_k^4\,g_0 ^
2\,\pi^2\,(t - t_0)\left[\,(m_d^2-m_k^2+t)\,a_0(m_d^2 ) -(m_d^2-m_k^2-t)\,a_0(m_k^2)\,\right ] { \nonumber}\\ & & \hspace{-3 mm } \times \,b_0(t , m_d^2,m_k^2 ) - 1024f_k^6\,g_0 ^ 2\,\pi^4\,t\,(t - t_0)\,b_0(t , m_d^2,m_k^2 ) \,,\end{aligned}\ ] ] @xmath234 { \nonumber}\\ & & \hspace{-3 mm } + ( t - t_0)\,\left[\ , ( t - m_d^2 + m_k^2)\ , a_0(m_d^2 ) + ( t + m_d^2 - m_k^2)\ , a_0(m_k^2)\ , \right]\ , a_0(m_d^2)\,a_0(m_k^2)\ , { \nonumber}\\ & & \hspace{-3 mm } + 32f_k^2\,\pi^2\,\left[\,2c\,f_k^2\,g_0 ^ 2 + ( m_d^2 + m_k^2 + t)\ , ( t - t_0)\,\right ] a_0(m_d^2)\ , a_0(m_k^2 ) - 32f_k^2\,\pi^2\ , \left\{\ , \right .
{ \nonumber}\\ & & \hspace{-3 mm } \left .
\times \left[c\,f_k^2\,g_0 ^ 2 + ( m_d^2 - t)\,(t - t_0 ) \right]\ , a_0(m_d^2)^2 + \left[c\,f_k^2\,g_0 ^ 2 + ( m_k^2 - t)\,(t - t_0 ) \right]\ , a_0(m_k^2)^2\ , \right\}\ , { \nonumber}\\ & & \hspace{-3 mm } + \left[\,(m_d^2 - m_k^2)^2 - t^2\,\right]\,(t - t_0)\ , a_0(m_d^2)\ , a_0(m_k^2)\ , b_0(t , m_d^2,m_k^2 ) { \nonumber}\\ & & \hspace{-3 mm } + 32f_k^2\,\pi^2\,\left\{(m_d^2 - m_k^2 + t)\,\left[c\,f_k^2\,g_0 ^ 2 + ( m_d^2 - t)\,(t - t_0)\,\right]\ , a_0(m_d^2 ) \right .
{ \nonumber}\\ & & \hspace{-3 mm } \left . -
( m_d^2 - m_k^2 - t)\,\left[c\,f_k^2\,g_0 ^ 2 + ( m_k^2 - t)\,(t - t_0)\,\right]\ , a_0(m_k^2 ) \right\ } \,b_0(t , m_d^2,m_k^2)\ , { \nonumber}\\ & & \hspace{-3 mm } + 256f_k^4\,\pi^4\ , \left\{\,4c\,f_k^2\,g_0 ^ 2\,t + \left[(m_d^2 - m_k^2)^2 + 2 ( m_d^2 + m_k^2)\,t - 3t^2 \right]\,(t - t_0 ) \right\}\ , { \nonumber}\\ & & \hspace{-3 mm } \times b_0(t , m_d^2,m_k^2 ) + 8192f_k^6\,\pi^6\ , t\ , ( t - t_0)\,.\end{aligned}\ ] ] b. aubert _ et al .
_ [ babar collaboration ] , phys .
lett . * 90 * , 242001 ( 2003 ) .
d. besson _ et al .
_ [ cleo collaboration ] , aip conf .
proc . * 698 * , 497 ( 2004 ) .
s. godfrey and r. kokoski , phys . rev .
* d 43 * , 1679 ( 1991 ) ; s. godfrey and n. isgur , phys . rev .
* d 32 * , 189 ( 1985 ) ; m. di pierro and e. eichten , phys . rev . *
d 64 * , 114004 ( 2001 ) .
k. g. chetyrkin and m. steinhauser , phys .
b * 502 * , 104 ( 2001 ) [ arxiv : hep - ph/0012002 ] .
k. g. chetyrkin and m. steinhauser , eur .
j. c * 21 * , 319 ( 2001 ) [ arxiv : hep - ph/0108017 ] .
w. a. bardeen , a. j. buras , d. w. duke , and t. muta , phys .
d*18 * , 3998 ( 1978 )
. j. h. kuhn , m. steinhauser and c. sturm , nucl .
b * 778 * , 192 ( 2007 ) .
s. bethke , prog .
. phys .
* 58 * , 351 ( 2007 ) .
s. narison , phys .
d * 74 * , 034013 ( 2006 ) .
w. m. yao _ et al . _
[ particle data group ] , j. phys .
g * 33 * , 1 ( 2006 ) .
j. bordes , j. penarrocha and k. schilcher , jhep * 0511 * , 014 ( 2005 ) [ arxiv : hep - ph/0507241 ] .
j. bordes , j. penarrocha and k. schilcher , jhep * 0412 * , 064 ( 2004 ) [ arxiv : hep - ph/0410328 ] .
m. jamin and m. munz , z. phys .
c * 60 * , 569 ( 1993 ) [ arxiv : hep - ph/9208201 ] .
d. j. broadhurst , phys .
b * 101 * , 423 ( 1981 ) .
s. c. generalis , j. phys .
g * 16 * , 785 ( 1990 ) .
v. p. spiridonov and k. g. chetyrkin , sov .
j. nucl .
* 47 * , 522 ( 1988 ) [ yad .
fiz . * 47 * , 818 ( 1988 ) ] . k. g. chetyrkin , c. a. dominguez , d. pirjol and k. schilcher , phys .
d * 51 * , 5090 ( 1995 ) .
k. g. chetyrkin , j. h. kuhn and m. steinhauser , nucl .
b * 482 * , 213 ( 1996 ) .
r. tarrach , nucl . phys .
b * 183 * , 384 ( 1981 ) .
n. gray , d. j. broadhurst , w. grafe and k. schilcher , z. phys .
c * 48 * , 673 ( 1990 ) .
g. passarino and m. j. g. veltman , nucl .
b * 160 * , 151 ( 1979 ) .
m. drees , _ based on a talk given at 2nd workshop on high energy physics phenomenology , calcutta , india , jan 2 - 15 , 1991 . | using the soft - pion theorem and the assumption on the final - state interactions , we include the contribution of @xmath0 continuum into the qcd sum rules for @xmath1 meson .
we find that this contribution can significantly lower the mass and the decay constant of @xmath2 state .
for the value of the current quark mass @xmath3 , we obtain the mass of @xmath2 @xmath4 in the interval @xmath5 , being in agreement with the experimental data , and the vector current decay constant of @xmath2 @xmath6 , much lower than those obtained in previous literature . |
for a given non - principal dirichlet character @xmath13 , where @xmath6 is an odd prime , let @xmath14 this quantity plays a fundamental role in many areas of number theory , from modular arithmetic to @xmath15-functions .
our goal in this paper is to understand how often @xmath0 is large , and to gain insight into the structure of those characters @xmath16 for which @xmath0 is large .
it makes sense to renormalize @xmath0 by defining @xmath17 and we believe that @xmath18 where @xmath19 is the euler mascheroni constant .
rnyi @xcite observed that @xmath20 with @xmath21 upper bounds on @xmath0 ( and hence on @xmath22 ) have a rich history . the 1919 plya
vinogradov theorem states that @xmath23 for all non - principal characters @xmath16 .
apart from some improvements on the implicit constant @xcite , this remains the state - of - the - art for the general non - principal character , and any improvement of this bound would have immediate consequences for other number theoretic questions ( see e.g. @xcite ) .
montgomery and vaughan @xcite ( as improved in @xcite ) have shown that the generalized riemann hypothesis implies @xmath24 whereas , for every prime @xmath6 , there are characters @xmath16 for which @xmath25 ( see @xcite , which improve on paley @xcite ) , so the conjectured upper bound is the best one could hope for .
however , for the vast majority of the characters @xmath16 , @xmath0 is somewhat smaller , and so we study the distribution function @xmath26 montgomery and vaughan @xcite showed that @xmath27 for all @xmath28 for any fixed @xmath29 ( they equivalently phrase this in terms of the moments of @xmath0 ) .
this was recently improved by bober and goldmakher @xcite , who proved that , for @xmath30 fixed and as @xmath31 through the primes , @xmath32 where @xmath33 is some positive constant and @xmath34 and @xmath35 is the modified bessel function of the first kind given by @xmath36 we improve upon these results and demonstrate that @xmath37 decays in a double exponential fashion : [ main thm ] let @xmath38 .
if @xmath6 is a prime number , and @xmath39 for some @xmath40 , then @xmath41 where @xmath42 the proof implies that @xmath43 of the characters for which @xmath44 are odd , so it makes sense to consider odd and even characters separately .
hence we define @xmath45 we outline the proof of theorem [ main thm ] in section [ outline ] , and fill in the details in subsequent sections .
the range of uniformity barely misses implying the upper bound in ; even so , it does show that @xmath22 is rarely large .
calculations reveal that @xmath46 each tend to a universal distribution function @xmath47 , and we will show this for @xmath7 later on . in figure [ figure - phiq - plot ]
we graph @xmath37 for a typical @xmath6 : , @xmath48 , and @xmath49 for @xmath50 ] notice that @xmath51 for all @xmath52 , and then decays quickly : @xmath53 and @xmath54 . for more computational data ,
the reader is invited to consult table [ minmax ] in section [ moredata ] .
the distribution of @xmath55 , for @xmath6 prime , decays similarly @xcite : @xmath56 uniformly for @xmath57 , @xmath58 .
this similarity is no accident , since for an odd , non - primitive character @xmath16 , the average of the character sum is @xmath59 where @xmath60 is the gauss sum , so that @xmath61 . in particular , if @xmath62 is large , then so is @xmath22 .
moreover , for @xmath8 as above , we have the pointwise formula @xmath63 which implies that @xmath64 a little larger than @xmath55 if @xmath65 .
the distribution of @xmath66 can be analyzed in the same way as the distribution of @xmath55 . however , even if @xmath67 , then we can show that the average of our character sum up to @xmath68 is slightly larger than @xmath55 when @xmath68 is close to @xmath11 .
this builds on ideas in @xcite .
[ comparison thm ] let @xmath6 be an odd prime , @xmath69 for a sufficiently large absolute constant @xmath70 .
there exists a subset @xmath71 of cardinality @xmath72 such that if @xmath73 and @xmath74 is sufficiently large , then @xmath75 \ll \sqrt{q } \cdot \frac{(\log \tau)^2}{\sqrt{\tau } } .
\end{split}\ ] ] we can deduce that @xmath76 this result , together with , suggests that the lower bound in theorem [ main thm ] should be sharp , and that @xmath77 when these quantities are large " .
however the numerical evidence indicates that @xmath22 is perhaps typically a little bigger : ( vertical axis ) and @xmath78 for the 13617 odd characters with @xmath79 , and @xmath6 prime , @xmath80 . ] the numerical comparison between @xmath81 and @xmath82 ( the @xmath55-distribution for odd characters @xmath8 ) is given in figure 3 : , @xmath50 , with a dashed line indicating @xmath83 . ]
the focus above has been on odd characters since almost all characters with @xmath44 are odd . however , we can obtain analogous results for even characters .
[ main thm even ] there exists an absolute constant @xmath84 such that if @xmath6 is an odd prime and @xmath85 for some @xmath58 , then @xmath86 the lower bound in the above theorem has the same shape as and it might be close to the truth , unlike the corresponding result for @xmath81 ( as explained by theorem [ comparison thm ] ) .
this is supported by computations , as it can be seen in figure [ figure - l - even - plot ] . , @xmath50 .
] classically , the statistical behavior of @xmath87 as @xmath8 varies over all non - principal characters mod @xmath6 is modelled by @xmath88 , where the @xmath89 are independent random variables , each uniformly distributed on the unit circle @xmath90 , and @xmath91 for @xmath92 .
this suggests modelling @xmath93 , @xmath16 , by @xmath94 .
indeed , in @xcite this was shown to be a very successful model .
similarly , one might guess that the distribution of @xmath95 could be accurately modelled by @xmath96 ( and thus @xmath0 by @xmath97 ) , but this seems unlikely since , for fixed @xmath98 , we have that @xmath99 \sim \alpha q \
\ \text { whereas } \ \
\frac{1}{\phi(q)}\sum_{\substack{\chi{{\ifmmode\text{\rm\,(mod\,$q$)}\else\discretionary{}{}{\hbox { } } \rm(mod~$q$)\fi } } \\
\chi\neq \chi_0 } } { \left\lvert\sum_{n\le \alpha q } \chi(n ) \right\rvert}^2 \sim \alpha(1-\alpha ) q\ ] ] as @xmath100 .
moreover harper @xcite recently showed that @xmath96 is not normally distributed , and subsequently we have no idea what its distribution should be ( though @xcite shows that @xmath101 is normally distributed provided @xmath102 is not too large ) .
so , what is the right way to model @xmath0 ?
one reason the above model failed is that it does not take account of characters periodicity .
the periodicity of @xmath8 , via the formula @xmath103 where @xmath104 , leads to plya s formula ( * ? ? ?
( 9.19 ) , p. 311 ) : @xmath105 ( which is periodic for @xmath106 ) . since @xmath107
, this suggests modelling @xmath108 by the random variable @xmath109 where @xmath110 is a random variable independent of the @xmath111 s , with probability 1/2 of being @xmath112 or @xmath9 , and we have set @xmath113 .
the infinite sum here converges with probability 1 , applied to @xmath88 in place of @xmath87 .
] so that @xmath114 is well defined , and we can consider its distribution function , @xmath115 the following result confirms our intuition that @xmath114 serves as a good model for @xmath116 .
[ distributiontheorem ] if @xmath117 and @xmath7 is continuous at every point of @xmath118 $ ] , then the sequence of functions @xmath119 converges to @xmath7 uniformly on @xmath118 $ ] .
in particular , the sequence of distributions @xmath119 converges weakly to @xmath7 .
theorem [ distributiontheorem ] will be shown in section [ prob ] .
since @xmath7 is a decreasing function , it has at most countably many discontinuities , and so theorem [ distributiontheorem ] implies that @xmath120 for almost all @xmath121 .
we conjecture that @xmath7 is a continuous function .
we can prove that , for any @xmath28 , @xmath122 by considering two points @xmath123 and @xmath124 where @xmath7 is continuous , with @xmath125 so that @xmath126 .
we can bound @xmath127 and @xmath128 by theorems [ distributiontheorem ] and [ main thm ] , and then the result follows by letting @xmath129 and @xmath130 . the authors would like to thank sandro bettin , adam harper , k. soundararajan , akshay venkatesh , john voight , and trevor wooley for some helpful discussions .
ag and dk are partially supported by discovery grants from the natural sciences and engineering research council of canada . given a positive integer @xmath131 , we let @xmath132 and @xmath133 be the largest and smallest prime divisors of @xmath131 , with the notational convention that @xmath134 and @xmath135 . furthermore , we write @xmath136 for the set of integers all of whose prime factors lie in @xmath137 $ ] .
the symbol @xmath138 denotes the number of ways @xmath131 can be written as a product of @xmath139 positive integers .
we denote with @xmath140 the number of distinct primes that divide @xmath131 , and with @xmath141 the total number of prime factors of @xmath131 , counting multiplicity .
finally , we use the symbol @xmath142 to indicate the characteristic function of the set @xmath143 .
for example , @xmath144 equals 1 if @xmath145 and 0 otherwise .
we now determine more precise information about the structure of characters with large @xmath0 [ structure thm ] let @xmath6 be an odd prime and @xmath146 for a sufficiently large absolute constant @xmath70 .
there exists a set @xmath147 of cardinality @xmath148 for which the following holds : if @xmath149 , then 1 .
@xmath8 is an odd character .
2 . let @xmath150 $ ] , and @xmath151 be the reduced fraction for which @xmath152 with @xmath153 .
let @xmath154 if @xmath155 is prime , and @xmath156 otherwise .
then @xmath157 @xmath158 granville and soundararajan showed in @xcite that if @xmath0 is large then @xmath8 must pretend to be a character @xmath159 of small conductor and of opposite parity ( that is , @xmath8 is `` @xmath159-pretentious '' ) .
building on the techniques of @xcite , theorem [ 1-pretentious ] shows that , for the vast majority of characters with large @xmath0 , the character @xmath159 is the trivial character , and @xmath8 must be odd ( and so @xmath9-pretentious as in ) .
the discrepancy between theorem [ main thm ] and means that the error term in can not be made @xmath160 .
inequality allows us to establish accurate estimates for @xmath161 for all @xmath68 . here and for the rest of the paper , for @xmath162 we set @xmath163 where @xmath164 is the dickman
de bruijn function , defined by @xmath165 for @xmath166 $ ] and via the differential - delay equation @xmath167 for @xmath168 .
then we have the following estimate : [ partial sums odd ] let @xmath169 and @xmath149 be as in theorem [ structure thm ] . given @xmath170
$ ] , let @xmath171 be the reduced fraction for which @xmath172 with @xmath173 . define @xmath174 by @xmath175 . 1 . if @xmath156 , then @xmath176 2 . if @xmath177 , then @xmath178 if @xmath179 then @xmath180 , so @xmath181 is a measure of the distance between @xmath182 and @xmath171 .
theorems [ structure thm ] and [ partial sums odd ] will be proved in section [ odd ] .
analogous results can also be proven for even characters : [ structure thm even ] let @xmath6 be an odd prime and @xmath183 for a sufficiently large constant @xmath70 .
there exists a set @xmath184 of cardinality @xmath185 for which the following statements hold . if @xmath186 then 1 .
if @xmath150 $ ] , and @xmath151 is the reduced fraction for which @xmath152 with @xmath153 , then @xmath187 .
we have that @xmath188 @xmath189 3 .
given @xmath170 $ ] , let @xmath171 be the reduced fraction for which @xmath172 with @xmath173 . define @xmath174 by @xmath190 .
then @xmath191 the proof of theorem [ structure thm even ] will be given in section [ even ] .
before we get started with the proofs of our main results , we state and prove some facts about smooth numbers .
as usual , we set @xmath192 we begin with the following simple estimates .
[ rho ] for @xmath193 , we have that @xmath194 the claimed estimates follow by corollary 2.3 in @xcite and corollary 8.3 in section iii.5 of @xcite .
[ smooth ] for @xmath195 and @xmath193 , we have that @xmath196 without loss of generality , we may assume that @xmath197 .
when @xmath198 $ ] , de bruijn @xcite showed that @xmath199 therefore , lemma [ rho ] yields that @xmath200 together with partial summation , this implies that @xmath201 finally , if @xmath202 , then we note that @xmath203 since @xmath204 for @xmath205 , the product over the primes is @xmath206 , which completes the proof of the lemma .
we also need the following result , where @xmath207 is defined by .
[ smooth2 ] if @xmath195 and @xmath174 , then @xmath208 in particular , @xmath209 .
if @xmath210 , then the lemma follows from the estimate @xmath211 .
assume now that @xmath193 , and set @xmath212 .
lemma [ smooth ] implies that @xmath213 we use de bruijn s estimate and partial summation to deduce that @xmath214 thus completing the proof of the lemma .
using the slightly more accurate approximation @xmath215 for @xmath216\cup[1+\epsilon,\sqrt{\log y}]$ ] ( see @xcite ) , one can similarly deduce the stronger estimate @xmath217 for all @xmath174 . finally , we have the following key estimate .
its second part is a strengthening of theorem 11 in @xcite .
[ around a / b ] let @xmath218 and @xmath219 . 1 .
we have that @xmath220 2 .
there is an absolute constant @xmath221 such that if @xmath222 , @xmath151 is a reduced fraction with @xmath223 and @xmath224 , then @xmath225 where @xmath226 is defined by @xmath227 and @xmath228 is von mangoldt s function .
\(a ) if @xmath229 , then @xmath230 , and so @xmath231 hence it suffices to show that @xmath232 note that if @xmath233 , then @xmath234 using partial summation on the estimate @xmath235 .
therefore , in order to complete the proof of , it suffices to show that @xmath236 where @xmath237 .
moreover , lemma [ smooth ] reduces the above relation to showing that @xmath238 where @xmath239 . in particular
, we may assume that @xmath240 . if @xmath241 $ ] , then theorem 10 in @xcite and relation imply that @xmath242 provided that @xmath243 , for @xmath70 sufficiently large .
( this lower bound on @xmath244 guarantees that the parameter @xmath6 in theorem 10 of @xcite is @xmath245 , as required . )
consequently , if we set @xmath246 , then partial summation implies that @xmath247 then follows from @xmath248 which is trivial unless @xmath249 , so that @xmath250 .
let @xmath251 for @xmath252 .
let @xmath253 be the largest integer for which @xmath254 .
then @xmath255 .
if @xmath256 $ ] for some @xmath257 and @xmath258 , then theorem 3 in @xcite implies that @xmath259 therefore , partial summation and lemma [ rho ] imply that @xmath260 summing the above relation over @xmath257 implies that @xmath261 finally , integration by parts gives us that @xmath262 by the definition of @xmath263 and lemma [ rho ] .
this completes the proof of and thus the proof of part ( a ) of the lemma .
\(b ) write @xmath264 and set @xmath265 .
we note that , since @xmath266 , by assumption , we must have that @xmath267 .
we start by estimating the part of the sum with @xmath268 . we claim that @xmath269 note that if @xmath270 , then @xmath271 by partial summation and the estimate @xmath272 .
therefore , it suffices to prove that @xmath273 where @xmath274 .
furthermore , lemma [ smooth ] reduces the above estimate to showing that @xmath275 where @xmath276 . for this sum to have any summands we need that @xmath277 .
next , we fix @xmath278 $ ] and estimate @xmath279 for @xmath280 $ ] .
let @xmath33 be the constant in theorem 10 of @xcite when the parameters @xmath143 and @xmath281 there equal 100 and @xmath282 , respectively , and set @xmath283 and @xmath284 .
there is some reduced fraction @xmath285 with @xmath286 and @xmath287 .
note that @xmath288 in particular , we must have that @xmath289 . if , in addition , @xmath290 , then ( a straightforward modification of ) theorem 11 in @xcite and relation imply that @xmath291 where @xmath258 .
therefore @xmath292 if @xmath293 .
consider now @xmath294 such that @xmath295 and set @xmath296 .
we claim that @xmath297 .
we argue by contradiction : assume , instead , that @xmath298
. then @xmath299 and thus @xmath300 and @xmath301 .
in particular , @xmath302 and thus @xmath303 , which implies that @xmath304 , a contradiction . using the above information
, we are going to show that @xmath305 when @xmath295 .
if this relation holds , then follows by a straightforward dyadic decomposition argument .
we use the stronger relation for the @xmath306 dyadic intervals corresponding to @xmath307 $ ] , and we use for the @xmath308 dyadic intervals with @xmath309 $ ] . it remains to prove when @xmath295 , in which case @xmath294 lies in the interval @xmath310 $ ] .
in addition , note that @xmath311 .
let @xmath312 for @xmath252 .
let @xmath253 be the largest integer for which @xmath313 . then @xmath314 .
consider @xmath315 $ ] for some @xmath257 and set @xmath258 .
since @xmath316 , theorems 2 and 5 in @xcite imply that @xmath317 since @xmath318 ( see , for example , lemma [ gauss sum ] below ) .
note that @xmath319 for @xmath320 .
therefore @xmath321 we apply theorem 3 in @xcite to find that @xmath322 for @xmath323 $ ] and @xmath324 .
consequently , @xmath325 for some constant @xmath326 .
since @xmath327 , applying partial summation as in the proof of part ( a ) , we deduce that @xmath328 setting @xmath329 and summing the above relation over @xmath257 implies that @xmath330 finally , we have that @xmath331 whence @xmath332 which completes the proof of and thus of . to conclude , we have shown that @xmath333 since @xmath334 , we further deduce that @xmath335 observe that @xmath336 where @xmath337 denotes the distance of @xmath338 from its nearest integer . note that @xmath339 by our assumption that @xmath340 . since we also have that @xmath341 , a consequence of the fact that @xmath267 , the error term in is @xmath342 by our assumption that @xmath340 .
hence , the lemma is reduced to showing that @xmath343 by lemma [ smooth ] , it suffices to show that @xmath344 where @xmath345 .
since @xmath346 , the argument leading to implies that @xmath347 for @xmath348 $ ]
. therefore partial summation yields @xmath349 arguing as in lemma [ smooth2 ] , we apply partial summation on relation to deduce that @xmath350 consequently , @xmath351 since @xmath352 and @xmath353 by lemma [ rho ] .
this completes the proof of and thus the proof of the lemma .
[ fund ineq lem ] let @xmath8 be a dirichlet character modulo @xmath6 and @xmath354 . for all @xmath355 , we have @xmath356 if @xmath8 is an even character , then @xmath357 , so that @xmath358 by the prime number theorem , as desired .
we may therefore assume that @xmath8 is an odd character , so that @xmath359 if @xmath360 , where @xmath361 is the constant from lemma [ around a / b](b ) , then lemma [ around a / b](a ) implies that @xmath362 which is admissible .
finally , assume that @xmath363 , and consider a reduced fraction @xmath151 with @xmath364 and @xmath365 .
then lemma [ around a / b](b ) and the prime number theorem imply that @xmath366 where @xmath367 . since @xmath368 and @xmath369 , we deduce that @xmath370 which concludes the proof .
we first deal with theorem [ main thm even ] . for the lower bound , note that
if @xmath8 is even and @xmath371 , then ( * ? ?
( 9.18 ) , p. 310 ) yields the inequality @xmath372 since we also have that @xmath373 we deduce that @xmath374 for all even characters @xmath8 .
the lower bound in theorem [ main thm even ] is a direct consequence of the above inequality and of the following result , whose proof is a straightforward application of the methods in @xcite : [ distr of euler products ] if @xmath159 is a character modulo some @xmath375 , @xmath376 is the set of odd or of even characters mod @xmath6 , and @xmath377 for some @xmath378 , then @xmath379 finally , the upper bound in theorem [ main thm even ] follows from theorems [ distr of euler products ] and [ structure thm even ] . (
the proof of the latter theorem is independent of the proof of the upper bound of theorem [ main thm even ] , as we will see . )
next , we turn to theorem [ main thm ] .
its lower bound is a direct consequence of theorems [ comparison thm ] and [ distr of euler products ] .
so it remains to outline the proof of the upper bound in theorem [ main thm ] , as well as to prove theorem [ comparison thm ] . in the heart of these two proofs
lies a moment estimate which implies that the bulk of the contribution in plya s fourier expansion comes from smooth inputs : @xmath380 for most @xmath8 and any @xmath244 . to state this
more precisely , we need some notation .
given a set of positive integers @xmath381 , set @xmath382 } \left| \sum _ { n\in { { \mathcal a } } } \frac{\chi(n)e(n\alpha)}{n } \right| ; \ ] ] in the special case when @xmath383 , write @xmath384 in place of @xmath385 . observe that @xmath386 } \left| \sum_{1\le |n|\le z } \frac{\chi(n)(1-e(n\alpha))}{n } - \sum_{\substack{1\le |n|\le z
\\ p^+(n)\le y } } \frac{\chi(n)(1-e(n\alpha))}{n } \right| \\ & \quad = \max_{\alpha\in[0,1 ] } \left| \sum_{\substack{1\le |n|\le z \\ p^+(n ) > y } } \frac{\chi(n)(1-e(n\alpha))}{n } \right| \le \left| \sum_{\substack{1\le |n|\le z \\
p^+(n ) > y } } \frac{\chi(n)}{n } \right| + \max_{\alpha\in[0,1 ] } \left| \sum_{\substack{1\le |n|\le z \\ p^+(n ) > y } } \frac{\chi(n)e(n\alpha)}{n } \right| \\ & \quad \le 2 \left| \sum_{\substack{n\le z \\
p^+(n ) > y } } \frac{\chi(n)}{n } \right| + 2 \max_{\alpha\in[0,1 ] } \left| \sum_{\substack{n\le z \\
p^+(n ) > y } } \frac{\chi(n)e(n\alpha)}{n } \right| \le 4 s_{y , z}(\chi ) .
\end{split}\ ] ] our next goal is to show that @xmath384 is small for most @xmath8 . to do this
, we will prove in section [ moments ] that high moments of @xmath384 are small . as a straightforward application of our moment bounds we will get the following theorem . [ tail bound cor ] if @xmath387 , @xmath388 and @xmath389 $ ] , then @xmath390 we now show how to complete the proof of theorems [ main thm ] and [ comparison thm ] .
let @xmath391 $ ] be such that @xmath392 .
by plya s expansion , and lemma [ fund ineq lem ] , we find that @xmath393 for all @xmath218 , where @xmath394 .
we set @xmath395 for some parameter @xmath389 $ ] to be chosen later .
theorem [ tail bound cor ] then implies that @xmath396 taking @xmath397 completes the proof of the upper bound in theorem [ main thm ] .
we conclude this section with the proof of theorem [ comparison thm ] .
the set @xmath398 in which we work is defined by @xmath399 where we have set @xmath400 for some constant @xmath74 .
if the constant @xmath70 in the statement of theorem [ comparison thm ] is large enough , then theorems [ tail bound cor ] and [ distr of euler products ] imply that does hold .
assume now that @xmath73 .
using partial summation on the plya
vinogradov inequality , then that @xmath401 , and finally lemma [ smooth ] , we obtain @xmath402 given that @xmath403 , we deduce that @xmath404 by mertens s estimate .
therefore @xmath405 taking logarithms , we find that @xmath406 that is , @xmath8 is `` @xmath9-pretentious '' . since @xmath407 for @xmath408 , the above inequality and the cauchy schwarz inequality imply that @xmath409 for all @xmath410 $ ] .
moreover , since @xmath411 we find that @xmath412 now , since @xmath8 is odd for @xmath413 , plya s fourier expansion implies that @xmath414 set @xmath415 if @xmath416 and @xmath417 is sufficiently large , then lemma [ around a / b ] and relation imply that @xmath418 now , lemma [ smooth ] and relation yield that @xmath419 for some complex numbers @xmath420 of modulus @xmath421 . finally ,
if @xmath422 , then we simply note the trivial bound @xmath423 , which follows by our assumption that @xmath424 for the @xmath8 we are working with .
so will certainly follow if we show that @xmath425 by cauchy
schwarz , it suffices to show that @xmath426 for convenience , set @xmath427 , and note that @xmath428 where @xmath429 therefore @xmath430 note that @xmath431 @xmath432 for @xmath433 , and @xmath434 using the bound @xmath435 , we conclude that @xmath436 since @xmath437 for @xmath438 , follows .
this completes the proof of .
finally , note that relation clearly implies that @xmath439 relations and with @xmath440 imply that @xmath441 if @xmath442 then @xmath443 .
therefore @xmath444 , and so @xmath445 , which completes the proof of the theorem .
in this section we show that for most @xmath8 we can limit the fourier expansion to a sum over very smooth numbers without much loss , which is the content of theorem [ tail bound cor ] .
we prove this theorem by showing that high moments of @xmath446 are small : [ tail bound ] let @xmath6 and @xmath139 be integers with @xmath447 . for @xmath448 , we have that @xmath449 one consequence of theorem [ tail bound ] is the desired conclusion that @xmath384 is usually small : we may assume that @xmath102 and @xmath6 are large .
let @xmath164 be the proportion of characters @xmath16 such that @xmath450 . moreover , set @xmath451 where @xmath417 is a constant to be determined .
then theorem [ tail bound ] implies that @xmath452 which completes the proof .
we prove theorem [ tail bound ] as an application of the following technical estimates .
[ tail bound prelim ] let @xmath6 and @xmath139 be two integers with @xmath453 , and @xmath454 , where @xmath102 and @xmath455 are two positive real numbers such that @xmath456 .
then @xmath457 [ tail bound prelim2 ] let @xmath6 and @xmath139 be two integers with @xmath453 , and @xmath454 , where @xmath102 and @xmath455 are two positive real numbers such that @xmath458 .
then @xmath459 before we proceed to the proof of these propositions , let us see how we can apply them to deduce theorem [ tail bound ] .
set @xmath460 .
then @xmath461 where @xmath462 is the set of integers all of whose prime factors lie in @xmath463 $ ] .
we let @xmath464 } \left|
\sum_{\substack{y < n \le w \\ n\in { { \mathcal p}}(y , y ) } } \frac{\chi(n ) e(n\alpha)}{n } \right| \quad\text{and}\quad s^{(2)}_w(\chi ) = \max_{\alpha\in[0,1 ] } \left| \sum_{\substack{y < n \le w \\
p^-(n)>y } } \frac{\chi(n ) e(n\alpha)}{n } \right| , \ ] ] so that @xmath465 we shall bound the moments of each summand appearing above , individually .
we start with the summand involving @xmath466 . here
we may assume that @xmath467 ( and thus @xmath468 ) , else @xmath469 and so @xmath470 for all @xmath263 .
set @xmath471 and note that @xmath472 by lemma [ smooth ] .
so minkowski s inequality and proposition [ tail bound prelim2 ] imply that @xmath473 hence , applying hlder s inequality and mertens estimate @xmath474 , we arrive at the estimate @xmath475 next , in order to bound the summand in involving @xmath476 , we observe that @xmath477 for all @xmath478 by proposition [ tail bound prelim ] .
therefore hlder s inequality implies that @xmath479 finally , relations , and , together with an application of minkowski s inequality , imply that @xmath480 we note that , for all @xmath481 , we have that @xmath482 indeed , if @xmath483 , then @xmath484 , whereas if @xmath485 , then @xmath486 . combining and completes the proof of theorem [ tail bound ] .
our next task is to show proposition [ tail bound prelim ] .
first , we demonstrate the following auxiliary lemma .
[ d_k^2 ] let @xmath487 $ ] and @xmath488 be an integer .
uniformly for @xmath489 and for @xmath490 , we have that @xmath491 lemma 3.1 in @xcite , which is a generalization of lemma 4 in @xcite , implies that @xmath492 where @xmath493 is defined by .
note that if @xmath494 , then @xmath495 , which implies that @xmath496 so we arrive at the estimate @xmath497 which in turn yields that @xmath498 by our assumptions that @xmath490 and @xmath499 . without loss of generality , we may assume that @xmath6 is large , else the result is trivially true . set @xmath500 $ ] , so that @xmath501 hlder s inequality with @xmath502 and @xmath503 implies that @xmath504 which reduces the problem to bounding @xmath505 for @xmath506 $ ] . in order to do this ,
we first decouple the @xmath507 , the point where the maximum @xmath508 occurs , from the character @xmath8 .
we accomplish this by noticing that for every @xmath509 and for every @xmath510 $ ] , there is some @xmath511 such that @xmath512 . then @xmath513 we choose @xmath514
. then minkowski s inequality implies that @xmath515 which reduces proposition [ tail bound prelim ] to bounding @xmath516 notice that @xmath517 where @xmath518 clearly , @xmath519 so opening the square in , and summing over @xmath16 , we find that @xmath520 the right hand side of is at most @xmath521 , where @xmath522 we shall bound each of these sums in a different way .
firstly , note that @xmath523 by lemma [ d_k^2 ] with @xmath524 , which is admissible .
next , we bound @xmath525 .
note that @xmath526 for @xmath131 and @xmath527 in the support of @xmath525 .
in particular , for @xmath525 to have any summands , we need that @xmath528 .
we choose @xmath529 such that @xmath530 then @xmath531 our goal is to bound @xmath532 for every @xmath533 that is coprime to @xmath6 .
first , assume that @xmath534 .
note that @xmath535 is supported on integers @xmath536 which can be written as a product @xmath537 with each of the factors @xmath538 lying in the interval @xmath539 $ ] and having all their prime factors @xmath540 . in particular , @xmath541 for all @xmath542 and consequently @xmath543 by .
in particular , @xmath544 by our assumptions that @xmath545 and that @xmath546 .
this inequality also holds when @xmath547 , since in this case @xmath548 for all @xmath549 and @xmath550 for the numbers @xmath536 in the range of @xmath535 .
so , no matter what @xmath551 is , we conclude that @xmath552 since @xmath553 . inserting into , we arrive at the estimate @xmath554 combining relations and with
, we deduce that @xmath555 together with , the above estimate implies that @xmath556 since we have assumed that @xmath557 .
together with , this implies that @xmath558 since @xmath559 and @xmath560 , proposition [ tail bound prelim ] follows . without loss of generality
, we may assume that @xmath102 is large enough .
we start in a similar way as in the proof of proposition [ tail bound prelim ] : we set @xmath561 and note that @xmath562 therefore , minkowski s inequality implies that @xmath563 we claim that @xmath564 for all @xmath565 .
proposition [ tail bound prelim2 ] follows immediately if we show this relation , since we would then have that @xmath566 by our assumption that @xmath567 and , which establishes proposition [ tail bound prelim2 ] . arguing as in the proof of proposition [ tail bound prelim ] and setting @xmath568 we find that @xmath569 where @xmath570 we shall bound each of these sums in a different way .
we start with bounding @xmath571 .
set @xmath572 for large enough @xmath139 , so that @xmath573 . also , let @xmath574 .
fix @xmath575 with @xmath576 , and note that if @xmath577 with @xmath578 , then @xmath526 .
therefore @xmath579 \\
n_1\cdots n_k > q \\
p^-(n_1\cdots n_k)>y
\\ n_1\cdots n_k\equiv m{{\ifmmode\text{\rm\,(mod\,$q$)}\else\discretionary{}{}{\hbox { } } \rm(mod~$q$)\fi } } } } \frac{1}{n_1\cdots n_k } \\ & \le \sum_{\substack{1\le r_1,\dots , r_k\le \log(z / y)+1 \\ r_1+\cdots+r_k>\log(q / y^k ) } } \sum_{\substack { ye^{r_\ell -1}<n_\ell\le ye^{r_\ell } \\ p^-(n_\ell)>y \ , ( 1\le \ell\le k )
\\ n_1\cdots n_k\equiv m{{\ifmmode\text{\rm\,(mod\,$q$)}\else\discretionary{}{}{\hbox { } } \rm(mod~$q$)\fi } } } } \frac{1}{n_1\cdots n_k } \\ & \le \sum_{\substack{1\le r_1,\dots , r_k\le \log(z / y)+1 \\ r_1+\cdots+r_k>\log(q / y^k ) } } \frac{e^k}{y^ke^{r_1+\cdots+r_k } } \sum_{\substack { ye^{r_\ell-1}<n_\ell\le ye^{r_\ell } \\ p^-(n_\ell)>y \ ,
( 1\le \ell\le k ) \\ n_1\cdots n_k\equiv n{{\ifmmode\text{\rm\,(mod\,$q$)}\else\discretionary{}{}{\hbox { } } \rm(mod~$q$)\fi } } } } 1 .
\end{split}\ ] ] we fix @xmath580 as above , set @xmath581 for all @xmath582 , and choose @xmath529 such that @xmath583 then , for any @xmath584 that is coprime to @xmath6 , the cauchy
schwarz inequality implies that @xmath585 so , applying lemma [ d_k^2 ] , we deduce that @xmath586 since @xmath587 and @xmath588 , by . inserting this estimate into
, we deduce that @xmath589 consequently , we immediately deduce that @xmath590 which is admissible .
it remains to bound @xmath591 .
this will be done in a very different way .
we observe that if @xmath592 are the random variables defined in the introduction , then @xmath593.\ ] ] we have that @xmath594 where @xmath595 therefore @xmath596^{\frac{1}{2k}}+ o(1/y),\ ] ] by minkowski s inequality .
fix @xmath597 $ ] .
when @xmath598 , then @xmath599 , whereas if @xmath600 , then we use the trivial bound @xmath601 , for any @xmath602 . therefore @xmath603^{\frac{1}{2k } } + 2\cdot { { \mathbb e}}\left [ e^{2k|t| } ( |t|/\epsilon)^{\ell } \right]^{\frac{1}{2k } } + o(1/y ) .\ ] ] we have that @xmath604 = \sum_{\substack{y < p_1,\dots , p_{2k}\le z \\
p_1\cdots p_k = p_{k+1}\cdots p_{2k } } } \frac{1}{p_1\cdots p_{2k } } \le k !
\sum_{y < p_1,\dots , p_k\le z } \frac{1}{(p_1\cdots p_k)^2 } \le \frac{e^{o(k/\log y ) } k^k}{(ey\log y)^k } , \ ] ] since @xmath605 and @xmath606 by the prime number theorem .
moreover , @xmath607 & = \epsilon^{-\ell}\sum_{m=0}^{\infty } \frac{(2k)^m}{m ! } { { \mathbb e}}\left [ |t|^{m+\ell } \right ] \le \epsilon^{-\ell}\sum_{m=0}^{\infty } \frac{(2k)^m}{m ! } { { \mathbb e}}\left [ |t|^{2m+2\ell } \right]^{1/2 } \\ & \le \epsilon^{-\ell}\sum_{m=0}^{\infty } \frac{(2k)^m \sqrt{(m+\ell)!}}{m ! } \left(\frac{e^{o(1)}}{y\log
y}\right)^{m/2+\ell/2}\\ & \le \left(\frac{e^{o(1)}\ell}{\epsilon^2y\log y}\right)^{\ell/2 } \sum_{m=0}^{\infty } \frac 1{(m/2 ) ! } \left(\frac{e^{o(1)}k^2}{y\log y}\right)^{m/2 } \ll \left(\frac{e^{o(1)}\ell}{\epsilon^2 y\log y}\right)^{\ell/2 } e^{o(k ) } \end{aligned}\ ] ] as @xmath608 and @xmath609 , with @xmath610 .
choosing @xmath611 for an appropriate small constant @xmath74 makes the left hand side @xmath612 for some @xmath613 .
we take @xmath614 to conclude that @xmath615 \le e^{o(k)-4c'y(\log\log y)^2/\log y } \le e^{-2kc'(\log\log y)^2 } , \ ] ] where we used our assumption that @xmath139 is large and @xmath616 .
we thus conclude that @xmath617 by .
together with and , this completes the proof of relation and , thus , of proposition [ tail bound prelim2 ] .
in this section , we prove theorem [ distributiontheorem ] . throughout this section we fix @xmath121 and a large odd prime number @xmath6 such that @xmath618 . set @xmath619 and @xmath620 } { \left\lvert\sum_{\substack{n\in{{\mathbb z}},\ , n\neq 0 \\ p^+(n)\le y } } \frac{\chi(n)(1-e(n\alpha))}{n } \right\rvert } .\ ] ] relation and lemma [ smooth ] imply that @xmath621 therefore , if @xmath6 is large enough , then theorem [ tail bound cor ] and lemma [ smooth ] imply that @xmath622 where the last relation follows by theorem [ main thm ] and our assumption that @xmath618 .
therefore we conclude that @xmath623 where @xmath624 we perform the same analysis on @xmath625 . with a slight abuse of notation ,
we set @xmath626 } { \left\lvert \sum_{\substack{n\in{{\mathbb z}},\ , n\neq 0 \\ p^+(n)\le y } } \frac{x_n(1-e(n\alpha))}{n } \right\rvert } .\ ] ] the analogue of theorem [ tail bound cor ] for the random variables @xmath88 is ( more easily ) proven by the same method with a few simple changes : we proceed exactly as in the proof of theorem [ tail bound ] but wherever we evaluated a sum @xmath627 there ( as @xmath628 , unless @xmath629 when it equals @xmath9 ) , we now evaluate an expectation @xmath630 $ ] which equals @xmath628 unless @xmath631 , when it equals @xmath9 .
( in fact this only happens in the proofs of and of . )
we then conclude that @xmath632 where @xmath633 ( we could have written @xmath634 in place of @xmath635 in but this would have required proving a lower bound for @xmath625 . in order to avoid this technical issue
, we use for comparison @xmath37 whose size we already know by theorem [ main thm ] . )
next , for each @xmath205 we fix a parameter @xmath636 of the form @xmath637 with @xmath638 to be chosen , and a partition of the unit circle into arcs @xmath639 of length @xmath636 .
moreover , we let @xmath640 be the point in the middle of the arc @xmath641 and we define @xmath642 to be the set of @xmath643-tuples @xmath644 with @xmath645 and @xmath646 for all primes @xmath205 . given such a choice of @xmath647 and @xmath648 , we set @xmath649 and @xmath650 . moreover ,
similarly to before , we let @xmath651 we will show that there exist choices of @xmath636 and a constant @xmath652 such that if @xmath653 and @xmath111 belongs to the arc @xmath641 centred at @xmath654 for all @xmath205 , then @xmath655 .
this immediately implies that @xmath656 where @xmath657 it remains to show that @xmath658 if @xmath636 is chosen appropriately .
we choose these parameters so that @xmath659 . then the condition that @xmath660 for each prime @xmath205 implies that @xmath661 for all @xmath662 with @xmath663 .
hence @xmath664 which proves our claim .
we now prove similar inequalities for @xmath81 , namely that @xmath665 theorem 9.3 of @xcite states that if @xmath666 is an arc of length @xmath667 for each @xmath205 , then @xmath668 the same methods can easily be adapted to also show that @xmath669 for @xmath670 .
since @xmath619 , the inequality @xmath667 is indeed satisfied for our choice of @xmath636 , thus completing the proof of . finally , combining relations , , and , we obtain @xmath671 if @xmath7 is continuous in @xmath118 $ ] , then it is also uniformly continuous .
it then follows immediately from the above estimate that @xmath672 as @xmath673 over primes , uniformly on @xmath118 $ ] . in order to see that @xmath81 converges also weakly to @xmath7 ,
consider a continuous function @xmath674 of bounded support .
fix @xmath675 .
since @xmath7 has at most countable many discontinuity points , we deduce that there is an open set @xmath676 of lebesgue measure @xmath677 that contains all discontinuities of @xmath7 .
if @xmath678 is a bounded closed interval containing the support of @xmath679 , then the set @xmath680 is compact and thus it can be written as a finite union of closed intervals . since @xmath681 converges uniformly to @xmath7 on each such interval , it does so on @xmath682 as well
. therefore @xmath683 since @xmath684 was arbitrary , we conclude that @xmath685 thus completing the proof of theorem [ distributiontheorem ] .
in this section , we develop some general tools we will use to prove theorems [ structure thm ] and [ structure thm even ] .
we begin by stating a result that allows us to concentrate on the case when @xmath244 is a rational number with a relatively small denominator . [ centering alpha ]
let @xmath195 , @xmath686 , @xmath8 be a dirichlet character , @xmath354 and @xmath687 $ ] .
let @xmath151 be a reduced fraction with @xmath688 and @xmath689 .
then @xmath690 where @xmath691 .
this follows immediately by the second part of lemma 4.1 in @xcite ( see also lemma 6.2 in @xcite ) .
when @xmath155 is large , we have the following result . [ mv ]
let @xmath692 , where @xmath693 . for all @xmath694
, we have that @xmath695 this is corollary 2.2 in @xcite , which is based on a result due to montgomery and vaughan @xcite . for smaller @xmath155
, we shall use the following formula .
[ formula ] let @xmath8 be a dirichlet character and @xmath693 . for @xmath696
we have that @xmath697 if @xmath698 , then @xmath699 so , writing @xmath700 for the greatest common divisor of @xmath131 and @xmath155 , we find that @xmath701 finally , observe that the innermost sum vanishes if @xmath702 is an even character , whereas if @xmath702 is odd it equals @xmath703 this concludes the proof of the lemma .
following granville and soundararajan @xcite , we are going to show that the all the terms in the right hand side in lemma [ formula ] are small unless @xmath159 is induced by some fixed character @xmath704 which depends at most on @xmath8 and @xmath102 . as in @xcite , in order to accomplish this , we define a certain kind of `` distance '' between two multiplicative functions @xmath679 and @xmath705 of modulus @xmath706 : @xmath707 then we let @xmath708 be a primitive character of conductor @xmath709 such that @xmath710 we need a preliminary result on certain sums of multiplicative functions , which is the second part of lemma 4.3 in @xcite .
[ general ub ] let @xmath711 be a multiplicative function . for @xmath696
we have that @xmath712 then we have the following `` repulsion '' result . [ repulsion ]
let @xmath8 , @xmath102 and @xmath704 be as above .
if @xmath159 is a dirichlet character modulo @xmath713 , of conductor @xmath714 , that is not induced by @xmath704 , then @xmath715 let @xmath716 be the primitive character inducing @xmath159 .
since @xmath717 lemma 3.4 in @xcite and the definition of @xmath704 imply that @xmath718 the claimed estimate then follows by lemma [ general ub ] above .
when applying lemma [ formula ] , we will need to evaluate the gauss sum that arises . in order to do this , we shall use the following classical result ( see , for example , theorem 9.10 in @xcite ) .
[ gauss sum ] let @xmath159 be a character modulo @xmath719 induced by the primitive character @xmath720 modulo @xmath721 . then @xmath722 we also need the following simple estimate , which we state below for easy reference .
[ change gcd ] let @xmath723 be a completely multiplicative function . for all @xmath584 , we have that @xmath724 and @xmath725 we write @xmath726 if @xmath727 for all primes @xmath728
. then @xmath729 the second part is proved similarly , starting from the identity @xmath730 combining the above results , we prove the following simplified version of lemma [ formula ] .
[ formula2 ] let @xmath8 , @xmath102 , @xmath704 and @xmath731 be as above , and consider a real number @xmath732 and a reduced fraction @xmath151 with @xmath733 and @xmath734 .
if either @xmath735 or @xmath736 is even , then @xmath737 whereas if @xmath738 and @xmath736 is odd , then @xmath739 by lemma [ repulsion ] , we see that if @xmath159 is a character modulo @xmath719 that is not induced by @xmath704 , then @xmath740 so the first result follows by lemma [ formula ] .
finally , if @xmath738 and @xmath741 is odd , then lemmas [ formula ] and [ gauss sum ] , and imply that @xmath742 writing @xmath743 , noticing that @xmath744 if @xmath745 , and using the trivial bound + @xmath746 to extend the sum over @xmath747 to a sum over @xmath748 , we find that @xmath749 setting @xmath750 and applying lemma [ change gcd ] with @xmath751 and @xmath527 in place of @xmath752 we deduce that @xmath753 since we have assumed that @xmath734 , we have that @xmath754 if @xmath755 with @xmath756 , then @xmath757 therefore the absolute value of the sum in is @xmath758 since we also have that @xmath759 , we deduce that @xmath760 where we used lemma [ general ub ] and the inequality @xmath761 for @xmath762 .
this completes the proof of the lemma .
finally , if @xmath8 pretends to be 1 in a strong way , then we can get a very precise estimate on the sum of @xmath763 over smooth number using estimates for such numbers in arithmetic progressions due to fouvry and tenenbaum @xcite .
[ remove exp ] let @xmath764 , @xmath8 be a character mod @xmath6 , and @xmath151 be a reduced fraction with @xmath765 .
then @xmath766 where @xmath767 is the largest divisor of @xmath155 with @xmath768 and @xmath769 , with @xmath770 first , note that the inequality @xmath771 follows by the cauchy
schwarz inequality and the fact that @xmath772 for @xmath773 .
we write @xmath774 .
then we have that @xmath775 and thus @xmath776 moreover , observe that , using lemma [ smooth ] , we may assume that @xmath777 .
we begin by estimating the sum @xmath778 for @xmath779 $ ] . set @xmath780 and
note that @xmath781 where we bounded trivially the sum over @xmath782 when @xmath783 ( note that @xmath784 for all @xmath785 ) .
next , we need an estimate for the sum @xmath786 when @xmath787 .
note that , since @xmath788 , theorems 2 and 5 in @xcite imply that @xmath789 when @xmath790 ; the same result also holds when @xmath791 by elementary techniques since @xmath792 for @xmath793 and @xmath787 .
so , using the identity @xmath794 we deduce that @xmath795 where we used .
we get the same right side no matter what the value of @xmath752 , as long as @xmath693 .
hence @xmath796 for some function @xmath797 satisfying the bound @xmath798 letting @xmath799 , and writing @xmath800 and @xmath801 so that @xmath802 , we deduce that @xmath803 by ( * ? ? ?
* eq ( 7 ) , p. 149 ) , for all @xmath804 $ ] .
therefore partial summation implies that @xmath805 integrating by parts and applying relation and our assumption that @xmath777 we conclude that @xmath806 since we also have that @xmath807\cup(z / d , z ] \\ ( m , c)=1,\ , p^+(m)\le y } } \frac{\chi(m)}{m } \ll\sum_{b = cd } \frac { \log\log y } { \phi(c)d } \ll\log\log y,\ ] ] we deduce that @xmath808 applying lemma [ change gcd ] with @xmath809 and @xmath155 in place of @xmath752 implies that @xmath810 inserting this formula into leads to an error term of size @xmath811 and a main term of @xmath812 times @xmath813 thus completing the proof of the lemma .
[ remove exp cor ] let @xmath6 be an integer that either equals 1 or is prime .
let @xmath814 , and @xmath151 be a reduced fraction with @xmath815
. then @xmath816
the goal of this section is to prove theorem [ structure thm even ] . throughout this section , we set @xmath817 for some constant @xmath417 .
we will show this theorem with @xmath818 where the quantity @xmath384 is defined as in section [ outline ] .
theorem [ tail bound cor ] and the lower bound in theorem [ main thm even ] , which we already proved in the beginning of section [ outline ] ( independently of the proof of theorem [ structure thm even ] ) , guarantee that the cardinality of @xmath819 satisfies , provided that the constant @xmath417 in the definition of @xmath102 and the constant @xmath70 in the statement of theorem [ structure thm even ] are large enough .
we fix a large @xmath820 and we consider a character @xmath186 .
let @xmath821 be such that @xmath822 then @xmath823 by , our assumption that @xmath824 for @xmath825 , and lemma [ smooth ] .
as in the statement of theorem [ structure thm even ] , we approximate @xmath244 by a reduced fraction @xmath151 with @xmath826 and @xmath152 .
we let @xmath827 and apply lemma [ centering alpha ] with @xmath828 to find that @xmath829 since @xmath830 , we must have that @xmath831 , which implies that @xmath832 we now proceed to show that @xmath8 satisfies properties ( 1 ) and ( 2 ) of theorem [ structure thm even ] .
we choose @xmath833 with @xmath834 to satisfy .
we claim that @xmath835 the first claim being equivalent to @xmath836 .
firstly , note that lemma [ mv ] in conjunction with implies that @xmath837 .
equation also tells us that we must be in the second case of lemma [ formula2 ] ( provided that @xmath30 is large enough ) , that is to say @xmath738 and @xmath736 is odd . since @xmath8 is even , we conclude that @xmath704 is odd .
thus @xmath838 .
moreover , the last inequality in lemma [ formula2 ] implies that @xmath839 comparing this inequality with , we deduce that @xmath840 and thus @xmath841 , the quadratic character modulo 3 , which completes the proof of our claim and hence of the fact that property ( 1 ) holds .
we start with the proof of .
note that @xmath842 where @xmath843 . then using , we find that @xmath844 so that , by , @xmath845 hence , we conclude that @xmath846 therefore @xmath847 which , in turn , implies that @xmath848 so that @xmath849 finally , we have that @xmath850 by the plya
vinogradov inequality , our assumption that @xmath830 and lemma [ smooth ] . inserting the above estimates into completes the proof of .
finally , we prove . for convenience , we set @xmath851 .
then and our assumption that @xmath852 for @xmath830 imply that @xmath853 since this lower bound is also an upper bound , we deduce that @xmath854 then , the argument leading to implies that @xmath855 thus completing the proof of slightly more than property ( 2 ) .
define @xmath263 via the relation @xmath856 .
note that @xmath857 as @xmath817 .
so we may show the theorem with @xmath263 in place of @xmath226 . arguing as at the beginning of section [ even ] , and applying lemma [ centering alpha ] with @xmath828
, we find that @xmath858 we note that inequality and the argument leading to imply that @xmath859 where we have set @xmath851 .
we write @xmath860 with @xmath861 , so that @xmath862 by and as @xmath863 . using the formula @xmath864
, we deduce that @xmath865 if @xmath782 is not a power of 3 , then @xmath866 for each @xmath551 , so corollary [ remove exp cor ] implies that @xmath867 as claimed . finally , if @xmath868 and @xmath869 , then @xmath870 , unless @xmath871 and @xmath872 , so that @xmath873
. therefore corollary [ remove exp cor ] and lemma [ smooth2 ] imply that @xmath874 this completes the proof of theorem [ structure thm even ] .
the goal of this section is to prove theorems [ structure thm ] and [ partial sums odd ] . throughout this section
, we set @xmath400 for some constant @xmath74 ( note that this a different value of @xmath102 than in the previous section ) .
we will show this theorem with @xmath875 where the quantity @xmath384 is defined as in section [ outline ] .
theorem [ tail bound cor ] and the lower bound in theorem [ main thm ] , which we already proved in the beginning of section [ outline ] , guarantee that the cardinality of @xmath876 satisfies , provided that the constant @xmath417 in the definition of @xmath102 and the constant @xmath70 in the statement of theorem [ structure thm ] are large enough .
we fix a large @xmath820 and we consider a character @xmath149 .
let @xmath821 be such that @xmath822 as in the statement of theorem [ structure thm ] , we pick a reduced fraction @xmath151 such that @xmath877 and @xmath878 , and we define @xmath879 to equal @xmath155 if @xmath155 is prime , and 1 otherwise . following the argument leading to , we deduce that @xmath880 where @xmath881 .
we choose @xmath833 with @xmath834 to satisfy .
since @xmath882 , we must have that @xmath831 , which implies that @xmath883 we claim that @xmath884 the first relation being property ( 1 ) .
we separate two cases .
first , assume that @xmath885 .
then we apply lemma [ mv ] to find that @xmath886 which , together with , implies that @xmath887 then we must have that @xmath888 .
furthermore , the first part of lemma [ formula2 ] implies that @xmath889 and thus @xmath890 .
finally , lemma [ general ub ] and imply that @xmath891 , which completes the proof of in this case . finally , assume that @xmath892 .
suppose that either @xmath8 is even or @xmath893 .
then lemma [ formula2 ] implies that @xmath894 so becomes @xmath895 thus we must be in the second case of lemma [ formula2 ] as far as the above sum is concerned , that is to say that @xmath738 and @xmath736 is odd .
then the second part of lemma [ formula2 ] implies that @xmath896 comparing this inequality with , we deduce that @xmath897 and thus @xmath890 , provided that @xmath30 is large enough .
but then @xmath8 has to be odd too , which contradicts our initial assumption .
so we conclude that our initial assumption must be wrong , that is to say @xmath8 must be odd and @xmath890 , so that @xmath889 .
it remains to show that @xmath891 in the case when @xmath892 .
we apply lemma [ general ub ] and the second part of lemma [ formula2 ] to deduce that @xmath898 combining the above inequality with yields the estimate @xmath891 , thus completing the proof of our claim and , consequently , of property ( 1 ) . in order to prove property ( 2 ) in theorem [ structure thm ] , we need an intermediate result : we set @xmath899 for @xmath900 , and @xmath901 as in lemma [ remove exp ] , where we used . additionally , we set @xmath902 and we write @xmath903 , where @xmath904 the intermediate result we need to show is that @xmath905 before proving this , we show how to use it to complete the proof of theorem [ structure thm ] .
we argue as in the proof of property ( 2 ) in theorem [ structure thm even ] . for , note that our assumption that @xmath882 implies , as @xmath424 , that , for all @xmath906 , @xmath907 therefore , @xmath908 for all @xmath906 , by the plya
vinogradov inequality , and lemma [ smooth ] .
so follows from but with the weaker error term @xmath909 in place of @xmath910 , where @xmath911 if @xmath912 , and @xmath913 otherwise , so that @xmath914 .
we argue much like we did getting to to : we have @xmath915 so that @xmath916 and therefore @xmath917 this implies that @xmath918 in particular , @xmath919 , so that we always have that @xmath920 and @xmath921 .
this completes the proof of property ( 2 ) in theorem [ structure thm ] .
we separate two main cases . * case 1 . *
assume that @xmath922 .
then and the fact that @xmath8 is odd imply that @xmath923 since @xmath852 , we find that @xmath924 consequently , @xmath925 , which in turn gives us that @xmath926 .
inserting this estimate into , we deduce that @xmath927 that is to say , holds ( with a stronger error term ) . * case 2 . *
assume that @xmath928 .
then and lemma [ remove exp ] implies that @xmath929 now lemma [ change gcd ] , applied with @xmath930 and @xmath155 in place of @xmath752 , implies that @xmath931 so , if we set @xmath932 then we have that @xmath933 our goal is to show , using formula , that @xmath934 for all @xmath793 .
indeed , if this were true , then @xmath935 . we start by observing that @xmath936 indeed , reversing the last steps of the proof of lemma [ remove exp ] , we find that @xmath937 moreover , @xmath938 combining the two above relations , we obtain . using and the inequality @xmath939 , we deduce that @xmath940 for @xmath941 . inserting this into , and
since @xmath942 we obtain that @xmath943 we also have @xmath944 by lemma [ change gcd ] .
together with , this implies that @xmath945 since this holds as an upper bound too , we deduce that @xmath946 substituting into we obtain @xmath947 and then comparing with the displayed line above , @xmath948 where we have set @xmath949 for @xmath950 . *
case 2a . *
assume that @xmath155 has at least two distinct prime factors .
if @xmath951 , then we can find two distinct primes @xmath952 and @xmath6 such that @xmath953 , @xmath954 and @xmath955 ( which only fails for @xmath956 or @xmath957 ) .
therefore , taking absolute values in , we find that @xmath958 note that @xmath959 therefore @xmath960 now @xmath961 when @xmath962 so that , for @xmath963 and @xmath964 , @xmath965 since @xmath966 and @xmath967 .
we therefore deduce that @xmath968 now @xmath969 , and so @xmath970 and therefore , by , @xmath971 substituting this into , and using , yields in this case , except if @xmath972 . when @xmath972 , we use relations and to deduce that @xmath973 since the left hand side is @xmath974 and @xmath975 , we deduce that @xmath976 , as before . proceeding now as in the case @xmath951 completes the proof when @xmath972 too .
* case 2b .
* now suppose that @xmath912 is a prime power with @xmath977 . using , we see that @xmath978 define @xmath979 so that @xmath980 and @xmath981
. then @xmath982 which is @xmath983 and @xmath984 by .
taking real parts , and noting that each re@xmath985 , we deduce that re@xmath986 re@xmath987 taking the @xmath988 and @xmath9 terms .
hence @xmath989 and therefore @xmath990 substituting this into , and using , yields the result in this case .
* case 2c .
* now suppose that @xmath991 is a prime , so that @xmath992 .
hence we can not use to gain information on @xmath993 .
now , using lemma [ change gcd ] , we have @xmath994 combining this with yields that @xmath995 ; that is @xmath996 taking real parts , we deduce that @xmath997 with the lower bound being trivial .
this implies that @xmath998 using lemma [ change gcd ] , we then conclude that @xmath999 which , together with , completes the proof of in this last case too .
note that @xmath1000 can be quite big if @xmath68 is small and if @xmath244 is not very close to @xmath151 .
so we can not say more than the last formula without more information on the location of @xmath244 .
we conclude the paper with the proof of theorem [ partial sums odd ] .
let @xmath400 for a large enough constant @xmath417 , as above , and define @xmath263 via the relation @xmath856 .
note that @xmath857 .
so we may show the theorem with @xmath263 in place of @xmath226 .
arguing as at the beginning of this section , and applying lemma [ centering alpha ] with @xmath828 , we find that @xmath1001 note that and the argument leading to imply that @xmath1002 for all @xmath675 .
hence @xmath8 is @xmath9-pretentious and so , substituting this in above , we obtain @xmath1003 now suppose that @xmath156 . if @xmath1004 then we bound the second sum using corollary [ remove exp cor ] . if @xmath1005 , then the result follows from lemma [ smooth2 ] .
this concludes the proof of part ( a ) .
finally , assume that @xmath155 is a prime number so that @xmath177 . writing @xmath1006 with @xmath1007 , we find that @xmath1008 by and the trivial estimate @xmath1009 .
if @xmath1005 , then we have that @xmath1010 by lemmas [ change gcd ] and [ smooth2 ] .
next , if @xmath1011 for all @xmath1012 , then @xmath1013 for all @xmath1014 .
so corollary [ remove exp cor ] implies that @xmath1015 as claimed .
finally , assume that @xmath1016 for some @xmath1017 .
the terms with @xmath1018 contribute @xmath1019 by lemmas [ change gcd ] and [ smooth2 ] .
when @xmath1020 , we apply corollary [ remove exp cor ] .
the total contribution of those terms is @xmath1021 by lemmas [ change gcd ] and [ smooth2 ] . putting the above estimates together
, we complete the proof the theorem .
.the minimum , maximum , mean , and @xmath1022-quantile for @xmath22 over even , odd , and all nontrivial @xmath8 mod @xmath6 for some selected values of @xmath6 . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] , _ extreme values of @xmath1025 . _
the riemann zeta function and related themes : papers in honour of professor k. ramachandra , 6580 , ramanujan math .
soc . lect .
notes ser . , 2 , ramanujan math .
soc . , mysore , 2006 .
g. tenenbaum , _ introduction to analytic and probabilistic number theory .
_ translated from the second french edition ( 1995 ) by c. b. thomas .
cambridge studies in advanced mathematics , 46 . cambridge university press ,
cambridge , 1995 . | let @xmath0 denote the maximum of @xmath1 for a given non - principal dirichlet character @xmath2 , and let @xmath3 denote a point at which the maximum is attained .
in this article we study the distribution of @xmath4 as one varies over characters @xmath5 , where @xmath6 is prime , and investigate the location of @xmath3 .
we show that the distribution of @xmath4 converges weakly to a universal distribution @xmath7 , uniformly throughout most of the possible range , and get ( doubly exponential decay ) estimates for @xmath7 s tail .
almost all @xmath8 for which @xmath0 is large are odd characters that are @xmath9-pretentious .
now , @xmath10 , and one knows how often the latter expression is large , which has been how earlier lower bounds on @xmath7 were mostly proved .
we show , though , that for most @xmath8 with @xmath0 large , @xmath3 is bounded away from @xmath11 , and the value of @xmath0 is little bit larger than @xmath12 . |
because the light from a distant galaxy comes from discrete but unresolved stars , poisson statistics lead to mottling of the galaxy s otherwise smooth surface brightness profile .
surface brightness is independent of distance , but the amplitude of the surface brightness fluctuations ( sbfs ) is not . as the distance ( @xmath4 ) to a galaxy increases , the number of stars ( @xmath5 ) in a given resolution element increases as @xmath6 , but the observed flux ( @xmath7 ) from each star is reduced by @xmath8 , making surface brightness ( @xmath9 ) independent of distance . on the other hand ,
the rms variation in observed flux from region to region is @xmath10 , which scales as @xmath11 .
the variance in surface brightness @xmath12 normalized by the mean galaxy brightness @xmath9 decreases with distance ; distant galaxies appear smoother than nearby galaxies . because the variance in surface brightness is proportional to the second moment of the stellar luminosity function , it is dominated by luminous red giant stars . although individual stars are not resolved , measuring sbfs probes the stellar population of the galaxy directly . with good theoretical models for stellar populations , the absolute magnitude of sbfs can be calculated , allowing a direct determination of the distance that is independent of the global dynamics or the environment of the galaxy ( to the extent that the stellar populations in old stellar systems are independent of these variables ) .
alternatively , measuring sbfs in galaxies with known distances provides insight into stellar populations , allowing comparison with stellar evolution models and providing an empirical calibration of the sbf distance scale .
good descriptions of the theory and practice of using sbfs as a distance measurement tool and stellar population probe can be found in several papers by tonry and collaborators ( tonry et al .
1997 ; jensen , luppino , & tonry 1996 , hereafter jlt ; tonry 1991 ; tonry , ajhar , & luppino 1990 ; tonry & schneider 1988 ) , in papers by sodemann & thomsen ( 1995 , 1996 ) , and in the review by jacoby et al .
the first @xmath13-band sbf studies are described by jlt , luppino & tonry ( 1993 ) , and pahre & mould ( 1994 ) .
j. tonry and coworkers have completed an extensive survey of @xmath14-band sbf distances in a sample of several hundred early - type ( dynamically hot elliptical and s0 ) galaxies which is more than 50% complete out to @xmath152800 ( tonry et al .
they find that the @xmath14-band absolute fluctuation magnitude is a linear function of v - i and has a universal zero point .
their @xmath14-band calibration , which is based empirically on cepheid distances , is in good agreement with worthey s ( 1993a , b ; 1994 ) simple stellar populations models .
worthey s models predict that the effects of age and metallicity in the @xmath14 band are largely degenerate , so that may be calibrated using a single parameter such as the v - i color .
the resulting intrinsic scatter of the @xmath14-band sbf distance scale is of order 0.07 mag .
the purpose of the current study is to examine the behavior of the sbf calibration in the near - ir band , where worthey s models predict a much weaker dependence on v - i , but potentially larger scatter as the effects of age and metallicity are no longer degenerate .
stellar surface brightness fluctuations are very red , since they are dominated by luminous red giant stars .
the advantages of observing ir sbfs are clear : fluctuations are @xmath16 times brighter at @xmath13 than at @xmath14 , making them observable to greater distances .
the sbf amplitude is inversely proportional to the seeing fwhm , and the seeing is typically much better in the ir than at optical wavelengths .
fluctuations are also red compared to the globular cluster ( gc ) population , so the contrast between stellar sbfs and gcs is higher at @xmath13 . finally , stellar population models predict that @xmath13-band sbf magnitudes have a much weaker dependence on v - i than at @xmath14 , reducing or eliminating the need to accurately measure the color of the galaxy ( worthey 1993a ) .
several studies have demonstrated the feasibility of measuring ir sbfs , including luppino & tonry ( 1993 ) , pahre & mould ( 1994 ) , and jlt .
these papers report results for a rather limited set of galaxies in the local group ( lg ) and the virgo cluster , but the consistency in the measured calibration of is encouraging ( @xmath17 luppino & tonry , @xmath18 pahre & mould , @xmath19 jlt ) .
all three of these papers report absolute fluctuation magnitudes that are consistent with worthey s ( 1994 ) predictions based on simple stellar population models . while the mean of jlt and pahre & mould ( 1994 ) are quite consistent , differences in apparent fluctuation magnitudes for individual galaxies are larger than the stated uncertainties allow .
jlt discuss several possible reasons for the disagreement in fluctuation magnitudes and showed that residual variance resulting from spatial variations in the detector sensitivity and dark current can contribute significantly to the power spectrum in low signal - to - noise ratio ( @xmath20 ) observations .
a residual pattern of only 0.1% of the sky level can significantly change the fluctuation magnitude measured . to use ir sbfs as a reliable distance indicator ,
it is critical that the observations be of sufficiently high @xmath20 ratio to avoid biases from residual variances .
jlt also demonstrated that properly subtracting the sky and carefully sampling the point spread function ( psf ) are crucial to accurately measuring the fluctuation amplitude .
an uncertainty in the background subtraction of 2% of the sky level can dominate the uncertainty in the fluctuation magnitude .
our virgo sample showed a large dispersion ( 0.29 mag ) due to the depth of the cluster and observational errors .
pahre & mould ( 1994 ) also addressed the dispersion in fluctuation magnitude in a similarly - sized sample of virgo ellipticals .
clearly a larger sample of galaxies is needed to calibrate the ir sbf distance scale .
to better understand the effects stellar populations have on the fluctuation amplitude , we observed five galaxies in the fornax cluster .
fornax is ideal for several reasons : first , it is much more centrally concentrated than the virgo cluster .
the reduced dispersion in distances allows us to better quantify and understand the uncertainty in our sbf measurements .
second , fornax contains a large number of giant elliptical galaxies with a wide range of v - i colors , metallicities , and globular cluster populations .
the virgo galaxies observed by jlt and pahre & mould ( 1994 ) spanned a very limited range in 2 index to avoid stellar population variations which may affect the calibration of .
we now extend the sample to calibrate fluctuation magnitudes across a wider range of metallicities and v - i colors to compare with theoretical stellar population models .
finally , cepheid distances to several virgo cluster galaxies and to the fornax spiral ngc 1365 have been measured using the hubble space telescope ( hst ) .
we can empirically anchor our -band sbf calibration both to the cepheid and the @xmath14-band sbf distance scales
. this paper reports results from our observations of four eridanus cluster elliptical galaxies in addition to the fornax galaxies .
eridanus is not as compact a cluster as fornax , but we can still compare our ir sbf results with the @xmath14-band results from tonry ( 1997 ) . finally , we return to our sbf measurements for the seven virgo ellipticals discussed by jlt and apply the improved techniques we describe in this paper .
we present the updated results for these galaxies in section [ virgorevisited ] , along with new observations of ngc 4365 .
this paper presents results derived from data obtained during several observing runs using the quirc camera mounted at the f/10 focus of the university of hawaii 2.24 m telescope .
quirc is a near - ir camera employing the first- and second - generation @xmath21-pixel hgcdte `` hawaii '' detectors ( hodapp et al .
these observations are compared to previous work which used the 256@xmath0-pixel nicmos detector ( jlt , luppino & tonry 1993 ) .
the second - generation science - grade hawaii array has far fewer bad pixels , higher quantum efficiency , and a much more spatially uniform response than the earlier hgcdte devices .
the dark current for these detectors is less than 1 e@xmath22s@xmath23 and the readnoise is @xmath1515 e@xmath22 .
the quirc camera has a plate scale of 01886 pixel@xmath23 , providing a field of view 32 on a side .
the university of hawaii filter used for these observations is centered at a slightly shorter wavelength ( 2.11 ) than the standard @xmath13 filter to reduce the thermal background ( wainscoat & cowie 1992 ) .
the mean sky background measured using the filter during this run was 13.4 mag arcsec@xmath24 .
these observations are directly comparable to those of jlt and luppino & tonry ( 1993 ) , which used the same filter .
cccccc 1993 apr 711 & nicmos & 0.375 & 1.6 & 30 & virgo 1995 may 1720 & quirc-1 & 0.19 & 3.2 & 6 & ngc 4365 , 4489 1995
oct 1417 & quirc-1 & 0.189 & 3.2 & 6 & fornax , eridanus 1996 mar 1 , apr 2728 & quirc-2 & 0.189 & 3.2 & 1.85 & ngc 4365 we observed five giant elliptical galaxies in the fornax cluster and four in the eridanus cluster over a period of four nights in 1995 october using the first - generation 1024@xmath0-pixel array . in this paper
we reanalyze the virgo cluster data originally presented by jlt .
most of the virgo observations were made using the smaller @xmath25-pixel nicmos3 array ( hodapp , rayner , & irwin 1992 ) .
two of the jlt galaxies were observed using the first - generation 1024@xmath0-pixel array in the quirc camera .
ngc 4365 was observed again in 1996 march and april using the second - generation @xmath21-pixel hgcdte array with improved electronics , and we compare these newer data to those presented by jlt .
details of the individual observing runs are presented in table [ fornaxobsdata ] .
cccccccccc fornax&&&&&&&&&1339 & @xmath26 & 1.676 & 0.13 & 0.68 & 13.57 & @xmath27 & 60 & 2160 & 0.00 1344 & @xmath28 & 1.635 & 0.10 & 0.74 & 13.15 & @xmath29 & 60 & 2220 & 0.00 1379 & @xmath30 & 1.906 & 0.13 & 0.83 & 13.47 & @xmath31 & 90 & 2070 & 0.00 1399 & @xmath32 & 1.888 & 0.13 & 0.83 & 13.21 & @xmath33 & 90 & 1710 & 0.00 1404 & @xmath28 & 1.901 & 0.10 & 0.77 & 13.37 & @xmath34 & 60 & 1980 & 0.00 eridanus&&&&&&&&&1395 & @xmath32 & 1.621 & 0.13 & 0.75 & 13.05 & @xmath35 & 90 & 1350 & 0.02 1400 & @xmath28 & 1.591 & 0.10 & 0.70 & 13.34 & @xmath36 & 60 & 1980 & 0.13 1407 & @xmath30 & 1.550 & 0.13 & 0.91 & 13.56 & @xmath37 & 120 & 2280 & 0.16 1426 & @xmath30 & 1.865 & 0.13 & 0.89 & 13.61 & @xmath38 & 90 & 1710 & 0.02 virgo w&&&&&&&&&4365 & @xmath39 & 1.000 & 0.00 & 0.59 & 13.30 & @xmath40 & 89 & 2670 & 0.00 the techniques we used to do the initial data reduction are described by jlt .
we adopted several key improvements for the new data described in this paper .
flat field images and bad pixel masks were created as described by jlt . to improve the noise characteristics of the background for our sbf analysis
, we constructed average ( rather than median ) sky frames from @xmath41 individual sky images which were interspersed with the galaxy observations .
stars were removed from individual images and bad pixels masked , and the remaining good pixels averaged to form the sky frame .
the average sky frame was subtracted from individual galaxy images prior to flat fielding and removal of bad pixels and cosmic rays .
the cleaned and flattened galaxy images were then registered to the nearest integer pixel and the good pixels averaged as described by jlt .
sub - pixel registration was not used because such procedures produce noise that is correlated from pixel to pixel and is therefore not constant with wavenumber in the power spectrum . to calibrate the photometry
, we frequently observed the faint ukirt standards ( casali & hawarden 1992 ) .
the majority of our 1995 october observations were made when conditions were not photometric .
the photometric zero point fluctuated by as much as 0.25 mag during periods of all but one of the nights . to better calibrate our photometry
, we compared our results to photoelectric aperture photometry ( persson , frogel , & aaronson 1979 , frogel et al .
1978 , persson et al .
we measured the difference between our magnitudes and the photoelectric magnitudes for all galaxies in common ( 7 of 9 ) and used the mean difference and standard deviation to adjust our photometric zero - point and uncertainty for each night .
atmospheric extinction coefficients were determined by combining standard star observations covering a range of air masses during photometric periods of different nights .
no color term to the photometry was measured .
photometric zero - points and extinction coefficients are listed in table [ fornaxphotdata ] .
next , we corrected for galactic extinction . from cohen
et al . ( 1981 ) we learn that @xmath42 following tonry et al .
( 1997 ) , we take @xmath43 for early type galaxies .
we assume that the extinction at @xmath13 and are the same , so combining the above relations gives @xmath44 .
values for @xmath45 are listed in table [ fornaxphotdata ] ( burstein & heiles 1984 ) .
no k corrections were needed for the low - redshift galaxies in this study . in section [ wortheymodels ]
we describe stellar population models used to predict fluctuation magnitudes ( worthey 1994 , 1997 ) .
the worthey models were also used to determine the redshift dependence of the @xmath13-band fluctuation magnitudes . at a redshift of @xmath46 appropriate for the virgo , fornax , and eridanus clusters ,
the k correction to the @xmath13-band fluctuation magnitude is @xmath47 mag for a 17 gyr stellar population and + 0.004 for an 5 gyr population ( [ fe / h]=0 ) . while these corrections are for the @xmath13 filter , the offsets are so small that it is safe to assume the corrections to the sbf magnitudes are negligible . even at redshifts of @xmath48 ( 7500 ) ,
worthey s models predict corrections no larger than @xmath49 mag between ages of 5 and 17 gyr .
the fact that @xmath13 sbfs change so little as a function of redshift is very encouraging .
the purpose of calibrating sbfs is to allow sbf measurements to be made to much larger distances than possible at @xmath14 band , and the absence of significant k corrections reduces the uncertainty in doing so .
we followed the same procedures described by jlt to measure sbf amplitudes in this data , with important modifications to account for the contributions from undetected globular clusters and background galaxies .
we used the vista image analysis software ( written by t. lauer and r. stover ) and additional programs written by j. tonry to perform the sbf analysis . the first step in
the process was to determine the residual sky background remaining after stacking individual images .
we assumed that this small adjustment ( typically @xmath50 of the sky level ) was constant over the entire frame and could be applied as an offset to the final flattened and registered image .
as jlt found , however , the fluctuation power is quite sensitive to residual patterns in sky subtraction , flat fielding , and non - zero background offsets , even at the level of @xmath51 of the sky level .
we estimated the residual sky background levels in the final images by fitting a devaucouleurs @xmath52 profile to each galaxy using elliptical apertures .
the larger quirc field of view made it possible to determine the residual sky level much more accurately than using nicmos .
the sky levels in the jlt data were also remeasured for consistency .
the next step in measuring the sbf magnitude was to fit and subtract the galaxy profile .
we used a routine which allows the center , ellipticity , and position angle to vary as we fit annuli of increasing radius , and we iterated the procedure to ensure that the fit was well - behaved .
point sources were masked before fitting the galaxy profile .
the resulting model of the galaxy was subtracted from the data .
large - scale residual background variations were also fitted and subtracted , leaving a smooth , flat background with stellar sbfs , globular clusters , background galaxies , foreground stars , and noise superimposed .
the noise varies with position in the image because of the subtraction of the galaxy , although in our ir images sky - subtraction noise dominates everywhere in the image except very near the center of the galaxy .
globular clusters and background galaxies were masked as described in the following section .
the residual image was then masked with a window function which selected the region of the galaxy to be analyzed .
the fourier transform and power spectrum were then computed for the data .
the variance arising from stellar sbfs and unresolved point sources was determined from the best fit of the data power spectrum with the function @xmath53 a linear combination of the fluctuation power @xmath54 in times the expectation power spectrum @xmath55 and a white noise component @xmath56 arising from photon shot noise .
the white noise component depends on the background sky level and the mean galaxy brightness : @xmath57 in the absence of sky background and blurring by the atmosphere , @xmath58 and @xmath54 is simply the power at @xmath59 . in real observations ,
the fluctuations are convolved with the psf .
we must also account for the contribution to the power spectrum caused by the sharp edges of our window function and point source mask .
the expectation power spectrum @xmath55 is the power spectrum of the psf modified slightly to include the power spectrum of the window function , scaled by the square root of the mean galaxy brightness within the window . to compute
, we started by extracting a bright psf star from the galaxy - subtracted image .
the region around the star was cleaned of other stars , if necessary , and we confirmed that the background was zero .
we used as large an aperture around the star as possible without allowing sky subtraction noise or uncertainties in the residual background level to corrupt the power spectrum of the psf at low wavenumbers .
this usually required masking the psf for radii greater than @xmath60 .
we checked the aperture photometry to confirm that the total flux of the psf star was consistent with the power measured within the aperture .
the fourier transform and power spectrum of the psf were computed and fitted with a third - order polynomial .
for small wavenumbers ( @xmath61 ) , the fit to the power spectrum was used to create a composite psf spectrum , and the flux at @xmath59 is normalized to unity .
next , we multiplied the window function mask by the point source mask which removed the detected globular clusters and galaxies brighter than the limiting magnitude .
the window mask was then multiplied by the square root of the galaxy profile model .
the scaling of the variance by the mean galaxy brightness was thereby included in the expectation power spectrum .
the fourier transform and power spectrum for the resulting image were computed .
the expectation power spectrum is the convolution of the normalized power spectrum of the psf and the power spectrum of the modified window function : @xmath62 in computing in this way , we are effectively multiplying the data by the mask prior to convolution with the psf .
the effect of this approximation on the expectation power spectrum is minimal , and is very nearly the power spectrum of the psf alone .
the fluctuation magnitude is computed from the fit to the data in the radial region of the galaxy where @xmath54 is mostly constant with wavenumber @xmath63 . near the center of the galaxy ,
the sampled area is small and the power spectrum varies due to the small number of fluctuations being measured . in the largest apertures , noise from sky subtraction dominates and the @xmath20 ratio is significantly reduced . to get the best fit for @xmath54 , we examine the power spectra in apertures centered on the galaxy , and include the largest area possible in the final fit that does not compromise the quality of the power spectrum . for the data described in this paper , we measured @xmath54 in apertures with inner and outer radii of 2 , 6 , 12 , 24 , 48 and 80 .
the fits to the power spectra were performed in radial regions where @xmath54 was nearly constant and for wavenumbers in the range @xmath64 .
once @xmath54 is known from the fit to the data ( equation [ sbffit ] above ) , the fluctuation magnitude is defined as @xmath65 where @xmath54 is a flux in and @xmath66 is the magnitude of a source yielding one at the top of the atmosphere .
pixel to pixel variance can arise from a number of sources other than the surface brightness fluctuations resulting from the poisson statistics of the stellar population .
foreground stars , globular clusters , other galaxies ( usually in the background ) and clumpy dust can all contribute to the fluctuation power we measure . by working in the near - infrared
, we minimize the effects of dust absorption .
faint galactic foreground stars beyond limiting magnitude of our observations are not numerous enough to add significantly to the sbf power measured ( wainscoat et al . 1992 ) . to remove the contribution from gcs , we first identify as many as possible .
gcs brighter than the completeness limit are masked .
assuming we know the form of the globular cluster luminosity function ( gclf ) , we can extrapolate beyond the completeness limit to determine the residual variance due to undetected clusters .
this technique has been shown to work successfully by decreasing the cutoff magnitude and confirming that the fluctuation magnitude does not change significantly ( tonry 1991 ) .
blakeslee & tonry ( 1995 ) used sbfs to measure the gclf of coma cluster elliptical galaxies without detecting the majority of the gcs individually .
the procedure for removing background galaxies is similar .
we determine the magnitudes of the galaxies brighter than the completeness limit and assume a power - law distribution of galaxies to estimate the contribution from undetected galaxies .
since stellar sbfs are much redder than gcs , we expect the relative contribution to the fluctuations from gcs to be smaller
. however , we do not sample the gclf as deeply at because the background is brighter and our sensitivity to faint point sources is reduced . to estimate the relative contributions to the fluctuation power from gcs , background galaxies , and stellar sbfs as a function of limiting magnitude and distance , we adapted blakeslee & tonry s ( 1995 ) models to the @xmath13 band . to calculate the models plotted in figure [ models ] , we derived the following expressions for the variances arising from stellar sbfs , gcs , and background galaxies . in general , the variance arising from a population with @xmath67 sources per unit flux per pixel is @xmath68 ( tonry & schneider 1988 , eq . 4 ) .
from the definition of the fluctuation magnitude @xmath69 , the stellar sbf contribution is @xmath70 where @xmath71 is the magnitude of an object yielding one unit of flux per total integration time . from blakeslee & tonry ( 1995 ) , equation 12 ,
we write the gc contribution as @xmath72}~{\rm erfc}(z ) \label{gc}\ ] ] where @xmath73 is the gclf width , @xmath74 is the turnover magnitude for the gclf , and @xmath75 is the cutoff magnitude . all gcs brighter than the cutoff magnitude are masked and do not contribute to @xmath76 .
the argument of the complementary error function is @xmath77 taking the ratio of equations [ sbf ] and [ gc ] gives @xmath78}~{\rm erfc}(z ) \label{gc / sbf}\ ] ] in the upper panel of figure [ models ] we plotted equation [ gc / sbf ] for several values of the cutoff magnitude @xmath75 and taking the width of the gclf to be 1.4 mag ( blakeslee & tonry 1995 , 1996 ) . the distance scale was tied to the virgo cluster , for which we used a distance modulus of 31.0 mag ( tonry 1997 ) .
the peak magnitude for the gclf was taken to be @xmath79 for the @xmath80 band at the distance of virgo ( secker & harris 1993 ) , and we assumed @xmath81 for gcs ( frogel et al .
1978 ) to get @xmath82 .
we adopted a gc specific frequency of @xmath83 , a typical value for giant ellipticals ( blakeslee 1997 ; blakeslee , tonry , & metzger 1997 ; blakeslee & tonry 1995 ) .
the galaxy color we adopted was @xmath84 ( frogel et al .
1978 ; glass 1984 ) . to calculate the relative contribution to the fluctuation power from unresolved background galaxies ,
we first assume a power - law distribution @xmath85 cowie et al .
( 1994 ) found @xmath86 galaxies degree@xmath24mag@xmath23 at @xmath87 and a power - law coefficient of @xmath88 .
this agrees nicely with the work of djorgovski et al .
( 1995 ) , who found @xmath89 and the same normalization at @xmath87 .
we adopt @xmath90 , which translates to one galaxy arcsec@xmath24mag@xmath23 at @xmath91 .
we can write equation 9 from blakeslee & tonry ( 1995 ) as @xmath92 the pixel scale @xmath93 is included so that @xmath94 is in units of flux per integration time per pixel@xmath0 . combining equations [ sbf ] and [ bg ] ,
we find that the relative contribution from background galaxies is @xmath95~{+}~(\gamma-0.8)m_c~{-}~29.38\gamma}\ ] ] where @xmath96 and @xmath97 ( jlt ) . to plot realistic models as a function of cutoff magnitude @xmath75 and distance modulus
, we need to estimate typical mean galaxy brightnesses and the magnitude of an object giving one unit of flux per integration . over the areas of the fornax galaxies in our sample
in which we perform the sbf analysis , the surface brightness ranges from 17.0 to 18.8 mag arcsec@xmath24 , which corresponds to a typical flux of @xmath41 . for the 1995 october observing run , @xmath98 for a source with a flux of 1 .
the resulting plots for the fornax observations are shown in the lower panel of figure [ models ] .
the models shown in figure [ models ] are useful for predicting the relative contributions of gcs and galaxies to our sbf measurement for a given exposure time . for example , at the distance of fornax ( marked with a line in figure [ models ] ) and assuming a cutoff magnitude of @xmath87 , the contribution from unexcised globular clusters is one - third that of the stellar sbfs , or 25% of @xmath54 . at the same limiting magnitude and distance
, the contribution from background galaxies is smaller than that of the stellar sbfs by a factor of 10 .
it is interesting to note that there is a maximum gc contribution to the fluctuation power measured which is approximately equal to the stellar sbf variance . in the @xmath14 band ,
the gc contribution is a factor of 10 greater compared to the stellar fluctuations .
even though the effect of gcs on @xmath54 is greatly reduced at @xmath13 , it is not negligible . when the contribution from gcs is larger than @xmath99 of the fluctuation power @xmath54 ( or the ratio in eq .
[ gc / sbf ] exceeds 0.5 ) , the uncertainty in the correction becomes the dominant source of error and the distance derived unreliable .
it is therefore crucial that ir sbf measurements be deep enough to adequately sample the gc and background galaxy populations .
inspection of figure [ models ] shows that the ability to remove the gc and galaxy populations from the sbf measurement will be a limiting factor in extending the sbf distance scale to @xmath100100 mpc .
in the following sections , we describe the practical techniques we used to measure and subtract the gc and galaxy fluctuations , and argue that identifying objects in deep optical images and removing them from ir images prior to measuring the sbf amplitude is potentially the most profitable way to proceed .
to measure the globular cluster and background galaxy contributions to the sbf power spectrum , we first identified and performed photometry on as many sources as possible in the image using the dophot program version 3 ( schechter , mateo , & saha 1993 ) .
dophot identifies sources in a galaxy - subtracted image and determines their magnitudes by scaling the psf . for optical sbf work , a modified version of dophot ( ver .
2 ) was developed which carefully accounts for the galaxy subtraction and sbf contributions to the noise ( tonry et al .
1990 ) . in the results presented here , we used the standard `` off - the - shelf '' dophot version 3 which does not correct for galaxy subtraction in modeling the noise . at , sky subtraction noise dominates at all radii except very near the center of the galaxy .
dophot performs well except within a few arcseconds of the center , and we examined the objects found near the center and edited them manually as appropriate . with the output list of objects and magnitudes from dophot , we constructed luminosity functions in radial bins and determined the completeness limit .
a mask was made to remove all objects brighter than the completeness limit , and we estimated the residual variance for the population fainter than the cutoff magnitude . to do this , we fitted the observed luminosity function with a sum of a gaussian gclf and a power - law function for the background galaxies .
objects brighter than @xmath101 were not included in the fits , nor were objects with @xmath102 .
the gclf width was assumed to be 1.4 mag ( blakeslee & tonry 1995 , 1996 ) , and the distance was varied to determine the best fit .
when the gclf was not well - sampled , the distance was fixed and the fits performed without distance as a free parameter .
the galaxies were assumed to have a power - law distribution with an exponential coefficient of 0.30 ( cowie et al .
1994 , djorgovski et al .
the normalization was left as a free parameter , and we confirmed that the fit predicted reasonable number counts at faint magnitudes . because we tried to constrain both the gc and galaxy populations with relatively few points at the bright end of the distributions , the covariance between the components was often significant .
for these observations , the completeness limit was 2 to 3.5 mag brighter than the peak of the gclf .
our ability to sample the luminosity function was not only limited by the bright sky background , but also by the small field of view .
we found that the fits were best constrained by fixing the distance to the galaxy and hence the gclf peak @xmath103 , either using cepheid or @xmath14-band sbf distances .
the fits using these assumed distances were good , but our measurements of the gclf do not constitute an independent measurement of the distance given our sampling of only the brightest few clusters . once the luminosity functions had been fitted , we extrapolated and integrated the contributions to the variance coming from sources fainter than the cutoff magnitude .
we then computed the residual power @xmath104 for the region of the galaxy being examined , and we subtracted @xmath104 from @xmath54 prior to computing the fluctuation magnitude .
ir sbf measurements are limited by the bright sky background rather than photon shot noise in the fluctuations themselves . because the background is so bright , integration times would have to be much longer at than at @xmath14 to reach comparable limiting magnitudes .
however , at fluctuations are much brighter than at @xmath14 , so it is not necessary to reach as faint a limiting magnitude to measure the fluctuations .
the result is that at we identify many fewer globular clusters than in comparable observations at @xmath14 .
furthermore , modern ccds typically cover a larger field of view than ir detectors , making it possible to sample more of the gclf than we can at .
a comparison of the luminosity functions from @xmath14 and -band sbf observations of ngc 1399 is shown in figure [ likefig2 ] .
the @xmath14 band data are taken from tonry ( 1997 ) . even though the peak of the gclf is brighter at than it is at @xmath14 , the deeper @xmath14-band images sample a much greater fraction of the gclf .
we can take advantage of the deeper @xmath14-band images by using them to identify gc and galaxy positions , and we use this information to mask sources in the image .
this allows us to remove objects that are not detectable in the image before performing the sbf analysis . in the case of ngc 1399 ( fig .
[ likefig2 ] ) , the @xmath14-band completeness limit is approximately 22.75 mag . given that the mean color of gcs is @xmath81 ( frogel et al .
1978 ) and @xmath105 ( ajhar & tonry 1994 ) , we adopt a mean @xmath106 .
the equivalent cutoff magnitude of the @xmath14-band observation shifted to is 21.5 , 2.5 mag fainter than the completeness limit of the observation alone .
examination of figure [ models ] and assuming a distance modulus of 31.22 to fornax ( tonry 1997 ) shows that this corresponds to a ratio of @xmath107 . on the other hand , had we relied upon the observations alone , the cutoff magnitude would be @xmath108 mag , and the corresponding ratio @xmath109 .
when gcs and galaxies make up such a large fraction of the sbf magnitude , the uncertainty in the @xmath104 correction is large . by combining @xmath14 and data we gain the important advantage of reducing the @xmath104 correction to a negligible level
. this paper reports sbf results from our data alone , applying the corrections determined by fitting the gc and galaxy luminosity functions as described above .
we also present sbf magnitudes determined using the @xmath14-band masks . in practice ,
we collected @xmath14-band images and object masks from the optical sbf survey ( tonry 1997 ) .
the @xmath14-band images and masks were scaled , rotated , and cropped appropriately to bring them into registration with the image . in the subsequent sbf analysis , we replaced the -band mask with the @xmath14-band mask , removing clusters and galaxies from the image which were not detected because of the sky background noise .
the residual variance @xmath104 was taken to be zero and the fluctuation magnitude computed .
the nicmos3 and first - generation hawaii 1024@xmath0-pixel hgcdte detectors used for these observations have banded sensitivity and dark - current patterns that are easily seen in the raw images ( hodapp et al .
1996 , fig . 4 ) .
jlt found that imperfect sky and galaxy subtraction can leave a residual variance which contributes to the spatial power spectrum .
in low-@xmath20 ratio measurements , the residual background variance can be a significant fraction of the fluctuation power measured . imperfectly fitting the galaxy profile and background leave patterns that add power to the spectrum .
the sky background is @xmath110 times brighter than the galaxy , even near the center , and imperfect sky subtraction affects the fit to the galaxy profile .
also , if an object is not masked prior to fitting the galaxy profile , the iterative elliptical profile fitting routine will find too much flux and overestimate the galaxy brightness at that radius , resulting in an elliptical ring in the galaxy profile .
unmasked objects brighter than the completeness limit that escaped detection by dophot are not accounted for in the fit to the luminosity function and also contribute to the residual power spectrum . in fitting the sbf power spectrum to determine @xmath54
, we avoid low wavenumbers @xmath111 which are contaminated by large - scale features in the image .
however , residual patterns are not smoothly varying , and are observed to contribute power on scales of @xmath112 , where the fit to @xmath54 is commonly performed .
we attempted to measure the contribution from residual patterns in several different ways .
first , we examined the radial behavior of @xmath54 .
we frequently found that @xmath54 increased with radius by as much as a factor of two within 1 of the center .
stellar population models would require a decrease in age of @xmath113 gyr or a increase in [ fe / h ] of at least 1 dex to account for a gradient this large ( worthey 1994 , 1997 ) .
galaxies that do show radial sbf gradients such as ngc 3379 ( sodemann & thomsen 1995 ) or maffei 1 ( luppino & tonry 1993 ) only show a change of @xmath114 mag within an arcminute of the galaxy center . furthermore , from dynamical arguments , we expect the outer envelopes of elliptical galaxies to be older and more metal - poor than the centers ( faber et al .
stellar population models predict that ir sbfs should be fainter in the outer regions , not brighter , if the outer regions of the galaxy are older and metal deficient .
we regard the increasing sbf amplitude with radius as evidence for residual variance , especially in relatively low-@xmath20 ratio observations .
variance remaining due to improper galaxy subtraction , incomplete point - source masking , and residual dark current patterns will affect the larger outer annuli more strongly , leading to an increasing fluctuation power with radius . assuming a modest residual power of 0.1% of the sky value usually brings all apertures to the same fluctuation power .
we conducted several experiments to see if a level of residual variance of 0.1% of the sky level was reasonable . in one trial , we took two averaged sky images observed on one night , separated in time by about half an hour .
the resulting difference image contains residual patterns at the level we might expect to accumulate during typical observing , since in our normal image reduction strategy we averaged individual sky frames taken during approximately half an hour .
the power spectrum of the difference image was measured using the same sbf procedures used to analyze the data images .
the spectrum rose steeply for @xmath115 , and gradually decreased with increasing @xmath63 . the residual variance measured in this way
should be considered an upper limit , since the patterns were added coherently ( sky frames are combined without registration ) . in the final registered galaxy images ,
residual patterns were partially cancelled because individual galaxy images had been dithered . in another experiment , we combined individual sky images with their relative offsets , so that residual patterns were not added coherently as in the previous example .
we then computed the power spectrum and measured the deviation from a flat spectrum .
the power spectra show the same behavior observed in the previous experiment but with a lower amplitude . since this measurement of the residual variance is for the sky subtraction only , and
does not account for any residual pattern from the galaxy subtraction or point source removal , it is a lower limit of the residual variance we expect in a galaxy power spectrum .
the variances measured in these two experiments fall between 0.02% and 0.2% of the sky value , confirming that 0.1% is a reasonable estimate of the residual spatial variance .
occasionally faint objects were not identified by dophot ; we measured the flux in these objects and found that they can contribute up to 0.1% of the sky level in variance .
we removed such objects and repeated the sbf analysis to confirm the reliability of our procedure for removing residual variance .
a few faint unidentified objects appear when using the @xmath14-band point source mask because very red galaxies detected at can be fainter than the completeness limit in the @xmath14-band image , and therefore remain unmasked .
our measurements of residual power in the sky images were used as guidelines for determining the best value of the residual variance ( @xmath116 ) , which we determined by eliminating any radial trend in @xmath54 . when radial gradients in @xmath54 could be attributed to residual variance , adopting a correction of @xmath51 of the sky value usually removed the effect . however , residual patterns were not always uniform , and low-@xmath20 ratio observations frequently showed variable radial behavior that could not be removed by simply subtracting a uniform residual variance .
experience with our data showed that the best way to deal with the residual variance was to use the value of @xmath116 that removed a slope in @xmath54 when a clear trend exists , and to adopt a uniform correction of 0.1% of the sky level prior to sky subtraction when the radial behavior of @xmath54 was uncertain . minimizing the radial variation in @xmath54
usually required a correction of @xmath117 at @xmath118 , or in general , @xmath119 .
the lower the background noise , the smaller the required correction for residual variance .
the sbf analysis was performed in the annular region that produced the cleanest power spectrum , and the residual variance correction @xmath116 was subtracted from the fluctuation power @xmath54 prior to computing the fluctuation magnitude ( in addition to any correction @xmath104 applied to account for faint globular clusters and galaxies ) .
since the value of @xmath116 is only an estimate of the residual spatial variance , its associated uncertainty is large when @xmath116 is a significant fraction of @xmath54 .
clumpy dust distributions can also add to the fluctuation power measured and cause the distance to the galaxy to be underestimated . by working at ,
we minimize the effects of dust extinction .
careful inspection of the images revealed no evidence of patchy dust , and we made no additional correction to . if clumpy dust were present in significant quantities , we would expect it to be concentrated toward the center of the galaxy . the resulting sbf amplitude would decrease with radius , which is opposite the trend arising from the other sources of residual variance described in the previous paragraph . for our data ,
we find that sbf amplitudes are larger in the outer annuli , and we conclude that patchy dust in the innermost regions of these galaxies does not affect the sbf magnitude .
we measured apparent fluctuation magnitudes for five giant elliptical galaxies in the fornax cluster using the techniques described in section [ howtosbf ] .
the power spectra are shown in figure [ powerspectra ] , along with the individual components of the fit @xmath120 and @xmath56 .
two values of are listed in tables [ fornaxmbarskp ] and [ fornaxmbarsi ] for each galaxy : the magnitudes in table [ fornaxmbarskp ] were derived using the data alone ; the magnitudes in table [ fornaxmbarsi ] were determined using the @xmath14-band images to identify and remove faint gcs and galaxies .
we compared the fluctuation magnitudes measured using the two different masks in figure [ imaskvskmask ] .
in the first set of magnitudes ( table [ fornaxmbarskp ] ) , we fitted the background galaxy and gc luminosity functions and integrated them to compute the power @xmath104 from objects fainter than the completeness limit .
because of the high background , the observations only sample the brightest gcs and galaxies ( fig .
[ likefig2 ] ) , and @xmath104 was frequently a significant fraction of the total power @xmath54 . when @xmath121 exceeds @xmath122 , the uncertainty in the sbf magnitude from the gc and galaxy contribution becomes the dominant source of error .
we did not apply an additional residual variance correction @xmath116 because of the significant uncertainty in @xmath104 , thus the uncertainties listed in table [ fornaxmbarskp ] are not directly comparable to those in table [ fornaxmbarsi ] . for the second set of fluctuation magnitudes ( table [ fornaxmbarsi ] ) , the deeper @xmath14-band images were used to identify and mask gcs and galaxies , and we assumed that their contribution to is negligible ( @xmath123 ) . as was shown in section [ imasks ] for ngc 1399 ,
the cutoff magnitude in the @xmath14-band images ( shifted to ) is usually 2 mag fainter than the typical cutoff magnitude of @xmath124 in the images . applying the models in section [ intheory ] ( fig .
[ models ] ) , we find that @xmath125 and can safely be ignored . without worrying about gcs and galaxies , we addressed the issue of residual variance in the images .
we followed the procedures outlined in section [ residvar ] and adopted a residual variance correction @xmath116 at a level which minimized radial variations in @xmath54 , usually near 0.1% of the sky brightness .
the relative sizes of the corrections ( @xmath126 ) are listed in table [ fornaxmbarsi ] .
figure [ imaskvskmask ] shows that sbf magnitudes measured using the different masks are consistent for the high-@xmath20 observations ( filled symbols ) .
the diagonal line with unit slope and zero intercept indicates perfect agreement , not a fit to the data .
four of the six low-@xmath20 measurements have significantly brighter values of using the mask than using the @xmath14 mask .
ccccccc fornax&&&&&&ngc 1339 & 224 & @xmath127 & 4.5 & 0.12 & 25.55 & 0.14 ngc 1344 & 248 & @xmath128 & 3.5 & 0.12 & 25.52 & 0.11 ngc 1379 & 224 & @xmath129 & 2.3 & 0.40 & 25.60 & 0.19 ngc 1399 & 248 & @xmath130 & 3.2 & 0.38 & 26.10 & 0.14 ngc 1404 & 1248 & @xmath131 & 3.4 & 0.29 & 25.91 & 0.14 eridanus&&&&&&ngc 1395 & 1248 & @xmath132 & 1.6 & 0.64 & 26.07 & 0.12 ngc 1400 & 248 & @xmath133 & 1.8 & 0.30 & 26.10 & 0.12 ngc 1407 & 1248 & @xmath134 & 3.8 & 0.53 & 26.60 & 0.16 ngc 1426 & 248 & @xmath135 & 1.4 & 0.30 & 25.22 & 0.33 virgo&&&&&&ngc 4365 & 248 & @xmath136 & 4.9 & 0.25 & 26.40 & 0.11 ngc 4406 & 248 & @xmath137 & 9.0 & 0.06 & 25.46 & 0.10 ngc 4472 & 248 & @xmath138&20.0 & 0.03 & 25.31 & 0.11 ngc 4489 & 248 & @xmath139 & 3.1 & 0.03 & 24.49 & 0.35 ngc 4552 & 248 & @xmath140&11.0 & 0.03 & 25.32 & 0.11 ngc 4578 & 1224 & @xmath141 & 1.3 & 0.02 & 25.45 & 0.30 ngc 4636 & 248 & @xmath142 & 4.1 & 0.09 & 25.38 & 0.15 cccccccc fornax&&&&&&&ngc 1339 & 1224 & @xmath143 & 1.8 & 0.34 & 0.7 & 25.96 & 0.15 ngc 1344 & 248 & @xmath144 & 3.3 & 0.22 & 1.5 & 25.58 & 0.12 ngc 1379 & 224 & @xmath145 & 3.2 & 0.15 & 1.8 & 25.44 & 0.13 ngc 1399 & 248 & @xmath146 & 3.0 & 0.32 & 1.1 & 25.98 & 0.16 ngc 1404 & 1248 & @xmath147 & 2.8 & 0.17 & 1.6 & 25.82 & 0.14 eridanus&&&&&&&ngc 1395 & 1248 & @xmath148 & 1.3 & 0.22 & 0.8 & 25.43 & 0.17 ngc 1400 & 248 & @xmath149 & 1.1 & 0.48 & 0.4 & 26.46 & 0.25 ngc 1407 & 248 & @xmath150 & 3.0 & 0.15 & 1.7 & 26.19 & 0.15 ngc 1426 & 248 & @xmath151 & 1.1 & 0.51 & 0.4 & 25.67 & 0.30 virgo&&&&&&&ngc 4365 & 248 & @xmath152 & 4.5 & 0.00 & 4.5 & 26.09 & 0.11 ngc 4406 & 248 & @xmath153 & 8.7 & 0.06 & 5.6 & 25.45 & 0.10 ngc 4472 & 248 & @xmath154&16.0 & 0.03 & 11.3 & 25.30 & 0.11 ngc 4489 & 248 & @xmath155 & 1.7 & 0.41 & 0.6 & 24.99 & 0.35 ngc 4552 & 248 & @xmath156 & 4.9 & 0.09 & 3.0 & 25.44 & 0.11 ngc 4578 & 1224 & @xmath157 & 1.4 & 0.41 & 0.5 & 25.54 & 0.30 ngc 4636 & 248 & @xmath133 & 3.5 & 0.13 & 2.0 & 25.59 & 0.15 fluctuation magnitudes were also measured for the four ellipticals in the eridanus cluster , following the same procedures as those used to measure sbfs in the fornax galaxies ( tables [ fornaxmbarskp ] and [ fornaxmbarsi ] ) .
because the fornax observations were our first priority , somewhat less time was allocated to the eridanus galaxies .
eridanus is also more distant , so the resulting observations have significantly lower in their @xmath20 ratios .
only ngc 1407 has a ratio of @xmath158 .
estimating uncertainties for can be difficult .
we proceeded by first identifying all the probable sources of error : the statistical uncertainty in the fit for @xmath54 , errors in measuring the sky offset , errors in fitting the galaxy profile , uncertainties in subtracting sources of residual variance , and so forth .
we modified the parameters individually through their full ranges and measured the change in @xmath54 . from the change in @xmath54 we estimated the 1-@xmath73 uncertainty arising from the parameter in question
we then combined the uncertainties in quadrature to get the uncertainties listed in tables [ fornaxmbarskp ] and [ fornaxmbarsi ] .
this procedure assumes that the different sources of uncertainty are independent , an assumption which is not strictly true .
for example , modifying the residual sky level affects the fit to the galaxy profile slightly .
errors in fitting the galaxy profile affect our estimate of residual spatial variance as well . in the procedure
used to fit the luminosity functions of the globular clusters and background galaxies , an error in one component obviously affects the fit to the other .
while the cross correlations are obviously not zero , they were found to be small compared to the principal errors .
the uncertainties computed are reasonable when we compare them with other distance estimates to calibrate the sbf distance scale .
we reanalyzed the data originally presented by jlt using the updated sbf techniques presented in this paper .
we improved our sbf measurements by determining the globular cluster and background galaxy contributions to the fluctuation amplitude .
we also calculated magnitudes using the masks generated from the @xmath14-band images and applied a residual variance correction , just as we did for the fornax and eridanus galaxies .
new fluctuation magnitudes are listed in tables [ fornaxmbarskp ] and [ fornaxmbarsi ] .
observations of five virgo cluster galaxies are higher in @xmath20 ratio than the fornax and eridanus data , and it is reassuring to find that the gc , galaxy , and residual pattern corrections are quite small .
the difference between fluctuation magnitudes measured using the different masks is negligible . in the nicmos data , the field of view was small enough that very few gcs were observed , and the sbf @xmath20 ratio was high enough that the gc contribution was negligible .
three of the virgo galaxies had low @xmath20 ratios , two of which appeared to have anomalously bright fluctuations ( jlt , pahre & mould 1994 ) . in our reanalysis of the jlt virgo data ,
ngc 4578 was found to have a typical for the virgo cluster but a large uncertainty due to the low @xmath20 ratio .
the fluctuation magnitudes reported for ngc 4365 by jlt and pahre & mould ( 1994 ) were anomalously bright , if this galaxy resides in the virgo w cloud behind the main virgo cluster core ( tonry et al .
1990 , forbes 1996 ) .
we re - observed this galaxy in 1996 using the science - grade hawaii detector and applied the full corrections for the globular cluster and background galaxy contributions .
the new observations confirmed that the earlier measurement was biased by a significant gc population masquerading as stellar sbfs . applying
the gc correction to the 1995 data ( jlt ) brought it into agreement with the 1996 observation reported here .
applying the @xmath14-band gc mask allowed us to further check the consistency of the measurement .
magnitudes listed in tables [ fornaxmbarskp ] and [ fornaxmbarsi ] are for the new 1996 data . when we used the @xmath14-band mask to remove gcs from the image , we measured a fluctuation magnitude of @xmath159 mag .
ngc 4365 illustrates that sbf amplitudes can be too large , making the galaxy appear significantly closer than it really is , if exposure times are not long enough to achieve a sufficiently high @xmath20 ratio and the globular cluster contribution is not adequately measured . with no _ a priori _ knowledge of the distance to ngc 4365 and follow - up observations , the jlt data would have lead to the incorrect conclusion that this galaxy resides in the main virgo cluster .
jlt and pahre & mould ( 1994 ) also observed an anomalously bright fluctuation magnitude in ngc 4489 .
we have not yet re - observed this galaxy , and reanalysis of the jlt data does not expose an obvious gc contribution that could be responsible for the fluctuation magnitude measured .
we have referred to 0/p1 as the @xmath20 ratio . with a better understanding of the variance arising from sources other than the stellar sbfs
, we computed a more realistic measure of the @xmath20 ratio : @xmath161 values for are listed in table [ fornaxmbarsi ] .
galaxies for which @xmath162 were not included in the calibration of .
comparison of the fluctuation magnitudes listed in tables [ fornaxmbarskp ] and [ fornaxmbarsi ] showed that the sbf magnitudes for these low-@xmath20 ratio galaxies are unreliable , in agreement with the conclusion reached by jlt from analysis of the virgo data alone .
the differences are plotted as a function of @xmath163 in figure [ snfig ] .
the mean difference between values for the low-@xmath20 ratio observations is 0.41 mag ; for the galaxies with @xmath164 the difference is only 0.15 mag , comparable to the individual uncertainties .
the highest-@xmath20 ratio observations show the smallest difference between values of measured with and without the @xmath14-band point - source masks .
it is clear that when either @xmath104 or @xmath116 exceeds @xmath165 , the resulting is unreliable . in the following section , only the observations with @xmath164 ( or @xmath166 )
were included in the statistical analysis . from our observations
we conclude that it is crucial that future ir sbf observations be deep enough to ensure that all sources of variance can be removed and that the @xmath20 ratio of the _ stellar _ sbfs is larger than 1 .
the definition of in equation [ snsbfeq ] is empirical , but we can also use the results of our survey to predict as a function of observational parameters . the relationship can be approximated as @xmath167 where @xmath168 is the telescope diameter in meters and @xmath169 is the seeing fwhm in arcseconds .
the relationship between and the observational parameters is not particularly good because a number of other parameters also affect the measured value of .
equation [ predictedsnsbf ] provides a basic guideline for the minimum integration required to make @xmath13-band sbf measurements .
the coefficient in equation [ predictedsnsbf ] is somewhat less than the value of 0.5 found by pahre & mould ( 1994 ) , and we emphasize that achieving a @xmath170 is marginal and assumes that residual variances have been subtracted . to make reliable @xmath13-band sbf distance measurements
, we recommend that on - source integration times be @xmath171
to calibrate the sbf distance scale , we first calculated absolute fluctuation magnitudes for the high-@xmath20 ratio galaxies in our sample .
we adopted three different ( but not independent ) sets of distances : cepheid distances to m31 and the virgo cluster , @xmath14-band sbf distances to individual galaxies , and cluster @xmath14-band sbf distances averaged over a larger sample of galaxies than we observed at .
our calibration of the ir sbf distance scale is strengthened significantly by the cepheid distances to several virgo spirals that have now been measured ( freedman et al .
1994 ; pierce et al .
1994 ; ferrarese et al .
1996 ; saha et al .
1996a , b ; sandage et al .
of the five reported cepheid distance measurements , four were made using data acquired using the hubble space telescope .
three of these hst measurements are remarkably consistent : ngc 4321 @xmath172 ( ferrarese et al . 1996 ) , ngc 4496a @xmath173 ( saha et al .
1996b ) , ngc 4536 @xmath174 ( saha et al . 1996a ) ; ngc 4639 is @xmath175 mag more distant than the other virgo cepheid distances ( sandage et al .
1996 ) . like ngc 4365 ,
it is probably a member of the background virgo w cloud , and has been excluded from the average virgo cepheid distance modulus used to calibrate the sbf distance scale .
we adopt a cluster distance of @xmath176 ( @xmath177 mpc ) for the virgo cluster .
recently the hst key project team has measured a cepheid distance to the fornax spiral galaxy ngc 1365 . at this time
the analysis is nearly complete , and the cepheid distance modulus to ngc 1365 is approximately @xmath178 ( silbermann 1998 ) . in the near future hst cepheid distances to three fornax cluster galaxies
will be measured by the key project team . until then , we use only the cepheid distances to m31 and the virgo cluster to calibrate the sbf distance scale .
j. tonry and collaborators are in the process of publishing the results of an extensive @xmath14-band sbf survey ( tonry et al .
the @xmath14-band sbf distance scale was calibrated using cepheid distances to a number of galaxies out to 20 mpc .
we can use @xmath14-band distances to calibrate the sbf scale using a larger sample of galaxies than using the sbf and cepheid distances alone .
we used the @xmath14-band data in two ways .
first , distances to individual galaxies ( tonry 1997 ) gave absolute fluctuation magnitudes directly for each galaxy .
second , mean distances to each cluster ( tonry et al .
1997 ) were used , allowing us to take advantage of a larger sample but requiring us to assume that all the galaxies in a cluster are at the same distance . in both cases
the @xmath14-band distances were calculated according to the calibration of tonry et al .
@xmath14-band distance moduli are listed in table [ fornaxm - mtab ] .
recent measurements by the hipparcos astrometric satellite have suggested that a systematic error in the cepheid distance scale may exist because of an error in the parallax distances to the nearest stars ( feast & catchpole 1997 ) .
if this is the case , all cepheid distances need to be increased by @xmath179 , although the exact change may depend on metallicity corrections to the period - luminosity relation for cepheid variables ( madore & freedman 1997 ) .
the cepheid zero - point correction also affects the @xmath14-band sbf distances , since they are also calibrated using the cepheid distance scale .
because this issue has not yet been resolved , we adopt the distances cited above with a caution to the reader that the sbf distance calibration presented in this paper may need to be revised in the future .
surface brightness fluctuations can also be calibrated using theoretical models by computing the second moment of the luminosity function for a stellar population and normalizing by the first moment .
if we know the absolute magnitudes of the brightest stars ( which dominate the second moment of the luminosity function ) from stellar population models , we can determine the distance directly from the apparent sbf magnitude .
alternatively , we can check our empirical calibration of the sbf distance scale against the synthesized stellar population models .
g. worthey produced a set of single - age , single - metallicity stellar population models and computed fluctuation magnitudes in the johnson @xmath180 , @xmath181 , and @xmath13 bands ( worthey 1994 , 1993b ) .
he kindly created new models specifically for the filter , and we compare our fluctuation magnitudes with these new models ( worthey 1997 ) .
the sbf models show only slight differences from the @xmath13 models , but are significantly different from the @xmath13 models transformed to using the standard relation from wainscoat & cowie ( 1992 ) .
sbfs are considerably redder than the stars upon which the wainscoat & cowie transformation is based .
the new model magnitudes are listed in table [ kpmodels ] .
the fact that the new models are similar to the @xmath13 models is reassuring .
the detailed stellar population models worthey created include the co absorption bandhead , which falls close to the edge of the @xmath13 window .
we were not certain to what extent differences in co absorption between the @xmath13 and filters would affect the measured sbf magnitudes .
now that worthey ( 1997 ) has produced models precisely for our filter , we can compare them to our observations with confidence .
the relatively small difference ( @xmath182 mag ) between @xmath13 and sbf magnitudes allows us to compare to pahre & mould ( 1994 , @xmath183 ) as well since @xmath183 is centered midway between @xmath13 and .
we make a detailed comparison of our data with worthey s models in the following section .
crrrr 5&@xmath184 & 0.185 & 1.121 & @xmath185 & @xmath186 & 0.215 & 1.186 & @xmath187 & @xmath188 & 0.268 & 1.291 & @xmath189 & @xmath190 & 0.316 & 1.355 & @xmath191 8&@xmath192 & 0.034 & 0.800 & @xmath193 & @xmath194 & 0.056 & 0.846 & @xmath195 & @xmath196 & 0.092 & 0.887 & @xmath197 & @xmath198 & 0.160 & 1.083 & @xmath199 & @xmath200 & 0.194 & 1.147 & @xmath201 & @xmath186 & 0.239 & 1.258 & @xmath202 & @xmath188 & 0.290 & 1.346 & @xmath185 & @xmath190 & 0.331 & 1.395 & @xmath203 12&@xmath204 & 0.042 & 0.857 & @xmath205 & @xmath194 & 0.064 & 0.892 & @xmath206 & @xmath196 & 0.100 & 0.930 & @xmath207 & @xmath198 & 0.171 & 1.110 & @xmath208 & @xmath200 & 0.214 & 1.204 & @xmath209 & @xmath186 & 0.258 & 1.304 & @xmath210 & @xmath188 & 0.314 & 1.396 & @xmath211 & @xmath190 & 0.350 & 1.427 & @xmath212 17&@xmath204 & 0.046 & 0.897 & @xmath213 & @xmath194 & 0.069 & 0.937 & @xmath214 & @xmath196 & 0.115 & 0.975 & @xmath197 & @xmath198 & 0.187 & 1.159 & @xmath208 & @xmath200 & 0.231 & 1.247 & @xmath215 & @xmath186 & 0.275 & 1.342 & @xmath216 & @xmath188 & 0.335 & 1.444 & @xmath211 & @xmath190 & 0.374 & 1.482 & @xmath217 in addition to single - age , single - metallicity populations , we also considered simple composite populations .
we used worthey s ( 1994 ) @xmath13-band models to show that models vary smoothly between grid points .
we tried combining a ( 8 , @xmath218 ) ( gyr , [ fe / h ] ) population with a ( 17 , @xmath219 ) population , with the younger population weighted by 5% to 50% ( by mass ) .
the point in halfway between the two models was reached when 30% of the population was younger .
we also combined a ( 12 , @xmath218 ) model with a ( 5 , 0.0 ) stellar population and found the midpoint to be at the same ratio of populations .
in other words , at least 30% of the stellar population by mass must be young for the young stars to dominate the @xmath13-band sbf magnitude .
an old population will not look like a significantly younger population in @xmath13-band fluctuation magnitude if only a few percent of the mass is in the young population .
stellar populations younger than 5 gyr were not considered . because fluctuations are dominated by the brightest stars in the population , model sbf magnitudes are sensitive to the treatment of the complex stellar physics of the luminous asymptotic giant branch ( agb ) stars .
fluctuations significantly brighter than model predictions may be evidence for young populations with brighter agb stars than the model population .
g. worthey s models ( 1997 ) also predict the redshift dependence of the sbf magnitudes as a function of age . in table
[ kcorr ] we list the k corrections in the @xmath13 band for redshifts between @xmath220 .
the models are all for solar metallicity [ fe / h]=0 .
the tabulated values must be added to apparent @xmath13-band sbf magnitudes to find the equivalent magnitude at @xmath221 .
the k corrections for @xmath13-band sbfs are _ very _ small , especially for the older stellar populations . for the galaxies in this study ,
k corrections can be ignored . even in future @xmath13-band sbf observations out to 10,000 , uncertainties due to k corrections will be negligible .
ccccc 0.005 & 0.004 & 0.003 & 0.002 & @xmath47 0.010 & 0.008 & 0.006 & 0.003 & @xmath222 0.015 & 0.011 & 0.008 & 0.003 & @xmath223 0.020 & 0.013 & 0.010 & 0.003 & @xmath224 0.025 & 0.015 & 0.012 & 0.003 & @xmath225 0.030 & 0.017 & 0.013 & 0.002 & @xmath226 broad - band colors are frequently used to constrain the properties of stellar populations . in the optical @xmath14 band ,
the effects of age and metallicity on sbfs are degenerate .
therefore the @xmath14-band sbf distance scale can be calibrated using the v - i color to parameterize both the age and metallicity of stellar populations ( tonry 1997 , worthey 1993a ) .
@xmath14-band sbfs are sensitive to galaxy color and has a fairly steep slope with v - i , but the scatter is small and @xmath14-band sbfs can be used to measure accurate distances over a wide range of stellar population . at ,
stellar population models ( worthey 1993a , 1997 ) predict a shallower slope with v - i , but more importantly , the slope is opposite in sign .
age and metallicity effects do not counteract each other as in the @xmath14-band .
because the age - metallicity degeneracy is partially broken , the dispersion in fluctuation magnitudes is also expected to be larger than at @xmath14 . to explore the behavior of sbfs with v - i color
, we plotted the fluctuation magnitudes as a function of v - i in figure [ vminusi ] .
filled circles denote those measurements for which the @xmath20 ratio was large ( @xmath164 or @xmath228 ) .
the magnitudes derived using the @xmath14-band masks were used to create the figures in this section .
we superimposed worthey s models ( 1997 ) , with grid lines labeled by age and metallicity .
to place the models , we used the mean cepheid distance for virgo and cluster @xmath14-band sbf distances for the fornax and eridanus clusters . from figure [ vminusi ]
we learn that the galaxies on the red end of the sample are mostly consistent with older , 12 to 17 gyr stellar populations with metallicities near [ fe / h]=@xmath218 .
the bluer galaxies with @xmath229 have fluctuation magnitudes consistent with 5 to 8 gyr population models and similar metallicities .
the trend is easily seen in figure [ mkvi ] , in which we plotted the absolute fluctuation magnitudes as a function of v - i for all the high-@xmath20 ratio observations in our sample .
the upper panel in figure [ mkvi ] used the cepheid distances to m31 and the virgo galaxies ; the lower panel was plotted using the @xmath14-band sbf distances to individual galaxies and includes galaxies in fornax and eridanus . for the bluer galaxies to be consistent with 12 to 17 gyr population models , the sbf magnitudes would have to be at least 0.5 mag fainter .
we see that the dispersion in v - i color appears to be due to variations in population age , not metallicity .
this conclusion contradicts the hypothesis that the dispersion in colors of elliptical galaxies is the result of metallicity variations ( buzzoni 1995 ) .
on the other hand , the age spread we infer from our data is consistent with the `` bursty '' formation scenario in which ellipticals form episodically , as predicted by the hypothesis that ellipticals form through mergers ( faber et al .
of course our sample is not very large , and this conclusion should be confirmed by additional high-@xmath20 ratio ir sbf measurements over a larger range of galaxy color .
we also learn something from the galaxies whose fluctuation magnitudes are not consistent with the stellar population models shown .
jlt and pahre & mould ( 1994 ) both noted two galaxies with anomalous sbf magnitudes in the virgo cluster ( ngc 4489 and ngc 4365 ) .
our latest observations of ngc 4365 reveal that this galaxy is indeed in the background w cloud .
the new fluctuation magnitude is fully consistent with the @xmath14-band distance ( tonry 1997 ) and with the distance derived from the peak of the gclf ( forbes 1996 ) . as stated in section
[ virgorevisited ] above , failure to remove significant residual variances ( from globular clusters in this case ) can result in significantly biased fluctuation measurements .
ngc 4489 s low-@xmath20 ratio fluctuation magnitude is @xmath175 mag brighter than the model predictions .
either it is a young galaxy with a possibly anomalous agb population , or the measurement is biased by residual variances which have not been adequately removed .
ngc 4489 does not appear to have an extensive gc population as does ngc 4365 , but it is unusual in other ways ( jlt ) . until deeper @xmath13-band images are obtained for ngc 4489 , our conclusions regarding an anomalous stellar population
will be uncertain .
observations of fornax and eridanus cluster galaxies have revealed other candidates with unusually bright sbf magnitudes ; all have low @xmath20 ratios @xmath162 .
comparison of the fluctuation magnitudes for ngc 1395 in tables [ fornaxmbarskp ] and [ fornaxmbarsi ] reveals a difference of 0.64 mag .
a reasonable correction for globular clusters from the observations brings the into agreement with the stellar population models .
nevertheless , the @xmath14-band mask should remove the gcs much better than using the data alone .
ngc 1426 is also seen to be anomalously bright in both tables [ fornaxmbarskp ] and [ fornaxmbarsi ] .
given that these observations all have low @xmath20 ratios , all appear too bright , and all have large ( and uncertain ) corrections for residual variances @xmath104 and @xmath116 , we can not draw any conclusions regarding their stellar populations .
when @xmath162 , fluctuation magnitudes are unreliable and are not included in the calibration of .
the high-@xmath20 ratio data plotted in figure [ mkvi ] can be fitted to find the relationship between and v - i .
the fits were performed using the three distance estimates and the two different sets of values described in sections [ fornaxanderidanus ] and [ distances ] above .
we used an iterative least - squares approach to weight each point according to the uncertainties in both and v - i .
the results are listed in table [ fornaxfitstable ] , which includes the number of points used in the fit and @xmath230 per degree of freedom .
the @xmath14-band gc masks produce calibrations of with consistent zero points . whether we use the individual @xmath14-band sbf distances or the mean distances for the clusters
makes little difference . somewhat greater scatter
is observed in the zero points of the fits which use the -mask data , confirming that the @xmath14-band gc and galaxy masks produce more consistent fluctuation magnitudes .
the @xmath230 values indicate that our uncertainties are reasonable .
the cepheid fits include only 5 points , and we believe it is small number statistics , rather than overestimated errors , that causes @xmath231 .
using the @xmath14-band sbf distances results in larger numbers of galaxies in the fits .
differences in age and metallicity between the galaxies produce a dispersion in somewhat larger than our uncertainties , and the @xmath230 per degree of freedom is larger than one .
the only statistically significant slopes we measured resulted from using the gc masks and the @xmath14-band sbf distances .
the positive slope suggests that sbfs are _ fainter _ in redder galaxies ,
opposite the trend predicted by the stellar population models ( figure [ mkvi ] ) .
no significant slope is seen in the data using the cepheid distances alone , but only 5 points were included in this fit , so the lack of a correlation is not conclusive either .
on the other hand , there is no convincing reason that a slope opposite that predicted by the models should exist .
our best sbf results are produced using the @xmath14-band masks , and the fits to these data are most consistent with a constant across the range in v - i spanned by our sample .
this is a very useful feature , since it implies that accurate v - i colors do not need to be measured to determine sbf distances .
future observations of a significantly larger sample including bluer galaxies will be required to determine whether or not sbfs get fainter as galaxies get bluer , as predicted by the stellar population models .
lccccccc cepheids & @xmath14 & @xmath232 & 0.06 & @xmath233 & 1.8 & 0.8 & 5@xmath14 sbf individual & & @xmath232 & 0.04 & 1.4 & 1.1 & 1.9 & 11@xmath14 sbf group & & @xmath234 & 0.04 & 1.7 & 1.1 & 1.2 & 9 cepheids & & @xmath235 & 0.06 & @xmath236 & 1.8 & 0.6 & 5@xmath14 sbf individual & & @xmath237 & 0.04 & 3.0 & 1.2 & 1.0 & 10@xmath14 sbf group & & @xmath238 & 0.05 & 3.0 & 1.2 & 2.0 & 9 the @xmath14-band masks produce the most reliable fluctuation magnitudes , and the @xmath14-band sbf and cepheid distance scale calibrations are equally good .
we therefore adopt as the calibration of the sbf distance scale @xmath239 with no v - i term for the limited range @xmath240 .
the statistical uncertainty is 0.06 mag ; adding 0.10 mag in quadrature for the cepheid zero point uncertainty results in a total uncertainty of 0.12 mag .
the statistical uncertainty in the zero point was derived by fitting a slope to v - i .
the scatter due to variations in stellar population results in somewhat greater uncertainties when no correction for v - i is made .
our calibration is consistent with the jlt measurement , which is not surprising given that the same virgo data were used in both studies .
this value for is the same as that measured by luppino & tonry ( 1993 ) for m31 , and is somewhat fainter than pahre & mould s ( 1994 ) measurement of @xmath18 , although the values are well within the errors .
our calibration was used to compute distance moduli for the galaxies in our survey , and we list distances in table [ fornaxm - mtab ] with the @xmath14-band sbf distance moduli for comparison . the comparison between @xmath14 and distance moduli is presented graphically in figure [ distmodfig ] .
the line shown is not a fit to the data , but simply the line with a slope of 1 which passes through the origin .
the high-@xmath20 measurements are consistent with the @xmath14-band sbf distances .
the errors listed for distance moduli are the individual errors in , and do not include the zero - point uncertainty .
mean sbf distances for each cluster are included with their standard deviations , where only the high-@xmath20 measurements were included in the mean .
once again , we note that since the same cepheid calibration was used for all distances tabulated , a systematic change in the cepheid distance scale will affect the @xmath14 and sbf distances equally .
ccccccccc fornax & & & & 31.30 & 31.22 & 0.06 & 31.32 & 0.24 ngc 1339 & 1.169 & 0.019 & 0.328 & & 31.34 & 0.14 & 31.57 & 0.15 ngc 1344 & 1.156 & 0.011 & 0.290 & & 31.17 & 0.09 & 31.19 & 0.12 ngc 1379 & 1.159 & 0.019 & 0.284 & & 31.30 & 0.08 & 31.05 & 0.13 ngc 1399 & 1.241 & 0.016 & 0.361 & & 31.42 & 0.14 & 31.59 & 0.16 ngc 1404 & 1.238 & 0.013 & 0.344 & & 31.27 & 0.08 & 31.43 & 0.14 eridanus & & & & & 31.71 & & 31.80&ngc 1395 & 1.239 & 0.012 & 0.340 & & 31.63 & 0.41 & 31.04 & 0.17 ngc 1400 & 1.216 & 0.009 & 0.336 & & 31.67 & 0.14 & 32.07 & 0.25 ngc 1407 & 1.264 & 0.010 & 0.354 & & 32.02 & 0.19 & 31.80 & 0.15 ngc 1426 & 1.176 & 0.009 & 0.303 & & 31.80 & 0.08 & 31.28 & 0.30 virgo & & & & @xmath241 & 30.96 & 0.05 & 31.06 & 0.12 ngc 4365 & 1.223 & 0.011 & 0.348 & & 31.70 & 0.09 & 31.70 & 0.11 ngc 4406 & 1.175 & 0.008 & 0.336 & & 31.19 & 0.07 & 31.06 & 0.10 ngc 4472 & 1.246 & 0.011 & 0.345 & & 30.94 & 0.06 & 30.91 & 0.11 ngc 4489 & 1.076 & 0.015 & 0.222 & & & & 30.60 & 0.35 ngc 4552 & 1.206 & 0.015 & 0.352 & & 30.86 & 0.08 & 31.05 & 0.11 ngc 4578 & 1.154 & 0.015 & 0.312 & & 31.25 & 0.08 & 31.15 & 0.30 ngc 4636 & 1.225 & 0.009 & 0.341 & & 30.83 & 0.08 & 31.20 & 0.15 local group&&&&&&&&m31 & 1.240 & 0.007 & 0.334&@xmath242 & 24.38 & 0.05 & 24.43 & 0.14 m32 & 1.145 & 0.007 & 0.199 & & 24.48 & 0.05 & 24.17 & 0.14 we also compared our fluctuation magnitudes with worthey s ( 1997 ) models as as function of the 2 spectral index .
values of the 2 index were taken from faber et al .
( 1989 ) , worthey et al .
( 1992 ) , and worthey ( 1993a ) .
there is a systematic offset between the faber et al . and the worthey et al .
measurements , the latter being larger by about 8% .
seven galaxies in our sample have 2 values listed by both worthey et al . and by faber et al . the offset of 0.027 mag between the two studies
was used to adjust the faber et al .
data to the worthey system for the fornax and eridanus cluster galaxies .
thus we compare the worthey ( 1997 ) models with the most compatible 2 data .
2 values are listed in table [ fornaxm - mtab ] .
we plotted as a function of 2 for the galaxies with @xmath164 in figure [ mg2 ] with the worthey ( 1997 ) models superimposed , as in the previous section .
very few of the observations fall within the range spanned by the models . for our observations to be consistent with the results of the previous section , 2 indices must be reduced by @xmath243 mag .
the reason for the discrepancy is not known , but may be the result of significantly enhanced [ mg / fe ] ratios ( 0.2 to 0.3 dex ) observed in giant ellipticals compared to stars in the solar neighborhood with similar metallicities ( worthey et al .
1992 , worthey 1994 ) .
we do not fit as a function of 2 to calibrate the sbf distance scale because of the small range in 2 spanned by our data and the inconsistency with the stellar population models .
radial gradients in fluctuation magnitudes potentially reveal changing ages and metallicities within the stellar population of a galaxy .
tonry ( 1991 ) showed that the color gradient in ngc 205 correlates with a change in the @xmath14-band fluctuation magnitude .
a younger stellar population at the center of ngc 205 gives rise to brighter @xmath14-band fluctuations at smaller radii than in the redder outer regions .
sodemann & thomsen ( 1995 ) discuss the implications of a radially increasing @xmath14-band fluctuation amplitude observed in ngc 3379 , opposite the trend seen in ngc 205 . at @xmath14
the effects of age and metallicity on are largely degenerate , making it difficult to interpret radial changes in fluctuation magnitude .
older stellar populations have fainter magnitudes , but populations with higher metallicity also have fainter magnitudes , making it impossible to distinguish between old , metal - poor and young , metal - rich populations . at , on the other hand , the degeneracy is partially broken . the worthey models ( 1993b , 1997 ) predict that ir fluctuations should fade with increasing age and decreasing metallicity , allowing us to separate young , metal - rich and old , metal - poor populations .
if an elliptical galaxy is formed by merging which triggers a burst of new star formation , we would expect the outer envelope of such a galaxy to be older and more metal - poor than the nuclear region ( faber et al .
the brighter @xmath14-band fluctuation magnitudes observed in the outer regions of ngc 3379 may be consistent with this scenario if the change is is principally due to a decrease in metallicity ; in ngc 205 the change in must be due to age differences between the populations .
pahre & mould ( 1994 ) observed no radial gradient in the @xmath13-band fluctuation magnitude in ngc 3379 , but a modest gradient may be hidden in the scatter of the individual measurements . while the prospects for using ir sbfs to reveal radial gradients in stellar populations are good
, the data presented here do not have a sufficiently high @xmath20 ratio to detect radial changes in fluctuation amplitude ; the uncertainties due to background subtraction and estimated residual variance mask the modest changes in that might result from differences in stellar populations within a galaxy .
if the outer regions of an elliptical galaxy are older and more metal - poor than the center , the fluctuation amplitude will decrease with radius . in our data
, uncorrected fluctuation amplitudes usually increase with radius , sometimes changing by a factor of two . even if we could imagine a scenario that creates a galaxy with the older stars at the center , the variations in stellar population that would be required to explain such large changes in fluctuation amplitude are not plausible .
instead , we interpret the observed sbf gradients in our data to be indicators of residual variations in the sky background
. using ir sbfs to reach definitive conclusions regarding stellar population gradients in nearby galaxies will require high-@xmath20 ratio data in which corrections for residual variance are negligible .
although we have not conclusively measured ir sbf gradients it should be emphasized that the breaking of the age and metallicity degeneracy potentially makes ir sbfs a powerful probe for distinguishing old , metal - poor populations from younger , metal - rich stars .
for example , the range in v - i in ngc 205 should give rise to a change in of @xmath244 mag , with becoming fainter with radius .
if an ir sbf gradient were detected in ngc 3379 , we could explain the very different optical sbf properties of these two galaxies in terms of the ages and metallicities of the stellar populations .
we have completed a ( 2.1 ) survey of surface brightness fluctuations in 16 giant elliptical galaxies in the virgo , fornax , and eridanus clusters .
we also observed m31 and m32 in the @xmath180 and @xmath181 bands . from this study
we draw the following conclusions : @xmath13-band sbfs can be measured reliably and used to determine distances to early - type galaxies . because sbfs are relatively bright in the ir and the seeing is typically very good , integration times can be quite modest . however , observations must be sufficiently deep to adequately sample the gc luminosity function and remove sources of residual variance .
low-@xmath20 ratio measurements are unreliable .
optical images can be used to identify and remove gcs and background galaxies that are not detected in the ir , improving the ir sbf measurement .
we empirically calibrated the sbf distance scale using cepheid distances to m31 and the virgo cluster .
the absolute fluctuation magnitude is @xmath245 ( statistical error ) with a total uncertainty of 0.12 mag .
no significant change in with v - i was observed over the range in color spanned by the galaxies in this sample .
accurate color measurements are not required to measure the sbf distance to a galaxy .
sbf magnitudes are consistent with predictions from simple stellar population models .
the lack of a correlation in with v - i is best explained by a spread in ages among the galaxies observed .
the redder ellipticals are consistent with 12 to 17 gyr stellar population models , while the bluer galaxies in our sample must have younger 5 to 8 gyr populations .
metallicities appear to vary less than ages , with the typical galaxy having [ fe / h]=@xmath246 .
the stellar population models show that the age - metallicity degeneracy is broken with @xmath13-band sbfs , allowing one to distinguish between old , metal - poor and young , metal - rich populations . examining the radial variation in ir sbfs will help distinguish between different galaxy formation scenarios .
our observations do not agree with the relationship between and the 2 index predicted by the stellar population models .
we are especially indebted to guy worthey for providing the custom stellar population models and the @xmath13-band k corrections .
many sections of this paper were strengthened by the addition of his models .
we also thank john tonry and his collaborators for allowing us to use @xmath14-band images from their sbf survey to identify and remove globular clusters and background galaxies from our images .
we gratefully acknowledge helpful conversations with michael pahre and michael liu .
we acknowledge the support of grants nsf ast9401519 and stsci go-06579.01 - 95a . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
ajhar , e. a. & tonry , j. t. 1994 , , 429 , 557 blakeslee , j. p. 1997
, , 481 , l59 blakeslee , j. p. & tonry , j. l. 1995 , , 442 , 579 blakeslee , j. p. & tonry , j. l. 1996 , , 465 , l19 blakeslee , j. p. , tonry , j. l. , & metzger , m. r. 1997 , , 114 , 482 burstein , d. & heiles , c. 1984 , , 54 , 33 buzzoni , a. 1995 , , 98 , 69 casali , m. & hawarden , t. g. 1992 , jcmt - ukirt newsletter , 4 , 33 cohen , j. g. , frogel j. a. , persson , s. e. , & elias , j. h. 1981 , , 249 , 481 cowie , l. l. , gardner , j. p. , hu , e. m. , songaila , a. , hodapp , k .- w . , & wainscoat , r. j. 1994 , , 434 , 114 djorgovski , s. et al . , 1995 , , 438 , 13 faber , s. m. , trager , s. c. , gonzalez , j. j. , & worthey , g. 1995 , in stellar populations , i.a.u . symp . 164 , eds .
p. c. van der kruit & g. gilmore ( kluwer , dordrecht ) , p. 249
faber , s. m. , wegner , g. , burstein , d. , davies , r. l. , dressler , a. , lynden - bell , d. , & terlevich , r. j. 1989 , , 69 , 763 feast , m. w. & catchpole , r. m. 1997 , , 286 , l1 ferrarese , l. et al . 1996 , , 464 , 568 forbes , d. 1996 , , 112 , 954 freedman , w. l. & madore , b. f. 1990 , , 365 , 186 freedman , w. l. et al . 1994 , , 371 , 757 frogel , j. a. , persson , s. e. , aaronson , m. , & matthews , k. 1978 , , 220 , 75 glass , i. s. 1984 , , 211 , 461 hodapp , k .- w .
1996 , new astronomy , 1 , 177 hodapp , k .- w . ,
rayner , j. , & irwin , e. 1992 , , 104 , 441 jacoby , g. h. , branch , d. , ciardullo , r. , davies , r. l. , harris , w. e. , pierce , m. j. , pritchet , c. j. , tonry , j. l. , & welch , d. l. 1992 , , 104 , 599 jensen , j. b. , luppino , g. a. , & tonry , j. l. 1996 , , 468 , 519 ( jlt ) luppino , g. a. & tonry , j. l. 1993 , , 410 , 81 madore , b. f. & freedman , w. l. 1997 , , 492 , 110 pahre , m. a. & mould , j. r. 1994 , , 433 , 567 persson , s. e. , cohen , j. g. , sellgren , k. , mould , j. , & frogel , j. a. 1980 , 240 , 779 persson , s. e. , frogel , j. a. , & aaronson , m. 1979 , , 39 , 61 pierce , m. j. , welch , d. l. , mcclure , r. d. , van den bergh , s. , racine , r. , & stetson , p. b. 1994 , , 371 , 385 saha , a. , sandage , a. , labhardt , l , tammann , g. a. , macchetto , f. d. , & panagia , n. 1996a , , 466 , 55 saha , a. , sandage , a. , labhardt , l , tammann , g. a. , macchetto , f. d. , & panagia , n. 1996b , , 107 , 693 sandage , a. , saha , a. , tammann , g. a. , labhardt , l. , panagia , n. , & macchetto , f. d. 1996 , , 460 , l15 schechter , p. l. , mateo , m. , & saha , a. 1993 , , 105 , 1342 secker , j. & harris , w. e. 1993 , , 105 , 1358 silbermann , n. 1998 , private communication sodemann , m. & thomsen , b. 1995 , , 110 , 179 sodemann , m. & thomsen , b. 1996 , , 111 , 208 tonry , j. l. 1991 , , 373 , l1 tonry , j. l. 1997 , private communication tonry , j. l. , ajhar , e. a. , & luppino , g. a. 1990 , , 100 , 1416 tonry , j. l. , blakeslee , j. p. , ajhar , e. a. , & dressler , a. 1997 , , 475 , 399 tonry , j. l. & schneider , d. p. 1988
, , 96 , 807 wainscoat , r. , cohen , m. , volk , k. , walker , h. j. , & schwartz , d. e. 1992 , , 83 , 111 wainscoat , r. & cowie , l. 1992 , , 101 , 332 worthey , g. 1993a , , 409 , 530 worthey , g. 1993b , , 418 , 947 worthey , g. 1994 , , 95 , 107 worthey , g. 1997 , private communication worthey , g. , faber , s. m. , & gonzalez , j. j. 1992 , , 398 , 69 | surface brightness fluctuations ( sbfs ) are much brighter in the infrared than they are at optical wavelengths , making it possible to measure greater distances using ir sbfs .
we report new ( 2.1 ) sbf measurements of nine galaxies in the fornax and eridanus clusters using a 1024@xmath0-pixel hgcdte array .
we used improved analysis techniques to remove contributions to the sbfs from globular clusters and background galaxies , and we assess the relative importance of other sources of residual variance .
we applied the improved methodology to our fornax and eridanus images and to our previously published virgo cluster data .
apparent fluctuation magnitudes were used in conjunction with cepheid distances to m31 and the virgo cluster to calibrate the sbf distance scale .
we find the absolute fluctuation magnitude @xmath1 , with an intrinsic scatter to the calibration of 0.06 mag .
no statistically significant change in is detected as a function of v - i .
our calibration is consistent with simple ( constant age and metallicity ) stellar population models .
the lack of a correlation with v - i in the context of the stellar population models implies that elliptical galaxies bluer than @xmath2 have sbfs dominated by younger ( 58 gyr ) populations and metallicities comparable to redder ellipticals .
significant contributions to the sbfs from anomalous populations of asymptotic giant branch stars are apparently uncommon in giant ellipticals .
sbfs prove to be a reliable distance indicator as long as the residual variance from globular clusters and background galaxies is properly removed . also , it is important that a sufficiently high signal - to - noise ratio be achieved to allow reliable sky subtraction because residual spatial variance can bias the measurement of the sbf power spectrum .
0/p1 2mg@xmath3 2mg@xmath3 v - i -m -mbar |
we thank p. brogueira , j.g . milhano and m. pimenta for helpful discussions on auger data and strings .
j. a - m . and
c. p. thank ministerio de educacin y ciencia ( fpa 2005 - 01963 ) and spanish consolider - ingenio 2010 programme cpan ( csd2007 - 00042 ) .
j. a - m .
is also supported by the ramn y cajal " programme ; by ministerio de educacin y ciencia ( fpa 2007 - 65114 ) ; by xunta de galicia ( 2005 pxic20604pn , pgidit 06 pxib 206184 pr ) , and by feder funds , spain .
we thank cesga ( centro de supercomputacin de galicia ) for computing resources .
m. roth for the pierre auger collaboration , proceedings of the @xmath64 international cosmic ray conference , icrc 2007 , mrida , mxico , arxiv:0706.2096 ; t. yamamoto for the pierre auger collaboration , proceedings of the @xmath64 international cosmic ray conference , icrc 2007 , mrida , mxico , arxiv:0707.2638 r. ulrich , j. blmer , r. engel , f. schssler and m. unger . proceedings for the blois07/eds07 ( @xmath65 international conference on elastic and diffractive scattering ) workshop , desy , hamburg ( 2007 ) , arxiv:0709.1392 and arxiv:0712.3633 a. capella , u. sukhatme , c.i
. tan , j. tran thanh van , phys .
b * 81 * ( 1979 ) 68 ; a. capella , u.p .
sukhatme , j. tran thanh van , z. phys .
c * 3 * ( 1980 ) 329 ; a. capella , j. tran thanh van , z. phys .
c * 10 * ( 1981 ) 210 ; a. capella _
et al . _ ,
c * 33 * ( 1987 ) 1452 h. sorge , h. stoecker and w. greiner , nucl .
a * 498 * ( 1989 ) 567c ; h. sorge , a. von keitz , r. mattiello , h. stoecker and w. greiner , phys . lett .
b * 243 * ( 1990 ) 7;. a. von keitz , h. sorge , h. stoecker and w. greiner , nucl .
a * 527 * ( 1991 ) 601c .
i. arsene _ et al . _
[ brahms collaboration ] , nucl .
. a * 757 * ( 2005 ) 1 ; j. adams _ et al . _ [ star collaboration ] , nucl .
phys . a * 757 * ( 2005 ) 102 ; k. adcox _ et al . _ [ phenix collaboration ] , nucl .
a * 757 * ( 2005 ) 184 ; b. b. back _ et al . _
[ phobos collaboration ] , nucl .
phys . a * 757 * ( 2005 ) 28 .
n. s. amelin , l. v. bravina , v. l. korotkikh , l. i. sarycheva and s. y. sivoklokov , sov .
j. nucl .
* 55 * ( 1992 ) 1414 [ yad . fiz .
* 55 * ( 1992 ) 2537 ] ; n. s. amelin , e. f. staubo and l. p. csernai , nucl .
* 24b * ( 1991 ) 269 . r.c .
hwa , c.b .
yang , phys .
c * 69 * ( 2004 ) ; v. greco , c.m .
ko , p. lvai , phys .
* 90 * ( 2003 ) 202302 ; r.j .
fries , b. mller , c. nonaka , s.a .
bass , phys .
* 90 * ( 2003 ) 202303 ; l. maiani , a.d .
polosa , v. riquer , c.a .
salgado , phys .
b * 645 * ( 2007 ) 138 .
kopeliovich , n.n .
nikolaev and i.k .
potashnikova , phys . rev .
d * 39 * ( 1989 ) 769 ; a. capella , u. sukhatme , c.i . tan and j. tran thanh van , phys .
reports * 236 * ( 1994 ) 225 ; j. ranft , phys .
d * 51 * ( 1995 ) 64 ; l. durand and h. pi , phys .
* 58 * ( 1987 ) 303 ; h.j .
drescher , a. dumitru and m. strikman , hep - ph/0408073 ; s. ostapchenko , hep - ph/0501093 ; i.n .
mishustin and j.l .
kapusta , phys .
* 88 * ( 2002 ) 112501 . | we argue that the increase of the ratio baryon / meson due to the presence of strong colour fields and percolation in ultra - high energy hadronic collisions , helps to explain some of the global features of ultra - high energy cosmic ray cascades at @xmath0 ev and , in particular the observed excess in the number of muons with respect to current models of hadronic interactions .
a reasonable agreement with the small value and slope of the average depth of shower maximum @xmath1 vs shower energy as seen in data collected at the pierre auger observatory can be obtained with a fast increase of the p - air production cross - section compatible with the froissart bound .
strong color fields , percolation , high energy cosmic rays , extensive air showers , muons , elongation rate 13.85.tp , 12.40.nn , 96.40.de , 96.40.pq , 96.40.tv the largest source of uncertainty in the prediction of observables of cosmic - ray induced atmospheric showers at ultra - high energy ( uhe ) above @xmath2
ev stems from our limited knowledge of the features of hadronic interactions in this energy range .
qcd - inspired models of multi - particle production need to be extrapolated to energies far beyond those reached in terrestrial accelerators , and in regions of phase space of the collisions usually not covered in collider or accelerator experiments . on the other hand inferring the properties of the hadronic interactions at uhe in cosmic ray experiments is not an easy task which is hampered by the low luminosity of the primary beam , and by the unknown nature of its constituents .
furthermore , data are relatively scarce at such high energies .
the composition of cosmic rays at high energies is still a matter of controversy , mainly because direct measurements are only possible up to @xmath3 ev @xcite . above this energy ,
attempts to infer the mass number @xmath4 of the primary particle are based on measured shower variables and are rather indirect @xcite .
observables such as the number of muons at ground level @xmath5 , or the depth at which the maximum in shower longitudinal development occurs @xmath6 , are sensitive to both the nature of the primary and the hadronic interaction model . in particular , the muon component of the shower is a powerful tool for the validation of hadronic interaction models .
very recently an analysis of the data collected with the hybrid detector of the pierre auger observatory @xcite , indicated that the number of muons produced in uhe showers is about 1.5 times larger than that predicted by the qgsjetii @xcite model assuming proton primaries .
this conclusion seems to be rather independent of the primary cosmic ray composition .
this deficit of muons " in the hadronic interaction models is consistent with the one found in the recent analysis of more direct data on muons collected with the yakutsk array @xcite .
the pierre auger collaboration has also reported that the arrival directions of 20 out of the 27 most energetic detected events above 57 eev ( 1 eev = @xmath7 ev ) correlate with the positions in the sky of nearby active galactic nuclei ( agn ) @xcite , which are candidate sources of cosmic - ray production and acceleration to ultra - high energies .
it has been pointed out @xcite that this result together with the observed suppression of the cosmic - ray flux above 60 eev @xcite , is consistent with the hypothesis that most of the cosmic rays reaching the earth are protons from nearby astrophysical sources @xcite .
there are also measurements of the average depth of maximum shower development as a function of energy @xcite .
the measured @xmath1 is smaller than what is predicted by current hadronic models assuming a proton dominated composition , and moreover it increases with energy slower than expected @xcite . in this work under the assumption of proton primaries ,
we show that the presence of strong colour fields in uhe collisions increases the ratio of baryons to mesons ( @xmath8 ) with respect to hadronic models currently used in the simulation of atmospheric showers .
this has the effect of increasing the number of muons , following the tendency seen in data .
we also find that the behaviour of @xmath6 is not strongly affected by strong colour fields , and conclude that available data can be explained with a rapid increase of the p - air production cross section , but sill compatible with the froissart bound .
the global features of particle production in hadron - hadron ( hh ) , hadron - nucleus ( ha ) and nucleus - nucleus ( aa ) collisions seem to be well described by models where particles are produced in the decay of longitudinal strings formed in the initial stage of the interaction @xcite . however , in order to explain some effects observed at sps and rhic energies , such as the limited growth of the rapidity particle density in the central region , or the increase of the average transverse momentum and strangeness with the number of participant nucleons @xcite , one requires the introduction of fusion of strings @xcite or more generally percolation of strings @xcite .
one should mention here that the results obtained within the string percolation model coincide to a large extent with those of the colour glass condensate model @xcite .
the key idea in the present work is that fusion and percolation of strings lead to the formation of strings with higher string tension , and this implies that mass effects formation of strange quarks , di - quarks , etc ... become less important . in the schwinger model of string fragmentation
@xcite , the probability of generating a set @xmath9 of quarks and anti - quarks and a conjugate set @xmath10 out of the vacuum , each one with mass @xmath11 , is @xmath12 where @xmath13 is the string tension .
as @xmath13 tends to infinity the probability ( [ eq : prob ] ) becomes less and less dependent on @xmath11 .
the role of strong colour fields to increase the fraction of heavier particles has also been emphasized in @xcite . if we consider a single @xmath14 string , with @xmath15 being the mass of the u and d quarks ( @xmath16 gev ) , and @xmath13 the 3-representation string tension @xmath17 , and neglect strangeness and higher mass flavours , we have that the probability @xmath18 of producing a baryon is given by @xmath19 and the corresponding probability @xmath11 for meson production is @xmath20 so that the ratio @xmath8 is @xmath21 this is our reference value for the ratio @xmath8 at low energies .
we shall next consider the effect of exchanging a quark by a diquark , as in eqs .
( [ eq : probb ] ) and ( [ eq : probm ] ) , in more complex systems . in @xmath22 @xmath22 fused strings , if they are in the su(3 ) 6-representation with @xmath23 , we obtain @xmath24 which is larger than eq.([eq : ratiobm ] ) . furthermore ,
if we consider @xmath22 @xmath22 in the 8-representation with @xmath25 , one obtains @xmath26 which is again larger than eq.([eq : ratiobm ] ) . in general , if we consider @xmath27 quarks in @xmath28 representations of su(3 ) ( see for instance @xcite ) , the highest dimension representation corresponds to @xmath29 with @xmath30 , and @xmath31 with @xmath32 being a number of the order of @xmath33 such that for the quark diquark interchange we obtain , @xmath34 and , in the @xmath35 limit , @xmath36 we take this result as an indication for the increase of the ratio @xmath8 as the energy or centrality of the collision increases .
it is also worth emphasizing that the enhancement of baryons or antibaryons over mesons is not only due to a mass effect as explained above .
in fact in addition to the larger color and string tension in the percolation model , the way the cluster formed from the overlapping of individual strings decays , favours the increase of the ratio @xmath8 with energy and/or density of strings .
consider a cluster formed by several @xmath37 strings .
this cluster behaves as a @xmath38 string , where @xmath39 is composed of the different flavours of the individual @xmath37 strings .
the fragmentation of a cluster occurs through the successive creation of @xmath38 complexes @xcite .
clearly this mechanism leads to an enhancement in the production of baryons over mesons , because the large number of quarks in the cluster of strings favours the formation of particles with higher number of constituents @xcite .
coalescence and recombination models @xcite proposed similar mechanisms to explain the suppression of pion yields relative to baryons and antibaryons observed in relativistic heavy - nuclei collisions @xcite . the increase with energy of @xmath8 helps to explain the excess of muons in data when compared to existing hadronic interaction models .
in fact , increasing the number of baryons and anti - baryons more muons are produced .
this effect has been shown with a simple toy model and also in the framework of the epos model of hadronic interaction in @xcite .
increasing the number of baryons decreases the number of neutral pions @xmath40 which decay into @xmath41 and initiate electromagnetic subshowers in which muon production is typically very small .
moreover a decrease in the amount of shower energy going into @xmath40s increases the total number of hadronic interactions in the shower , in which copious muon production typically occurs .
also @xmath40s tend to stretch shower development due to the production of electromagnetic subshowers with comparatively larger @xmath6 than hadronic subshowers of the same energy . as a consequence
one would expect that decreasing the number of @xmath40s increasing the @xmath8 ratio would limit the rate of increase of @xmath6 with energy , a tendency seen in cosmic ray data .
it is important to note that an increase of the @xmath8 ratio also leads to a reduction in the number of charged pions , but as a first approximation charged pions play the same role as baryons in shower development and hence this has no consequences for either pion production or the behaviour of @xmath6 . in order to test these effects in a more quantitative way and make contact with auger data ,
we have simulated sets of 100 proton induced showers at energies ranging from @xmath42 ev up to @xmath43 ev .
the hybrid , one - dimensional shower simulation described in @xcite was used to obtain the average number of muons at auger ground level @xmath44 , and the average depth of shower maximum @xmath1 .
firstly , we used the sibyll 2.1 hadronic interaction model described in @xcite which includes a ratio @xmath45 constant with energy .
then we modified the sibyll 2.1 model and implemented a value of @xmath8 according to the following relation @xmath46 normalized to the sibyll 2.1 value at low energy @xmath47 ev where accelerator data exists . in fig .
[ fig : muons ] we show the relative increase in the number of muons when increasing the ratio @xmath8 .
the relative difference between the average number of muons at ground predicted by sibyll 2.1 with @xmath48 , and the number of muons predicted by sibyll 2.1 but with @xmath8 following eq .
( [ bm ] ) is shown as a function of shower energy .
one can see that at the highest energies the number of muons can increase by as much as @xmath49 in agreement with auger data @xcite .
it is important to remark that the choice of the underlying hadronic model is not relevant for our results , since we are only interested in the relative change of the number of muons when increasing the ratio @xmath8 .
constant with energy @xmath50 , and with the sibyll 2.1 model but with a modified ratio @xmath8 following eq.([bm ] ) @xmath51 .
the relative difference is calculated as @xmath52/\langle n_\mu({\rm sib})\rangle$ ] and is shown as a function of shower energy .
the lines joining the points are just to guide the eye . ] in fig .
[ fig : xmax ] we show the behaviour of @xmath6 with energy as obtained in proton - induced showers with the sibyll 2.1 model , and with the sibyll 2.1 model with the ratio of @xmath8 modified according to eq .
( [ bm ] ) . as explained above , increasing the ratio @xmath8 decreases the slope of @xmath53 vs @xmath54 curve and the value of @xmath53 due to the decrease in the production of @xmath40 , which would otherwise induce electromagnetic subshowers evolving deep into the atmosphere .
one can see that the decrease of @xmath53 is not large enough to explain the trend observed in the data collected at the pierre auger observatory @xcite also shown in fig .
[ fig : xmax ] . with energy as obtained in proton - initiated showers simulated with the sibyll 2.1 model with ratio @xmath48 constant with energy .
dashed line with empty squares : @xmath53 with the sibyll 2.1 model but with a modified ratio @xmath8 following eq.([bm ] ) . dotted line with empty circles : @xmath53 predicted by the sibyll 2.1 model but with a modified ratio @xmath8 following eq .
( [ bm ] ) , using the inelasticity @xmath55 predicted by the percolation model and with the p - air and @xmath56air cross section modified according to eq .
( [ xsection ] ) .
the filled squared points without lines are the data on @xmath53 collected at the pierre auger observatory @xcite . ]
another important consequence of the presence of strong colour fields in the region where the interaction occurs is the percolation of strings .
more details on the percolation model can be found in @xcite .
of special importance for shower development is the prediction of the percolation model on the inelasticity @xmath55 , defined as one minus the fraction of momentum carried by the fastest ( leading ) particle .
essentially , all existing high energy strong interaction models based on qcd and qcd evolution , predict an increase with energy of the inelasticity @xcite .
the same is true for the hadronic generators sibyll @xcite and qgsjet @xcite , widely used in the analysis of cosmic ray data .
however as shown in @xcite , above the percolation threshold the inelasticity decreases with energy .
this has important consequences for shower development as studied in @xcite , among them a decrease of the inelasticity tends to propagate the primary energy deeper into the atmosphere and increase @xmath6 in contrast to what is seen in auger data .
note that the observed decrease of @xmath53 with energy @xcite is at odds with the reported reduction of @xmath4 at high energies @xcite as we shall see next , the behaviour of @xmath6 with energy may be accommodated with an increase of the p - air cross section compatible with the froissart bound .
we have implemented the inelasticity predicted by the percolation model @xcite in the sibyll 2.1 hadronic generator . also we decreased the position of the first interaction and subsequent hadronic collisions by increasing the p - air and @xmath57-air production cross section in a similar manner as was done in @xcite .
in particular we have changed the sibyll 2.1 energy dependence of the production cross section @xmath58 using the relation @xmath59 which we apply for @xmath60 ev , so that at lower energies one reproduces accelerator data . moreover , since the sibyll cross section behaves as @xmath61 the modified cross section in eq .
( [ xsection ] ) clearly behaves as @xmath62 , saturating the froissart bound but not violating it .
the p - air cross section in eq .
( [ xsection ] ) and the sibyll 2.1 p - air cross section are shown in fig .
[ fig : xsection ] . ) .
note that the sibyll cross section behaves as @xmath61 , while the modified cross section behaves as @xmath62 , saturating the froissart bound but not violating it . ]
our results for @xmath1 are shown in fig .
[ fig : xmax ] where a fair agreement with auger data can be seen .
this is mainly due to the changes implemented in the sibyll 2.1 cross - section , while modifying the inelasticity and the @xmath8 ratio according to what strong colour fields induce , affects @xmath53 much less .
it is also interesting to note that an increase of the p - air cross section also induces a decrease of the rms of the @xmath6 distribution with respect to what current models of hadronic interaction predict for proton - initiated showers .
also we have checked in our shower simulations , that the increase in the number of muons is fairly insensitive to both the decrease of inelasticity due to percolation of strings , and to the increase of the p - air and @xmath56air cross - section , the dominant effect being the increase of the ratio @xmath8 .
we have shown that strong colour fields associated to high dimension representations of @xmath38 strings , together with the @xmath38 string breaking mechanism , lead to the increase of the @xmath8 ratio with respect to current hadronic models in which the @xmath8 ratio is @xmath63 and constant with energy .
for uhe cosmic ray - induced showers this results in an increase of the average number of muons in the shower at ground level , and in a ( small ) decrease of @xmath53 . in order to further decrease @xmath53 and approximately agree with data , a large , but still compatible with the froissart bound increase of the p - air and @xmath57-air cross sections with energy
is required . |
over the last several years event - by - event fluctuations of many observables have been studied in relativistic heavy ion collisions @xcite .
the measurement of these fluctuations may reveal separate event classes , fluctuations associated with phase - transitions etc .
( for a recent review see @xcite ) .
the first measurement of event - by - event hadron ratio fluctuations has been carried out by the na49 collaboration , which analyzed the fluctuations of the kaon to pion ( @xmath1 ratio in lead - lead collisions at a center of mass energy of @xmath2 @xcite .
this measurement found surprisingly little fluctuations , of the order of a few percent , which could be explained by a combination of bose - einstein correlations together with resonance decays @xcite .
subsequently , the na49 collaboration has measured event - by - event ratio fluctuations for several values of the center of mass energy , ranging from @xmath3 to @xmath4 @xcite .
in addition , at the relativistic heavy ion collider ( rhic ) the star , phenix , and phobos collaborations have measured fluctuations up to center of mass energies of @xmath5 so that excitation functions of many fluctuation observables are now available over a wide range of energies . while most excitation functions show only little energy dependence , that for the kaon to pion ratio exhibits a steep increase towards the lower energies .
so far this increase could not be reproduced in either thermal model calculations @xcite nor with the microscopic transport model urqmd @xcite and thus has sparked quite some interest and speculations concerning the qcd phase transition @xcite .
another calculation using the hsd event - generator @xcite can describe the general increase of the fluctuations towards lower energies but fails to reproduce the very steep rise exhibited in the na49 data . in this note , we will point out that the observable generally used in the discussion of dynamical fluctuations,@xmath6 has an inherent dependence on the multiplicity of particles that are used in the experimental analysis - these are the particles located within the phase space domain covered by experimental acceptance with particle identification capability .
this has some nontrivial consequences for excitation functions extracted with fixed target experiments , such as na49 , where the acceptance changes considerably with beam energy .
specifically , in case of @xmath0 - ratio fluctuations , the dynamical fluctuations depend on the inverse of the _ accepted _ number of pions and kaons , and since their number decreases with beam energy this may lead to non - negligible corrections , as we shall show . indeed in @xcite a significant centrality dependence of these ratio fluctuations was found at top rhic energies , consistent with an inverse multiplicity scaling .
in addition for a fixed target experiment the actual acceptance changes with beam energy . since it is the multiplicity of the _ accepted _ particles which determines the fluctuations , this needs to be taken into account if one studies an excitation function .
it is the purpose of this note to discuss the multiplicity dependence of particle ratio fluctuations and in particular those of the kaon - to - pion ratio . after a brief review of the underlying formalism governing fluctuations of ( particle ) ratios
, we will discuss several ways to remove the inherent multiplicity dependence in the definition of the dynamical fluctuations , @xmath7 , eq .
[ eq : sig_dynamic ] .
we then apply them to the fluctuations of the kaon - to - pion ratio and discuss the energy dependence of this observable in this context .
following @xcite the fluctuations of a particle ratio @xmath8 are given to leading order by@xmath9 with the definition@xmath10 since@xmath11 eq .
[ eq : variance ] can also be written as@xmath12 in absence of any correlation , @xmath13 and @xmath14 , and the scaled variance ( eq . [ eq : variance ] ) reduces to@xmath15 the difference between the scaled variance @xmath16 and the uncorrelated scaled variance @xmath17 is commonly referred to as the dynamical fluctuations , @xmath7,@xmath18 where @xmath19 is the variable usually used by the star collaboration . obviously in the absence of any correlations , @xmath20 , by construction . introducing the scaled correlations@xmath21 the dynamical fluctuations , @xmath7 ,
can be written as@xmath22 we note that the scaled correlations , @xmath23 , typically do not or only weakly depend on the multiplicity as we will show explicitly in the context of a resonance gas . consequently , eq . [ eq : sig_dyn_scal ] already suggests , that a simple scaling of @xmath7 with number of charged particles may not be sufficient . depending on which of the scaled correlations dominates ,
@xmath7 may either scale with @xmath24 , @xmath25 or with @xmath26 or some combination of those .
furthermore , if the particle abundances differ considerably , say @xmath27 , as it is the case for the kaon to pion ratio , the dynamical fluctuations will be dominated by the least abundant particle , even if the scaled correlations are of the same magnitude .
this follows directly from eq .
[ eq : sig_dyn_scal ] . in this case
a scaling with @xmath24 should work best .
finally , quantum statistics gives rise to additional correlations @xcite @xmath28 with@xmath29 ^ 2\ ] ] where @xmath30 stands for bosons and @xmath31 for fermions and @xmath32 is the number of particles of type @xmath33 in momentum bin @xmath34 .
the correction term due to quantum statistics is typically of the order of a @xmath35 for systems of consideration , resulting in @xmath36 effects for the dynamical fluctuations @xcite .
after these general remarks about fluctuations of particle ratios and their scaling with multiplicity let us turn to the specific case of kaon - to - pion ratio fluctuations .
let us now turn to the specific case of @xmath0 fluctuations . in this case the scaled variance , eq .
[ eq : variance ] , is given by @xmath37 and the dynamical fluctuations , eq . [ eq : sig_dyn_scal ] , are @xmath38 since @xmath39 and @xmath40 @xmath41 so that the scaled correlations , eq .
[ eq : scaled_corr ] , are given by @xmath42 we note , that the `` diagonal '' scaled correlations , @xmath43 and @xmath44 , also contain cross - correlations between the positively and negatively charged kaons and pions respectively .
therefore , correlations introduced by resonances such as the @xmath45-meson and the @xmath46-meson will _ enhance _ the `` diagonal '' scaled correlations and thus the dynamical fluctuations .
resonances decaying into a kaon and a pion , such as the @xmath47-meson will contribute to the off - diagonal scaled correlation , @xmath48 and will reduce the dynamical fluctuations .
this will be different if charge specific ratios such as @xmath49 are considered . in this case only resonances which decay in either two @xmath50-mesons or two @xmath51-mesons will contribute to the diagonal terms , while the @xmath47-mesons do contribute to the off - diagonal scaled correlation .
consequently , the fluctuations of the charge specific ratio is not necessarily the same as that of the ratio of sums of negative and positive kaons / pions .
this is also seen in the data by the star collaboration @xcite .
it may be instructive to discuss the above equations in the context of a simple model system which contains @xmath52 charged pions and @xmath53 charged kaons as well as @xmath54 _ neutral _ rho - mesons , @xmath55 omega - mesons , @xmath56 phi - mesons , and @xmath57 neutral @xmath58 and @xmath59 @xmath60-mesons . via their decay channels
these resonances will give rise to the various correlation terms in eq .
[ eq : scal_corr_kp ] . in reality
, of course , there are many other resonances contributing to @xmath61 and to the scaled correlations .
the ones chosen here are the lightest and most abundant ones which lead to correlations and thus should provide a rough idea on how the different terms contribute -mesons . while they contribute via the kaon and pion multiplicities , they do not introduce any correlations , as their decay channels involve always a neutral kaon or pion . ] .
a detailed investigation would involve a full study in the hadron resonance gas model , which is not the purpose of this paper .
in addition to correlations due to resonances there are effects due to quantum statistics @xcite , as already discussed . here , for simplicity we work with classical ( boltzmann ) statistics .
noting that the branching ratio @xmath62 and @xmath63 the average particle numbers in our simple model are given by@xmath64 and we obtain the following expression @xcite for the ( co)-variances in eq .
[ eq : numerators ] @xmath65 leading to ( see eq .
[ eq : numerators ] ) @xmath66 the corresponding scaled correlations , eq .
[ eq : scaled_corr ] , are@xmath67 we find , that the scaled correlations in our simple model indeed depend only weakly ( if at all ) on the multiplicity , since both the number of resonances and the number of kaons and pions are expected to scale roughly with the multiplicity or volume of the system . in a thermal system at fixed temperature
they would be constant .
furthermore , as already discussed the correlations introduced via the resonances affect all three scaled correlations . while the @xmath47-mesons control the off - diagonal correlation term , @xmath68 , both the @xmath45-meson and the @xmath46 and @xmath69 contribute to the diagonal parts .
the former reduce the dynamical fluctuations while the latter increase them . putting everything together ,
the scaled variance , eq . [ eq : variance ] , is given by@xmath70 leading to @xmath71 or in terms of the scaled correlations@xmath72 evaluating scaled correlations for our simple model , eq .
[ eq : scaled_corr_model ] , for a temperature of @xmath73 and vanishing chemical potential , we get @xmath74 obviously , in our simple model all the scaled correlations are of the same order of magnitude .
in addition , adding more resonances will likely reduce @xmath44 as there are many resonances decaying into only one pion , which add to the denominator , @xmath75 , but not the numerator of @xmath44 .
while it would be worthwhile to study these scaled correlation coefficients in a full hadron gas model , here instead we want to concentrate on simple phenomenological scaling rules , with a special emphasis on the effect of a varying acceptance .
in this section we want to discuss several ways to scale out the multiplicity dependence of the dynamical fluctuations . of course , if all the relevant scaled correlations are known and if they , as we argued , depend only weakly on the multiplicity and beam energy , the appropriate scaling is simply given by eq .
[ eq : sig_dyn_scaled_corr ] .
this is equivalent to having full understanding of all the sources for the fluctuations in which case this discussion is mute . in general , however , we do not have a full understanding of all the sources contributing to the fluctuations . in this case the multiplicity / energy dependence may provide additional information such as signals for a possible phase transition etc . in order to extract this information , one needs to understand `` trivial '' dependencies on the multiplicity , as exhibited e.g. in eq .
[ eq : sig_dyn_scaled_corr ] . in the following
we will discuss several `` trivial '' scaling prescriptions which we believe should be applied before conclusions about new phenomena can be drawn .
we will focus on the rather interesting energy dependence of the @xmath0-fluctuations , referring to @xcite for a discussion on the centrality dependence .
in addition to applying appropriate scaling prescriptions to the dynamical fluctuations , it is essential to realize that the multiplicities which control the fluctuations , such as @xmath76 and @xmath75 are that of the _ identified _ particles and _ not _ the extrapolated total multiplicities or mid - rapidity multiplicities . this is especially important when studying the energy dependence of fixed target data , as the acceptance varies with beam energy .
we will discuss this point in detail below after we have introduced the various scaling prescriptions .
as suggested in @xcite , one way to avoid any scaling of the dynamical fluctuations with the number of accepted particles would be to take the ratio of the measured scaled variance @xmath16 over that of mixed events @xmath77 , @xmath78 instead of the difference as it is done in the definition of @xmath7 , eq .
[ eq : sig_dynamic ] .
this , however , has the disadvantage , that correlations and fluctuations due to the the detector do not cancel out . in order to remove the multiplicity dependence in the same fashion
, one can simply divide the dynamical fluctuations , @xmath7 , by that of uncorrelated particles , @xmath17 , eq .
[ eq : variance_uncorr ] , evaluated for the same number of particles .
@xmath79 with @xmath80 for the case at hand . this scaling we will subsequently denote as `` poisson '' scaling , as we scale with the scaled variance of a poisson distribution based on the observed multiplicities .
the advantage of this scaling is that it is unbiased in the sense that one does not need to make any assumptions about the relative magnitude of the scaled correlations .
another unbiased scaling would be to scale with the number of particles involved , i.e. @xmath81 .
this is similar in spirit of @xcite where a scaling with the number of charged particles was studied .
we shall henceforth refer to this as particle number scaling , @xmath82 finally the expression for @xmath7 in terms of the scaled correlations , eq .
[ eq : sig_dyn_scaled_corr ] , suggests the scaling with the kaon or pion number or with the geometric mean of both , depending on which of the scaled correlations dominates .
@xmath83 since the number of kaons is much smaller than the number of pions , at least for the lower energies the kaon - number scaling is similar to the poisson scaling , @xmath84 for @xmath85 . alternatively
the above scaling relations allow to relate the dynamical fluctuations at a given center of mass energy with those at another energy .
specifically , given the dynamical fluctuations at top rhic energies , @xmath5 , the dynamical fluctuations at any other center of mass energy is given by * _ _ poisson scaling:__@xmath86 * particle number scaling@xmath87 * @xmath88__-scaling : _ _ @xmath89 * @xmath90__-scaling:__@xmath91 * _ _ geometric scaling:__@xmath92 the resulting _ scaled _ dynamical fluctuations are shown in fig . [
fig : star_scale ] where we find that the energy dependence of @xmath93 is reasonably reproduced by any of the scaling rules discussed above .
the values for the multiplicities of identified particles taken from @xcite as well as the result for the rescaled fluctuations according to eqs .
( [ eq : poisson_scaling]-[eq : geom_scaling ] ) . the essential point for the success of these scaling rules
is that we have used the number of _ identified _ kaons and pions for the mean values , @xmath76 and @xmath75 , entering the scaling formulae , eqs .
[ eq : poisson_scaling]-[eq : geom_scaling ] .
this leads to an additional energy dependence especially for the na49 data .
since na49 is a fixed target experiment , the actual acceptance and thus the fraction of identified particles of the total number of particles may vary considerably with the beam energy .
this is illustrated in fig .
[ fig : star_scale_dndy ] , where we applied the same scaling formulae , eqs .
[ eq : poisson_scaling]-[eq : geom_scaling ] , but used the mid - rapidity yields , @xmath94 for the respective mean particle numbers , @xmath76 and @xmath75 @xcite .
obviously the energy dependence is not reproduced , especially for the highest sps energies , @xmath95 .
-[eq : geom_scaling ] based on the @xmath5 data from star , scaledwidth=70.0% ] looking more closely at fig .
[ fig : star_scale ] , we see that the star data at @xmath96 , which have a rather small error - bar are not well reproduced .
instead of arguing for new physics in this energy regime , we note that in ref .
@xcite the two most central values for @xmath97 at this energy do not agree very well with the systematics developed by the star collaboration either . as already discussed in section [ sec : fluctuations - of - particle ] , since @xmath98 the dynamical fluctuations are dominated by the kaons . therefore , the kaon - number or poisson scaling should work better than pion - number or particle number scaling .
this is indeed the case . instead of the actual number of identified particles contributing to @xmath97 .
otherwise same scaling formulae are used as in fig . [
fig : star_scale ] . , scaledwidth=70.0% ] one may be tempted to use the quality of agreement of the various scaling prescriptions to draw conclusions about the importance of the various contributions to the dynamical fluctuations .
given the experimental error - bars and the quality of agreement of the various scaling rules the value of such an exercise is not obvious .
at least , a global fit based on the various scaling prescriptions needs to be carried out before any more detailed conclusions about the strength of the various contributions can be drawn .
finally let us comment on the fact that calculations based on the urqmd model do not reproduce the rise of the dynamical fluctuations as observed by na49 , although the na49 acceptance has been applied @xcite .
first of all , while urqmd seems to do reasonably well for mid - rapidity abundances , it is not clear if urqmd does reproduce the correct multiplicity of identified particles _ within _ the acceptance , especially at small center of mass energies .
second , it is not at all obvious , if urqmd contains indeed all the relevant sources for fluctuations and correlations .
for instance quantum statistics is not taken into account .
the fact that the above scaling relations are able to connect the dynamical fluctuations over a wide range of beam energies does not say anything about the nature and origin of these fluctuations except that they seem to be the same at all energies . while this considerably weakens the argument for the @xmath0-fluctuations being a signature for the qcd critical point , the disagreement with urqmd may very well point to a yet to be discovered new source for fluctuations / correlations .
this however , requires that all `` trivial '' sources , such as e.g. quantum statistics , are systematically taken into account .
finally , let us conclude this section by proposing an observable , which should , to leading order , be independent of the multiplicity .
as already discussed at the beginning of this section , the ratio @xmath99 proposed in @xcite has the disadvantage that correlations and fluctuations due to the detector do not cancel out .
we , therefore , propose to study instead the ratio , eq .
[ eq : possion ] @xmath100 or equivalently @xmath101 , which has the same limit as @xmath102 in the absence of correlations . in fig .
[ fig : f_poisson+1 ] we show @xmath101 for both the @xmath0 fluctuations as well as for the @xmath103-fluctuations , which have been measured by na49 as well .
we see a rather weak energy dependence of the @xmath0-fluctuations , whereas the @xmath103-fluctuations exhibit a variation with energy .
we note , however , that the energy dependence of the @xmath103-fluctuations in the scaled variable is opposite to that in @xmath104 while @xmath7 decreases with energy , the magnitude of @xmath105 increases , suggesting in increase of the the strength of the correlations with energy , which seems to level off at top sps energies .
this suggests that the correlations , mostly due to baryon resonances , increase with energy up to @xmath106 . using the chemical freeze - out parameters of ref .
@xcite and assuming that only the delta resonance contributes to the scaled correlation coefficient @xmath107 we find indeed an increase of the correlations from ags up to sps energies .
for @xmath0 ( black ) and @xmath103-fluctuations ( red).,scaledwidth=70.0% ] .table of the results for the various scaling scenarios eqs .
( [ eq : poisson_scaling]-[eq : geom_scaling ] ) using the star data at @xmath5 as reference . the data for identified multiplicities , @xmath108 and @xmath109 , are from @xcite . [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ]
in this article we have reviewed the multiplicity dependence of particle ratio fluctuations .
we have provided several scaling prescriptions which correct for the inherent dependence of the dynamical fluctuations @xmath7 on the number of _ identified _ particles .
we have demonstrated that these scaling rules naturally reproduce the trend seen in the energy dependence of the kaon - to - pion fluctuations .
consequently any interpretation of the rise of @xmath93 towards lower energies in the context of a possible qcd critical point , needs to account for the `` trivial '' effect due to the multiplicity dependence .
we propose that future studies of energy and / or multiplicity dependencies should correct for the `` trivial '' multiplicity dependencies inherent in @xmath110 . in our view , the simplest and least biased scaling is the poisson scaling , @xmath105 , which could serve as a benchmark .
of course a more complete analysis of the energy dependence would include a thorough global fit of the available data , which we have not carried out in this paper .
we have further applied the scaling to the @xmath103-fluctuations measured by the na49 collaboration .
we find that the properly scaled observable still exhibits a strong energy dependence , which , however , is opposite to the unscaled one . in this context
it would be interesting to include the star data , which unfortunately are not available in a final version yet .
finally we point out that any multiplicity scaling needs to be based on the mean multiplicity of the actual identified particles used for the fluctuation measurement instead of an extrapolated multiplicity such as @xmath111 .
the authors would like to thank g. westfall and r. stock for useful discussions .
this work was initiated during the program `` the qcd critical point '' at the institute for nuclear theory ( int ) in seattle , wa .
the authors would like to thank the int for the hospitality during this program .
this work is supported by the director , office of energy research , office of high energy and nuclear physics , divisions of nuclear physics , of the u.s .
department of energy under contract no .
de - ac02 - 05ch11231 .
the authors also acknowledge support by the deutsche forschungsgemeinschaft ( dfg ) and the helmholtz research school on quark matter studies . | in this note we will discuss the energy dependence of particle ratio fluctuations in heavy ion collisions .
we study how the inherent multiplicity dependence of ratio fluctuations is reflected in the excitation function of the dynamical fluctuations .
specifically , we will show that the observed excitation function of dynamical @xmath0 fluctuations is consistent with the expected dependence on the number of accepted pions and kaons in both the star and na49 experiments . |
suspensions of hard spherical colloids have been extensively studied theoretically as well as experimentally as a model system for simple liquids in and out - of equilibrium @xcite . however , even weak anisotropies of the particle shape can have dramatic effects on the structure , dynamics , and phase behavior of colloidal suspensions while the details of their molecular origins are not well understood .
particularly interesting is the question at which elongation what dynamic and structural properties start to deviate from the limiting hard sphere reference case . to address such a fundamental question , well - defined experimental and theoretical model systems are in need .
one of the simplest realizations of mildly anisotropic colloids are dumbbell - shaped particles ( that is , a dimer of two fused colloidal spheres ) which can be synthesized nowadays by different routes in well - controlled and monodisperse ways @xcite .
the phase diagram of hard - core dumbbells is theoretically known , where , at high packing fractions , the colloids can form the so - called plastic crystal phase @xcite . here , the translational degrees of freedom are essentially frozen at high packings as in the hard sphere system but the particles are free to rotate .
experimental realizations of dumbbell suspensions have been shown to fit well into the respective regions of the theoretical phase diagram @xcite .
thanks to these recent advances , viable , purely repulsive monodisperse model systems are experimentally available which crystallize in equilibrium conditions without the help of external fields or long - range interactions .
colloidal dumbbells are thus among the most promising experimental model systems for a fundamental understanding of structural and dynamic effects of increasing the elongation with respect to the hard sphere reference case , in particular in the ( plastic ) crystal phase where new phenomena are expected @xcite . in this work
, we systematically study the structure and fluctuations of suspensions of steeply repulsive dumbbell particles in equilibrium by means of brownian dynamics ( bd ) computer simulations .
we focus on the dense regime of the to transition region of the phase diagram which is well accessible for experimental dumbbell suspensions and promises interesting insight into the influence of elongation on steric correlations and structural properties in both fluid and crystal . apart from simple spatial correlation functions
, we also systematically explore the dynamic properties in the linear response regime at high volume fractions measured by equilibrium fluctuations .
all results for the elongated dumbbells are discussed with respect to the well - explored hard sphere reference system .
in our model a dumbbell particle consists of two spherical beads of diameter @xmath2 rigidly constrained at a center - to - center distance @xmath3 , see fig .
[ fig : model_sketch ] . with
that we define a dimensionless elongation , or , aspect ratio , as @xmath4 . for @xmath5
, we recover the reference case of one spherical colloid . .
] the site - site ( bead - bead ) pair interaction is modeled by a yukawa ( or screened coulomb ) potential which interpolates between the long - ranged electrostatic coulomb and the purely excluded volume interaction .
the potential between two beads at distance @xmath6 is given by @xmath7 the parameter @xmath8 is a screening constant , which may be used to tune the softness of the pair interaction . in the limit of @xmath9 one obtains the purely hard - core interaction .
we choose @xmath10 as a compromise between a steep hard - core like repulsion at the same time being smooth enough to not cause numerical problems in the integration .
the parameters @xmath11 and @xmath12 define the length and energy scale of the interaction in our model , respectively .
since we compare our results to the hard - dumbbell and hard - sphere reference system , we define the effective bead diameter @xmath2 via the barker - henderson relation @xcite @xmath13 the effective volume fraction @xmath14 at dumbbell number density @xmath15 is defined accordingly .
the function @xmath16 is a geometric factor accounting for the bead overlap volume @xcite .
the total instantaneous force @xmath17 exerted by particle @xmath18 on particle @xmath19 is the sum of the forces @xmath20 acting between the @xmath21-th site of the @xmath18-th particle and the @xmath22-th site of the @xmath18-th particle , via @xmath23 the vector of the @xmath19-th particle is decomposed with respect to the axial symmetry as @xmath24 { \mathrm{\mathbf{\omega } } } _ i(t ) , \end{aligned}\ ] ] where @xmath25 is the unit vector pointing in the direction of the connection between the centers of the beads belonging to the @xmath19-th particle .
the total force @xmath26 is decomposed analogously : the parallel and perpendicular parts are given by @xmath27{\mathrm{\mathbf{\omega } } } _ i(t ) .
\end{aligned}\ ] ] given this framework , we implement the algorithm for the above defined coordinates as follows : the particle positions are updated using an explicit forward euler method with the finite time step @xmath28 .
starting from the configuration @xmath29 at time @xmath30 , we find new coordinates @xmath31 at @xmath32 following the scheme @xmath33 where @xmath34 and @xmath35 are unit vectors oriented perpendicular to the director @xmath36 .
the random variates @xmath37 have zero mean and the variances @xmath38 and @xmath39 , respectively .
the torque acting on symmetric dumbbells is given by @xmath40 , where @xmath41 is the total instantaneous force on the @xmath21-th bead of the @xmath19-th particle .
the internal director of the @xmath19-th particle is updated according to @xmath42 where @xmath43 and @xmath44 are zero mean gaussian distributed random variates with variance @xmath45 .
the single particle mobility is given by parallel @xmath46 , perpendicular @xmath47 and rotational @xmath48 diffusion coefficients
. we neglect hydrodynamic interactions between the colloids which is justified for the high packing fractions where steric correlations dominate the equilibrium structure and linear response behavior @xcite .
the infinite - dilution ( single - particle ) self - diffusion coefficients of a dumbbell particle were calculated previously using the @xcite . in this method ,
the particle surface is represented by a number of mini - beads which act as sources of hydrodynamic friction , where no - slip boundary conditions are assumed .
the mobility tensors are calculated on the oseen - level using an increasing number of mini - beads .
using these tensors , the values are extrapolated to zero mini - bead size , e.g. , infinite number of friction sources on the particle surface .
we are using the parallel @xmath49 , perpendicular @xmath50 and rotational @xmath51 diffusion coefficients obtained by the above method as input for our simulations .
the free single - sphere @xmath52 translational @xmath53 and rotational @xmath54 diffusivities may be obtained from the respective stokes - einstein relations .
the translational diffusion coefficient of a free dumbbell is given by @xmath55 .
the brownian time is accordingly given by @xmath56 @xcite . in our simulations ,
the systems are subject to in all three cartesian dimensions .
the presented results have been obtained from simulation runs with @xmath57 particles in a constant volume @xmath58 , so that the number density @xmath59 .
the total simulation time is @xmath60 with the time step of @xmath61 .
the ordered with directors pointing in the @xmath62 directions have been set as initial configuration . to ensure equilibrium conditions results
have only been obtained for @xmath63 .
all statistical correlations have been calculated for @xmath64 using every @xmath65-th time step . )
indicate the state points investigated in this work , according to a barker - henderson mapping ( [ eq : diameter ] ) of the site - site yukawa pair potential ( [ eq : potential ] ) onto hard spheres . ] the expansion of the full radial distribution function @xcite yields two partial correlation functions of particular interest : we calculate the radial distribution function of the @xmath66 and the orientational radial distribution function @xmath67 where @xmath68 denotes the separation vector of the @xmath19-th and @xmath18-th particles , and @xmath69 is the angle between the respective directors .
the orientational distribution function @xmath70 describes the spatial correlation of the particle directors with @xmath71 denoting the second legendre polynomial .
the function @xmath70 equals @xmath72 for parallel aligned particles , @xmath73 for perpendicular alignment , and vanishes for uncorrelated orientations .
we calculate the translational diffusion in the linear response regime from green - kubo - type of relations . in the limit of linear response
, we thus obtain the long - time self - diffusion coefficient @xmath74 from the time correlation functions in equilibrium from the integral of the as @xmath75 where @xmath76 is the instantaneous velocity .
the long - time limit of this function is the long - time self - diffusion coefficient @xmath77 the linear response orientational relaxation is quantified in terms of the defined as @xmath78 we assume that the particle orientations become diffusive for long times with the constant @xmath79 , therefore asymptotically the following relation holds @xcite : @xmath80 in the limit of zero strain , we may obtain the dynamic viscosity of the suspension from equilibrium fluctuations . in particular
, we calculate the relative dynamic viscosity difference from the off - diagonal @xmath81 where the symmetric stress tensor is defined as @xmath82 and @xmath83 denote cartesian components . from our simulations ,
the potential part of the stress tensor is available , therefore ( [ eq : stress ] ) does not contain a momentum part .
the steady shear viscosity difference is obtained from the integral of the as @xmath84
in fig . [ fig : rdf_g_com_phi0.55_0.40 ] the of dumbbell suspensions at the volume fraction @xmath85 are displayed for various elongations .
the signature of the phase transition from to may be readily observed here . at state points with @xmath86
the system is in the phase , cf .
[ fig : marechal2008_phasediagram ] , where the positions are frozen in a crystal lattice and the directors do not show any long - distance correlations . hence , the corresponding show strong correlations and long - ranged oscillations .
they vanish at @xmath87 , where the system is in the isotropic fluid phase .
the inset of fig .
[ fig : rdf_g_com_phi0.55_0.40 ] shows the @xmath88 at the fixed colloidal packing fraction @xmath89 .
all shown curves are in the phase . with increasing elongation
, the extrema get less pronounced and are shifted to greater distances due to the larger effective particle sizes . for the dumbbell center - of - mass at a volume fraction @xmath85 : black , blue , and purple are in the , and red in the phases , respectively .
inset : the same for a packing fraction @xmath89 , where all systems are in the phase . ] the orientational @xmath90 at the volume fractions @xmath85 and @xmath89 are depicted in fig .
[ fig : rdf_gp2_phi0.55_0.40 ] and its inset , respectively , for various representative elongations .
as expected , the orientational correlation in space is essentially flat for suspensions of nearly spherical particles ( @xmath91 ) at all volume fractions . at roughly @xmath92 , the @xmath90 for non - zero elongations
show maxima indicating a strong correlation of nearest neighbors at contact .
this first correlation peak moves slightly away from @xmath93 as the elongation increases .
at the first maxima , the functions are positive , representing a preferably parallel orientation of dumbbell particles at close contact . for a bit larger distances , @xmath94 , negative dips indicating orientational anti - correlations
are observed which tend to higher distances as the elongation is increased .
hence , in the phase at @xmath85 , there are non - vanishing next neighbor correlations which grow with elongation . with further increasing aspect ratio ,
the orientational correlation becomes non - zero over a higher distance as the system crosses to the dense fluid phase at @xmath89 .
here , we observe a new correlation peak at in the dense fluid state for @xmath87 close to @xmath95 and 2.8 , indicating growing second and third neighbor correlations .
thus , in the fluid phase the correlations are longer ranged due to the higher disorder and collisions when compared to the phase . at a volume fraction @xmath85 : the black , blue , and purple curves present data for the systems in the phase and red in the phase , respectively .
inset : the same for a packing fraction @xmath89 where all systems are in the phase . ] from the at ( a ) @xmath96 and ( b ) @xmath87 for different packing fractions .
the dashed line in ( b ) is at state point @xmath97 for a direct comparison . ]
figure [ fig : diffoftime_vacf_lstar0.20_lstar0.40 ] a ) and b ) show the time - dependent diffusion coefficients @xmath98 at constant elongation @xmath96 and @xmath87 for different colloidal volume fractions , respectively . at early times ( @xmath99 ) , before the particles feel the interacting neighbors in a non - dilute suspension , the diffusion coefficients are close to the short - time limits of a single , free particle . within times
@xmath100 , a continuous transition to long - time diffusion starts which is reached at about the brownian time scale @xmath101 . with increasing volume fraction @xmath102 , the cross - over from short - time to long - time behavior sets in earlier and
is steeper . at the elongation @xmath96 the systems with packing fractions @xmath103 and @xmath104 are in the phase . due to the translational constraints in the crystal , the diffusion vanishes at about a tenth of a brownian time .
the comparison of the diffusion between @xmath96 and @xmath87 at @xmath89 in the dense fluid in panel ( a ) demonstrates that the influence of the aspect ratio is very small in the range of focus of this study .
since all the shown curves for the elongation @xmath87 are in the phase ( panel ( b ) ) , the diffusion coefficients are not vanishing for all volume fractions .
obtained from the via relation ( [ eq : diffoftime_vacf ] ) . ] in fig .
[ fig : dinfvslstar_vacf ] the values of the long - time self - diffusion coefficients @xmath105 are presented for various elongations and packing fractions . as expected ,
the diffusion is significantly decreased for larger packer fractions , as known for the spherical reference case .
surprisingly , the dependence of the diffusion on elongation is weak and we find that this is the case for both parallel and perpendicular long - time diffusion as well .
apparently , the anisotropy is not large enough to substantially change the long - time ( single colloid ) friction in these systems .
furthermore , the data shows a discontinuous transition to vanishing diffusion at elongations where the transition to the phase takes place , see the curves for the large packing fractions @xmath106 .
[ fig : drotinfvsphi_dacf ] shows the rotational diffusion coefficients obtained from exponential fits to the . at small elongations @xmath107
the rotational diffusion is almost constant with respect to volume fraction , albeit the system crosses from the to the phase at high fractions .
hence , for these small elongations the particles indeed rotate almost freely in the phase .
the obvious statistical outliers in the rotational diffusion data for small packing fractions @xmath108 are within the statistical error of @xmath109 , estimated from block averages over independent trajectories of length of @xmath110 .
however , suspensions of dumbbells with @xmath111 show a notable drop of the rotational diffusion coefficients for densities higher than @xmath112 and packing effects clearly influence the rotation correlations in the phase . on increasing the elongation further , now a clear non - linear density - dependence of
the rotational relaxation emerges which becomes obvious for @xmath87 for packings larger than @xmath113 . here , the system is in the fluid phase only and the closer contacts in the disordered systems alter the rotational dynamics substantially . at the highest investigated packing fraction ( @xmath85 )
the rotational diffusion drops down by more than @xmath114 . at various elongations from exponential fits to on volume fraction @xmath102 .
the statistical error of the data is estimated to be @xmath115 , calculated from block averages over independent trajectories of length of @xmath110 . ]
figure [ fig : sacf_mult_lstar ] shows the stress autocorrelation function at various aspect ratios @xmath116 . for the smaller packing fractions @xmath117
, the correlation functions decay slower for higher densities while the elongation dependence is rather weak .
however , at the volume fraction @xmath85 the system is in the phase for the shown elongations , except for @xmath87 . here
an interesting behavior can be observed : at @xmath91 the decays more rapidly in the state point than in the fluid states .
however , for the next two larger elongations @xmath96 and @xmath111 , the decays slower and develops a new slow time scale for times about 10@xmath118 in the phase . this must be assigned to slow stress relaxations in the dense crystal phase likely stemming from more pronounced translational - rotational couplings @xcite , since they are absent for the small elongations at the same density .
this long - time tail occurring in the phase seems to diminish when crossing into the phase by increasing the aspect ratio to @xmath87 .
figure [ fig : etavsphi ] a ) shows the packing fraction dependence of the relative steady shear viscosity @xmath119 for different elongations @xmath116 , with @xmath120 being the viscosity of the suspending medium .
these data are obtained from integration of the in fig .
[ fig : sacf_mult_lstar ] . as an important comparison , also the reference data following the empirical krieger - dougherty relation @xcite for the _ purely fluid _
phase is shown .
as we readily see , there is a difference of the viscosity for a fluid of when compared to our minimal elongation data ( at @xmath91 ) .
this can be understood as a result of the sensitivity of viscosities to the softness of the chosen pair potential for repulsive spheres @xcite . in the present case ,
the @xmath121 system with minute anisotropy can be considered equal to the purely spherical system . at
higher packing the system is in the crystal phase and the linear shear response drops to zero ( and not accounted for in the fluid - state krieger - dougherty approach ) .
the shear viscosity of dumbbells is similar to the system in the fluid phase below packings of @xmath122 but deviates strongly for larger packings for all elongations where the dumbbell shear response substantially increases with @xmath123 .
we explain the increase by the long time tails in the in figure [ fig : sacf_mult_lstar ] due to the relaxation of the rotational degrees of freedom missing in the case .
the large increase of the shear response is somewhat unexpected since the translational degrees of freedom are still frozen , except for the largest elongation @xmath87 for which all shown data are in the fluid phase .
we note that the error bars of this data are hard to estimate due to the long - time tails in the . while we believe from the systematic behavior of the data that the trends are reliable , small fluctuations of the data as for @xmath89 are very probable within the statistical uncertainty .
( a ) as a function of the volume fraction @xmath102 and ( b ) versus elongation @xmath116 at high densities .
the line ( @xmath124 ) refers to the phenomenological krieger - dougherty theory @xcite . ]
figure [ fig : etavsphi ] b ) shows the dependence of the relative steady shear viscosity on the elongation at the high packing fractions @xmath125 .
this view reveals a transition from the hard spherical case of almost vanishing viscosity to a finite level at about a critical elongation of about @xmath126 .
this is most evident at the highest packing fraction where the transition is relatively steep .
( the fluctuations of the data between @xmath127 and 0.3 are most certainly within the statistical uncertainty . )
hence , it may be inferred from the equilibrium linear response data that a small deviation from the spherical shape already to an aspect ratio of @xmath128 induces a dramatic collective response that manifests itself in a large increase of the linear response shear viscosity .
in summary , we have investigated the equilibrium structure and fluctuations of colloidal dumbbells by means of brownian dynamics computer simulations .
we focused on high packing fractions and on the weak elongation regime ( @xmath129 ) where the dumbbells are predominantly in the plastic crystal phase .
our systematic investigation revealed the expected structural changes with larger elongation with respect to the hard sphere reference case and very localized orientational correlations , typically just involving next - neighbor correlations .
these relatively weak correlations are also reflected in only minor changes in the translational and rotational diffusion coefficients for most of the investigated elongations , except for the highest elongation in the fluid phase where the rotational diffusion drops by ca .
@xmath114 when compared to the free rotation .
however , the linear response shear viscosity exhibits a dramatic increase even in the plastic crystal phase at high packing fractions ( @xmath0 ) beyond a critical elongation of about @xmath126 .
this result is surprising in view of the relatively weak effects of elongation found before . here
one should rationalize that the linear response shear viscosity expresses collective relaxation time scales and not ( more local ) single molecule dynamic properties . apparently beyond a critical , surprisingly small anisotropy
, newly occurring rotational - translational couplings must be made responsible for the slow time scales appearing at higher elongations in the crystal @xcite .
hence , more detailed calculations and modeling of these rotational - translational correlations in weakly anisotropic model systems shall be interesting for future studies , in particular beyond linear response where probably more substantial dynamical effects of increased anisotropy may occur @xcite .
the authors thank matthias ballauff and fangfang chu for helpful discussions .
jd would like to express his gratitude to jean - pierre hansen for being a great teacher and mentor , the plenty of inspiring scientific exchanges , and his kind hospitality in cambridge and paris . | we investigate the structure and equilibrium linear - response dynamics of suspensions of hard colloidal dumbbells using brownian dynamics computer simulations .
the focus lies on the dense fluid and plastic crystal states of the colloids with investigated aspect ( elongation - to - diameter ) ratios varying from the hard sphere limit up to 0.39 , which is roughly the stability limit of the plastic crystal phase .
we find expected structural changes with larger elongation with respect to the hard sphere reference case and very localized orientational correlations , typically just involving next - neighbor couplings .
these relatively weak correlations are also reflected in only minor effects on the translational and rotational diffusion coefficients for most of the investigated elongations . however , the linear response shear viscosity exhibits a dramatic increase at high packing fractions ( @xmath0 ) beyond a critical anisotropy factor of about @xmath1 which is surprising in view of the relatively weak changes found before on the level of colloidal self - dynamics .
we suspect that even for the small investigated anisotropies , newly occurring , collective rotational - translational couplings must be made responsible for the slow time scales appearing in the plastic crystal .
[ coms]centers of masses to whom correspondence should be addressed .
e - mail : [email protected] ] dense fluids ; plastic crystals ; colloidal dumbbells ; anisotropic colloids ; linear response dynamics ; brownian dynamics simulation ; |
various binary systems of particles with pairwise interactions , confined to a plane , have been studied intensively .
they are known to exhibit a richer variety of ground state ( zero - temperature ) crystal structures than the one - component ones.@xcite 2d binary like - charge crystals with yukawa ( screened coulomb ) interactions were analyzed in detail by using monte - carlo simulations,@xcite a plenty of triangle , square and rhombic lattices were identified .
such structures were found also experimentally,@xcite especially in colloidal suspensions . the coulomb potential is a limiting case of the yukawa one , so it is not surprising that similar structures appear in 2d binary coulomb systems as well ,
as was shown in numerical simulations @xcite based on a combination of ewald summation techniques@xcite and the evolutionary algorithm ( ea).@xcite some binary systems tend to separate if the parameters ( size , charge , ) of the two components are similar .
this was observed in the case of hard disks when their two diameters are comparable,@xcite for the yukawa binary systems @xcite as well as for long - ranged coulomb systems of like charges @xcite when the charges do not differ too much .
analogous behavior , called demixing , was reported also for lenard - jones mixtures @xcite and binary mixtures with dipole interactions @xcite .
the impact of the interaction type on the resulting ground - state structures is still far from being understood . in this work ,
we study the lowest energy configurations of a binary mixture of classical pointlike particles with charges @xmath0 and @xmath1 of the same sign .
we introduce the parameter @xmath8 due to the symmetry @xmath9 , one can restrict itself to the interval @xmath10 $ ] , i.e. @xmath0 is larger than @xmath1 .
we consider the equimolar case , i.e. the numbers of species @xmath11 , where @xmath12 is the total number of particles .
the particles , confined to a 2d plane with a neutralizing background charge density , interact via the 3d coulomb @xmath13 pairwise potential .
six fully mixed structures were identified for @xmath3 , while the separation of pure @xmath0 and @xmath1 hexagonal phases minimizes the energy for @xmath4.@xcite our main contribution to the field consists in the introduction of a novel structure which consists in the separation of two phases , the pure hexagonal one formed by a fraction of particles with the larger charge @xmath0 and the other completely mixed one containing different numbers of @xmath0 and @xmath1 charges . using an analytic method based on an expansion of the interaction energy in misra functions
we show that this novel structure possesses the lowest energy in two intervals of @xmath5 , @xmath6 and @xmath7 .
the paper is organized as follows . in sec .
[ sec2 ] , we introduce the structures found by numerical simulation.@xcite there are six completely mixed structures , i. e. those with the same number of the two components in one periodic cell .
we calculate their energies per particle using analytic formulas , with an extremely high precision .
the energies are compared to the energy of the separation of two pure phases . in sec .
[ sec3 ] , we introduce the novel structure , i.e. the separation of a pure @xmath0-phase and a specifically mixed @xmath14 phase and calculate its energy . the results for the phase diagram are presented in sec .
the impact of our study to numerical simulations and other possible applications of the method are indicated in conclusion .
long formulas , with a sketch of their derivation , are presented in appendices .
the equimolar binary system of @xmath15 @xmath0-charges and @xmath16 @xmath1-charges is confined to a plane of surface @xmath17 , the particle density being @xmath18 . the particle charge density @xmath19
is given by @xmath20 , i.e. @xmath21 the particle system is immersed in a neutralizing homogeneous background of the charge density @xmath22 .
the coulomb potential energy of charges @xmath5 and @xmath23 at distance @xmath24 is given by @xmath25 .
we express the energy of all structures using the method based on a sequence of transformations for coulomb lattice sums.@xcite in contrast to the energies of bilayer wigner crystals , which are expressed as series of generalized misra functions , the energies of the present planar wigner structures require the introduction of the standard misra functions@xcite , see eq .
( [ misra ] ) of appendix [ apa ] .
the particles can always be divided into infinite lattice subsets ( with the corresponding decomposition of the neutralizing background ) in such a way that the energy per particle is expressible by using two types of lattice summations @xmath26 and @xmath27 , defined and expressed as series of misra functions in appendix [ apb ] .
these series are extremely quickly convergent and in our calculations we take first five expansion terms .
the minimization of the energy with respect to lattice parameters reproduces the exact ground - state energy up to 15 decimal digits ; this was checked on the well - known madelung constant of the hexagonal lattice and , for a few chosen structures , by including also the sixth and seventh expansion terms .
next we recall the six completely mixed structures , which have the same number of @xmath0 and @xmath1 particles per cell , found in ea numerical simulations.@xcite since some of the structures were described only partially , we provide a detailed definition of the relevant structures in order to specify their energies with a high accuracy .
this is needed because energy differences among structures are very small , especially in the region of small values of @xmath5 . at the end
, we write down the energy per particle for the separation of two pure phases . , the empty ones to @xmath1 .
dashed lines in one of the left triangles meet at the circumcenter passing through three @xmath0 particles.,scaledwidth=34.0% ] for small values of @xmath5 , the structure i was detected .
the larger charges @xmath0 constitute a slightly deformed hexagonal lattice , see fig .
[ fig : structurei ] . on the left side of this figure , each isosceles triangle has base of length @xmath28 and two legs of length @xmath29 with @xmath30 .
the isosceles triangles on the right side have the same base of length @xmath28 and two legs of length @xmath31 with @xmath32 .
the energy minimization fixes the smaller charges @xmath1 inside each of the left triangles , just on their symmetry axis ( no vertical offset ) , at distance @xmath33 from the center of the triangle ; altogether they form a zig - zag line . in the limit @xmath34
, it holds that @xmath35 and @xmath36 , restoring the perfect hexagonal ( equilateral triangle ) lattice .
the periodic rectangle has the vertical side @xmath37 and the horizontal one @xmath38 .
the lattice constant @xmath28 can be expressed in terms of the particle density @xmath39 as follows .
there are two particles of each type per rectangle , with the total charge @xmath40 .
this charge compensates the surface charge density @xmath19 within the rectangle via @xmath41 , so using the relation ( [ sigro ] ) we get for the reciprocal lattice constant @xmath42 the energy per particle of structure i is found to be @xmath43 \nonumber \\ & & + q \sigma_2\left[\psi,\frac{1}{\psi}\left(\frac{\delta_1 ^ 2 - 1/2 } { \sqrt{4\delta_1 ^ 2 - 1}}-d\right),\frac{1}{2}\right ] + \frac{q^2}{2}\sigma_1(\psi ) \nonumber \\ & & + \frac{q^2}{2}\sigma_2\left[\psi,\frac{1}{\psi}\left ( -\frac{1}{2\sqrt{4\delta_1 ^ 2 - 1}}+2d\right),\frac{1}{2}\right ] \bigg\},\end{aligned}\ ] ] where @xmath44 . here ,
the terms of order @xmath45 , @xmath46 and @xmath47 originate from @xmath48 , @xmath14 and @xmath49 interactions , respectively . , the empty ones to @xmath1 .
the cross denotes the center of the longest side of left triangles.,scaledwidth=34.0% ] structure ii resembles the structure i , but now the hexagon of particles with larger charge @xmath0 is even more deformed , all sides of the triangles have different lengths : @xmath29 and @xmath31 on the left , @xmath50 and @xmath51 on the right , see fig . [ fig : structureii ] .
the common vertical sides have length @xmath28 .
the @xmath1 charges are placed symmetrically with respect to the center of the longest sides @xmath31 .
their coordinates measured from that center are @xmath52 and @xmath53 .
this choice of only two free parameters is supported by numerical simulations and mirror symmetry considerations .
[ fig : structureii ] does not show the whole periodic rectangle .
one must imagine another hexagon added , say on the right hand side , but with lengths of the sides reversed , i. e. @xmath54 and @xmath55 .
the vertical side of the rectangle is still @xmath37 , the horizontal one @xmath56 is nontrivial , see below .
there are four particles of each type per rectangle and the reciprocal lattice constant is given by @xmath57 note that this structure can continuously go over into the structure i , if @xmath58 , @xmath59 and @xmath60 .
the energy of structure ii reads @xmath61 + q\bigg[\sigma_2\left(\psi,\frac{x}{2\psi}+d_1 , \frac{\delta_1 ^ 2-\delta_2 ^ 2 - 1}{4}+d_2 \right ) + \sigma_2\left(\psi,\frac{x}{2\psi}-d_1,\frac{\delta_1 ^ 2-\delta_2 ^ 2 - 1}{4}-d_2 \right ) \nonumber \\ & & + \sigma_2\left(\psi,\frac{1}{2}+\frac{x}{2\psi}-d_1 , \frac{\delta_1 ^ 2-\delta_2 ^ 2 + 1}{4}-\frac{\delta_3 ^ 2-\delta_4 ^ 2}{2}+d_2 \right ) + \sigma_2\left(\psi,\frac{1}{2}+\frac{x}{2\psi}+d_1 , \frac{\delta_1 ^ 2-\delta_2 ^ 2 + 1}{4}-\frac{\delta_3 ^ 2-\delta_4 ^ 2}{2}-d_2 \right ) \bigg ] \nonumber \\ & & + \frac{q^2}{2 } \left [ \sigma_2\left(\psi,-2 d_1,-2 d_2 \right ) + \sigma_2\left ( \psi,\frac{1}{2}-2 d_1,\frac{\delta_4 ^ 2-\delta_3 ^ 2 + 1}{2 } \right ) + \sigma_2\left(\psi,\frac{1}{2},\frac{\delta_4 ^ 2-\delta_3 ^ 2 + 1}{2}-2 d_2 \right ) \right ] \bigg\ } .\end{aligned}\ ] ] the altitude @xmath62 of the left triangle in fig .
[ fig : structureii ] is given by @xmath63 a similar expression holds for the right triangle .
consequently , @xmath64 reads as @xmath65
the most simple complete mixture is structure iii .
it is composed of two intertwining square lattices with the same lattice constant @xmath28 .
the larger @xmath0 charges occupy the vertices of one square lattice , the @xmath1 charges occupy the other one shifted by vector @xmath66 . the parameter @xmath67 .
there is one particle of each type per square .
the reciprocal lattice constant is given by @xmath68 .
for the energy per particle , we get the simple formula @xmath69.\ ] ] this phase has no internal parameters , so no optimization is needed .
we mention structures iv and v only briefly.@xcite they appear in a narrow interval of @xmath5 values , namely @xmath70 .
they are characterized by a large number of internal parameters and one can not go smoothly neither between themselves , nor to the neighboring iii and vi structures , hence all transitions must be of first order .
due to the absence of interesting phenomena , we decided not to calculate structures iv and v by our method and in the phase diagram we shall use values got numerically by antlanger and kahl.@xcite , the empty ones to @xmath1.,scaledwidth=40.0% ] structure vi is presented in fig .
[ fig : structurevi ] . the charges @xmath0 occupy hexagons of a deformed honeycomb lattice with two sides of length @xmath28 and four sides of length @xmath71 ; @xmath72 varies over a small interval @xmath73 $ ] .
two smaller charges @xmath1 enter the interior of each hexagon and are placed symmetrically at horizontal distances @xmath74 with respect to the hexagon center .
the sides @xmath75 and @xmath76 of the basic periodic rectangle are marked in the figure .
the horizontal side @xmath77 , where @xmath78 for a fully symmetric hexagon and we have theoretical bounds @xmath79 for a deformed hexagon . the vertical side @xmath75 is easily deduced in the form @xmath80 there are four particles of each type per rectangle .
the reciprocal lattice constant is found to be @xmath81^{1/2 } .\ ] ] the energy is obtained in the form @xmath82 + \sigma_2\left [ \psi , d,-\frac{1}{a ' } \left ( \frac{1}{2}+\sqrt{\delta^2-\frac{\delta^2}{4}}\right ) \right ] \biggr\ } \nonumber \\ & & + \frac{q^2}{2 } \left [ \sigma_2(\psi,2d,0 ) + \sigma_2\left ( \psi,\frac{1}{2},\frac{1}{2}\right ) + \sigma_2\left ( \psi,\frac{1}{2}+2d,\frac{1}{2}\right ) \right ] \biggr ) , \end{aligned}\ ] ] where @xmath83 and @xmath84 . for large enough @xmath85 ,
the energy of the nontrivial structure vi becomes higher than that of the pure phase separation in which the charged species @xmath0 and @xmath1 occupy spatially separated infinite areas , each in a hexagonal arrangement ( which is known to give the energy minimum for the planar coulomb particles of one type ) immersed in a neutralizing background @xmath19 , with the lattice constant given by the requirement of electroneutrality .
first we sketch the formula for the energy per particle of charges @xmath86 , localized on vertices of the hexagonal lattice and immersed in a neutralizing background @xmath19 .
there is one particle per two triangles , each with area @xmath87 where @xmath28 is the lattice constant .
the electroneutrality requires that @xmath88 , so that @xmath89 the distance of two interacting particles is proportional to @xmath28 and the energy per particle including the neutralizing background energy is@xcite @xmath90 where @xmath91 is the madelung constant of the hexagonal lattice . for the separation of the pure phases of @xmath92 particles with charge @xmath0 and @xmath92 particles with charge @xmath1 , each constituting the corresponding hexagonal structure , the total energy @xmath93 is@xcite @xmath94 \nonumber \\ & = & \frac{n}{2 } \sqrt{\sigma } c_{\rm m } \left ( q_1^{3/2}+q_2^{3/2 } \right ) .
\label{et0}\end{aligned}\ ] ] with regard to eq .
( [ sigro ] ) , the energy per particle @xmath95 is given by @xmath96 as was mentioned above , the separation is energetically favorable for higher values of @xmath5 , but its energy will serve as a reference for any @xmath5 .
now we introduce a novel structure which consists in the separation of a pure @xmath75 and a mixed @xmath76 phases .
the pure phase @xmath75 is composed of @xmath97 charges @xmath0 in a hexagonal arrangement ; there are no @xmath1 charges in phase @xmath75 , i.e. @xmath98 . the mixed phase @xmath76 is composed of the remaining @xmath99 charges @xmath0 and all @xmath100 charges @xmath1 .
the charges are distributed in a specific way within a basic hexagonal lattice .
we denote the primitive translation vectors of this basic hexagonal lattice as @xmath101 see fig .
[ fig:1 - 1 ] .
let the vector @xmath102 be defined as @xmath103 where @xmath104 are non - negative integers , at least one of them being nonzero , i.e. @xmath105 , @xmath106 , @xmath107 , @xmath108 , etc .
for a given @xmath104 , charges @xmath0 occupy a hexagonal subset of the basic hexagonal lattice with the lattice constant @xmath109 ; it is a special property of any hexagonal lattice @xmath110 that joining arbitrary two vertices of @xmath110 implies a side of the new hexagonal lattice @xmath111 whose all points also belong to @xmath110 .
the remaining sites are occupied by charges @xmath1 .
the special case @xmath106 is pictured in fig .
[ fig:1 - 1 ] .
the particle densities of fully occupied hexagonal structures are proportional to the reciprocal of the squared lattice constant .
in particular , @xmath112 and @xmath113 since the the basic hexagonal lattice contains both types of particles . consequently , @xmath114 using the fact that @xmath115 , the number of @xmath0 particles in phase @xmath76 is equal to @xmath116 to calculate the total energy for the pure hexagonal phase @xmath75 with @xmath117 particles of charge @xmath0 , we use eq .
( [ ehex ] ) and obtain the result @xmath118 of the separation \{1,1}. full disks correspond to particles with charge @xmath0 , the empty ones to @xmath1 . the basic hexagonal structure ( solid lines )
has lattice constant @xmath28 , particles @xmath0 form another hexagonal lattice with lattice constant @xmath119.,scaledwidth=40.0% ] to calculate the total energy for the mixed phase @xmath76 , we introduce the average charge @xmath120 as follows @xmath121 in analogy with eq .
( [ aqs ] ) , the lattice constant @xmath28 is given by @xmath122 we apply a trick which consists in splitting each charge @xmath0 into two charges , @xmath123 , and then performing three summations over hexagonal lattices only .
the first contribution sums the mutual interactions of @xmath124 charges @xmath1 on the hexagonal lattice with lattice constant @xmath28 , @xmath125 the second one sums the interactions of @xmath126 @xmath127 charges with all other charges @xmath1 on the hexagonal lattice with lattice constant @xmath28 , @xmath128 note the factor 2 for the interaction of different species .
the third contribution sums mutual interactions of @xmath126 charges @xmath127 on the hexagonal lattice with lattice constant @xmath109 , @xmath129 to get the total energy of the phase separation , coined by @xmath130 , we sum all three terms with the energy ( [ rega ] ) of the pure phase @xmath75 . the final formula for the energy per particle becomes @xmath131 .\end{aligned}\ ] ] note that the limit @xmath132 or @xmath133 of this relation reproduces correctly eq .
( [ e0 ] ) . the separation @xmath106 with the mixed @xmath76 phase presented in fig .
[ fig:1 - 1 ] is special : each charge @xmath0 is in the middle of an hexagon formed by six @xmath1 charges and each charge @xmath1 is in the middle of a triangle formed by three @xmath0 charges . due to this symmetry ,
the total force acting on particles vanishes and the particles are in the true ground state with the energy per particle given by eq .
( [ ejk ] ) as follows @xmath134 . \end{aligned}\ ] ] for other separations @xmath104 , the particles are in general in unstable positions and should relax to the optimal ground - state positions . for the structure @xmath105 in fig .
[ fig:0 - 2 ] , let us make the lattice of ( black ) @xmath0 charges rigid .
the ( white ) @xmath1 charges experience an asymmetric force with a saddle point at the center of the rhombus . fig .
[ fig:0 - 2 ] documents one of possibilities how @xmath1 charges can relax within the equilateral triangles . for small @xmath5 ,
the distance of @xmath1 charges to @xmath0 increases , since its reciprocal enters the energy term linear in @xmath5 , whereas the mutual @xmath49interactions are only of the order @xmath47 , allowing them to get closer .
symmetry considerations suggest that the distance to each pair of @xmath0 particles should be equal , which is represented by the dashed lines and the @xmath1 particles form a smaller equilateral triangle .
for these relaxed configurations , we indeed get slightly lower energies than for the non - relaxed ones .
there are other possibilities , but we do not go into details since for any @xmath5 all these relaxed structure have higher energies than the simplest rigid @xmath106 structure . of the separation @xmath105 .
full disks correspond to particles with charge @xmath0 , the empty ones to @xmath1 .
there are three @xmath1 particles within each rhombus formed by @xmath0 vertices.,scaledwidth=37.0% ]
we have used the above formulas for various structures to optimize the minimum energy in the whole interval @xmath135 . the approach taken from our former works@xcite allowed us to achieve at least 15 decimal digits precision , compared to typical 7 decimal digits in numerics.@xcite we included the novel separated @xmath136 phase which turns out to have the lowest energy both in the region of small and intermediate values of @xmath5 .
the differences in energies between competing structures are often very small , hence we will subtract the reference value @xmath137 , see eq .
( [ e0 ] ) , which makes them more visible .
the quantity in our figures will be @xmath138/(\sqrt{\rho}q_1 ^ 2)$ ] , where @xmath139 is the energy of the considered structure .
it stands to reason that when the separation of the two pure phases becomes dominant , the corresponding energy difference is 0 .
it was claimed@xcite that the energy @xmath140 of structure i gives the lowest energy for small values of @xmath5 , up to @xmath141 when structure ii takes the place .
we found that the energy of the separated ( pure and mixed ) structure @xmath142 is lower than @xmath140 in the whole interval of structure i dominance .
it is even lower than @xmath143 within a short interval of @xmath5-values , see fig .
[ fig : small - q ] .
the difference between @xmath144 and @xmath142 is tiny , of order @xmath145 which is on the border of accuracy of numerical methods . in this work , we take advantage of analytic formulas and calculate the small-@xmath5 expansions of both the optimized @xmath144 and @xmath142 . from the representation ( [ e1 ] ) , we deduce the expansion @xmath146 + { \cal o}\left ( q^4 \right ) \end{aligned}\ ] ] with @xmath147 being negative . here
, the coefficients up to the @xmath47 term were derived analytically in appendix [ apc ] , the coefficient at @xmath148 comes from numerical fittings .
this expression should be compared to the analogous expansion of eq .
( [ e11 ] ) : @xmath149 + { \cal o}\left ( q^4\right )
\nonumber\\ & = & e_0(0 ) \big [ 1+(\sqrt{3}-1)q + 1.375q^2 \nonumber \\ & & - 0.79246825 q^3 \big ] + { \cal o}\left ( q^4\right ) .\end{aligned}\ ] ] the two expansions ( [ e1exp ] ) and ( [ e11exp ] ) coincide up to the term linear in @xmath5 . since @xmath150 and the prefactor to the @xmath47 term in ( [ e11exp ] ) is larger than that in ( [ e1exp ] ) , the separation of the pure and mixed phases @xmath136 is dominant in the region of small @xmath5 .
numerically one finds that @xmath151 for @xmath152 . /(\sqrt{\rho}q_1
^ 2)$ ] for completely mixed structures i , ii and for the separation of the pure and mixed phases @xmath136 in the region of small values of @xmath5 .
the transition point between structures i and ii is depicted by the full disc .
this point is ineffective since the structure @xmath153 is dominant well beyond it .
structure ii becomes the one with the lowest energy at @xmath154.,scaledwidth=45.0% ] we mentioned the possibility that the structure i goes smoothly into structure ii via a second - order phase transition . if our separation @xmath106 is not taken into account , the transition between structures i and ii at @xmath155 would be of first order , i.e. there is a change in the slope of the upper line in fig .
[ fig : small - q ] and the structural parameters change discontinuously .
but the structure i is wiped out by the separation @xmath106 and the latter performs a first - order transition directly into the structure ii at @xmath156 .
the next transition between structures ii and iii takes place at @xmath157 . /(\sqrt{\rho}q_1
^ 2)$ ] for intermediate values of @xmath5 .
the interplay among mixed phase vi , our separation of pure and mixed phases @xmath136 and the separation of pure phases @xmath158 .
the dashed lines would be valid if the separation @xmath106 is ignored .
the abbreviation `` pps '' means `` pure phase separation '' , its energy @xmath137 serving as a reference.,scaledwidth=45.0% ] it was found@xcite that the completely mixed structure vi goes over into the separation of two pure phases , represented by the zero value of @xmath159 , via a first - order transition at @xmath160 .
we found surprisingly that the separation @xmath106 provides the lowest energy also for intermediate values of @xmath5 , namely for @xmath161 , see fig .
[ fig : medium - q ] . the lower value of @xmath162 marks the first - order transition between structures vi and @xmath106 , the higher @xmath163 the first - order transition between @xmath136 and the separation of two pure phases .
note extremely small differences among energies .
finally , we summarize the minimum energies of dominant phases for all values of @xmath164 in fig .
[ fig : all - q ] .
the energies are calculated by using the series formulas given above , with the exception of marginal phases iv and v ; the corresponding data were taken from the previous numerical study.@xcite the open circles denote the first - order transition points .
note that the separated structures emerge when @xmath1 particles are almost negligible or comparable to @xmath0 .
the phase separation of pure hexagonal phases was observed also in the case of yukawa interaction,@xcite for @xmath85 .
/(\sqrt{\rho}q_1 ^ 2)$ ] for all values of @xmath5 .
open circles mark the first - order transition points .
the abbreviation `` pps '' means `` pure phase separation''.,scaledwidth=45.0% ]
we have shown that besides the perfectly mixed structures and the separated pure phase , a special separation of pure and mixed phases @xmath136 can appear in the phase diagram of a planar , equimolar binary mixture of classical charged particles .
the energy of the separation structure @xmath106 turns out to be the lowest one in two regions of @xmath5 : it dominates the completely mixed i and ii phases for @xmath6 and the completely mixed vi phase and the separation of two pure phase for @xmath7 .
the energy differences among the considered phases are very small .
we obtained the results by using an analytic method based on an expansion of the coulomb interaction energy in misra functions@xcite which enables us to get the energies with an accuracy by a dozen digits better than that of ea numerical simulations.@xcite as concerns possible variations of the model , one could consider different numbers of two particle species , like it was done in monte carlo approach of yukawa binary mixtures.@xcite another interesting problem is to confine the binary coulomb mixture between two ( symmetrically or asymmetrically ) charged plates ; with regard to the planar case , there should exist a larger variety of bilayer wigner crystals .
our method permits one to study inverse - power - law or lenard - jones particles@xcite as well
. this paper might inspire physicists oriented in numerical computation to test more general separations of binary mixtures , going beyond a simple separation of two pure phases .
we have proposed a specific separation of a pure and a mixed phase which provide the lowest energy in comparison with the known structures in two intervals in the regions of small and intermediate values of parameter @xmath5 .
but there may exist other separations , say of two or more completely mixed phases , with even lower energy .
it is not clear how to test numerically all possibilities on a reasonable time scale .
we would like to thank moritz antlanger and gerhard kahl for sending us their numerical data of energy,@xcite partially used in fig .
[ fig : all - q ] .
the support received from the grant vega no .
2/0015/2015 is acknowledged .
the misra functions are defined by@xcite @xmath165 in this paper , we use the misra functions with the half - integer argument @xmath166 which are expressible in terms of the complementary error function@xcite @xmath167 for the first misra functions , we have@xcite @xmath168 , \nonumber \\
z_{7/2}(y ) & = & \frac{\sqrt{\pi}}{4 y^{5/2 } } \left [ 2 { \rm e}^{-\pi y}\sqrt{y}\left(3 + 2\pi y\right ) + 3\ { \rm erfc}{\left(\sqrt{\pi y}\right ) } \right ] . \nonumber \\ & & \label{znu0y}\end{aligned}\ ] ] for a small @xmath169 , we shall need the following expansion @xmath170 \nonumber\\ & = & z_\nu(y ) -\delta y\
z_{\nu+1}(y ) + \frac{(\delta y)^2}{2}\ z_{\nu+2}(y ) . \phantom{aaa } \label{zexp}\end{aligned}\ ] ]
using a sequence of transformations for coulomb lattice sums,@xcite we derive the interaction energy as a series of misra functions for two geometries presented in fig .
[ fig : rectangles ] . with ( i ) the reference site @xmath171 being on the rectangle ; ( ii ) the reference site @xmath171 being shifted from the rectangle.,scaledwidth=37.0% ]
let the periodic rectangular lattice of sides @xmath172 first includes the reference site @xmath171 .
taking the particles on the rectangle vertices to have charge @xmath173 , the interaction energy of the reference particle with all others ( but not with itself ) can be written as @xmath174 this energy can be expressed as@xcite @xmath175 where @xmath176 + 8 \sum_{j , k=1}^\infty z_{3/2}\left(\frac{j^2}{\psi}+k^2\psi\right ) - 4\sqrt{\pi } .\ ] ] if the reference site @xmath171 is shifted with respect to the rectangular lattice as indicated in fig . [
fig : rectangles ] , the energy @xmath177 is expressible as@xcite @xmath178 where @xmath179 \phantom{aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa } \nonumber\\ + 4\sum_{j , k=1}^\infty \cos(2\pi j \alpha_1)\cos(2\pi k \alpha_2 ) z_{3/2}\left(\frac{j^2}{\psi}+k^2\psi\right ) + \sum_{j , k=-\infty}^\infty z_{3/2}\left[(j+\alpha_1)^2\psi+\frac{(k+\alpha_2)^2}{\psi}\right ] - 2\sqrt{\pi } .
\label{s2}\end{aligned}\ ] ] from the representation ( [ znu0y ] ) of the misra functions @xmath180 it is simple to show that @xmath181 = 2 \sqrt{\pi } + \sigma_1(\psi ) , \ ] ] where the subtraction of the singular term from @xmath27 corresponds to the interaction of the reference site @xmath171 with its position on the rectangular lattice when @xmath182 .
we aim at calculating the small-@xmath5 expansion of the optimized energy of structure i , @xmath144 , from eq .
( [ e1 ] ) , up to the @xmath47 term .
the absolute term can be obtained straightforwardly by setting the limiting @xmath34 value of the parameters @xmath183 @xmath184 in the first two summands , to get @xmath185 = \frac{2}{3 } e_0(0 ) + \frac{1}{3 } e_0(0 ) = e_0(0).\ ] ] the contributions of these first two summands to the term linear in @xmath5 cancel with each other , so that the @xmath5 term is given by the next two summands in ( [ e1 ] ) , where we set @xmath186 , @xmath187 and @xmath188 , @xmath189 q = \left ( \sqrt{3}-1\right ) e_0(0 ) q.\ ] ] in order to derive the @xmath47 term , we use the taylor expansion in @xmath5 of the internal parameters : @xmath190 expanding eq .
( [ e1 ] ) in @xmath5 and using the expansion ( [ zexp ] ) , we get @xmath191 q^2 + { \cal o}\left(q^3\right ) , \ ] ] where @xmath192 -\frac{16\pi}{9 } \sum_{j=1}^{\infty } j \sin\left(\frac{\pi j}{3}\right ) z_{3/2}\left ( \frac{j^2}{\sqrt{3}}\right ) \nonumber\\ & & - \frac{32\pi}{9 } \sum_{j , k=1}^{\infty } j
\sin\left(\frac{\pi j}{3}\right ) ( -1)^k z_{3/2}\left(\frac{j^2}{\sqrt{3}}+k^2\sqrt{3}\right ) - \frac{8\sqrt{3}}{9 } \sum_{j , k=-\infty}^{\infty } \left ( j+\frac{1}{6}\right ) z_{5/2}\left[\left(j+\frac{1}{6}\right)^2\sqrt{3}+\frac{(k+1/2)^2}{\sqrt{3 } } \right ] \bigg\ } \nonumber\\ & \approx & -0.9822578883980 , \label{a1}\end{aligned}\ ] ] @xmath193 \nonumber\\ & & + \frac{4}{3}\sum_{j , k=-\infty}^{\infty } \left(j+\frac{1}{2}\right)^2 z_{7/2}\left[\left(j+\frac{1}{2}\right)^2\sqrt{3}+\frac{1}{\sqrt{3 } } \left ( k+\frac{1}{2 } \right)^2 \right ] \bigg\ } \nonumber\\ & \approx & 0.62793330025966\end{aligned}\ ] ] and @xmath194 ( the series expression for @xmath195 is also at our disposal ) . here
, the terms proportional to @xmath196 and @xmath197 vanish since their prefactors contain a sum from @xmath198 to @xmath199 of the summands @xmath200 $ ] which cancel in pairs @xmath201 , @xmath202 , etc .
the same holds for the mixed term proportional to @xmath203 whose prefactor contains the sum of @xmath204 $ ] .
the terms proportional to @xmath205 , @xmath206 , @xmath207 and @xmath208 vanish as well , though the cancellation is more tricky .
the minimization of the contribution @xmath209 to the energy @xmath210 implies @xmath211 because @xmath195 is positive .
we are left with the only optimization parameter @xmath212 .
taking @xmath213 yields @xmath214 inserting the values of @xmath215 , @xmath216 , @xmath212 and @xmath211 into eq .
( [ e1exq2 ] ) , we get with a high precision the value of the coefficient at @xmath47 in eq .
( [ e1exp ] ) . | we study the lowest energy configurations of an equimolar binary mixture of classical pointlike particles with charges @xmath0 and @xmath1 , such that @xmath2 $ ] .
the particles interact pairwisely via 3d coulomb potential and are confined to a 2d plane with a homogeneous neutralizing background charge density . in a recent paper by m. antlanger and g. kahl [ cond .
mat .
phys .
* 16 * , 43501 ( 2013 ) ] , using numerical computations based on evolutionary algorithm , six fully mixed structures were identified for @xmath3 , while the separation of @xmath0 and @xmath1 pure hexagonal phases minimizes the energy for @xmath4 . here
, we introduce a novel structure which consists in the separation of two phases , the pure hexagonal one formed by a fraction of particles with the larger charge @xmath0 and the other mixed one containing different numbers of @xmath0 and @xmath1 charges . using an analytic method based on an expansion of the interaction energy in misra functions
we show that this novel structure provides the lowest energy in two intervals of @xmath5 values , @xmath6 and @xmath7 .
this fact might inspire numerical methods , for both coulomb and yukawa interactions , to test more general separations which go beyond the separation of two pure phases . |
we analyze the current through a one - dimensional array of @xmath2 metallic nanoparticles placed between two electrodes .
we consider the classical coulomb blockade regime with @xmath3 .
@xmath4 is the level spacing , @xmath5 the temperature and @xmath6 the charging energy of the islands , with capacitance @xmath7 . here and in the following the electronic charge @xmath8
we assume that each nanoparticle has a continuum level spectrum ( @xmath9 ) and a constant density of states at the fermi level , but a gap @xmath10 for adding charge .
we restrict electrostatic interactions to those charges on the same conductor : capacitive coupling vanishes .
this limit is referred to as short - range or onsite interaction limit .
the nanoparticles are separated by high tunneling barriers with a resistance much larger than the quantum of resistance . in these conditions the charge in the islands can be assumed fixed and quantized .
eventually we allow tunneling processes between nearest neighbors , and treat the transport at the sequential tunneling level .
a single charge is involved in the tunneling process .
we assume that when a charge hops , the charge density in the final state of the array immediately relaxes to the electrostatic equilibrium configuration .
we take into account that the electrodes are not ideal voltage sources , but have a finite self - capacitance . in equilibrium , and before the tunneling event the electrodes are held at a given potential due to the charge provided by a battery .
we assume that the tunneling time , i.e. the time needed by the electron to cross the tunnel barrier , is smaller than the circuit characteristic time that determines how quickly the battery can transfer charge to the leads in order to restore the voltage at the electrodes . as a consequence , just after the tunneling process the electrodes will not necessarily be at the same potential at which they were at before the tunneling event because the charge , provided by the battery , necessary to restore their initial potentials has not arrived yet .
the voltage is restored to the nominal value before the next tunneling event . for finite - range interactions the potentials on the leads will thus fluctuate in response to all tunneling events , even those that do not directly involve the electrodes . in the short - range case , considered here
, they will fluctuate only when an electron jumps into or out of the leads .
the current is calculated numerically by means of a monte carlo simulation , described in the appendix , which depends on the tunneling rates .
the probability of a tunneling process@xcite is given by @xmath11 with @xmath12 the tunneling resistance of the junction .
we will restrict the discussion to zero temperature for which @xmath13 .
@xmath14 is the difference between the energy of the system before and after the tunneling event , with the sign convention that @xmath14 is negative if the energy decreases .
it excludes the work done by the battery to recharge the electrodes , as explained before .
the energy gained by tunneling is assumed to be dissipated .
only changes in energy with electrostatic origin are considered .
the energy of our system is given by @xmath15 labels @xmath16 and @xmath17 refer to source and drain electrodes and @xmath18, ... ,@xmath2 to the islands .
in the following , latin capital and lower case letters are used to denote electrodes and islands respectively .
greek indexes will be used when the labels refer to both islands and electrodes . @xmath19 and @xmath20 are respectively the conductor capacitance and the charge in it .
the charges provided by the battery at the source and drain electrodes , which maintain their potentials at @xmath21 and @xmath22 are @xmath23 and @xmath24 .
the capacitances of the electrodes are much larger than those of the nanoparticles .
@xmath25 is a random potential at each island due to randomly dispersed charges within the substrate and within the material surrounding the nanoparticles@xcite .
clean arrays will be characterized by @xmath26 for every @xmath27 .
the random potential is included only at the islands because a similar term at each electrode is compensated by the battery and thus has no effect on transport . in the case of disordered arrays , the disorder potentials
can , in principle , take values larger than the charging energy @xmath10 .
however , for large values of the disorder potential , charges flow to compensate for these large fluctuations . in the case of short - range interactions , except if the original disorder potential is very weak , once the screening of the potential due to the mobile charges is taken into account , the set of disorder potentials is uniformly distributed in the interval @xmath28@xcite , and in the following we consider this distribution . the relevant quantity for the transport is the change in energy due to a tunneling event .
the tunneling process can be seen as the creation of a hole in the conductor @xmath1 from which the charge leaves , @xmath29 , and the addition of an electron in @xmath30 at which the charge arrives , @xmath31 . here and thereafter , we let @xmath32 denote the charge of an electron ( hole ) .
in fact , the change in energy can be rewritten as the energy to create an electron - hole ( also called in the following excitonic energy ) plus the difference in potential between the sites involved in the process before the tunneling event . @xmath33
the first term gives the energy to create an electron - hole pair in an uncharged clean array and is given by @xmath34 this energy does not depend on the direction of tunneling ( from @xmath1 to @xmath30 or from @xmath30 to @xmath1 ) and in the following it will be denoted @xmath35 with @xmath27 running from @xmath18 to @xmath17 .
index @xmath27 , when used to label a junction will refer to the one between conductors @xmath36 and @xmath27 .
we will use the term _ contact _ junction for those junctions which connect an island and an electrode , and _ bulk _ or inner junction for those ones in between two nanoparticles . for the contact junctions
@xmath37 @xmath38 as @xmath39 while for the bulk junctions @xmath40 to @xmath41 @xmath42 . the second term in ( [ change ] )
can be seen as the change in potential between the conductors involved in the tunneling .
the potential at each site depends on the charge state of the array prior to the tunneling event . at the electrodes @xmath43 , @xmath44 . at the islands
, the potential can be decomposed into two terms : @xmath25 , a random potential due to random charges in the substrate and @xmath45 a potential due to the charges in the islands . @xmath46 , with @xmath47 analogously we can define the potential drop at each junction @xmath48 with the corresponding disorder and charging terms @xmath49 and @xmath50 .
the potential drop at a contact junction depends on the disorder and charge state of the nanoparticle and on the applied bias voltage . on the contrary
the potential drop at a bulk junction is not affected by the bias voltage , except via a change in the charge state . in the following
we rewrite the potential at the electrodes as @xmath51 and @xmath52 .
the total potential drop through the array is @xmath53 . in our model
some measurable properties depend on the value of @xmath1 , which characterizes how the bias voltage is partitioned between source and drain chemical potential shifts . in several previous works the value of @xmath1
was chosen either as @xmath54 , correspondingly to a symmetrically biased array , or as @xmath55 corresponding to completely asymmetric biasing .
@xmath56 has been also called the forward bias condition@xcite both values have been used in the literature , mostly without discussion . in the symmetrically biased case the potential drop at both contact junctions is equally modified by the bias voltage . on the contrary for @xmath57 only the drain ( source ) junction
is affected by the bias .
since no physical properties depend on the overall zero of energy , varying @xmath1 in our model is entirely equivalent to rigidly shifting all impurity potentials by @xmath58 .
the dependence on @xmath1 discussed below , corresponds in part to a dependence on the alignment of the equilibrium source and drain chemical potentials with respect to the addition and removal energies of the electron .
for example for a single nanoparticle on whether the chemical potential shift required to add or remove an electron is larger . since in our model all transport occurs by transfer between adjacent nanoparticles , the evolution of a nanoparticle array as the bias voltage is applied is sensitive to @xmath1 . for a given nanoparticle array with a fixed set of disorder potentials , we believe that the dependence on @xmath1 discussed below should in principle be observable .
whenever not specified we assume that all the junction resistances @xmath59 are equal and given by @xmath60 .
the effect of non homogeneous resistances will be studied in two ways .
one of the junction resistances at a given position is larger than the other ones ( given by @xmath60 ) or resistances , varying in between two values are randomly assigned to the junctions . to mimic that disorder in resistances originates in variations in distances between the islands and the exponential dependence of the junction resistance on the distance between islands
the junction resistance is given by @xmath61 with @xmath62 and @xmath63 input parameters and @xmath64 . here
@xmath65 is a random number between @xmath16 and @xmath18 . in the paper , we have used @xmath66 and @xmath67 . with these values
the resistance changes respectively between ( 5 - 11)@xmath60 , ( 8 - 21)@xmath60 and ( 23 - 83)@xmath60 .
in this section we analyze the dependence of the threshold voltage on the array parameters .
the threshold voltage is controlled by changes in energy in tunneling and not affected by the resistance of the junctions .
thus , we do not address the case of disorder in resistance in this section , as it does not modify the threshold compared to the equal - resistances case .
the threshold voltage is the minimum bias voltage at which current can flow through the array .
a finite bias voltage can assist the entrance of charge to the array from the leads , as it creates a potential drop at the contact junction , which can overcome the excitonic energy .
a finite current requires that charges are able to be transferred from one electrode to the other one across the entire array .
if charge flow can occur between the leads , the threshold voltage is the minimum voltage which permits the entrance of an electron or hole into the array .
however it is possible for charge to become stacked inside the array due to the disorder potential configuration or due to the lack of potential drops across the bulk junctions . in this case
the threshold will be controlled by the flow of charges . here for disordered arrays we recover previously predicted@xcite values for the average threshold @xmath68 and its root mean square deviation @xmath69 .
@xmath68 is proportional to the number of particles and @xmath69 to @xmath70 . on the contrary , for clean arrays and strictly onsite interactions
the threshold voltage differs from the one expected by extrapolating to zero coupling the value obtained for weakly coupled nanoparticles .
in particular , for finite coupling ( or its zero coupling extrapolation ) an @xmath2-independent threshold voltage is predicted for large arrays , while for strictly zero coupling we obtain @xmath71 .
we also show that its value depends on @xmath1 .
the threshold corresponding to the clean case is plotted in fig .
1(a ) for the case of symmetrically biased arrays ( @xmath72 ) , antisymmetrically biased arrays ( @xmath73 ) , and an intermediate biasing ( @xmath74 ) . in the symmetrically biased case
, @xmath0 shows a step - like dependence on @xmath2 . there is a clear even - odd effect . the even - odd effect in the threshold voltage with the number of particles is absent for @xmath56 and @xmath75 with a threshold voltage @xmath76 .
at strictly zero temperature the tunneling rate @xmath77 and vanishes when @xmath14 is positive or zero . at the inner junctions
@xmath14 is independent of the bias voltage and zero or negative if the two islands differ just by just a single charge . thus the tunneling rate vanishes .
@xmath78 junctions prevent the flow of charges .
a charge gradient at each bulk junction has to be created to allow flow of charge . in the symmetrically biased case ,
increasing the potential at the electrodes allows positive and negative charges to enter from the source and the drain respectively .
these charges accumulate on the array and create potential drops across the bulk junctions . at voltages just below the threshold , the accumulated charges at the first and last islands are equal in number and opposite in sign .
current starts to flow at voltages larger than @xmath79 when @xmath2 is odd and @xmath80 when @xmath2 is even , as these values allow for the build up of @xmath81 and @xmath82 charges at the first and last islands for odd and even @xmath2 respectively .
this corresponds to a charge gradient @xmath83 , across all bulk junctions for odd @xmath2 and across all bulk junctions except one for even @xmath2 . for @xmath55 , charge can only enter the array from one lead and the energy barriers across _ all _
@xmath78 bulk junctions must be overcome by accumulated charges .
when @xmath54 the first and last junctions are equivalent and charge can enter from both leads so it is possible in some cases for the energy barrier across one of the bulk junctions to be overcome by the potential drop due to two injected charges of opposite sign from the two leads , i.e. one of the junctions can be uncharged .
the absence of this possibility is what removes the even - odd effect when @xmath75 and @xmath73 .
an intermediate situation is found for @xmath84 .
complementary information can be obtained by looking at the threshold voltage as a function of @xmath1 for different values of @xmath2 in fig .
the threshold voltage changes in a periodic way with @xmath1 .
the periods of the features in @xmath85 depend on the number of barriers in the array .
dependence of @xmath0 on @xmath2 and periodic features in @xmath0 with respect to @xmath1 reflect the number of charges which have to accumulate in the first and last island prior to current flow .
1(b ) includes curves corresponding to @xmath21 and @xmath22 for specific values of @xmath86 and @xmath87 . as the assymmetry of the array increases ( increasing @xmath88 ) ,
the threshold voltage is alternately determined by the cost of injecting a charge unto the array from the source and the drain . at all
@xmath1 except for the values that lead to the minimum @xmath0 for a given array length , the difference in the charge occupying the first and last islands , @xmath89 , equals @xmath78 .
at the minimum values of @xmath0 , this charge difference equals @xmath90 . for different values of the asymmetry bias parameter @xmath1 .
a clear even - odd effect is present for symmetrically biased arrays , @xmath91 .
( b)threshold voltage as a function of @xmath1 . from top
to bottom lines with symbols correspond to clean arrays with 6 and 5 islands and to disordered arrays with 6 and 5 islands .
the peak and valley dependence reflect the change of contact junction which control the entrance and the number of charges which have to accumulate in the corresponding contact island . for comparison thin solid and dashed lines
give , for clean arrays , the bias voltage at which 2 and -2 charges can be placed at the first and last islands .
( c ) main figure : average threshold voltage for disordered arrays as a function of the number of islands .
the dependence of @xmath0 on @xmath1 disappears on average and a linear dependence on @xmath2 is recovered .
fluctuations in threshold voltage follow @xmath92 as predicted@xcite , width=3 ] the clean case of a related system was studied by hu and oconnell@xcite .
they analyzed a one - dimensional array of n gated junctions with equal junction capacitances @xmath93 and equal gate capacitances @xmath94 .
due to the finite value of @xmath93 charges in a given island interact with charges in other islands and with charges in the electrodes . with an applied bias voltage
the interaction between charges in the electrodes and in the islands results in a bias induced potential drop at the bulk junctions .
once a charge is injected unto the array , it will have no difficulty in traveling through it , and the threshold voltage equals the voltage required for injection of a charge from the electrodes .
as the ratio @xmath95 increases the threshold voltage of a long array tends to an n - independent value of the order of the charging energy .
the onsite case discussed here corresponds to @xmath96 .
if one extrapolates the case discussed by hu and oconnell@xcite to @xmath97 an n - independent threshold voltage would be expected for onsite interactions .
as shown above , the threshold voltage of clean arrays does not satifies this prediction , as at zero temperature the charges can not travel freely through the array , and the threshold voltage increases with the number of islands . in the case of disordered arrays , each array has a threshold voltage @xmath0 dependent on the given configuration of disorder @xmath98 .
the threshold voltage depends on @xmath1 , in a way similar to the clean case .
however , as shown in fig .
1(c ) , this dependence on @xmath1 disappears in the average value and we recover middleton and wingreen prediction@xcite . for the disordered case middleton and
wingreen@xcite predicted a linear dependence of the threshold voltage on the array length .
upward steps in the disorder potential @xmath99 prevent the flow of charge . the downward steps @xmath100 facilitate it .
in average there are @xmath101 upward steps . to overcome such steps a charge gradient has to be created in those junctions . for onsite interactions this results in@xcite @xmath102 .
in this limit , they also argued that fluctuations in the values of @xmath0 over many random configurations of disorder are analogous to the fluctuations in the net distance traveled by a 1-d random walk with @xmath2 steps , resulting in @xmath92 .
compared to arrays with no disorder , the threshold of disordered arrays in the onsite limit tends to be smaller because in the clean case , all the bulk junctions inhibit charge flow whereas in the disordered case , only the bulk junctions with positive @xmath49 block charge flow . as shown in the inset of fig .
1(c ) , we also recover the relationship for the fluctuations in the threshold predicted by middleton and wingreen@xcite , @xmath103 .
for bias voltages larger than threshold the current @xmath104 can flow , but it is a strongly non - linear function of voltage . the current depends on the charging energy and number of islands , the presence or not of charge disorder in the array , the resistances of the junctions and on the asymmetry of the applied bias voltage .
linearity and independence on @xmath1 is recovered at large voltages . as discussed by middleton and wingreeen@xcite a power - law dependence of the current on voltage
is found close to threshold . in this section
we discuss the different regimes which can be differentiated in an i - v characteristics and its dependence on the input parameters . at very low voltages
by comparing analytical and numerical results we resolve the controversy on the exponent of the power - law and show it to be linear with a slope which depends on @xmath1 and the resistance of the contact junctions , but that is independent on the array length ( except in a particular case in which an even - odd effect is found ) .
the linearity is however restricted to very small values of @xmath105 .
we also clarify the dependence of the coulomb staircase profile on the bias parameter @xmath1 and estimate the asymptotic current at high voltages , and the bias voltage at which this high voltage regime is found .
very large values of the bias voltage have to be applied to reach this linear dependence .
there has been some controversy regarding the power - law of the current with @xmath106 through one dimensional disordered arrays for voltages close to @xmath0 , @xmath107 .
middleton and wingreen@xcite predicted linear behavior for both the long and short range interaction .
reichardt and reichardt@xcite found a square root behavior using a model with a @xmath108 interaction between the charges in the islands .
they argued that @xmath109 is the exponent corresponding to an sliding charge - density wave .
they pointed out that the larger values of the exponent obtained by middleton and wingreen@xcite are a consequence of using voltages which are not small enough .
kaplan et al.@xcite found @xmath110 in the long - range limit of an array of dots capacitively coupled to their nearest neighbors . finally jha@xcite and
middleton argued that the dependence of the current of disordered arrays in the onsite limit on @xmath106 for voltages marginally greater than @xmath0 is linear with an slope inversely proportional to the length of the array .
numerically they found an approximate linear behavior only in the case of very long arrays but not for the smallest voltages analyzed ( where they found a sublinear dependence ) but in an intermediate voltage regime .
so far , there are no experiments available in completely one - dimensional arrays , but there are a few in quasi - one dimensional systems .
the approximate power - law measured@xcite at voltages @xmath111 is larger than unity which has been attributed to the fact that the system is not strictly one - dimensional . in this subsection , we show that the current varies linearly with respect to @xmath105 for very small @xmath105 and that the slope is not inversely proportional to @xmath2 .
the slope is sensitive to the resistance of the contact junctions and to the degree of symmetry between the applied bias voltages on the source and drain leads . while this linear dependence is found analytically and in the numerical calculations the value of @xmath106 at which it disappears is very small .
because of the small value of the current at such voltages , the linear dependence most probably can not be seen experimentally . ) of the curves plotted in the main figures.the linear dependence of current on voltage ( constancy of the derivative )
is clearly seen in all the plots but it disappears for @xmath112 which for the cases shown corresponds to @xmath113 .
( a ) to ( c ) show i - v curves corresponding to arrays without charge disorder .
all junction resistances are equal in ( a ) and ( b ) . as shown in ( a ) for clean arrays the slope of the linear dependence does not depend on the number of islands except for @xmath114 which shows an even - odd effect with a slope equal to unity and @xmath115 for odd and even number of particles , respectively .
( b ) the dependence of slope on @xmath1 can be non - monotonous if there is a change in the contact junction which acts as a bottle - neck .
( c ) i - v curves for arrays with junction resistances randomly assigned varying between @xmath116 ( upper curve ) , @xmath117 ( middle curve ) and @xmath118 ( bottom curve ) .
the different slopes are due to different resistances at the bottle - neck junction .
( d ) i - v curves of disordered arrays with homogeneous contact resistance .
the @xmath119 curves correspond to different realizations of disorder .
the even - odd effect present for @xmath114 in the clean case has disappeared as just one contact junction acts as bottle - neck.,width=3 ] the linearity above but very close to threshold can be easily understood .
charges can enter only through the contact junctions from the leads .
the current through the array is equal to the average charge transferred per unit time .
the average time necessary to transfer a charge through the array is the sum of the time involved in all the processes in the sequence of tunneling events from the moment in which charge enters the array from one electrode until it leaves the array to the other one . if the time associated to tunneling at a given junction is much larger than the time involved in the rest of processes , this junction acts as bottleneck and the time necessary to transverse the array is approximately equal to the inverse of the scattering rate through this junction .
thus , the current can be approximated by the tunneling rate across the bottle - neck junction .
below threshold , but close to it , the tunneling at the contact junctions costs finite energy and transport is suppressed .
this cost in energy is reduced by the applied bias voltage and at threshold is zero at the entrance junction .
when @xmath120 but very close to it , the junction from which the charge enters the array acts as bottle - neck for transport , once it is allowed .
at zero temperature , the tunneling rate is @xmath121 .
the dependence on bias voltage of the energy for tunneling compared to the one at threshold , at the source and drain junction is , respectively @xmath122 and @xmath123 .
if charge enters through a single junction , the current approximately equals @xmath124 if charge enters from the source , and @xmath125 if holes enter unto the array from the drain . both source and drain junctions
have to be taken into account in the clean symmetrically biased case when @xmath2 is odd .
@xmath126 when @xmath2 is even and @xmath114 it is necessary that charge enters through both junctions for current to flow and current is approximately equal to @xmath127 this behavior is observed in fig . 2 . the linear behavior is clearly appreciated in both the log - log scale in which the main figures are plotted as well as in the constancy of the derivatives in the insets .
the dependence of the slope of the i - v curves is better seen in the conductance plotted in the insets . as seen in ( a )
the @xmath128 is equal for the @xmath114 , @xmath129 and @xmath130 curves , i.e. independent on the array length . on the contrary ,
it is double for @xmath131 .
this behavior originates in the even - odd alternancy predicted by eqs.([ultimaeq ] ) and ( [ nuevaeq ] ) .
the dependence on @xmath1 is studied in fig.2(b ) . from top to bottom , the change of slope with @xmath1 is smooth and given by @xmath132 as predicted by ( [ primeq ] ) , but turns non monotonously if currents starts being controlled by ( [ segundaeq ] ) with slope @xmath133 . as also expected from the discussion above the slope
is affected by disorder in resistances in fig .
2(c ) , but not by charge disorder in fig .
2(d ) , except in the lose of the even - odd effect present in the clean case for @xmath114 .
the linear behavior in fig .
2 appears for several orders in magnitude . in spite of this
, it disappears for @xmath134 or @xmath113 .
the magnitude of the current is probably too small for this linearity to be detected experimentally .
we have found that linearity gives rise to sublinear behavior when the time of the other tunneling processes become relevant compared to the time spent at the bottle - neck . upon increasing @xmath135 the tunneling rates of the different processes involved in the transport
become more homogeneous . to obtain sublinear behavior , it is just necessary than the two slowest processes in a sequence have comparable rates .
the lost of linearity can depend on the resistance of the junctions when a non bottle - neck junction has a resistance much larger than the bottle - neck one .
the bottle - neck character of a process at a junction at the leads can disappear faster for longer arrays as there are more tunneling processes which will contribute to the total time . in the disordered case , the energy gain of some of the tunneling processes is smaller than in the clean case and the contact junctions can stop being the bottle - neck earlier , i.e. for smaller @xmath105 .
but , in general we have not found very significative differences for different array parameters in the value of the bias voltage at which the linearity disappears . as the energy of a tunneling process through an inner junction
does not depend on the applied voltage ( except via the charge accumulated on it ) , when a bulk junction controls the transport , the current is independent on voltage showing a characteristic staircase profile .
the existence of the coulomb staircase has been known for a long time@xcite .
early claims reported a coulomb staircase only in the asymmetrically biased case@xcite .
more recent results in clean capacitively coupled nanoparticle arrays , show that a staircase also emerges in a symmetric array under symmetric bias@xcite , but claim that the i - v characteristic for an n - dot array under forward bias is identical to that for a 2n - dot one under symmetric bias .
we show here that while the appearance of the staircase is generic , the last statement is not correct .
the current has kinks at those voltage values which change the maximum number of charges which can be accumulated at the first or last island , allowing new transport processes .
these voltage values depend on the asymmetry @xmath1 and the existence of charge disorder in the array but not on resistance disorder . to allow the addition of an extra charge in the
first or last nanoparticle requires an increase in bias voltage in the adjacent electrode of approximately @xmath136 . in the clean case , when @xmath137 only one electrode changes its potential and the width of the steps in bias voltage is @xmath136 . on the other hand ,
when @xmath54 the change in potential of a given electrode is just the half of the bias voltage and steps appear in intervals of @xmath138 . with charge disorder and @xmath55
the position of the kinks slightly depends on voltage , but the width of the voltage intervals between the kinks does not change , as new charges are added through a single junction .
if @xmath54 , charges enter from both contact junctions but the corresponding kinks in the current do not appear at the same position .
while the width of a kink corresponding to a given junction remains @xmath138 in a general case in the i - v characteristic there will be two kinks in each @xmath138 interval in bias voltage due to the alternative position of the kinks of both contact junctions . except in very special cases the separation of kinks does not do equal @xmath136 .
the position of the kinks in the clean and disorder cases can be observed in fig .
two main features can be observed . for clean arrays ( insets in ( a ) and ( b ) ) at the onset
the current shows a big jump . once the charge gradient is created and charge can transverse the array , it flows easily .
the steps at higher voltages are much smaller but have the width predicted above . in the main figures corresponding to disordered arrays , the large big jump has disappeared and steps are more clearly observed .
two stepwidths , which add @xmath139 are seen for @xmath114 and just steps with width @xmath140 appear for @xmath56 . and @xmath56 .
the current shows a big jump at threshold and very weak small steps at higher voltages.the width of the steps is @xmath138 in ( a ) and @xmath136 in ( b ) .
the staircase structure is completely washed out with increasing bias voltage .
main figures in ( a ) and ( b ) show i - v curves for arrays with charge disorder but homogeneous resistances .
@xmath114 in ( a ) and @xmath56 in ( b ) . the first step height is reduced and the staircase structure is more pronounced than in the clean case because tunneling processes in the bulk can have small energy gain and become the bottle - neck more easily .
the width of the steps in ( b ) remains equal to @xmath136 as in the clean case . in ( a )
there are two different step lengths which alternate and sum to @xmath141 , corresponding to the entrance of charge through both electrodes .
the i - v curves in ( c ) correspond to arrays with resistance disorder and @xmath114 without charge disorder in the inset , and with charge disorder in the main figure .
disorder in resistance in general favors that bulk junctions act as bottle - neck for the current , but for @xmath114 it affects differently to the steps associated to any of the contact junctions.,width=3 ] previously , in the analysis of the threshold voltage , we saw that while the value of @xmath1 has an effect the value for the onset of current , it does not seem possible to determine @xmath1 from a measurement of @xmath0 in the , experimentally relevant case , of a disordered array .
however if the onsite interaction case discussed here can be experimentally reproduced , the width of the steps in the i - v curve can differentiate @xmath1 values .
disorder in the resistances does not modify the voltages at which kinks in the current appear but it does affect the staircase profile .
the staircase profile is modified in fig .
2(c ) , compared to ( a ) due to the disorder in resistances .
a very large resistance in a bulk junction can sharpen the steps , as it creates a bottleneck for the current at a junction with an associated energy for tunneling which does not directly depend on bias voltage , but the opposite behavior can also take place if the large resistance if found at any of the contact junctions . note that the particular way in which the i - v curve is affected by disorder in resistance depends on the particular resistance and charge - disorder distributions . at larger voltages linear dependence
is recovered .
the asymptotic linear i - v does not extrapolate to zero current at zero voltage , but it cuts the zero current axis at a finite offset voltage .
this high - voltage linear regime can be understood analytically .
we obtain that the slope of the i - v curve depends only on the sum of the junction resistances in series and the offset voltage is given by the sum of the excitonic energies of all the junctions .
contrary to what was found at low voltages , close to threshold , the offset voltage for purely onsite interactions recovers the zero inter - island capacitive coupling value@xcite calculated starting from a finite value of the inter - island capacitance . to assist comparison with experiments
we compute the voltage at which this linear regime is reached . for clean and charge - disordered arrays .
both curves approach the same asymptotic curve at high voltages , even if the threshold voltage is quite different .
theoretical prediction is included for comparison.(b ) computed i - v curves for charge - disordered arrays with different junction resistances ( solid lines ) with their theoretical asymptotic predictions ( dashed lines ) . from top
to bottom a curve corresponding to an array with all - equal junction resistances , an array with randomly assigned resistances and an array with all - equal random resistances , except the first one which is ten times @xmath60 .
the slope differs but all the curves have equal offset voltage .
( c ) slope ( main figure ) and offset voltage ( inset ) which give better fitting to the numerically computed current at high - voltages as a function of the number of islands in the array , for several array parameters , all of them with homogeneous junction resistances . for comparison the theoretical prediction ( dashed - line ) is included .
the offset voltage is proportional to the number of islands and given by @xmath142 , while the slope goes like the inverse of the sum of the junction resistances added in series .
it is inversely proportional to the number of junctions , when all resistances are equal .
independence on the value of @xmath1 and the presence or absence of charge disorder is observed .
( d ) voltage at which the high - voltage asymptotic behavior is reached , estimated as the value at which @xmath143 is smaller than a given value , 1% and 5% in the figures.it is slightly larger for disordered arrays , increases linearly with the number of islands and it is approximately three and 2.5 times the offset voltage . in large arrays
a very large bias voltage can be required to reach this linear dependence.,width=3 ] at very high voltages , the charge gradient ensures that all the tunneling processes to the right decrease the energy .
the corresponding tunneling rates are @xmath144 and the total tunneling rate @xmath145 for no resistance disorder is @xmath146 .
having in mind that @xmath147 , @xmath148 .
this rate is independent on the selected tunneling process . to transfer a charge from the source to the drain
requires in average @xmath149 tunneling events .
the average current is thus @xmath150 this prediction is compared in fig . 4 ( a ) and ( c ) with numerical results .
the slope of the current does not depend on @xmath1 or the existence of charge disorder , but only on the number of junctions @xmath149 .
the slope is the same as obtained adding all the resistances in series .
asymptotically , at low voltage , this curve cuts the @xmath151 axis at the offset voltage @xmath152 .
this value is , in general , different to the threshold voltage and independent of the resistance of the junction and the asymmetry of the bias voltage .
previous derivation is valid even if there is inhomogeneity in the value of island capacitances but relies on the homogeneity of the junction resistances through the array .
if this is not the case the total tunneling rate of each step in a sequence is @xmath153 .
one could argue that on average , the charge gradient would be such that it ensures a uniform tunneling rate @xmath154 through all the junctions .
the potential drop which gives such a tunneling rate is @xmath155 and @xmath156 with @xmath157 .
there are @xmath149 possible tunneling events at each step in a sequence and @xmath149 steps , thus both @xmath149 factors cancel our .
the resulting average current is @xmath158 as in the uniform resistance case , the slope in the current corresponds to the addition in series of all the resistances . the predicted asymptotic high - voltage behavior is observed in fig . 4 for arrays with different parameters .
note that as longer is the array larger voltages have to be applied to reach this voltage .
the voltage @xmath159 at which the linear behavior is reached is estimated in fig .
it is approximately three times the offset voltage and slightly larger in the presence of charge disorder . in long arrays
@xmath159 can become very large and the linear behavior will not be easily reached experimentally .
the potential through the array can nowadays be measured@xcite , but to our knowledge it has not been studied theoretically . in conventional ohmic systems with a linear current - voltage relation @xmath160 the potential drops homogeneously through the array if the resistivity of the system is homogeneous .
when the proportionality constant between voltage and current is given by the sum of the resistances in series , but these resistances are not all equal , the voltage drop at each point is proportional to the local resistance .
the nanoparticle array i - v characteristics are highly non - linear and in general it is not obvious how the potential drops through it . at the islands the potential is the sum of the disorder and charge terms , while at the electrodes
the potential is controlled by the applied bias . as seen above , there are two linear regimes which can be identified . at high voltages
the differential conductance equals the inverse of the sum of the resistances in series .
naively , a potential drop proportional to the resistance at each junction could be expected at these voltages , but we will see that this is not exactly the case . in the low - voltage regime
the slope of the current does not correspond to the addition of the resistances in series , but it is proportional to a single one ( to the sum of two of them in the symmetric case ) and it is not clear that the potential drop should be proportional to the junction resistance . in this section
we study the potential drop through the array at low , intermediate and high voltages and show that in none of these regimes the potential drop at a given junction is strictly proportional to its resistance . in the linear regimes we show that deviations from this proportionality are related to the offset voltages and filling of the array . in the intermediate regime
the voltage drop oscillates with position and reflects correlations between the charges .
we start with the high - voltage linear regime , as it is the easiest to understand .
5 shows the average potential drop in a clean and a disordered array for a given bias voltage in the high - voltage linear regime .
all the junction resistances are equal in the top figures .
the average voltage drop is equal in both the clean and disordered case , and at first sight it seems linear . a linear potential drop through the array implies a homogeneous average junction potential drop@xmath161 .
however , at the contact junctions @xmath162 is approximately @xmath163 times smaller than at the bulk junctions . the voltage drop at each junction is not equal to the current divided by the junction resistance either , as could be naively expected .
the reasons for these deviations can be found in the offset voltage in the i - v curve and in eq .
( [ change ] ) which gives the change in energy for tunneling . at high voltages
the i - v curve is linear but the total voltage drop through the array does not equal @xmath164 .
as , seen in equation ( [ currenthigh ] ) , there is an offset voltage @xmath165 .
this offset voltage reflects the excitonic energy cost for tunneling .
the excitonic energy is not equal at each junction .
it is @xmath136 at the bulk junctions and approximately @xmath163 at the contact ones .
only the extra potential drop @xmath166 at each junction gives a finite contribution to current through it . on average @xmath167 from current conservation at high voltages the average potential drop through the array @xmath168 it
is not affected by the presence of charge disorder in the array ( but it would change if capacitances are not homogeneous , via @xmath169 ) . as observed in fig . 5
, eq ( [ potdrophigh ] ) gives a good estimate of the potential drop .
the validity of eq([potdrophigh ] ) is better seen when @xmath170 is plotted .
it is proportional to the resistance of the junction and equal in every junction if all resistances are the same .
this statement is valid independently on the position of the resistance , as shown in the bottom figure of fig . 5 and its inset and on the asymmetry @xmath1 of the applied voltage ( not shown ) .
the dependence of @xmath162 on the junction resistance is easily understood .
the tunneling probability through a junction depends on its resistance .
it is inversely proportional to it .
when the resistance is very large , the charge has a lesser tendency to jump from an island to its neighbor and it will spend more time in the island producing a dependence of the time - averaged potential drop on the junction resistance distribution .
as a function of position for a disordered array with @xmath129 , @xmath114 and all junction resistances equal .
the average potential drop through the array is close to be linear .
as shown in the inset the average potential drop at the junctions @xmath162 is homogeneous only once the excitonic energy is substracted .
the value subtracted is smaller at the contact junctions where @xmath171 is smaller .
error bars give an estimation of the fluctuations of the potential drop .
bottom : main figure ( inset ) show the average potential drop , with error bars giving its root mean square , at the junctions with the excitonic energy substracted corresponding to a clean array with the first ( middle ) junction ten times larger than the rest .
@xmath172 is proportional to the junction resistance .
note that this proportionality holds only once the excitonic energy is substracted .
, width=3 ] as seen above , in this high - voltage linear regime the current can be obtained from the average tunneling rate and correspondingly from the average potential drop .
deviations of the average value @xmath173 , i.e. the root mean square ( r.m.s . ) , are shown in the inset in fig.5 , in the form of error bars .
they are slightly smaller at the contact junctions as the potential at the electrodes is restored to its nominal value via a battery prior to any tunneling event and larger at junctions with a larger resistance .
fluctuations in the local voltage drop @xmath174 increase with applied bias voltage as the number of possible charge states and the width of the distribution of hopping energies do .
fluctuations are larger at those junctions with a larger resistance , but @xmath175 is smaller . close to threshold the current depends linearly on @xmath105 . here
we show that in this linear regime the average potential drop mainly reflects the charge state of the array at threshold .
this charge state depends on the asymmetry of the voltage drop @xmath1 and disorder and in a symmetrically biased clean array on the even or odd number of islands . for @xmath56 , charges enter from the source and @xmath176 bulk junctions prevent charge motion .
if an electron reaches the last nanoparticle , it can freely jump onto the drain at zero potential .
there is no charge gradient at the drain junction .
consequently , the potential drop at this junction vanishes at threshold . on the contrary , at the @xmath176 bulk junctions there is a charge gradient equal to unity , with the corresponding potential drop @xmath136 . to allow current @xmath177 equals the excitonic energy of the first junction , approximately equal to @xmath163 .
close to threshold , as the bottle - neck for the current is the entrance of electrons from the source the charge state of the array is most of the time equal to the one at threshold , and only perturbed by the fast passage of charges .
the average potential drop , plotted in fig .
6(a ) for a clean array with all junction resistances equal is almost the same as the static potential drop at threshold . .
notice that the potential drop vanishes at the last junction , as the source is at zero potential and there is no charge gradient at this junction .
at the first junction is equal to the excitonic energy , which at a contact junction equals @xmath163 .
( b ) average potential drop for a symmetrically biased n=49 array in main figure ( n=50 in the inset ) with the excitonic energy substracted .
once the excitonic energy , associated to the charge gradient at threshold has been substracted the average potential drop almost vanishes for @xmath178 .
the homogeneous and positive value for @xmath129 reflects that every junction is uncharged with equal probability .
( c)average potential drop at the junctions @xmath162 corresponding to a disordered array with @xmath129 islands and @xmath54 .
( d ) main figure ( inset ) @xmath179 corresponding to a @xmath129 clean array with the middle ( first ) junction resistance 10.8 larger than the other ones .
when the larger resistance is in the middle its effect on the average potential drop at these small voltages is barely visible and the potential drop almost equals the one found in the inset in ( b ) .
this is the generic behavior found with resistance disorder very close to threshold .
a special case is a n - even clean @xmath114 array with the first resistance larger than the other ones , shown in the inset .
the average potential drop differs considerably with respect to the one found in the inset in ( b ) .
the presence of the larger resistance at a contact junction modifies the average charging of the array and selects the opposite contact junction as the one which lacks charge gradient .
, width=3 ] for @xmath54 and odd number of particles , at threshold the charge gradient and charge potential drop in a clean array are respectively one and @xmath136 at the bulk junctions . at the contact
junctions the potential drop is @xmath163 . as in the @xmath56 case
discussed above , the average potential drop in the linear regime close to threshold , is very close to the one found at threshold , which equals the excitonic energy at each junction , @xmath180 , what can be seen in the main figure in fig .
when the number of particles is even , the charge gradient at one of the junctions vanishes . as shown in the inset of fig .
6(b ) , @xmath181 in the bottleneck regime is positive and equal for all the junctions .
this reflects that every junction is uncharged with equal probability . in the disordered case ,
only those junctions with upward steps in the disorder potential are charged , and this is reflected in the average potential drop , in fig.6 which adds disorder , charge and bias potential .
the threshold voltage does not depend on the resistance of the junctions , but the flow of charge does .
this is reflected on the average potential drop on such very small scale that even if the threshold voltage potential drop is substracted at each junction and for reasonably large changes in resistance it is not visible ( see main figure in fig .
this is different to the dependence observed in the high - voltage regime . for extremely large values of the resistance disorder a weak effect on the average voltage at bias close to threshold
can be seen ( not shown ) .
in this case , the potential drop at a junction with a larger resistance is slighltly larger than at the rest . at the adjacent junctions it is slightly smaller , what reflects the average charge state of the nanoparticles joined by the large resistance .
the r.m.s of the junction potential drop at low bias voltages are very small in some of the cases analyzed , of order @xmath182 in main figures in fig .
6(a ) and ( b ) .
error bars are small because most of the time the array is at the threshold charge state .
this is not the case in the inset in 6(b ) where fluctuations are or the order of @xmath183 , or in the inset in 6(d ) where they are small everywhere except at the last two junctions where they are of order @xmath163 .
disorder also increases the fluctuations in the average voltages to values of the order of the charging energy . the most interesting regime to analyze the voltage drop
is at intermediate voltages where the i - v curve show the coulomb staircase . for the case of a clean array with
capacitively coupled nanoparticles stopa@xcite showed that the steps in the i - v characteristic correspond to alternation of the charge density between distinct wigner crystalline phases . in our case , the interaction is short - range , but the possibility of such a wigner state with charges periodically ordered to minimize their repulsion , if present , could be observed via the potential drop .
periodic charge ordering should lead to oscillations in the potential drop along the array .
such an observation would be a clear evidence of correlated motion . at the junctions as a function of position at several values of the bias voltage , for which the current is in the coulomb staircase regime , corresponding to a clean @xmath129 array and @xmath54 . from top to bottom @xmath184 curves have been vertically displaced to avoid overlap .
the potential drop show almost regular oscillations which reflect a stationary ( but not static ) charge ordering , resembling the one expected in a wigner crystal ( see discussion in ref.@xcite ) .
the number of maxima and minima does not change within a step and increases in two ( for @xmath54 ) from one steps to the next one .
the position of maxima is slightly adjusted within a step .
they start appearing close to the electrodes and reflect the entrance of charge from them .
, width=3 ] the average potential drop ( with the excitonic energy subtracted ) through the array for several voltages corresponding to clean n=50 nanoparticle arrays is shown in fig .
clear oscillations are seen . comparing the values of the bias voltage chosen with the position of the steps in the corresponding i - v curve in fig .
7 , the number of maxima / minima in the potential drop do not change in a given step in the coulomb staircase . for symmetrically biased arrays
, they increase in pairs from a step to the next one . for odd / even number of particles ,
there is always a minimum / maximum at the center of the array .
the other maxima and minima tend to be as equally spaced as possible , but this is not exact .
inconmensurability beween the period of the oscillations and the lattice can distort equal spacement . also , when new maxima or minima appear they are closer to the source and drain electrodes and move inwards , producing a movement of the other maxima and minima , with increasing voltage .
this can be taken as a finite size effect of the wigner crystal state . as the number of charges in the array increase with increasing bias voltage the amplitude and period of the oscillations decreases , approaching the high - voltage regime for which @xmath181 is homogeneous .
islands , @xmath54 and no resistance disorder at intermediate voltages .
the average potential drop at the contact junction shows clear oscillations as a function of the bias , while the potential drop at the center of the array depends monotonously on @xmath135 .
( c ) and ( d ) show @xmath185 for charge and resistance disordered 49-islands arrays respectively and @xmath54 .
the coulomb staircase is much more pronounced in the presence of disorder . however , as shown here , the oscillatory potential drop structure found in fig . 7 and characteristic of wigner - crystal like physics is destroyed . ,
width=3 ] the anomalous potential drop can be also seen in the potential drop at a given junction as a function of the bias voltage , shown in fig 8(a ) and 8(b ) for junctions 1 and 25 for a clean symmetrically biased 50-islands array at the first junction the potential drop show clear oscillations as a function of the bias voltage , which reflect that new charges state at the first island are allowed .
the potential drop increases until an extra charge can be accumulated at the first nanoparticle , for larger voltages the average occupation of first island increases and the voltage decreases smoothly until a new value at which it increases again as the increase in occupation of first island can not compensate the increase in the electrode potential .
oscillations , but less regular and less pronounced due to the movement of maxima and minima discussed above , are also observed at intermediate junctions .
potential drop is much more homogeneous at the middle of the array , where there is always a minimum ( or maximum ) in the potential drop .
charge or resistance disorder alters the charge motion and frustrates the formation of this wigner crystal like state , as seen in fig .
8(c ) and 8(d ) .
this is the opposite behavior that would be naively expected if one just associates the appearance of plateaux with the oscillations in voltage drop and emphasizes that the step profile is just a consequence of the dependence on the bias voltage of the tunneling rate of the processes which control the current .
the r.m.s . of the junction potential drop is larger than at low voltages and smaller than at high voltages .
it is of the order of the excitonic energy and reflects the variation in occupation of the island .
it slowly increases with voltage .
in this paper we have studied , analytical and numerically , the current through an array of @xmath2 metallic islands placed in between two large electrodes , the source and the drain at voltages @xmath21 and @xmath22 respectively . the applied bias voltage @xmath135 satisfies @xmath186 .
interactions are restricted to those charges in the same conductor .
the capacitive coupling between different conductors vanishes .
the nanoparticle level spacing is assumed negligible and transport is treated at the sequential tunneling level . in the model
used we take into account that the electrodes are not ideal voltage sources , but have a finite self - capacitance .
this means that the voltage at the electrodes fluctuates in response to tunneling processes , but we assume that prior to the next tunneling event the nominal voltage is restored . due to the large value of the electrode capacitance numerical results
are barely modified and its effect is neglected in the discussion .
the probability of each process depends on the change in energy involved , which can be written as the energy to create an electron - hole pair plus the difference in potential between the sites involved in the tunneling , see eq.([change ] ) . we have analyzed and clarified how the transport properties depend on the number of particles , the presence of charge or resistance disorder and the bias voltage , including how symmetrically it is applied .
to quantify this symmetry we have introduced a parameter @xmath1 as @xmath51 .
we have shown that in the purely onsite interaction limit the dependence of the threshold voltage @xmath0 of clean arrays on the number of nanoparticles differs qualitatively of the dependence predicted for weakly coupled islands , as in the onsite case studied here a charge can not flow freely through an empty array .
for symmetrically biased arrays @xmath0 equals @xmath187 for odd @xmath2 and @xmath188 for even @xmath2 .
the even - odd effect disappears for forward biasing ( @xmath137 ) and @xmath189 , see fig .
the threshold voltage is not affected by disorder in the junction resistances but it depends on the selected disorder configuration if charge disorder is present . with charge disorder ,
the average threshold voltage is independent on @xmath1 and we recover previously predicted values @xmath190 , see fig .
1(b ) . at voltages marginally close to threshold ,
current is linear on @xmath106 with a slope independent on the number of particles @xmath2 but which depends on the resistance of the contact junctions and on the bias asymmetry @xmath1 , see eqs .
( [ primeq ] ) to ( [ ultimaeq ] ) and fig .
2 . this dependence has been obtained both numerical and analytically and resolves previous controversy on the power - law close to threshold .
it reflects that the junction through which charges enter into the array acts as a bottle neck .
the range of voltages at which this linear dependence holds is probably too small to be observed experimentally .
linearity is lost when the scattering rate for tunneling through the contact junctions is comparable to other tunneling processes scattering rates .
the linear regime is followed by a coulomb staircase at intermediate voltages .
the width of the steps depends on @xmath1 and on the presence of charge disorder . for clean arrays
the bias voltage step width is @xmath136 for forward bias and @xmath138 for symmetric bias .
the stepwidth changes if disorder is present but still depends on the value of @xmath1 .
the staircase profile depends on the junction resistances values
. see fig .
3 . at high voltages
current depends linearly on bias voltage .
the asymptotic i - v characteristic is given by ( [ currenthigh ] ) and cuts the zero current axis at a finite offset voltage , see fig .
the slope of the asymptotic linear i - v is given by the inverse of the sum of the junction resistances in series and the offset voltage takes into account charging effects .
the high - voltage linear behavior is reached for bias voltages @xmath159 approximately three times larger than the offset .
@xmath159 can be very large long arrays .
we have studied the potential drop through the array and showed that in none of the transport regimes studied it drops completely linearly through the array . in the low - voltage linear regime
the average potential drop mainly reflects the charge gradient created at threshold , see fig . 6 .
the effect of disorder in resistances is extremely weak , except for symmetrically biased arrays with even number of particles . in the coulomb staircase regime
the potential drop along the array show almost periodic oscillations in disorder - free arrays which reflect the correlated motion and the tendency to form a wigner - crystal like state , see fig
such periodicity is destroyed by charge or resistance disorder in fig .
8 . in the high - voltage regime
the ohmic dependence , and the associated proportionality between the potential drop and the junction resistance , is only recovered once the excitonic energy is substracted , as seen in ( [ potdrophigh ] ) and fig .
the mean value of the potential drop serves to compute the i - v characteristic in this regime .
we numerically determine the time evolution of the state of an array of nanoparticles sandwiched between two large metallic leads symmetrically biased at potentials @xmath21 and @xmath22 , with @xmath191 .
the state of the array consists of the set of charges @xmath192 that occupy the array islands and the leads .
the island charges take on integer values .
the charge of the nanoparticles is modified when an electron tunnels between two adjacent sites .
the charges of the source and drain can take on any real value because they can be modified discretely via the tunneling of charges or continuously via the charging of the leads by a battery .
the evolution of the system is computed by means of a kinetic monte carlo simulation@xcite . at each iteration
a single tunneling event takes place .
the time involved in this event @xmath193 depends on the tunneling rates of all the possible tunneling processes .
each iteration starts from an initial charge configuration .
first , it is computed the change in energy and the tunneling rate of the @xmath194 possible hopping events , corresponding to the tunneling of a single electron , to the left or to the right , through any of the @xmath149 junctions .
the probability of changing the initial configuration varies with time like @xmath195 with @xmath196 the time at which the preceding tunneling process took place and @xmath197 . @xmath198 and @xmath199 are the tunneling rates through the @xmath27 junction to the left or to the right , respectively , and are calculated from ( [ tunnrate ] ) . to sample the time interval between two hopping events we generate random numbers between @xmath200 $ ] to mimic @xmath201 and obtain @xmath202 from ( [ pmove ] ) .
as the average of @xmath203 is the unity , if one is interested only in the average values of the charge or the current , and not on its fluctuations , the time step could be fixed@xcite to @xmath204 .
this option is numerically faster .
the relative probability of each tunneling event is @xmath205 . to determine the hopping process which changes the charge state , the relative probabilities are consecutively arranged in the interval @xmath200 $ ] . a second random number in this interval is generated to select the tunneling process .
then , the charge configuration is updated .
after we modify the state of the system , we allow the external circuit to return the leads to their applied bias values prior to the selection of the next hop . this effect is simulated by resetting the charges on the source and the drain to the values that restore the nominal applied bias . in order to remove all sensitivity to initial conditions ,
before we track the evolution of @xmath206 as a function of time at any voltage , we perform @xmath207 iterations to equilibrate the system .
following these iterations , we track the evolution of the charge state until the net number of electrons that arrive at the drain , @xmath208 , equals a very large number ( @xmath209 ) . the average calculated current
is given by @xmath210 where @xmath211 is the sum of all time intervals between hopping processes in the evolution runs .
if a tunneling event involving the drain is selected , an amount @xmath212 is added to @xmath208 depending on whether an electron hopped to or from the drain .
current conservation ensures that the average current is the same through any junction .
the minimum numbers of equilibration cycles , @xmath213 , and evolution cycles ( set by @xmath208 ) depend on the voltage .
financial support from the swiss national foundation , nccr manep of the swiss national fonds , the spanish science and education ministry through ramn y cajal contract and fis2005 - 05478-c02 - 01 grant and the direccin general de universidades e investigacin de la consejera de educacin de la comunidad de madrid and csic through grant 200550m136 is gratefully acknowledged . work at ut austin was supported by the welch foundation by the nsf under grant dmr-0606489 and by the aro under grant w911nf-07 - 1 - 0439 .
99 r.l.whetten , j.t .
khoury , m.m .
alvarez , s.murthy , i. vezmar , z.l .
wang , p.w .
stephens , c.l .
cleveland , w.d .
luedtke and u. landman , adv .
mater . * 8 * , 428 ( 1996 ) . c.p .
collier , r.j .
saykally , j.j .
shiang , s.e .
henrichs , j.r . heath ,
science * 277 * , 1978 ( 1997 ) .
g. markovich , c. p. collier , s.e .
henrichs , f. remacle , r.d .
levine and j.r .
heath , acc .
32 * , 415 ( 1999 ) .
r. parthasarathy , x - m lin and h. m. jaeger , phys .
lett . * 87 * , 186807 ( 2001 ) j.d .
le , y. pinto , n.c .
seeman , k. musier - forsyth , t.a .
taton and r.a .
kiehl , nanoletters * 4 * , 2343 ( 2004 ) .
y. lin , a. bker , j. he , k. sill , h. xiang , c. abetz , x. li , j. wang , t. emrick , s. long , q. wang , a. balazs and t.p .
russell , nature * 434 * , 55 ( 2005 ) .
k. elteto , x .-
m . lin and h.m .
jaeger , phys .
b * 71 * , 205412 ( 2005 ) .
j. zhang , y.liu , y. ke and h. yan , nanoletters * 6 * , 248 ( 2006 ) . t.p .
bigioni , x .-
m.lin , t.t .
nguyen , e.i .
corwin , t.a .
witten and h. m. jaeger , nature materials * 5 * , 265 ( 2006 ) .
j. liao , l. bernard , m. langer , c. schnenberger and m. calame , adv . mater.*18 * , 2803 ( 2006 ) .
( ojo comprobar si es 2803 o 2444 ) .
s.s . mark , m. bergkvist , x. yang , l.m .
teixeira , p. bhatnagar , e. r. angert and c.a .
batt , langmuir * 22 * , 3763 ( 2006 )
murray , c.r .
kagan and m.g .
bawendi , science * 270 * , 1335 ( 1995 ) .
d.v.talapin , e.v .
shevchenko , a. kornowski , n. gaponik , m. haase , a.l .
rogach and h. weller , adv .
* 13 * , 1868 ( 2001 ) .
morgan , c.a .
leatherdale , m. drndc , m.v .
jarosz , m.a .
kastner and m. bawendi , phys .
rev . b*66 * , 075339 ( 2002 ) .
v.j . porter , t. mentzel , s. charpentier , m.a .
kastner and m.g .
bawendi , phys .
b * 73 * , 155303 ( 2006 ) .
d. yu , c. wang , b.l .
wehrenberg and p. guyot - sionnest , phys
lett . * 92 * , 216802 ( 2004 ) .
h. romero and m. drndic , phys .
rev . lett . * 95 * , 156801 ( 2005 ) .
t. feng , h. yu , m. dicken , j.r . heath and h.a .
atwater , app .
. lett . * 86 * , 033103 ( 2005 ) .
s. sun , c.b .
murray , d. weller , l. folks and a. moser , science * 287 * , 1989 ( 2000 ) .
black , c.b .
murray , r.l . sandstrom and s.
sun , science * 290 * , 1131 ( 2000 ) v. f. puntes , p. gorostiza , d.m .
aruguete , n.g .
bastus and a.p .
alivisatos , nature materials * 3 * , 263 ( 2004 ) .
redl , k .- s .
cho , c.b .
murray and s. obrien , nature ( london ) * 423 * , 968 ( 2003 ) .
a.e . saunders and b. a. korgel , chem .
phys . chem .
* 6 * , 61 ( 2005 ) .
schevchenko , c.v .
talapin , n.a .
kotov , s. obrien and c.b .
murray , nature ( london ) * 439 * , 55 ( 2006 ) . , _ nato asi series , vol b _ , 294 , eds . h. grabert and m.h .
devoret . , new york , plenum press 1992 .
p. sheng and b. abeles , phys .
* 28 * , 34 ( 1972 ) .
p. sheng , b. abeles and y. arie , phys .
* 31 * , 44 ( 1973 ) .
j.s . helman and b. abeles ,
lett . * 37 * , 1429 ( 1976 ) .
shiang , j.r .
heath , c.p .
collier and r.j .
saykally , j. phys .
b * 102 * , 3425 ( 1998 ) . s .-
kim , g. medeiros - ribeiro , d.a.a .
ohlberg , r.s .
williams and j.r .
heath , j. phys .
b * 103 * , 10341 ( 1999 ) .
sampaio , k.c .
beverly and j.r .
heath , j. phys .
b , * 105 * , 8797 ( 2001 ) i.s . weitz , j.l .
sample , r.ries , e.m .
spain and j.r .
heath , j. phys .
chem b * 104 * , 4288 ( 2000 ) .
quinn , i. prieto , s. k. haram and a.j .
bard , j. phys .
b * 105 * , 7474 ( 2001 ) .
doty , h. yu , c.k .
shih and b.a .
korgel , j. phys .
b * 105 * , 8291 ( 2001 ) .
a. courty , a. mermet , p.a .
albouy , e. duval and m.p .
pileni , nature materials * 4 * , 395 ( 2005 ) .
a.a . middleton and n.s .
wingreen , phys .
rev . lett . * 71 * , 3198 ( 1993 ) .
matsuoka and k.k .
likharev , phys .
b * 57 * , 15613 ( 1998 )
kaplan , v.a .
sverdlov and k.k .
likharev , cond - mat/0303477 .
m.shin , s. lee , k.w . park and e.h .
lee , phys .
lett . * 80 * , 5774 ( 1998 ) .
kaplan , v.a .
sverdlov and k.k .
likharev , phys .
b * 65 * , 193309 ( 2002 ) .
kaplan , v.a .
sverdlov and k.k .
likharev , phys . rev .
b * 68 * , 045321 ( 2003 ) .
kinkhabwala , v.a .
sverdlov , k.k .
likharev , j , of phys . condens .
matter * 18 * 2013 ( 2006 ) .
r.kotliar and s. das sarma , superlatt . and microst .
* 20 * , 641 ( 1996 ) .
beloborodov , a.v .
lopatin , v.m .
vinokur and k.b .
efetov , rev .
79 * , 469 ( 2007 ) .
beloborodov , a. glatz and v.m .
vinokur , phys .
b * 75 * , 052302 ( 2007 )
tran , i.s .
beloborodov , x.m .
lin , t.p .
bigioni , v.m .
vinokur and h.m .
jaeger , phys .
lett . * 95 * , 076806 ( 2005 ) .
a. atland , l. glazman , a. kamenev , phys .
lett . * 92 * , 026801 ( 2004 ) .
meyer , a. kamenev and l.i .
glazman , phys .
b * 70 * , 045310 ( 2004 ) .
y.l . loh , v. tripathi and m. turlakov , phys .
b * 72 * , 233404 ( 2005 ) .
efetov and a. tschersich , europhys . lett . * 59 * , 114 ( 2002 ) . j. zhang and b.i .
shklovskii , phys .
b * 70 * , 115317 ( 2004 ) .
r. parthasarathy , x .-
lin , k. elteto , t.f .
rosenbaum and h.m .
jaeger , phys .
* 92 * , 076801 ( 2004 ) .
k. elteto , e. g. antonyan , t. t. nguyen , and h. m. jaeger , phys .
b * 71 * , 064206 ( 2005 ) .
beverly , j.f .
sampaio and j.r .
heath , j. phys .
chem b * 106 * , 2131 ( 2002 ) .
f. remacle , c.p .
collier , g. markovich , j.r .
heath , u. banin , r.d .
levine j. phys .
chem b * 102 * , 7727 ( 1998 ) .
f. remacle , k.c .
beverly , j.r .
heath and r.d .
levine , j. phys .
chem b * 106 * 4116 ( 2002 ) .
cordan , a. goltzen , y. herv , m. mejas , c. vieu and h. launois , jour . of app
. phys . * 84 * , 3756 ( 1998 ) .
a. ppin , c. vieu , m. mejias , y. jin , f. carcenac , j. gieraz , c. david , l. couraud , a. s. cordan , y. leroy and a. goltzen , app .
. lett . * 74 * , 3047 ( 1999 ) .
cordan , y. leroy , a. goltzen , a. pepin , c. vieu , m. mejias and h. launois , jour . of app
. phys . * 87 * 345 ( 2000 ) .
y. leroy , a. s. cordan and a. goltzen , jour . of app
. phys . * 90 * 953 ( 2001 ) .
bakhalov , g.s .
kazacha , k.k .
likharev and s.i .
serdyukova , sov .
jetp * 68 * , 581 ( 1989 ) .
s. semrau , h. schoeller and w. wenzel , phys .
b * 72 * , 205443 ( 2005 )
ancona , w. kruppa , r.w .
rendell , a.w .
snow , d. park and j.b .
boos , phys .
b * 64 * , 033408 ( 2001 ) .
a. bezryadin , r.m .
westervelt , and m. tinkham , app .
. lett . * 74 * , 2699 ( 1999 ) .
hu and r.f .
oconnell , phys .
b * 49 * , 16773 ( 1994 ) .
melsen , u. hanke , h.o .
mller and k.a .
chao , phys .
b * 55 * , 10638 ( 1997 ) .
berven and m.n .
wybourne , app .
. lett . * 78 * , 3893 ( 2001 ) . c. kurdak , a.j .
rimberg , t.r
. ho and j. clarke , phys . rev .
b * 57 * , r6842 ( 1998 ) .
s. jha and a.a .
middleton , cond - mat/0511094 c.reichhardt and c.j .
o. reichhardt , phys .
. lett . * 90 * , 046802 ( 2003 ) .
e. bascones , j.a .
trinidad , v. estvez and a.h .
macdonald , cond - mat some previous work on capacitance disorder can be found in@xcite y. xue and m.a .
ratner , phys .
b * 68 * , 235410 ( 2003 ) .
m. stopa phys .
b * 64 * , 193315 ( 2001 ) .
d.v . averin and k.k .
likharev , _ mesoscopic phenomana in solids _ , edited by b.l .
altshuler , p. a. lee , and r.a .
webb ( elsevier , amsterdam , 1991 ) .
m. amman , r. wilkins , e. ben - jacob , p.d .
maker , and r.c .
jaklevic , phys .
b * 43 * , 1146 ( 1991 ) .
hanna and m. tinkham , phys .
b * 44 * , 5919 ( 1991 ) .
u. geigenmller and g. schn , europhys .
* 10 * , 765 ( 1989 ) .
bakhalov , g.s .
kazacha , k.k .
likharev and s.i .
serdyukova , physica b * 173 * , 319 ( 1991 ) .
j.l . sample , k.c .
beverly , p.r .
chaudhari , f. remacle , j.r .
heath and r.d .
levine , adv .
* 14 * 124 ( 2002 ) .
in@xcite the average threshold voltage is predicted to be linear on the number of junctions , which in our case is @xmath17 .
it is straightforward to argue for both @xmath214 that @xmath215 is linear in the number of bulk islands @xmath2 as observed in fig.1(b ) . in large arrays
the difference between both predictions is negligible . d.n
tsigankov , e. pazy , b.d .
laikhtman and a.l .
efros , phys .
b * 68 * , 184205 ( 2003 ) | we analyze and clarify the transport properties of a one - dimensional metallic nanoparticle array with interaction between charges restricted to charges placed in the same conductor .
we study the threshold voltage , the i - v curves and the potential drop through the array and their dependence on the array parameters including the effect of charge and resistance disorder . we show that very close to threshold the current depends linearly on voltage with a slope independent on the array size . at intermediate bias voltages , for which a coulomb staircase is observed we find that the average potential drop through the array oscillates with position . at higher voltages
i - v curves are linear but have a finite offset voltage .
we show that the slope is given by the inverse of the resistances added in series and estimate the voltage at which this linear regime is reached .
we also calculate the offset voltage and relate it to the potential drop through the array .
nanoparticle arrays made of metallic@xcite , semiconducting@xcite , magnetic@xcite or combined@xcite materials and with radii of the order of 2 - 7 nm can be now synthesized .
the transport properties of these systems are influenced by the ratio between the energy level spacing , the charging energy of the nanoparticles , and the temperature .
the first two quantities depend on the material and the size of the nanoparticle . in the case of metallic nanoparticles , at not too low temperatures , the level spacing is much smaller than the temperature and does not play any role in the transport@xcite . on the contrary ,
the charging energy is of the order of 0.1 ev .
strong interactions between the electric charges and the possibility of tuning interparticle coupling make nanoparticles arrays an ideal system to study correlated motion@xcite .
experimentally , these arrays are strongly influenced by disorder@xcite .
local charging disorder is present in all arrays due to randomly dispersed charged impurities lodged in the substrate or in the materials that separate and surround the nanoparticles . because of the exponential dependence of the tunneling resistance , even a small dispersion in the distance between nanoparticles results in large variations in the tunneling resistances of the junctions .
differences in the island sizes and voids in the lattice can be other sources of disorder@xcite . due to the combination of disorder and charging effects the current in voltage biased arrays
is blocked up to a threshold voltage@xcite @xmath0 . for bias voltages larger than @xmath0
current is in general non - linear in voltage with a power - law dependence@xcite close to threshold , a linear dependence recovered at high - voltages and frequently a step - like behavior , called a coulomb staircase , at intermediate voltages .
most studies have focused on the statistical analysis of the threshold voltage and on the power - law behavior of the current close to this threshold .
this exponent depends on the dimensionality of the array , but there is controversy between different theoretical approaches in the one - dimensional case , with both linear@xcite and square - root@xcite predictions .
much experimental work has been concentrated on two and three dimensional arrays , but some quasi - one dimensional systems have also been fabricated@xcite .
comparison between experiments and theory is not yet well settled . in this paper
we provide a complete description of the zero - temperature transport properties of one - dimensional metallic nanoparticle arrays for the case in which interactions are restricted to charges in the same nanoparticle ( onsite limit ) . in a separate work@xcite
we discuss the effect of the long - range character of the interactions on these properties .
we resolve the controversy on the power - law , analyze carefully dependencies on different array parameters and estimate the voltage which delimits each of the transport regimes to favor comparison with experiment .
we discuss arrays with and without charge disorder .
although clean arrays are mainly of academic interest , their analysis will help us to understand the main features of the experimentally more relevant , disordered arrays .
the effect of variations in the junction resistances is also analyzed .
disorder in capacitances ( nanoparticle size variations ) is not considered as it is less important in present experiments .
nanoparticles synthesized nowadays are monodispersed in size to a few percent . in any case
, the effect of capacitance disorder@xcite in most of the properties studied here can be deduced from the analytic approximations provided in the text . due to the one - dimensionality of the array and the nearest neighbor tunneling
considered we assume that there are no nanoparticle voids in the array , as this would completely prevent current flow .
we have analyzed the threshold voltage , the i - v characteristics , close to threshold but also at larger bias voltages and the potential drop through the array . among the results presented
we show that the threshold voltage of clean arrays in the onsite case does not approach the one found for capacitively coupled junctions in the limit of weak coupling@xcite .
the controversy in the power - law dependence of current on bias voltage close to threshold is clarified and shown to be linear , but with a slope different from the one previously predicted@xcite .
we find signatures of correlated charging of the system in the potential drop through the array , especially at intermediate values of the bias voltage .
we show that even at high voltages , with the conductance controlled by the sum of the resistances in series , the potential drop through the array is not completely linear , but it has a contribution due to charging effects .
we identify an asymmetry external parameter @xmath1 which controls the bias voltage drop .
the influence of @xmath1 has barely been discussed in previous works .
we show that its effect can _ a priori _ be observed experimentally in the width of the coulomb staircase steps .
calculations are perfomed numerically , but analytic approximations are given in several limits and compared with numerical results . |
systems locating in between one dimension and two dimensions , ladders , have attracted great attention both theoretically and experimentally.@xcite this line of research has been pursued both with the hope of finding some clues to the understanding of the high - temperature superconductivity , and because of the experimental discoveries of new ladder compounds , such as a cuprate spin ladder@xcite srcu@xmath3o@xmath4 and a superconducting ladder@xcite sr@xmath5ca@xmath6cu@xmath7o@xmath8 .
another spin ladder system of current interest is the organic material cu@xmath3(c@xmath9h@xmath10n@xmath3)@xmath3cl@xmath11 whose spin - gap behavior was observed by the measurements of the spin susceptibility , the magnetization curve , and the nmr relaxation rate.@xcite the nonzero magnetization is observed once the external magnetic field @xmath12 exceeds the lower critical field @xmath1 which is equal to the spin gap ( in units where the product of the bohr magneton and the @xmath13-factor is set to be unity ) .
this is a quantum phase transition driven by the external magnetic field from the gapped spin liquid to the gapless tomonaga - luttinger liquid state .
these experiments motivated further theoretical works on the various properties of the gapless regime : the magnetization process,@xcite the spin - spin correlation functions,@xcite and the spin - peierls instability.@xcite in this paper we extend the theory of shelton _
et al._@xcite to the two - leg heisenberg ladder in the gapless regime ( @xmath14 ) and calculate the dynamical spin - spin correlation functions and structure factors in the ground state.@xcite we obtain the dynamical structure factors containing massive spin excitations as well as massless excitations .
the latter contribution is characteristic of the tomonaga - luttinger liquid and is commonly found in the @xmath15 heisenberg chain.@xcite this component may be interpreted as coming from the bose - condensate of @xmath16 magnons.@xcite on the other hand , the massive excitations we are interested in have @xmath17 magnons in their origin . below @xmath1
the massive magnons give rise to @xmath18-function peaks in the dynamical structure factor . in the gapless regime ( @xmath14 )
the @xmath18-function peaks of the massive excitations turn into power - law singularities , because the massless excitations introduce algebraically decaying prefactors to the exponentially decaying correlation functions .
an interesting feature of the antiferromagnetically coupled heisenberg ladder is that these massless and massive modes appear in different sectors of the structure factors , characterized by the momentum @xmath19 in the direction of the rungs ( @xmath19 can take only 0 and @xmath2 ) and by the momentum @xmath20 in the direction of the chains . our study of the massive excitations in the gapless regime is also motivated to better understand the @xmath2-resonance mode@xcite in the so(5 ) symmetric ladder model introduced recently by scalapino _
et al_.@xcite in this model there is a quantum phase transition driven by the chemical potential from the spin - gap mott insulator to the @xmath21-wave - like superconducting phase .
it was shown in the strong - coupling limit that the spin - gap magnon mode of the mott insulator evolves continuously into the @xmath2-resonance mode of the superconducting phase.@xcite a complementary weak - coupling approach@xcite was taken to study a general two - leg ladder model of weakly interacting electrons .
in particular , lin _ et al._@xcite showed that the model is renormalized to a fixed point where a global so(8 ) symmetry is realized that contains the so(5 ) .
they also obtained the ground - state phase diagram and low - energy excitation spectra at half filling which are qualitatively in agreement with the strong - coupling picture of the so(5 ) symmetric ladder model .
yet the energy and the spectral weight of the @xmath2-resonance mode in the dynamical spin structure factor are not completely understood even for the ideally constructed so(5 ) symmetric ladder .
an extensive numerical exact - diagonalization study is performed to clarify this and related issues,@xcite but clearly it is desirable to develop an analytic theory to discuss the power - law singularity which is expected to appear in the structure factor.@xcite the field - driven quantum phase transition in the heisenberg ladder we study in this paper is analogous to the quantum phase transition in the so(5 ) symmetric ladder .
a @xmath16 boson in the heisenberg ladder corresponds to a hole pair in the latter model , and a @xmath17 boson to a magnon or the @xmath2-resonance mode .
therefore our heisenberg spin ladder may be viewed as a toy model for the so(5 ) symmetric ladder . using the analogy
, we can deduce the correlation functions for the @xmath21-wave superconductivity , the charge density wave , and the spin correlations in the so(5 ) ladder from the spin - spin correlators of the heisenberg ladder .
this paper is organized as follows . in sec .
ii we introduce the model and analyze it in the limit of strong interchain coupling , where it is easier to get intuitive pictures of the physics . detailed calculation of the dynamical spin - spin correlation functions and the structure factors are presented in sec .
here we take the opposite limit of the weak interchain coupling , and employ the bosonization method . in sec .
iv we briefly discuss implications to the @xmath2-resonance in the so(5 ) symmetric ladder model and summarize the results .
the hamiltonian for the two - leg heisenberg ladder we study in this paper is @xmath22 where @xmath23 is spin-@xmath0 operator , and the intrachain coupling @xmath24 is positive ( antiferromagnetic ) .
the interchain coupling @xmath25 is also assumed to be positive , unless otherwise noted . in this section
we briefly discuss static correlations above the lower critical field @xmath1 in the strong - coupling limit , @xmath26 . when @xmath27 , the ladder is decomposed into independent rungs , each rung consisting of two spins @xmath28 and @xmath29 .
there are four eigenstates in each rung , the singlet @xmath30 , and the triplet states @xmath31 , @xmath32 , and @xmath33 .
their energies are @xmath34 , @xmath35 , @xmath36 , and @xmath37 , respectively .
it is then natural to introduce boson operators @xmath38 , @xmath39 , @xmath40 , and @xmath41 , which create the singlet state and the triplet states at the @xmath42th rung.@xcite they obey the constraint , @xmath43 . with these hard - core bosons
we represent the spin operators @xmath44 the suffixes @xmath45 and @xmath2 refer to the momentum @xmath19 .
note that @xmath46 and @xmath47 do not contain the @xmath48 boson . in the presence of a high magnetic field @xmath49
, we may ignore the @xmath48 and @xmath50 bosons as they have higher energies than the @xmath51 boson .
in lowest order in @xmath24 the effective hamiltonian thus becomes @xmath52 with the simplified constraint @xmath53 . as pointed out in ref .
, this effective hamiltonian may be written as @xmath54 where @xmath55 is a spin-@xmath0 operator @xmath56 we notice that eq .
( [ h_xxz ] ) is just the hamiltonian of the spin-@xmath0 @xmath57 chain with the external magnetic field @xmath58 , whose properties are well understood .
when @xmath59 , the @xmath57 chain is not fully polarized.@xcite this means that the ladder has unsaturated magnetization @xmath60 ( @xmath61 ) when @xmath62 .
we have thus obtained the lower and upper critical fields @xmath63 and @xmath64 in lowest order in @xmath65.@xcite the magnetization curve for @xmath66 can also be obtained from the known result for the @xmath57 chain:@xcite @xmath67 a similar result holds for @xmath68 .
the square - root behavior is a well - known universal behavior.@xcite when @xmath61 the ground state is a superfluid of the @xmath51 bosons . within our approximation
the following equal - time spin correlation functions are readily obtained from those of the @xmath57 chain:@xcite @xmath69 where @xmath70 s are numerical constants . for @xmath66
the parameter @xmath71 is given by@xcite @xmath72 in general @xmath73 for @xmath61 , and @xmath74 when @xmath75.@xcite thus , the leading term in @xmath76 , besides the constant @xmath77 , is @xmath78 , which reflects the fact that the hard - core bosons @xmath51 have a tendency to form a density wave with a period equal to @xmath79 . on the other hand , the correlation functions @xmath80 and
@xmath81 involve the massive @xmath48 and @xmath50 bosons , and therefore decay exponentially for @xmath82 .
we conclude that the correlators @xmath83 and @xmath84 show quasi - long - range order while @xmath85 and @xmath86 are short - ranged .
these correlation functions are discussed in more detail in the next section . before closing this section
, we make a few comments .
first , eqs .
( [ static < s^z_0s^z_0 > ] ) and ( [ static < s^x_pis^x_pi > ] ) have the same form as the correlators @xmath87 and @xmath88 of the @xmath89 heisenberg chain in a magnetic field larger than the haldane gap.@xcite
an important difference is that @xmath90 in the @xmath89 chain.@xcite we will come back to this point in the next section .
second , from the knowledge of the exact propagator of hard - core bosons,@xcite we may expect that the correlation function @xmath91 should have terms proportional to @xmath92 with @xmath93 .
third , as noted in introduction and will be discussed in sec .
iv , the @xmath51 boson and @xmath48 boson correspond to a pair of holes sitting on a rung and to a magnon in the so(5 ) symmetric ladder , respectively .
thus , we may compare the hole - pair correlation and the spin correlation functions with @xmath94 and @xmath80 of our model .
in this section we calculate the dynamical spin - spin correlation functions in the weak - coupling limit , @xmath95 .
we use the abelian bosonization method which has been successfully applied to the heisenberg ladder.@xcite in this approach we first bosonize two independent spin-@xmath0 heisenberg chains and then treat the interchain coupling @xmath25 perturbatively .
it is a relevant perturbation and is renormalized to a strong - coupling regime , generating a mass gap in the excitation spectrum .
this behavior does not depend on the sign of @xmath25 , and therefore the model describes both the antiferromagnetic heisenberg ladder for @xmath96 and the @xmath89 heisenberg chain for @xmath97.@xcite the spin gap in the latter case is nothing but the haldane gap.@xcite since the gapless phase in a magnetic field is a single phase for @xmath98 , we expect the correlation functions in the weak - coupling limit to have the same structure as in the strong - coupling limit .
we follow the formulation of shelton _
et al._,@xcite which we explain below to establish the notation .
we begin with the bosonized hamiltonian of the ladder in the continuum limit .
it consists of three parts : @xmath99 , where @xmath100 -\frac{m}{\pi a_0}\cos(\sqrt{4\pi}\phi_+ ) -\frac{h}{\sqrt{\pi}}\frac{d\phi_+}{dx}\right\ } , \label{h_+}\\ { \cal h}_-&=&\int dx\left\ { \frac{v}{2}\left [ \left(\frac{d\phi_-}{dx}\right)^2 + \left(\frac{d\theta_-}{dx}\right)^2 \right ] + \frac{2m}{\pi a_0}\cos(\sqrt{4\pi}\theta_- ) + \frac{m}{\pi a_0}\cos(\sqrt{4\pi}\phi_- ) \right\ } , \label{h_-}\\ { \cal h}_\perp&= & \frac{j_\perp}{2\pi^2a_0}\int dx\cos(\sqrt{4\pi}\theta_- ) \left[\cos(\sqrt{4\pi}\phi_+)-\cos(\sqrt{4\pi}\phi_-)\right ] + \frac{j_\perp a_0}{4\pi}\int dx \left[\left(\frac{d\phi_+}{dx}\right)^2 -\left(\frac{d\phi_-}{dx}\right)^2\right ] . \label{h_perp}\end{aligned}\ ] ] here @xmath101 is the spin - wave velocity , @xmath102 is a lattice constant , and @xmath103 with @xmath104 being a numerical constant . the bosonic fields @xmath105 and @xmath106 obey the commutation relations @xmath107=[\phi_-(x),\theta_-(y)]=i\theta(y - x)$ ] and @xmath108=[\phi_+(x),\theta_-(y ) ] = [ \phi_-(x),\theta_+(y)]=[\theta_+(x),\theta_-(y)]=0 $ ] , where @xmath109 is the step function .
it is important to note that @xmath12 appears in @xmath110 only .
thus , the external magnetic field changes the dynamics of the fields @xmath111 and @xmath112 , while the other fields @xmath113 and @xmath114 are not directly influenced by the uniform field @xmath12 .
the excitations involving these latter fields remain gapful even above the lower critical field @xmath1 .
the spin operators with @xmath115 and @xmath2 , defined in eqs .
( [ s^z_0,i])([s^-_pi , i ] ) , are written in terms of the bosonic fields as @xmath116 , \nonumber\\ & & \label{s^+_0(x)}\\ s^+_\pi(x)&= & \frac{2}{\pi}e^{i\sqrt{\pi}\theta_+ } \left[i(-1)^{x / a_0}\lambda\sin(\sqrt{\pi}\theta_- ) + \cos(\sqrt{\pi}\theta_-)\cos(\sqrt{\pi}\phi_+)\sin(\sqrt{\pi}\phi_- ) + i\sin(\sqrt{\pi}\theta_-)\sin(\sqrt{\pi}\phi_+)\cos(\sqrt{\pi}\phi_- ) \right ] , \nonumber\\ & & \label{s^+_pi(x)}\end{aligned}\ ] ] where @xmath117 for @xmath118 ( @xmath119 ) .
equations ( [ s^z_0(x)])-([s^+_pi(x ) ] ) directly follow from eq .
( 30 ) and appendix a of ref . .
et al_. showed that @xmath110 and @xmath120 can be greatly simplified by fermionization .
the hamiltonian @xmath120 then becomes that of free massive majorana fermions @xmath121 and @xmath122 having the mass gap @xmath123 and @xmath124 , respectively .
this is equivalent to the two - dimensional ising model above or below the critical temperature .
this observation allowed them to obtain the dynamical spin - spin correlation functions@xcite from the known results of the ising model.@xcite physically the @xmath125 fermion describes the @xmath17 magnon excitation , whereas the @xmath126 corresponds to a singlet excitation with much higher energy .
on the other hand , the @xmath111 and @xmath112 fields represent the @xmath127 magnon excitations . as the @xmath16 bosons condense above @xmath1
, these bosonic fields have massless excitations
. we will first integrate out the massive majorana fermions and concentrate on the massless bosonic fields .
to proceed , here we introduce an approximation for dealing with @xmath128 .
this interaction hamiltonian has three components .
the first component involving only @xmath113 and @xmath114 , i.e. , @xmath129 and @xmath130 , has two major effects on the dynamics of @xmath120 .
one effect is to renormalize the bare mass @xmath123 to @xmath131 , where @xmath132 is a high - energy cutoff , as noted by shelton _
et al._@xcite this can be absorbed by redefining the mass .
the other effect is a strong two - particle collision described by a @xmath133 matrix having a superuniversal form , as recently discussed by damle and sachdev.@xcite since we are only concerned with processes in which at most one @xmath17 magnon is created , this strong scattering effect may be irrelevant for our discussion of the dynamical correlations at zero temperature .
the second component is a coupling term , @xmath134 . when integrating out the @xmath114 field perturbatively , we find that the leading term @xmath135 gives the renormalization of the mass of the @xmath111 and @xmath112 fields , @xmath136 , as expected from the so(3 ) symmetry , where the average is taken in the ground state of @xmath120 .
this is again taken care of by redefining the mass .
the higher - order terms will generate , through gradient expansions , both irrelevant terms like @xmath137 with @xmath138 , which we can safely ignore , and a marginal operator @xmath139 , which should be kept .
the third component is the term @xmath140 already present in @xmath128 and will be kept in the following calculation .
hence we reduce @xmath128 to the form @xmath141 where the coupling constant @xmath142 may be modified from its bare value because of the emergence of the operator @xmath139 in the higher - order terms discussed above .
we believe , however , that the sign of the coupling constant is not changed by the renormalization.@xcite having made this approximation , we now integrate out the @xmath113 and @xmath114 fields to get the spin - spin correlation functions . within our approximation
the fields @xmath113 and @xmath114 are independent of @xmath111 and @xmath112 , and therefore the correlation functions of @xmath113 and @xmath114 are independent of @xmath12 .
we use eq .
( 33 ) of ref . to represent @xmath143 , @xmath144 , @xmath145 , and @xmath146 in terms of the order and disorder parameters of an ising model .
we then use eqs .
( 38 ) and ( 39 ) of ref . to obtain their correlation functions .
the correlation functions @xmath86 involve @xmath147 and @xmath148 .
they are equivalent to free massive majorana fermions , @xmath125 , whose correlators are easily obtained .
finally , the correlator @xmath149 decays exponentially [ @xmath150 with @xmath151 and is ignored .
hence , we arrive at the following expression of the dynamical spin - spin correlation functions : @xmath152\cos[\sqrt{\pi}\phi_+(0,0 ) ] \rangle_+ , \label{<s^z_pis^z_pi>}\\ \langle s^+_\pi(x,\tau)s^-_\pi(0,0)\rangle&= & ( -1)^{x / a_0}\left(\frac{2\lambda}{\pi}\right)^2a^2_1 \left\langle e^{i\sqrt{\pi}\theta_+(x,\tau)}e^{-i\sqrt{\pi}\theta_+(0,0 ) } \right\rangle_+ , \label{<s^+_pis^-_pi>}\\ \langle s^+_0(x,\tau)s^-_0(0,0)\rangle&= & -\frac{a_0}{2\pi^2v } \left\langle e^{i\sqrt{\pi}\theta_+(x,\tau)}e^{-i\sqrt{\pi}\phi_+(x,\tau ) } e^{i\sqrt{\pi}\phi_+(0,0)}e^{-i\sqrt{\pi}\theta_+(0,0 ) } \right\rangle_+ \partial_+k_0(mr / v)\nonumber\\ & & -\frac{a_0}{2\pi^2v } \left\langle e^{i\sqrt{\pi}\theta_+(x,\tau)}e^{i\sqrt{\pi}\phi_+(x,\tau ) } e^{-i\sqrt{\pi}\phi_+(0,0)}e^{-i\sqrt{\pi}\theta_+(0,0 ) } \right\rangle_+ \partial_-k_0(mr / v)\nonumber\\ & & + \frac{a_0m}{2\pi^2v}k_0(mr / v ) \left [ \left\langle e^{i\sqrt{\pi}\theta_+(x,\tau)}e^{-i\sqrt{\pi}\phi_+(x,\tau ) } e^{-i\sqrt{\pi}\phi_+(0,0)}e^{-i\sqrt{\pi}\theta_+(0,0 ) } \right\rangle_+ \right.\nonumber\\ & & \hspace*{2.7cm}\left . + \left\langle e^{i\sqrt{\pi}\theta_+(x,\tau)}e^{i\sqrt{\pi}\phi_+(x,\tau ) } e^{i\sqrt{\pi}\phi_+(0,0)}e^{-i\sqrt{\pi}\theta_+(0,0 ) } \right\rangle_+ \right ] , \label{<s^+_0s^-_0>}\end{aligned}\ ] ] where @xmath153 is the imaginary time , @xmath154 is a numerical constant , @xmath155 , and @xmath156 is the average with respect to the hamiltonian @xmath157 . in eqs .
( [ < s^z_0s^z_0>])([<s^+_0s^-_0 > ] ) we have ignored the terms decaying much faster than @xmath158 . therefore we discarded the contribution from processes involving more than one massive magnons .
these equations are valid for @xmath96 .
when @xmath97 , on the other hand , the strongly renormalized interchain coupling combines the spins @xmath28 and @xmath29 into a single spin , and the ladder behaves as a @xmath89 heisenberg chain . as explained in ref . ,
when taking average over the massive majorana fermions @xmath125 and @xmath126 , we only need to exchange the order and disorder parameter of the ising model . using the correlators of @xmath113 and @xmath114 for @xmath97 in ref .
, we find @xmath159 where the correlators in the right - hand side are those in eqs .
( [ < s^z_0s^z_0>])([<s^+_0s^-_0 > ] ) .
when @xmath97 , the @xmath160 correlators decay much faster than the @xmath115 correlator and are thus negligible .
hence , the dynamical spin - spin correlation functions of the @xmath89 haldane chain are linear combinations of those @xmath115 and @xmath2 correlators of the heisenberg ladder ( @xmath96 ) .
now our task is to calculate the correlators of @xmath111 and @xmath112 in the presence of the magnetic field .
the hamiltonian @xmath110 is identical to the one used to study the commensurate - incommensurate transition in classical two - dimensional systems.@xcite in fact some of the correlation functions in eqs .
( [ < s^z_0s^z_0>])([<s^+_0s^-_0 > ] ) have been discussed in this context.@xcite in particular , the leading term of the correlation function corresponding to @xmath161 is obtained by schulz@xcite including its universal exponent in the limit @xmath162 .
these results are used to calculate the spin - spin correlation functions by chitra and giamarchi,@xcite who unfortunately seem to have overlooked some terms including the leading term [ @xmath163 in @xmath164 .
we think therefore that it is still worthwhile to describe the calculation of the correlations of @xmath111 and @xmath112 in eqs .
( [ < s^z_0s^z_0>])([<s^+_0s^-_0 > ] ) in some detail , despite the fact that the hamiltonian @xmath110 has been analyzed in many literatures . to our knowledge ,
our results concerning the massive modes are new . following ref .
, we fermionize @xmath110
: @xmath165 , \nonumber\\ & = & \int^\infty_{-\infty}dk \left [ vk\left(c^\dagger_{r , k}c_{r , k}-c^\dagger_{l , k}c_{l , k}\right ) -im\left(c^\dagger_{r , k}c_{l , k}-c^\dagger_{l , k}c_{r , k}\right ) -h\left(c^\dagger_{r , k}c_{r , k}+c^\dagger_{l , k}c_{l , k}\right ) \right ] , \label{h_+2}\end{aligned}\ ] ] where @xmath166 ( @xmath167 ) is the right - going ( left - going ) complex fermion field , and @xmath168 .
the fermion fields are related to the bosons by the standard relations : @xmath169},\quad : \!\psi^\dagger_r(x)\psi_r(x)\!:\,= \frac{1}{2\sqrt{\pi}}\frac{d}{dx}[\phi_+(x)-\theta_+(x ) ] , \label{psi_r}\\ \psi_l(x)&= & \frac{1}{\sqrt{2\pi a_0}}e^{-i\sqrt{\pi}[\phi_+(x)+\theta_+(x)]},\quad : \!\psi^\dagger_l(x)\psi_l(x)\!:\,= \frac{1}{2\sqrt{\pi}}\frac{d}{dx}[\phi_+(x)+\theta_+(x ) ] .
\label{psi_l}\end{aligned}\ ] ] it is important to note that the normal ordering in the above equations is defined with respect to the ground state of @xmath170 .
the fermionized hamiltonian @xmath110 is easily diagonalized : @xmath171 , \label{diagonalh_+}\ ] ] where @xmath172 with @xmath173 .
the magnetic field plays a role of the chemical potential to the fermions .
we are concerned with the case where @xmath12 is slightly above the lower critical field @xmath174 such that @xmath175 .
the ground state is obtained by filling the upper band ( @xmath176 ) up to the fermi points ( @xmath177 ) and the lower band ( @xmath178 ) completely ; see fig . [
fig : band ] .
the fermi wavenumber @xmath179 is related to the magnetization : @xmath180 .
this follows from @xmath181 where @xmath182 is the length of the ladder and @xmath183 . in calculating long - distance correlations
, we can safely ignore the lower band and keep only the low - energy excitations around the fermi points in the upper band . in the fermion representation
the interaction term @xmath128 reads @xmath184 where @xmath185 and we have dropped the contribution from the lower band .
the total hamiltonian for the fermions in the upper band , @xmath186 , consists of the kinetic energy , eq .
( [ diagonalh_+ ] ) , and the short - range scattering term , eq .
( [ hint ] ) .
the coupling constant of the latter term is proportional to @xmath25 in lowest order .
thus , the interaction is repulsive for the antiferromagnetically coupled ladder , while it is attractive for the @xmath89 chain.@xcite obviously the scattering term has only negligible effects in both limits @xmath162 and @xmath187 , where we will get the correlation functions of free fermions.@xcite the low - energy physics of @xmath188 can be easily solved by the abelian bosonization.@xcite we first linearize the dispersion around @xmath189 ( fig .
[ fig : band ] ) .
we then bosonize the fermions in the upper band : @xmath190 } + e^{-i\pi mx / a_0-i\sqrt{\pi}[\phi(x)+\theta(x ) ] } \right ) , \label{bosonize}\ ] ] where the bosonic fields @xmath191 and @xmath192 obey @xmath193=i\theta(y - x)$ ] . using these fields , we write the hamiltonian @xmath188 as @xmath194 , \label{h_+boson}\ ] ] where @xmath195 is the fermi velocity , and @xmath13 is a parameter determined by the interaction : @xmath196 ( @xmath197 ) when @xmath96 ( @xmath97 ) , and @xmath198 as @xmath75 . incidentally , @xmath13 is related to the compactification radius @xmath199 of the field @xmath200 by @xmath201 .
we now need to express @xmath111 and @xmath112 in terms of @xmath200 and @xmath202 .
once this is done , it is straightforward to calculate the correlation functions since @xmath188 is a free - boson hamiltonian .
first we note that for states near the fermi surface we have @xmath203 where we used the approximation @xmath204 . using eqs .
( [ bosonize ] ) and ( [ approximation ] ) and discarding the @xmath178 fermions , we find @xmath205 . \label{dphi_+/dx}\ ] ] it follows that @xmath206\cos[\sqrt{4\pi}\phi(0,0)]\rangle_a , \label{<dphi_+dphi_+>}\end{aligned}\ ] ] where @xmath207 represents the average taken in the ground state of @xmath188 .
the averages are found to be @xmath208\cos[\sqrt{4\pi}\phi(0,0)]\rangle_a= \frac{1}{2 } \left(\frac{\tilde
a_0 ^ 2}{x^2+\tilde v^2\tau^2}\right)^g , \label{<cos(4piphi)cos(4piphi)>a}\end{aligned}\ ] ] where @xmath209 is a short - distance cutoff of order @xmath102 .
integration of eq .
( [ dphi_+/dx ] ) then yields @xmath210 we have neglected the contribution from the oscillating term .
we thus get @xmath211\cos[\sqrt{\pi}\phi_+(0,0)]\rangle_+ & = & \cos(\pi mx / a_0 ) \langle\cos[\sqrt{\pi}\phi(x,\tau)]\cos[\sqrt{\pi}\phi(0,0)]\rangle_a \nonumber\\ & = & \frac{1}{2}\cos(\pi mx / a_0 ) \left(\frac{\tilde a_0 ^ 2}{x^2+\tilde v^2\tau^2}\right)^{g/4}. \label{<cos(piphi)cos(piphi)>a}\end{aligned}\ ] ] we next consider @xmath212 . from eqs .
( [ psi_r ] ) and ( [ psi_l ] ) we can express it as @xmath213 .
\label{e^(-ipitheta)}\ ] ] using the same approximation as in the derivation of eqs .
( [ approximation ] ) and ( [ dphi_+/dx ] ) , we get @xmath214 and @xmath215 . here
we have made a further approximation @xmath216 which is valid for @xmath175 . from eqs .
( [ bosonize ] ) , ( [ phi_+ ] ) , and ( [ e^(-ipitheta ) ] ) we find @xmath217 \right\ } e^{-i\sqrt{\pi}\theta(x)+i\pi/4}. \label{e^(-ipitheta)-2}\ ] ] from eq .
( [ e^(-ipitheta)-2 ] ) we may write the correlation functions in eqs .
( [ < s^+_pis^-_pi > ] ) and ( [ < s^+_0s^-_0 > ] ) as @xmath218 @xmath219 and @xmath220 the averages in the above equations are given by @xmath221 combining these results together , we finally obtain the dynamical spin - spin correlation functions : @xmath222 + c_1\cos(2\pi mx ) \left(\frac{\tilde
a_0 ^ 2}{x^2+\tilde v^2\tau^2}\right)^g , \label{<s^z_0s^z_0>a}\\ \langle s^z_\pi(x,\tau)s^z_\pi(0,0)\rangle&= & c_2(-1)^x\cos(\pi mx)k_0(mr / v ) \left(\frac{\tilde
a_0 ^ 2}{x^2+\tilde v^2\tau^2}\right)^{g/4 } , \label{<s^z_pis^z_pi > a}\\ \langle s^+_\pi(x,\tau)s^-_\pi(0,0)\rangle&= & ( -1)^x \left\ { c_3\left(\frac{\tilde a_0 ^ 2}{x^2+\tilde v^2\tau^2}\right)^{1/4 g } -c_4 \left(\frac{\tilde a_0 ^ 2}{x^2+\tilde v^2\tau^2}\right ) ^{(1/2\sqrt{g}-\sqrt{g})^2 } \left [ \frac{\tilde a_0 ^ 2e^{2\pi imx}}{(x+i\tilde v\tau)^2 } + \frac{\tilde a_0 ^ 2e^{-2\pi imx}}{(x - i\tilde v\tau)^2 } \right ] \right\ } , \label{<s^+_pis^-_pi > a}\\ \langle s^\pm_0(x,\tau)s^\mp_0(0,0)\rangle&= & \frac{i}{8\pi^2v } \left(\frac{\tilde a_0 ^ 2}{x^2+\tilde v^2\tau^2}\right ) ^{(1/\sqrt{g}-\sqrt{g})^2/4 } \left [ \left(\frac{2\tilde a_0e^{\pm i\pi mx}}{x - i\tilde v\tau } -\frac{\tilde a_0e^{\mp i\pi mx}}{x+i\tilde v\tau}\right ) \partial_-k_0(mr / v ) + ( x\to - x)\right ] \nonumber\\ & & \pm\frac{im}{4\pi^2v}k_0(mr / v ) \left(\frac{\tilde a^2_0}{x^2+\tilde v^2\tau^2}\right ) ^{(1/\sqrt{g}-\sqrt{g})^2/4 } \left ( \frac{\tilde a_0e^{\mp i\pi mx}}{x+i\tilde v\tau } -\frac{\tilde a_0e^{\pm i\pi mx}}{x - i\tilde v\tau } \right ) , \label{<s^+_0s^-_0>a}\end{aligned}\ ] ] where @xmath223 s are positive numerical constants , and we have set @xmath224 .
the correlation function @xmath225 is obtained by replacing @xmath226 with @xmath227 in eq .
( [ < s^+_pis^-_pi > a ] ) . in eq .
( [ < s^+_0s^-_0>a ] ) we have discarded the term proportional to @xmath228 decaying much faster than the kept terms .
we should therefore regard eqs .
( [ < s^z_0s^z_0>a])([<s^+_0s^-_0>a ] ) as listing only the leading terms . as noted in sec .
ii , we may expect that @xmath229 should contain algebraically decaying terms that are proportional to @xmath230 with any integer @xmath231 . from ref .
, we expect that the correlator @xmath164 should also have terms proportional to @xmath230 which decay as @xmath232 .
the appearance of the term proportional to @xmath233 suggests that @xmath234 should have terms proportional to @xmath235 $ ] .
we note that the equal - time correlations @xmath236 and @xmath237 agree with eqs .
( [ static < s^z_0s^z_0 > ] ) and ( [ static < s^x_pis^x_pi > ] ) if we identify @xmath71 with @xmath238 . as is well known , the strongest correlation is @xmath239 .
note also that the value of the exponent is consistent between the weak- and strong - coupling approach : @xmath196 ( @xmath197 ) for @xmath96 ( @xmath97 ) and @xmath198 as @xmath75 .
another interesting finding is that the exponentially decaying equal - time correlation functions have different phases by @xmath240 : @xmath241 and @xmath242 .
now we are in a position to calculate the dynamical spin structure factors defined by @xmath243 where @xmath244 is a real time and the correlation functions in the real time are obtained by replacing @xmath245 .
we may also calculate it from @xmath246 for @xmath247 . from the obvious relation @xmath248
, we assume @xmath249 in the following discussion .
we first consider the correlation functions showing the quasi - long - range order .
using eq .
( [ formula1 ] ) in appendix , we get from eqs .
( [ < s^z_0s^z_0>a ] ) and ( [ < s^+_pis^-_pi > a ] ) @xmath250 ^ 2 } \theta\biglb(\omega-\tilde v(q-2\pi m)\bigrb ) \theta\biglb(\omega+\tilde v(q-2\pi m)\bigrb ) \left(\frac{4\tilde v^2/\tilde a_0 ^ 2 } { \omega^2-\tilde v^2(q-2\pi m)^2}\right ) ^{1-g } , \label{s^zz_0(q , omega)}\\ s^{+-}_\pi(q,\omega)&= & \frac{\pi c_3\tilde a_0 ^ 2}{\tilde v[\gamma(1/4g)]^2 } \theta\biglb(\omega+\tilde v(q-\pi)\bigrb ) \theta\biglb(\omega-\tilde v(q-\pi)\bigrb ) \left(\frac{4\tilde v^2/\tilde a_0 ^ 2 } { \omega^2-\tilde v^2(q-\pi)}\right ) ^{1 - 1/4g}\nonumber\\ & & + \frac{\pi c_4\tilde a_0 ^ 2 } { \tilde v\gamma(\eta_+)\gamma(\eta_- ) } \biggl [ \theta\biglb(\omega-\tilde v[q-\pi(1 - 2m)]\bigrb ) \theta\biglb(\omega+\tilde v[q-\pi(1 - 2m)]\bigrb ) \nonumber\\ & & \hspace*{3.2cm}\times \left(\frac{\omega-\tilde v[q-\pi(1 - 2 m ) ] } { 2\tilde v/\tilde a_0}\right)^{\eta_+-1 } \left(\frac{2\tilde v/\tilde a_0 } { \omega+\tilde v[q-\pi(1 - 2m)]}\right)^{1-\eta_- } \nonumber\\ & & \hspace*{2.5 cm } + \theta\biglb(\omega-\tilde v[q-\pi(1 + 2m)]\bigrb ) \theta\biglb(\omega+\tilde v[q+\pi(1 + 2m)]\bigrb ) \nonumber\\ & & \hspace*{3.2cm}\times \left(\frac{\omega+\tilde v[q-\pi(1 + 2 m ) ] } { 2\tilde v/\tilde a_0}\right)^{\eta_+-1 } \left(\frac{2\tilde v/\tilde a_0 } { \omega-\tilde v[q-\pi(1 + 2m)]}\right)^{1-\eta_- } \biggr ] , \nonumber\\ & & \label{s^+-_pi(q , omega)}\end{aligned}\ ] ] where @xmath251 ^ 2 $ ] . the structure factor @xmath252 can be obtained from @xmath253 by @xmath254 . being obtained from the long - distance asymptotic expansions ( [ < s^z_0s^z_0>a ] ) and ( [ < s^+_pis^-_pi > a ] ) , each term in eqs . ( [ s^zz_0(q , omega ) ] ) and ( [ s^+-_pi(q , omega ) ] ) describe the behavior of the structure factors correctly only near its low - energy threshold .
for example , the last term ( @xmath255 ) in eq .
( [ s^zz_0(q , omega ) ] ) is valid only for @xmath256 ( @xmath257 ) and can not be extended to @xmath258 .
the supports of these structure factors are shown in figs .
[ fig : s^zz_0(q , omega)][fig : s^-+_pi(q , omega ) ] .
they are essentially the same as those of the @xmath15 @xmath57 chain@xcite except that the small @xmath226 in the ladder corresponds to the nearly polarized state in the @xmath15 chain through the relation @xmath259 .
the strongest divergence is at @xmath260 of @xmath261 : @xmath262^{-1 + 1/4g}$ ] .
the exponent approaches @xmath263 as @xmath75 .
we note that the boundaries of the supports of these structure factors became all straight lines because of our linearization of the dispersion relation in the continuum limit .
this is an artifact of the approximation , and the true boundary lines should be given by some nonlinear functions .
furthermore , some of the boundary lines may be parts of a single curve .
we next consider the massive components .
using eqs .
( [ < s^z_pis^z_pi > a ] ) , ( [ s(q , omega)-2 ] ) , and ( [ formula2 ] ) , we get @xmath264 ^ 2}\int^\infty_{-\infty}dk \frac{\theta\biglb(\omega-\varepsilon(q - k-\pi+\pi m)-\tilde v|k|\bigrb ) } { \varepsilon(q - k-\pi+\pi m ) \{[\omega-\varepsilon(q - k-\pi+\pi m)]^2-\tilde v^2k^2\}^{1-g/4 } } \nonumber\\ & & + ( m\to - m ) , \label{s^zz_pi(q , omega)}\end{aligned}\ ] ] where @xmath265 . the minimum energy above which @xmath266
is @xmath267 around @xmath268 . near this threshold energy
the structure factor reduces to @xmath269 where @xmath270 and @xmath271 are assumed .
the support of @xmath272 is shown in fig .
[ fig : s^zz_pi(q , omega ) ] .
we see that @xmath272 diverges at the low - energy threshold as @xmath273^{-1+g/2}$ ] .
since the two thresholds @xmath274 intersect at @xmath260 , we expect to have a peak at @xmath260 and @xmath275 .
the exponent approaches @xmath276 as @xmath75 , and the singularity is even stronger for @xmath61 , where @xmath196 .
note that the exponent jumps from @xmath277 to @xmath276 when @xmath12 crosses @xmath1 from below . in the strong - coupling limit of the ladder
, the square - root divergence may be understood in the following way .
the correlation function @xmath80 is a propagator of the @xmath48 bosons .
if we ignored the interaction with the @xmath51 bosons , we would get the massive free - particle propagator , @xmath278 . due to the interaction
the motion of the @xmath48 boson is necessarily accompanied by a superfluid flow of the @xmath51 bosons .
its main effect in the low - density limit amounts both to multiplying the free propagator by that of the hard - core @xmath51 bosons @xmath279 and to shifting the momentum by @xmath280 .
the fourier transform of the product has the square - root divergence at the threshold .
finally we consider @xmath281 .
this can be obtained from eqs .
( [ < s^+_0s^-_0>a ] ) and ( [ s(q , omega)-2 ] ) as described in appendix .
the result is @xmath282^{\eta_0 } } { \varepsilon(q - k\mp\pi m ) [ \omega-\varepsilon(q - k\mp\pi m)+\tilde vk]^{1-\eta_0 } } \nonumber\\ & & \hspace*{5 cm } \times\left
[ \frac{3}{2}\varepsilon(q - k\mp\pi m ) -\frac{1}{2}v(q - k\mp\pi m)\pm m\right ] \nonumber\\ & & + ( q\to - q ) , \label{s^+-_0(q , omega)}\end{aligned}\ ] ] where @xmath283 ^ 2/4 $ ] . near the lower edge @xmath284 , it may be approximated by @xmath285 in general the exponent @xmath286 and approaches 0 as @xmath75 .
the support of @xmath287 is shown in fig .
[ fig : s^+-_0(q , omega ) ] . like @xmath272
, @xmath287 has a peak at @xmath288 , where the two thresholds @xmath289 cross .
the structure factor @xmath290 is approximately equal to @xmath291 near the low - energy threshold , @xmath292 and @xmath293 .
that @xmath290 is much smaller than @xmath287 is consistent with the result in the strong - coupling limit where the @xmath48 bosons are absent in the ground state [ see eqs .
( [ s^+_0,i ] ) and ( [ s^-_0,i ] ) ] .
we shall discuss implications of the results we obtained for the heisenberg ladder to the @xmath2-resonance mode in the so(5 ) symmetric ladder model . as pointed out in ref .
, there is an analogy between the quantum phase transition driven by the chemical potential in the so(5 ) symmetric ladder model and the field - induced phase transition in the heisenberg ladder . obviously , the chemical potential plays the role of the magnetic field .
the analogy is most clearly seen in the strong - coupling limit.@xcite at half filling the ground state of the @xmath294 phase or the mott insulating phase discussed in ref .
is a state in which all the rungs are in the spin singlet state .
when the chemical potential is zero , there are fivefold degenerate low - lying massive modes above the ground state .
the five modes consist of a @xmath89 magnon triplet , a hole pair state where two holes are placed on a single rung , and a state where two additional electrons are put on a rung .
when the chemical potential is turned on , the energy of the hole - pair ( electron - pair ) excitation decreases ( increases ) while the magnon triplet is not directly affected by the chemical potential .
thus , we see that the hole pair corresponds to the @xmath16 magnon or the @xmath51 boson in the heisenberg ladder .
the triplet magnon in the so(5 ) model is an analog of the @xmath17 magnon ( @xmath48 boson ) in the heisenberg ladder .
furthermore , the low - energy effective hamiltonian for the hole - pair excitations in the strong - coupling limit is similar to the effective hamiltonian for the @xmath51 bosons .
that is , hole pairs may be viewed as hard - core bosons which repel each other when two hole pairs sit on neighboring rungs.@xcite let us find operators playing the role of the spin operators in the heisenberg ladder .
first , the operator corresponding to @xmath295 should change a singlet rung @xmath296 into a hole pair or @xmath297 . here
@xmath298 and @xmath299 are creation operators of an electron with spin @xmath300 on the @xmath42th rung of upper ( @xmath70 ) and lower ( @xmath21 ) chains .
obviously the @xmath21-wave pair operator @xmath301 is such an operator : @xmath302 .
second , the operator corresponding to @xmath303 should be a number operator of hole pairs .
it is given by @xmath304 . finally , from the relation @xmath305 , we find that @xmath306 creates a triplet magnon from a rung singlet .
thus we conclude that the spin operator @xmath306 is an analog of @xmath295 . when the chemical potential is increased beyond the charge gap which equals the spin gap in the so(5 ) symmetric model , the ladder is doped with the charge carrier ( holes ) and
becomes superconducting with the @xmath21-wave - like symmetry . for spatial dimension greater than or equal to two ,
the superconducting order is long - ranged in the ground state , and this gives rise to a @xmath18-function peak or the @xmath2-resonance in the dynamic spin structure factor . in one dimension ,
however , the order is quasi - long - ranged , and therefore the peak is expected to be replaced by a power - law singularity.@xcite the threshold energy at @xmath260 is also shown to be equal to the chemical potential.@xcite these features are readily reproduced from our results for the heisenberg ladder model . from the approximate
mapping we discussed above , the @xmath21-wave pair correlation function is expected to show the quasi - long - range order corresponding to the @xmath307 order in the heisenberg ladder , eq .
( [ static < s^x_pis^x_pi > ] ) : @xmath308 where the exponent @xmath309 is presumably smaller than 2 and approaches 2 in the limit where the hole density @xmath18 goes to zero .
the correlator has no @xmath310 factor because the hole - pair mode has a minimum energy at @xmath311 .
the charge density correlation is related to @xmath312 , eq .
( [ static < s^z_0s^z_0 > ] ) , and is also quasi - long ranged : @xmath313 where we have used the relation between the hole density and the fermi wave number , @xmath314 . the result that the correlations of the @xmath21-wave superconductivity and @xmath315 charge density wave show power - law decay with the exponents whose product is 1 was also obtained by nagaosa for a generic two - chain model.@xcite finally the spin correlation of the so(5 ) ladder is expected to be @xmath316 where @xmath317 is the correlation length determined by the spin gap .
the spin structure factor is then @xmath318^{-1+\tilde\eta/4 } , \label{structurefactor}\ ] ] where @xmath319 is the magnon dispersion at @xmath320 .
from fig .
[ fig : s^zz_pi(q , omega ) ] we see that the threshold energy at @xmath260 is determined by the chemical potential , as expected .
although the exponent @xmath309 depends on the detail of the model , we can generally conclude that it is @xmath321 in the low - density limit of holes ( @xmath322 ) , where we may regard the hard - core bosons as free fermions ( @xmath323 ) .
this universal exponent was independently found by ivanov and lee@xcite and by schulz@xcite for the @xmath244-@xmath24 ladder and was also obtained by konik _
et al . _ for the so(8 ) gross - neveu model.@xcite we thus find that the spin structure factor has a universal square - root divergence at the critical point .
when the superfluid density is finite , the interaction between bosons becomes important and modify the exponent , as we saw in the heisenberg ladder model .
we expect that the square - root singularity is a universal feature for the spectral weight of a gapped excitation generated by injecting a massive particle to a superfluid in the low - density limit .
although our argument above is based on the analogy and approximate mapping , our results should be valid as long as the weak - coupling and strong - coupling limits are in the same phase .
we notice that eq .
( [ < ss > ] ) is the same as the `` mean - field result given in sec .
vii of ref . .
the validity of this result is , however , questioned by lin _
_ as it misses the existence of the bound states such as the cooper pair - magnon bound states found in the so(8 ) gross - neveu model
. on the other hand , we did nt find such a bound state in our weak - coupling calculation .
it is not clear at the moment whether this is due to the approximation we made , for example , concerning the interaction term in @xmath128 .
it was shown by damle and sachdev @xcite that this term can indeed lead to a bound state of two magnons when @xmath170 .
the fate of the bound state in the gapless phase is an open question .
finally we conclude this paper by summarizing our results on the spin correlations in the gapless phase of the two - leg heisenberg ladder . we have obtained the dynamical spin - spin correlation functions and the structure factors , extending the bosonization theory of shelton _
_ to the gapless regime .
the correlation functions are classified into two categories : algebraically decaying ones , @xmath324 and @xmath229 , and exponentially decaying ones , @xmath325 and @xmath234 .
we have also found that the terms @xmath326 ( @xmath231 : integer ) are quasi - long - ranged , while the terms @xmath327 $ ] are short - ranged .
the exponents of the correlation functions are controlled by the single parameter @xmath13 , which is smaller ( larger ) than 1 for @xmath96 ( @xmath97 ) .
the parameter @xmath13 approaches 1 in the limits @xmath75 .
the structure factors have power - law singularities at the lower edges , and the strongest divergence is at @xmath328 in @xmath261 due to the dominant @xmath307 spin correlation .
the next strongest singularity is found at the lower edge of @xmath272 : @xmath329 ^ 2+m^2\}^{1/2}$ ] . the exponent is universally given by @xmath276 in the limits @xmath75 .
we thank i. affleck , l. balents , e. demler , m. p. a. fisher , w. hanke , h. h. lin , m. oshikawa , and d. scalapino for useful discussions .
is supported by the monbusho overseas research grant .
this work is supported by nsf under grant numbers dmr-9400372 and dmr-9522915 .
in this appendix we list integral formulas we used to calculate the dynamical structure factors . for the structure factor @xmath272
we first take the fourier transform of the correlation function in the imaginary time : @xmath331 ^ 2+\tilde v^2k^2\}^{1-\gamma } } + \int^\infty_{\tilde v|k|}d\nu \frac{2\sin(\pi\gamma)(\nu^2-\tilde v^2k^2)^{\gamma-1 } } { ( \tilde\omega - i\nu)^2+\varepsilon^2(q - k ) } \right ) .
\nonumber\\ & & \label{convolution}\end{aligned}\ ] ] after the analytic continuation we take the imaginary part to find @xmath332 ^ 2-\tilde v^2k^2\}^{1-\gamma } } \label{formula2}\ ] ] for @xmath247 .
when @xmath333 , @xmath334 . in this case we may approximate the last integral as @xmath335 ^ 2-\tilde v^2k^2\}^{1-\gamma } } = \theta\biglb(\omega-\varepsilon(q)\bigrb ) \frac{b(\gamma,1/2)}{\tilde vm } [ \omega-\varepsilon(q)]^{2\gamma-1 } , \label{formula3}\ ] ] where @xmath336 is the beta function . from eqs .
( [ formula2 ] ) and ( [ formula3 ] ) we finally obtain @xmath337 which is valid for @xmath338 .
we next consider @xmath281 .
according to eq .
( [ < s^+_pis^-_pi > a ] ) , we need the following fourier transform : @xmath339^\gamma } { \varepsilon(q - k)[\omega - i\tilde vk+i\varepsilon(q - k)]^{1-\gamma } } -2\sin(\pi\gamma)\int^\infty_{\tilde v|k|}d\nu \frac{(\nu+\tilde vk)^\gamma(\nu-\tilde vk)^{\gamma-1 } } { ( \omega - i\nu)^2+\varepsilon^2(q - k ) } \right ) .
\nonumber\\ & & \label{formula5}\end{aligned}\ ] ] after the analytic continuation we obtain @xmath340^\gamma } { \varepsilon(q - k)[\omega-\varepsilon(q - k)+\tilde vk]^{1-\gamma}}. \label{formula6}\ ] ] using the same approximation as in eq . ( [ formula3 ] )
, we obtain @xmath341 for @xmath342 and @xmath333 . in the same way we get @xmath343^\gamma } { \varepsilon(q - k)[\omega-\varepsilon(q - k)-\tilde vk]^{1-\gamma } } , \label{formula8}\end{aligned}\ ] ] which reduces to eq .
( [ formula7 ] ) for @xmath342 and @xmath333 .
present address : yukawa institute for theoretical physics , kyoto university , kyoto 606 - 8502 , japan .
for a review , see for example e. dagotto and t. m. rice , science * 272 * , 618 ( 1996 ) .
m. azuma , z. hiroi , m. takano , k. ishida , and y. kitaoka , phys .
lett . * 73 * , 3463 ( 1994 ) .
m. uehara , t. nagata , j. akimitsu , h. takahashi , n. mori , and k. kinoshita , j. phys .
. jpn . * 65 * , 2764 ( 1996 ) .
g. chaboussant , p. a. crowell , l. p. lvy , o. piovesana , a. madouri , and d. mailly , phys .
b * 55 * , 3046 ( 1997 ) ; g. chaboussant , y. fagot - revurat , m .- h .
julien , m. e. hanson , c. berthier , m. horvati , l. p. lvy , and o. piovesana , phys .
lett . * 80 * , 2713 ( 1998 ) .
p. r. hammar , d. h. reich , c. broholm , and f. trouw , phys .
b * 57 * , 7846 ( 1998 ) . c. a. hayward , d. poilblanc , and l. p. lvy , phys . rev .
b * 54 * , 12649 ( 1996 ) .
d. c. cabra , a. honecker , and p. pujol , phys .
lett . * 79 * , 5126 ( 1997 ) .
r. chitra and t. giamarchi , phys .
b * 55 * , 5816 ( 1997 ) .
r. eder , phys .
b * 57 * , 12832 ( 1998 ) .
n. nagaosa and s. murakami , j. phys .
jpn . * 67 * , 1876 ( 1998 ) .
r. calemczuk , j. riera , d. poilblanc , j .-
boucher , and o. piovesana , eur .
j. b * 7 * , 171 ( 1999 ) .
d. g. shelton , a. a. nersesyan , and a. m. tsvelik , phys .
b * 53 * , 8521 ( 1996 ) . for finite temperatures
the dynamical structure factors are discussed in a more general context in s. sachdev , t. senthil , and r. shankar , phys .
b * 50 * , 258 ( 1994 ) .
g. mller , h. thomas , h. beck , and j. c. bonner , phys .
b * 24 * , 1429 ( 1981 ) .
j. c. talstra and f. d. m. haldane , phys .
b * 50 * , 6889 ( 1994 ) ; * 54 * , 12594 ( e ) ( 1996 ) . s. c. zhang , science * 275 * , 1089 ( 1997 ) ; e. demler and s. c. zhang , phys .
* 75 * , 4126 ( 1995 ) .
d. scalapino , s. c. zhang , and w. hanke , phys .
b * 58 * , 443 ( 1998 ) .
d. g. shelton and d. snchal , phys .
b * 58 * , 6818 ( 1998 ) .
e. arrigoni and w. hanke , cond - mat/9712143 .
h. h. lin , l. balents , and m. p. a. fisher , phys . rev .
b * 58 * , 1794 ( 1998 ) .
r. eder , a. dorneich , m. g. zacher , w. hanke , and s. c. zhang , phys .
b * 59 * , 561 ( 1999 ) .
s. sachdev and r. n. bhatt , phys .
b * 41 * , 9323 ( 1990 ) .
s. gopalan , t. m. rice , and m. sigrist , phys .
b * 49 * , 8901 ( 1994 ) .
f. mila , eur .
j. b * 6 * , 201 ( 1998 ) . c. n. yang and c. p.
yang , phys . rev . * 150 * , 327 ( 1966 ) .
the next - order term is found in m. reigrotzki , h. tsunetsugu , and t. m. rice , j. phys . :
condens . matter * 6 * , 9235 ( 1994 ) .
even higher - order terms are given in ref . .
g. i. dzhaparidze and a. a. nersesyan , pisma zh .
* 27 * , 356 ( 1978 ) [ jetp lett . * 27 * , 334 ( 1978 ) ] .
a. m. tsvelik , phys .
b * 42 * , 10499 ( 1990 ) .
i. affleck , phys .
b * 43 * , 3215 ( 1991 ) .
t. sakai and m. takahashi , phys .
b * 43 * , 13383 ( 1991 ) .
n. m. bogoliubov , a. g. izergin , and v. e. korepin , nucl .
b * 275 * , 687 ( 1986 ) .
f. d. m. haldane , phys .
* 45 * , 1358 ( 1980 ) . t. sakai and m. takahashi , j. phys .
. jpn . * 60 * , 3615 ( 1991 ) . h. g. vaidya and c. a. tracy , j. math
. phys . * 20 * , 2291 ( 1979 ) . m. jimbo , t. miwa , y. mri , and m. sato , physica * 1d * , 80 ( 1980 ) .
j. schulz , phys .
b * 34 * , 6372 ( 1986 ) .
s. p. strong and a. j. millis , phys .
b * 50 * , 9911 ( 1994 )
. f. d. m. haldane , phys . lett .
* 93a * , 464 ( 1983 ) ; phys . rev . lett . * 50 * , 1153 ( 1983 ) .
the equal - time correlation function obtained from the majorana fermion theory shows an excellent agreement with the monte carlo calculation : m. greven , r. j. birgeneau , and u. j. wiese , phys .
lett . * 77 * , 1865 ( 1996 ) .
t. t. wu , b. mccoy , c. a. tracy , and e. barouch , phys .
b * 13 * , 316 ( 1976 ) . k. damle and s. sachdev , phys .
b * 57 * , 8307 ( 1998 ) .
this is supported by the strong - coupling analysis in sec .
ii and by a recent numerical work by m. usami and s. suga , phys .
b * 58 * , 14401 ( 1998 ) . v.
l. pokrovsky and a. l. talapov , phys .
lett . * 42 * , 65 ( 1979 ) ; zh . eksp .
. fiz . * 78 * , 269 ( 1980 ) [ sov .
jetp * 51 * , 134 ( 1980 ) ] .
y. okwamoto , j. phys .
. jpn . * 49 * , 8 ( 1980 ) .
j. schulz , phys .
b * 22 * , 5274 ( 1980 ) .
f. d. m. haldane , j. phys .
a * 15 * , 507 ( 1982 ) . v. j. emery , in _ highly conducting one - dimensional solids _ , edited by j. t. devreese _
( plenum , new york , 1979 ) .
j. slyom , adv .
* 28 * , 209 ( 1979 ) .
n. nagaosa , solid state commun .
* 94 * , 495 ( 1995 ) .
d. a. ivanov and p. a. lee , phys .
b * 57 * , 2118 ( 1998 ) .
h. j. schulz , phys .
b * 59 * , r2471 ( 1999 ) .
r. konik , f. lesage , a. w. w. ludwig , and h. saleur , cond - mat/9806334 . | the zero - temperature dynamical spin - spin correlation functions are calculated for the spin-@xmath0 two - leg antiferromagnetic heisenberg ladder in a magnetic field above the lower critical field @xmath1 .
the dynamical structure factors are calculated which exhibit both massless modes and massive excitations .
these modes appear in different sectors characterized by the parity in the rung direction and by the momentum in the direction of the chains .
the structure factors have power - law singularities at the lower edges of their support .
the results are also applicable to spin-1 heisenberg chain .
the implications are briefly discussed for the various correlation functions and the @xmath2-resonance in the so(5 ) symmetric ladder model . |
similar to the relevant role of the q-@xmath8 potential for meson properties , the three - quark ( 3q ) potential @xcite is directly responsible to the structure and properties of baryons . in spite of the importance of the 3q potential in the hadron physics
, there were only a few preliminary lattice - qcd studies for the 3q potential done in 80 s @xcite .
the 3q ground - state potential @xmath0 can be measured in a gauge - invariant manner using the 3q wilson loop @xmath11 as shown in fig.1 , @xmath12 similar to the derivation of the q-@xmath13 potential from the wilson loop . in principle , @xmath0 can be obtained in the large @xmath14 limit , however , the practical lattice - qcd calculation of @xmath15 becomes severe for large @xmath14 , because @xmath15 decreases exponentially with @xmath14 .
physically , the true ground state of the 3q system is expected to consist of three flux tubes , and the 3q state expressed by the three strings generally includes many excited - state components such as flux - tube vibrational modes .
therefore , for the accurate measurement of the 3q ground - state potential @xmath0 , the ground - state enhancement or the excitation - component reduction by the smearing technique @xcite is practically indispensable .
( this smearing was not applied in refs .
@xcite . ) in this paper , we study the 3q ground - state potential @xmath0 using the ground - state enhancement by the gauge - covariant smearing technique for the link - variable in the su(3)@xmath1 lattice qcd with the standard action with @xmath16=5.7 ( @xmath17 0.19fm ) and @xmath18 24 at the quenched level @xcite .
we consider 16 patterns of the 3q configuration where the three quarks are put on @xmath19 , @xmath20 and @xmath21 in @xmath22 with @xmath23 in the lattice unit .
the junction point @xmath24 , which does not affect @xmath0 , is set at the origin @xmath25 in @xmath22 .
the smearing technique is actually successful for accurate measurements of the q-@xmath8 potential in the lattice qcd @xcite .
the standard smearing for link - variables is performed by the iterative replacement of the spatial link - variable @xmath26 ( @xmath27 ) by the obscured link - variable @xmath28 @xcite which maximizes @xmath29 \right\ } , \quad u_{-\mu}(s ) \equiv u_{\mu}^{\dagger}(s-\hat \mu ) , \ ] ] with a real smearing parameter @xmath30 .
the @xmath31-th smeared link - variable @xmath32 is defined as @xmath33 as an important feature , this smearing procedure keeps the gauge covariance of the smeared link - variable @xmath32 properly .
in fact , the gauge - transformation property of @xmath32 is just the same as that of the original link - variable @xmath34 , and therefore the gauge invariance of @xmath35 is ensured whenever @xmath36 is gauge invariant .
since the smeared link - variable @xmath32 includes a spatial extension , the `` line '' expressed with @xmath32 physically corresponds to a `` flux tube '' with a spatial extension .
therefore , if a suitable smearing is done , the `` line '' of the smeared link - variable is expected to be close to the ground - state flux tube . here , the overlap between the ground - state operator and the 3q - state operator at @xmath37 in the 3q wilson loop @xmath11 is estimated by @xmath38.\ ] ] to get the ground - state - dominant 3q system , we investigate the ground - state overlap @xmath39 for the 3q wilson loop composed of the smeared link - variable @xmath32 in the lattice qcd , and we adopt the smearing with @xmath40 and the iteration number @xmath41 , which largely enhances @xmath39 for all of the 3q configurations in consideration as shown in fig.2 .
in this section , we consider the potential form of @xmath0 based on qcd . in the short - distance limit , the perturbative qcd is applicable and the coulomb - type potential appears as the one - gluon - exchange ( oge ) result . in the long - distance limit at the quenched level , the flux - tube picture would be applicable from the argument of the strong - coupling qcd @xcite , and hence a linear - type confinement potential is expected to appear .
actually , the q-@xmath42 potential @xmath43 is well reproduced by the coulomb - plus - linear potential in the lattice qcd .
then , we conjecture that the 3q ground - state potential @xmath0 is also expressed by a sum of the short - distance oge result and the long - distance flux - tube result as @xmath44 where @xmath7 denotes the minimal length of the total flux tubes linking the three quarks .
next , we consider the diquark limit , where the 3q system becomes equivalent to a q-@xmath42 system , which leads to a physical requirement on the relation between @xmath0 and @xmath43 . here , the constant term is to be considered carefully , because there appears a divergence from the coulomb term in @xmath0 as @xmath45 in the continuum diquark limit . in the lattice regularization ,
this ultraviolet divergence is regularized to be a finite constant with the lattice spacing @xmath46 as @xmath47 , where @xmath48 is a dimensionless constant satisfying @xmath49 and @xmath50 .
thus , we get the diquark - limit requirement on the lattice , @xmath51
we measure the 3q ground - state potential @xmath0 from 210 gauge configurations in the su(3)@xmath1 lattice qcd using the smearing technique , and compare the lattice data with the theoretical form of eq.(5 ) .
table 1 shows the best fit coefficients in @xmath0 and @xmath43 in eq.(5 ) .
table 2 shows the comparison between the lattice qcd data @xmath52 and the fitting function @xmath53 in eq.(5 ) with the three coefficients listed in table 1 .
as shown in table 2 , the three - quark ground - state potential @xmath0 is well described by eq.(5 ) with accuracy better than a few % . from table 1 , we find a universal feature of the string tension , @xmath54 , as well as the oge result for the coulomb coefficient , @xmath55 .
all the diquark - limit requirements in eq.(6 ) are satisfied for @xmath56 .
another fitting with the @xmath57-type flux - tube ansatz @xcite seems rather worse , because of unacceptably large @xmath58 .
however , as an approximation , @xmath0 seems described by a simple sum of an effective two - body q - q potential with a reduced string tension as @xmath59 .
this reduction factor can be naturally understood as a geometrical factor rather than the color factor , since the ratio between @xmath7 and the perimeter length @xmath60 satisfies @xmath61 , which leads to @xmath62 with @xmath63 . .
the string tension , the coulomb coefficient and the constant term for the 3q ground - state potential @xmath0 and the q-@xmath8 potential @xmath64 in the lattice unit with @xmath65 0.19fm . [ cols= " < ,
< , < , < " , ] 9 t.t .
takahashi , h. matsufuru , y. nemoto and h. suganuma , proc . of the tmu - yale symp . on `` dynamics of gauge fields '' ,
tmu , tokyo , dec . 1999 ,
a. chodos et al .
( eds . ) , universal academy press , tokyo ( 2000 ) in press .
s. capstick and n. isgur , phys . rev .
* d34 * ( 1986 ) 2809 .
n. brambilla , g.m .
prosperi and a. vairo , phys . lett .
* b362 * ( 1995 ) 113 .
r. sommer and j. wosiek , phys.lett.*149b * ( 1984 ) 497 ; nucl.phys.*b267 * ( 1986 ) 531 .
thacker , e. eichten and j.c .
sexton , nucl .
* b * ( proc . suppl . ) * 4 * ( 1988 ) 234 .
ape collaboration , m. albanese et al .
, phys . lett .
* b192 * ( 1987 ) 163 .
bali , c. schlichter and k. schilling , phys . rev .
* d51 * ( 1995 ) 5165 .
j. kogut and l. susskind , phys . rev .
* d11 * ( 1975 ) 395 . | with the smearing technique , the three - quark ( 3q ) ground - state potential @xmath0 is numerically extracted in the su(3)@xmath1 lattice qcd monte carlo simulation with @xmath2 and @xmath3 at the quenched level . with accuracy better than a few %
, @xmath0 is well described by a sum of a constant @xmath4 , the two - body coulomb part @xmath5 and the three - body linear confinement part @xmath6 , where @xmath7 denotes the minimal length of the color flux tube linking the three quarks . by comparing with the q-@xmath8 potential , we find a universal feature of the string tension as @xmath9 and the one - gluon - exchange result for the coulomb coefficient as @xmath10 .
all our results including the constant term are consistent with the requirement on the diquark limit in the lattice regularization . |
at light transmission through a qw some characteristics occur in reflected and transmitted light , on which it is possible to judge about electronic processes proceeding in a qw [ 1 - 4 ] .
most interesting results turn out when energy levels of electronic system are discrete .
it takes place in a quantizing magnetic field directed perpendicularly to a qw plane or at taking into account excitonic levels in a zero magnetic field .
two close located energy levels arise in a case of a magnetophonon resonance [ 5 ] , when @xmath1 where @xmath2 is the longitudinal optical @xmath3 phonon frequency , @xmath4 is the integer and @xmath5 is the cyclotron frequency , @xmath6 is the electron charge , @xmath7 is the electron ( hole ) effective mass .
the modern semiconductor technologies allow to make qws of high quality , when radiative broadening of an absorption line may be comparable to contributions of non - radiative mechanisms or to exceed them .
in such situation it is impossible to be limited by linear interaction of an electron with electromagnetic field and it is necessary to take into account all the orders of this interaction [ 6 - 22 ] .
light pulses and monochromatic radiation were considered as stimulating light .
one , two and large number of exited levels were taken into account .
results of all previous works , except [ 19,20 ] , are fair for rather narrow qws , when the condition @xmath8 is satisfied ( @xmath9 is the module of a light wave vector , @xmath10 is the qw width ) .
as calculations show , in case of narrow qws only positions of reflection and absorption peaks depend on the qw width , but not their height and form . in [ 19,20 ]
the theory of light reflection and absorption is constructed for wider qws for which @xmath11 in both works the interaction of light with one exited level was considered , in [ 19 ] - at a monochromatic irradiation , in [ 20 ] - at a pulse irradiation . under condition
( 4 ) results begin to depend on the qw width @xmath12 . in the present work we investigate theoretically reflection , absorption and transmission of monochromatic light through a wide qw in a case , when the interaction of light with two close located energy levels is essential .
we compare new results to conclusions of [ 21 ] devoted to research of similar problems in narrow qws .
the case of normal incidence of light on a semiconductor qw surface , located in a plane @xmath13 , is considered
. qw may be in a zero or quantizing magnetic field perpendicular to a qw surface .
temperature is close to 0 .
light excites electron - hole pairs ( ehps ) . in the theory interband matrix elements
@xmath14 , describing an electron transition from valence band in a conductivity band ( i.e. the ehp creation ) are essential . as well as in previous works , the following model will be used .
vectors @xmath15 for two degenerated valence bands @xmath16 and @xmath17 are @xmath18 where @xmath19 and @xmath20 are the unite vectors along @xmath21 and @xmath22 axis , @xmath23 is the real value .
this model corresponds to heavy holes in a semiconductor with the zinc blend structure , if the @xmath24 axis is directed along a 4-th order axis [ 23,24 ] .
if to use circular polarization vectors of a stimulating light @xmath25 a property of preservation of a polarization vector @xmath26 is preserved . thus ,
neither ehp wave functions , nor energy levels depend on band numbers i or ii . a class of a wave function @xmath27 at @xmath28 is essential in the theory in the effective mass approximation ( @xmath29 is the electron ( hole ) radius - vector ) [ 22 ] . we suppose that it is possible to write down @xmath30 the separation of variables is possible , if the coulomb interaction poorly influences on the movement of these particles in @xmath13 plane .
it occurs , if the quantizing magnetic field is directed along @xmath24 axis and the condition @xmath31 is carried out , where @xmath32 is the radius of a wannier - mott exciton in absence of a magnetic field , @xmath33 is the static permittivity , @xmath34 is the reduced effective mass , @xmath35 is the magnetic length . for @xmath36
it is obtained ( the parameters from [ 26 ] are used ) : @xmath37 where @xmath38 corresponds to magnetic field @xmath39 , which is obtained from ( 1 ) at @xmath40 for a magnetopolaron , containing an electron . according to ( 8b )
we obtain @xmath41 i.e. the condition ( 8 ) is carried out .
influence of the coulomb interaction on the movement of particles in @xmath13 plane was considered in [ 27 ] . in case of free ehps ( without an account of coulomb forces )
@xmath42 where @xmath43 is the electron ( hole ) wave function corresponding to a size quantization quantum number @xmath44 .
for qws of finite depth functions @xmath45 are given , for example , in [ 25 ] .
for infinitely deep qw ( when there is no any tunnel penetration of electrons and holes in a barrier ) @xmath46 @xmath47 in order to neglect by coulomb forces at the description of particles movements along @xmath24 axis , the performance of the condition @xmath48 is required . whether the conditions ( 4 ) and ( 10 ) are compatible ?
for example , for @xmath36 the energy gap @xmath49 , the frequency of stimulating light owes to exceed this value .
the module @xmath50 of a light wave vector corresponds to the frequency @xmath51 , where @xmath52 is the refraction index , @xmath53 is the light velocity in vacuum .
for @xmath36 we have @xmath54 , i. e. if @xmath55 , then @xmath56 , what is comparable to unit .
thus , for @xmath36 it is possible to neglect by coulomb forces at movement along @xmath24 axis to combine with condition @xmath57 for `` wide qws '' only with some stretch .
but even if there are essential deviations from ( 9 ) we obtain the same qualitative results for frequency dependence of light absorption and reflection using unknown function @xmath58 ( see below ) .
we write down the stimulating electrical field extending along @xmath59 axis as @xmath60 where @xmath52 is the refraction index ( identical inside and outside of a qw ) , @xmath61 @xmath62 may correspond to a light pulse of any form [ 22 ] and at excitation by monochromatic light with frequency @xmath63 looks like @xmath64 let us spread out a true field @xmath65 in the fourier integral @xmath66 in [ 22 ] the equation ( see also [ 19 ] ) @xmath67 for the fourier - component @xmath68 is obtained , where @xmath69 , the sum on indexes @xmath70 means summation on excitation levels in a qw .
a set of indexes @xmath71 is separable on two groups @xmath72 concerning , according to ( 8) , to cross and longitudinal wave functions . in case of free ehps the set @xmath73 contains two indexes : @xmath74 and @xmath75 . in a quantizing magnetic field
( far from magnetophonon resonance ) the set @xmath76 includes only one index @xmath77 where @xmath78 is the landau quantum number concerning to electron ( hole ) .
close to the magnetophonon resonance a level with an index @xmath79 splits in two levels with indexes @xmath80 and @xmath81 [ 25,21 ] .
the equality ( 17 ) is caused by the law of preservation of quantum number @xmath79 at ehp creation in case of normal light incidence on qw surface . far away from the magnetophonon resonance energy levels in a quantizing magnetic field
are @xmath82 @xmath83 @xmath84 is the magnetic field , @xmath85 in case of free ehps @xmath86 where @xmath87 is the size quantized electron ( hole ) energy level and in approximation of infinitely deep qw ( which is used at obtaining of ( 15 ) ) @xmath88 close to the magnetophonon resonance for a polaron a ( fig.1 ) , according , for example , to [ 21 ] , we have @xmath89 @xmath90 @xmath91 @xmath92 is the polaron splitting in the exact resonance , when @xmath93 .
the top sign in ( 23 ) corresponds to the top polaron level @xmath80 , bottom - to the bottom polaron level @xmath81 .
the values @xmath94 represent some factors which are included in expressions for radiative broadenings of electronic excitations ( see ( 24 ) ) . in [ 21 ]
these broadenings @xmath95 are calculated in the distance and near the magnetophonon resonance . at @xmath96 ( where @xmath97 is the cross component of a quasi - momentum of an electronic excitation ) we have @xmath98 where @xmath99 far from the magnetophonon resonance @xmath100
i.e. it does not depend on an index @xmath101 .
near the resonance for an excitation , consisting of the polaron @xmath102 ( see fig.1 ) and a hole with @xmath103 @xmath104 where @xmath105 and the top sign corresponds to the term @xmath80 , bottom - to the term @xmath81 . precisely in the resonance @xmath93 , and @xmath106 is identical for terms @xmath80 and @xmath81 and is equal 1/2 .
the factor @xmath107 depends strongly from the value @xmath108 deviation of the cyclotron frequency from the resonant value @xmath109 ( see fig.3 in [ 21 ] ) . in expressions ( 26 ) and
( 27 ) the approximation @xmath110 is used , what corresponds to the effective mass method . at last , the values @xmath111 are non - radiative broadenings of excitations with an index @xmath70 . in [ 21 ] their estimations from below for terms @xmath80 and @xmath81 are given .
dependence of radiative damping @xmath112 ( continuous lines ) and radiative shift @xmath113 of ehp energy ( dashed lines ) on a qw width @xmath12 . on an axis of ordinates
@xmath114 and @xmath115 are shown , @xmath116 is calculated in ( 65 ) .
@xmath117 is the size quantization quantum number of electron ( hole ) , @xmath9 is the module of a light wave vector . ]
let us limit the sum in ( 15 ) on numbers of exited levels by two terms : @xmath118 .
it is possible , if two levels 1 and 2 are located closely to each other and other levels are far from them- on the distance @xmath119 .
thus , @xmath120 the level 1 is characterized by indexes @xmath121 , level 2 - by indexes @xmath122 .
having entered new designations , we rewrite ( 15 ) for a case of two levels as : @xmath123 @xmath124 where @xmath125 is the fourier - component of a field , which we name `` induced '' as against components of an exciting field @xmath126 , @xmath127 @xmath128
@xmath129 @xmath130 the equation ( 31 ) describes electric fields at any @xmath131 , i.e. to the left of qw at @xmath132 , inside qw at @xmath133 and to the right of qw at @xmath134 .
it follows from ( 31 ) and ( 33 ) that to the left of qw the component @xmath135 is proportional to @xmath136 , what corresponds to reflected light ; to the right of qw the component @xmath125 is proportional to @xmath137 , what corresponds to wave extending from left to right .
inside qw the solution is more complicated . the equations ( 30 ) - ( 31 )
may be solved by a method of iterations , if the interaction of a stimulating field with an electronic system is weak .
further we show that it is possible to be limited by a lowest approximation under the condition @xmath138 but we solve the equations ( 30 ) - ( 31 ) precisely , calculating factors @xmath139 and @xmath140 . for this purpose
we multiply ( 30 ) consecutively on @xmath141 and @xmath142 and integrate on @xmath24 from @xmath143 up to @xmath144 .
we obtain the system of two equations @xmath145 where @xmath146 @xmath147 @xmath148 @xmath149 @xmath150 @xmath151 solving the equation system ( 36 ) we obtain @xmath152 let us note the important properties of factors @xmath153 and @xmath154 .
having substituted ( 12 ) in ( 37 ) , we obtain @xmath155 where @xmath156 having substituted ( 33 ) in ( 38 ) , we have @xmath157 what is easy to transform to @xmath158 it follows from ( 41 ) and ( 42 ) @xmath159 @xmath160 in particular , @xmath161 we enter also a designation @xmath162 thus , having substituted ( 39 ) in ( 31 ) and using the formula @xmath163 we basically have solved the task about calculation of induced fields in case of two exited energy levels in a wide qw . further we consider some special cases .
dimensionless light reflection @xmath164 as function of light frequency @xmath165 in case of two excitation energy levels in a wide qw under condition @xmath166 .
continuous and dashed lines are the permitted transitions , dot - and - dash lines are the forbidden transitions .
@xmath167 is the radiative ( non - radiative ) broadening of an exited state . ]
if to put @xmath168 it follows from ( 31 ) that it remains an interaction of light only with one level .
let us enter designations @xmath169 and obtain following expressions for fields at the left and to the right of qw : @xmath170 @xmath171 @xmath172 where @xmath173 if to use ( 9 ) and the function ( 10 ) , it is possible to show the following . if the size quantization quantum numbers @xmath74 and @xmath75 are of identical parity , @xmath174 and @xmath175 , and if they are different , @xmath176 and @xmath177 .
dimensionless light absorption @xmath178 as function of light frequency @xmath165 in case of two excitation energy levels in a wide qw under condition @xmath166 .
continuous and dashed lines are the permitted transitions , dot - and - dash lines are the forbidden transitions .
@xmath167 is the radiative ( non - radiative ) broadening of an exited state . ]
if to reject in expression @xmath179 the non - resonant term @xmath180 , then from ( 49 ) and ( 49a ) we obtain @xmath181 @xmath182 comparing the first equality from ( 50 ) to the expression ( 24 ) , we find that @xmath183 is the radiative broadening of the exited state with energy @xmath184 .
the value @xmath185 means a certain shift of the exited level in qw caused by interaction with light .
if in [ 19,20 ] to put @xmath186 ( where @xmath187 is the refraction index of a qw ( barrier ) substance ) , we obtain the results ( 51 ) and ( 52 ) . in formulas ( 47 ) and
( 48 ) contain misprints .
instead of @xmath188 it follows @xmath189 . ] in case of narrow qws ( @xmath190 ) we obtain @xmath191 and for free electrons and holes ( when ( 9 ) is carried out ) @xmath192 the expressions ( 51 ) and ( 52 ) pass in used , for example , in [ 16 ] .
let us emphasize that in case of wide qws there is a dependence of induced fields on a qw width @xmath12 ( through values @xmath193 and @xmath185 ) , and also an opportunity of interaction of light with ehps , which have @xmath194 , appears .
in narrow qws ( @xmath190 ) under the condition ( 9 ) we obtain from ( 37 ) @xmath195 and it follows from ( 43 ) @xmath196 i.e. light interacts with two levels , which size quantization quantum numbers of electron and hole are identical , i.e. @xmath197 we obtain also @xmath198}.\end{aligned}\ ] ] having substituted ( 56 ) in ( 31 ) , we find @xmath199}.\end{aligned}\ ] ] having substituted ( 57 ) in ( 46 ) , we determine the induced electric fields at the left and to the right of qw : @xmath200 @xmath201 where @xmath202 @xmath203.\ ] ] for an electric field inside a qw we obtain the result @xmath204\}^{-1}\nonumber\\ & \times&\left\{\gamma_{r1}l_1(\omega)\left[e^{i\kappa z } \int _ { -d/2 } ^zdz ' \phi_1 ( z ' ) \right.\right.\nonumber \\ & + & \left.e^{-i\kappa z } \int ^ { d/2 } _ zdz ' \phi_1 ( z ' ) \right ] \nonumber \\ & + & \gamma_{r2}l_2(\omega)\left[e^{i\kappa z } \int _ { -d/2 } ^zdz ' \phi_2 ( z ' ) \right.\nonumber \\ & + & \left.\left.e^{-i\kappa z } \int ^ { d/2 } _ zdz ' \phi_2 ( z ' ) \right ] \right \}.\end{aligned}\ ] ] obviously , formulas ( 58 ) , ( 58a ) and ( 61 ) may be generalized on any number of exited levels in narrow qws ( see , for example , [ 18 ] )
. expressions ( 58 ) , ( 58a ) were used in [ 17,21 ] .
let us consider a special case , when functions @xmath205 for two levels coincide , i.e. @xmath206 if ( 9 ) is carried out , ( 62 ) is right , in particular , at @xmath207 let us admit that only @xmath208 and @xmath209 , energy @xmath210 and @xmath211 ( see [ 21 ] ) and quantum numbers @xmath74 and @xmath75 may differ each from other .
two close located levels of a system consisting of an usual magnetopolaron and a hole concern to such cases . with the help of ( 31 ) and
( 39 ) we obtain @xmath212f(z)\over 1+{i\over 2 } [ \gamma_{r1}l_1(\omega)+\gamma_{r2}l_2(\omega)]j(\kappa)},\end{aligned}\ ] ] @xmath213 is determined in ( 34 ) , @xmath214 is determined in ( 33 ) , @xmath215 @xmath216 for induced electrical fields on the left and to the right of qw with the help ( 46 ) and ( 64 ) we find : @xmath217 @xmath218 @xmath219\ ] ] @xmath220 @xmath221\}^{-1},\ ] ] where @xmath222 we reject non - resonant contributions to @xmath223 and @xmath224 and obtain @xmath225 where @xmath226 ( @xmath227 and @xmath228 are real by definition ) satisfy to the equation @xmath229 and are equal @xmath230 the sign plus concerns to the subscript 1 , and the sign minus - to the subscript 2 , @xmath231 @xmath232 the expression in the root in ( 71 ) , generally speaking , is complex .
designations are entered also @xmath233 @xmath234 \over \omega_1-\omega_2+i(g_2-g_1)/2}\nonumber\\ + { \tilde{\gamma}_{r2}[\omega_1-\omega_1-i(g_1-\gamma_1)]\over \omega_1-\omega_2+i ( g_2-g_1 ) /2}.\end{aligned}\ ] ] let us compare results ( 67 ) and ( 67a ) with substitution ( 69 ) to the formulas ( 51 ) and ( 52 ) for a case of one excitation level in a wide qw .
it is visible that levels with numbers 1 and 2 influence each other , that results into a renormalization of @xmath235 and @xmath236 : they are being replaced accordingly by @xmath237 and @xmath238 . in narrow qws at @xmath239 and at performance ( 9 ) , i.e. without the coulomb forces account , @xmath240 and we obtain results of [ 21 ] for two levels in a narrow qw .
knowing expressions for electric fields on the left and to the right of qw , let us calculate light reflection and absorption .
let us introduce umov - pointing vectors @xmath241 on the left ( to the right ) from qw @xmath242 @xmath243 @xmath244 @xmath245 dimensionless light reflection is defined as @xmath246 light absorption is defined as @xmath247 light transmission is equal @xmath248 ^ 2 \nonumber \\ & + & { 1\over 4}(\tilde{\gamma}_{r1}\gamma_2+\tilde{\gamma}_{r2}\gamma_1)^2\},\end{aligned}\ ] ] @xmath249\nonumber\\ & + & \tilde{\gamma}_{r2}\gamma_2[(\omega_l-\omega_1)^2+\gamma_1 ^ 2/4]+(\delta_1\tilde{\gamma}_{r2}-\delta_2\tilde{\gamma}_{r1 } ) \nonumber\\ & \times&[(\omega_l-\omega_2)\gamma_1- ( \omega_l-\omega_1 ) \gamma_2]\},\end{aligned}\ ] ] where @xmath250[(\omega_l-\omega_2)^2+g_2 ^ 2/4].\end{aligned}\ ] ] using ( 71 ) - ( 73 ) , we transform ( 86 ) to @xmath251 let us emphasize that expressions ( 84 ) and ( 85 ) for light reflection and absorption , as well as expressions ( 67 ) and ( 67a ) for induced electric fields , are right under condition ( 8) , but performance of the condition ( 9 ) is not necessarily .
the coulomb forces may be taken into account , therefore functions @xmath252 will differ from ( 9 ) . in narrow qws at @xmath253
we obtain @xmath254 let us consider some extrem cases . \1 . under condition @xmath190 ,
i.e. for narrow qws , we use ( 76 ) .
then the expressions ( 84 ) and ( 85 ) pass in appropriate formulas from [ 21 ] .
let us emphasize that in the limit @xmath255 for free ehps only @xmath256 is admitted .
having put @xmath257 , we exclude interaction of light with the level 2
. then we obtain results for one level @xmath258},\ ] ] @xmath259}.\ ] ] 3 .
let us consider an extreme case @xmath260 when the perturbation theory on light interaction with an electronic system is applicable . from ( 84 )
and ( 85 ) we find @xmath261[(\omega_l-\omega_2)^2+(\gamma_2/2)^2]},\end{aligned}\ ] ] @xmath262 since @xmath263 and reflection ( 92 ) is square on values @xmath264 , in the rhs of ( 92 ) the factor @xmath265 may be taken out of brackets . similarly in ( 93 ) it is possible to bear out brackets @xmath266 . in brackets
@xmath267 should be replaced by @xmath268 .
thus , dependence of @xmath269 and @xmath270 on qw width is determined by factors @xmath271 and @xmath266 , respectively . for free ehps in a case @xmath272
there appears an opportunity of light interaction with ehps for which @xmath273 . according to ( 93 )
light absorption @xmath270 is the sum of contributions from levels 1 and 2 .
each of these contributions may be obtained from ( 90 ) at @xmath274 . according to ( 92 )
light reflection @xmath269 is square on @xmath275 and @xmath276 and consequently , except of contributions of separate levels , contains an interference contribution .
the following limiting case is opposite to previous and is determined by conditions @xmath277 let us assume in the rhs of ( 84 ) and ( 85 ) @xmath278 then @xmath279 , @xmath280 ^ 2(\omega_l-\omega_0)^2\nonumber\\ & \times&\{[(\omega_l-\omega_1)(\omega_l-\omega_2)\nonumber\\ & - & \delta_1(\omega_l-\omega_2)-\delta_2(\omega_l-\omega_1)]^2\nonumber\\&+&[(\tilde{\gamma}_{r1 } + \tilde{\gamma}_{r2})/2]^2(\omega_l-\omega_0)^2\}^{-1},\end{aligned}\ ] ] where @xmath281 having determined roots of the equation @xmath282 @xmath283 let us transform ( 97 ) to @xmath284 ^ 2(\omega_l-\omega_0)^2\nonumber\\ & \times&\{(\omega_l-\omega_{d1})^2(\omega_l-\omega_{d2})^2 \nonumber\\&+&[(\tilde{\gamma}_{r1 } + \tilde{\gamma}_{r2})/2]^2(\omega_l-\omega_0)^2\}^{-1},\end{aligned}\ ] ] where @xmath285 it follows from ( 99 ) that @xmath286 at @xmath287 , i.e. there is a point of total light transmission through a qw . at @xmath288 or @xmath289 @xmath290 , i.e. light
is completely reflected .
let us compare this result ( 99 ) with the result of [ 21 ] for narrow qws ( @xmath190 ) .
we find that at transition to a case of wide qws values @xmath291 begin to depend on a qw width , the interaction with ehps for which @xmath292 is admitted and points of total reflection are displaced ( at @xmath190 these points are @xmath293 and @xmath294 ) .
but existence of one point of total transmission and two points of total reflection in case of wide qws are preserved ( see fig.7 below ) .
\5 . following extreme case is the smallness of broadenings and energy shifts in comparison to distance @xmath295 between energy levels , i.e. @xmath296 thus , a ratio between non - radiative broadening @xmath297 and values @xmath291 and @xmath298 may be anyone .
let frequency @xmath165 is close to a resonance with a level 1 , i.e. in addition to ( 101 ) the conditions are carried out @xmath299 then from ( 84 ) and ( 85 ) we obtain results ( 89 ) and ( 90 ) , i.e. the second level poorly influences @xmath269 and @xmath270 .
\6 . at last ,
we consider a case of merging levels , when @xmath300 from ( 71 ) we obtain @xmath301 then it follows from ( 84 ) - ( 86 ) @xmath302 @xmath303 comparing obtained results with ( 89 ) and ( 90 ) , we find that in case of twice degenerated exited level the formulas for one non - degenerated level with the doubled values of @xmath304 and @xmath305 are right .
in fig.1 terms of an electron - phonon system in a qw , concerning to a size quantization quantum number @xmath306 , are represented by continuous lines .
it is supposed that phonons , essential at formation of magnetopolarons ( confined or of interface ) , have one frequency @xmath2 without the account of dispersion . on an abscissa axis
the relation @xmath307 is shown , on an ordinates axis the relation @xmath308 is shown , where @xmath309 is the electron ( hole ) energy , counted from energy @xmath310 , appropriate to a size quantization energgy level @xmath306 .
the polaron states correspond to crossing points of terms .
twofold polarons , appropriate to crossing only of two terms , are designated by black circles .
let us consider some point of terms crossing , to which an integer @xmath311 corresponds ( see ( 1 ) ) .
let @xmath79 is the landau quantum number of a level , passing through the given point of terms at @xmath312 ( see fig.1 ) .
then the conditions should be carried out @xmath313 for a twofold polaron existence .
it is easy to see that to @xmath40 one twofold polaron ( designated by the letter @xmath102 ) corresponds .
two twofold polarons @xmath314 and @xmath309 correspond to @xmath315 ( i.e. @xmath316 ) , three double polarons @xmath317 and @xmath318 correspond to @xmath319 ( i.e. @xmath320 ) , et cetera . in fig.1 polarons , located to the left of @xmath321 are not designated . above of twofold polarons threefold polarons are located , appropriate to crossing of three terms , more higher fourfold polarons are located and so on .
number of polarons of each sort is equal @xmath311 at given @xmath322 .
threefold polarons in bulk crystals are considered for the first time in [ 28 ] , in qws - in [ 29 - 31 ] .
let us note that for crossing three and more terms in one point the equidistance of landau levels is necessary . in order the theory
[ 31 ] of threefold polarons should be applicable , it is necessary that the amendments to energy , caused by non - parabolicity of bands or by excitonic effect , should be less than splitting of terms .
but in case of twofold polarons infringement of parabolisity is not an obstacle , since crossing of two terms all the same exists .
all above mentioned polarons correspond to the integer @xmath311 .
besides in fig.1 there are other crossings of lines with quantum number @xmath306 ( continuous lines ) , designated by empty circles .
they correspond to fractional @xmath311 .
as the terms crossed in these points are characterized by @xmath323 , real direct transitions between them with emitting of one phonon are impossible .
let us name such polarons as weak polarons . as the terms are crossed , their splitting is inevitable , but for calculation of splitting we need to take into account transitions between crossing terms through virtual intermediate states or to take into account the small two - phonon contributions in the operator of electron - phonon interaction . in result
splittings @xmath324 of terms in case of weak polarons should be much less , than in a case of integer @xmath311 .
the contributions of transitions through intermediate states in @xmath325 are of more high order than @xmath326 on the frhlich dimensionless coupling constant @xmath327 . at the account of two or more values of size
quantization quantum number @xmath306 the picture of crossing of terms becomes considerably complicated . besides usual polarons , appropriate to a level @xmath328 ( for example polaron @xmath329 ) , occur `` combined '' polarons for which the electron - phonon interaction connects two electronic levels with different numbers @xmath306
. the landau quantum numbers may coincide or be different [ 32,33 ] .
in fig.1 for an example three terms concerning to quantum number @xmath330 ( dot - and - dash lines ) and a position of two combined polarons @xmath331 and @xmath332 are shown . in fig.1
it would be necessary to carry out more dot - and - dash lines and to obtain the greater number of combined polarons .
however , it would strongly complicate figure . for an example in fig.1 the polaron @xmath333
is designated , which is combined and weak .
interesting feature of combined polarons is that that the appropriate resonant values of magnetic fields depend on distance @xmath334 between size quantized levels @xmath306 and @xmath328 and hence on qw depth and width .
really , with the help of fig.1 it is easily to obtain @xmath335 one more kind of combined polarons [ 32 ] is not represented in fig.1 , as it exists only under the condition @xmath336 when , for example , terms @xmath337 and @xmath338 coincide at any magnitudes of a magnetic field . for performance of the resonant condition ( 109 )
a certain distance between levels @xmath306 and @xmath328 is required , what may be reached only by selection of the qw width and depth . in order the picture of fig.1 should be applicable
it is necessary that the distances between the neighbour levels @xmath339 should be much more , than @xmath92 of polaron splittings .
since the distances between levels decrease with growth of the qw width @xmath12 , the restriction from above on @xmath12 is imposed ( numerical estimations see in [ 34 ] ( fig.2 and 3 ) . ) .
it follows from above mentioned classification of magnetopolarons that the technique of calculation of light reflection and absorption stated in section 7 is inapplicable only to combined polarons in a qw , since in case of twofold combined polarons in each of polaron states @xmath340 or @xmath341 the states with different numbers @xmath306 and @xmath328 of size quantization `` are mixed '' and as a result the expression ( 8) is inapplicable . to all other twofold polarons - usual ( black circles ) and weakened ( empty circles ) basic formulas ( 84 ) and ( 85 ) are applicable . as excitation 1 a pair consisting from a hole and a state @xmath340 of polaron appears , as excitation 2 a pair of a hole and state @xmath341 appears .
in narrow qws characterized by an inequality @xmath253 , light interacts only with those pairs for which the size quantization numbers coincide , i.e. @xmath342 . let us name a creation of such ehps by permitted transitions . for wider qws ( when @xmath343 ) forbidden transitions , for which @xmath344 , appear . since energies @xmath345 and @xmath346 , determined in ( 21 ) , depend on numbers @xmath306 , permitted transitions with indexes @xmath347 and @xmath348 are separated on frequency on the value @xmath349 for example , for @xmath36 , using parameters [ 26 ] @xmath350 and qw width @xmath351 , we obtain @xmath352 for wide qws in an interval between the permitted transitions @xmath353 and @xmath354 there are two forbidden transitions at @xmath355 on distance @xmath356 from transition @xmath347 and at @xmath355 on distance @xmath357 from transition @xmath347 . to each of transitions at resonant value @xmath358
there corresponds a doublet of close located maxima of light reflection and absorption .
comparative intensity of reflection and absorption for cases permitted and forbidden transitions is shown in fig.3 - 5 .
same that in fig.3 under condition @xmath359 . ]
let us show still numerical estimations . in [ 34 ] the values @xmath360 of polaron splittings are calculated for some sorts of ordinary double polarons ( see fig.1 ) and some substances of qws and barriers .
dependencies @xmath92 on a qw width @xmath12 are constructed . at @xmath12 from @xmath361
up to @xmath362 @xmath363 for @xmath347 and @xmath348 .
@xmath364 ( according to ( 26 ) with use parameters of @xmath36 from [ 26 ] ) is equal @xmath365 ( see also [ 21 ] ) . at @xmath366 in fig.2 the dependence of @xmath367 and radiative energy shifts @xmath305 on qw width @xmath12 for two permitted transitions @xmath347 and @xmath348 and
two forbidden transitions @xmath368 and @xmath369 is represented ; results for two forbidden transitions precisely coincide , since according to ( 9 ) and ( 10 ) the function @xmath370 does not vary at replacement of an index @xmath74 by @xmath75 and vice verse . at @xmath371
the non - radiative damping @xmath372 , but their values are unknown , in [ 21 ] an attempt is made only to estimate these values from below .
same that in fig.4 under condition @xmath359 . ]
fig.3 corresponds approximately to above mentioned numerical estimations @xmath92 and @xmath373 at @xmath371 .
two arbitrary values @xmath374 are used : @xmath375 and @xmath376 , both of them exceed @xmath377 .
so fig.3 corresponds to a case @xmath378 .
it is visible that @xmath379 and @xmath380 for permitted and for forbidden transitions . comparing the used polaron splittings @xmath381 and @xmath382 for fig.3 we find that @xmath188 is two orders less than @xmath381 .
but it is right only for usual twofold polarons ( black circles in fig.1 ) . in case of weak polarons ( the empty circles ) polaron splittings
are much smaller .
figs.5 , 6 and 7 , 8 may concern to weak polarons , when @xmath383 and @xmath188 may be comparable . in figs.5 , 6 dependencies @xmath384 and @xmath385 are represented in a case @xmath386 . for permitted transition at equality of radiative and non - radiative damping @xmath269 and @xmath270 reach the greatest magnitudes comparable among themselves and comparable with unit , what is visible in figs.5 , 6 . as to forbidden transitions , for them
the condition @xmath387 is satisfied and appropriate @xmath269 and @xmath270 are small . at last , the figs.7 , 8 correspond to an inequality @xmath388 , for which the most interesting results are obtained .
in both cases @xmath389 and @xmath390 for a permitted transition on figs.7 , 8 we obtain results appropriate to ( 99 ) , i.e. equal to zero reflection in a point @xmath391 and total reflection @xmath392 in points @xmath288 and @xmath393 , which position depends on a qw width . to the forbidden transition in figs.7 , 8 at @xmath394
large @xmath269 correspond in maxima .
it occurs because @xmath395 and @xmath396 are comparable .
but peaks are very narrow , since @xmath397 and @xmath398 . in fig.8 , where dependencies @xmath399 are represented , to forbidden transition at @xmath394
the same high and narrow peaks correspond for the same reason .
same that in fig.3 under condition @xmath400 . ]
the dimensionless reflection @xmath269 and absorption @xmath401 are calculated at normal incidence of monochromatic light on a qw surface , which width is comparable to length of a light wave . it is considered a resonance of stimulating radiation with two close located energy levels in a qw , which are two terms of a system consisting from a magnetopolaron and hole .
results are compared with what were obtained earlier in [ 21 ] for narrow qws ( which width is much less than a light waves length ) .
firstly , it is shown that if a light wave length is comparable to a qw width , height and form of peaks of reflection @xmath402 and absorption @xmath403 begin to depend on a qw width .
secondly , it is established , that in case of wide qws appears an interaction of light with excitations , which are characterized by various size quantization numbers of electrons and holes ( @xmath404 ) .
dependence of @xmath405 and @xmath406 on a qw width @xmath12 is various for cases @xmath256 and @xmath404 .
thirdly , it is shown that in case of wide qws an original behaviour of reflection @xmath405 on stimulating light frequency @xmath165 is preserved , if non - radiative broadenings of excitations are much less than radiative broadenings . there
exist two points @xmath288 and @xmath407 of total reflection ( @xmath408 ) and one point @xmath409 between them of total transmission ( @xmath410 ) .
however , the position of points @xmath411 and @xmath412 depends on a qw width . all sequence of processes of absorption and reradiation of light quanta is taken into account , what means an exit outside the perturbation theory on a coupling constant of light and electrons .
it is shown that the perturbation theory is unsuitable , when radiative and non - radiative broadenings are comparable .
same that in fig.4 under condition @xmath400 . ]
this work has been partially supported by the program `` solid state nanostructures physics '' and by the russian scientific - educational programme `` integration '' .
the authors are grateful to m. harkins for critical reading of the manuscript .
i. g. lang , l. i. korovin , a. contreras - solorio , s. t. pavlov , + _ change in the shape of a symmetrical light pulse passing through a quantum well _ , + fiz .
tela , 2000 , * 42*. pp .
2230 - 2244 ( physics of the solid state ( st . petersburg ) , 2000 , * 42 * , n 12 , p. 2300 - 2313 . ) ; cond - mat/0006364 .
i. g. lang , l. i. korovin , a. contreras - solorio , s. t. pavlov , + _ reflection and absorption of light by a quantum well in a quantizing magnetic field under pulsed irradiation _ , + fiz .
tela , 2001 , * 43 * , n 6 , pp . 1117 - 1125 ( physics of the solid state ( st .
petersburg ) , 2001 , * 43 * , n 6 , p. 1159 - 1168 . ) ; cond - mat/ 0004178 .
l. i. korovin , i. g. lang , d. a. contreras - solorio , s. t. pavlov , + _ a role of a spatial dispersion of an electromagnetic wave coming through a quantum well _ , + fiz .
tela , 2001 , * 43 * , n 11 , pp .
2091 - 2199 ( physics of the solid state ( st .
petersburg ) , 2001 , * 43 * , n 11 , pp .
2182 - 2191 ) ; cond - mat/0104262 .
i. g. lang , l. i. korovin , d. a. contreras - solorio , s. t. pavlov , _ transmission of a symmetric light pulse through a wide qw _ , + fiz .
tela , 2002 , * 44 * , n 9 , pp .
1681 - 1689 ( physics of the solid state ( st .
petersburg ) , 2002 , * 44 * , n 9 , pp .
1759 - 1768 . )
; cond - mat/0203390 .
i. g. lang , l. i. korovin , a. contreras - solorio , s. t. pavlov , _ magnetopolaron effects in light reflection and absorption of a three - level system in a quantum well _ , + fiz .
tverd.tela , 2002 , * 44 * , n 11 , pp .
2084 - 2096 ( physics of the solid state ( st .
petersburg ) , 2002 , * 44 * , n 44 , pp .
2181 - 2195 ) ) ; cond - mat/ 0001248 .
l. i. korovin , i. g. lang , s. t. pavlov , + _ combined polaron states in magnetooptical effects in a quantum well _ , + zh .
fiz . , 1999 ,
* 116 * , n 4(10 ) , p. 1419 - 1439 ( jetp , 1999 , * 89 * , n 4 , p. 764 - 784 ) .
l. i. korovin , i. g. lang , s. t. pavlov , + _ magnetopolaron states involving confined phonons in a semiconductor quantum well . spatial correlation of electron and hole _ , zh . eksp .
fiz . , 1999 ,
* 115 * , n 1 , p. 187 - 204
( jetp , 1999 , * 88 * , n 1 , p. 105 | light reflection and absorption spectra by a semiconductor quantum well ( qw ) , which width is comparable to a light wave length of stimulating radiation , are calculated .
a resonance with two close located exited levels is considered .
these levels can arise due to splitting of an energy level of an electron - hole pair ( ehp ) due to magnetopolaron effect , if the qw is in a quantizing magnetic field directed perpendicularly to the qw plane .
it is shown that unlike a case of narrow qws light reflection and absorption depend on a qw width @xmath0 .
the theory is applicable at any ratio of radiative and non - radiative broadenings of electronic excitations . |
the successful prediction of anomalous magnetic moments in early days of particle physics was one of the first triumphs of quantum field theory @xcite . since then it has served as a critical test of the standard model of particle physics , which is based on quantum field theory , guided by gauge principle .
recent measurement of anomalous magnetic moment of muon ( e821 at bnl ) @xcite provides the most stringent test so far , at the level of sub parts per million , of the standard model , @xmath1 which exhibits currently @xmath2 deviation from the standard model estimate : @xmath3 if the discrepancy persists in more improved measurements , such as the e969 experiment , planned at bnl , or other experiments at the fermi lab , it will certainly indicate a hint of new physics .
therefore it is quite necessary to understand the standard model predictions more precisely to pin down the possible hint of new physics at the level of @xmath4 or more .
the theoretical prediction of muon @xmath5 in the standard model consists of three different contributions : @xmath6 among them the qed contribution is most dominant one and has been calculated to be @xmath7 at 4.5 loops by kinoshita et al @xcite , while the weak interaction corrections are found to be @xmath8 at the two - loop level @xcite . as hadrons ( or quarks ) contribute to the anomalous magnetic moment of muon only through quantum fluctuations , the strong interaction contribution is suppressed , compared to that of qed .
the current estimate of the hadronic contributions is @xmath9 the most uncertainty in the sm estimate is however coming from the hadronic contributions , since we poorly understand strong dynamics , while the electroweak corrections can be calculated accurately in perturbation
. direct calculation of hadronic contributions from qcd requires lattice calculations , which are currently not accurate enough to give a meaningful result due to large systematic uncertainties @xcite . in this talk
i present new estimates of the hadronic contributions , based on a holographic model of qcd .
holographic models have been proposed recently for qcd @xcite , inspired by the gauge / gravity duality , found in string theory @xcite , which states that certain strongly coupled gauge theories are equivalent to weakly coupled gravity in one - higher dimension , the extra dimension being the renormalization scale .
the holographic models of qcd were found to be quite successful to account for the properties of hadrons at low energy and give relations to their couplings and also new sum rules @xcite , offering theoretical understanding of phenomenological rules for hadrons found in 60 s such as vector meson dominance , proposed by sakurai @xcite .
holographic qcd ( hqcd ) is a model for 5d gravity dual theory of qcd in the large @xmath10 and large t hooft coupling limit ( @xmath11 ) , describing qcd directly with hadrons . as a gravity dual of qcd with three light flavors we consider @xmath12 flavor gauge theory in a slice of @xmath13 , @xmath14 where the metric is given as
, taking the ads radius @xmath15 , @xmath16 we take the ultraviolet ( uv ) regulator @xmath17 and introduce an infra - red ( ir ) brane at @xmath18 to implement the confinement of qcd , breaking the conformal symmetry @xcite . the bulk scalar ( @xmath19 ) and the bulk gauge fields ( @xmath20 ) are dual to @xmath21 and @xmath22 , respectively . following katz and schwartz @xcite ,
we have introduced a flavor - singlet bulk scalar ( @xmath23 ) for the @xmath24 meson , which is dual to @xmath25 ( @xmath26 ) and is described by @xmath27\label{5}\,.\ ] ] finally we introduce a chern - simons term to reproduce the qcd flavor anomaly @xcite , @xmath28\,,\label{cs}\ ] ] where @xmath29 .
our calculations can be easily applied to other holographic models like sakai - sugimoto model , which will give similar results
. gauge / gravity duality at large @xmath10 and large @xmath30 implies that the generating functional of one - particle irreducible green s functions is given by the classical gravity action of the dual theory : @xmath31=s_{5d{\rm eff}}[\phi(x,\epsilon)]\quad{\rm with}\quad \phi(x,\epsilon)=\phi_{0}(x)\,.\label{ads - cft}\ ] ] using this equality we will be able to calculate the hadronic contributions at the leading order in @xmath32 and @xmath33 expansion .
the strong interaction contributions to the muon magnetic moment consist of three pieces ; the hadronic vacuum polarization ( hlo ) , the higher - order hadronic vacuum - polarization effect , and the hadronic light - by - light ( lbl ) scattering .
the hlo contribution ( see fig . [ hlo ] ) is given as @xcite @xmath34 where the vacuum polarization is defined as @xmath35 and the kernel is @xmath36 from the ads / cft formula ( [ ads - cft ] ) the vector current correlator @xmath37 can then be expressed in terms of the infinite set of the vector meson wave - functions @xmath38 , as shown in fig .
[ vmd ] , which is the holographic realization of the vector meson dominance proposed by sakurai , @xmath39 ^ 2}{(q^2 - m_{v_n}^2 ) m_{v_n}^2 } \ , , \label{piv : res}\ ] ] where the dot denotes the derivative with respect to @xmath40 , @xmath41 . using the holographic renormalization to take care of the uv divergence and keeping only the first four low - lying states
, we find for euclidean momentum @xmath42
@xcite @xmath43 where @xmath44 is the decay constant of @xmath45-th excited vector mesons .
plugging the holographic vacuum polarization ( [ bar : pi ] ) into the formula ( [ g-2 ] ) , we obtain @xcite @xmath46 which agrees , within 30% errors , with the currently updated value @xcite , estimated from new 2009 babar data @xcite @xmath47 |_{\rm babar } = ( 514.1 \pm 3.8 ) \times 10^{-10 } \ , .
\label{babar}\ ] ] for the hadronic light - by - light corrections , shown in fig .
[ lbl diagram ] , we need to calculate 4-point functions of flavor currents .
since there is no quartic term for @xmath48 ( @xmath49 ) , there is no 1pi 4-point function for the em currents in hqcd , because higher order terms like @xmath50 or @xmath51 terms are suppressed . in hqcd
the lbl diagram is therefore dominated by vva or vvp diagrams ( see fig .
[ lbl diagram1 ] ) , which come from the chern - simons term , eq . [ cs ] : @xmath52 where the gauge fields satisfy in the axial gauge , @xmath53 , @xmath54_{\perp}&=&0\,,\label{1}\\ \left[\partial_z\left(\frac{1}{z}\partial_za_\mu^{\hat a}\right)+\frac{q^2}{z}a_\mu^{\hat a}-\frac{g_5 ^ 2v^2}{z^3}a_\mu^{\hat a}\right]_{\perp}&=&0\,,\label{2}\end{aligned}\ ] ] where @xmath55 denotes the projection onto the transversal components . for two flavors
the longitudinal components , @xmath56 , and the phase of bulk scalar @xmath19 are related by eom as @xmath57 the anomalous form factor is then given as @xcite , with @xmath58 and @xmath59 @xmath60,\label{pionff}\ ] ] where we take as boundary conditions @xmath61 and @xmath62 .
-0.2 in for lower part ; @xmath63 for upper part ( brodsky - lepage).,title="fig:",scaledwidth=55.0% ] to calculate the hadronic lbl contribution to the muon anomalous magnetic moment we expand the photon line in the anomalous form factor ( [ pionff ] ) as @xmath64 the result is shown in table [ result ] for several choices of vector - mode truncations @xcite .
[ result ] our results are comparable with recent results by a. nyffeler @xcite , obtained in the lmd+v model ; @xmath65
in the era of electroweak precision , it becomes more important to understand precisely qcd corrections to the electroweak processes . as being non - perturbative strong dynamics , qcd corrections are often difficult to estimate and one resorts to lattice calculations , which are , however , not precise enough for certain measurements such as muon anomalous magnetic moment .
recent development in gauge / gravity duality shows that in the large @xmath10 and large @xmath30 limit the estimate of qcd contributions can be made precisely in holographic qcd , thus may be useful in assessing the new physics effects .
i present recent estimates @xcite of anomalous magnetic moment of muon in holographic qcd , which are found to be in consistent with other calculations .
d.k.h . is grateful to the organizers of scgt09 for a very stimulating meeting and thanks doyoun kim and shinya matsuzaki for the collaborations upon which this talk is based on .
this work is supported by the korea research foundation grant funded by the korean government ( moehrd , basic research promotion fund ) ( krf-2007 - 314- c00052 ) .
99 j. s. schwinger , phys .
82 * , 664 ( 1951 ) .
g. w. bennett _ et al . _ [ muon g-2 collaboration ] , phys .
d * 73 * , 072003 ( 2006 ) . for recent five - loop calculations , see t. aoyama , m. hayakawa , t. kinoshita and m. nio , phys .
d * 78 * , 113006 ( 2008 ) .
a. czarnecki , b. krause and w. j. marciano , phys .
lett . * 76 * , 3267 ( 1996 ) .
t. blum , phys . rev
. lett . * 91 * , 052001 ( 2003 ) .
t. sakai and s. sugimoto , prog .
phys . * 113 * , 843 ( 2005 ) .
j. erlich , e. katz , d. t. son and m. a. stephanov , phys .
rev . lett . * 95 * , 261602 ( 2005 ) ; l. da rold and a. pomarol , nucl .
b * 721 * , 79 ( 2005 ) .
j. m. maldacena , adv .
theor . math .
* 2 * , 231 ( 1998 ) [ int .
j. theor .
phys . * 38 * , 1113 ( 1999 ) ] ; s. s. gubser , i. r. klebanov and a. m. polyakov , phys .
b * 428 * , 105 ( 1998 ) ; e. witten , adv .
* 2 * , 253 ( 1998 ) .
d. k. hong , t. inami and h. u. yee , phys .
b * 646 * , 165 ( 2007 ) .
d. k. hong , m. rho , h. u. yee and p.
yi , phys .
d * 76 * , 061901 ( 2007 ) ; jhep * 0709 * , 063 ( 2007 ) ; phys . rev .
d * 77 * , 014030 ( 2008 ) .
j. j. sakurai , phys .
* 22 * , 981 ( 1969 ) .
b. aubert _ et al . _ [ babar collaboration ] , phys .
lett . * 103 * , 231801 ( 2009 ) .
h. r. grigoryan and a. v. radyushkin , phys .
d * 77 * , 115024 ( 2008 ) .
d. k. hong and d. kim , phys .
b * 680 * , 480 ( 2009 ) .
a. nyffeler , phys .
d * 79 * , 073012 ( 2009 ) . | i present recent calculations of the hadronic contributions to muon anomalous magnetic moment in holographic qcd , based on gauge / gravity duality .
the holographic estimates are compared well with the analysis based on recently revised babar measurements of @xmath0 cross - sections and also with other model calculations for the light - by - light scattering contributions . |
since the discovery of the first gravitationally lensed quasar q0957 + 561 @xcite , about 80 strong lens systems have been found so far .
all of the lensed quasars have image separations smaller than @xmath7 , and they are lensed by massive galaxies ( sometimes with small boosts from surrounding groups or clusters of galaxies ) . the probability that distant quasars are lensed by intervening galaxies was originally estimated by @xcite to be 0.1%1% , assuming that galaxies can be modeled as singular isothermal spheres ( sis ) .
this prediction has been verified by several optical and radio lens surveys , such as the _ hubble space telescope ( hst ) _ snapshot survey @xcite , the jodrell bank / very large array astrometric survey ( jvas ; * ? ? ?
* ) , and the cosmic lens all sky survey ( class ; * ? ? ?
the lensing probability is sensitive to the volume of the universe , so it can be used to place interesting constraints on the cosmological constant @xmath8 ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* but see @xcite ) .
in contrast , lenses with larger image separations should probe a different deflector population : massive dark matter halos that host groups and clusters of galaxies .
such lenses therefore offer valuable and complementary information on structure formation in the universe , including tests of the cold dark matter ( cdm ) paradigm @xcite .
so far the observed lack of large separation lensed quasars has been used to infer that , unlike galaxies , cluster - scale halos can not be modeled as singular isothermal spheres @xcite .
the difference can probably be ascribed to baryonic processes : baryonic infall and cooling have significantly modified the total mass distribution in galaxies but not in clusters @xcite . as a result , large separation lenses may constrain the density profiles of dark matter halos of cluster more directly than small separation lenses @xcite . alternatively , large separation lensed quasars may be used to place limits on the abundance of massive halos if the density profiles are specified @xcite .
better yet , the full distribution of lens image separations may provide a systematic diagnostic of baryonic effects from small to large scales in the cdm scenario .
the fact that clusters have less concentrated mass distributions than galaxies implies that large separation lensed quasars should be less abundant than small separation lensed quasars by one or two orders of magnitude .
this explains why past surveys have failed to unambiguously identify large separation lensed quasars @xcite .
for instance , class found 22 small separation lenses but no large separation lenses among @xmath315000 radio sources @xcite .
although several large separation lensed quasar candidates have been found ( e.g. , * ? ? ?
* ) , they are thought to be physical ( unlensed ) pairs on the basis of individual observations ( e.g. , * ? ? ? * ) or statistical arguments @xcite .
recently @xcite found 6 candidate lens systems with image separations @xmath9 among @xmath320000 quasars in the two - degree field ( 2df ) quasar sample .
given the lack of high - resolution spectra and deep imaging for the systems , however , it seems premature to conclude that they are true lens systems .
we note that because the expected number of lenses with such large image separations in the 2df sample is much less than unity @xcite , these systems would present a severe challenge to standard models if confirmed as lenses .
to find a first unambiguous large separation lensed quasar , we started a project to search for large separation lenses in the quasar sample of the sloan digital sky survey ( sdss ; * ? ? ? * ) .
this project complements ongoing searches for small separation lenses in sdss ( e.g. , * ? ? ?
* ; * ? ? ?
the sdss has completed less than half of its planned observations , but already it contains more than 30000 quasars and is superior to previous large separation lens surveys in several ways .
the full sdss sample will comprise @xmath3100000 quasars , so we ultimately expect to find several large separation lensed quasars @xcite .
one of the most important advantages of the sdss in searching for large separation lensed quasars is that imaging in five broad optical bands allows us to select lens candidates quite efficiently .
recently we reported the discovery of the large separation lensed quasar sdss j1004 + 4112 at @xmath10 @xcite in the sdss .
the quasar itself turned out to be previously identified in the _ rosat _ all sky survey @xcite and the two - micron all - sky survey @xcite , but was not recognized as a lensed system .
@xcite showed that sdss j1004 + 4112 consists of four quasar images with the same redshift from the keck spectroscopy .
the colors of galaxies found by subaru imaging follow - up observations indicated the presence of a cluster of galaxies at @xmath2 .
moreover , the configuration of the four images was successfully reproduced by a simple lens model based on a singular isothermal ellipsoid mass distribution .
all these results strongly implied that sdss j1004 + 4112 is the first quasar lens system due to a massive cluster - scale object . in this paper , we describe photometric and spectroscopic follow - up observations of sdss j1004 + 4112 in detail .
we discuss the spectra of lensed quasar components , including puzzling differences between emission lines seen in the different images .
we analyze deep multicolor imaging data to show the existence of a lensing cluster more robustly .
we also present detailed mass modeling of the lens , and discuss the implications of this system for the statistics of large separation lenses .
the paper is organized as follows .
section [ sec : select ] describes the method used to identify large separation lens candidates in the sdss data .
the results of follow - up observations are summarized in
[ sec : data ] . in
[ sec : model ] we perform mass modeling of sdss j1004 + 4112 , and in [ sec : stat ] we consider the statistical implications of the discovery of sdss j1004 + 4112 .
we summarize our results and conclusions in [ sec : discussion ] . throughout this paper
, we assume the popular `` concordance '' cosmology with @xmath11 , @xmath12 , and @xmath13 km s@xmath14 mpc@xmath14 ( e.g. , * ? ? ?
the sdss is a survey to image a quarter of the celestial sphere at high galactic latitude and to measure spectra of galaxies and quasars found in the imaging data @xcite .
the dedicated 2.5-meter telescope at apache point observatory ( apo ) is equipped with a multi - ccd camera @xcite with five broad bands centered at @xmath15 , @xmath16 , @xmath17 , @xmath18 , and @xmath19 @xcite .
the imaging data are automatically reduced by a photometric pipeline @xcite .
the astrometric positions are accurate to about @xmath20 for sources brighter than @xmath21 @xcite .
the photometric errors are typically less than 0.03 magnitude @xcite . the sdss quasar selection algorithm is presented in @xcite . the sdss spectrographs are used to obtain spectra , covering @xmath22@xmath23 at a resolution of @xmath24@xmath25 , for the quasar candidates .
the public data releases of the sdss are described in @xcite and @xcite .
large separation lens candidates can be identified from the sdss data as follows .
first , we select objects that were initially identified as quasars by the spectroscopic pipeline . specifically , among sdss spectroscopic targets we select all objects that have spectral classification of spec_qso or spec_hiz_qso with confidence z_conf larger than @xmath26 ( see * ? ? ? * for details of the sdss spectral codes ) .
next we check the colors of nearby unresolved sources to see if any of those sources could be an additional quasar image , restricting the lens search to separations @xmath27 .
we define a large separation lens by @xmath28 so that it exceeds the largest image separation lenses found so far : q0957 + 561 with @xmath29 @xcite and rx j0921 + 4529 with @xmath30 @xcite , both of which are produced by galaxies in small clusters .
we regard the stellar object as a candidate companion image if the following color conditions are satisfied : @xmath31 @xmath32 @xmath33 where @xmath34 , @xmath35 , @xmath36 , and @xmath37 , ) are used to denote still - preliminary 2.5m - based photometry ( see * ? ? ?
and @xmath38 denotes the difference between the spectroscopically identified quasar and the nearby stellar object .
note that this selection criterion is tentative ; we still do not know much about large separation lenses , so selection criteria may evolve as we learn more .
our full sample contains 44269 quasars with the redshift distribution shown in figure [ fig : z_qso ] . for the lens search we select the subset of 29811 quasars with @xmath39 , making the redshift cuts for four reasons : ( 1 ) at @xmath40 quasars
are often extended , which can complicated both lens searches and also lens statistics analyses ; ( 2 ) at @xmath40 the sample is contaminated by narrow emission line galaxies ; ( 3 ) at @xmath41 we may miss a number of quasar candidates because of large color errors ; and ( 4 ) lens statistics calculations for high redshift quasars are not very reliable because of uncertainties in the quasar luminosity function @xcite .
lens surveys of high - redshift quasars are of course very interesting for insights into the abundance and formation of distant quasars ; a search for lenses among high - redshift sdss quasars is the subject of a separate analysis by @xcite .
sdss j1004 + 4112 was first selected as a lens candidate based on a pair of components , a and b ( see figure [ fig : sdss1004_i ] ) , where b is the sdss spectroscopic target .
components c and d were identified by visual inspection and found to have colors similar to those of a and b ( even though they do not match the above color criteria ) .
table [ table : sdssphoto ] summarizes the photometry for the four components , and table [ table : astrometry ] gives the astrometry for the four components as well as the galaxy g1 seen in figure [ fig : sdss1004_i ] .
the reason that components c and d have somewhat different colors from b is still unclear , but it must be understood in order to discuss the completeness of the lens survey .
the difference might be ascribed to differential absorption or extinction by intervening material @xcite , or to variability in the source on time scales smaller than the time delays between the images ( e.g. , * ? ? ?
* ) , both of which are effects that become more important as the image separation grows .
since only component b has a spectrum from sdss , we obtained spectra of the other components to investigate the lensing hypothesis .
the first spectroscopic follow - up observations were done on 2003 may 2 and 5 with the double imaging spectrograph of the astrophysical research consortium ( arc ) 3.5-m telescope at apo .
all four components have a prominent emission line ( @xmath42 ) at @xmath43 , indicating that they are quasars with very similar redshifts .
spectra with higher resolution and longer wavelength range were taken on 2003 may 30 with the low - resolution imaging spectrometer ( lris ; * ? ? ?
* ) of the keck i telescope at the w. m. keck observatory on mauna kea , hawaii , usa . the blue grism is 400 line mm@xmath14 , blazed at @xmath44 , @xmath45 pixel@xmath14 , covering @xmath46 to @xmath47 .
the red grating is 300 line mm@xmath14 , blazed at @xmath47 , @xmath45 pixel@xmath14 , covering @xmath47 to the red limit of the detector .
the spectra were obtained with 900 sec exposures and a @xmath48 slit in @xmath49 seeing .
the data were reduced in a standard way using iraf .
the keck / lris spectra are shown in figure [ fig : spec ] .
all four components show emission lines of ly@xmath50 , , , , and .
they have nearly identical redshifts of @xmath1 , with velocity differences less than @xmath51 ( see table [ table : sdssphoto ] ) .
the flux ratios between the images ( see figure [ fig : spec ] ) are almost constant over the wavelength range 30008000 , indicating that these are actual lensed images . from the spectra
, we conclude that the color differences found in the sdss images are mainly caused by differences in the emission lines ( discussed below ) and by slightly different continuum slopes .
several absorption line systems are seen in the spectra .
components a and d have intervening / absorption systems at @xmath52 ; this redshift is similar to that of the foreground galaxies ( [ sec : spec_gal ] ) , suggesting that this absorption system is associated with the lensing galaxies .
component d has additional absorption systems at @xmath53 , @xmath54 , @xmath55 , @xmath56 , and @xmath57 .
figure [ fig : spec_abs ] identifies the various absorbers .
we also note that all four components have absorption lines just blueward of emission lines ( see figure [ fig : spec_line ] ) .
the velocity difference between the emission and absorption lines is @xmath58 , so the absorption system is likely to be associated with the quasar .
the fact that all four components have absorption lines offers further evidence that sdss j1004 + 4112 is indeed a gravitational lens .
figure [ fig : spec_line ] shows notable differences in the emission line profiles in the different components .
one possible explanation is the time delay between the lensed images ; at any given observed epoch , the images represent different epochs in the source frame . however , the fact that the emission lines in components a and b differ seems to rule out the time delay explanation : the expected delay ( see [ sec : modelresult ] ) is shorter than the month or year time scale on which emission lines typically vary ( e.g. , * ? ? ? * ) .
other possible explanations include differences between the viewing angles probed by the images , microlensing amplification of part of the quasar emission region , significant errors in the predicted time delay between a and b , or just that the quasar is extremely unusual .
understanding the puzzling line profile differences will require further observations , preferably including measurement of the time delays and spectroscopic monitoring to identify any variability in the lines .
the spectrum of the galaxy g1 , the brightest object near the center of the quasar configuration ( see figure [ fig : sdss1004_i ] ) , was acquired on 2003 may 30 with lris . the spectrum measured from a 900 sec exposure
is shown in figure [ fig : spec_g ] .
we confirm the break and h&k lines at @xmath59 .
the g - band also appears in the spectrum . from the h&k and mg lines we derive the redshift of g1 as @xmath60 .
the spectra of two additional galaxies near g1 ( see [ sec : imaging ] ) were taken on 2003 june 20 with the faint object camera and spectrograph ( focas ; * ? ? ?
* ) on the subaru 8.2-m telescope of the national astronomical observatory of japan on mauna kea , hawaii , usa .
we used the 300b grism together with the sy47 filter , and took optical spectra covering @xmath61 to @xmath62 with resolution @xmath63 pixel@xmath14 .
the seeing was @xmath64 , and the exposure time was 1740 sec for both galaxies .
the data were reduced in a standard way using iraf .
the spectra are shown in figure [ fig : spec_focas ] .
both galaxies , denoted as g2 and g3 , have @xmath65 , only @xmath66 from the redshift of g1 . a deep @xmath67-band image of sdss j1004 +
4112 was taken on 2003 may 5 with the seaver prototype imaging camera of the arc 3.5-m telescope at apo .
the image shows rich structure , with many galaxies between and around the quasar components suggesting a possible galaxy cluster in the field . for a further check
, we obtained deeper multi - color ( @xmath68 ) images on 2003 may 28 with the subaru prime focus camera ( suprime - cam ; * ? ? ?
* ) on the subaru 8.2-m telescope .
the exposure times and limiting magnitudes of the suprime - cam images are given in table [ table : subaruobs ] .
suprime - cam has a pixel scale of @xmath69 , and the seeing was @xmath70@xmath71 .
the frames were reduced ( bias - subtracted and flat - field corrected ) in a standard way .
the resulting images are shown in figure [ fig : subaru_griz ] .
it is clear that there are many red galaxies around the four images .
moreover , we find three possible lensed arclets ( distorted images of galaxies behind the cluster ) , which are shown in more detail in figure [ fig : subaru_center ] . the fact that the arclets are relatively blue compared with the brighter galaxies in the field ( see figure 2 in * ? ? ?
* ) suggests that the arclets may be images of distant galaxies ( e.g. , * ? ? ?
* ) . confirming that they are lensed images will require higher resolution images and measurements of the arclets redshifts
. if the hypothesis is confirmed , the arclets will provide important additional constraints on the lens mass distribution .
the colors of galaxies in the vicinity of sdss j1004 + 4112 can help us search for the signature of a cluster .
the central regions of clusters are dominated by early - type galaxies ( e.g. , * ? ? ?
* ) that show tight correlations among their photometric properties @xcite .
these correlations make it possible to search for clusters using color - magnitude and/or color - color diagrams @xcite .
we measure the colors of galaxies using the deep suprime - cam @xmath68 images .
object identifications are performed using the source extractor algorithm ( sextractor ; * ? ? ?
* ) ; we identify objects with sextractor parameter class_star smaller than 0.6 in the @xmath72 band image as galaxies .
note that this star / galaxy separation criterion is successful only for objects with @xmath73 .
the magnitudes in the images are calibrated using nearby stars whose magnitudes are taken from the sdss photometric data .
since the red galaxies in clusters are dominant in the central regions , and the center of the cluster is thought to be near g1 , we divide the galaxies in the field into three categories : galaxies inside a @xmath74 ( corresponding to @xmath75 at @xmath2 ) box centered on g1 ; galaxies inside a @xmath76 ( @xmath77 ) box ( except for those in the first category ) ; and galaxies inside a @xmath78 ( @xmath79 ) box ( except for those in the first two categories ) .
figure [ fig : colormag ] shows color - magnitude diagrams for the three categories .
it is clear that the color - magnitude relations , particularly @xmath80 and @xmath81 , show tight correlations for galaxies inside the @xmath74 box .
ridge lines at @xmath82 and @xmath83 strongly suggest a cluster of galaxies at @xmath84 @xcite .
the result is consistent with the keck and subaru spectroscopic results showing that the redshifts of galaxies g1 , g2 , and g3 are all @xmath85 .
we identify cluster members by their location in color - color space @xcite .
we show @xmath86 and @xmath87 color - color diagrams in figures [ fig : cc_gri ] and [ fig : cc_riz ] , respectively .
we restrict the plots to galaxies brighter than @xmath88 because of the limitation of the star / galaxy separation .
we make color - color cuts based on the colors expected of elliptical galaxies @xcite : @xmath89 , @xmath90 , and @xmath91 for elliptical galaxies at @xmath92 . the galaxy distributions with and without the color cuts are shown in figure [ fig : colorcut ] .
the galaxies that survive the color cuts are concentrated around g1 , so we conclude that they are candidate members of a cluster of galaxies at @xmath2 whose center is near g1 .
we note that the distribution of candidate cluster members is not spherical and appears to be elongated north south .
to search for mass models that can explain the image configuration of sdss j1004 + 4112 , we use standard lens modeling techniques as implemented in the software of @xcite .
the main constraints come from the image positions .
we also use the flux ratios as constraints , although we broaden the errorbars to 20% to account for possible systematic effects due to source variability and time delays , micro- or milli - lensing , or differential extinction ( see table [ table : posflux ] for the full set of constraint data ) .
in particular , the different colors of the images and the different absorption features seen in figure [ fig : spec ] suggest that differential extinction may be a significant effect . in this section
we do _ not _ use the position of the main galaxy as a constraint , because we want to understand what constraints can be placed on the center of the lens potential from the lens data alone .
we first consider the simplest possible models for a 4-image lens : an isothermal lens galaxy with a quadrapole produced either by ellipticity in the galaxy or by an external shear .
a spherical isothermal lens galaxy has surface mass density @xmath93 where @xmath94 is the einstein radius of the lens , and @xmath95 is the critical surface mass density for lensing , with @xmath96 , @xmath97 , and @xmath98 being angular diameter distances from the observer to the lens , from the observer to the source , and from the lens to the source , respectively .
the einstein radius is related to the velocity dispersion @xmath99 of the galaxy by @xmath100 for an elliptical model we replace @xmath67 with @xmath101^{1/2}$ ] in the surface density , where @xmath102 and @xmath103 are the axis ratio and position angle of the ellipse .
simple models using either pure ellipticity or pure shear fail miserably , yielding @xmath104 values no better than @xmath105 for @xmath106 degrees of freedom .
this failure is not surprising : most 4-image lenses require _ both _ ellipticity and external shear ( e.g. , * ? ? ?
* ) , and such a situation is likely in sdss j1004 + 4112 since the main galaxy is observed to be elongated and the surrounding cluster surely contributes a shear .
we therefore try models consisting of a singular isothermal ellipsoid ( sie ) plus an external shear @xmath107 .
even though such models are still comparatively simple , they can fit the data very well with a best - fit value of @xmath108 for @xmath109 .
the best - fit model has an einstein radius @xmath110 kpc corresponding to a velocity dispersion of 700 km s@xmath14 , an ellipticity @xmath111 at position angle @xmath112 , and an external shear @xmath113 at position angle @xmath114 . among other known lenses , such
a large shear is found only in lenses lying in cluster environments @xcite .
figure [ fig : sieg - crit ] shows the critical curves and caustics for the best - fit model .
the inferred source position lies very close to the caustic and fairly near a cusp , implying that the total magnification is @xmath357 .
figure [ fig : sieg - params ] shows the allowed ranges for the position of the deflector and the ellipticity and external shear in the model .
the models indicate a small but significant offset of @xmath115 kpc between the center of the lens potential and the main galaxy , although it remains to be seen whether that offset is real or an artifact of these still simple lens models .
even though the simple sie+shear model provides a good fit to the data , we believe that it is not physically plausible because the system clearly has multiple mass components and it seems unlikely that all of the mass is associated with a single @xmath3700 km s@xmath14 isothermal component . the next level of complication is to add a mass component representing the cluster halo .
we still model the galaxy g1 explicitly , treating it as an isothermal ellipsoid constrained by its observed position . at this point
we do not further complicate the model by attempting to treat the other galaxies within the lens explicitly .
we model the cluster component with an nfw profile which has been predicted in cosmological @xmath116-body simulations @xcite : @xmath117 where @xmath118 is a scale radius , @xmath119 is a characteristic overdensity ( which depends on redshift ) , and @xmath120 is the critical density of the universe .
although the nfw density profile appears to deviate from the results of more recent @xmath116-body simulations in the innermost region @xcite , we adopt this form for simplicity .
the lensing properties of a spherical nfw halo are described by the lens potential @xcite @xmath121,\ ] ] where the lensing strength is specified by the parameter @xmath122 since asphericity in the cluster potential is important in modeling this system , we generalize the spherical model by adopting elliptical symmetry in the potential . making the potential ( rather than the density ) elliptical
makes it possible to compute the lensing properties of an nfw halo analytically @xcite
. we may still be over - simplifying the mass model , because the cluster profile may have been modified from the nfw form by baryonic processes such as gas cooling @xcite , and the cluster may have a complex angular structure if it is not relaxed ( e.g. , * ? ? ?
* ) . to allow for the latter possibility
, we still include a tidal shear in the lens model that can approximate the effects of complex structure in the outer parts of the cluster .
overall , our goal is not to model all of the complexities of the lens potential , but to make the minimal realistic model and see what we can learn .
even with our simplifying assumptions , we still have a complex parameter space with 11 parameters defining the lens potential : the mass , ellipticity , and position angle for the galaxy g1 ; the position , mass , scale radius , ellipticity , and position angle for the cluster ; and the amplitude and position angle of the shear .
there are also three parameters for the source ( position and flux ) . with just 12 constraints ( position and flux for each of four images ) , the models are under - constrained .
we therefore expect that there may be a range of lens models that can fit the data . to search the parameter space and identify the range of models , we follow the technique introduced by @xcite for many - parameter lens modeling .
specifically , we pick random starting points in the parameter space and then run an optimization routine to find a ( local ) minimum in the @xmath104 surface . repeating that process numerous times
should reveal different minima and thereby sample the full range of models .
many of the recovered models actually lie in local minima that do not represent good fits to the data , so we only keep recovered models with @xmath123 ( which represents the 3@xmath99 limit relative to a perfect fit when examining two - dimensional slices of the allowed parameter range ; see * ? ? ?
we make one further cut on the models . from the previous section , we know that an sie+shear lens model can give a good fit to the data .
thus , there are acceptable two - component models where most or all of the mass is in the galaxy component and the cluster contribution is negligible .
to exclude such models as physically implausible , we impose an upper limit on the velocity dispersion of the model galaxy .
specifically , we only keep models with @xmath124 km s@xmath14 , because there are essentially no galaxies in the observed universe with larger velocity dispersions , even in rich clusters ( e.g. , * ? ?
* ; * ? ? ?
* ; * ? ? ?
formally , we impose this cut as an upper limit @xmath125 on the einstein radius of the galaxy g1 . we first consider models where the scale radius of the cluster is fixed as @xmath126 ( we shall justify this choice below ) .
figure [ fig : clus - params ] shows the allowed parameter ranges for acceptable models .
first , panel ( a ) shows that the cluster component is restricted to a fairly small ( but not excessively narrow ) range of positions near the center of the image configuration .
this is mainly a result of our upper limit on the mass of the galaxy component ; there is a certain enclosed mass implied by the image separation , and if the galaxy can not contain all of that mass then the cluster component must lie within the image configuration to make up the difference .
it is interesting to note that even in these more complicated models there still seems to be a small offset between the center of the cluster component and the brightest cluster galaxy g1 , although the lower limit implied by our models is just @xmath127kpc . figure [ fig : clus - params]b shows that the allowed values for the ellipticity and position angle of the galaxy g1 basically fill the parameter space , so these parameters are not constrained by the lens data .
we might want to impose an external constraint , however .
analyses of other lens systems show that the lensing mass is typically aligned with the projected light distribution @xcite .
we may therefore prefer lens models where the model galaxy is at least roughly aligned with the observed galaxy , which has a position angle of @xmath128 . to illustrate this possible selection , we show all models but highlight those where the position angle of the model galaxy is in the range @xmath129 .
the broad @xmath130 uncertainties prevent this constraint from being too strong . figure [ fig : clus - params]c shows that there are some acceptable models where the cluster potential is round , but most models have some ellipticity that is aligned roughly north south .
this is in good agreement with the distribution of member galaxies which is also aligned roughly north south ( see figure [ fig : colorcut ] ) .
the ellipticity @xmath1310.4 is actually quite large , considering that this parameter describes the ellipticity of the potential , not that of the density .
figure [ fig : clus - params]d shows that all of the acceptable models require a fairly large tidal shear @xmath132 , and models where the galaxy is aligned with the observed galaxy have a strong shear @xmath133 .
the shear tends to be aligned east west .
the fact that the models want both a large cluster ellipticity and a large tidal shear strongly suggest that there is complex structure in the cluster potential outside of the image configuration
. it would be interesting to see whether there is any evidence for such structure in , for example , x - rays from the cluster .
figure [ fig : clus - crit ] shows critical curves and caustics for sample lens models .
the critical curves are not well determined .
the outer , tangential critical curve can point either northeast ( panel e ) or northwest ( panel d ) , or it can have a complex shape ( panel a ) .
sometimes there is just one inner , radial critical curve ( panel e ) , but often there are two ( panel c ) .
the distance of the source from the caustic ( and of the images from the critical curve ) varies from model to model , so the total magnification can range from @xmath350 to several hundred or even more . finally , perhaps the most interesting qualitative result is that even the image parities are not uniquely determined . in most models ( e.g. , panels a
f ) images a and d lie inside the critical curve and have negative parity while b and c lie outside the critical curve and have positive parity .
however , in some models ( e.g. , panels g h ) the situation is reversed .
having ambiguous image parities is very rare in lens modeling .
so far we have only discussed models where the cluster has a scale radius @xmath134 .
we have also computed models with @xmath135 arcsec and we find that all of the results are quite similar .
to understand what value of the scale radius is reasonable , we must consider which ( if any ) of the models have physically plausible cluster parameters . even though nfw models are formally specified by two parameters @xmath118 and @xmath136 , @xmath116-body simulations reveal that the two parameters are actually correlated .
nfw models therefore appear to form a one - parameter family of models , although with some scatter which reflects the scatter of the concentration parameter @xmath137 ( @xmath138 is a virial radius of the cluster ) .
figure [ fig : nfwnorm ] shows the predicted relation between @xmath118 and @xmath139 , including the scatter . for comparison
, it also shows the fitted values of @xmath139 in lens models with different scale radii .
models with @xmath140 or @xmath141 require @xmath139 much larger than expected , corresponding to a halo that is too concentrated .
models with @xmath142 , by contrast , overlap with the predictions and thus are physically plausible .
we can therefore conclude very roughly that the cluster component must have @xmath143 and a total virial mass @xmath144 .
finally , we can use the models to predict the time delays between the images .
the models always predict that the time delay between images c and d is the longest and the delay between a and b is the shortest .
however , there is no robust prediction of the temporal ordering : most models predict that the sequence should be c
b a d , but a few models predict the reverse ordering d a b
this is a direct result of the ambiguity in the image parities , because the leading image is always a positive - parity image ( e.g. , * ? ? ?
we note , however , that all of the models where the model galaxy is roughly aligned with the observed galaxy have the c b a d ordering .
figure [ fig : clus - tdel ] shows the predictions for the long and short time delays . the long delay between c and d can be anything up to @xmath145 days , while the short delay between a and b can be up to @xmath146 days . for the models where
the galaxy is roughly aligned with the observed galaxy , the two delays are approximately proportional to each other with @xmath147 .
these results have several important implications .
first , the a b time delay should be on the order of weeks or months , so it should be very feasible to measure it , provided that the source has detectable variations .
measuring the a
b delay would be very useful because it would determine the temporal ordering , and thereby robustly determine the image parities .
in addition , it would allow a good estimate of the long c d delay and indicate whether attempting to measure that delay would be worthwhile .
second , the enormous range of predicted time delays means that constraining the hubble constant with this system @xcite will be difficult because of large systematic uncertainties in the lens models .
although @xcite recently showed that it is possible to obtain useful constraints on the hubble constant even in a complex system with two mass components , the analysis is very complex and requires extensive data including not just the image positions and all of the time delays , but also an einstein ring image and the velocity dispersion of one of the mass components . even if we obtain such data for sdss j1004 + 4112 in the near future , it seems likely that it will be difficult to obtain reliable constraints on the hubble constant given the complexity of the lens potential in sdss j1004 + 4112 .
the time delays , however , would still be extremely useful , because they would determine the temporal ordering and hence the image parities , and they would provide constraints that can rule out many of the models that are currently acceptable .
in this section , we calculate the expected rate of large separation lensing in the sdss quasar sample .
the discovery of sdss j1004 + 4112 allows us to move past the upper limits obtained from previous large separation lens searches , although at present the main thing we can do is test whether the detection of one large separation lens in the current sample is consistent with standard theoretical models in the cdm scenario .
we calculate lensing probabilities using spherical models for simplicity .
although halos in cdm simulations are in fact triaxial ( e.g. , * ? ? ?
* ) , the spherical assumption is often adopted in lens statistics calculations because deviations from spherical symmetry mainly affect image multiplicities , not image separations or the total optical depth for lensing ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
while this result has been obtained for isothermal lens potentials , checking it for more general halos is beyond the scope of this paper and is the subject of a follow - up analysis ( m. oguri et al .
, in prep . ) .
let the physical image position in the lens plane and physical source position in the source plane as @xmath148 and @xmath149 , respectively .
consider the probability that a quasar at @xmath150 with luminosity @xmath151 is strongly lensed .
the probability of lensing with image separation larger than @xmath152 is given by @xmath153 where @xmath154 is the cross section for lensing , with @xmath155 being the physical radius of the radial caustic in the source plane .
the lower limit of the mass integral is the mass @xmath156 that corresponds to the image separation @xmath152 ; this can be computed once the density profile of the lens object is specified .
the magnification bias @xmath157 is @xcite @xmath158 where @xmath159 is the luminosity function of source quasars .
note that the magnification factor @xmath160 may be interpreted as the total magnification or the magnification of the brighter or fainter image , depending on the observational selection criteria @xcite . in this paper
, we adopt the magnification of the fainter image , because we concentrate on the large separation lenses for which the images are completely deblended . the lensing probability distribution at large separation reflects the properties of dark halos , rather than galaxies @xcite .
for the statistics calculation , the debate over the inner slope of the density profile seen in @xmath116-body simulations leads us to consider the generalized version @xcite of the nfw density profile ( eq . [ [ nfw ] ] ) : @xmath161 while the correct value of @xmath50 is still unclear , the existence of cusps with @xmath162 has been established in recent @xmath116-body simulations @xcite .
the case @xmath163 corresponds to the original nfw profile , while the case @xmath164 resembles the profile proposed by @xcite .
the scale radius @xmath118 is related to the concentration parameter as @xmath137 .
then the characteristic density @xmath165 is given in terms of the concentration parameter : @xmath166 where @xmath167 is @xmath168 with @xmath169 being the hypergeometric function ( e.g. , * ? ? ?
mean _ overdensity @xmath170 can be computed using the nonlinear spherical collapse model ( e.g. , * ? ? ?
we define @xmath171 and @xmath172 .
then the lensing deflection angle @xmath173 is related to the dark halo profile as follows : @xmath174 the lensing strength parameter @xmath139 was defined in equation ( [ kappas ] ) . for sources inside the caustic ( @xmath175 ) , the lens equation has three solutions @xmath176 , where image # 1 is on the same side of the lens as the source and images # 2 and # 3 are on the opposite side .
the lens image separation is then @xmath177 where @xmath178 is a radius of the tangential critical curve @xcite .
the magnification of the fainter image may be approximated by @xcite @xmath179 these approximations are sufficiently accurate over the range of interest here ( see * ? ? ? * ) .
although we adopt a selection criterion that the flux ratios should be smaller than @xmath180 , this condition does not affect our theoretical predictions because the flux ratios of strong lensing by nfw halos are typically much smaller than @xmath180 @xcite .
the concentration parameter @xmath181 depends on a halo s mass and redshift .
moreover , even halos with the same mass and redshift show significant scatter in the concentration which reflects the difference in formation epoch @xcite , and which is well described by a log - normal distribution . for the median of this distribution ,
we adopt the mass and redshift dependence reported by @xcite as a canonical model : @xmath182 where @xmath183 is the mass collapsing at redshift @xmath184 ( defined by @xmath185 ) . to study uncertainties related to the concentration distribution we also consider other mass and redshift dependences , e.g. , @xmath186 from @xcite , and @xmath187 from @xcite , with @xmath188 being the collapse redshift of the halo of mass @xmath189 .
note that these relations were derived under the assumption of @xmath163 .
we can extend them to @xmath190 by multiplying the concentration by a factor @xmath191 @xcite .
the statistics of large separation lenses are highly sensitive to the degree of scatter in the concentration @xcite .
* see also @xcite ) found @xmath192 in their simulations .
@xcite found a smaller scatter @xmath193 among well relaxed halos , but @xcite found @xmath194 if all halos are considered .
therefore we use @xmath195 throughout the paper . for the comoving mass function of dark matter halos , unless otherwise specified we adopt equation ( b3 ) of @xcite : @xmath196 where @xmath197 , @xmath198 , and @xmath199 .
we use the approximation of @xmath200 given by @xcite , and the shape parameter presented by @xcite . note that this mass function is given in terms of the mean overdensity @xmath201 instead of @xmath170 .
therefore , the mass function should be converted correctly ( e.g. , * ? ? ?
* ) . to study uncertainties related to the mass function we also consider two other possibilities : the mass function derived in the hubble volume simulations , @xmath202 , which is given by equation ( [ mf_jenkins ] ) with @xmath203 , @xmath204 , @xmath205 in terms of the mean overdensity @xmath206 @xcite ; and the mass function given by @xcite @xmath207\nonumber\\ & & \times\sqrt{\frac{2a}{\pi}}\frac{\delta_{\rm c}}{\sigma_m(z)}\frac{d\ln\sigma_m^{-1}}{dm}\exp\left(-\frac{a\delta_{\rm c}^2}{2\sigma_m^2(z)}\right),\end{aligned}\ ] ] with @xmath208 , @xmath209 , @xmath210 in terms of the mean overdensity @xmath201 @xcite .
because the lensing probability depends on the source redshift and luminosity , we compute the predicted number of lenses in redshift and luminosity bins and then sum the bins .
specifically , let @xmath211 by the number of quasars in a redshift range @xmath212 that have a magnitude in the range @xmath213 .
then the predicted total number of lensed quasars is @xmath214 we adopt bins of width @xmath215 and @xmath216 .
the quasar sample we used comprises 29811 quasars with mean redshift @xmath217 ( see figure [ fig : z_qso ] ) , and roughly corresponds to a sample with magnitude limit @xmath218 @xcite . to calculate the @xmath219-band absolute luminosity @xmath220 corresponding to observed magnitude @xmath221
, we must estimate the cross - filter k - correction @xmath222 . the k - correction calculated from the composite quasar spectrum created from the sdss sample by @xcite
is shown in figure [ fig : kcor ] . as a simplification , one might use the following approximation : @xmath223 where the offset 0.12 mainly arises from the difference between @xmath224 and @xmath219 magnitudes ( calculated assuming @xmath225 ; * ? ? ?
throughout the paper , however , we use the k - correction directly calculated from composite quasar spectrum .
the luminosity function of quasars is needed to compute magnification bias .
we adopt the standard double power law @xmath219-band luminosity function @xcite @xmath226^{\beta_l}+[l / l_*(z_{\rm s})]^{\beta_h}}\frac{dl}{l_*(z_{\rm s})}.\ ] ] as a fiducial model of the evolution of the break luminosity , we assume the form proposed by @xcite , @xmath227 where a power - law spectral distribution for quasar spectrum has been assumed , @xmath228 .
@xcite determined the parameters so as to reproduce the low - redshift luminosity function as well as the space density of high - redshift quasars for a model with @xmath229 below @xmath230 , @xmath231 above @xmath230 , and @xmath232 .
the resulting parameters are @xmath233 , @xmath234 , @xmath235 , @xmath236 , and @xmath237 .
we call this model lf1 . to estimate the systematic effect
, we also use another quasar luminosity function ( lf2 ) derived by @xcite : @xmath238 , @xmath239 and an evolution of the break luminosity @xmath240 with @xmath241 , @xmath242 , and @xmath243 .
first we show the conditional probability distributions @xmath244 @xmath245 in order to identify the statistically reasonable ranges of redshift and mass for the lensing cluster .
figure [ fig : dist_z ] shows the conditional probability distribution of the lens redshift , and figure [ fig : dist_m ] shows the conditional probability distribution of the lens mass , given that the gravitational lens system sdss j1004 + 4112 has image separation @xmath246 , source redshift @xmath247 , and apparent magnitude @xmath248 .
we find the most probable lens redshift to be @xmath249 , but the distribution is broad and the measured redshift @xmath250 is fully consistent with the distribution .
we also find a cluster mass @xmath251@xmath252 to be most probable for this system .
this result is in good agreement with the mass estimated from the lens models ( see figure [ fig : nfwnorm ] ) .
note that we do not include information on the measured redshift @xmath250 in figure [ fig : dist_m ] , which might cause a slight underestimate of the lens mass .
next we consider the statistical implications of sdss j1004 + 4112 .
although our large separation lens search is still preliminary , and we have several other candidates from the current sdss sample that still need follow - up observations , we can say that the current sample contains _ at least _
one large separation lens system .
this is enough for useful constraints because of the complementary constraints available from the lack of large separation lenses in previous lens surveys . among the previous large separation lens surveys
, we adopt the class @xmath253 survey comprising a statistically complete sample of 9284 flat - spectrum radio sources @xcite . for the class sample
, we use a source redshift @xmath254 @xcite and a flux distribution @xmath255 @xcite to compute the expected number of large separation lenses .
figure [ fig : cont ] shows contours of the predicted number of large separation lenses with @xmath28 in the sdss quasar sample .
since the number of lenses is very sensitive to both the inner slope of the density profile @xmath50 and the mass fluctuation normalization @xmath6 ( e.g. , * ? ? ?
* ) , we draw contours in the @xmath256 plane . constraints from the existence of sdss j1004 + 4112 together with the lack of large separation lenses in the class sample are shown in figure [ fig : limit ] .
to explain both observations , we need a relatively large @xmath50 or @xmath6 , such as @xmath4 ( 95% confidence ) for @xmath164 .
this value is consistent with other observations such as cosmic microwave background anisotropies @xcite .
by contrast , if we adopt @xmath163 then the required value of @xmath6 is quite large , @xmath257 .
thus , our result might be interpreted as implying that dark matter halos have cusps steeper than @xmath163 .
alternatives to collisionless cdm , such as self - interacting dark matter @xcite or warm dark matter @xcite , tend to produce less concentrated mass distributions which are effectively expressed by low @xmath50 ; such models would fail to explain the discovery of sdss j1004 + 4112 unless @xmath6 is unexpectedly large .
this result is consistent with results from strong lensing of galaxies by clusters ( i.e. , giant arcs ) , which also favors the collisionless cdm model ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* but see @xcite for different conclusion ) .
we note that the abundance of large separation lenses produces a degeneracy between @xmath50 and @xmath6 seen in figure [ fig : limit ] , but additional statistics such as the distribution of time delays can break the degeneracy @xcite .
table [ table : lens ] summarizes the sensitivity of our predictions to various model parameters .
the uncertainties in our predictions are no more than a factor of 23 , dominated by uncertainties in the concentration parameter and the matter density @xmath258 .
this error roughly corresponds to @xmath259 , and so does not significantly change our main results .
there may be larger systematic errors associated with effects we have not considered in this paper . for example , triaxiality in cluster halos is very important in arc statistics because it can dramatically increase the length of the tangential caustic that gives rise to giant arcs @xcite .
the effect would seem to be less important in the statistics of lensed quasars , which depend mainly on the area enclosed by the radial caustic , but it still needs to be examined quantitatively .
the presence of a central galaxy is thought to have a small impact on arc statistics @xcite , but the complexity of the lens potentials we found in our mass modeling suggests that the effect of the galaxy on the statistics of lensed quasars also needs to be considered .
we have presented the candidate selection and follow - up observations of the first cluster - scale lensed quasar , sdss j1004 + 4112 .
the system consists of four components with image separation @xmath260 , and was selected from the large separation lens search in the sdss .
the spectroscopic and photometric follow - up observations confirm sdss j1004 + 4112 to be a lens system ; spectroscopic observations of four components showed that they have nearly identical spectra with @xmath1 .
deep images and spectroscopy of nearby galaxies indicate that there is a cluster of galaxies with @xmath2 , whose center is likely to be among the four components .
we conclude that the cluster is responsible for this large separation lens .
differences between the emission line profiles in the four images remain puzzling , and it will be interesting to reobserve the profiles at several epochs to search for variability that might explain the differences . we have shown that reasonable mass models can successfully reproduce the observed properties of the lens . when we consider models that include both the cluster potential and the brightest cluster galaxy , we find a broad range of acceptable models . despite the diversity in the models , we find several general and interesting conclusions .
first , there appears to be a small ( @xmath261 kpc ) offset between the brightest cluster galaxy and the center of the cluster potential .
such an offset is fairly common in clusters ( e.g. , * ? ? ?
second , the cluster potential is inferred to be elongated roughly north south , which is consistent with the observed distribution of apparent member galaxies .
third , we found that a significant external shear @xmath262 is needed to fit the data , even when we allow the cluster potential to be elliptical .
this may imply that the structure of the cluster potential outside of the images is more complicated than simple elliptical symmetry .
fourth , given the broad range of acceptable models , we can not determine even the parities and temporal ordering of the images , much less the amplitudes of the time delays between the images .
measurements of any of the time delays would therefore provide powerful new constraints on the models .
we note that the complexity of the lens potential means that the time delays will be more useful for constraining the mass model than for trying to measure the hubble constant .
our modeling results suggest that further progress will require new data ( rather than refinements of current data ) .
the interesting possibilities include catalogs of confirmed cluster members , x - ray observations , and weak lensing maps , not to mention measurement of time delays and confirmation of lensed arcs ( either the possible arclets we have identified , or others ) .
for instance , with an estimated cluster mass of @xmath263 , the estimated x - ray bolometric flux is @xmath264 , which means that the cluster should be accessible with the _ chandra _ and _ xmm - newton _ x - ray observatories ; the excellent spatial of _ chandra _ may be particularly useful for separating the diffuse cluster component from the bright quasar images ( which have a total x - ray flux @xmath265 in the _ rosat _ all sky survey )
. the confirmation of lensed arclets would be very valuable , as they would provide many more pixels worth of constraints on the complicated lens potential . in principle , mapping radio jets in the quasar images could unambiguously reveal the image parities ( e.g. , * ? ? ? * ; * ? ? ?
* ) , but unfortunately the quasar appears to be radio quiet as it is not detected in radio sky surveys such as the first survey @xcite . although the large separation lens search in the sdss is still underway , we can already constrain model parameters from the discovery of sdss j1004 + 4112 .
the existence of at least one large separation lens in sdss places a lower limit on the lensing probability that complements the upper limits from previous surveys .
both results can be explained if clusters have the density profiles predicted in the collisionless cdm scenario and moderate values of the mass fluctuation parameter @xmath6 .
in particular we find @xmath4 ( 95% confidence ) assuming the inner density profile of dark matter halos has the form @xmath266 with @xmath164 .
the value of @xmath6 is , however , degenerate with @xmath50 such that smaller values of @xmath50 require larger values of @xmath6 .
various systematic errors are estimated to be @xmath267 , dominated by uncertainties in the distribution of the concentration parameter @xmath181 and in the matter density parameter @xmath258 .
other systematic effects , such as triaxiality in the cluster potential and the presence of a central galaxy , remain to be considered .
still , our overall conclusion is that the discovery of sdss j1004 + 4112 is fully consistent with the standard model of structure formation ( i.e. , cdm with @xmath268 ) . in summary ,
sdss j1004 + 4112 is a fascinating new lens system that illustrates how large separation lenses can be used to probe the properties of clusters and test models of structure formation .
the full sdss sample is expected to contain several more large separation lenses .
the complete sample of lenses , and the distribution of their image separations , will be extremely useful for understanding the assembly of structures from galaxies to clusters . more immediately
, the discovery of a quasar lensed by a cluster of galaxies fulfills long - established theoretical predictions and resolves uncertainties left by previously unsuccessful searches .
we thank don york for valuable comments on the manuscript , and an anonymous referee for many useful suggestions .
part of the work reported here was done at the institute of geophysics and planetary physics , under the auspices of the u.s .
department of energy by lawrence livermore national laboratory under contract no . w-7405-eng-48 .
funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , the university of pittsburgh , princeton university , the united states naval observatory , and the university of washington .
this work is based in part on data collected at subaru telescope , which is operated by the national astronomical observatory of japan .
some of the data presented herein were obtained at the w. m. keck observatory , which is operated as a scientific partnership between the california institute of technology , the university of california , and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w. m.keck foundation .
this work is also based in part on observations obtained with the apache point observatory 3.5-meter telescope , which is owned and operated by the astrophysical research consortium .
we thank the staffs of subaru , keck , and apo 3.5-meter telescopes for their excellent assistance .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
abazajian , k. , et al .
2003 , , 126 , 2081 bahcall , j. n. , maoz , d. , doxsey , r. , schneider , d. p. , bahcall , n. a. , lahav , o. , & yanny , b. 1992 , , 387 , 56 barkana , r. , leh ' ar , j. , falco , e. e. , grogin , n. a. , keeton , c. r. , & shapiro , i. i. 1999 , , 520 , 479 barkhouse , w. a. , & hall , p. b. 2001 , , 121 , 2843 bartelmann , m. 1996 , , 313 , 697 becker , r. h. , white , r. l. , & helfand , d. j. 1995 , , 450 , 559 bernardi , m. , et al . 2003 , , 125 , 1817 bertin , e. , & arnouts , s. 1996 , , 117 , 393 blanton , m. r. , lin , h. , lupton , r. h. , maley , f. m. , young , n. , zehavi , i. , & loveday , j. 2003 , , 125 , 2276 blumenthal , g. r. , faber , s. m. , flores , r. , & primack , j. r. 1986 , , 301 , 27 bode , p. , ostriker , j. p. , & turok , n. 2001 , , 556 , 93 bower , r. g. , lucey , j. r. , & ellis , r. s. 1992 , , 254 , 601 boyle , b. j. , shanks , t. , croom , s. m. , smith , r. j. , miller , l. , loaring , n. , & heymans , c. 2000 , , 317 , 1014 boyle , b. j. , shanks , t. , & peterson , b. a. 1988 , , 235 , 935 bullock , j. s. , kolatt , t. s. , sigad , y. , somerville , r. s. , kravtsov , a. v. , klypin , a. a. , primack , j. r. , & dekel , a. 2001 , , 321 , 559 burud , i. , et al .
1998 , , 501 , l5 cao , l. , wei , j .- y .
, & hu , j .- y . 1999 , , 135 , 243 cen , r. , gott , j. r. , ostriker , j. p. , & turner , e. l. 1994 , , 423 , 1 chae , k .- h .
2003 , , 346 , 746 chae , k .- h . , et al .
2002 , phys .
lett . , 89 , 151301 chen , d. 2003 , , 587 , l55 chiba , m. , & yoshii , y. 1999 , , 510 , 42 colley , w. n. , tyson , j. a. , & turner , e. l. 1996 , , 461 , l83 cooray , a. , hu , w. , & miralda - escud ' e , j. 2000 , , 535 , l9 dalal , n. , holder , g. , & hennawi , j. f. 2004 , , submitted ( astro - ph/0310306 ) de vries , w. h. , becker , r. h. , & white , r. l. 2003 , , 126 , 1217 dressler , a. 1980 , , 236 , 351 dressler , a. , & gunn , j. e. 1992 , , 78 , 1 evrard , a. e. , et al .
2002 , , 573 , 7 falco , e. e. , et al . 1999 , , 523 , 617 flores , r. a. , & primack , j. r. 1996 , , 457 , l5 fukugita , m. , futamase , t. , & kasai , m. 1990 , , 246 , 24p fukugita , m. , ichikawa , t. , gunn , j. e. , doi , m. , shimasaku , k. , & schneider , d. p. 1996
, , 111 , 1748 fukugita , m. , shimasaku , k. , & ichikawa , t. 1995 , , 107 , 945 fukushige , t. , & makino , j. 1997 , , 477 , l9 fukushige , t. , & makino , j. 2001 , , 557 , 533 fukushige , t. , & makino , j. 2003 , , 588 , 674 fukushige , t. , kawai , a. , & makino , j. 2004 , , submitted ( astro - ph/0306203 ) garrett , m. a. , calder , r. j. , porcas , r. w. , king , l. j. , walsh , d. , & wilkinson , p. n. 1994 , , 270 , 457 gavazzi , r. , fort , b. , mellier , y. , pello , r. , & dantel - fort , m. 2003 , , 403 , 11 ghigna , s. , moore , b. , governato , f. , lake , g. , quinn , t. , & stadel , j. 2000 , , 544 , 616 gladders , m. d. , & yee , h. k. c. 2000 , , 120 , 2148 golse , g. & kneib , j .-
2002 , , 390 , 821 gorenstein , m. v. , cohen , n. l. , shapiro , i. i. , rogers , a. e. e. , bonometti , r. j. , falco , e. e. , bartel , n. , & marcaide , j. 1988 , , 334 , 42 goto , t. et al .
2002 , , 123 , 1807 green , p. j. , et al .
2002 , , 571 , 721 gunn , j. e. , et al .
1998 , , 116 , 3040 hayashi , e. , et al .
2004 , , submitted ( astro - ph/0310576 ) hinshaw , g. , & krauss , l. m. 1987 , , 320 , 468 hogg , d. w. , finkbeiner , d. p. , schlegel , d. j. , & gunn , j. e. 2001 , , 122 , 2129 huterer , d. , & ma , c .- p .
2004 , , in press ( astro - ph/0307301 ) inada , n. , et al . 2003a , , 126 , 666 inada , n. , et al . 2003b , , 426 , 810 jenkins , a. , frenk , c. s. , white , s. d. m. , colberg , j. m. , cole , s. , evrard , a. e. , couchman , h. m. p. , & yoshida , n. 2001 , , 321 , 372 jing , y. p. 2000 , , 535 , 30 jing , y. p. , & suto , y. 2000 , , 529 , l69 jing , y. p. , & suto , y. 2002 , , 574 , 538 kashikawa , n. , et al .
2002 , , 54 , 819 keeton , c. r. 1998 , ph .
d. thesis , harvard university keeton , c. r. 2001a , , 561 , 46 keeton , c. r. 2001b , preprint ( astro - ph/0102340 ) keeton , c. r. 2002 , , 575 , l1 keeton , c. r. , kochanek , c. s. , & falco , e. e. 1998 , , 509 , 561 keeton , c. r. , kochanek , c. s. , & seljak , u. 1997 , , 482 , 604 keeton , c. r. , & madau , p. 2001
, , 549 , l25 keeton , c. r. , & winn , j. n. 2003 , , 590 , 39 kelson , d. d. , zabludoff , a. i. , williams , k. a. , trager , s. c. , mulchaey , j. s. , & bolte , m. 2002 , , 576 , 720 kitayama , t. , & suto . y. 1996 , , 469 , 480 klypin , a. , kravtsov , a. v. , bullock , j. s. , & primack , j. r. 2001 , , 554 , 903 kochanek , c. s. 1995 , , 453 , 545 kochanek , c. s. 1996 , , 466 , 638 kochanek , c. s. 2002 , in proc .
workshop , the shapes of galaxies and their halos , ed .
p. natarjan ( singapore : world scientific ) , p. 62
( astro - ph/0106495 ) kochanek , c. s. , falco , e. e. , & mu~ noz , j. a.1999 , , 510 , 590 kochanek , c. s. , falco , e. e. , & schild , r. 1995 , , 452 , 109 kochanek , c. s. , & white , m. 2001 , , 559 , 531 komatsu , e. , & seljak , u. 2002 , , 336 , 1256 koopmans , l. v. e. , treu , t. , fassnacht , c. d. , blandford , r. d. , & surpi , g. 2003 , , 599 , 70 kuhlen , m. , keeton , c. r. , & madau , p. 2004
, , in press ( astro - ph/0310013 ) li , l. x. , & ostriker , j. p. 2002
, , 566 , 652 li , l. x. , & ostriker , j. p. 2003
, , 595 , 603 lopes , a. m. , & miller , l. 2004 , , in press ( astro - ph/0306174 ) lupton , r. , gunn , j. e. , ivezi , z. , knapp , g. r. , kent , s. , & yasuda , n. 2001 , in asp conf .
238 , astronomical data analysis software and systems x , ed
. f. r. harnden , jr .
, f. a. primini , and h. e. payne ( san francisco : astr .
soc . pac . ) , p. 269
( astro - ph/0101420 ) ma , c .- p .
2003 , , 584 , l1 madau , p. , haardt , f. , & rees , m. j. 1999 , , 514 , 648 maoz , d. , rix , h .- w . , gal - yam , a. , & gould , a. 1997 , , 486 , 75 marlow , d. r. , rusin , d. , jackson , n. , wilkinson , p. n. , & browne , i. w. a. 2000 , , 119 , 2629 meneghetti , m. , yoshida , n. , bartelmann , m. , moscardini , l. , springel , v. , tormen , g. , & white s. d. m. 2001 , , 325 , 435 meneghetti , m. , bartelmann , m. , & moscardini , l. 2003a , , 340 , 105 meneghetti , m. , bartelmann , m. , & moscardini , l. 2003b , , 346 , 67 miller , l. , lopes , a. m. , smith , r. j. , croom , s. m. , boyle , b. j. , shanks , t. , & outram , p. 2004
, , in press ( astro - ph/0210644 ) miralda - escud , j. 2002 , , 564 , 60 miyazaki , s. , et al .
2002 , , 54 , 833 moore , b. , quinn , t. , governato , f. , stadel , j. , & lake , g. 1999 , , 310 , 1147 mortlock , d. j. , & webster , r. l. 2000 , , 319 , 872 mortlock , d. j. , webster , r. l. , & francis , p. j. 1999 , , 309 , 836 mu~ noz , j. a. , et al .
2001 , , 546 , 769 myers , s. t. , et al .
1995 , , 447 , l5 nakamura , t. t. , & suto , y. 1997 , prog .
phys . , 97 , 49 narayan , r. , & white , s. d. m. 1988 , , 231 , 97p navarro , j. f. , frenk , c. s. , & white , s. d. m. 1996 , , 462 , 563 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , , 490 , 493 ofek , e. o. , maoz , d. , prada , f. , kolatt , t. , & rix , h .- w .
2001 , , 324 , 463 ofek , e. o. , rix , h .- w . , maoz , d. , & prada , f. 2002 , , 337 , 1163 oguri , m. 2002 , , 580 , 2 oguri , m. 2003 , , 339 , l23 oguri , m. , lee , j. , & suto , y. 2003 , , 599 , 7 oguri , m. , taruya , a. , suto , y. , & turner , e. l. 2002 , , 568 , 488 oke , j. b. , et al .
1995 , , 107 , 375 patnaik , a. r. , browne , i. w. a. , wilkinson , p. n. , & wrobel , j. m. 1992 , , 254 , 655 phillips , p. m. , browne , i. w. a. , & wilkinson , p. n. 2001a , , 321 , 187 phillips , p. m. , et al . 2001b , , 328 , 1001 pier , j. r. , munn , j. a. , hindsley , r. b. , hennessy , g. s. , kent , s. m. , lupton , r. h. , & ivezi ' c , .
2003 , , 125 , 1559 pindor , b. , turner , e. l. , lupton , r. h. , & brinkmann , j. 2003 , , 125 , 2325 porciani , c. , & madau , p. 2000 , , 532 , 679 postman , m. , & lauer , t. r. 1995 , , 440 , 28 power , c. , navarro , j. f. , jenkins , a. , frenk , c. s. , white , s. d. m. , springel , v. , stadel , j. , & quinn , t.2003 , , 338 , 14 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992
, numerical recipes in fortran ( cambridge : cambridge univ . press ) rees , m. j. , & ostriker , j. p. 1977
, , 179 , 541 refsdal , s. 1964 , , 128 , 307 richards , g. t. , et al . 2002 , , 123 , 2945 richards , g. t. , et al .
2004 , , submitted ( astro - ph/0309274 ) rusin , d. 2002 , , 572 , 705 sand , d. j. , treu , t. , simth , g. p. , & ellis , r. s. 2004 , , in press ( astro - ph/0309465 ) sarbu , n. , rusin , d. , & ma , c .- p .
2001 , , 561 , l147 sasaki , s. , & takahara , f. 1993 , , 262 , 681 schmidt , m. , schneider , d. p. , & gunn , j. e. 1995 , , 110 , 68 schneider , p. , ehlers , j. , & falco , e. e. 1992 , gravitational lenses ( new york : springer ) sheth , r. k. , et al .
2003 , , 594 , 225 sheth , r. k. , & tormen , g. 1999 , , 308 , 119 smith , g. p. , kneib , j. , ebeling , h. , czoske , o. , & smail , i. 2001 , , 552 , 493 smith , j. a. , et al .
2002 , , 123 , 2121 spergel , d. n. , & steinhardt p. j. 2000 , , 84 , 3760 spergel , d. n. , et al .
2003 , , 148 , 175 stoughton , c. , et al .
2002 , , 123 , 485 sugiyama , n. 1995 , , 100 , 281 takahashi , r. , & chiba , t. 2001 , , 563 , 489 turner , e. l. 1980 , , 242 , l135 turner , e. l. 1990 , , 365 , l43 turner , e. l. , ostriker , j. p. , & gott , j. r. 1984 , , 284 , 1 vanden berk , d. e. , et al .
2001 , , 122 , 549 vanden berk , d. e. , et al .
2004 , , in press ( astro - ph/0310336 ) walsh , d. , carswell , r. f. , & weymann , r. j. 1979 , , 279 , 381 wambsganss , j. , cen , r. , ostriker , j. p. , & turner , e. l. 1995 , science , 268 , 274 wambsganss , j. , bode , p. , & ostriker , j. p. 2004
, , submitted ( astro - ph/0306088 ) wechsler , r. h. , bullock , j. s. , primack , j. r. , kravtsov , a. v. , & dekel , a. 2002 , , 568 , 52 white , m. 2002 , , 143 , 241 wyithe , j. s. b. , & loeb , a. 2002 , , 417 , 923 wyithe , j. s. b. , & loeb , a. 2002 , , 577 , 57 wyithe , j. s. b. , turner , e. l. , & spergel , d. n. 2001 , , 555 , 504 york , d. g. , et al .
2000 , , 120 , 1579 zhao , h. s. 1996 , , 278 , 488 zhdanov , v. i. , & surdej , j. 2001 , , 372 , 1 crrrrrr a & @xmath269 & @xmath270 & @xmath271 & @xmath272 & @xmath273 & @xmath274 + b & @xmath275 & @xmath276 & @xmath277 & @xmath278 & @xmath279 & @xmath280 + c & @xmath281 & @xmath282 & @xmath271 & @xmath283 & @xmath284 & @xmath285 + d & @xmath286 & @xmath287 & @xmath270 & @xmath288 & @xmath289 & @xmath290 + cccrrc a & 10 04 34.794 & @xmath29141 12 39.29 & @xmath292 & @xmath292 + b & 10 04 34.910 & @xmath29141 12 42.79 & @xmath293 & @xmath294 + c & 10 04 33.823 & @xmath29141 12 34.82 & @xmath295 & @xmath296 + d & 10 04 34.056 & @xmath29141 12 48.95 & @xmath297 & @xmath298 + g1 & 10 04 34.170 & @xmath29141 12 43.66 & @xmath299 & @xmath300 + crrcc a & @xmath302 & @xmath302 & @xmath303 & + b & @xmath304 & @xmath305 & @xmath306 & + c & @xmath307 & @xmath296 & @xmath308 & + d & @xmath309 & @xmath310 & @xmath311 & + g1 & @xmath312 & @xmath313 & & @xmath314 + ccccc fiducial model & @xmath315 & @xmath316 & @xmath317 & @xmath318 + @xmath319 & @xmath320 & @xmath321 & @xmath322 & @xmath323 + @xmath324 & @xmath325 & @xmath326 & @xmath327 & @xmath328 + @xmath329 & @xmath330 & @xmath331 & @xmath332 & @xmath333 + @xmath334 & @xmath335 & @xmath336 & @xmath337 & @xmath338 + lf1@xmath339lf2 & @xmath340 & @xmath341 & @xmath342 & @xmath343 + @xmath344 & @xmath345 & @xmath346 & @xmath347 & @xmath323 + @xmath348 & @xmath349 & @xmath350 & @xmath351 & @xmath352 + | we study the recently discovered gravitational lens sdss j1004 + 4112 , the first quasar lensed by a cluster of galaxies .
it consists of four images with a maximum separation of @xmath0 .
the system was selected from the photometric data of the sloan digital sky survey ( sdss ) , and has been confirmed as a lensed quasar at @xmath1 on the basis of deep imaging and spectroscopic follow - up observations .
we present color - magnitude relations for galaxies near the lens plus spectroscopy of three central cluster members , which unambiguously confirm that a cluster at @xmath2 is responsible for the large image separation .
we find a wide range of lens models consistent with the data , and despite considerable diversity they suggest four general conclusions : ( 1 ) the brightest cluster galaxy and the center of the cluster potential well appear to be offset by several kiloparsecs ; ( 2 ) the cluster mass distribution must be elongated in the north south direction , which is consistent with the observed distribution of cluster galaxies ; ( 3 ) the inference of a large tidal shear ( @xmath30.2 ) suggests significant substructure in the cluster ; and ( 4 ) enormous uncertainty in the predicted time delays between the images means that measuring the delays would greatly improve constraints on the models .
we also compute the probability of such large separation lensing in the sdss quasar sample , on the basis of the cold dark matter model .
the lack of large separation lenses in previous surveys and the discovery of one in sdss together imply a mass fluctuation normalization @xmath4 ( 95% confidence ) , if cluster dark matter halos have an inner density profile @xmath5 .
shallower profiles would require higher values of @xmath6 .
although the statistical conclusion might be somewhat dependent on the degree of the complexity of the lens potential , the discovery of sdss j1004 + 4112 is consistent with the predictions of the abundance of cluster - scale halos in the cold dark matter scenario . |
gauge symmetries and their systematic construction , while being of great significance , is fraught with subtleties . they must be off - shell , i.e. the action should remain invariant under them without use of any equations of motion and there are systematic algorithms for construction of such off - shell symmetries @xcite . in models of gravity obtained by gauging poincare symmetries , it is however seen that such systematic approaches yield the poincare symmetries only on - shell .
so , at the off - shell level , it appears that there are two independent set of symmetries .
this is unsettling .
we would then be required to find additional independent first - class constraints to support these symmetries . however , physically the only independent symmetries present are the space - time translations and lorentz boosts , i.e. the poincare symmetries .
we set up the model in the poincare gauge theory ( pgt ) formalism @xcite .
the global poincare symmetries on a set of global coordinates @xmath0 are localised at every spacetime point by introducing local frames having coordinates @xmath1 .
both sets are 3-dimensional for our model , though the construction can be carried out in any dimensions .
to change physical variables from one set of coordinates to another , triad fields are introduced , which are defined through @xmath2 here ` @xmath3 ' and ` @xmath4 ' are the metrics of the global and local sets of coordinates , respectively and ` @xmath4 ' is the standard minkowskian metric with signature @xmath5 .
now , to keep the invariance of the theory , a gauge connection field ` @xmath6 ' is introduced through a newly defined gauge covariant derivative . for an arbitrary vector @xmath7 , the derivative is written as @xmath8 the corresponding field strengths give rise to gravitational fields @xmath9 b^i_{\ \rho } = \epsilon^i_{\ jk } r^j_{\ \mu\nu } b^k_{\ \rho}\,,\end{aligned}\ ] ] where @xmath10 defines the riemann .
similarly , the commutator of @xmath11 derivatives give rise to the torsion @xmath12 it must be emphasised that this is a setting in the riemann - cartan spacetime rather than the riemann spacetime of usual general relativity .
in general it contains torsion , which goes to zero only after imposition of the connection variable s equation of motion , giving back the torsion free riemannian manifold of general relativity .
the poincare symmetries of the basic fields are @xcite @xmath13 where @xmath14 parametrizes local lorentz rotations and @xmath15 local coordinate translations .
they together constitute the @xmath16 independent poincare symmetries in 3d .
these symmetries are independent of the particular action being considered , so long as they are constructed in a proper tensorial manner .
now , symmetries of an action can be constructed via a hamiltonian analysis through completely off - shell procedures , as stated before .
after this is carried out , we find the symmetries which keeps the action invariant , as expected . however they can not be identified with the poincare symmetries , even after an appropriate field - dependent redefinition of the symmetry parameters .
the clue of identification , as we show , lies in ` trivial ' gauge symmetries @xcite .
they are symmetries of the form @xmath17 which keep any action _ off - shell _ invariant due to antisymmetry in the coefficients @xmath18 .
thus @xmath19 without depending on the form of the euler derivatives @xmath20 and without requiring the imposition of the equations of motion @xmath21 in what follows , we take up the mielke - baekler model , summarise its hamiltonian constraints , outline construction of an off - shell gauge generator and generate gauge symmetries .
these are then compared with the poincare symmetries to see how the two may actually be equivalent .
the mielke - baekler 3d topological gravity model with torsion @xcite along with a cosmological term is described by @xmath22\end{aligned}\ ] ] whose equations of motion are found by setting the euler derivatives to zero @xmath23 = 0 \\ \frac{\delta s}{\delta \omega^i_{\ \mu } } & = \epsilon^{\mu\nu\rho } \left [ \alpha_3\ , r_{i\nu\rho } + a\ , t_{i\nu\rho } + \alpha_4\ , \epsilon_{ijk}b^j_{\ \nu}b^k_{\ \rho } \right ] = 0 . \end{aligned}\end{aligned}\ ] ] the constraints in the above model are @xcite listed in table [ tab : constraints mb ] l c c & first class @xmath24 & second class @xmath25 + primary & @xmath26 & @xmath27 , @xmath28 + secondary & @xmath29 & + with the definitions @xmath30 - \nabla_{\!\alpha } \phi_i^{\ \alpha } + \epsilon_{ijk}\ , b^j_{\ \alpha } \left ( p\,\phi^{k\alpha } + q\ , \phi^{k\alpha } \right ) \\
\bar{\mathcal{k}}_i & = -\left [ \epsilon^{0\alpha\beta } \left ( a\,t_{i\alpha\beta } + \alpha_3\,r_{i\alpha\beta } + \alpha_4\,\epsilon_{ijk } b^j_{\ \alpha } b^k_{\ \beta } \right ) \right ] - \nabla_{\!\alpha } \phi_i^{\ \alpha } - \epsilon_{ijk}\ , b^j_{\ \alpha } \phi^{k\alpha}\\ p & = \frac{\alpha_3\lambda + \alpha_4a}{\alpha_3 \alpha_4 - a^2}\,;\qquad q = -\frac{\alpha_4 ^ 2 + a\lambda}{\alpha_3\alpha_4 - a^2}. \end{aligned}\end{aligned}\ ] ] the gauge generator @xmath31 which generates gauge symmetries is , by the dirac principle , a linear combination of all first class constraints @xmath32 the sum in @xmath33 running over the complete first - class sector .
the symmetries of the basic variables are then obtained through @xmath34 where @xmath35 indicates dirac brackets obtained through elimination of the second class sector @xmath25 @xcite .
now , we know that all the gauge parameters @xmath36 are not independent , as the number of independent gauge parameters must equal the number of independent primary first - class constraints . in the off - shell method that we adopt @xcite , the restriction on @xmath36 is obtained from the commutativity of gauge variations and time evolution @xmath37 explicit calculations @xcite yield the generator @xmath38\right . \\
\left.+ \dot{\sigma}^i\pi_i^{\ 0 } + \sigma^i\left[\bar{\mathcal{k}_i}-\varepsilon_{ijk}\big(b^j_{\ 0}\,\pi^{k0 } + \omega^j_{\ 0}\,\pi^{k0}\big)\right]\right ) \end{aligned}\end{aligned}\ ] ] where @xmath39 and @xmath40 are six independent gauge parameters .
we are now ready to study the gauge symmetries of the model utilising this generator .
the generator yields the following symmetries of the basic fields @xmath41 a first comparison of the two sets of symmetries and reveals that the hamiltonian symmetries are explicitly dependant on the coupling parameters of the various terms in the action , in contrast to the pgt symmetries . also , we note that the gauge parameters of the two symmetries are different .
so we need to map the gauge parameters of into the pgt gauge parameters .
this is achieved through the field dependent map @xcite @xmath42 using this map , the hamiltonian symmetries can be written in the form @xmath43 it may seem that the two symmetries are thus equivalent only on - shell . to show that this is not the case , we write out their form explicitly @xmath44 to cast the balance terms of @xmath45 into the form , we first write down the appropriate @xmath46 matrix : @xmath47 the anti - symmetry of the above structure can now be easily demonstrated .
as an example , @xmath48 so , we see that indeed ` @xmath49 ' is of the required form for trivial gauge symmetries .
thus the invariance of the action can now be achieved under these ` balance ' symmetries @xmath50 without the use of any equations of motion .
hence we see , that from a canonical point of view , @xmath51 as their difference is just an un - physical ` trivial ' symmetry .
99 henneaux m , teitelboim c and zanelli j 1990 nucl .
b * 332 * , 169 - 188 .
banerjee r , rothe h j and rothe k d 1999 phys .
b * 463 * 248 .
banerjee r , rothe h j and rothe k d 2000 phys .
b * 479 * , 429 .
utiyama r 1956 phys .
* 101 * 1597 .
kibble t w b 1961 j. math .
* 2 * 212 .
sciama d w 1962 _ recent developments in general relativity , festschrift for leopold infeld _
( new york : pergamon press ) p 415 .
banerjee r , gangopadhyay s , mukherjee p and roy d 2010 _ j. high energy phys .
_ jhep * 1002 * , 075 ( 2010 ) henneaux m and teitelboim c 1992 _ quantization of gauge systems _ ( princeton , usa : univ .
henneaux m , kleinschmidt a and gomez g l 2010 phys .
d * 81 * , 064002 .
castellani l 1982 annals phys . * 143 * , 357 .
baekler p , mielke e w and hehl f w 1992 nuovo cim .
b * 107 * 91 .
blagojevic m and cvetkovic b 2006 _ trends in gr and qc _ * 2 * ed ch benton ( new york : nova science ) 103 .
banerjee r and roy d 2011 phys .
d * 84 * , 124034 . | we study the gauge symmetries in a mielke - baekler type model of gravity in 2 + 1 dimensions .
the model is built in a poincare gauge theory framework where localisation of poincare symmetries lead to gravity .
however , explicit construction of gauge symmetries in the model through a hamiltonian procedure yields an apparently different set of symmetries , as has been noted by various authors . here ,
we show that the two sets of symmetries are actually equivalent in a canonical sense , their difference being just a set of trivial symmetries . |
rotation - powered pulsars are known to produce relativistic magnetized winds responsible for a substantial fraction of the pulsar s spin - down energy loss .
high angular resolution observations with the _ chandra x - ray observatory _ ( _ cxo _ ) have shown that the winds are intrinsically anisotropic ( both polar and equatorial outflow components are often seen , with the latter dominating in most cases ; see kargaltsev & pavlov 2008 ) .
the outflow morphology is also influenced by the interaction with the ambient medium , which can introduce further anisotropy .
the motion effects become particularly pronounced when the pulsar speed exceeds the sound speed in the ambient medium leading to the formation of a bow shock . in this case , the outflow from the pulsar is expected to acquire a cometary shape with an extended tail behind the pulsar .
some examples of such outflows are the pulsar wind nebulae ( pwne ) associated with psrs b0355 + 54 ( mcgowan et al .
2006 ) , j03573205 ( de luca et al .
2010 ) , and j17412054 ( auchettl et al . 2015 ) .
moreover , there has been growing evidence that pwne around high - speed pulsars ( such as psr b2224 + 65 , igr j110146103 ) exhibit puzzling elongated features in addition to cometary tails ( hui & becker 2007 ; pavan et al .
2014 , 2015 ) . to understand the origin of the diverse morphologies of pulsar outflows and their interaction with ambient medium , deep x - ray observations with high spatial resolution are particularly valuable .
psr j15095850 ( j1509 hereafter ) is a relatively young pulsar with spin - down age @xmath6 kyr and energy loss rate @xmath7 erg s@xmath2 ( other parameters are listed in table [ tbl - parameters ] ) .
an x - ray nebula associated with j1509 was discovered in a 40 ks _
_ observation ( kargaltsev et al . 2008 ; k+08 hereafter ) .
it was found to consist of a compact nebula ( cn ) in the pulsar vicinity and a 6 pc long elongated structure interpreted as a tail behind the pulsar moving in the northeast direction .
the pwn s non - thermal spectra were fit with an absorbed ( @xmath8 @xmath9 ) power - law ( pl ) model with photon indices @xmath10 and @xmath11 ( here and everywhere below the uncertainties are given at 68% confidence unless specified otherwise ) .
the luminosity of the tail was found to be @xmath12 erg s@xmath13 in the 0.58 kev band ( at the assumed distance of 3.8 kpc ; see table 1 ) .
observations with the australia telescope compact array ( atca ) detected the tail in the radio and revealed a high degree of linear polarization whose orientation suggested a helical magnetic field ( ng et al .
additionally , a faint h@xmath14 halo around j1509 and a bow shock cavity were detected ( brownsberger & romani 2014 ) .
the x - ray spectrum extracted from an @xmath15 aperture around the pulsar position was fitted with an absorbed pl model with a best - fit @xmath16 . since no x - ray pulsations could be detected ( because the 3.2 s frame time was much longer than the pulsar period )
, the unresolved point - like emission could include a compact pwn contribution .
the pulsed emission was detected in @xmath17-rays by the _ fermi _
large area telescope ( lat ) up to energies of 3 gev , with a pl spectrum ( @xmath18 ) and high @xmath17-ray efficiency , @xmath19 , higher than those of 75% of @xmath17-ray pulsars ( see weltevrede et al . 2010 and abdo et al .
2013 ) .
in this paper we report on the results of a series of four deep observations of j1509 and its pwn . in section 2
we describe the observations and data reduction . in section 3
we present the images and spectral fit results . in section 4
we discuss our findings , and in section 5 we present our conclusions .
( j2000.0 ) & 15 09 27.13 + decl .
( j2000.0 ) & 58 50 56.1 + epoch of position ( mjd ) & 51463 + galactic longitude ( deg ) & 319.97 + galactic latitude ( deg ) & 0.62 + spin period , @xmath20 ( ms ) & 88.9 + period derivative , @xmath21 ( 10@xmath22 ) & 0.92 + dispersion measure , dm ( pc @xmath23 ) & 137.7 + distance than observed , and , based on the j1509 s galactic coordinates , would place the pulsar in between the sagittarius and scutum - centaurus arms , whereas the distance of 3.8 kpc places j1509 in the middle of the scutum - centaurus arm . ] , @xmath24 ( kpc ) & 3.8 + distance from the galactic plane , @xmath25 ( kpc ) & 0.04 + surface magnetic field , @xmath26 ( 10@xmath27 g ) & 0.91 + spin - down power , @xmath28 ( 10@xmath29 erg s@xmath2 ) & 5.1 + spin - down age , @xmath30 ( kyr ) & 154 +
we conducted four @xmath3190 ks observations of j1509 over the course of 15 months , from 2013 june 14 ( mjd 56457 ) to 2014 september 23 ( mjd 56923 ) , totaling 374 ks ( see table [ tbl - obs ] ) .
the target was placed on the acis - i detector , operated in very faint mode with a 3.2 s time resolution .
we used the interactive analysis of observations ( ciao ) software package ( ver . 4.6 ) and the _ cxo _ calibration data base ( caldb , ver .
4.6.3 ) for data processing , and restricted the energy range to 0.58 kev for images and spectral analysis .
we created exposure maps for each observation and an exposure - corrected image from the merged observation data using the ciao routine merge_obs .
ciao s wavdetect tool ( a mexican - hat wavelet source detection routine ; freeman et al .
2002 ) was used to detect field point sources , which were used for image alignment and then subtracted for diffuse emission analysis .
we used the positions of 12 nearby stationary point sources in conjunction with the ciao tool wcs_match to correct slight discrepancies in the astrometry of the event files .
spectra were extracted using ciao s specextract tool and analyzed using xspec ( ver .
12.8.2 ) . in all spectral
fits we used the xspec wabs model with absorption cross sections from morrison & mccammon ( 1983 ) . to study the morphology of the emission in the immediate vicinity of the pulsar
, we simulate the point spread function ( psf ) using chart ( ray trace ) ver . 2 and marx ver . 5.1 .
we also analyzed two _ xmm - newton _ observations ( obsids 0500630101 and 0500630301 ) carried out on 2008 january 29 and 2008 march 1 .
the epic ( european photon imaging camera ) observations of psr j15095850 employed the pn , mos1 , and mos2 cameras in full frame imaging mode with the medium filter .
we used the _ xmm - newton _ science analysis software ( sas , ver .
14.0 ) for the reprocessing of the observations and data reduction . after removal of background flares ,
the effective exposure times are 45ks ( pn ) , 73ks ( mos1 ) , and 73ks ( mos2 ) . because of the high background
, we only use _ xmm - newton _ data to analyze the pulsar spectrum ( which , however , is more contaminated by the cn contribution than the _ chandra _ spectrum due to the lower spatial resolution of _ xmm - newton _ ) . to minimize the contamination from the cn and the background , we use circular extraction regions with radii of @xmath32 ( pn ) and @xmath33 ( mos1&2 ) centered on the pulsar ( brightness profile peak ) .
these extraction regions correspond to @xmath34% of the encircled energy ( figures 6 and 7 in section 3.2.1.1 of the _ xmm - newton _ user handbook ) .
we use an @xmath35 circular background region close to the pulsar ( on the same chip ) but away from the cn .
the resulting net source counts fraction is @xmath36% for all instruments .
cccccc 14523 & 2013 jun 14 & 93.87 & 26 & 2 & 303@xmath37 + 14524 & 2013 sep 15 & 93.87 & 17 & 3 & 237@xmath37 + 14525 & 2014 may 9 & 92.01 & 17 & 3 & 2@xmath37 + 14526 & 2014 sep 23 & 93.86 & 17 & 3 & 232@xmath37 +
the merged image produced from four acis - i observations ( see figure [ image - raw ] ) reveals a bright cn around the pulsar ( 514 net source counts ) and two fainter large - scale elongated structures .
in addition to the extended tail ( hereafter , southern outflow ; 3530 net counts ) previously reported by k+08 , the much deeper image shows , for the first time , another fainter structure ( hereafter , the northern outflow ; 1,760 net counts ) apparently originating from the pulsar / cn area and visible up to 7@xmath38 ( or 7.7 pc at @xmath39 kpc ) north of the pulsar . below we consider the individual pwn elements in detail .
the inset in figure [ image - raw ] shows the cn structure .
the peak of the x - ray brightness distribution ( r.a.=@xmath40 , decl.=@xmath41 ) coincides ( within @xmath42 ) with the radio position of the pulsar ( r.a.=@xmath43 , decl.=@xmath44 at mjd 51463 , i.e. , about 14 years before our x - ray observations ) .
the centroiding uncertainty of the x - ray position , @xmath45 ( calculated by celldetect ) , is much smaller than the typical uncertainty of the _ chandra _ absolute astrometry ( @xmath46 at 68% confidence ) .
this puts an upper limit of @xmath47 on the pulsar position shift during 14 years , corresponding to an upper limit @xmath48 mas yr@xmath2 on proper motion .
the psf modeling and subtraction does not reveal any additional structure within the core region .
we also attempted psf deconvolution using the ciao tool arestore , but it did not produce anything new .
the rest of the cn structure bears much resemblance to the geminga pwn .
therefore , following pavlov et al .
( 2006 ) , we will call the three distinct features of the cn the lateral tails and the axial tail .
the lateral tails are noticeably longer ( @xmath317@xmath49 , or [email protected] pc at @xmath39 kpc ) than the axial tail ( @xmath313@xmath51 , or 0.05 pc at @xmath39 kpc ) .
a wind from a fast - moving pulsar is expected to produce a termination shock with a bow - like ( parabolic ) structure ahead of the pulsar .
however , we see no extended emission ahead of the point - like source at the leading tip of the cn ( see inset in figure 1 ) . to constrain the stand - off distance @xmath52 of the possible unresolved shock apex ahead of the pulsar
, we assume that the shape of the lateral tails represent the shock surface , fit the shape with a parabola , and estimate @xmath52 as the distance between the vertex of the fitted parabola and the position of the brightness peak .
this approach gave us only upper limits on the stand - off distance , @xmath5310 , depending on the region chosen for fitting .
these fits , however , allowed us to estimate the radius of curvature at the parabola s vertex , @xmath54@xmath55 .
examination of the cn images from individual observations suggests variability associated with the lateral tails .
this variability might be caused by the motion of blobs moving along the tails ( see figure [ image - blobs - jets]kargaltsev/ + j1509movie.mov ] ) ] ) .
to quantify the changes in the positions of the blobs ( shown with green and yellow ellipses in figure [ image - blobs - jets ] ) , we measure their positions ( relative to the pulsar ) in the images from four separate observations and plot them as a function of time in figure [ image - blobs - plot ] .
the positions are measured by running ciao s tool celldetect values range between 0102 ) .
therefore , we conservatively take the actual uncertainties to be 05 for all blobs and use these in the fits . ] .
it is apparent from these plots that the green data points can be approximately connected by straight lines corresponding to steady motion .
if we fit a steady motion model to the green data points ( omitting the others from the fitting procedure ) , the resulting blob speeds are @xmath56 for the western lateral tail and @xmath57 for the eastern lateral tail
. this result should be taken with some caution because the significance of the individual blobs in the individual observations is low , @xmath58 , when using a local background ( i.e. within the lateral outflows ) as calculated by celldetect ) .
however , it increases to @xmath59@xmath60 when the external ( i.e. , outside the cn ) background region is used .
moreover , under the assumption of steady motion , the same blob is seen in multiple observations and hence the combined blob significances increase to 3@xmath61 for the local background choice , and to @xmath62@xmath63 for the external background choice .
we have also compared the merged image of the cn with the older observation ( obsid 3515 , acis - s , 2003 feb 9 ) .
that observation was significantly shorter ( 40 ks ) , and the acis - s3 background contribution was much higher than that of the acis - i , making it difficult to resolve the individual components of the cn ( see figure 4 in k+08 ) .
figure [ image - mid - scale ] shows the transition from the cn to the extended outflow south of the pulsar .
one can see the lateral tails fading and apparently merging at @xmath3127@xmath51 from the pulsar . beyond this distance
the flow expands in the transverse direction and gives rise to the fainter _ southern outflow _ seen in figure [ image - large - scale ] that presents the merged image of the large - scale diffuse emission with field point sources subtracted and exposure - map correction applied .
unexpectedly , the same image clearly reveals a similarly shaped emission north - northwest of the pulsar , although this _ northern outflow _ is not discernible in the pulsar vicinity ( figure [ image - mid - scale ] ) .
the shapes of both large - scale structures , extending up to @xmath64 from the pulsar , can be crudely described as two wedges with @xmath65 opening angles and tips at ( or near ) the pulsar position .
the southern outflow is approximately aligned with the cn symmetry axis , while the northern outflow is inclined to the cn symmetry axis by @xmath66
. although the sizes and shapes of these outflows are similar , the distributions of surface brightness within the two structures are markedly different .
the averaged ( over the transverse direction ) surface brightness of the southern outflow ( see figure [ image - radio ] for the profile extraction regions ) appears to decrease almost linearly with distance from the pulsar ( see figure [ figure - tail - brightness ] ) , while the averaged surface brightness of the northern outflow does not appear to change significantly over its length ( shown in figure [ figure - headlight - brightness ] ) .
moreover , a substantially brighter region ( hereafter referred to as the `` spine '' ) is clearly seen within the southern outflow up to @xmath67 from the pulsar , while no such a feature is seen in the northern outflow .
figure [ image - radio ] also shows the extended radio emission reported by ng et al .
( 2010 ) , which stretches along the southern outflow .
it becomes wider than the x - ray outflow beyond @xmath68 from the pulsar and then narrows down again beyond @xmath69 .
the radio outflow is seen up to @xmath70 from of the pulsar , whereas the visible southern x - ray outflow becomes comparable with the background at about @xmath3 from the pulsar .
no radio emission is seen from the northern outflow region in the same atca observation .
we measure the cn spectra in three regions shown in the figure [ image - raw ] inset : ( 1 ) the bright pulsar / head region ( within the @xmath71 radius white circle around the pulsar ) , which contains an unresolved cn `` head '' in addition to the pulsar , ( 2 ) the lateral tails ( cyan polygons ) , and ( 3 ) the axial tail ( the green polygon ) .
for the southern outflow , we define the following regions ( shown in figure [ image - large - scale ] ) : the entire southern outflow ( combination of green polygons ) , the spine ( magenta polygon ) , and the southern outflow excluding the spine .
we also divide the entire southern outflow region into 4 segments , with segment 1 being the closest to the pulsar and segment 4 being the farthest . for the northern outflow , we define two regions : the proximal and distal halves ( cyan polygons ) .
we fit all spectra with an absorbed pl model ( unless specified otherwise ) and list the fitting parameters table [ spatially - resolved - spectra ] . for all regions , the spectra from different observations
are fitted simultaneously , with the all parameters being tied together .
fitting the spine of the southern outflow yields an absorbing column density @xmath72 @xmath9 and a photon index @xmath73 , with reduced chi - squared @xmath74 for @xmath75 degrees of freedom ( d.o.f . ) .
since the spectrum from this region fits the pl model well , and has a relatively small background contribution and a good statistic , we fix @xmath76 at the best - fit value obtained for this region while fitting the spectra from other regions which have poorer statistics by fitting the spectra from the lateral tails and the pulsar / head region . ] . for the cn regions
, we obtained very similar pl slopes for the pulsar / head and the lateral tails ( @xmath77 and @xmath78 , respectively ) while the axial tail appears to have a harder spectrum ( @xmath79 ) .
it is plausible that a substantial fraction of the pulsar / head emission comes from the neutron star and hence it may have a thermal component .
therefore , we attempted to fit the pulsar / head spectrum with a pl + blackbody ( bb ) model ( with a fixed bb radius of 12 km ) .
this resulted in a slightly better fit , with @xmath80 , @xmath81 ev , and @xmath82 for 53 d.o.f .
however , the best - fit bb+pl model is preferred over the pl model at a 93% ( @xmath83 ) confidence , according to the f - test ( i.e. , the bb component is not required by the data ) .
we find an upper limit on the neutron star surface temperature by gradually increasing the temperature of the bb component ( and re - fitting the pl component at each new temperature chosen ) until the model becomes inconsistent with the data at a @xmath59 confidence level .
this yields an upper limit @xmath84 ev ( or @xmath85 k ) .
although the quality of the fits is good for all the cn regions ( see table [ spatially - resolved - spectra ] ) , the fits of the pulsar / head spectrum ( see figure [ image - psr - spectra ] ) shows some systematic residuals centered at 3.5 kev ( between 3.33.6 kev ) . to examine plausibility of the alleged spectral feature
, we jointly fit the _ xmm - newton _
spectra from two observations ( see figure [ image - xmm - psr - spectra ] ) extracted from circular regions ( @xmath86 and @xmath33 for the pn and mos detectors ; see section 2 ) centered on the pulsar with the absorbed pl model and @xmath76 fixed at the same value as for the acis spectral fits .
this gave a similar spectral slope , @xmath87 , but a hint of an emission feature at @xmath88 kev is seen in the pn spectrum from only one observation while the pn spectrum from the other observation shows a hint of absorption , which means that the feature is likely a statistical fluctuation .
the mos spectra do not have enough counts to look for spectral features .
the spectrum of the spine ( @xmath73 ) appears somewhat softer than that of the rest of the southern outflow ( @xmath89 ) . for the southern outflow segments 14 , we obtain @xmath90 , @xmath91 , @xmath92 , and @xmath93 .
we do not see any systematic softening of the spectrum along the southern emission despite its large extent ; rather there is a hint of hardening from segment 2 to 4 .
almost all fits are of good quality ( the poorest quality fit , @xmath94 for 46 d.o.f . , being for segment 1 , its spectrum is shown in figure [ image - tail - seg-1-spectra ] ) .
the slope of the spectrum for the entire northern outflow ( @xmath95 ) is similar to the spectrum of the southern outflow excluding the spine ( @xmath89 ) although , unlike the southern outflow , the northern outflow spectrum shows a hint of softening with increasing distance from the pulsar ( @xmath96 changes from @xmath97 to @xmath98 ) .
this may explain the somewhat poorer quality of the pl fit to the spectrum of the entire northern outflow ( see table [ spatially - resolved - spectra ] ) .
the photon indices for all regions as a function of distance from the pulsar are shown in figure [ image - gamma - fn - dist ] .
lrccccccc[h ] pulsar / head & 8.96 & @xmath99 & @xmath100 & @xmath101 & 0.82 ( 54 ) & @xmath102 & @xmath103 + lateral tails & 88.1 & @xmath104 & @xmath78 & @xmath105 & 1.01 ( 35 ) & @xmath106 & @xmath107 + axial tail & 23.7 & @xmath108 & @xmath109 & @xmath110 & 790 ( 2046 ) & @xmath111 & @xmath112 + southern outflow ( entire ) & 36778 & @xmath113 & @xmath114 & @xmath115 & 0.90 ( 105 ) & @xmath116 & @xmath117 + spine & 4799 & @xmath118 & @xmath119 & @xmath120 & 1.01 ( 73 ) & @xmath121 & @xmath122 + s. outflow ( excl .
spine ) & 31979 & @xmath123 & @xmath124 & @xmath125 & 1.09 ( 83 ) & @xmath126 & @xmath127 + s. outflow seg . 1 & 1805 & @xmath128 & @xmath129 & @xmath130 & 1.26 ( 46 ) & @xmath131 & @xmath132 + s. outflow seg . 2 & 2224 & @xmath133 & @xmath134 & @xmath135 & 1.06 ( 35 ) & @xmath136 & @xmath137 + s. outflow seg . 3 & 9463 & @xmath138 & @xmath139 & @xmath140 & 0.71 ( 42 ) & @xmath141 & @xmath142 + s. outflow seg . 4 & 23162 & @xmath143 & @xmath144 & @xmath145 & 0.95 ( 73 ) & @xmath146 & @xmath147 + n. outflow ( entire ) & 21714 & @xmath148 & @xmath149 & @xmath150 & 1.14 ( 60 ) & @xmath151 & @xmath152 + n. outflow ( proximal ) & 5960 & @xmath153 & @xmath97 & @xmath154 & 0.95 ( 47 ) & @xmath155 & @xmath156 + n. emission ( distal ) & 15754 & @xmath157 & @xmath158 & @xmath159 & 1.09 ( 67 ) & @xmath160 & @xmath161 +
in addition to the previously detected cn and southern outflow ( k+08 ) , the deep _ chandra _ acis - i observations have revealed the previously unresolved fine structure of the cn .
the overall shape of the cn and the southern outflow supports the earlier suggestion that they are due to the shocked wind of the pulsar moving in the north - northeast direction with a supersonic speed .
the deeper image has also revealed a previously undetected faint structure , the northern outflow , extending @xmath162 to the north from the pulsar position .
below we consider possible interpretations for the different components of the j1509 pwn and compare them to those seen in other pwne .
the structure of the j1509 cn is very similar to that of the geminga pwn ( pavlov et al . 2006 , 2010 ; see figure [ bowshocks ] ) , both in shape ( one axial and two lateral tails ) and in ( projected ) physical size ( @xmath163 cm for the lateral tails , and @xmath164 cm for the axial tail , for @xmath165 pc and @xmath166 kpc ) .
it is reasonable to expect that the physical sizes of compact nebulae with similar structures are proportional to @xmath167 , where @xmath168 is the ambient density , and @xmath169 is the pulsar velocity . assuming also similar angles @xmath170 between the line of sight and the pulsar motion direction , and taking into account that @xmath171 , we can crudely estimate the speed of the j1509 pulsar : @xmath172 however , it is likely that geminga lies within a local high temperature bubble , with @xmath173 @xmath23 ; in contrast , j1509 is in a high density , partly neutral medium ( namely the h@xmath14 halo ; see brownsberger & romani 2014 ) with @xmath174 @xmath23 .
an upper limit on the pulsar velocity follows from the lack of any significant shift of the pulsar position over @xmath3114 years , @xmath175 ( see section 3.1.1 ) , which gives @xmath176 km s@xmath2 . a possible lower limit on the pulsar velocity can be estimated from the lack of resolved extended emission ahead of the pulsar . if there is a ( quasi-)isotropic component of the pulsar wind , it should produce a bullet - like termination shock with the apex ( bullet head ) distance @xmath177 . for @xmath178 , the upper limit on the angular apex distance of @xmath179
( see section 3.1.1 ) implies @xmath180 cm , @xmath181 dyn @xmath9 , and @xmath182 km s@xmath2 , where @xmath183 is the ism number density . despite the morphological similarities , there are also notable differences between the two pwne .
the spectra of the lateral tails are much softer in the j1509 cn than in the geminga pwn ( @xmath184 vs. @xmath185@xmath186 , respectively ) , while the lateral tails luminosity , @xmath187 erg s@xmath2 for j1509 , is a factor of @xmath188 higher .
the higher luminosity could be attributed to the larger @xmath28 , but the radiative x - ray efficiency for j1509 s lateral tails is also significantly higher : @xmath189 for j1509 vs. @xmath190 for geminga .
another difference is that there is the presence of faint diffuse emission between the lateral tails ( beyond the extent of the shorter axial tail ) in the j1509 cn ( see the inset in figure [ image - raw ] ) , while such emission is not detected in the geminga pwn ( posselt et al .
, in prep . ;
pavlov et al . 2010 ) .
numerical simulations of the synchrotron brightness distributions from pwne of supersonically moving pulsars with isotropic wind ejection ( bucciantini et al .
2005 ) predict filled structures with the brightness reaching its maximum ahead of the moving pulsar and gradually decreasing behind the pulsar ( see figures 4 and 5 in bucciantini et al .
these simulated synchrotron brightness maps do not resemble the observed structures of the cne around j1509 and geminga .
one reason could be that the magnetic field structure is substantially different from that assumed in the simulations .
bucciantini et al .
( 2005 ) mention that the magnetic field may be compressed in narrow layers near the contact discontinuity surface , which , in principle , could produce limb - brightened structures , depending on the amount of doppler boosting ( caused by a high bulk flow velocity ) and the magnetic field geometry .
another reason could be that the assumption of isotropic wind injection used in the simulations is not valid for real pulsars .
for instance , one could interpret the lateral tails as jets or an equatorial outflow bent back by the ram pressure .
however , at least the jet interpretation looks unrealistic as it requires stretching some of the parameters to their extremities . indeed , a jet produced by a moving pulsar bends because the jet flow acquires a momentum from the ambient medium ; it becomes comparable to the original jet flow momentum over the bending length scale ( curvature radius ) , it is likely that the jets are inside the contact discontinuity , in which case the deflection is a result of the post - shock sweep back , and not from ballistic impact from the unperturbed ism .
* ] @xmath191 , where @xmath192 is the fraction of the spin - down power @xmath28 that goes into the jet , @xmath193 is the jet radius , and @xmath14 is the angle between the initial jet direction and the pulsar s velocity . for j1509 , we obtain @xmath194 cm .
therefore , for a jet with the angular width @xmath195 , a significant bending on a scale of the observed curvature radius @xmath54@xmath55 ( see section 3.1.1 ) implies that most of the spin - down power goes into the two jets .
however , observations of pwne with both jet and equatorial components suggest @xmath196 ( e.g. , kargaltsev & pavlov 2008 ; pavlov et al . 2003 ) .
therefore , the jet interpretation of the lateral tails would imply that in some pulsars jets can be more powerful than equatorial outflows .
more extensive modeling and simulations are needed to see whether it is possible to reproduce the observed cn morphologies for other assumptions on the angular distribution of the pulsar wind injection .
we also note that the morphologies of the geminga and j1509 cne are markedly different from those of other bow - shock pwne .
in particular , others exhibit a cn with a local minimum in brightness in the pulsar vicinity ( i.e. , behind the pulsar ; pwne of psrs j0357 + 3205 and j11016101 ) , or exhibit a filled cn morphology ( pwne of psrs j17472958 , j17412054 , and b0355 + 54 ) . the differences , however , are not limited to x - ray morphologies only .
for instance , the radio counterparts of the pwne associated with psrs j15095850 ( ng et al . 2010 ) , j11016101 ( pavan et al .
2014 ) , and j17472958 ( the mouse ; yusef - zadeh & gaensler 2005 ) , the radio brightness distributions are also quite different . while the radio brightness in the mouse is maximal near the pulsar ( similar to x - rays ; gaensler et al .
2004 ) , in the other two pwne the radio brightness is low near the pulsar and becomes higher further away ( see figure [ image - radio ] ) .
we also note that the x - ray morphologies of most of the other cne in figure [ bowshocks ] do not bear much resemblance to the simulated images , except perhaps for the mouse and j17412054 .
ng et al .
( 2010 ) mentioned an open cluster , ruprecht 112 ( c1453623 ; alter et al .
1970 ) , about 4@xmath197 southwest of the current pulsar position , as a possible birth place of the pulsar . at the distance of 3.8 kpc and
the pulsar s spin - down age of 150 kyr , the required pulsar velocity would be @xmath198 km s@xmath2 , much higher than the upper limit of 640 km s@xmath2 on the transverse velocity inferred from the lack of detectable change of the pulsar position in 14 years . based on the comparison of the older x - ray image with the shape of the termination shock `` bullet '' seen in numerical simulations , k+08 suggested that the pulsar velocity should be @xmath199 km s@xmath2 , which is still within the allowed velocity range ( see above ) .
however , since the deeper x - ray images reveal substantial differences between the cn structure and the simulations , this velocity estimate should be taken with caution .
the most puzzling new feature of the j1509 pwn is the large - scale northern outflow seen up to @xmath200 almost ahead of the pulsar .
one could assume that the northern outflow is a powerful relativistic jet piercing the bow shock near the pulsar .
the lack of a counter - jet could be explained by doppler boosting , but there are other problems with the jet interpretation . according to the estimates in section 4.1.1 , the bending length for a jet is very small , even at @xmath201 and @xmath202 , compared to the length of the northern outflow that shows no bending .
it is also difficult to explain the relative faintness of the northern outflow closer to the pulsar ( but outside the cn ) in the jet scenario .
finally , the x - ray radiation efficiency of the northern outflow , @xmath203 , substantially exceeds the efficiencies of jets of younger pwne with clear torus - jet morphology .
therefore , we discard the jet interpretation for the northern outflow .
similar arguments have led bandiera ( 2008 ) to conclude that the extended x - ray feature accompanying the guitar nebula ( which also exhibits a large - scale outflow misaligned with the pulsar motion direction ; see figure [ image - guitar ] and johnson & wang 2010 ) can not be interpreted as a hydrodynamical jet .
following bandiera ( 2008 ) , we consider a scenario in which particles with the highest energies leak from the termination shock apex into the ism , and travel along the ordered ism magnetic field .
bandiera ( 2008 ) suggested that the length of the guitar pwn feature , @xmath204 pc , is on the order of the radiation length ( i.e. , the distance over which particles lose a substantial fraction of their energy to synchrotron radiation ) .
however , this assumption implies a propagation speed of only a few thousands km s@xmath2 , instead of the speed of light , as expected for particles with large ( bulk ) lorentz factors .
bandiera ( 2008 ) argued that the turbulent component of the magnetic field would provide a scattering mechanism and slow down particle propagation ( which can then be considered as diffusion ) along the ism magnetic field to the required non - relativistic speed .
the observed hint of softening of the x - ray spectrum in j1509 s northern outflow can be considered as supporting this scenario .
a problem with this scenario is that in all three cases with prominent misaligned ( with respect to pulsar motion ) x - ray features ( the guitar , j1509 , and j11016101 ) , the features appear to be one - sided or show very faint `` counter - features '' , which means that there is a preferred direction for the escape of the high - energy particles .
the cause of this asymmetry is currently unclear . on the other hand ,
since the particles escaping from the termination shock apex into the ism must be the most energetic ones ( those with large gyration radii ; see bandiera 2008 ) , the lack of radio synchrotron emission can be naturally explained by the deficit of lower energy particles .
the southern outflow is aligned with the cn symmetry axis and clearly represents a continuation of the cn ( see figure [ image - mid - scale ] ) .
therefore , we can assume that at least the brighter part of the southern outflow can be interpreted as a tail filled with the shocked pulsar wind confined by the ram pressure .
the lack of evidence for synchrotron burn - off in the x - ray spectra suggests a high flow speed and/or a low magnetic field .
alternatively , there could be some `` reheating '' due to in - situ conversion of magnetic field energy into particle energy , e.g. , via turbulent processes and accompanying reconnection , which might explain the hint of spectral hardening at large distances from the pulsar ( see figure [ image - gamma - fn - dist ] . ) indeed , ng et al .
( 2010 ) reported the presence of substantial turbulent magnetic field components in the tail using radio polarimetry data .
there are , however , some puzzling aspects associated with the southern outflow .
one is that the radio emission brightens with distance from the pulsar and becomes broader than the x - ray emission beyond @xmath68 ( see figure [ image - radio ] ) .
such behavior could possibly be caused by entrainment of ambient matter ( see also morlino et al .
2015 ) , which decreases the flow speed , increases the transverse size of the tail , and leads to adiabatic cooling of the outflowing matter .
this tentative explanation remains to be checked with detailed modeling of entrainment effects .
another puzzling aspect is the similarity of the wedge - like shapes of the fainter part of the southern outflow ( i.e. , the southern outflow excluding the spine ) and the northern outflow .
given the qualitative differences in the above interpretations of the northern and southern outflows , these similarities could be attributed to merely a chance coincidence . alternatively , the structural and spectral similarities could be explained if both the northern emission and the southern emission ( excluding the spine ) are caused by very energetic particles that leaked from the termination shock apex in opposite directions and diffused along the curved ism magnetic field ( the curvature of the magnetic field is required to explain the different orientations of the `` wedges '' ) . in this scenario , the spine ( whose spectrum is slightly softer than that of the rest of the southern outflow ) is still interpreted as the pulsar tail , coincidentally projected onto the sky within the southern outflow .
this interpretation may look somewhat artificial , but it can not be firmly excluded with the currently available data .
the spectra of the large - scale outflows measured in several regions allow one to crudely estimate magnetic fields in these regions . for a pl synchrotron spectrum with photon index @xmath96 , the magnetic field at a given magnetization parameter @xmath205 ( ratio of energy densities of magnetic field and relativistic electrons )
depends on the ratio of the luminosity @xmath206 , measured in the @xmath207 frequency range , to the radiating volume @xmath208 .
a simple generalization of equation ( 7.14 ) in pacholczyk ( 1970 ) ( which minimizes the total energy density @xmath209 ) and assumes @xmath210 and @xmath211 . ] to the case of arbitrary @xmath61 gives @xmath212^{2/7}. \label{synch_magn_field}\ ] ] in this equation , @xmath213 and @xmath214 are the characteristic synchrotron frequencies ( @xmath215 ) corresponding to the boundary energies ( @xmath216 and @xmath217 ) of the electron spectrum ( @xmath218 ; @xmath219 ) , and @xmath220 .
the photon indices and 0.58 kev luminosities have been measured from the spectral fits ( see table [ spatially - resolved - spectra ] ) .
we estimate the radiating volumes from the images assuming conical shapes of the outflows . since @xmath221 and @xmath222 , the magnetic field estimated from equation ( [ synch_magn_field ] ) weakly depends on the assumed distance , @xmath223 .
the boundary frequencies @xmath213 and @xmath214 are unknown .
the exact value of @xmath214 is not really important as long as @xmath224 , so we choose a plausible ( but arbitrary ) value @xmath225 kev .
the unknown frequency @xmath213 is the main source of uncertainty of the magnetic field at a given @xmath61 , for the measured spectral slopes . for the southern outflow
, we can at least estimate a lower limit on @xmath213 from the requirement that extrapolation of the x - ray pl spectrum to lower frequencies does not exceed the harder radio spectrum ( @xmath226 ; ng et al .
2010 ) , but even the limit , @xmath227@xmath228 hz , is not very certain because of the uncertainty of the x - ray photon indices ( and because the slope of the radio spectrum was measured in a broader region ) .
since the northern outflow was not detected in the radio , we can only guess that @xmath213 is higher there than in the southern outflow ( which would be consistent with our hypothesis that the northern outflow emission is due to leaked particles with very high energies ) .
given these uncertainties , we estimate the magnetic fields assuming @xmath229 ghz for all the regions . if future multiwavelength observations yield different estimates of @xmath213 , the magnetic field could be scaled ( approximately ) as @xmath230 ( if @xmath224 ) .
the equipartition magnetic fields , @xmath231 , for different parts of the j1509 southern and northern outflows are given in table [ b - fields ] , for the above - described choice of boundary frequencies and @xmath39 kpc .
although the actual magnetic field strengths can differ from those in table [ b - fields ] by a factor of a few , we can at least hope that their relative values reflect the spatial dependence of the magnetic field in the emitting regions .
surprisingly , the magnetic field in the southern outflow shows a hint of non - monotonic dependence on distance from the pulsar : the field grows from segment 1 to segment 2 , and then decreases outwards ( in concert with the photon index @xmath96 and slope @xmath232 of electron spectrum ) .
the reason for such behavior is unclear , and we can not exclude that it is simply due to a variation of unknown parameters ( e.g. , @xmath213 and/or @xmath61 along the outflow ) . since the southern outflow was detected in radio ( ng et al .
2010 ) , the radio spectrum ( @xmath233 ) can also be used for magnetic field estimates .
if the radio and x - ray spectra are generated in different regions ( e.g. , the radio emission comes from a `` sheath '' around an x - ray emitting `` spine '' of the outflow ) , the magnetic field in the radio - emitting region should be estimated solely from the radio spectrum , which gives @xmath234 @xmath235 g for @xmath236 erg s@xmath2 , @xmath237 @xmath238 , @xmath239 mhz , @xmath240 ghz ( @xmath241 is the magnetization in this region ) .
this value of @xmath242 is close to those determined from the x - ray spectra in segments 3 and 4 ( assuming @xmath243 ) .
if the radio spectrum in the x - ray detected part of the southern outflow is generated in the same region as the x - ray emission , then the mutiwavelength ( radio to x - rays ) spectrum can be considered as a broken pl .
the magnetic field from such a spectrum could be estimated for a given magnetization @xmath244 , where @xmath241 and @xmath245 ( each of them is greater than @xmath61 ) are magnetizations corresponding to the energy densities of particles responsible for radiation below and above the break frequency , respectively . in this case
, the equipartition field would be slightly higher than those estimated separately from the x - ray and radio spectra , but the difference would be within the uncertainties . according to table [ b - fields ] ,
the difference of the equipartition fields in the two segments of the northern outflow is within the @xmath246 uncertainties caused by the uncertainties of the spectral slopes . the characteristic equipartition field , @xmath24710 @xmath235 g , is lower than typical values for the southern outflow value higher than 100 ghz for the northern outflow . ] . however
, if the x - ray emission from the northern outflow is indeed due to particles of very high energy leaked from the termination shock apex , then @xmath61 can be very different from unity , and @xmath248 very different from @xmath246 .
the magnetic field strengths can be used to estimate the lorentz factors , gyration radii , and characteristic synchrotron cooling times of electrons radiating synchrotron photons with energy @xmath249 : @xmath250 @xmath251 and @xmath252 the spatially - resolved x - ray spectra of the outflows allow one to crudely estimate ( or constrain ) characteristic flow velocities .
for instance , if the apparent spectral softening in the northern outflow is real , the flow velocity can be estimated as @xmath253 km s@xmath2 , where @xmath254 is the spatial scale of the alleged softening .
this is , of course , just an order - of - magnitude estimate , but the fact that @xmath255 is consistent with the scenario by bandiera ( 2008 ) , which suggests that propagation of the leaked high - energy particles is slowed down by diffusion in a turbulent magnetic field ( see above ) . knowing the equipartition magnetic field and the flow velocity , we can estimate the energy density , @xmath256(\sigma^{4/7}+\sigma^{-3/7})^2 \\
\nonumber & \sim & 2\times 10^{-12 } ( b_{\rm eq}/7\,\mu{\rm g})^2 ( \sigma^{4/7}+\sigma^{-3/7})^2\,\,{\rm erg}\,{\rm cm}^{-3 } , \label{energy_density_north}\end{aligned}\ ] ] and the energy transfer rate ( @xmath257 energy injection rate ) , @xmath258 for the northern outflow , * where @xmath259 is the enthalpy density , @xmath260 is the pressure of the relativistic particles and magnetic field , and @xmath261 is the characteristic cross - sectional area . * as expected , @xmath262 is a small fraction of the pulsar s spin - down power ( @xmath263 for @xmath264 ) . on the other hand ,
@xmath262 exceeds the 0.58 kev luminosity of the northern outflow ( by a factor of 7 at the parameters chosen and @xmath264 , but the value of this factor strongly depends on the poorly known outflow parameters ) .
in the southern outflow we see no spectral softening with increasing distance from the pulsar ( i.e. , no synchrotron cooling ) . if there is no in - situ particle acceleration in the outflow , the lack of cooling sets a lower limit on the flow velocity of a few thousand km s@xmath2 ( k+08 ) , well above the upper limit on the pulsar velocity .
the value of the limiting speed is somewhat uncertain because of the magnetic field uncertainties .
moreover , the hint of spectral hardening with increasing distance from the pulsar ( see figure [ image - gamma - fn - dist ] ) suggests that particle acceleration may occur in the southern outflow ( e.g. , due to magnetic field reconnection ) , which would invalidate the velocity estimates based on passive cooling .
an upper limit of a few tens of thousands km s@xmath2 on the characteristic flow velocity can be estimated from the condition @xmath265 , where @xmath266 , similar to equation ( 7 ) .
lcccc[h ] s. outflow seg . 1 & @xmath267 & 1.3 & 2.6 & @xmath268 + s. outflow seg . 2 & @xmath269 & 1.4 & 7.4 & @xmath270 + s. outflow seg . 3 & @xmath271 & 3.4 & 42 & @xmath272 + s. outflow seg . 4 & @xmath273 & 3.2 & 125 & @xmath274 + n. outflow ( proximal ) & @xmath275 & 1.0 & 9.1 & @xmath276 + n. outflow ( distal ) & @xmath277 & 3.3 & 76 & @xmath278 + as we mentioned above , the j1509 and guitar pwne are not the only ones which exhibit puzzling asymmetric structures misaligned with their apparent pulsar motion directions .
two other examples are the outflows associated with psr j11016101 ( the lighthouse nebula ; pavan et al .
2014 ) and psr j2055 + 2539 ( marelli et al . 2015 ) .
the former one shows a particularly remarkable structure . the deep 300 ks acis image of the lighthouse nebula shown in figure [ image - lighthouse ] reveals a long ( @xmath279 pc ) jet - like outflow that originates near the pulsar , bends back in the direction opposite to that of the pulsars proper motion , and then sharply turns in a direction nearly orthogonal to the proper motion direction ( see also pavan et al .
this outflow can possibly be interpreted in the framework of the bandiera ( 2008 ) scenario , similar to guitar and j1509 misaligned outflows . * in this scenario , the external ( ism ) magnetic field lines are illuminated by the escaping wind particles traveling along the distorted magnetic field ( cf
. magnetic `` draping '' ; lyutikov 2006 and dursi & pfrommer 2008 ) .
* additionally , the pwn of psr b0355 + 54 also shows hints of a misaligned outflow ( see figure [ bowshocks ] ) , although in this case it is rather faint and is seen on both sides of the bow shock ( perhaps due to the lower speed of the pulsar ) .
an in - depth discussion of this pwn will be given in a separate paper ( klingler et al .
, in prep . )
the most important findings from our deep acis - i observations of j1509 can be summarized as follows .
we were able to resolve the fine structure of the cn in the immediate pulsar vicinity .
the cn consists of two lateral tails , apparently emerging from the pulsar and resolved up to 0.2 pc behind the pulsar , and a short , 0.05 pc , axial tail along the cn symmetry axis .
the cn structure and size are remarkably similar to those of the geminga , pwn but different from several other bow shock pwne .
such pwn morphology challenges the existing numerical simulations of synchrotron emission from the pwne of supersonically moving pulsars .
the morphological differences from other pwne could be attributed to different initial wind magnetizations or , more likely , to anisotropic wind ejection , which affects the geometry of the magnetic field downstream of the termination shock and the pwn morphology .
we found that the cn narrows beyond 0.20.3 pc , reaching the minimum width ( where the lateral tails apparently merge ) at about 0.5 pc behind the pulsar . beyond this distance the flow expands and forms the diverging 7 pc long southern outflow , discovered by k+08 who interpreted it as a tail of shocked pulsar wind confined by the ram pressure of the ambient medium .
the pl spectra of the southern outflow do not show softening ( synchrotron cooling ) with increasing distance from the pulsar , which is expected for such long tails .
the lack of softening ( with a hint of hardening ) may suggest that particle acceleration occurs within the tail .
we also discovered a previously undetected wedge - like northern outflow , seen up to about 6 pc , with the wedge axis inclined at an angle of about @xmath280 from the pulsar s proper motion direction , inferred from the cn symmetry axis .
this outflow is possibly formed by energetic pulsar wind particles escaping from the apex of the bow shock and diffusing along the ism magnetic field .
the equipartion magnetic field in this outflow is about a few @xmath235 g , the flow speed is a few thousand km s@xmath2 , and the energy injection rate is @xmath281 erg s@xmath2 .
the discovery of the northern outflow puts j1509 into the ( so far ) small group of supersonically moving pulsars with misaligned outflows ( i.e. , outflows inclined to the pulsar trajectory at substantial angles ) . to better understand the nature of the misaligned outflows , further deep multiwavelength observations with high angular resolution are needed .
we would like to thank maxim lyutikov , andrei bykov , giovanni morlino , marina romanova , and john hewitt for the very insightful discussions , as well as our referee , roger romani , for the helpful suggestions .
support for this work was provided by the national aeronautics and space administration through _ chandra _ award number g03 - 14082 issued by the _
chandra _ x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of the national aeronautics space administration under contract nas8 - 03060 .
the work was also partly supported by nasa grant nnx08ad71 g . | psr j15095850 is a middle - aged pulsar with the period @xmath0 ms , spin - down power @xmath1 erg s@xmath2 , at a distance of about 3.8 kpc .
we report on deep _ chandra x - ray observatory _ observations of this pulsar and its pulsar wind nebula ( pwn ) .
in addition to the previously detected tail extending up to @xmath3 southwest from the pulsar ( the southern outflow ) , the deep images reveal a similarly long , faint diffuse emission stretched toward the north ( the northern outflow ) and the fine structure of the compact nebula ( cn ) in the pulsar vicinity .
the cn is resolved into two lateral tails and one axial tail pointing southwest ( a morphology remarkably similar to that of the geminga pwn ) , which supports the assumption that the pulsar moves towards the northeast .
the luminosities of the southern and northern outflows are about @xmath4 and @xmath5 erg s@xmath2 , respectively .
the spectra extracted from four regions of the southern outflow do not show any softening with increasing distance from the pulsar .
the lack of synchrotron cooling suggests a high flow speed or in - situ acceleration of particles .
the spectra extracted from two regions of the northern outflow show a hint of softening with distance from the pulsar , which may indicate slower particle propagation .
we speculate that the northern outflow is associated with particle leakage from the bow shock apex into the ism , while the southern outflow represents the tail of the shocked pulsar wind behind the moving pulsar .
we estimate the physical parameters of the observed outflows and compare the j15095850 pwn with pwne of other supersonically moving pulsars . |
symmetric matrix inversion is one of the most important problem for many practical tasks e.g. analysis of electrical circuits with inductance elements @xcite , synthesis of kalman or wiener filters @xcite , using of finite element method @xcite .
existing symmetric matrix inversion methods are cholesky decomposition , ldl decomposition @xcite , bordering method @xcite , and the most efficient krishnamoorthy - menon s method ( based on cholesky decomposition , and requires @xmath0 operations with @xmath1 square roots computation ) @xcite .
the aim of this paper is to propose a symmetric matrix inversion method , which reduces machine time spending compared to krishnamoorthy - menon s method by avoiding of square root computations .
moreover , this fact allow us to use the proposed method for efficient inverse of symmetric matrix not only with strict diagonal dominance , but without diagonal dominance as well .
in this section shows modified gaussian elimination , which proposed method based on @xcite-@xcite .
let there is a system of linear equations ( [ sle ] ) , @xmath2 where @xmath3 , @xmath4 , @xmath5 . during modified gaussian elimination an addition matrix @xmath6 changes instead of matrix @xmath7 , but
addition memory is not necessary in this case @xcite .
let vector @xmath8 consist of two types of variables : required , which should be find during elimination , and unrequired , which is not interesting for researcher .
re - solution of ( [ sle ] ) after some changes of @xmath7 could be done with reduced number of multiplications and divisions , using formulae from @xcite-@xcite .
let @xmath9 be an @xmath10 column of @xmath7 ; let @xmath11 be a matrix @xmath12 after @xmath13 change ; let @xmath14 be an @xmath10 row of @xmath11 , @xmath15 ; let @xmath16 , where @xmath17 is an identity matrix
. then solution of ( [ sle ] ) could be done with formulae ( [ modification ] ) below : * @xmath18 for @xmath19 , @xmath20 , and @xmath21 is an unrequired variable ; * @xmath22 for @xmath23 . if @xmath24 , then two rows of @xmath12 should be permitted .
such permutation is always possible for non - singular matrix @xmath7 ( see @xcite-@xcite ) ; * @xmath25 for other required @xmath26 ; * @xmath27 for any required @xmath21 . @xmath28 it can easily be checked that @xmath29 , @xmath30 , for any required @xmath21 and for @xmath31 , @xmath32 .
further , if all elements of @xmath8 are required variables , then @xmath33 .
matrix inversion using modified gaussian elimination requires @xmath34 multiplications and divisions . if only the last element @xmath35 of @xmath8 is a required variable , then gaussian elimination requires @xmath36 multiplies and divisions .
let @xmath37 be a number of required variables in @xmath8 : @xmath38 is a required variable .
then number of multiplies and divisions for gaussian elimination could be determined with formula @xmath39 ( see @xcite ) .
in this section shows a method for efficient symmetric matrix inversion based on modified gaussian elimination . the first variant of the proposed method consist of two different stages .
on the first stage we use formulae ( 2 ) for @xmath40 , and only @xmath35 is a required variable ( indeed , it is a valid proposition for @xmath41 as well ) . on the second stage we use addition formulae described below .
let us introduce some notation .
let @xmath42 be an element from @xmath10 row and @xmath43 column of @xmath44 ; let @xmath45 be an element from @xmath10 row and @xmath43 column of @xmath7 ; let @xmath46be a submatrix of @xmath44 , such that , @xmath47 , @xmath48 ; let @xmath49 be a submatrix of @xmath7 , such that @xmath50 , @xmath47 , @xmath48 .
it is easily shown that after the first stage @xmath12 is a lower triangular matrix .
after the second stage @xmath12 became a matrix @xmath51 , as if we use ( 2 ) for @xmath52 . for symmetric matrix inversion using modified gaussian elimination that is enough to use only lower triangular matrix .
to prove this statement , we need a lemma [ lemma_1 ] .
[ lemma_1 ] let @xmath44 is a matrix from ( [ modification ] ) for , . then @xmath53 . using ( [ modification ] ) , let us consider 4 cases .
let @xmath54 . then @xmath55 case 2
let @xmath56 . then @xmath57 case 3 .
. then @xmath59 it can be checked easily for @xmath60 , @xmath23 ; case 4 .
. then @xmath62 it can be checked easily for @xmath63 : @xmath64 , @xmath32 , .
since lemma [ lemma_1 ] , it follows that @xmath46 is a symmetric matrix .
this statement is needed for the second stage of the method .
let @xmath65 be a matrix @xmath12 after @xmath66 change , @xmath67 , where @xmath68 is a lower triangular matrix after the first stage , and @xmath33 .
let @xmath69 is an element from @xmath10 row and @xmath43 column of @xmath65 .
then the second stage describes with formulae below : * @xmath70 for @xmath71 , @xmath31 , @xmath72 , and for @xmath73 , @xmath74 , @xmath75 ; * @xmath76 for @xmath73 , @xmath77 , .
@xmath78 it is easily to prove that @xmath79 , where @xmath80 is a diagonal matrix , such that @xmath81 , @xmath82 .
number of multiplications and divisions for the first stage of the method describes with formula @xmath36 , as we note earlier .
it is easily shown that number of multiplications and divisions for the second stage describes with formula .
both stages for symmetric matrix inversion using the first variant of the proposed method requires @xmath83 multiplications and divisions .
it is less then requirements of cholesky decomposition or ldl decomposition ( see table 1 ) .
let us remark that proposed method avoid square root computations ; this considerably reduce machine time spending , and make it possible to use proposed method not only for positive determined , but for any invertable symmetric matrices as well . the second variant of the proposed method consist of only one stage .
suppose , that @xmath84 where @xmath85 , but all elements from @xmath86 row of @xmath87 are zeros .
let @xmath88 be an element of @xmath10 row and @xmath43 column of @xmath11 ; let @xmath89 be a @xmath43 column of @xmath11 . for explanation of the following formulae
, we need a lemma [ lemma_2 ] .
[ lemma_2 ] let @xmath44 is a matrix from ( [ modification ] ) for @xmath52 , @xmath23 .
then @xmath90 from ( [ second_variant ] ) be @xmath91 . from ( [ second_variant ] ) @xmath92 .
using ( [ modification ] ) , we get \1 . @xmath93 \2 . @xmath94 if we combine lemma [ lemma_1 ] with lemma [ lemma_2 ] , we get formulae below : * @xmath95 for @xmath23 , , * @xmath96 for @xmath97 , , * @xmath98 for @xmath20 , @xmath99 , @xmath100 , @xmath101 , * @xmath102 for @xmath20 , @xmath99 , * @xmath103 for @xmath104 , @xmath105 , , * @xmath106 for @xmath97 , @xmath82 , @xmath107 .
@xmath108 the formulae ( [ second_a ] ) describe an idea of the method in detail , but for practical tasks it is better to use different formulae : * @xmath109 for @xmath23 , , * @xmath96 for @xmath97 , , * @xmath98 for @xmath110 , @xmath111 , , * @xmath112 for @xmath20 , @xmath113 , @xmath77 , * @xmath103 for @xmath97 , @xmath82 , , * @xmath106 for @xmath20 , @xmath113 , .
@xmath114 it can easily be shown that ( [ second_a ] ) and ( [ second_b ] ) are equivalent . number of multiplications and divisions for the second variant of the proposed method describes with formulae .
it is less then requirements of cholesky decomposition , ldl decomposition or krishnamoorthy - mennon method .
it is the same requirements as for bordering method , but it should be noted that bordering method could not be use for inversion of matrix with @xmath115 , @xmath82 , where @xmath116 is a minor of @xmath7 , @xmath82 , @xmath31 . at the same time , proposed method could be use for inversion of matrix with @xmath117 , if @xmath7 is not a singular matrix .
let us remark that the second variant of the proposed method avoid square root computations as well .
in order to demonstrate advantages of the proposed algorithms , we use matlab based simulation via different cpus . we give results for intel core i5 - 3230 m 2.60 ghz below ( intel pentium dual core t 2390 1.86 ghz and intel atom n450 1.67 ghz gives familiar results ) .
we compare proposed algorithms with the most efficient notable algorithms for symmetric matrix inversion : cholesky decomposition @xcite , ldl decomposition @xcite , and krishnamoorthy - mennon method @xcite , @xcite , @xcite . we generate table with full equations , which describes number of multiplications , divisions and square roots computation for every noted method . the first row of the table 1 describes cholesky decomposition and solving of systems of linear equations @xmath118 , and @xmath119 .
the second row describes ldl decomposition and solving of sle @xmath120 , @xmath121 , and @xmath122 .
the third row describes matrix inversion using krishnamoorthy - mennon method , based on cholesky decomposition @xcite , @xcite .
the fourth row describes the first variant , and the fifth row describes the second variant of the proposed method .
experiment 1 .
inversion of a real symmetric matrix with strict diagonal dominance .
let @xmath123 be a number of multiplications and divisions , and @xmath124 be a number of square root computations , determined with formulae from the table 1 .
let @xmath125 and @xmath126 be numbers of operations , determined with counter variables from matlab scripts .
results of the experiment are given in tables 2 and 3 .
experiment 2 .
inversion of real symmetric matrices of order @xmath1 with strict diagonal dominance .
let @xmath127 be a time for matrix inversion ; let @xmath128 be a second @xmath129 , where @xmath130 is a matrix , inverted via one of described methods , and @xmath131 is a matrix , inverted via matlab function @xmath132 .
results of the experiment are given in tables 4 and 5 .
experiment 3 .
inversion of real symmetric matrices of order @xmath1 without diagonal dominance .
results of the experiment are given in the table 6 .
from tables 2 and 3 we can conclude that formulae from the table 1 are correct . from tables 4 and 5
we can conclude that both variants of the proposed method provide notable reduction of machine time spending and has a good accuracy . from the table 6
we can conclude that both variants of the proposed method increase advantages for matrices without diagonal dominance .
let us remark that it is especially important for inductance matrix inversion .
we propose a new method for symmetric matrix inversion based on modified gaussian elimination with avoiding of square root computations .
proposed method could be useful for any scientific and technical problem with symmetric matrix inversion , especially if matrix has not a diagonal dominance .
chua , l. o. ; lin , p .-
: `` _ _ machine analysis of electronic circuits _ _ '' , in russian , the energy , moscow , 1980 .
happonen , a. , burian , a. , hemming , e. : `` a reconfigurable processing elements implementation for matrix inversion using cholesky decomposition '' _ world academy of science , engineering and technology international journal of computer , information , systems and control engineering _ , 2007 , vol:1 no:3 .
streng , g , fix , g. : `` _ _ an analysis of the finite element method _ _ '' , in russian , the world , 1977 .
golub , g. h. ; loan , c. f. v. : `` _ _ matrix computations _ _ '' , in russian , the world , moscow , 1999 .
krylov , v. i. , bobkov , v. v. , monastyrskii , p. i. : `` _ _ numerical methods in applied mathematics _ _ '' , in russian , the higher school , 1972 .
krishnamoorthy , a. ; menon , d. : matrix inversion using cholesky decomposition , _ arxiv.org : url : http://arxiv.org / abs/1111.4144_. krishnamoorthy , a. , menon d. : `` matrix inversion using cholesky decomposition '' , _ ieee signal processing : algorithms , architectures , arrangements , and applications ( spa ) conference .
_ 26 - 28th of september , 2013
. poznan , poland .
savelov , n. s. : `` transient analysis for preliminarily ordered electrical circuits '' , in russian , _ russian electromechanic _ , 1985 ,
no:4 , p. 85 - 92 .
savelov , n. s. : `` forming of state equation for changing electrical circuits '' , in russian , _ russian electromechanic _ , 1987 ,
no:12 , p. 13 - 18 .
savelov , n. s. , kochubei , p. m. : `` fast electrical circuit analysis with multiprocessor system '' , in russian , _ russian electromechanic _ , 2012 , no:5 ,
savelov , n. s. , khlyntsev s. g. : `` efficient method for parametric synthesis of electrical circuits '' , in russian , _ russian electromechanic _
, 2013 , no:5 , p. 9
savelov , n. s. , kochnev , a. v. : `` fast electrical circuit analysis with reduced machine time and memory spending '' , in russian , _ russian electromechanic _ , 2013 ,
no:4 , p. 3
samarskii a. a. , gulin , a. v. : `` _ _ numerical methods _ _ '' , in russian , the science , moscow , 1989 .
matlabcentral / 1994 - 2014 the mathworks , inc . : _ http://www.mathworks.com/matlabcentral/fileexchange/41957-matrix-inversion-using-cholesky-decomposition _ | in this paper we present two different variants of method for symmetric matrix inversion , based on modified gaussian elimination .
both methods avoid computation of square roots and have a reduced machine time s spending .
further , both of them can be used efficiently not only for positive ( semi- ) definite , but for any non - singular symmetric matrix inversion .
we use simulation to verify results , which represented in this paper . |
microfluidic devices are becoming a means of performing low cost and high - throughput chemical and biochemical analysis on chip .
the range of applications include measuring dynamics of protein folding @xcite , kinetics of enzyme reactions @xcite and surface patterning of cells and proteins @xcite .
some of these applications generally involve cross - stream interaction of two or more fluids flowing side by side in a channel .
however , such systems are characterized by laminar flow due to the small dimensions involved and thus fluids flowing side by side can only mix or interact by molecular diffusion . in pressure driven flow through rectangular microchannels , where the fluid motion through the channel is actuated by pressure pumps ,
the channel velocity profile is approximately parabolic across the shortest dimension , as dictated by the balance between the gradient of the applied pressure and the viscous shear stress in combination with the no - slip boundary condition at the walls . due to slow fluid motion close to the walls , tracer particles
are hardly advected by the flow , thereby spending longer time at the walls . at the center of the channel ,
on the other hand , the flow velocity is quite high and tracer particles are transported more efficiently .
this results in a fairly wide distribution of the amount of time spent by tracer particles in the channel , the so - called residence time . as demonstrated
experimentally @xcite , this wide distribution of residence times gives rise to a non - uniform diffusive broadening across the channel .
ismagilov et al .
@xcite , for example , showed using confocal fluorescence microscopy that in contrast to the center line of the channel , where the extent ( along @xmath0 direction ) of the interdiffusion zone exhibits normal diffusive behavior @xmath1 ( @xmath2=distance from the inlet ) , it scales as one third power of the axial distance , @xmath3 , close to the top and bottom walls .
these observations are shown to be in line with results of a scaling analysis , which makes use of a certain similarity with the general lvque problem @xcite .
numerical solution of the advection - diffusion problem @xcite confirms the existence of the experimentally observed scaling behavior with an exponent of 1/3 in the proximity of the walls as compared to the exponent 1/2 at the center of the channel .
furthermore , these calculations also show that the exponent 1/3 can only be observed if the distance ( along the flow ) from the entrance of the channel is not too large . at sufficiently large distances from the inlet , on the other hand
, the exponent for diffusive broadening close to the walls approaches 1/2 and thus becomes identical to the scaling exponent in the center of the channel .
the cross over distance , @xmath4 , is identified as the distance at which the tracer concentration along the shortest channel dimension ( @xmath5 direction ) becomes homogeneous ( note that @xmath6 is measured along the @xmath0 direction ) .
the above mentioned numerical approaches assume a one dimensional parabolic velocity profile and neglect any dependence of the fluid velocity on the distance @xmath2 from the inlet as well as on the `` neutral '' direction , @xmath0 .
the first assumption is valid at axial distances @xmath7 ( @xmath8=width and @xmath9=height of the channel ) , where the flow is fully developed .
the dependence on @xmath0 , on the other hand , can only be neglected if the width of the channel is large compared to the height @xmath9 , i.e. in the case of large aspect ratio , @xmath10 . in the experiments @xcite , however , the fully developed fluid velocity is _ two _ dimensional , i.e. it depends both on @xmath0 and @xmath5 due to the finite aspect ratios investigated ( @xmath11 ) @xcite . despite this fact ,
the agreement between theory and experiment is quite good suggesting that the specific form of the velocity profile does not play a crucial role as long as a linear regime close to the walls and an approximately uniform flow at the channel center can be assumed . in the present work
, we are going to focus on this aspect via a systematic study of the effect of finite aspect ratio on the spreading dynamics of a tracer field entering the y - junction micromixer through one of the arms ( see fig . [
fig : y - junction ] ) . for this purpose ,
we solve , via the lattice boltzmann ( lb ) method @xcite , the full three dimensional advection - diffusion problem for rectangular channels with various aspect ratios .
our previous studies of wall roughness effects on the chaotic mixing of passive tracers in a 2d channel showed the flexibility of the lb method in dealing with advection - diffusion problem in microchannels @xcite .
the present work extends this approach to 3d with a particular focus on various scaling laws within the _ laminar _ flow regime using smooth walls ( no wall roughness effects ) . .
the origin of the co - ordinate system is at the middle of the entrance .
measurements are taken after reaching a fully developed flow ] in addition to a systematic study of the effects related to a finite aspect ratio , our studies differ from numerical calculations of salmon and ajdari in that we do not make any assumption about the shape of the velocity profile . rather , the velocity profile results from the solution of the navier - stokes equations for the problem of interest . by doing this
, we remain as close as possible to real experiments and realize a range of velocity profiles from practically parabolic to those with strong deviations from parabolic dependence . in the following section ,
we briefly introduce the simulation scheme and also provide some benchmark tests for our lb simulation by comparing our results with known analytical solutions for the spreading of a point source in a 3d microchannel both with and without walls . excellent agreement with the analytical solutions
we then apply in section [ sec : results ] the lb method to study the extent of the inter - diffusion zone between two fluids flowing side by side in a y - shape micromixer .
we study the combined effect of number ( defined as the ratio @xmath12 , where @xmath13 is the maximum fluid velocity , @xmath9 the channel height and @xmath14 the diffusion coefficient of the tracer field ) and channel s aspect ratio on the broadening .
our studies include both the upstream zone where the non - uniform broadening along the @xmath0 direction occurs ( with an extent @xmath3 close to the walls and @xmath1 at the center of the channel ) and downstream region where diffusion has had enough time to homogenize the concentration distribution along the vertical @xmath5 direction whereby leading to homogeneous broadening with a one - half power law both at the walls and in the center of the channel .
a summary compiles our results .
the lattice boltzmann method @xcite is a mesoscopic approach that approximately solves the continuum boltzmann equation in a discrete form by dividing space into a regular lattice , time into discrete steps and velocities into a finite number of vectors @xcite .
the number and direction of the velocities are chosen such that the resulting lattice is symmetric so as to easily reproduce the isotropy of the fluid @xcite .
the density of the fluid at each lattice site is accounted for by a one particle probability distribution @xmath15 , where the subscript @xmath16 represents one of the finite velocity vectors @xmath17 , @xmath18 the lattice site and @xmath19 the time . during each time step particles stream along each velocity vector @xmath20 to a neighboring lattice site and
collide locally , conserving mass and momentum in the process .
the lb equation describing propagation and collision of the particles is given by : @xmath21 where @xmath22 is the collision operator .
the most widely used lb method is the lattice bgk model @xcite which approximates the collision operator by simplifying it to a single time relaxation towards the local equilibrium distribution @xmath23 .
the lattice bgk model is given as : @xmath24 where @xmath25 is relaxation time and the equilibrium distribution @xmath26 is closely related to the low mach number expansion of the maxwellian velocity distribution given as @xcite @xmath27 \label{eq : maxwellian}\ ] ] in eq .
( [ eq : maxwellian ] ) , @xmath28 is the sound speed on the lattice and @xmath29 is a set of weights normalized to unity . near equilibrium and in the limit of small knudsen number (=
mean free path / characteristic length of problem ) the macroscopic navier stokes equation can be recovered using chapman - enskog multiscale analysis @xcite .
the relaxation time @xmath25 is then found to be related to the kinematic viscosity as @xmath30 .
the density @xmath31 , velocity @xmath32 and pressure @xmath33 are computed from the distribution function using : @xmath34 the weights @xmath29 in the equilibrium distribution depend on the number of velocities used for the lattice . in this work
we use the d3q15 model which has fifteen velocities with weights @xmath29 given as : @xmath35 the approach can be extended to simulate diffusing solute by introducing passive tracers into the flow field .
the tracers are advected with the flow velocity @xmath32 and follow similar relaxation and propagation as the fluid medium . in this study , the tracer field @xmath36 does not have any effect on the velocity field ( passive tracer limit ) .
the lb equation governing propagation and collision of the density ( concentration ) distribution of the passive tracers is given as : @xmath37 with the equilibrium distribution function , @xmath38 $ ] .
here , @xmath39 is the concentration distribution of the tracers given as @xmath40 . using chapman - enskog multiscale analysis ,
the advection - diffusion equation can also be recovered in the limit of low mach number and near equilibrium situation .
the relaxation time @xmath41 is then found to be related to the tracer diffusion coefficient as @xmath42 . in the stokes flow regime , assuming a steady flow with a constant pressure gradient @xmath43 along the channel
, the navier - stokes equation can be simplified as @xmath44 . for a rectangular channel of length @xmath45 , width @xmath8 and height @xmath9 with a fully developed flow @xmath46
, this equation can be solved via the method of separation of variables with no - slip boundary condition on the walls of the channel .
one thus obtains the following result for the velocity field @xcite : @xmath47 .
\label{eq : velocity_profile}\ ] ] far away from the side walls and close to the center of the channel @xmath48 , the fluid velocity follows the well known 2d parabolic profile . in the y - shape micromixer
studied here , the velocity profile at the two inlets is quite different from that in the mixing channel . assuming that the channel width @xmath8 is larger than its height , @xmath49 , it takes approximately a distance of the order of @xmath8 from the junction for velocity profile to become fully developed . in the steady state ,
the mass transport of a solute of concentration @xmath39 with a constant isotropic diffusion coefficient @xmath14 in a unidirectional and one dimensional velocity field , @xmath50 , can be describe by @xmath51 given that the initial concentration at the junction , where the two fluids meet , is @xmath52 and the characteristic fluid velocity is @xmath13 , eq . ( [ eq : mass_transfer ] ) can be put into a non - dimensional form using relevant characteristic length scale .
a convenient form is to use @xmath53 and @xmath54 .
equation ( [ eq : mass_transfer ] ) then becomes @xmath55 the first term on the right hand side of eq .
( [ eq : non - dimensional ] ) represents the contribution of axial diffusion .
it can be neglected if _ two _ conditions are fulfilled .
the first condition is that the first term on the right hand side must be small compared to the other two terms i.e. @xmath56 .
obviously , this happens when @xmath57 .
the second condition is that the axial convective term must be large compared to the axial diffusive term , i.e. @xmath58 . in this case , using order of magnitude analysis , the regimes ( @xmath59 ) where this condition is valid can easily be identified to be @xcite , @xmath60 . assuming the validity of both the conditions stated above , eq . ( [ eq : non - dimensional ] ) can be simplified to @xmath61 salmon and ajdari @xcite performed a numerical integration of eq .
( [ eq : reduced_eq ] ) assuming a parabolic velocity profile across the channel , i.e. @xmath62 $ ] .
the lattice boltzmann method , on the other hand , solves the full eq .
( [ eq : mass_transfer ] ) incorporating all relevant 3d features such as the fact that , in general , the fluid velocity is not parabolic and depends on all the three coordinates @xmath63 and @xmath5 , i.e. @xmath64 .
, however , can be neglected after a length of the order of @xmath8 , necessary for the establishment of a fully developed flow profile across the channel . ]
all simulations reported in this work are conducted using the d3q15 lb model on a parallel machine .
the lattice boltzmann code is written in c++ with mpi ( message passing interface ) routines that enable exchange of information between processes .
the code is run on smp 8 x 2 core opteron 2.4 ghz .
typical channel dimensions used vary in the range @xmath65 to @xmath66 as well as a channel of size @xmath67 ( lattice units ) , thereby allowing the study of aspect ratios between 1 and 20 . at the walls of the channel , the bounce - back scheme @xcite is implemented .
populations streamed to the walls are simply reversed back along the directions where they came from .
starting with the fluid at rest , a flow is imposed by the application of a body force .
note that it takes a finite amount of time ( of the order of the momentum diffusion time @xmath68
@xcite ) until steady state is reached with respect to the fluid velocity , i.e. until the velocity field becomes independent of time . in order to avoid this transient effects
, we wait a time of @xmath69 before injecting the tracer field .
after a stationary flow has been reached , we start to continuously inject passive tracers into the channel .
this is simply achieved by setting at each time step the concentration of the tracers at the inlet to 1 ( i.e. @xmath70 for all times @xmath19 ) .
we then monitor the time evolution of the tracer concentration field along the channel .
computation of the extent of diffusive broadening of the concentration distribution at a given cross section is done after a time independent concentration profile has been reached at the area of interest ( corresponding to a dynamic balance between the incoming and outgoing tracer populations ) .
the modeled diffusing analyte in this work has a diffusion coefficient of @xmath71m@xmath72/s corresponding to @xmath73 , where we have chosen the lattice units of @xmath74 m and @xmath75ns .
the kinematic viscosity of the fluid was set to @xmath76 corresponding to the viscosity of water ( @xmath77m@xmath78/s at room temperature ) .
the parameters chosen are readily comparable with that of the experiments .
+ before presenting in section [ sec : results ] results of our simulations on laminar micromixer , we provide here a simple test of our lattice boltzmann approach . for this purpose , we compute the fundamental problem of diffusion of a point source placed at point @xmath79 at time @xmath80 on a 3d square lattice using the lbgk model . assuming a constant and isotropic diffusion coefficient @xmath14 , the system is described by the well known diffusion equation @xmath81 ( @xmath82=the laplace operator ) with the initial condition @xmath83 .
the fundamental solution or green function of this equation is well known to be @xcite @xmath84 in the presence of non - absorbing impermeable walls ( equivalent to no - flux boundary condition ) , the analytical solution for the diffusion of a point source can be obtained by using the principle of superposition and method of images @xcite . the positions @xmath85 of the infinite number of images formed on each side of the box is obtained from the real point source according to the relations @xmath86 , @xmath87 and @xmath88 .
the final result for the concentration profile in the presence of non - absorbing walls is thus @xmath89 using a square lattice of dimension @xmath90=80 , @xmath91=80 , @xmath92=80 and diffusion coefficient @xmath93 ( lattice units ) , we perform lattice boltzmann simulation with a simple bounce back scheme @xcite at the walls .
figure [ fig : lb_and_analytical](a ) compares the simulation results for the concentration profile along the @xmath2-direction with the analytical solution obtained from eq .
( [ eq : solution ] ) . in the same figure we also plot the gaussian function given by eq .
( [ eq : green_function ] ) , which represents the solution of the same problem in the absence of the walls , thus emphasizing the non - triviality of the obtained solution in the presence of the walls .
our simulation results are in agreement with the analytical results obtained from eq .
( [ eq : solution ] ) within an error of 0.2% .
this agreement shows that lb model is able to capture the effect of the wall correctly .
it is straight forward to extend the situation to the case of diffusion in a uniform steady unidirectional velocity field @xmath94 .
the results for this situation are shown in fig .
[ fig : lb_and_analytical](b ) for the case of an infinite system ( no walls ) . here
, the point source exhibits a normal diffusive broadening ( @xmath95 ) , while at the same time being advected by the flow .
the analytic curves shown in fig . [ fig : lb_and_analytical](b ) are flow advected version of gaussian distributions given in eq .
( [ eq : green_function ] ) , where @xmath96 is replaced by @xmath97 . noting that the center of the gaussian distribution along the flow direction obeys @xmath98 , the normal diffusive broadening can also be expressed as @xmath99 .
for a number of @xmath100 , we plot in fig .
[ fig : images ] , the cross - sectional image of the concentration field at various axial positions @xmath101 along the channel for aspect ratios of @xmath102 and 5 respectively .
first note that the width of the interdiffusion zone increases with increasing axial distance ( compare panels ( a ) , ( b ) and ( c ) ) .
this is reminiscent of the role played by time in diffusion process . indeed , recalling that the tracer concentration at a height @xmath5 is advected with the velocity @xmath103 and neglecting the effect of shear for the moment , a rough correspondence between diffusion time and the distance @xmath2 from the inlet can be obtained via @xmath104 .
a larger distance from the inlet thus corresponds to a larger diffusion time . a survey of fig .
[ fig : images ] allows a second important observation , namely that the extent @xmath6 of interdiffusion zone increases when going from the center of the channel towards the walls , the so called `` butterfly effect '' ( see e.g. panel ( b ) ) . again , making use of the above estimate of diffusion time as @xmath104 , a qualitative understanding of this behavior can be gained by noting that @xmath103 is maximum at the center of the channel ( @xmath105 ) and decreases to zero at the walls ( @xmath106 ) .
thus , at a given distance @xmath2 from the inlet , the time available for diffusion is larger in the proximity of the walls as compared to the center of the channel . even though being able to describe some important qualitative features of diffusive broadening , the above argument is too crude to capture the different scaling laws discussed above ( @xmath107 close to the walls as compared to @xmath108 at the channel center ) .
indeed , an argument based only on an estimate of the effective diffusion time would yield @xmath109 .
in other words , one would obtain diffusive broadening with an exponent of @xmath110 for all distances from the wall .
the @xmath5-dependence being reflected in an effective diffusion coefficient @xmath111 . an adequate description of experimental observation , therefore , requires taking into account the effect of the non - uniformity of the flow on tracer distribution .
interestingly , assuming a linear velocity profile seems to be sufficient for a derivation of the observed scaling exponent of @xmath112 close to the walls @xcite .
the exponent @xmath110 , on the other hand , results from the presence of a quasi - uniform flow in the central region of the channel . a third important aspect in fig .
[ fig : images ] is the effect of aspect ratio . a comparison of the left and right panels of fig .
[ fig : images ] at a given distance from the inlet clearly shows that the interdiffusion zone is broader in the case of lower aspect ratio . as a consequence , also the inhomogeneity of the tracer concentration field along the @xmath5 direction
is more pronounced when the aspect ratio is decreased .
one could therefore expect that a longer diffusion time is required in order to homogenize the concentration field across the vertical direction .
using the rough correspondence between time and axial distance along the lines discussed above , the position at which the concentration distribution across @xmath5 direction becomes homogeneous should increase for smaller aspect ratios .
a comparison of panels ( c ) and ( f ) in fig .
[ fig : images ] confirms this expectation .
both these panels correspond to the same distance from the inlet but different aspect ratios . while in fig .
[ fig : images](f ) ( aspect ratio=5 ) the solute concentration is already homogeneous along the @xmath5-direction , it exhibits significant inhomogeneity in fig .
[ fig : images](c ) ( aspect ratio=2 ) . an important consequence of this feature will be worked out below . in order to proceed with a more quantitative analysis
, we adopt a simple definition of the extent of interdiffusion zone @xmath6 as the width of the concentration profile at which the concentration is reduced to 20% of the maximum value at the inlet .
it is to be stressed that the results presented here are insensitive to the specific definition of @xmath6 .
other definitions such as using the integral over the second moment of the spatial derivative of the concentration field , @xmath113 , lead essentially to the same conclusions
. the present simple definition , however , is numerically more robust since it does not require the computation of numerical derivative of the concentration field at discrete intervals .
as discussed in section [ sec : method ] , @xmath1 for diffusion in a uniform velocity field . in the case of the inhomogeneous velocity profile considered here ,
the relation between the extent of the broadening @xmath6 and the axial distance @xmath2 takes the more general form @xmath114 , where @xmath115 is the exponent characterizing the diffusive broadening .
it is shown in ref .
@xcite that @xmath115 in general depends both on @xmath2 and @xmath5 . in order to examine this property within our simulations
, we survey in fig .
[ fig : scalings ] the change of @xmath6 downstream along the @xmath2 direction .
the local slope of the ( log - log ) plots gives the scaling exponent @xmath115 . in @xmath116
cross sections of a rectangular channel for aspect ratios @xmath102 ( left panels ) and 5 ( right panels ) at various reduced distances @xmath117 from the inlet . from top to bottom : @xmath118 and @xmath119 . for all cases shown ,
the number is @xmath100 and the height of the channel is @xmath120 lattice units .
the width of the channel is @xmath121 ( left panels ) and @xmath122 ( right panels ) lattice units . ] versus @xmath123 for aspect ratios of 2 ( upper panels ) and 5 ( lower panels ) at numbers of @xmath100 ( left panels ) and @xmath124 ( right panels ) .
the local slope of the lines gives the scaling exponents .
the @xmath2-axis is non - dimensionalized via @xmath101 .
the channel height is @xmath120 ( left panels ) and @xmath125 ( right panels ) .
the channel widths are chosen such that the aspect ratio is @xmath102 in the case of upper panels and @xmath126 in the case of lower panels ( @xmath121 ( top left ) , @xmath127 ( top right ) , @xmath122 ( bottom left ) and @xmath128 ( bottom right ) ) .
in all the panels shown , the vertical dashed line indicates @xmath129 as obtained from a dimensional estimate of the distance for vertical homogenization of the solute concentration . ] versus the aspect ratio for different numbers as indicated .
right : the same data as in the left panel , rescaled via @xmath130 .
the horizontal solid line marks the value of @xmath131 obtained from an estimate of the time necessary for vertical homogenization in a channel with infinite aspect ratio ( @xmath132 ) : @xmath133 . using the cross - over distance @xmath4 and the mid - channel fluid velocity @xmath13 to estimate the cross - over time , @xmath134 , one obtains @xmath135 and hence @xmath131 ( recall our definition of the number @xmath136 ) .
this estimate is improved significantly by taking , @xmath137 , where @xmath138 is the average fluid velocity across the channel ( dashed horizontal line ) .
+ + , title="fig:",height=245 ] versus the aspect ratio for different numbers as indicated .
right : the same data as in the left panel , rescaled via @xmath130 .
the horizontal solid line marks the value of @xmath131 obtained from an estimate of the time necessary for vertical homogenization in a channel with infinite aspect ratio ( @xmath132 ) : @xmath133 . using the cross - over distance @xmath4 and the mid - channel fluid velocity @xmath13 to estimate the cross - over time , @xmath134 , one obtains @xmath135 and hence @xmath131 ( recall our definition of the number @xmath136 ) .
this estimate is improved significantly by taking , @xmath137 , where @xmath138 is the average fluid velocity across the channel ( dashed horizontal line ) . + + , title="fig:",height=245 ] , and shear rate , @xmath139 , versus @xmath0 for aspect ratios of @xmath102 and 5 .
the distance from the bottom wall is equal to @xmath140 ( = 2 lattice units ) for all the data shown . for the aspect ratio of 5
, there is a wide range of @xmath0-values for which both @xmath141 and @xmath142 are roughly constant thus justifying the assumption @xmath143 .
this assumption , however , fails at the smaller aspect ratio shown . in this case
, there is practically no @xmath0 independent region .
vertical dashed lines mark @xmath144 and @xmath145 for which velocity profiles along the vertical ( @xmath5 ) direction are depicted in the adjacent panel .
( b ) velocity versus @xmath5 for @xmath146 and @xmath145 for aspect ratios of @xmath102 and 5 . in the case of an aspect ratio of @xmath147 ,
the velocity profile is identical for both values of @xmath0 .
this is in accordance with the panel ( a ) where @xmath141 hardly varies in the @xmath0 range delimited by the two vertical dashed lines . at a lower aspect ratio of 2 ,
on the other hand , the effect of the side wall is quite significant .
the number is @xmath100 in all the cases shown.,title="fig:",height=275 ] , and shear rate , @xmath139 , versus @xmath0 for aspect ratios of @xmath102 and 5 .
the distance from the bottom wall is equal to @xmath140 ( = 2 lattice units ) for all the data shown . for the aspect ratio of 5
, there is a wide range of @xmath0-values for which both @xmath141 and @xmath142 are roughly constant thus justifying the assumption @xmath143 .
this assumption , however , fails at the smaller aspect ratio shown . in this case
, there is practically no @xmath0 independent region .
vertical dashed lines mark @xmath144 and @xmath145 for which velocity profiles along the vertical ( @xmath5 ) direction are depicted in the adjacent panel .
( b ) velocity versus @xmath5 for @xmath146 and @xmath145 for aspect ratios of @xmath102 and 5 . in the case of an aspect ratio of @xmath147 , the velocity profile is identical for both values of @xmath0 .
this is in accordance with the panel ( a ) where @xmath141 hardly varies in the @xmath0 range delimited by the two vertical dashed lines . at a lower aspect ratio of 2 ,
on the other hand , the effect of the side wall is quite significant .
the number is @xmath100 in all the cases shown.,title="fig:",height=275 ] within each panel in fig .
[ fig : scalings ] , the extent @xmath6 of the interdiffusion zone is shown versus axial distance both for the center of the channel ( @xmath148 ) as well as in the proximity of one of the walls @xmath149 .
the following features can be observed in the two left panels shown in fig .
[ fig : scalings ] corresponding to a number of @xmath100 . at small distances from the inlet , @xmath150 at the center of the channel while @xmath151 close to the wall .
however , increasing the lattice resolution of our simulation domain in the @xmath5 direction and correspondingly the number reveals another short regime , where @xmath150 across the entire cross section of the channel .
this regime , as shown in fig .
[ fig : scalings ] for number of @xmath124 , occurs earlier in the flow at the entrance of the channel and changes to the 1/3 regime over a short time interval .
the transition of this 1/2 regime to the 1/3 regime can be understood using an analytical argument similar to lvque analysis by assuming a linear velocity profile very close to the top / bottom wall @xcite .
qualitatively , this cross - over represents the enhancement in diffusive broadening arising from a homogeneous shear rate .
since , in the case of the poiseuille flow studied here , the shear rate is practically zero in the center of the channel , the effect appears only close to the walls , where the velocity profile is approximately linear . at larger distances from the inlet , on the other hand , the scaling exponent close to the wall gradually increases towards the value of the exponent at the center of the channel , the latter being very close to @xmath110 .
this latter behavior can be understood by assuming that the cross over from an exponent of @xmath152 to @xmath153 corresponds to a homogeneous tracer distribution along the @xmath5-axis .
a criterion for vertical homogenization via diffusion is obtained from @xmath154 which , using the maximum fluid velocity @xmath13 to estimate the cross over time , @xmath134 , yields the cross over distance @xmath155 . in terms of the reduced distance
, this relation translates to @xmath131 @xcite .
the above estimate of @xmath4 does not take into account the effect of aspect ratio .
in fact , as discussed above , the vertical homogenization takes place at larger axial distances when the aspect ratio is decreased ( compare panels ( f ) and ( c ) in fig .
[ fig : images ] ) . in order to investigate this issue , data in fig .
[ fig : scalings ] are plotted for two different aspect ratios of @xmath102 ( upper panels ) and @xmath126 ( lower panels ) .
indeed , a comparison of the panels ( a ) and ( b ) [ as well as ( c ) and ( d ) ] in fig . [ fig : scalings ] suggests that @xmath4 increases when the aspect ratio is decreased .
the left and right panels in fig .
[ fig : scalings ] differ in number investigated .
this is to underline the fact that the observed trend with regard to aspect ratio is not related to the specific choice of the number . in order to quantify the effect of aspect ratio further , we determine for two aspect ratios of @xmath102 and @xmath126 the values of the exponent @xmath115 for each position @xmath2 and @xmath5 along the channel by performing a running fit on the curves in fig . [
fig : scalings ] as proposed in @xcite .
the results are shown in fig .
[ fig : exponents ] . the dotted black lines in fig .
[ fig : exponents](a ) indicate concentration boundary layer which grows with the axial distance at a rate @xmath156 @xcite .
the growth of the concentration boundary layer results from the diffusive flux of solute from the top / bottom wall to the center of the channel .
the white area of the plot corresponds to the high exponent region of the channel .
our resolution is limited by the small number of points in the @xmath157 direction .
the difference between the exponents obtained for the two investigated aspect ratios is shown in fig . [
fig : exponents](c ) confirming the retarded cross over to normal diffusive behavior in the case of smaller aspect ratio . in a more systematic study ,
we varied the aspect ratio from 1 to 20 and determined the cross over distance @xmath4 .
the left panel of fig .
[ fig : xc ] illustrates the behavior of the thus obtained cross over point within the range of aspect ratios investigated and for different numbers of @xmath158 and @xmath159 . in line with the results presented above
, we observe an increase in the cross over point as the aspect ratio decreases .
the right panel of fig .
[ fig : xc ] depicts a non - dimensionalized version of the same data .
interestingly , the data for different numbers collapse onto a master curve in the limit of high aspect ratios .
this is an important observation , since it suggests that , in this limit , the effect of number is indeed a mere rescaling of time or axial distance . in an attempt to better understand the reason for dependence of @xmath4 on aspect ratio , we plot in fig .
[ fig : conc_and_vprofile](a ) fluid velocity @xmath141 as well as its spatial derivative , @xmath139 , as a function of the transverse coordinate @xmath0 for aspect ratios of @xmath102 and @xmath126 .
the plot is done for a distance of 2 lattice units from the bottom wall . for the aspect ratio of 5
, there is a wide range of @xmath0-values for which both @xmath141 and @xmath142 are roughly constant thus justifying the assumption that @xmath160 ( independent of @xmath0 ) .
this assumption , however , fails at the smaller aspect ratio shown .
the data shown in the right panel of fig . [ fig : conc_and_vprofile ] further underline the importance of taking into account the dependence of the fluid velocity both on @xmath5 and on @xmath0 when the aspect ratio is small .
the data shown in fig . [ fig : conc_and_vprofile ] suggest that the increase in @xmath4 is probably due to the side wall shear effect on solute distribution . at a lower aspect ratio ,
the shear rate @xmath161 decreases strongly close to the side wall .
an estimate from fig .
[ fig : conc_and_vprofile](a ) shows that for an aspect ratio of 2 , the @xmath161 at position @xmath162 is about 40% less than that at the center of the channel ( @xmath144 ) .
therefore , solute diffusing towards the side walls samples a strongly decreasing shear rate .
given that the extent of broadening @xmath6 at the top / bottom wall takes the form @xmath163 @xcite , the interdiffusion becomes faster when approaching the side walls , whereby enhancing the inhomogeneity of diffusion ( the so called `` butterfly effect '' ) which is already present at infinite aspect ratio .
this is exactly what we observe in the concentration profile images shown in fig .
[ fig : images ] . in general , solute diffusing towards the side walls from the center of the channel samples constant velocity gradient @xmath161 up to a distance @xmath9 from the side wall .
the time spent by solute before sampling a substantial decrease in the velocity gradient is , therefore , @xmath164 . for @xmath165 ,
this time scale is greater than the time scale to diffuse from the top / bottom wall to the center of the channel denoted as @xmath166 .
thus , at high aspect ratio , solute concentration becomes homogeneous along the vertical direction long before the side walls are `` felt '' . consequently , the cross over point @xmath4 becomes independent of aspect ratio .
in the case where @xmath167 , on the other hand , one can not neglect the additional enhancement of inhomogeneity of diffusive broadening due to the side wall effect .
this leads to @xmath168 for the effect of aspect ratio being significant . as a survey of @xmath169 in fig .
[ fig : xc](b ) reveals , this simple estimate ( which is based on a dimensional argument only ) lies within a factor of two of the result obtained within our computer simulations .
we study the effect of a finite aspect ratio on transverse diffusive transport of miscible solutes flowing in a pressure driven microchannel using the lattice boltzmann method .
the lattice boltzmann method incorporates the essential 3d features such as the non - parabolicity of the velocity profile and the velocity gradient due to the side walls .
we observe the previously reported @xcite different exponents characterizing the extent of the broadening both at the early stage of mixing of the two fluids and at the later stage downstream .
interestingly , we observe the same scaling laws regardless of the channel aspect ratio
. however , extent of diffusive broadening and the position , @xmath4 , at which the broadening becomes uniform and finally reverts to the 1/2 behavior vary remarkably with the channel aspect ratio . the number , on the other hand ,
is found to play the role of a scale factor in @xmath4 in a way that @xmath170 is independent of aspect ratio .
this is inline with the general structure of the advection - diffusion equation , upon neglection of axial diffusion .
a qualitative understanding of the effect of aspect ratio is provided invoking the influence of the shear stress in the proximity of the side walls on diffusive broadening .
this is based on the idea that side wall shear stress is non - negligible in a region of width @xmath9 close to the side walls .
the corresponding effect on inhomogeneous diffusive broadening will be felt if the solutes at the center of the channel have enough time to reach this region _ before _
vertical homogenization takes place .
this allows to derive a simple criterion to decide whether a given aspect ratio is `` large '' or `` small '' . within a factor of two ,
this estimate correctly reproduces the behavior observed within our lattice boltzmann computer simulations
. 9 e.a .
lipman , b. schuler , o. bakajin , and w.a .
eaton , science , * 301 * , 1233 ( 2003 ) . g. h. seong , jinseok heo , and r.m .
crooks , anal .
, * 75 * , 3161 ( 2003 )
david , a.m. glennys , and g.m .
walker annu rev .
eng . * 4 * , 261(2002 ) .
kalmholz , b.h .
weigl , b.a .
finlayson , and p. yager , anal . chem . *
71 * , 5340 ( 1999 ) .
ismagilov , a.d .
strock , p.j.a .
kennis , g. whitesides , and h.a .
stone , appl .
. lett . * 76 * , 2376 ( 2000 ) .
lvque , ann .
mines , * 13 * , 201 ( 1928 ) .
kalmholz , and p. yager , sensors , actuators b * 82 * , 117 ( 2002 ) . jean - baptiste salmon , and a. adjari , j. appl .
* 101 * , 074902 ( 2007 ) .
j. jimnez j. fluid mech . * 535 * , 245 ( 2005 ) .
p. gondret , n. rakotomalala , m. rabaud , d. salin , and p. watzky , phys .
fluids * 9 * , 1841 ( 1997 ) .
xu zheng , and zhan - hua silber - li , exp .
fluids * 44 * , 951 ( 2008 ) .
qian , d. dhumires , and p. lallemand , euro .
* 17 * , 479 ( 1992 ) . f. varnik , and d. raabe , modelling simul .
* 14 * , 857 ( 2006 ) .
f. varnik , d. dorner , and d. raabe , j. fluid mech .
* 573 * , 191 ( 2007 ) .
f. varnik , and d. raabe , molecular simulation * 33 * , 583 ( 2007 ) .
g. mcnamara and g. zanetti , phys .
lett . * 61 * , 2332 ( 1988 ) . f. higuera , s. succi , and r. benzi , europhys .
* 9 * , 345 ( 1989 ) .
y. qian , d. dhumieres , and p. lallemand , europhys . lett . * 17 * , 479 ( 1992 ) .
s. succi , the lattice boltzmann equation : for fluid dynamics and beyond , oxford university press , ( 2001 ) .
x. he , and l.s .
luo , phys .
e * 56 * , 6811 ( 1997 ) .
r. rubinstein , and l.s .
luo , phys .
e * 77 * , 036709 ( 2008 ) .
u. frisch , d. dhuimres , b. hasslacher , p. lallemand , y. pomeau , and j .-
rivet , complex systems * 1 * , 649 , ( 1987 ) .
koelmann , euro .
lett * 15 * , 603 ( 1991 ) . h. schlichting , boundary - layer theory , mcgraw hill , london ( 1958 ) j. mathews , and r.l .
walker , mathematical methods of physics , second ed . addison - wesley ( 1970 ) .
j. crank , the mathematics of diffusion , oxford , london ( 1956 ) . | we study scaling laws characterizing the inter - diffusive zone between two miscible fluids flowing side by side in a y - shape laminar micromixer using the lattice boltzmann method .
the lattice boltzmann method solves the coupled 3d hydrodynamics and mass transfer equations and incorporates intrinsic features of 3d flows related to this problem .
we observe the different power law regimes occurring at the center of the channel and close to the top / bottom wall .
the extent of the inter - diffusive zone scales as square root of the axial distance at the center of the channel . at the top / bottom wall
, we find an exponent 1/3 at early stages of mixing as observed in the experiments of ismagilov and coworkers [ appl .
. lett . * 76 * , 2376 ( 2000 ) ] . at a larger distance from the entrance , the scaling exponent close to the walls changes to 1/2 [ j
.- b . salmon et al j. appl .
phys .
* 101 * , 074902 ( 2007 ) ] . here
, we focus on the effect of _ finite _ aspect ratio on diffusive broadening .
interestingly , we find the same scaling laws regardless of the channel s aspect ratio
. however , the point at which the exponent 1/3 characterizing the broadening at the top / bottom wall reverts to the normal diffusive behavior downstream strongly depends on the aspect ratio .
we propose an interpretation of this observation in terms of shear rate at the _ side _ walls .
a criterion for the range of aspect ratios with non - negligible effect on diffusive broadening is also provided . |
this paper deals with the calculation of the spectral zeta function z(s ) = _ n=1^ , [ spectral ] associated with the eigenvalues @xmath0 of the inhomogeneous helmholtz equation on a @xmath1-dimensional domain @xmath2 ( -_d ) _
n(x_1 , , x_d ) = e_n ( x_1 , , x_d ) _
n(x_1 , , x_d ) , [ helmholtz_1 ] where @xmath3 is the laplacian operator in @xmath1-dimensions and @xmath4 are its eigenfunctions ( the index @xmath5 refers to the entire set of quantum numbers that specify a solution . ) .
@xmath6 is a density ( @xmath7 for @xmath8 ) .
in general , spectral zeta functions can be built using the eigenvalues associated to a given schrdinger operator : for example voros has studied the zeta function of anharmonic oscillators @xcite , steiner has obtained sum rules for confinement potentials of the form @xmath9 , with @xmath10 and @xmath11 @xcite , berry has studied the zeta function of aharonov - bohm quantum billiards at integer values of @xmath12 @xcite , extending the results of itzykson , moussa and luck@xcite , who had derived an explicit expression for the sum of inverse powers of the laplacian on a simply connected region with dirichlet boundary conditions at the border ; this problem has been later studied by steiner @xcite , obtaining exact expressions for the spectral zeta function at positive integer values , and by elizalde , leseduarte and romeo@xcite , who derived exact sum rules for the zeros of the bessel functions @xmath13 using a recursive formula ( these sum rules were already obtained by in @xcite , considering an infinite square well of radius @xmath14 ) ; steiner has also obtained exact expressions for the selberg s zeta function for compact riemann surfaces@xcite ( in particular the case of the two - torus is discussed in detail in the review article @xcite by arendt et al . ) ; crandall@xcite has calculated the exact values of the spectral zeta functions at integer values for several one dimensional potentials for the power - law oscillator obtained by voros holds for arbitrary power - laws with positive real exponents .
see also ref .
@xcite . ] ; kvitsinsky @xcite has obtained sum rules for two - dimensional domains which are close to the unit disc ; mezincescu @xcite , and later bender and wang@xcite , have used the spectral zeta functions to obtain strong evidence of the reality of the spectrum of certain @xmath15-symmetric hamiltonians .
spectral zeta functions are also useful tools in the calculation of the casimir effect , since the casimir energy is obtained as @xmath16 . the series given in eq.([spectral ] )
diverges in this case and one needs to analytically continue it to negative values of @xmath12 to obtain a finite physical value for the casimir energy : this is the essence of zeta function regularization method @xcite .
applications of these techniques may be found in @xcite .
unfortunately , the direct use of eq.([spectral ] ) is possible only if the eigenvalues @xmath0 are known exactly ; in this paper we develop a perturbative approach to the calculation of @xmath17 that allows one to obtain explicitly the corrections to the spectral zeta function to a given order in the perturbation , bypassing eq .
( [ spectral ] ) . in our calculation
we assume that @xmath2 is a @xmath1-cube with a density @xmath18 which is a perturbation of a uniform density ( in two dimensions @xmath2 may also be obtained from the small deformation of a square , a circle or a domain where a basis is available ) .
we have derived a general expression for the spectral zeta function for problems with small inhomogeneities and/or deformations , which we have worked out explicitly to second order in the perturbation .
we have applied our formula to different problems , obtaining in many cases the analytic continuation to all values of @xmath12 ( this has been generally accomplished only to first order , and to second order in a specific example a one dimensional string ) . for the case of a string with piecewise constant density , we have reproduced to first order in the perturbation the results obtained by hadasz , lambiase and nesterenko@xcite .
although most calculations of casimir energies for inhomogeneous systems concern the special case of piecewise constant densities@xcite , our formalism applies to more general densities .
in particular , we have observed that the casimir energy of an inhomogeneous string , calculated to first order in perturbation theory , in some cases contains irremovable divergences , when either dirichlet or neumann boundary conditions are used , but that these singularities cancel in the sum of the two zeta functions corresponding to the different boundary conditions .
this paper is organized as follows : in section [ sec : rspt ] we briefly introduce the rayleigh - schrdinger perturbation theory for a general hermitian operator ; in section [ sec : zeta ] we obtain an explicit formula for the spectral zeta function using the perturbative expansion of section [ sec : rspt ] ; in section [ sec : applications ] we discuss several applications in one , two and three dimensions ; finally , in section [ conclusions ] we draw our conclusions .
rayleigh - schrdinger perturbation theory is a standard tool of quantum mechanics : it allows to obtain perturbative expressions for the eigenvalues and the eigenstates of a hermitian operator @xmath19 , defined as the sum of a solvable hermitian operator @xmath20 and a of perturbation @xmath21 ( @xmath22 ) , which typically does not commute with @xmath20 . in this framework , using the basis of the eigenstates of @xmath20 , one obtains an explicit expression for the eigenvalues of @xmath23 , as a power series in @xmath24 , which to second order reads c_n & & a_n +
n | | n + ^2 _ k n + . [ pert ] the formula above assumes that @xmath25 is a non - degenerate eigenstate of @xmath20 . assuming that for an arbitrary small ( but finite ) @xmath24 , @xmath21 is a perturbation _ for all the states _ belonging to the spectrum of @xmath23 , we may write the trace of @xmath26 as a power series in @xmath24 using the explicit perturbative expression for @xmath27 : ^s = _ n , [ tracec ] where the last term has been symmetrized with respect to the summation indices .
notice that this term is well defined even when @xmath28 and @xmath29 are degenerate states , even though the corresponding expression in eq.([pert ] ) diverges in this case . in a similar fashion one may extend these results to the operator @xmath30 , obtaining f ( ) & = & _ n .
we work with the spectral zeta function of eq.([spectral ] ) associated with the eigenvalues @xmath0 of the hermitian operator @xmath31 , on the d - dimensional cube @xmath2 ( @xmath32 is the d - dimensional laplacian and @xmath33 is a positive function on @xmath2 ) .
the spectrum of @xmath34 is bounded from below , with strictly positive eigenvalues for dirichlet boundary conditions for @xmath35 .
if @xmath18 is a physical density , then the eigensolutions of @xmath34 provide the normal modes of an inhomogeneous d - dimensional cube ; in two dimensions , @xmath18 may also be obtained by conformally mapping the eigenvalue problem defined on an arbitrary two dimensional domain onto the square
. the specific form of @xmath34 given above is obtained from the helmholtz equation ( [ helmholtz_1 ] ) after recasting it into the equivalent form ( -_d ) _
n(x_1 , , x_d ) = e_n _ n(x_1 , , x_d ) , [ helmholtz_2 ] where @xmath36 . the advantage of working with eq.([helmholtz_2 ] ) rather than with eq .
( [ helmholtz_1 ] ) resides in the manifestly hermitian form of the operator @xmath37 .
alternatively , one may obtain a still equivalent equation ( x_1 , , x_d ) _
n(x_1 , , x_d ) = _ n(x_1 , , x_d ) , [ helmholtz_3 ] where @xmath38 . in this case one works with the hermitian operator @xmath39 , which being the inverse of @xmath34 is isospectral to @xmath40 . ] .
notice that the eigenvalues of eq .
( [ helmholtz_3 ] ) are the reciprocal of the eigenvalues of eqs.([helmholtz_1 ] ) and ( [ helmholtz_2 ] ) . in @xmath1 dimensions , the asymptotic behavior of the energies of eq .
( [ helmholtz_1 ] ) for @xmath41 is described by weyl s law , @xmath42 , where @xmath5 is the number of states with energy smaller than @xmath0 . as a result
, the series @xmath43 will converge to a finite value only for @xmath44 . using the invariance of the trace with respect to unitary transformations , for @xmath44
, we may express the spectral zeta function directly in terms of the diagonal matrix elements of the operators @xmath45 and @xmath46 : z ( s ) & = & _ n_1,
,n_d n_1 , , n_d | ^-s |n_1 , , n_d = _ n_1, ,n_d
n_1 , , n_d | ^s |n_1 , , n_d , s > d/2 [ zeta2 ] where @xmath47 is an eigenstate of @xmath48 on the d - cube @xmath2 and @xmath49 the corresponding eigenvalue ( when possible we will use the compact notation @xmath25 and @xmath50 to refer to @xmath47 and to @xmath49 ) . for integer values of @xmath12 , with @xmath44 , eq.([zeta2 ] ) provides exact sum rules .
we will now discuss the case in which @xmath51 , with @xmath52 , corresponding to a small perturbation of a uniform problem ; taking into account this fact we write : ^s = ( + ) ^s and treat the second term in the parenthesis as a perturbation .
after identifying @xmath53 and @xmath54 we then obtain z(s ) & = & tr ^s = _ n ( ) ^s - _ n _ k n n | | k ^2 + o [ spectralzeta ] where @xmath55 $ ] refers only to terms containing off - diagonal matrix elements , since the purely diagonal contributions have been included to all orders .
we define the first term in the expression above z^(diag)(s ) & & _ n ( ) ^s , to be the component of @xmath17 which only contains diagonal matrix elements .
there is an important aspect concerning eq.([spectralzeta ] ) : since the matrix elements of @xmath56 are always finite in the limit @xmath57 and @xmath58 for a fixed @xmath59 , with @xmath60 ; in ref .
@xcite explicit expressions for these matrix elements for dirichlet bc in one dimension are derived . ] , the perturbative hierarchy in the expression for @xmath17 is not destroyed by an unbounded growth of the perturbative corrections with respect to the quantum numbers and therefore eq.([spectralzeta ] ) provides a genuine perturbative expansion in this case . on the other hand
, we can easily find an example where the perturbative expansion for the spectral zeta function breaks down : in the case of the dimensional anharmonic oscillator , with hamiltonian @xmath61 ( @xmath62 ) , the perturbative corrections to the eigenvalues of @xmath63 are polynomials in the quantum number @xmath5 of increasing order e_n = ( n+1/2 ) + ( n^2 + n + ) + ^2 ( -n^3-n^2-
n - ) + and therefore for any arbitrary small @xmath64 the expansion breaks down for sufficiently large @xmath5 .
therefore , for any finite value of @xmath64 one can not apply perturbation theory to all the spectrum of @xmath63 .
notice however that our formalism applies to the hamiltonian @xmath65 ( @xmath66 ) , where v(x ) = \ { cc v(x ) & x_min x x_max + & x
< x_min , x > x_max + . ,
and @xmath67 is a potential bounded from below . in this case
the asymptotic behavior of the spectrum is determined by the confinement of the particle between the infinite walls and does not depend on the perturbation .
this property is sufficient to make @xmath68 a perturbation for all the spectrum , for a given infinitesimal @xmath24 .
let us now consider the operator : ^-1 = , which is isospectral to @xmath69 used in ref.@xcite .
the heat kernel of @xmath70 is defined as k(t ) = _ n e^-_n t , where @xmath71 are the eigenvalues of @xmath70 . in order to apply eq.([tracef ] ) we write = ( -_d ) + and identify @xmath72 and @xmath73\sqrt{(-\delta_d)}$ ] . using eq.([tracef ] )
the heat kernel reads k(t ) _
n e^- ( _ n + n | | n ) t - _ n _ k n ( ) n | | k^2 + . [ heatkernel ]
notice that for @xmath74 we have - _ n _ k n ( ) n | the perturbative expansion for the heat kernel : k(t ) _ n e^-_n t .
the spectral zeta functions associated with the eigenvalues of @xmath70 may be obtained from the heat kernel as z(s ) = _
0^t^s-1 k(t ) dt .
[ heatzeta ]
in this section we consider several applications of the formulas obtained in the previous sections . before discussing specific examples of inhomogeneous strings , it is useful to make some general considerations : the casimir energy of an inhomogeneous string is obtained calculating the value of its spectral zeta function at @xmath75 . assuming that the density of the string is a perturbation of a constant density ( @xmath76 ) and working to first order in perturbation theory
, we have @xmath77 , where @xmath78 are the eigenvalues of the homogeneous string subject to specific boundary conditions and @xmath79 are its eigenstates .
the analytic continuation of this series allows one to obtain the casimir energy of the string , directly evaluating @xmath80 : however , if @xmath17 contains terms which are proportional to @xmath81 , the corresponding @xmath80 , calculated to first order , diverges , making the physical interpretation problematic .
previous examples of casimir energies containing irremovable divergences are known : for example , sen calculated the casimir energy of a circle in two dimensions , finding that its casimir energy is infinite @xcite ; bender and milton have calculated the casimir energy of a massless scalar field in a hyperspherical shell in d spatial dimensions and observed that it diverges when d was a positive even integer @xcite . it is easy to see that for an inhomogeneous string , to first order in the perturbative expansion a divergent casimir energy is obtained when the matrix element @xmath82 contains a term going as @xmath83 . in the case of dirichlet boundary conditions , this matrix element may be expressed in terms of the series @xcite n | | n & = & _ -l^+l ( x ) dx - _ k=0^ ( -1)^k , [ eq_dir_1 ] where @xmath84 ; therefore the casimir energy of an inhomogeneous string is finite only if @xmath85 .
this condition is fulfilled if the perturbation is either an odd function of @xmath86 or if @xmath87 is flat at the border : @xmath88 .
therefore the casimir energy of the inhomogeneous string contains an irremovable divergence when none of these conditions is met ; on the other hand , irremovable divergences may still show up to higher orders , even when these are absent to first order , as we will see in an example .
the casimir energy of a string with piecewise constant density and four different sets of boundary conditions at its ends ( dirichlet - dirichlet , neumann - neumann , dirichlet - neumann and neumann - dirichlet ) has been calculated in ref.@xcite ; the density of the string is ( x ) & = & \ { ccc _ 1 & , & -l < x x_0 + _ 2 & , & x_0 <
x < l + . . to uniform our notation to that of ref.@xcite , we define @xmath89 , @xmath90 , @xmath91 and @xmath92 .
@xmath93 , @xmath94 and @xmath95 are defined as in ref.@xcite : , + , - . for dirichlet boundary conditions ,
the exact eigenvalues are the solutions of the trascendental equation + = 0 , which corresponds to the first of eqns .
( 2.4 ) of ref .
@xcite . for a string of arbitrary density ,
the energies go asymptotically as @xmath96 , where @xmath97 ( see for instance @xcite ) : in the present case @xmath98 and therefore @xmath99 for @xmath100 , which is the behavior of the spectrum of a uniform string with constant density @xmath101
. we may therefore write ( x ) | + ( x ) and apply our formulas to first order in @xmath102 , for which only diagonal matrix elements of the perturbation are needed . for dirichlet boundary conditions , the basis is _
n^(d)(x ) = , with @xmath103 .
correspondingly , the eigenvalues of the negative laplacian with dirichlet bc are _
n^(dd ) =
. the matrix elements of @xmath18 in this basis are @xmath104 notice that @xmath105 does not contain terms going like @xmath83 for @xmath100 and therefore we expect that the casimir energy of this string will be finite . a simple calculation , carried out to order @xmath106 , yields z^(dd)(s ) & & |^s _
n=1^\ { + s } + & = & ^-2 s-1 ^2 s , where @xmath107 and @xmath108 is the polylogarithmic function .
we observe that this equation provides an analytic continuation of the series to negative values of @xmath12 ; in particular , for @xmath109 this formula returns the casimir energy of the string , which reads e_c^(dd ) = z^(dd)(-1/2 ) = -+ .
[ piecewise ] the same result can be obtained regularizing the series , evaluated at @xmath75 , with a cutoff function , @xmath110 , and taking the limit @xmath111 at the end of the calculation : e_c^(dd ) & = & z^(dd)(-1/2 ) = _
n=1^ e^-a n + & = & + - , a 0^+ . notice that the finite part of this expression agrees with the result of eq.([piecewise ] ) .
we can compare this result with the analogous result of eq.(3.5 ) and ( 3.6 ) of ref.@xcite ; taking @xmath112 and calculating ( 3.5 ) to order @xmath106 we obtain the result of our eq.([piecewise ] ) .
we may also calculate the exact value of the spectral zeta function for this string at @xmath113 : z^(dd)(1 ) = _ n=1^ = + i ( _ 3(e^-)-_3(e^ ) ) .
this expression holds to all orders in @xmath106 . the calculation for the cases of neumann - neumann , neumann - dirichlet , dirichlet - neumann and periodic - periodic boundary conditions can be done in a similar fashion .
here we just report the results .
the spectral zeta functions for these cases , calculated to first order , are z^(nn)(s ) & = & ^-2 s ^2 s ( 2 s ) + i ( 2 ) ^-2 s-1 s e^- ^2 s ( ( e^-,2 s+1 , ) .
+ & - & . e^ ( e^,2 s+1 , ) -e^ _ 2 s+1 ( e^-)+e^ _ 2s+1(e^ ) ) , + z^(dn)(s ) & = & ( 4^s-1 ) ^-2 s ^2 s ( 2 s ) + & + & ^-2 s-1 s ^2 s ( e^- ( -e^-,2 s+1 , ) + e^ ( -e^,2 s+1 , ) ) , + z^(pp)(s ) & = & 2 ^ 1 - 2 s ^-2 s ^2 s ( 2 s ) , where @xmath114 is the lerch transcendent function . the case of neumann - dirichlet bc is simply obtained from the case dirichlet - neumann bc with the substitution @xmath115 .
the casimir energies for these cases are then e_c^(nn ) & = & - - , + e_c^(dn ) & = & - , + e_c^(nd ) & = & + , + e_c^(pp ) & = & - .
these results agree to order @xmath106 with the expressions calculated in ref.@xcite , with the exception of the pp case which was not studied there .
exact sum rules can also be obtained using eq.([zeta2 ] ) : z^(nn)(1 ) & = & -i ( _ 3(e^-)- _ 3(e^ ) ) , + z^(dn)(1 ) & = & + + & & ( e^- ( e^-,3,)-e^ ( e^,3 , ) ) , + z^(pp)(1 ) & = & .
notice that z^(pp)(1 ) = ( z^(dd)(1 ) + z^(nn)(1 ) ) .
we consider a string of length @xmath116 ( @xmath117 ) and with density ( x ) = 1 + 1 + , where @xmath118 ( this condition is needed to enforce @xmath119 for @xmath120 ) . in this case
the matrix elements of @xmath102 in the basis of the homogeneous string are particularly simple n | | m = - , and allow one to calculate exactly the value of the spectral zeta function of the string at the first few positive integers values of @xmath12 : z(1 ) & = & l^2 + z(2 ) & = & + - + z(3 ) & = & + + - .
using our eq.([tracef ] ) we may calculate the spectral zeta function of this string to second order : z(s ) & = & 4^s ( ) ^2 s ( 2s ) + ^2 s 4^s-1 ( ) ^2 s _ n=1^ + o .
let us define : ^(1)(a ) & & _
n=1^ = _ j=0^ ( j+1-a ) = + _ j=0^ ( ( j+1-a ) -1 ) , + ^(2)(a ) & & _
n=1^ = _ j=0^ ( ( j+1-a)-1 ) , and ( a ) & & ^(1)(a ) -^(2)(a ) = - _ j=0^ ( ( 2j+2-a ) -1 ) . [ psis ] notice that @xmath121 is singular at odd values of @xmath122 , @xmath123 and that @xmath124 .
thus the spectral zeta function reads z(s ) & = & 4^s ( ) ^2 s ( 2s ) + ^2 s 4^s-1 ( ) ^2 s ( 2 - 2s ) .
in particular , around @xmath75 we have : z(s ) & & - + ( 3 ( ) + 3 -31 ) + o ( ( s+1/2)^2 ) , where @xmath125 is the euler mascheroni constant . in this case a divergence in the casimir energy shows up to second order in perturbation theory , since @xmath85 .
clearly the divergences which emerge at a given order in the perturbative expansion can not be canceled by divergent contributions stemming from higher orders , given the different dependence on the expansion parameter .
the cancellation of these divergences therefore requires considering equally divergent terms which may show up in the spectral zeta function corresponding to different boundary conditions evaluated to the same order in perturbation theory .
we will now prove that the sum @xmath126 does not contain a singularity at @xmath75 to first order in perturbation theory .
to prove this point consider z^(dd+nn)(s ) & & z^(dd)(s ) + z^(nn)(s ) + & & |^s _ n=1^\ { + s } + & + & |^s _ u=1 ^ 2 _ n=1^\ { + s } , where _
n , u^(nn ) & = & \ { ccc & , & n=0,1,2 , , u = 1 + & , & n=1,2 ,
, u = 2 + .
are the eigenvalues of the negative 1d laplacian with neumann - neumann boundary conditions and @xmath127 .
notice that _
2n^(dd ) = _ n,1^(nn ) , _
2n-1^(dd ) = _ n,2^(nn ) . moreover _ 2n^(dd)(x)^2 + _ n,1^(nn)(x)^2
2n-1^(dd)(x)^2 + _ n,2^(nn)(x)^2 & = & , where _
n , u^(nn)(x ) = \ { ccc & , & n = 0 , u=1 + & , & n = 1,2 , , u=1 + & , & n = 1,2 , , u=2 + . , are the eigenfunctions of the negative 1d laplacian with neumann - neumann boundary conditions . using these identities we may write the spectral zeta function for the one - dimensional string to first order as z^(dd+nn)(s ) & & 2 |^s ( 1 + _ -l^l dx ) _ n=1^ , [ zetaem ] where @xmath128 is the spectral zeta function of the homogeneous string
. therefore the only singularity of the expression in eq .
( [ zetaem ] ) is the singularity of the riemann zeta function at @xmath129 and the corresponding casimir energy is _ always _ finite , as anticipated . of eq.([psis]).,width=302 ]
[ figure ] we consider a string of unit length ( @xmath130 ) with density ( x ) = , where @xmath131 and dirichlet boundary conditions at its ends .
@xmath93 is a free parameter ( @xmath132 ) .
it has been proved by borg that this string is isospectral to a homogeneous string of the same length and unit density @xcite : therefore the exact energies are e_n = ^2 n^2 and the corresponding spectral zeta functions are easily calculated : z_borg(s ) = .
it is possible to obtain an explicit expression for the matrix elements of @xmath18 in the basis with dirichlet boundary conditions ; for instance , the diagonal matrix elements read n | | n & = & , where @xmath133 and @xmath134 are the cosine and sine integrals . using these matrix elements
we obtain the identity z_borg(1 ) & = & = _ n=1^
+ & = & _ n=1^ , which we have numerically verified with high accuracy .
the isospectrality implies that the spectral zeta function of the borg string is independent of @xmath93 : using this property and working to a given order in @xmath93 , we may obtain non trivial mathematical relations .
for example , if we select the contribution of order @xmath135 in the perturbative expression for the spectral zeta function we obtain the identity : @xmath136 which provides the analytic continuation of the double series to negative values of @xmath12 ( see fig.[figure2b ] ) . of eq.([borg_analytic ] ) as function of @xmath12 ( solid line).,width=302 ] [ figure2b ] we may also obtain non trivial identities working with the heat kernel : in this case we need to evaluate the matrix elements of @xmath137 : n | | m & = & \ { ccc -+ & , & n = m + & , & n m . if we use eq .
( [ heatkernel ] ) , to order @xmath135 we obtain the identity : _
n=2^_m=1^n-1 = - t _ 3(0,e^-^2 t ) + - _ 3(0,e^-^2 t ) , + [ heat ] where @xmath138 is the jacobi theta function .
we have tested this identity numerically for several values of @xmath139 . using eq.([heatzeta ] )
inside eq.([heat ] ) we have _ n=1^_kn=1^ = ( 2 s)- ( 2 ( s+1 ) ) . after combining this identity with eq.([borg_analytic ] )
we obtain an alternative series representation for the riemann zeta function : ( s ) = _ n=2^_k=1^n-1 .
[ zetanew ] we consider now a string with rapidly oscillating density ( x ) = 2 + ( 2(x+|l)/ ) , where @xmath117 and @xmath140 .
this particular problem has been studied by castro and zuazua in ref .
@xcite and more recently by myself in ref .
the solutions to this problem with wavelength comparable to the typical size of the density oscillations are localized at the ends of the string @xcite . in particular , in ref .
@xcite we have obtained both numerical and analytic approximations to the low part of the spectrum of this string ( the analytic result improves the results previously obtained by castro and zuazua in ref .
@xcite ) . in this case
the non - zero matrix elements of the density are n | | m & = & \ { ccc 2- ( -1)^n ( ) & , & n = m , n = 1/| + 2 + & , & n = m , n 1/| + - & , &
n m , ( nm ) 2/| + . , where @xmath141 . as in the example of the piecewise constant string we may calculate ( l ) = , where @xmath142 is the incomplete elliptic integral of second kind . the asymptotic behavior of the spectrum of this string is @xmath143 , which is also the spectrum of a homogeneous string with density @xmath144 .
working to first order in @xmath24 we have z(s ) & & |^s _ n=1^\ { + s } + & & _ n=1^\ { ( ) ^s + s ( ) ( ) } .
this expression contains the series ( s,| ) _
n=1^ , [ phis ] which we may express as ( s,| ) & = & _ n=1^[1/| ] + _
n=[1/|]+1^ ^(1)(s,| ) + ^(2)(s,| ) , where @xmath145 $ ] is the integer part of @xmath122 .
the function @xmath146 may be cast in the form ^(2)(s,| ) & = & _ n=[1/|]+1^_k=0^ = _
k=0^ ( 2 ( k+s),+1 ) where @xmath147 is the hurwitz zeta function .
notice that @xmath148 has an infinite number of singularities , located at semi - integer values @xmath149 , with @xmath150 of eq.([phis ] ) as function of @xmath12 for @xmath151 ; the horizontal line is @xmath152 .
, width=302 ] [ figure3 ] for @xmath153 the spectral zeta of this string is z(s ) & & ( ) ^s ( 2 s ) + s ( ) ( ) ( s,| ) . for @xmath154 , this spectral zeta function to first order in @xmath24 reads z(s ) & & - - , where @xmath155 is the harmonic number of order @xmath5 .
notice that the term corresponding to @xmath156 in the series is treated separately because of the singularity of the hurwitz zeta function .
it is also useful to write : ^(1)(-1/2,| ) - h _
= _ n=1^ [ ] = + ---h _ .
since @xmath80 diverges for @xmath157 or for @xmath158 non integer , the casimir energy of this string is infinite in these cases .
moreover , in this example , there is an infinite sequence of singularities of the spectral zeta function located at negative semi - integer values of @xmath12 .
it is easy to see that this divergence is absent if the original density is replaced by a piecewise constant curve , for instance @xmath159 for @xmath160 . in this case
the diagonal matrix elements of @xmath161 do not contain the @xmath83 term and the resulting spectral functions are finite at @xmath162 . calling @xmath163 , this result holds for arbitrary small ( but finite ) @xmath164 ( or equivalently for arbitrary large @xmath165 ) .
here @xmath164 acts as a cutoff of the highly excited states with quantum numbers @xmath166 .
let us discuss in detail this approach ; in this case we have ( l ) = 2 l + ( ) ( ) ( ) + o(^2 ) and @xmath144 .
the diagonal matrix elements of @xmath167 may be evaluated explicitly and read : n | | n = 2 + .
working to first order in @xmath24 we have z(s ) & & 2 ^ 3 s-1 ^-2 s l^2 s ( 2 + s ( ) ( ) ( ) ) ( 2 s ) + & + & ( -2 ^ 3 s-2 ) ^-2 s-1 s ( ) l^2 s ( ) _ n=1^n^-2 s-1 ( ) ( ( ) + ( ) ) .
the function @xmath168 is periodic with period @xmath165 , @xmath169 , and therefore we may write z(s ) & & 2 ^ 3 s-1 ^-2 s l^2 s ( 2 + s ( ) ( ) ( ) ) ( 2 s ) + & + & ( -2 ^ 3 s-2 ) ^-2 s-1 s ( ) l^2 s ( ) n^-2 s-1 + & & _ n=1^n-1 ( ) ( ( ) + ( ) ) ( 2 s+1 , ) . for a finite @xmath165 , this expression is finite at @xmath75 : the divergent behavior of the casimir energy of the string with a smooth density is recovered only in the limit @xmath170 , where the sum over @xmath5 diverges . as a further example , we may calculate the spectral sum rule at @xmath113 for a string with an arbitrary density of period @xmath171 and length @xmath172 : ( x ) = + _ j=1^ , where a_0 = - _
j=1^ and @xmath173 is the total mass of the string .
the diagonal matrix elements of the density in this case are n | | n & = & - _ j=1^ and z(1 ) & = & _ n=1^ = + _ j=1^ .
notice that @xmath174 reduces to the sum rule of a homogeneous string for @xmath175 with @xmath165 integer . in this section
we will discuss the calculation of the spectral zeta functions of inhomogeneous drums .
we consider a square of side @xmath116 : in this case the eigenvalues of the negative laplacian ( with dirichlet boundary conditions at the border ) are known exactly and read e_n_x , n_y = ( n_x^2+n_y^2 ) , with @xmath176 .
the series defining the spectral zeta function for this problem , z_(s ) = _ n_x , n_y=1^ , may be analytically continued to all @xmath12 .
ziff @xcite and steiner @xcite report the spectral zeta function of the square : using the notation of ziff it reads z_(s ) = ( ) ^2s ( ( s ) ( s)-(2 s ) ) , where @xmath177 .
notice that @xmath129 and @xmath113 are the only singularities of @xmath178 . around @xmath113
the zeta function behaves as z_(s ) & & \ { + ( ( ) + -+ ) + } + & & \ { + g _ + } , where @xmath179 is the area of the square .
@xmath125 is the euler mascheroni gamma and @xmath180 is the stieltjes constant defined as @xmath181 ( @xmath182 is the hurwitz zeta function ) .
let @xmath183 be an arbitrary simply connected domain of the plane and call @xmath0 the eigenvalues of the helmholtz equation on this domain , assuming dirichlet boundary conditions on @xmath184 .
riemann s theorem grants the existence of a conformal map from @xmath183 to any other simply connected region of the plane and in particular to @xmath185 , @xmath186 : w = u + i v = f(z ) , @xmath187 . under this map , the original helmholtz equation on @xmath183 transforms into the helmholtz equation on @xmath2 , in presence of a variable density @xmath188 ( see ref .
. we will refer to @xmath189 as a _ conformal density _ , to underline that it does not correspond to a physical density of the membrane .
we observe that , if @xmath189 is a conformal density , its integral over the square is just the area of the region @xmath183 , @xmath190 . for @xmath191 we may obtain a general behavior of the spectral zeta function to order @xmath192 as .
z(s ) |_s1^+ & & _ n=1^ ( ) ^s + o(s-1 ) + & & _ n=1^ ( ) ^s + _
n=1^+ o(s-1 ) , where the second series converges for @xmath193 .
the first series , on the other hand , reproduces the spectral zeta function of a square of side @xmath116 , apart for a multiplicative factor @xmath194 . for this reason , we may obtain the general behavior of the spectral zeta function at @xmath113 as : .
z(s ) |_s1^+ & & \ { + g + } , where g & & g _ + + _
n=1^ + & = & ( ( ) + -+ ) + _ n=1^ .
itzykson , moussa and luck have found this behavior long time ago ( eq.(6 ) of ref .
@xcite ) , obtaining an expression for @xmath195 in terms of an integral . as a specific example
, we consider the domain obtained from the map f(z ) = for @xmath196 ( we have studied this problem for @xmath197 in a previous paper , ref.@xcite , calculating the lowest energies of this domain in perturbation theory and comparing them with the numerical values obtained using a collocation approach ) .
the domain corresponding to @xmath198 is shown in fig.[figure4 ] .
.,width=151 ] [ figure4 ] the map is defined to allow @xmath199 ; the conformal density in this case is ( x , y ) = and its matrix elements in the basis of the square are n_x , n_y | | n_x , n_y & = & \ { lll 1 - & , & n_x = n_x ; n_y = n_y + & , & n_x = n_x ; n_y n_y + & , & n_x n_x ; n_y = n_y + 0 & , & n_x n_x ; n_y n_y + . using the diagonal matrix elements above
, one finds in this case g & = & ( ( ) + -+ ) - . in table
[ table-1 ] we report the values of the spectral zeta function at @xmath200 for the deformed square for different values of @xmath93 : the exact value , @xmath201 , is compared with the value obtained considering only the diagonal contributions , @xmath202 , and with the values obtained from the first @xmath203 numerical eigenvalues obtained with a collocation approach with a grid with @xmath204 points ( adding the contribution of the higher excited states via the weyl s law ) : z(2)^(num ) _
n=1 ^ 3000 ( ) ^2 + _
n=3001^()^2 , where @xmath205 is the area of the domain and @xmath206 its perimeter ; in the last column we report the value obtained only using weyl s law : z(2)^(weyl ) _
n=1^()^2 . notice that @xmath202 accounts almost completely for the value of @xmath201 , up to @xmath207 .
.spectral zeta function at @xmath200 for the deformed square of area @xmath205 for different values of @xmath93 .
[ cols="^,^,^,^,^",options="header " , ] notice that for @xmath208 the off - diagonal contributions account only for about @xmath209 of the value of @xmath201 , confirming the observation made in ref.@xcite ; the value obtained using weyl s law on the other hand largely overestimates @xmath201 , since this approximation fails to describe the low energy part of the spectrum .
it is interesting to obtain the quantity z(1 ) _
s1^+ , which is finite .
we have z(1 ) & = & _ n_x=1^ + _
n_x=1^ _ n_y=1^n_x , ny |
| n_x , ny ^2
+ & = & + _ n_x=1^ + & = & + & + & ( 1-r^2 ) _ j=0^ , where @xmath210 is the incomplete beta function .
this expression is exact and holds for all values @xmath211 , @xmath212 : in particular , around @xmath213 we have z(1 ) & & ( 1-r)^2
. we will now discuss the analytic continuation of the spectral zeta function of the annulus , in the limit @xmath214 ( @xmath215 ) . in this limit ,
it is possible to approximate the conformal density as @xmath216 and @xmath217 working to this order we have that @xmath218 and the spectral zeta function of thin annulus may be approximated as @xmath219 an analogous expression can be obtained for the case of a thin annulus with a trasversal cut , which corresponds to choosing dirichlet boundary conditions along the two directions . in this case
n_x , n_y^(dd ) = + , where the superscript ( dd ) indicates dirichlet - dirichlet boundary condintions , and @xmath176 .
the spectral zeta function in this case reads @xmath220 the analytic continuation of these expressions is discussed explicitly by kirsten in ref .
@xcite : following @xcite we define _ ( l_2,l_3,s ) _ _ 2=1^__3=1^ ( ( ) ^2 + ( ) ^2)^-s .
the analytic continuation of this function is given in eq.(4.31 ) of @xcite : _
( l_2,l_3,s ) & = & - ( ) ^2s ( 2s ) + ( ) ^2s-1 ( 2s-1 ) + & + & _ _ 2=1^__3=1^()^s-1/2 k_1/2-s ( )
. therefore @xmath221 \nonumber \\
z^{({\rm dd})}(s ) & \approx & ( 1 - 2sl ) \
\zeta_{\mathcal{c}}(2\pi,2l , s ) \nonumber \ .\end{aligned}\ ] ] around @xmath75 , in the limit @xmath214 , we have z^(dp)(-1/2 ) z^(dd)(-1/2 ) - + , which provides the transverse electric contribution to the casimir energy of an arbitrarily thin annulus : e_c^(te ) = z^(dp)(-1/2 ) - + it is easy to calculate the transverse magnetic contribution to the casimir energy keeping in mind that @xmath222 and @xmath223 .
e_c^(tm ) = z^(np)(-1/2 ) - +
thus the electromagnetic casimir energy of a thin annulus is e_c - + gosdzinsky and romeo@xcite have related the spectral zeta function of a circle to the spectral zeta function of a cylinder of circular section ; nesterenko and pirozhenko @xcite have also used this formula to obtain the casimir energy of a circle using the results for a cylinder . using our conventions , the relation between the two spectral zeta functions , eq.(3 ) of @xcite and eq.(6 ) of @xcite ) can be cast in the form z_cyl(s ) = b ( , s- ) z_circle(s- ) .
[ cylinder ] although the authors of refs .
@xcite have applied this equation to the cylinder of circular section , the formula is general and it applies to cylinders of arbitrary section : therefore we can obtain the spectral zeta functions for these cases , working in two dimensions with a domain corresponding to the transversal section of the cylinder and then using eq.([cylinder ] ) . we write eq.([cylinder ] ) in the general form as z_cyl(s ) = b ( , s- ) z(s- ) .
[ cylinder2 ] we can apply eq.([cylinder2 ] ) to calculate the casimir energy of an infinite cylinder with annular section , in the limit where the two radii are very close ( @xmath224 , @xmath225 ) .
the spectral zeta function of the two dimensional domain has been already calculated to leading order in the previous example , and thus the calculation only requires the straightforward application of eq.([cylinder2 ] ) : z^(dp)_cyl(- ) z^(dd)_cyl(- ) - .
the total electromagnetic casimir energy will therefore be e_c = = - .
the casimir energy for concentric cylinders has been calculated by mazzitelli and collaborators in @xcite : our result reproduces eq.(61 ) of that paper . ] .
this provides a useful check of the correctness of our approach .
the same approach can also be applied to the calculation of the casimir energy of cylinders and annular cylinders with slightly nonuniform density .
in this paper we have discussed a novel approach to the calculation of the spectral zeta function and of the heat kernel associated with the eigenvalues of a slighlty inhomogeneous system in a @xmath1-dimensional cube within perturbation theory .
the method that we have devised relies on the standard rayleigh - schrdinger perturbation theory for hermitian operators and provides an explicit expression for the spectral zeta function associated to these operators that converges for @xmath44 . to the best of our knowledge ,
this is the first calculation where the spectral zeta function of a system with arbitrary density perturbation has been obtained . the analytic continuation of this perturbative series allows one to obtain an approximation to the casimir energy of the system under consideration .
we have illustrated our method with some examples , evaluating the casimir energy of different systems in one , two and three dimensions .
in particular , for a string with piecewise constant density and of two perfectly conduting , concentric cylinders of similar radius we have reproduced results already published @xcite .
this provides a useful check of our approach .
an interesting outcome of our analysis is the appearance of irremovable divergences already to first order perturbation theory in the casimir energy of one dimensional inhomogeneous systems , when either dirichlet - dirichlet or neumann - neumann boundary conditions are applied at the ends of the string , unless the density @xmath226 of the string is such that @xmath227 .
the physical interpretation of these singularities is not clear . on the other hand
, we also have found that the sum of dirichlet - dirichlet and neumann - neumann zeta functions is free of divergences , to first order . for positive integer values of @xmath12 , with @xmath44 ,
our formalism provides exact ( non - perturbative ) sum rules that can be useful to test numerical and analytic approximations to the eigenvalues of the problem .
the calculations presented in this paper have been mostly performed to first order in perturbation theory : the analytic continuation of the higher order expressions , which involve multiple series , is a non trivial task for general densities ( in the paper we have discussed a one dimensional problem where the second order calculation is easily obtained ) .
we are currently studying a non - perturbative extension of the present approach that may allow to describe a larger class of problems .
a. voros , nucl .
phys.*b * 165 , 209 ( 1980 ) a. voros , j.phys.*a * 32 , 1301 - 1311 ( 1999 ) f. steiner , phys.lett.*b * 159 , 397 - 402 ( 1985 ) m.v.berry , j.phys.*a * 19 , 2281 - 2296 ( 1986 ) c. itzykson , p. moussa and j.m.luck , j.phys.*a * 19 , l111-l115 ( 1986 ) f. steiner , fortschr .
phys.*35 * , 87 - 114 ( 1987 ) e.elizalde , s.leseduarte and a.romeo , j.phys.*a * 26 , 2409 - 2419 ( 1993 ) f. steiner , phys.lett.*b * 188 , 447 - 454 ( 1987 ) w. arendt , r. nittka , w. peter and f.steiner , `` weyl s law : spectral properties of the laplacian in mathematics and physics '' , in `` mathematical analysis of evolution , information and complexity '' , edited by w. arendt et al .
, wiley - vch ( 2009 ) , 1 - 71 r.e .
crandall , j.phys.*a * 29 , 6795 - 6816 ( 1996 ) j.m.borwein , d.m .
bradley and r.e .
crandall , j.comp .
appl . math .
* 121 * , 247 - 296 ( 2000 ) a.a .
kvitsinsky , j.phys.*a * 29 , 6379 - 6393 ( 1996 ) g.a .
mezincescu , j.phys.*a * 33 , 4911 ( 2000 ) c.m.bender and q.wang , j.phys.*a * 34 , 3325 ( 2001 ) s. hawking , comm . math.phys . *
55 * , 133 - 148 ( 1977 ) j.s .
dowker and r. critchley , phys.rev.*d * 13 , 3224 - 3232 ( 1976 ) e. elizalde , s. d. odintsov , a. romeo , a. a. bytsenko , and s. zerbini , zeta regularization techniques with applications ( world scientific , singapore , 1994 ) .
k. kirsten , spectral functions in mathematics and physics ( chapman & hall , london , 2001 ) .
e. elizalde , ten physical applications of spectral zeta functions , lecture notes in physics , 2nd edition , springer - verlag , berlin ( 2012 ) e. elizalde , j.phys.*a * 22 , 931 ( 1989 ) m. bordag , e. elizalde and k. kirsten , j.math.phys .
* 37 * , 895 ( 1996 ) e. elizalde , m. bordag and k.kirsten , j.phys.*a * 31 , 1743 ( 1998 ) e. elizalde , j.phys.*a * 39 , 6299 ( 2006 ) l. hadasz , g. lambiase and v. v. nesterenko , phys
d * 62 * , 025011 ( 2000 ) i. brevik and h.b .
nielsen , phys.rev.*d * 41 , 1185 ( 1990 ) x. li , x. shi and j.zhang , phys.rev.*d * 44 , 560 ( 1991 ) i. brevik and e. elizalde , phys.rev.*d * 49 , 5319 ( 1994 ) i. brevik , h.b .
nielsen and s.d.odintsov , phys.rev.*d * 53 , 3224 ( 1996 ) p. amore , annals of physics * 326 * , 2315 - 2355 ( 2011 ) p. amore , j. math . phys . * 51 * , 052105 ( 2010 ) s. sen , phys . rev .
* d * 24 , 869 ( 1981 ) .
s. sen , j. math .
22 , 2968 ( 1981 ) . c. m. bender and k. a. milton , phys
d * 50 , 6547 ( 1994 ) .
bender and s.a .
orszag , advanced mathematical methods for scientists and engineers , mcgraw - hill ( 1978 ) g. borg , acta mathematica * 78 * , 1 - 96 ( 1946 ) c.castro and e.zuazua , siam journal on applied mathematics * 60 * , 1205 - 1233 ( 2000 ) m. avellaneda , c. bardos and j. rauch , asymptotic analysis 5 , 481 - 494 ( 1992 ) c. castro and e.zuazua , european journal of applied mathematics * 11 * , 595 - 622 ( 2000 ) r.m.ziff , j.phys.*a * 19 , 3923 - 3924 ( 1986 ) c. alvarado and p. amore , j. math
. phys . * 52 * , 063516 ( 2011 ) k. kirsten , a window into zeta and modular physics , msri publications * 57 * ( 2010 ) p. gosdzinsky and a. romeo , phys . lett .
* b * 441 , 265 ( 1998 ) v.v.nesterenko and g.pirozhenko , j. math . phys.*41 * , 4521 ( 2000 ) f.d .
mazzitelli , m.j.snchez , n.n .
scoccola and j. von stecher , phys.rev.*a * 67 , 013807 ( 2003 ) b. derjaguin , kolloid z. * 69 * , 155 ( 1934 ) c.d .
fosco , f.c .
lombardo and f.d .
mazzitelli , phys . rev . * d * 84 , 105031 ( 2011 ) g. bimonte , t. emig , r. l. jaffe and m. kardar , epl * 97 * , 50001 ( 2012 ) | we show that the spectral zeta functions of inhomogeneous strings and drums can be calculated using rayleigh - schrdinger perturbation theory .
the inhomogeneities that can be treated with this method are small but otherwise arbitrary and include the previously studied case of a piecewise constant density . in two dimensions
the method can be used to derive the spectral zeta function of a domain obtained from the small deformation of a square .
we also obtain exact sum rules that are valid for arbitrary densities and that correspond to the values taken by the spectral zeta function at integer positive values ; we have tested numerically these sum rules in specific examples .
we show that the dirichlet or neumann casimir energies of an inhomogeneous string , evaluated to first order in perturbation theory , contain in some cases an irremovable divergence , but that the combination of the two is always free of divergences .
finally , our calculation of the casimir energies of a string with piecewise constant density and of two perfectly conduting concentric cylinders , of similar radius , reproduce the results previously published . |
knowledge of the stellar mass in galaxies is important for a wide range of science problems . the best way to determine
the stellar mass is through examination of the stellar populations , from either a star - by - star census or spectral energy distribution ( sed ) fitting of luminosities and colors based on stellar population synthesis ( sps ) models .
however , frequently , such data are not available , and we turn instead to a mass - to - light ratio , @xmath1 , coupled with information on the luminosity or surface brightness , to derive a stellar mass or mass density .
many studies have explored various calibrations of @xmath1 to color relations ( mlcrs ) for johnson - cousins optical bands ( bell & de jong 2001 ; portinari et al.2004 , hereafter p+04 ; mcgaugh & schombert 2014 , hereafter ms14 ) , for sloan digital sky survey ( sdss ) optical filters @xcite , and even some for both sets of filters ( bell et al.2003 , hereafter b+03 ; gallazzi & bell 2009 ; zibetti et al.2009 , hereafter z+09 ; into & portinari 2013 , hereafter ip13 ) . in particular , ms14 tested the self - consistency of mlcrs by applying relations from various studies ( b+03 , p+04 , z+09 , ip13 ) and various bands ( specifically @xmath1 in @xmath3 , @xmath4 , @xmath5 , and 3.6 @xmath6 m all as functions of @xmath7 ) to estimate masses for a sample of disk galaxies spanning over 10 mag in luminosity .
they found reasonable agreement between the four studies in @xmath3 , but determined revised mlcrs for @xmath4 , @xmath5 , and 3.6 @xmath6 m to improve self - consistency . for a large sample of galaxies from the two micron all sky survey ( 2mass ) and the sdss ,
b+03 used models to construct linear fits of @xmath8 and sdss colors as well as @xmath7 and @xmath9 .
the fits take the form : @xmath10 from there , one can go on , for example , to determine the stellar mass density profile , @xmath11 , from the surface brightness , @xmath6 , and color profiles using the following from @xcite : @xmath12 where @xmath13 is the absolute magnitude of the sun at wavelength @xmath14 and @xmath15 is measured in @xmath16 pc@xmath17 . however , a reliable relation between some color and the @xmath1 is essential , and the mlcrs in the literature have largely been determined from models appropriate for spirals .
it is questionable if any of these linear fits is suitable for dwarf galaxies , with lower metallicities and potentially different star formation histories ( sfhs ) than spirals .
therefore , we have determined relations between @xmath1 and colors for dwarf irregular ( dirr ) galaxies .
these relations , presented here , are based on empirical sed fitting of multi - wavelength surface photometry of a sample of 34 dirrs @xcite . in [ sec - data ]
we describe the data on which our new mlcrs are based . in [ sec - formula ] we present our mlcrs between stellar @xmath1 from the sed fitting and @xmath18 colors and compare them to several mlcrs from the literature .
we explore metallicity effects in [ sec - zbins ] by breaking our data into four metallicity bins .
the sample of galaxies is taken from little things ( local irregulars that trace luminosity extremes , the h@xmath19i nearby galaxy survey , * ? ? ?
this is a multi - wavelength survey of 41 nearby ( @xmath20 mpc ) dirr galaxies and blue compact dwarfs , which builds on the things project @xcite , whose emphasis was on nearby spirals .
the little things galaxies were chosen to be gas - rich so they have the potential to form stars , although a few do not currently have h@xmath21 emission . they were also chosen to be fairly isolated , or at least not companions to a giant galaxy or obviously interacting with another system .
the little things data set includes _ galaxy evolution explorer _ satellite ( _ galex _ ; * ? ? ?
* ) images at fuv ( 1516 ) and nuv ( 2267 ) wavelengths , @xmath0 and some @xmath22 images from @xcite , h@xmath21 images from @xcite , and _
infrared array camera ( * ? ? ?
* irac ) 3.6 @xmath6 m images .
this yields 7 - 10 passbands for each galaxy from the fuv to the nir .
@xcite used azimuthally averaged surface photometry in these passbands and performed a sed analysis of each annulus for a subsample of 34 of the little things dwarfs .
they modeled the data with a library of four million different sfhs based on the unpublished 2007 version of galaxev ( bruzual & charlot 2003 ) stellar population models and allowing dust extinction , metallicity , and relative star formation rate among different age bins to vary uniformly among physically reasonable ranges .
they used the stellar initial mass function ( imf ) of @xcite and took into account h@xmath21 line emission , but not contamination from polycyclic aromatic hydrocarbon lines since nebular emission is arguably not important for the normal star - forming dirr galaxies in the optical broadband images that were used . from the fits to the data , they derived stellar mass surface density distributions for the 34 dirr galaxies .
@xcite used two slightly different methods to determine the mass profile values at outer radii : ( 1 ) applying the @xmath1 from the most distant point where the sed analysis is applicable and ( 2 ) using the 3.6 @xmath6 m light for some of the outermost points .
we used the latter mass profiles for the 34 dwarfs in our averaging analysis .
the sed fitting analysis yielded robust mass profiles as a function of radius in a large sample of dwarf galaxies . @xcite and @xcite have determined surface brightness and color profiles for these same galaxies , as well as 107 other dwarfs .
with these data , we have the information we need to determine @xmath1 trends with @xmath23 and @xmath7 colors , as well as the transformed color @xmath24 ( using @xmath25 , @xmath26 , and @xmath27 from smith et al.2002 ) , that are appropriate specifically for dirrs .
cccccccccccccc [ tab - fits ] @xmath23 & @xmath28 & @xmath29 & 0.081 & 1.147 & 0.260 & & & & & & & &
+ @xmath23 & @xmath30 & @xmath31 & 0.081 & 0.747 & 0.260 & & & & & & & & + @xmath7 & @xmath30 & @xmath32 & 0.157 & 1.452 & 0.393 & @xmath33 & 1.737 & @xmath34 & 1.690 & @xmath35 & 2.237 & @xmath36 & 2.027 + @xmath7 & @xmath3 & @xmath37 & 0.157 & 1.053 & 0.393 & @xmath38 & 1.305 & @xmath39 & 1.290 & @xmath40 & 1.837 & @xmath41 & 1.627 + @xmath24 & @xmath42 & @xmath43 & 0.090 & 1.294 & 0.401 & @xmath44 & 1.519 & & & @xmath45 & 2.053 & @xmath46 & 1.930 + @xmath24 & @xmath47 & @xmath48 & 0.090 & 0.894 & 0.401 & @xmath49 & 1.097 & & & @xmath50 & 1.654 & @xmath51 & 1.530 + @xmath52 m & 3.6@xmath6 m & @xmath53 & 0.380 & @xmath54 & 0.180 & & & & & & & &
+ @xmath7 & 3.6@xmath6 m & @xmath55 & 0.148 & 0.776 & 0.367 & @xmath56 & @xmath57 & @xmath58 & 0.467 & @xmath59 & 1.289 & @xmath60 & 0.849
the relations between @xmath1 and colors were determined via a minimizing @xmath61 fitting routine @xcite iteratively reweighted to obtain a linear fit that was characteristic of as much of the data as possible , but not overly affected by outlying influential points .
the weights were 1.0 for @xmath62 where @xmath63 was the residual and @xmath64 was the weighted standard deviation , both from each previous fit , and @xmath65 for more discrepant data , following an l1 procedure for non - gaussian residuals .
the @xmath61 fitting routine was iterated until the slope changed by @xmath66
. figures [ m_l ] and [ new ] show the optical and 3.6 @xmath6 m data and fits , respectively , as well as the parameters of the final fits , which are also listed in columns 3 and 5 of table [ tab - fits ] .
stellar mass profiles are best calculated using the @xmath7 or @xmath24 relations ; the @xmath1 ratio is not correlated as strongly with @xmath23 or @xmath67 m as with @xmath7 or @xmath24 , as seen by the larger scatter in the two @xmath23 panels of figure [ m_l ] and the @xmath67 m panel of figure [ new ] .
this makes sense because redder passbands generally trace the mass better than bluer passbands .
fuv@xmath68nuv colors would not have been useful at all for estimating @xmath1 , though @xmath1 in 3.6 @xmath6 m is not as strongly correlated with @xmath7 as the @xmath1 in @xmath30 or @xmath3 . since our relations
( and those of z+09 ) are based on a @xcite stellar imf , we have applied zero - point offsets to raise or lower mlcrs determined using other imfs for a better comparison due to having one fewer variable between the different studies .
@xcite ( and b+03 ) used a `` diet '' salpeter imf from 0.1 to 125 m_@xmath69 by modifying the @xmath1 by a factor of 0.7 .
this global factor was equivalent to reducing the contribution from low - mass stars , and was done to bring the absolute normalization of their @xmath1 in line with observational constraints of maximum disks determined from rotation curves of spiral galaxies .
b+03 noted that @xmath8 must be reduced by 0.15 dex to convert from their `` diet '' salpeter imf to a @xcite imf . @xcite specified a similar correction ( -0.093 dex ) to convert from the `` diet '' salpeter imf to a @xcite imf , which we applied to the mlcr of b+03 . since p+04 and ip13 both used kroupa imfs , we added 0.057 dex to adjust to a chabrier imf . to convert to a @xcite imf ,
add a constant of + 0.243 to the right side of equation 1 .
for a @xcite imf , with a shallower imf than salpeter for low mass stars , subtract 0.057 dex .
with all the literature mlcrs adjusted ( as needed ) to a chabrier imf , the b+03 optical relations almost overlap our linear fits and the data , except our optical fits for the mlcrs of dirrs are consistently slightly shallower .
actually , each cyan line ( the dwarf best fit ) in figure [ m_l ] has the _ shallowest _ slope of all the colored lines in each panel , and correspondingly column 5 ( dwarf @xmath70 slopes ) of table 1 contains the smallest slope in each of the first six rows . the z+09 mlcrs provide a poor fit to our dirr galaxy data for all five bands in the comparison .
the p+04 mlcrs are slightly high for all three @xmath7 panels whereas the ip13 mlcrs fit the dwarf data in the @xmath7 panels fairly well ( except for being too steep ) but fall below the dwarf data in both transformed @xmath24 panels .
the dwarf fit values for @xmath71 as a function of @xmath24 ( @xmath72 and @xmath73 ) are similar to those used by @xcite for their spiral analysis ( @xmath74 and @xmath75 ) although the latter @xmath76 becomes @xmath49 after being adjusted from the `` diet '' salpeter imf to a chabrier imf .
the magenta asterisks in the lower central ( @xmath7 , @xmath3 ) panel of figure [ m_l ] show @xmath77 for a single starburst population of 0.1 - 100 @xmath16 stars @xcite reduced by 0.243 dex to adjust the data from their original salpeter imf to a chabrier imf .
interestingly , the dwarf data fall somewhere between the 1 gyr ( @xmath789 ) and 10 gyr ( @xmath7810 ) points since the typical luminosity - weighted age of dirrs is @xmath791 gyr whereas spirals are generally older than dwarfs in terms of their luminosity - weighted ages .
figure [ fit_params ] displays the fit parameters ( @xmath80-intercepts , @xmath81 , and slopes , @xmath70 ) of the mlcrs from this work as well as those from b+03 , p+04 , z+09 , and ip13 .
the left panel shows the fit parameters from each study with the b+03 data based on the `` diet '' salpeter imf , the p+04 and ip13 data based on a kroupa imf , and the data from this work and z+09 based on a chabrier imf .
the non - dwarf @xmath3 , @xmath4 , and @xmath5 data are directly from table 2 of ms14 .
the right panel shows the @xmath82 @xmath80-intercepts adjusted as needed to a chabrier imf ( the cyan and green data points are unchanged ) and the @xmath4 and 3.6 @xmath6 m fits from table 7 of ms14 , revised to be self - consistent with the @xmath3 band masses and corrected for @xmath83 as a second color term in addition to a 3.6 @xmath6 m data point for dwarfs .
the @xmath84 and 3.6 @xmath6 m fits are all with respect to @xmath7 ; that is , they are of the form : @xmath85 whereas the @xmath86 fits are with respect to @xmath24 . in this work
only , all @xmath86 data have been transformed from @xmath87 images . for completeness
, we have included @xmath88 mlcrs from the literature in figure [ fit_params ] even though the little things survey does not have enough observations in these bands for us to determine @xmath88 mlcrs for dwarfs .
the optical bands ( but not the @xmath5 or 3.6 @xmath6 m data ) in both panels display a fairly smooth progression from steep slopes and most negative @xmath80-intercepts in @xmath30 ( triangles , lower right ) up to more shallow slopes and less negative @xmath80-intercepts in @xmath47 ( circles ) and @xmath4 ( open diamonds , both in the upper middle ) .
however , in the right panel , the data points are more streamlined and the @xmath4-band points are all located in the upper middle instead of paralleling the @xmath30 to @xmath47 points in the left panel . furthermore , within each optical band , the shape obtained from connecting the data from lower right to upper left follows roughly the same pattern from z+09 ( green ) , ip13 ( lavender ) , b+03 ( pink ) , this work ( cyan ) , to p+04 ( gold ) .
the differences between the various studies probably occur due to the variety of sample galaxy sfhs or assumed sfhs since sed masses and luminosities depend strongly on sfhs .
why do the coefficients @xmath81 , the @xmath80-intercepts , and @xmath70 , the slopes , vary systematically with each other and with wavelength ?
the fact that @xmath81 decreases ( the @xmath1 gets smaller ) when @xmath70 increases ( the color dependence gets stronger ) for bluer passbands means that younger regions with small @xmath1s have a greater dependence between @xmath1 and @xmath7 color than older regions .
this must occur since all the mlcr lines in figure 1 have similar @xmath1 at the reddest colors .
that is , if we think of @xmath1 versus @xmath7 as pinned at a particular @xmath1 for a particular large @xmath7 , and all the other curves varying as straight lines around that pinning point , then naturally the @xmath80-intercept @xmath81 will get smaller as the slope @xmath70 gets larger .
physically this means that for red colors , the imf and sfhs do not matter much because only old red stars remain and the @xmath1 and colors are universal for very old populations .
the variety of sed models and galaxy fits is generally due to different histories or assumptions about _ young _ stars .
since our mlcrs for dwarfs are shallower than the mlcrs in the literature and a lower metallicity is one characteristic that separates dwarfs from spirals , we broke our dwarf data into four bins based on oxygen abundance ( @xmath89 , @xmath90 , @xmath91 , and @xmath92 ) and fit the mlcr for each bin .
figure [ z_params ] shows the results , including an apparent steepening of the mlcrs with higher metallicity ( see table [ tab - z ] for the fit parameters ) but little metallicity dependence of the @xmath80-intercepts , especially in @xmath42 and @xmath47 . since the @xmath42 and @xmath47 analyses are not based on images but instead on transformations of @xmath30 and @xmath3 , it is not surprising that the @xmath42 and @xmath47 fits have similar slopes to the @xmath30 and @xmath3 fits .
unfortunately , the metallicities of the samples of spiral galaxies explored to determine existing mlcrs are not available for comparison .
however , for our slope trends from the left panel of figure [ z_params ] to match the b+03 mlcr slopes , the metallicity for the spiral sample would need to be somewhere between 8.5 and 8.8 , which might be reasonable for a galaxy - wide average considering that the solar value is 8.69 @xcite . perhaps the shallower slope with lower metallicity is caused by less line blanketing occurring at lower metallicity @xcite .
the color changes with changing line blanketing ( bluer for less line blanketing ) _
faster _ than the brightness ( brighter for less line blanketing ) of a population .
cccccc [ tab - z ] @xmath7 & @xmath30 & 0.258 & @xmath93 & 0.004 & 0.026 + @xmath7 & @xmath3 & 0.258 & @xmath94 & 0.004 & 0.026 + @xmath24 & @xmath42 & 0.264 & @xmath95 & 0.004 & 0.029 + @xmath24 & @xmath47 & 0.263 & @xmath96 & 0.004 & 0.028
we have used the stellar mass density radial profiles of @xcite produced from sed fitting to 7 - 10 passbands from the fuv to the nir and the surface brightness and color profiles of @xcite to determine relations between @xmath97 and @xmath23 and @xmath7 colors as well as the transformed color @xmath24 given by @xcite .
these mlcrs are specifically for dirr type galaxies .
the @xmath7 relationship in particular can be used to determine stellar masses in dirrs in situations where sed fitting of multi - wavelength photometry or a census of individual stars is not available .
our mlcrs are consistently shallower than those reported in the literature and the differences could be due at least partly to a steepening of mlcrs with metallicity , perhaps from more line blanketing occurring at higher metallicity .
we would like to thank the anonymous reviewer for helpful suggestions .
kah acknowledges funding from pennsylvania state - mont alto that enabled extended visits to lowell observatory where some of this work was carried out .
dah also appreciates funding from the lowell research fund and is grateful to john and meg menke for funding for page charges .
hz acknowledges support from fondecyt postdoctoral fellowship project no .
3160538 and earlier from the chinese academy of sciences ( cas ) through a cas - conicyt postdoctoral fellowship administered by the cas south america centre for astronomy ( cassaca ) in santiago , chile .
asplund , m. , grevesse , n. , sauval , a.j . , & scott , p. 2009 ,
ara&a , 47 , 481 bakos , j. , trujillo , i. , & pohlen , m. 2008 , , 683 , l103 bell , e.f .
, & de jong , r.s . 2001 , apj , 550 , 212 bell , e.f . , mcintosh , d.h . , katz , n. , & weinberg , m.d .
2003 , , 149 , 289 ( b+03 ) bruzual , g. , & charlot , s. 2003 , , 344 , 1000 chabrier , g. 2003 , , 586 , 133 fazio , g.g . , hora , j.l . , allen , l.e . ,
2004 , , 154 , 10 gallazzi , a. , & bell , e.f .
2009 , , 185 , 253 gallazzi , a. , brinchmann , j. , charlot , s. , & white , s.d.m .
2008 , , 383 , 1439 hayden , m.r . ,
bovy , j. , holtzman , j.a . , et al .
2015 , , 808 , 132 herrmann , k.a . , hunter , d.a . , & elmegreen , b.g .
2013 , , 146 , 104 herrmann , k.a . , hunter , d.a . , & elmegreen , b.g .
2016 , , 151 , 145 hunter , d. a. , & elmegreen , b.g .
2004 , , 128 , 2170 hunter , d. a. , & elmegreen , b.g .
2006 , apjs,162 , 49 hunter , d.a .
, ficut - vicas , d. , ashley , t. , et al . 2012 , , 144 , 134 into , t. , & portinari , l. 2013 , , 430 , 2715 ( ip13 ) kroupa , p. 2001
, , 322 , 231 martin , d.c . , fansom , j. , schiminovich , d. , et al .
2005 , , 619 , l1 mcgaugh s.s . , &
schombert j.m .
2014 , , 148 , 77 ( ms14 ) portinari , l. , sommer - larsen , j. , & tantalo , r. 2004 , , 347 , 691 ( p+04 ) press , w.h . ,
teukolsky , s.a . , vetterling , w.t . , & flannery , b.p.2002 , numerical recipes in c++ : the art of scientific computing , second edition , ( new york , ny : cambridge university press ) roediger , j.c . , & courteau , s. 2015 , , 452 , 3209 salpeter , e.e .
1955 , apj , 121 , 161 smith , j.a . ,
tucker , d.l . , kent , s. , et al .
2002 , , 123 , 2121 taylor , e.n . ,
hopkins , a.m. , baldry , i.k .
, et al . 2004 , , 418 , 1587 walter , f. , brinks , e. , de blok , w.j.g . , et al . 2008 , , 136 , 2563 zhang , h .- x . , hunter , d.a . ,
elmegreen , b.g .
, gao , y. , & schruba , a. 2012 , , 143 , 47 zibetti s. , charlot s. , & rix h .- w .
2009 , , 400 , 1181 ( z+09 ) | we have determined new relations between @xmath0 colors and mass - to - light ratios ( @xmath1 ) for dwarf irregular ( dirr ) galaxies , as well as for transformed @xmath2 .
these @xmath1 to color relations ( mlcrs ) are based on stellar mass density profiles determined for 34 little things dwarfs from spectral energy distribution fitting to multi - wavelength surface photometry in passbands from the fuv to the nir .
these relations can be used to determine stellar masses in dirr galaxies for situations where other determinations of stellar mass are not possible .
our mlcrs are shallower than comparable mlcrs in the literature determined for spiral galaxies .
we divided our dwarf data into four metallicity bins and found indications of a steepening of the mlcr with increased oxygen abundance , perhaps due to more line blanketing occurring at higher metallicity . |
in this study , we consider the generalized @xmath0-component reaction diffusion system with @xmath2:@xmath3 in @xmath4 , where @xmath5 is an open bounded domain of class @xmath6 in @xmath7 with boundary @xmath8 .
the diffusion matrix @xmath9 is assumed to be of the form@xmath10{cccccc}\alpha_{1 } & \gamma_{1 } & 0 & \cdots & \cdots & 0\\ \beta_{1 } & \alpha_{2 } & \gamma_{2 } & \ddots & & \vdots \\ 0 & \beta_{2 } & \alpha_{1 } & \gamma_{1 } & \ddots & \vdots \\ \vdots & \ddots & \beta_{1 } & \alpha_{2 } & \gamma_{2 } & 0\\ \vdots & & \ddots & \beta_{2 } & \ddots & \ddots \\ 0 & \cdots & \cdots & 0 & \ddots & \ddots \end{array } \right ) _ { m\times m } , \label{trimat1}\ ] ] with @xmath11 being positive real numbers representing the self and cross diffusion constants and satisfying the inequality@xmath12 the laplacian operator @xmath13 has a spatial dimension of @xmath14 and @xmath15 is a polynomially growing functional representing the reaction terms of the system .
the boundary conditions and initial data for the proposed system are assumed to satisfy@xmath16 or@xmath17 and@xmath18 respectively . for generality , we will consider three types of boundary conditions in this paper : 1 .
nonhomogeneous robin boundary conditions , corresponding to@xmath19 2 .
homogeneous neumann boundary conditions , corresponding to@xmath20 3 .
homogeneous dirichlet boundary conditions , corresponding to@xmath21 note that @xmath22 denotes the outward normal derivative on @xmath8 and the vectors @xmath23 , @xmath24 , and @xmath25 are defined as@xmath26 the initial data is assumed to be in the region given by@xmath27 subject to@xmath28 the study at hand builds upon numerous previous works found in the literature . among the most relevant studies
is that of abdelmalek in @xcite where he considered an @xmath0-component tridiagonal matrix of the form@xmath10{ccccc}\alpha & \gamma & 0 & \cdots & 0\\ \beta & \alpha & \gamma & \ddots & \vdots \\ 0 & \beta & \ddots & \ddots & 0\\ \vdots & \ddots & \ddots & \ddots & \gamma \\ 0 & \cdots & 0 & \beta & \alpha \end{array } \right ) _ { m\times m},\ ] ] and proved the global existence of solutions subject to the parabolicity condition@xmath29 which can be easily shown to fall under the general condition in ( [ para1 ] ) with @xmath30 , @xmath31 , and @xmath32 .
another important study is that of kouachi and rebiai in @xcite where the authors established the global existence of solutions for a @xmath33 tridiagonal @xmath1toeplitz matrix of the form@xmath10{ccc}\alpha_{1 } & \gamma_{1 } & 0\\ \beta_{1 } & \alpha_{2 } & \gamma_{2}\\ 0 & \beta_{2 } & \alpha_{1}\end{array } \right ) , \ ] ] subject to the parabolicity condition@xmath34 note that this condition is weaker than@xmath35 which is obtained from ( [ para1 ] ) for @xmath36 .
although the work carried out in @xcite is important to us here , it is necessary to note that the authors failed to identify all the invariant regions of the proposed system and settled for only @xmath37 of them .
this paper will build upon the work of these two studies by assuming the diffusion matrix to be @xmath0component tridiagonal @xmath1toeplitz and determining all the possible invariant regions for the system .
a lyapunov functional will be used to establish the global existence of solutions in these regions .
the remainder of this paper is organized as follows : section [ seceigen ] uses the three point chebyshev recurrence relationhip of polynomials to derive the eigenvalues and eigenvectors of the transposed dffusion matrix for the odd and even dimension cases , respectively .
section [ secpara ] derives the parabolicity conditions for the proposed system , which is essential for the diagonalization process , which follows in section [ secmain ] .
section [ secmain ] shows how the invariant regions of the equivalent digonalized system can be identified and proves the local and global existence of solutions .
the last section of this paper will present a confirmation and validation of the findings through the use of numerical examples solved by means of the finite difference approximation method .
for reasons that will become apparent in the following section , we will first derive the eigenvalues and eigenvectors of matrix @xmath38 with @xmath9 being the proposed tridiagonal @xmath1toeplitz diffusion matrix .
we refer to the work of gover in @xcite where the characteristic polynomial of a tridiagonal @xmath1toeplitz matrix was shown to be closely connected to polynomials that satisfy the three point chebyshev recurrence relationship .
first , we have@xmath39{cccccc}\alpha_{1 } & \beta_{1 } & 0 & \cdots & \cdots & 0\\ \gamma_{1 } & \alpha_{2 } & \beta_{2 } & \ddots & & \vdots \\ 0 & \gamma_{2 } & \alpha_{1 } & \beta_{1 } & \ddots & \vdots \\ \vdots & \ddots & \gamma_{1 } & \alpha_{2 } & \beta_{2 } & 0\\ \vdots & & \ddots & \gamma_{2 } & \ddots & \ddots \\ 0 & \cdots & \cdots & 0 & \ddots & \ddots \end{array } \right ) _ { m\times m}. \label{2.0}\ ] ] the exact shape and characteristics of @xmath38 differ for odd and even values of the dimension @xmath0 .
hence , we will consider the two cases separately .
before we present the main findings of @xcite , let us define the constants@xmath40 we also define the polynomials@xmath41{l}q_{0}\left ( \mu \right ) = 1,\text { } q_{1}\left ( \mu \right ) = \mu+\beta \\ q_{n+1}\left ( \mu \right ) = \mu q_{n}\left ( \mu \right ) -q_{n-1}\left ( \mu \right ) , \end{array } \right .
\label{2.1}\ ] ] and@xmath41{l}p_{0}\left ( \mu \right ) = 1,\text { } p_{1}\left ( \mu \right ) = \mu \\ p_{n+1}\left ( \mu \right ) = \mu p_{n}\left ( \mu \right ) -p_{n-1}\left ( \mu \right ) , \end{array } \right .
\label{2.2}\ ] ] whose zeros are denoted by @xmath42 and @xmath43 , respectively , for @xmath44 .
we note that @xmath45 is a chebyshev polynomial of the first kind , whereas @xmath46 is not .
as shown in @xcite , the zeros of @xmath45 can be given by @xmath47 whereas for @xmath42 no explicit form was found .
let us now summarize the eigenvalues and eigenvectors for the odd and even cases separately .
first , for @xmath48 , we obtain the following results : [ oddeigval]the eigenvalues of the matrix @xmath38 of order @xmath48 given in ( [ 2.0 ] ) are @xmath49 along with the solutions of the quadratic equations@xmath50 for @xmath51 .
note that for every @xmath43 there exist two eigenvalues for matrix @xmath38 , which along with @xmath49 yields @xmath48 eigenvalues . for notational purposes ,
let us define a duplicated set of zeros given by@xmath52 for @xmath51 .
[ oddeigvec]the eigenvector of the matrix @xmath38 of order @xmath48 given in ( [ 2.0 ] ) associated with the eigenvalue @xmath53 , for @xmath54 , is given by@xmath55 where@xmath56{ll}s^{\frac{\ell-1}{2}}q_{^{\frac{\ell-1}{2}}}\left ( p_{r}^{^{\prime}}\right ) , & \ell \text { is odd}\\ -\frac{1}{\beta_{1}}s^{\frac{\ell}{2}-1}\left ( \alpha_{1}-\lambda_{r}\right ) p_{\frac{\ell}{2}-1}\left ( p_{r}^{^{\prime}}\right ) , & \ell \text { is even,}\end{array } \right .
\label{2.5.1}\ ] ] for @xmath57 .
the eigenvector associated with the eigenvalue @xmath49
is@xmath58 with@xmath59{ll}\left ( -\frac{\gamma_{1}}{\beta_{2}}\right ) ^{\frac{\ell-1}{2 } } , & \ell \text { is odd}\\ 0 , & \ell \text { is even,}\end{array } \right .
\label{2.5.3}\ ] ] for @xmath57 .
the second case is where the matrix @xmath38 ( [ 2.0 ] ) has an even dimension @xmath60 .
the following holds : [ eveneigval]the eigenvalues of the matrix @xmath38 of order @xmath48 given in ( [ 2.0 ] ) denoted by @xmath53 are the solutions of the quadratic equations@xmath61 for @xmath51 , where @xmath42 are the zeros of @xmath46 .
similar to @xmath43 , there exist two eigenvalues for matrix @xmath38 associated with every value of @xmath42 , which yields @xmath0 eigenvalues . in order to simplify the notation , we define the duplicated set of zeros given by@xmath62 for @xmath51 .
[ eveneigvec]the eigenvector of the matrix @xmath38 of order @xmath48 given in ( [ 2.0 ] ) associated with the eigenvalue @xmath53 is given by@xmath63 with@xmath56{ll}s^{\frac{\ell-1}{2}}q_{^{\frac{\ell-1}{2}}}\left ( q_{r}^{^{\prime}}\right ) , & \ell \text { is odd}\\ -\frac{1}{\beta_{1}}s^{\frac{\ell}{2}-1}\left ( \alpha_{1}-\lambda_{r}\right ) p_{\frac{\ell}{2}-1}\left ( q_{r}^{^{\prime}}\right ) , & \ell \text { is even,}\end{array } \right .
\label{2.7.1}\ ] ] for @xmath57 .
in this section , we will derive the parabolicity condition for the proposed system .
parabolicity is crucial to the diagonalization process , which we will be discussed later on in section [ secmain ] . in order to ensure the parabolicity of the system , we examine the positive definiteness of the proposed diffusion matrix . generally speaking ,
a matrix is said to be positive definite if and only if its top - left corner principal minors are all positive . to this end ,
andelic and da fonesca @xcite and others examined the parabolicity condition for a tridiagonal symmetric matrix .
the following theorem holds .
[ andelicprop]let @xmath64 be the tridiagonal matrix defined as@xmath65{ccccc}a_{1 } & b_{1 } & 0 & \cdots & 0\\ b_{1 } & a_{2 } & b_{2 } & & \vdots \\ 0 & b_{2 } & \ddots & \ddots & \vdots \\ \vdots & & \ddots & \ddots & b_{m-1}\\ 0 & \cdots & \cdots & b_{m-1 } & a_{m}\end{array } \right)\ ] ] with positive diagonal entries . if @xmath66 for @xmath67
, then @xmath64 is positive definite .
since the diffusion matrix considered here is not symmetric , proposition [ andelicprop ] does not apply directly to it .
however , we know that if a matrix is not symmetric , its quadratic form @xmath68 , with @xmath69 being an arbitrary column vector , is said to be positive definite if and only if the principal minors in the top left corner of @xmath70 are all positive . in order to derive sufficient conditions for matrix @xmath9 in ( [ trimat1 ] ) , we apply proposition [ andelicprop ] to produce the following theorem . let @xmath9 be the tridiagonal @xmath1toeplitz matrix defined in ( [ trimat1 ] ) .
the quadratic form of @xmath9 is positive definite iff condition ( [ para1 ] ) is satisfied .
it follows that subject to ( [ para1 ] ) , the reaction diffusion system ( [ sys1 ] ) satisfies the parabolicity condition .
condition ( [ andeliccond ] ) can be rearranged to the form@xmath71 the symmteric counterpart of @xmath9 as defined in ( [ trimat1 ] ) can be given by@xmath72{cccccc}\alpha_{1 } & \frac{\beta_{1}+\gamma_{1}}{2 } & 0 & \cdots & \cdots & 0\\ \frac{\beta_{1}+\gamma_{1}}{2 } & \alpha_{2 } & \frac{\beta_{2}+\gamma_{2}}{2 } & \ddots & & \vdots \\ 0 & \frac{\beta_{2}+\gamma_{2}}{2 } & \alpha_{1 } & \frac{\beta_{1}+\gamma_{1}}{2 } & \ddots & \vdots \\ \vdots & \ddots & \frac{\beta_{1}+\gamma_{1}}{2 } & \alpha_{2 } & \frac { \beta_{2}+\gamma_{2}}{2 } & 0\\ \vdots & & \ddots & \frac{\beta_{2}+\gamma_{2}}{2 } & \ddots & \ddots \\ 0 & \cdots & \cdots & 0 & \ddots & \ddots \end{array } \right ) .
\label{para3}\ ] ] now , substituting ( [ para3 ] ) in ( [ para2 ] ) yields the set of @xmath73 conditions@xmath41{ll}\text{for } i=1 : & \sqrt{\alpha_{1}\alpha_{2}}>\left ( \beta_{1}+\gamma _ { 1}\right ) \cos \left ( \frac{\pi}{m+1}\right ) \\ \text{for } i=2 : & \sqrt{\alpha_{1}\alpha_{2}}>\left ( \beta_{2}+\gamma _ { 2}\right ) \cos \left ( \frac{\pi}{m+1}\right ) \\ \text{for } i=3 : & \sqrt{\alpha_{1}\alpha_{2}}>\left ( \beta_{1}+\gamma _ { 1}\right ) \cos \left ( \frac{\pi}{m+1}\right ) \\ \multicolumn{1}{c}{\vdots } & \multicolumn{1}{c}{\vdots}\\ \text{for } i = m-1 : & \left \ { \begin{array } [ c]{ll}\sqrt{\alpha_{1}\alpha_{2}}>\left ( \beta_{2}+\gamma_{2}\right ) \cos \left ( \frac{\pi}{m+1}\right ) , & \text{if } m\text { is odd}\\ \sqrt{\alpha_{1}\alpha_{2}}>\left ( \beta_{1}+\gamma_{1}\right ) \cos \left ( \frac{\pi}{m+1}\right ) , & \text{if } m\text { is even.}\end{array } \right . \end{array } \right.\ ] ] however , we notice that the @xmath73 conditons reduce to only @xmath1 , which can be combined to form condition ( [ para1 ] ) .
this section shows how the proposed system can be diagonalized using the eigenvectors derived in section [ seceigen ] above .
we start by examining the invariant regions of the system and then move to diagonalize the system and establish the local and global existence of solutions given the initial data lies within the invariant regions .
let us denote the positive and descendingly ordered eigenvalues of matrix @xmath38 by @xmath74 , with @xmath57 , and the corresponding eigenvectors by @xmath75 , where @xmath76 .
assuming the proposed system satisfies the parabolicity condition ( [ para1 ] ) , matrix @xmath38 is guaranteed to have strictly positive eigenvalues , and thus is unitarily diagonalizable .
generally , the diagonalizing matrix can be formed containing as its columns the normalized eigenvectors of @xmath9 . recalling that for every eigenvalue there exist two eigenvectors with unit norm and opposite directions
, we can define the diagonalizing matrix as@xmath77 where each power @xmath78 is either equal to @xmath79 or @xmath1 . in order to simplify the notation ,
let us consider the two disjoint sets @xmath80 and@xmath81 which satisfy the properties@xmath82 each permutation of @xmath83 and @xmath84 satisfying ( [ 3.3 ] ) yields a valid diagonalizing matrix .
the total number of possible permutations is thus @xmath85 , which is also the number of invariant regions @xmath86 for the proposed system .
these regions may be written as@xmath87 subject to@xmath88 for simplicity , we will only consider one of the invariant regions which corresponds to the sets @xmath89 and @xmath90 and is defined in ( [ invreg1 ] ) and ( [ invreg2 ] ) .
this yields the diagonalizing matrix@xmath91 note that the work carried out in the following subsections can be trivially extended to the remaining @xmath92 regions . in order to establish the local existence of solutions for the proposed system ( [ sys1 ] ) , we start by diagonalizing the system by means of the diagonalizing matrix defined in ( [ 3.6 ] ) .
we follow the same work performed in @xcite to obtain the equivalent diagonal system . first , let@xmath93
where@xmath94{ll}\left \langle v_{\ell},u\right \rangle , & \ell \in \mathfrak{l}\\ \left \langle \left ( -1\right ) v_{\ell},u\right \rangle , & \ell \in \mathfrak{z.}\end{array } \right.\end{aligned}\ ] ] let us also define the functional@xmath95 with each function@xmath96 fulfilling the following conditions : 1 . must be continuously differentiable on @xmath97 for all @xmath57 , satisfying @xmath98 , for all @xmath99 @xmath57 .
2 . must be of polynomial growth ( see the work of hollis and morgan @xcite ) , which means that for all @xmath57:@xmath100 3 .
must satisfy the inequality:@xmath101 where@xmath102 for all @xmath103 @xmath104 .
all the constants @xmath105 satisfy @xmath106 @xmath57 where @xmath107 @xmath104 are sufficiently large positive constants .
note that @xmath108 and @xmath109 are uniformly bounded positive functions defined on @xmath110 .
finally , let@xmath111 now , by observing the similarity transformation@xmath112 we can propose the following : [ diagprop]diagonalizing system ( [ sys1 ] ) by means of @xmath113 yields@xmath114 with the boundary condition@xmath115 or@xmath116 and the initial data@xmath117 the proof of proposition [ diagprop ] is trivial and can be looked up in @xcite .
the diagonal system in ( [ eqsys ] ) is equivalent to ( [ sys1 ] ) in the invariant region given in ( [ invreg1 ] ) and ( [ invreg2 ] ) . by considering the equivalent diagonal system in ( [ eqsys ] )
, we can now establish the local existence and uniqueness of solutions for the original system ( [ sys1 ] ) with initial data in @xmath118 or @xmath119 , @xmath120 using the basic existence theory for abstract semilinear differential equations ( friedman @xcite , henry @xcite and pazy @xcite ) .
it simply follows that the solutions are classical on @xmath121 , with @xmath122 denoting the eventual blow up time in @xmath123 .
the local solution is continued globally by _ apriori _ estimates .
the aim here is to establish the global existence of solutions for the equivalent system ( [ eqsys ] ) and consequently the original system ( [ sys1 ] ) subject to the parabolicity condition ( [ para1 ] ) through the use of an appropriate lyapunov functional .
the results obtained here are similar to those of @xcite .
hence , no detailed proofs will be given here .
let us define@xmath124 ^{2},\text { } r=3, ... ,l\text { , } \label{mycond}\ ] ] where@xmath125 \right ) ^{2^{\left ( r - k-2\right ) } } , \text { } r=3, ... ,l-1\text{,}\]]@xmath126 , \ ] ] and@xmath127 .\ ] ] the term @xmath128 denotes the determinant of the @xmath129 square symmetric matrix obtained from @xmath130 by removing the @xmath131 rows and the @xmath132 columns . where @xmath133 , ... , \det \left [ m\right ] $ ] are the minors of the matrix @xmath134 the elements of the matrix are:@xmath135 where @xmath74 in ( [ 2.2])-([2.3 ] ) .
note that @xmath136 for all @xmath137 , and @xmath138 @xmath139 are positive constants .
[ globaltheo]suppose that the functions @xmath140 @xmath57 are of polynomial growth and satisfy condition ( [ 4.2.2 ] ) for some positive constants @xmath141 @xmath57 sufficiently large . let @xmath142 be a solution of ( [ eqsys ] ) and@xmath143 where@xmath144 with @xmath145 a positive integer and @xmath146.also suppose that the following condition is satisfied@xmath147 it follows from these conditions that the functional @xmath148 is uniformly bounded on the interval @xmath149 , $ ] @xmath150 .
[ corollary1]under the assumptions of theorem [ globaltheo ] , all solutions of ( [ eqsys ] ) with positive initial data in @xmath151 are in @xmath152 for some @xmath153 .
[ proposition2]under the assumptions of theorem [ globaltheo ] and given that the condition ( [ para1 ] ) is satisfied , all solutions of ( [ eqsys ] ) with positive initial data in @xmath154 are global for some @xmath155 .
in order to put the findings of this study to the test , let us consider the following @xmath156-component system@xmath157 where the transposed diffusion matrix is given by@xmath39{ccccc}1 & 0.5 & 0 & 0 & 0\\ 0.3 & 1.5 & 0.7 & 0 & 0\\ 0 & 0.25 & 1 & 0.5 & 0\\ 0 & 0 & 0.3 & 1.5 & 0.7\\ 0 & 0 & 0 & 0.25 & 1 \end{array } \right ) , \label{6.2}\ ] ] and the reaction functional @xmath15 is of the form@xmath158{ccccc}f_{1 } & f_{2 } & f_{3 } & f_{4 } & f_{5}\end{array } \right ) ^{t},\ ] ] with@xmath159 for the purpose of this example , let @xmath160 be the symmetric matrices given by@xmath161{ccccc}0.0146 & -0.0257 & 0.0073 & -0.0088 & 0\\ -0.0257 & 0.0202 & -0.004 & 0 & 0.0044\\ 0.0073 & -0.004 & 0.0005 & 0.0011 & -0.0015\\ -0.0088 & 0 & 0.0011 & -0.0043 & 0.0027\\ 0 & 0.0044 & -0.0015 & 0.0027 & -0.0007 \end{array } \right ) , \]]@xmath162{ccccc}0.1142 & 0.228 & 0.0571 & 0.1293 & 0\\ 0.228 & -0.2281 & -0.0153 & 0 & -0.0646\\ 0.0571 & -0.0153 & 0.0041 & 0.0158 & -0.0122\\ 0.1293 & 0 & 0.0158 & 0.0489 & -0.0244\\ 0 & -0.0646 & -0.0122 & -0.0244 & -0.0061 \end{array } \right ) , \]]@xmath163{ccccc}0.3702 & -0.1245 & 0.1851 & 0.0194 & 0\\ -0.1245 & 0.0371 & -0.0817 & 0 &
-0.0097\\ 0.1851 & -0.0817 & 0.0132 & 0.0364 & -0.0397\\ 0.0194 & 0 & 0.0364 & -0.0079 & 0.0133\\ 0 & -0.0097 & -0.0397 & 0.0133 & -0.0198 \end{array } \right ) , \]]@xmath164{ccccc}-0.1316 & -0.1013 & -0.0658 & -0.0743 & 0\\ -0.1013 & 0.1177 & 0.0236 & 0 & 0.0371\\ -0.0658 & 0.0236 & -0.0047 & -0.0154 & 0.0141\\ -0.0743 & 0 & -0.0154 & -0.0252 & 0.0108\\ 0 & 0.0371 & 0.0141 & 0.0108 & 0.0070 \end{array } \right ) , \ ] ] and@xmath165{ccccc}-0.1651 & 0.5295 & -0.0825 & 0.2108 & 0\\ 0.5295 & -0.4429 & 0.0539 & 0 & -0.1054\\ -0.0825 & 0.0539 & -0.0059 & -0.0081 & 0.0177\\ 0.2108 & 0 & -0.0081 & 0.0949 & -0.0567\\ 0 & -0.1054 & 0.0177 & -0.0567 & 0.0088 \end{array } \right ) .\ ] ] also , suppose that@xmath41{l}\sigma_{1}=\left ( \begin{array } [ c]{ccccc}0.0795 & 0.0303 & -0.0243 & -0.014 & 0.0059 \end{array } \right ) ^{t}\\ \sigma_{2}=\left ( \begin{array } [ c]{ccccc}-0.6466 & -0.6144 & 0.0798 & 0.2844 & 0.0572 \end{array } \right ) ^{t}\\ \sigma_{3}=\left ( \begin{array } [ c]{ccccc}0.4549 & -0.2791 & -0.2846 & 0.1292 & 0.1635 \end{array } \right ) ^{t}\\ \sigma_{4}=\left ( \begin{array } [ c]{ccccc}0.2682 & 0.3879 & 0.0097 & -0.1796 & -0.0618 \end{array } \right ) ^{t}\\ \sigma_{5}=\left ( \begin{array } [ c]{ccccc}-1.6033 & -0.8159 & 0.4251 & 0.3777 & -0.0608 \end{array } \right ) ^{t}. \end{array } \right.\ ] ] the system clearly satisfies the parabolicity condition ( [ para1 ] ) as@xmath166 we have from ( [ 2.3])@xmath167 hence , we can form the polynomial @xmath45 as@xmath41{l}p_{0}\left ( \mu \right ) = 1,\text { } p_{1}\left ( \mu \right ) = \mu \\ p_{2}\left ( \mu \right ) = \mu(\mu)-1=\mu^{2}-1 , \end{array } \right.\ ] ] with solutions@xmath168 now , the eigenvalues are @xmath49 along with the solutions of the following two equations derived from ( [ 2.4])@xmath41{l}\frac{\left ( 1-\lambda \right ) \left ( 1.5-\lambda \right ) } { \sqrt { 0.5\times0.7\times0.3\times0.25}}-\frac{1}{1.0801}-1.0801=1\\ \frac{\left ( 1-\lambda \right ) \left ( 1.5-\lambda \right ) } { \sqrt { 0.5\times0.7\times0.3\times0.25}}-\frac{1}{1.0801}-1.0801=-1 , \end{array } \right.\ ] ] which can be simplified to@xmath41{l}6.1721\left
( \lambda-1\right ) \left ( \lambda-1.5\right ) -3.0059=0\\ 6.172\,1\left ( \lambda-1\right ) \left ( \lambda-1.5\right ) -1.0059=0 .
\end{array } \right.\ ] ] solving the two quadratic equations in @xmath169 yields the four eigenvalues of @xmath9 , which in descending order can be given by@xmath41{l}\lambda_{1}=1.9913\\
\lambda_{2}=1.7248\\ \lambda_{3}=1\\ \lambda_{4}=0.77516\\ \lambda_{5}=0.50871 .
\end{array } \right . \label{6.6}\ ] ] hence,@xmath170 similarly , formula ( [ 2.7])-([2.7.1 ] ) can be used to derive the eigenvectors of @xmath38 , which are arranged according to the corresponding eigenvalues to form the diagonalizing matrix@xmath171{ccccc}0.3848 & -0.5265 & -0.5632 & -0.9063 & -0.8769\\ 0.7629 & -0.7633 & 0.5534 & 0.0000 & 0.3943\\ 0.3705 & -0.0195 & -0.5423 & 0.3884 & -0.0325\\ 0.3531 & 0.3534 & 0.2562 & 0.0000 & -0.1825\\ \underset{v_{1}}{\underbrace{0.0891 } } & \underset{v_{2}}{\underbrace{0.1219 } } & \underset{v_{3}}{\underbrace{-0.1303 } } & \underset{v_{4}}{\underbrace { -0.1665 } } & \underset{v_{5}}{\underbrace{0.2030}}\end{array } \right ) .
\label{6.7}\ ] ] matrix @xmath113 is used to diagonalize the system yields the equivalent system@xmath41{l}\frac{\partial w_{1}}{\partial t}-1.9913\delta w_{1}=-0.5w_{1}w_{5}+0.65w_{2}\\ \frac{\partial w_{2}}{\partial t}-1.7248\delta w_{2}=0.5w_{1}w_{5}-0.65w_{2}\\ \frac{\partial w_{3}}{\partial t}-\delta w_{3}=-0.32w_{3}w_{5}+0.41w_{4}\\ \frac{\partial w_{4}}{\partial t}-0.77516\delta w_{4}=0.32w_{3}w_{5}-0.41w_{4}\\ \frac{\partial w_{5}}{\partial t}-0.50871\delta w_{5}=-0.5w_{1}w_{5}+0.65w_{2}-0.32w_{3}w_{5}+0.41w_{4}. \end{array } \right .
\label{ex2diag}\ ] ] note that for simplicity , we have neglected small terms and rounded the polynomial coefficients to four decimal points .
the resulting reaction terms clearly satisfy conditions ( a1 ) through ( a3 ) as discussed in section [ subseclocal ] above .
observe that the proposed system has @xmath172 invariant regions where the resulting @xmath173 is guaranteed to be positive .
we consider one of these regions corresponding to @xmath174 and given by@xmath175 which yields five inequalities@xmath41{l}0.3848u_{01}+0.7629u_{02}+0.3705u_{03}+0.3531u_{04}+0.0891u_{05}\geq0\\ -0.5265u_{01}-0.7633u_{02}-0.0195u_{03}+0.3534u_{04}+0.1219u_{05}\geq0\\ -0.5632u_{01}+0.5534u_{02}-0.5423u_{03}+0.2562u_{04}-0.1303u_{05}\geq0\\ -0.9063u_{01}+0.3884u_{03}-0.1665u_{05}\geq0\\ -0.8769u_{01}+0.3943u_{02}-0.0325u_{03}-0.1825u_{04}+0.2030u_{05}\geq0 , \end{array } \right
. \label{6.8}\ ] ] with@xmath176 solving this system of inequalities yields the first region where the initial data is assumed to lie .
we will consider for instance the initial data@xmath177 the equivalent diagonalized system ( [ ex2diag ] ) was solved numerically by means of the finite difference ( fd ) method .
figures [ fig1 ] and [ fig2 ] show the solutions to the diagonalized system ( [ ex2diag ] ) and the original system ( [ ex2 ] ) , respectively , in the diffusion free case . in the one dimensional case
, a sinusoidal perturbation is added to the initial data to introduce spatial diversity into the model .
the solutions are shown in figures [ fig3 ] and [ fig4 ] .
s. abdelmalek , _ existence of global solutions via invariant regions for a generalized reaction
diffusion system with a tridiagonal toeplitz matrix of diffusion coefficients _ , accepted for publication in func .
& apps .
s. kouachi and b. rebiai , _ invariant regions and the global existence for reaction diffusion systems with a tridiagonal matrix of diffusion coefficients _ , memoirs on diff .
eqs . and math .
( 2010 ) : 93108 . | this paper is concerned with the local and global existence of solutions for a generalized @xmath0-component reaction diffusion system with a tridiagonal @xmath1toeplitz diffusion matrix and polynomial growth .
we derive the eigenvalues and eigenvectors and determine the parabolicity conditions in order to diagonalize the proposed system .
we , then , determine the invariant regions and utilize a lyapunov functional to establish the global existence of solutions for the proposed system .
a numerical example is used to illustrate and confirm the findings of the study .
reaction diffusion systems , invariant regions , diagonalization , global existence , lyapunov functional . |
clusters of galaxies , as the most massive gravitationally bound material structures , are of basic importance in the study of both baryonic and dark - matter density distributions in the universe . in practice ,
measurement of line - of - sight velocity dispersions of the galaxies in a cluster ( see e.g. @xcite and @xcite ) , observation of x - ray emission from the hot gas in a cluster s gravitational potential well ( see e.g. @xcite ; ; @xcite and @xcite ) , observation at microwave frequencies of the sz @xcite effect ( see e.g. @xcite and ) and measurement of the gravitational lensing of background galaxies by the cluster potential ( see e.g. @xcite and @xcite ) are all key in assessing the matter distribution in clusters .
however , each method has different strengths and weaknesses .
historically , the majority of the studies on the measurement of galaxy cluster masses have been in the x - ray band , and we consider x - ray measurement and analysis in this paper . the motivation for our work is to augment the high resolution x - ray observations of clusters with an analysis pipeline that comprises ( a ) a cluster model consistent with both numerical simulations and real observations of clusters , and ( b ) a bayesian statistical method .
this provides an interesting and very powerful way to investigate the constraints on the cluster physical parameters imposed by x - ray data .
the great majority of x - ray ( and indeed of sz ) measurements of cluster masses in the literature assume parameterised functional forms for the radial distribution of two independent cluster thermodynamic properties , such as electron density and temperature , to model the x - ray surface brightness ( see e.g. @xcite ; @xcite ; @xcite ; @xcite and ) .
these radial profiles ( e.g. @xmath3-model ) have an amplitude normalisation parameter and two or more shape parameters .
in bayes - x we use our recently developed cluster model @xcite to parameterise the radial x - ray surface brightness profile and explore the constraints on both model parameters and physical parameters of simulated _ chandra_- like observations of four clusters .
the model , hereafter model ( i ) , assumes that the dark matter density follows a navarro , frenk and white ( nfw ) profile @xcite and the icm plasma pressure is described by the generalised nfw ( gnfw ) profile , both in accordance with numerical simulations that take into account radiative cooling , star formation and energy feedback from supernova explosions ( see e.g. ; @xcite and ) .
computational advances allow us to compare the model with the data in a fully bayesian fashion ( see e.g. @xcite and @xcite ) , which has the well - known advantages of employing prior information and in dealing with probability distributions of any shape rather than just a mean value and an error - bar .
this is linked to ability of the method to explore degeneracies that would otherwise be hidden .
the importance of this property will become clear in this paper .
a bayesian approach has been previously used in the x - ray analysis of galaxies and galaxy clusters ( see e.g. @xcite , @xcite , @xcite and @xcite ) . in all of these studies ,
the spectra are extracted in concentric annuli around the cluster centre requiring annular widths to be large enough to give significant counts .
the data also need to be binned in the pi ( pulse invariant ) channels . in bayes - x
we extract the spectra in a 3-dimensional grid , two spatial dimensions and one energy dimension .
we do not bin the counts in pi energy channels either .
the previous studies differ from bayes - x in the models they assume for the cluster total and gas mass , the sampling parameters and the prior probability distributions
. however , the assumptions of spherical geometry and hydrostatic equilibrium are common to all approaches .
further , to perform a bayesian model selection , bayes - x substitutes the nfw density profile in model i with the einasto density profile @xcite , hereafter model ii , to parameterise the radial profiles of the gas density and temperature .
since one of the output results of bayes - x is the bayesian evidence , we can then calculate the bayes factor to compare the two models in describing the data .
the full details of the model ii are given in appendix a. throughout , we assume a @xmath4 cosmology with @xmath5 .
@xmath6 , @xmath7 , and @xmath8 represent the value of the cluster total mass , gas mass and gas mass fraction _ internal _ to the overdensity radius of @xmath9 , respectively , whereas @xmath10 represents the gas temperature _ at _ radius @xmath9 . the inner and outer contours in 2d
marginalised posterior probability distributions indicate the areas enclosing @xmath11 and @xmath12 of the probability distributions .
bayesian inference allows the estimation of a set of parameters @xmath13 within a model ( or hypothesis ) @xmath14 using the data @xmath15 .
bayes theorem states that : @xmath16 where @xmath17 is the posterior probability distribution of the parameters @xmath18 , given @xmath14 and @xmath15 , @xmath19 is the likelihood , @xmath20 is _ a priori _ `` prior '' probability distribution of @xmath18 given @xmath14 , and @xmath21 is the bayesian evidence .
bayesian inference in practice often divides into two parts : parameter estimation and model selection . in parameter estimation , the normalising evidence factor is usually ignored , since it is independent of the parameters @xmath18 , and inferences are obtained by searching the unnormalised posterior distributions using sampling techniques .
the posterior distribution can be subsequently marginalised over each parameter to give individual parameter constraints .
in contrast to parameter estimation , in model selection the evidence takes the central role and is simply the factor required to normalise the posterior over @xmath13 : @xmath22 where @xmath23 is the dimensionality of the parameter space . according to occam s razor ( see e.g , @xcite and @xcite ) , a simple theory with compact parameter space will have a larger evidence than a more complicated one , unless the latter is significantly better at explaining the data .
the question of the selection between two models @xmath24 and @xmath25 is decided by comparing their respective posterior probabilities , given the observed data set @xmath15 , via the model selection ratio @xmath26 where @xmath27 is the bayes factor @xcite and @xmath28 is the prior probability ratio for the two models . determining the bayes factor then provides a scale for comparing the posterior model odds of the two models where there is no prior reason to prefer one model versus another .
( i.e.@xmath29 ) .
table [ tab : bayesfactor ] lists jeffreys scale of the strength of evidence for bayes factors . .jeffreys
scale for an interpretation of the strength of the evidence for bayes factor @xcite .
[ tab : bayesfactor ] [ cols="<,^",options="header " , ] model i model ii model i model ii
by performing a bayesian analysis of simulated and real _ chandra _ data we have investigated the capability of our model ( where we assume that the dark matter density follows a nfw - profile and that the gas pressure is described by a gnfw profile ) and bayes - x to return the cluster input quantities and constrain the cluster physical parameters . we simulated _
chandra_-like observations of four clusters in a redshift range of @xmath30 all with the same @xmath31 .
we have performed four sets of analyses including prior - only analysis and assuming different types of priors on @xmath1 and model parameters @xmath2 .
we have demonstrated that bayes - x faithfully recovers the input values of the model parameters used in the simulations and can constrain clusters positions on the sky and clusters physical parameters including @xmath32 , @xmath33 and @xmath34 .
we find that we can still constrain @xmath35 as well as other cluster parameters even when we assume a wide uniform prior on @xmath36 . by letting @xmath2 vary in the analysis
we have shown that bayes - x is able to reveal the degeneracy among these parameters which must be taken into account for an unbiased estimate of cluster parameters .
we did this by assuming gaussian and uniform prior probability distributions on @xmath2 respectively .
the results also show no correlation between @xmath34 and @xmath37 , @xmath38 , or @xmath39 as one would expect .
the results of prior - only analyses show that we recover the assumed prior probability distributions for cluster positions , model parameters and physical parameters .
we find that the results of the analyses do not depend on the choice of prior probability distributions on the sampling parameters for these high signal - noise simulations and in all cases we were able to recover the input values of the simulated clusters and expected degeneracies among the cluster parameters .
the results of bayes - x analysis of _ chandra _ data on a262 also show the expected variation of the parameters as a function of radius .
the inferred cluster parameters at two overdensities of @xmath40 and @xmath41 are in general agreement with the results presented in the literature .
the results of the bayesian model selection favour model i decisively for the x - ray simulated data as expected but prefers model ii for a262 .
we therefore conclude that bayes - x is robust and can be used to analyse x - ray data and in future multi - wavelength analysis of clusters of galaxies .
the authors thank the two referees , stefano andreon and fabio gastaldello for their useful comments .
the analysis work was conducted on the darwin supercomputer of the university of cambridge high performance computing service supported by hefce and cosmos uk national supercomputer at damtp , university of cambridge .
the authors thank stuart rankin and andrey kaliazin for their computing assistance and dave green for his invaluable help with latex .
mo acknowledges a research fellowship from sidney sussex college , cambridge . 99 allen s. w. , rapetti d. a. , schmidt r. w. , ebeling h. , morris r. g. , fabian a. c. , 2008 , mnras , 383 , 879 allen s. w. , evrard a. e. , mantz a. b. , 2011 , ara&a , 49 , 409 ami consortium : hurley - walker et al . , 2012 , mnras , 419 , 2921 ami consortium : olamaie et al . , 2012 , mnras , 421 , 1136 anders e. , grevesse n. , 1989 , gecoa , 53 , 197 arnaud m. , pratt g. w. , piffaretti r. , bhringer h. , croston j. h. , pointecouteau e. , 2010 , a & a , 517 , a92 arnaud , k. , smith , r. , & siemiginowska , a. 2011 , handbook of x - ray astronomy , by edited by keith arnaud , randall smith , aneta siemiginowska , cambridge , uk : cambridge university press , 2011 , balucinska - church m. , mccammon d. , 1992 , apj , 400 , 699 bartalucci i. , mazzotta p. , bourdin h. , vikhlinin a. , 2014 , arxiv:1404.3587 bautz m. w. , et al . , 2009 , pasj , 61 , 1117 b hringer h. et al . , 2007 , a&a , 469 , 363 bonamente m. , joy m. k. , carlstrom j. e. , reese e. d. , laroque s. j. , 2004 , apj , 614 , 56 borgani s. , et al . , 2004 , mnras , 348 , 1078 borgani s. , 2004 , ap & ss , 294 , 51 broos p. s. , townsley l. k. , feigelson e. d. , getman k. v. , bauer f. e. , garmire g. p. , 2010 ,
apj , 714 , 1582 carlberg r. g. , et al . , 1997 , apj , 485 , l13 chemin l. , de blok w. j. g. , mamon g. a. , 2011 , aj , 142 , 109 corless v. l. , king l. j. , clowe d. , 2009 , mnras , 393 , 1235 carter c. , karovska m. , jerius d. , glotfelty k. , beikman s. , 2003 , aspc , 295 , 477 davis j. e. , 2001 , apj , 548 , 1010 de luca a. , molendi s. , 2004 , a&a , 419 , 837 dutton a. a. , macci a. v. , 2014 , mnras , 441 , 3359 einasto j. , 1965 , tralm , 5 , 87 feroz f. , hobson m. p. , 2008 ,
mnras , 384 , 449 feroz f. , hobson m. p. , bridges m. , 2009 , mnras , 398 , 1601 feroz f. , hobson m. p. , zwart j. t. l. , saunders r. d. e. , grainge k. j. b. , 2009 , mnras , 398,2049 feroz f. , hobson m. p. , cameron e. , pettitt a. n. , 2013 , arxiv:1306.2144 feroz f.,2013,ieee 13th international conference , 10.1109/icdmw.2013.21 fruscione a. , et al . , 2006 , spie , 6270 , gao l. , navarro j. f. , cole s. , frenk c. s. , white s. d. m. , springel v. , jenkins a. , neto a. f. , 2008 , mnras , 387 , 536 gastaldello f. , buote d. a. , humphrey p. j. , zappacosta l. , bullock j. s. , brighenti f. , mathews w. g. , 2007 , apj , 669 , 158 green d. a. , 2011 , basi , 39 , 289 gronenschild e. h. b. m. , mewe r. , 1978 , a&as , 32 , 283 holder g. p. , mccarthy i. g. , babul a. , 2007 , mnras , 382 , 1697 humphrey p. j. , buote d. a. , 2006 , apj , 639 , 136 humphrey p. j. , buote d. a. , brighenti f. , gebhardt k. , mathews w. g. , 2009 , apj , 703 , 1257 humphrey p. j. , buote d. a. , canizares c. r. , fabian a. c. , miller j. m. , 2011 , apj , 729 , 53 humphrey p. j. , buote d. a. , brighenti f. , flohic h. m. l. g. , gastaldello f. , mathews w. g. , 2012 , apj , 748 , 11 jaynes e. t. , 1986 , bayesian methods : an introductory tutorial , cambridge university press jeffreys h. , 1961 , theory of probability , 3rd ed .
oxford : university press .
kass robert e. and raftery adrian e. , 1995 , journal of the american statistical association , 90 , 430 kaastra j. s. , verbunt f. , 2010 , high energy astrophysics .
koutroumpa d. , acero f. , lallement r. , ballet j. , kharchenko v. , 2007 , a&a , 475 , 901 koutroumpa d. , lallement r. , kharchenko v. , dalgarno a. , 2009 , ssrv , 143 , 217 krolik j. h. , kallman t. r. , 1984 , apj , 286 , 366 laroque s. j. , bonamente m. , carlstrom j. e. , joy m. k. , nagai d. , reese e. d. , dawson k. s. , 2006 , apj , 652 , 917 mahdavi a. , hoekstra h. , babul a. , sievers j. , myers s. t. , henry j. p. , 2007 ,
apj , 664 , 162 mccarthy i. g. , babul a. , bower r. g. , balogh m. l. , 2008 , mnras , 386 , 1309 mewe r. , 1972 , soph , 22 , 459 mewe r. , 1975 , soph , 44 , 383 mewe r. , lemen j. r. , van den oord g. h. j. , 1986 , a&as , 65 , 511 mewe , r. , kaastra , j.s .
, liedahl , d.a . , 1995 ,
legacy 6 , 16 mroczkowski t. , et al .
, 2009 , apj , 694 , 1034 nagai d. , kravtsov a. v. , vikhlinin a. , 2007 , apj , 668 , 1 navarro j. f. , frenk c. s. , white s. d. m. , 1996 , apj , 462 , 563 navarro j. f. , frenk c. s. , white s. d. m. , 1997 , apj , 490 , 493 navarro j. f. , et al . , 2010 , mnras , 402 , 21 neto a. f. , et al . , 2007 , mnras , 381 , 1450 olamaie m. , hobson m. p. , grainge k. j. b. , 2012 , mnras , 423 , 1534 olamaie m. , hobson m. p. , grainge k. j. b. , 2013 , mnras , 430 , 1344 piffaretti r. , valdarnini r. , 2008 , a&a , 491 , 71 plagge t. , et al .
, 2010 , apj , 716 , 1118 planck collaboration , ade , p. a. r. , aghanim , n. , et al .
2011 , a&a , 536 , a8 planck collaboration , et al . , 2013 ,
a&a , 550 , a128 planck collaboration , et al . , 2013 ,
a&a , 550 , a131 pointecouteau e. , arnaud m. , pratt g. w. , 2005 , a&a , 435 , 1 pratt g. w. , et al . , 2010 , a&a , 511 , a85 retana - montenegro e. , van hese e. , gentile g. , baes m. , frutos - alfaro f. , 2012 , a&a , 540 , a70 rines k. , geller m. j. , diaferio a. , 2010 , apj , 715 , l180 russell h. r. , et al .
, 2012 , mnras , 423 , 236 sanders j. s. , fabian a. c. , frank k. a. , peterson j. r. , russell h. r. , 2010 , mnras , 402 , 127 sanders j. s. , fabian a. c. , 2013 , mnras , 429 , 2727 sanders j. s. , fabian a. c. , hlavacek - larrondo j. , russell h. r. , taylor g. b. , hofmann f. , tremblay g. , walker s. a. , 2014 , mnras , 444 , 1497 sarazin c. l. , 1988 , x ray emission from clusters of galaxies , cambridge university press sato k. , matsushita k. , gastaldello f. , 2009 , pasj , 61 , 365 shimwell a. c. t. w. , et al . , 2013 ,
mnras , 433 , 2036 siemiginowska a. , burke d. j. , aldcroft t. l. , worrall d. m. , allen s. , bechtold j. , clarke t. , cheung c. c. , 2010 , apj , 722 , 102 sifn c. , et al . , 2013 ,
apj , 772 , 25 sivia d. s. , skilling j. , 2005 , data analysis a bayesian tutorial , oxford university press snowden s. l. , mushotzky r. f. , kuntz k. d. , davis d. s. , 2008 , a&a , 478 , 615 snowden s. l. , 2009 , ssrv , 143 , 253 struble m. f. , rood h. j. , 1999 , apjs , 125 , 35 sun m. , voit g. m. , donahue m. , jones c. , forman w. , vikhlinin a. , 2009 , apj , 693 , 1142 sunyaev r. a. , zeldovich y. b. , 1970 , coasp , 2 , 66 tawa n. , et al . , 2008 , pasj , 60 , 11 van dyk d. a. , connors a. , kashyap v. l. , siemiginowska a. , 2001 , apj , 548 , 224 vikhlinin a. , markevitch m. , murray s. s. , jones c. , forman w. , van speybroeck l. , 2005 , apj , 628 , 655 vikhlinin a. , kravtsov a. , forman w. , jones c. , markevitch m. , murray s. s. , van speybroeck l. , 2006 , apj , 640 , 691 wachter k. , leach r. , kellogg e. , 1979 , apj , 230 , 274 wise m. , 1997 , chnew , 5 , 22 wise m. w. , huenemoerder d. p. , davis j. e. , 1997 , aspc , 125 , 477
in this model we first assume that the cluster matter density follows the einasto profile , @xmath42 \right\},\ ] ] where @xmath43 known as scale radius is the radius where the logarithmic slope of the density profile is @xmath44 , @xmath45 is the density at the scale radius and @xmath46 is the shape parameter .
the halo concentration parameter is also defined as @xmath47 .
figure [ fig : einastoplot ] shows the einasto density profile for different shape parameter versus nfw density profile . as for model
i , our second assumption is that @xmath48 follows the gnfw profile , @xmath49 the gas pressure is then defined by @xmath50 the third assumption concerns the dynamical state of the cluster , which we take to be in hydrostatic equilibrium throughout .
thus , the total cluster mass internal to radius @xmath51 is related to the gas pressure gradient at that radius by @xmath52 assuming spherical geometry and using equation [ eq : eindensity ] , the total mass enclosed within radius @xmath51 has the analytical solution , @xmath53,\end{aligned}\ ] ] where @xmath54 is the lower incomplete gamma function . substituting this form and the expressions ( [ eq : eingnfw ] ) and ( [ eq : einpgas ] ) for the gas pressure into the condition ( [ eq : einhse ] ) for hydrostatic equilibrium , one may derive the gas density profile @xmath55}\times \nonumber\\ & & \left(\frac{r}{r_{\rm p}}\right)^ { { -c}}\left[1 + \left(\frac{r}{r_{\rm p}}\right)^ { a}\right]^{-\left(\frac { { { a + b - c } } } { { a}}\right)}\left [ { b } \left(\frac{r}{r_{\rm p}}\right)^ { a } + { c } \right].\end{aligned}\ ] ] the radial profile of the electron number density is then trivially obtained using @xmath56 . assuming an ideal gas equation of state , this in turn yields the electron temperature profile @xmath57 , given by @xmath58\times \nonumber\\ & & \left [ \frac{\mbox{\huge $ \gamma$}\left[\frac{3}{\alpha},\frac{2}{\alpha}\left ( \frac{r}{r_{-2}}\right)^ \alpha\right]}{r } \right ] \times \nonumber\\ & & \left [ 1 + \left(\frac{r}{r_{\rm p}}\right)^ { a } \right]\left[{b } \left(\frac{r}{r_{\rm p}}\right)^{a } + { c } \right]^{-1}.\end{aligned}\ ] ] the only fundamental cluster property for which the radial profile can not be expressed in an explicit analytical form is the gas mass enclosed within radius @xmath51 , @xmath59 for the gas density profile in ( [ eq : einrhogas ] ) , we have been unable to evaluate this expression analytically , and so @xmath60 must be obtained using numerical integration .
consequently , the enclosed gas mass fraction profile @xmath61 also can not be written in closed form .
similar to nfw profile , in order to perform the bayesian analysis and determine the radial profiles of quantities of interest for a given cluster , one must first calculate the values of the model parameters @xmath62 and @xmath63 .
bayes - x assumes the same sampling parameters given in equation [ eq : prior ] for model ii as well .
thus for a given @xmath64 and @xmath65 , it calculates @xmath66 , assuming spherical geometry for the cluster , @xmath67 @xmath68 is also calculated using equation [ eq : c200m200 ] and hence @xmath69 .
the value of @xmath70 is then obtained by equating the input value of @xmath71 with the rhs of ( [ eq : einmass ] ) evaluated at @xmath72 , and is given by @xmath73}.\ ] ] by equating equations [ eq : einmass ] and [ eq : sphmass ] at @xmath74 and @xmath75 , bayes - x calculates both radii and hence @xmath76 .
finally , @xmath63 is obtained by substituting ( [ eq : einrhogas ] ) into ( [ eq : einmgas ] ) , evaluating the rhs at @xmath72 and equating the result to @xmath77 .
this yields @xmath78m_{\rm { g}}(r_{200})\times \nonumber\\ & & \frac{1 } { { \displaystyle \int_{0}^{r_{\rm 200 } } } r^{'3 } { \rm d}r ' \frac{\left[b \left(\frac{r'}{r_{\rm p}}\right)^a + c \right ] } { \mbox{\huge $ \gamma$}\left[\frac{3}{\alpha},\frac{2}{\alpha } \left(\frac{r}{r_{-2}}\right)^\alpha\right ] \left(\frac{r'}{r_{\rm p}}\right)^c \left[1 + \left(\frac{r'}{r_{\rm p}}\right)^a\right]^{\left(\frac{{a + b - c}}{a}\right ) } } } , \end{aligned}\ ] ] which is evaluated numerically . | we present the first public release of our bayesian inference tool , bayes - x , for the analysis of x - ray observations of galaxy clusters .
we illustrate the use of bayes - x by analysing a set of four simulated clusters at @xmath0 as they would be observed by a _ chandra_-like x - ray observatory . in both the simulations and the analysis pipeline
we assume that the dark matter density follows a spherically - symmetric navarro , frenk and white ( nfw ) profile and that the gas pressure is described by a generalised nfw ( gnfw ) profile .
we then perform four sets of analyses .
these include prior - only analyses and analyses in which we adopt wide uniform prior probability distributions on @xmath1 and on the model parameters describing the shape and slopes of the gnfw pressure profile , namely @xmath2 . by numerically exploring the joint probability distribution of the cluster parameters given simulated _
chandra_-like data , we show that the model and analysis technique can robustly return the simulated cluster input quantities , constrain the cluster physical parameters and reveal the degeneracies among the model parameters and cluster physical parameters .
we then use bayes - x to analyse _ chandra _ data on the nearby cluster , a262 , and derive the cluster physical and thermodynamic profiles .
the results are in good agreement with other results given in literature for the cluster .
to illustrate the performance of the bayesian model selection , we also carried out analyses assuming an einasto profile for the matter density and calculated the bayes factor .
the results of the model selection analyses for the simulated data favour the nfw model as expected . however , we find that the einasto profile is preferred in the analysis of a262 .
the bayes - x software , which is implemented in fortran 90 , is available at http://www.mrao.cam.ac.uk / facilities / software / bayesx/. galaxies : clusters cosmology : observations methods : data analysis |
during the late phases of stellar evolution of massive stars , electron capture processes on heavy nuclei dominate the time evolution of the lepton - to - baryon ratio ( @xmath0 ) of the core material primarily due to the low entropy of the stellar core but also due to the resulting dominance of heavy nuclei over free nucleons @xcite . inside the iron core of massive stars ,
the electron capture occurs via the gamow - teller , gt@xmath2 , transitions changing protons in the @xmath3 level into neutrons in the @xmath4 levels . on the other hand
the gt@xmath5 strength is responsible for transforming a neutron into a proton ( resulting in positron capture processes ) .
it is conjectured that the electron captures on proton and positron captures on neutron play a very crucial role in the supernovae dynamics @xcite .
due to the weak interaction processes , mainly electron and positron captures onto nuclei and @xmath6-decays of nuclei , the value of @xmath0 for a massive star changes from @xmath7 ( during hydrogen burning ) to roughly @xmath8 ( at the beginning of carbon burning ) and finally to around @xmath9 just before the collapse leading to a supernova explosion @xcite .
the temporal variation of @xmath0 within the core of a massive star has a pivotal role to play in the stellar evolution and a fine - tuning of this parameter at various stages of presupernova evolution is the key to generate an explosion @xcite .
furthermore , the electron capture rates is assumed to determine the mass of the core and thus the fate of the shock wave formed by the supernova explosion @xcite .
the calculation of electron and positron capture rates is very sensitive to the distribution of the gt@xmath10 strength functions @xcite . during the early phases of presupernova evolution ,
the electron chemical potential is of the same order of magnitude as the nuclear @xmath11 value and the capture rates are sensitive to the details of the gt@xmath10 strength functions in daughter nuclei @xcite . with proceeding collapse the stellar density increases by orders of magnitude and
the resulting electron chemical potential is significantly higher than the nuclear @xmath11 value and then the capture rates are largely determined by the centroid and the total strength of the gt@xmath10 strength functions .
a microscopic calculation of ground and excited state gt@xmath10 strength functions is the key to reliably estimate the stellar capture rate @xcite .
the positron captures are of great importance at high temperature and low density locations @xcite . in such conditions ,
a rather high concentration of positron can be reached from @xmath12 equilibrium favoring the @xmath13 pairs .
the electron captures on proton and positron captures on neutron are considered important ingredients in the modeling of type ii supernovae @xcite .
the pn - qrpa theory @xcite is an efficient way to microscopically generate gt@xmath10 strength functions which constitute a primary and nontrivial contribution to the weak interaction mediated rates among iron - regime nuclide . in this model
a quasiparticle basis is first constructed with a pairing interaction .
the random phase approximation ( rpa ) equation is then solved with a schematic gt residual interaction .
the pn - qrpa model has been developed by halbleib and sorensen @xcite by generalizing the usual rpa to describe charge - changing transitions of the type @xmath14 .
the model is then extended to deformed nuclei ( using nilsson wave functions @xcite ) by krumlinde and mller @xcite .
extension of the model to treat odd - odd nuclei and transitions from nuclear excited states has been done by muto and collaborators @xcite .
the pn - qrpa is a simple and microscopic model .
the pn - qrpa model has two big advantages for performing weak interaction calculations in stellar matter .
firstly it can handle any arbitrarily heavy system of nucleons since the calculation is performed in a luxurious model space of up to @xmath15 major oscillator shells .
the second advantage is even more important for the calculation of weak interaction rates in stellar environment .
the prevailing temperature of the stellar matter is of the order of a few hundred kilo - electron volts to a few million - electron volts and gt@xmath10 transitions occur not only from nuclear ground state , but also from excited states .
as experimental information about excited state strength functions seems inaccessible , aufderheide @xcite stressed much earlier the need to probe these strength functions theoretically . today
the pn - qrpa theory can be utilized to calculate gt@xmath10 strength distribution of _ all _ excited states of parent nucleus in a microscopic fashion and this feature of the pn - qrpa model greatly enhances the reliability of the calculated rates in stellar matter @xcite . in other words , the pn - qrpa model seems to allow a microscopic state - by - state calculation of all stellar weak rates and the brink hypothesis is not assumed in this model .
brink hypothesis @xcite states that gt strength distribution on excited states is _ identical _ to that from ground state , shifted _ only _ by the excitation energy of the state .
recent calculations have pointed towards the fact that brink hypothesis is not a safe approximation to use for calculation of stellar weak interaction rates @xcite .
recently , weak interaction rates of isotopes of iron , @xmath1fe , have been calculated using the pn - qrpa model @xcite . there , the author reported that the calculated electron capture rates on these iron isotopes were overall in fair agreement with the large - scale shell model ( lssm ) results @xcite .
however , it has been found that the calculated beta decay rates are suppressed by three to five orders of magnitude . in present paper
, we present the fine - grid calculation of stellar electron and positron capture rates for iron - isotopes @xmath1fe using pn - qrpa model .
recently , the improved pn - qrpa calculation has been introduced @xcite , where it was reported that the betterment resulted mainly from the incorporation of measured deformation values for these nuclei . a detailed analysis of the calculated ground and excited state gt@xmath10 strength distributions for @xmath1fe and the capture rates has been presented in ref .
@xcite . due to
the extreme conditions prevailing in the cores of massive stars , interpolation of calculated rates within large intervals of temperature - density points posed some uncertainty in the values of capture rates for collapse simulators @xcite .
as such the calculation is done on a detailed temperature and density grid pertinent to presupernova and supernova environment and should prove more suitable for running on simulation codes . in present work , a detailed calculation of electron and positron
capture rates on @xmath1fe is being presented for the first time at temperature - density intervals suitable for simulation purposes .
further , we also compare the pn - qrpa capture rates with previous key calculations for the astrophysically important range of stellar temperatures and densities .
it is worthwhile to mention that the presence of @xmath1fe in the stellar core is a result of the last phase of silicon burning .
the paper is organized as follows .
section [ sec:2 ] briefly discusses the formalism of the pn - qrpa calculations . in section [ sec:3 ] ,
the calculated results are present .
comparison with earlier calculations during presupernova evolution of massive stars is also included in this section .
we summarize the main conclusions in section [ sec:4 ] .
the fine - grid calculation of the stellar electron and positron capture rates on @xmath1fe is presented in table [ tab.1 ] .
the electron capture ( ec ) and positron capture ( pc ) rates of a transition from the @xmath16-th state of the parent to the @xmath17-th state of the daughter nucleus is given by @xmath18\left[f_{ij } ( t,\rho , e_{f } ) \right]\left[b(f)_{ij } + \left({\raise0.7ex\hbox{$ g_{a } $ } \!\mathord{\left/ { \vphantom { g_{a } g_{v } } } \right .
\kern-\nulldelimiterspace}\!\lower0.7ex\hbox{$ g_{v } $ } } \right)^{2 } b(gt)_{ij } \right].\end{aligned}\ ] ] the value of d is reported in ref .
@xmath19 are the sum of reduced transition probabilities of the fermi b(f ) and gt transitions b(gt ) . whereas for @xmath20fe phonon transitions contribute , in the case of @xmath21fe ( odd - a case ) two kinds of transitions are possible within the framework of the pn - qrpa model .
one are the phonon transitions , where the odd quasiparticle acts as spectator and the other is the transitions of the odd quasiparticle itself . in the later case phonon correlations have been introduced to one - quasiparticle states in first - order perturbation @xcite .
the functions @xmath22 give the phase space integrals .
details of the calculations of phase space integrals and reduced transition probabilities in the pn - qrpa model can be found in refs .
@xcite .
therefore , the total electron ( positron ) capture rate per unit time per nucleus reads @xmath23 the summation over all initial and final states is to be carried out until satisfactory convergence in the rate calculations is achieved . here @xmath24 is the probability of occupation of parent excited states and follows the normal boltzmann distribution .
the pn - qrpa theory allows a microscopic state - by - state calculation of both sums present in eq .
( [ eq:2a ] ) .
as discussed earlier in previous section , this feature of the pn - qrpa model greatly increases the reliability of the calculated rates in stellar matter where there exists a finite probability of occupation of excited states . in order to further increase the reliability of the calculated capture rates experimental data were incorporated in the calculation wherever possible .
in addition to the incorporation of the experimentally adopted value of the deformation parameter , the calculated excitation energies ( along with their log @xmath25 values ) were replaced with an experimental one when they were within 0.5 mev of each other .
the calculation presented in this work takes into consideration certain aspects .
the missing measured states are inserted and inverse and mirror transitions are also taken into account .
no theoretical levels have been replaced with the experimental ones beyond the excitation energy for which experimental compilations had no definite spin and/or parity .
a state - by - state calculation of gt@xmath10 strength has been performed for a total of @xmath26 parent excited states in @xmath27fe , @xmath28 states for @xmath21fe and @xmath29 states for @xmath30fe .
for each parent excited state , transitions are calculated to @xmath31 daughter excited states .
the band widths of energy states are chosen according to the density of states to cover an excitation energy of ( @xmath32 ) mev in parent and daughter nuclei .
the summation in eq .
( [ eq:2a ] ) has been done to ensure satisfactory convergence .
the use of a separable interaction assisted in the incorporation of a luxurious model space of up to @xmath15 major oscillator shells which in turn made possible to consider these many excited states both in parent and daughter nucleus .
recently , the deformation parameter is being argued as an important parameter for qrpa calculations _ at par _ with the pairing parameter by stetcu and johnson @xcite .
as such rather than using deformations from some theoretical mass model ( as used in earlier calculations of pn - qrpa rates e.g. @xcite ) , the experimentally adopted value of the deformation parameters for @xmath20fe , extracted by relating the measured energy of the first @xmath33 excited state with the quadrupole deformation , has been taken from raman _
_ @xcite . for the case of @xmath21fe , where such measurement apparently lacks ,
the deformation of the nucleus has been calculated as @xmath34 where @xmath35 and @xmath36 are the atomic and mass numbers , respectively .
@xmath37 is the electric quadrupole moment taken from ref .
q - values are taken from the recent mass compilation of audi _
_ @xcite .
the incorporation of measured deformations for @xmath20fe and a smart choice of strength parameters led to an improvement of the calculated gt@xmath10 distributions compared to the measured ones @xcite . in table 1 of ref .
@xcite , it has been shown that the present pn - qrpa calculated gt@xmath10 centroids and total @xmath38 strengths were in good agreement with available data for the even - even isotopes of iron .
the table also shows marked improvement in the reported pn - qrpa calculation over the previous one @xcite .
the fine - grid calculation of stellar electron and positron capture rates on @xmath1fe as a function of stellar temperature and density is given in table [ tab.1 ] .
the calculated rates are tabulated in logarithmic ( to base @xmath39 ) scale .
it is simply a matter of convention to give @xmath40 for instance to express that the rate is smaller than @xmath41 .
the first column gives log(@xmath42 ) in units of g@xmath43 @xmath44 , where @xmath45 is the baryon density and @xmath0 is the ratio of the electron number to the baryon number .
stellar temperatures ( @xmath46 ) are given in @xmath47 k. stated also are the values of the fermi energy of electrons in units of mev .
@xmath48(@xmath49 ) are the electron ( positron ) capture rates in units of @xmath50 , eq .
( [ eq:2a ] ) .
core - collapse simulators may find it useful to employ the reported electron and positron capture rates on iron isotopes in their codes .
the ascii file of table [ tab.1 ] is also available and can be received from the corresponding author upon request .
a detailed comparison of reported capture rates with previous calculations over a wide range of temperature and density has been presented earlier @xcite . here
, we would like to present how the reported rates compare with previous calculations during temperature and density domains of astrophysical interest .
for the sake of comparison , we took into consideration the pioneer calculations of ffn @xcite , those performed using the large - scale shell model ( lssm ) @xcite and the recently reported thermal qrpa ( tqrpa ) approach @xcite .
an analysis of the contribution of excited states to the total capture rates for these iron isotopes was performed earlier by nabi @xcite .
there , it was shown that the electron capture rates have a significant contribution from parent excited states for the case of @xmath27fe .
the total capture rate is dominated by ground state contribution alone for the case of @xmath21fe and the case of @xmath30fe had a @xmath51-@xmath51 contribution . whenever ground state and low - lying states ( @xmath52 1 mev ) dominate the total capture rates the microscopic calculations of pn - qrpa and lssm are found to be in very good agreement .
figure [ figure1 ] depicts the comparison of electron capture rates on @xmath27fe with lssm and ffn calculations .
the graph is drawn for temperature and density domain of astrophysical importance ( oxygen shell burning and silicon core burning phases ) for the case of @xmath27fe .
the upper panel displays the ratio of calculated rates to the lssm rates , @xmath53(qrpa / lssm ) , while the lower panel shows a similar comparison with the ffn calculation , @xmath53(qrpa / ffn ) . at lower stellar temperatures and densities the calculated electron capture rates are bigger than the corresponding lssm rates by as much as a factor of four .
the results agree at high temperatures and densities . during the early stages of the collapse
the capture rates are very sensitive to the excited gt@xmath2 strength distributions .
it is reminded that lssm approach allows for detailed nuclear spectroscopy but partially employs the brink hypothesis .
the pn - qrpa model , on the other hand , performs a microscopic calculation of all low - lying gt@xmath2 strength distributions . at high densities the capture rates depend more on the centroid and total strength of the gt@xmath2 functions .
it was shown in table 1 of ref .
@xcite that the pn - qrpa calculated the centroid at a slightly higher energy in daughter @xmath27mn nucleus as against lssm centroid .
but the effect was compensated by the calculation of a much higher total strength value as compared to lssm strength .
accordingly the two results match well at high densities .
the ffn electron capture rates are on the average bigger by an order of magnitude ( lower panel of figure [ figure1 ] ) as compared to pn - qrpa rates .
there are two main reasons for this enhancement in ffn rates .
firstly ffn did not take into consideration the quenching of the gt strength .
ffn also did not take into effect the process of particle emission from excited states and accordingly their parent excitation energies extended well beyond the particle decay channel .
for these reasons ffn rates were also bigger than lssm rates by around an order of magnitude .
the electron capture rates on @xmath21fe are argued to be most effective during the oxygen shell burning till around the ignition of the first stage of convective silicon shell burning of massive stars . figure [ figure2 ] depicts the comparison of the calculated electron capture rates on @xmath21fe with lssm ( upper panel ) and ffn ( lower panel ) calculations for the corresponding physical conditions . the upper panel shows a very good comparison of pn - qrpa and lssm rates ( within a factor of @xmath54 ) .
primarily the ground state but also a few low - lying excited gt@xmath2 strength distributions play a key role in the calculation of total capture rates for this odd - a nucleus .
lssm performs a microscopic calculation of these low - lying strength functions ( up to around @xmath7 mev ) .
as mentioned earlier the ground state gt@xmath2 strength function dominates the total capture rates on @xmath21fe and the two calculations agree very well .
the ffn rates are again bigger by an order of magnitude at high temperatures .
the probability of occupation of high - lying excited states increases with stellar temperature and ffn rates are much bigger for reasons given above .
the comparison of electron capture rates on @xmath30fe with previous calculations is shown in figure [ figure3 ] .
the capture rates on this isotope are very important for the pre - supernova phase of massive stars . as such figure [ figure3 ] shows the comparison up to @xmath55 = 30 $ ] and @xmath56 = 10^{9}$ ] .
once again the comparison between pn - qrpa and lssm is very good ( for reasons mentioned earlier ) .
ffn rates are again bigger for obvious reasons .
there are two key facts to keep in mind before we present a comparison of positron capture rates for @xmath1fe .
first , of all it is to be noted that positron capture rates are tens of orders of magnitude smaller than the corresponding electron capture rates . secondly , these small numbers are fragile functions of the available phase space and can change appreciably by a mere change of @xmath8 mev in phase space calculations .
the positron capture rates are more indicative of the uncertainties in calculation of the energy eigenvalues ( for both parent and daughter states ) .
further it was shown in ref .
@xcite that the ground state rate contributed at the maximum by only about @xmath57 to the total positron capture rates .
figure [ figure4 ] shows the comparison of the calculated positron capture rates on @xmath27fe with the lssm and ffn calculation for similar physical conditions as depicted in figure [ figure1 ] .
here one sees that the pn - qrpa rates are suppressed by up to five orders of magnitude at low temperatures .
the comparison improves as temperature increases .
the comparison is similar in both upper and lower panels and does not change by increasing density by two orders of magnitude ( from @xmath58 = 10^{6}$ ] to @xmath56 = 10^{8}$ ] ) .
the comparison between pn - qrpa and lssm improves for the case of @xmath21fe ( figure [ figure5 ] ) where the rates differ at the maximum by three orders of magnitude .
the positron capture rates are in very good agreement with lssm rates at high temperatures .
still better is the comparison case for @xmath30fe ( figure [ figure6 ] ) .
the ffn rates are bigger by up to four orders of magnitude .
recently dzhioev and collaborators @xcite applied the pn - qrpa model extended to finite temperature by the thermofield dynamics formalism ( referred to as tqrpa ) and calculated electron capture rates on @xmath20fe .
it was reported by the authors that tqrpa and lssm results matched well at low temperatures and high densities . at high temperatures
the tqrpa rates were bigger than lssm electron capture rates .
table [ tab.2 ] presents a mutual comparison between pn - qrpa , tqrpa and lssm electron capture rates on these even - even isotopes of iron .
it can be seen from table [ tab.2 ] that tqrpa electron capture rates on @xmath27fe are smaller by two ( six ) orders of magnitude at @xmath55 = 1 $ ] and @xmath59 = 10^{7 } ( 10^{8})$ ] as compared to pn - qrpa and lssm rates .
the comparison improves as temperature and density increase and tqrpa rates are doubled at @xmath55 = 10 $ ] .
as far as comparison of electron capture rates on @xmath30fe is concerned , one again notes that tqrpa rates are smaller by up to seven orders of magnitude at low temperatures and densities .
however by merely changing @xmath55 = 1 $ ] to @xmath55 = 1.5 $ ] the tqrpa rates gets up to @xmath60 orders of magnitude bigger than the corresponding pn - qrpa and lssm rates .
it looks that tqrpa rates are very sensitive to temperature changes for the case of @xmath30fe at low densities .
the comparison improves with increasing stellar temperatures and densities . once again at @xmath55 = 10 $ ] the tqrpa rates are roughly double the pn - qrpa and lssm rates .
table [ tab.2 ] shows that the comparison between pn - qrpa and lssm is comparatively far better and that tqrpa model requires further refinement .
precise and reliable nuclear data can lead to deciphering the elemental and isotopic abundances in nature with more confidence and thereby derive meaningful constraints on the associated astrophysical models .
the pn - qrpa model has a good track record of calculation of weak interaction rates both in terrestrial and stellar domains .
the model has access to a huge model space making it possible to calculate weak rates for arbitrarily heavy system of nucleons .
further the model gets rid of the brink hypothesis and calculates a _ state - by - state _ calculation of stellar capture rates which greatly increases the reliability of calculated rates .
incorporation of experimental deformation lead to a much improved version of this calculation .
the model was used recently to calculate weak interaction mediated rates on iron isotopes , @xmath1fe @xcite .
electron capture on these isotopes of iron are mainly responsible for decreasing the electron - to - baryon ratio during the oxygen and silicon burning phases of massive stars .
the corresponding positron capture rates are orders of magnitude smaller and accordingly less important for late stellar evolution of massive stars .
the calculated electron and positron capture rates are also compared against previous key calculations of weak rates . during the oxygen and silicon core burning phase of massive stars , the pn - qrpa electron capture rates on @xmath27fe are around three times bigger than those calculated by lssm .
the comparison is very good for the case of @xmath61fe .
ffn electron capture rates are bigger by an order of magnitude .
orders of magnitude differences in electron capture rates exist between tqrpa and lssm / pn - qrpa model .
the former calculations of positron capture rates on @xmath1fe are @xmath62-@xmath63 orders of magnitude bigger than the pn - qrpa rates . the lssm positron capture rates on @xmath30fe is in good agreement with reported rates . the main idea of reporting
this work is to present a fine - grid calculation of electron and positron capture rates on @xmath1fe in stellar matter .
such tables ( e.g. table ii of ref .
@xcite ) are of great utility for core - collapse simulators and also suitable for interpolation purposes .
the fine - grid calculation is presented in table [ tab.1 ] .
the ascii file of this table can also be requested from the author .
core - collapse simulators are encouraged to employ these rates in simulation codes to check for possible interesting outcomes .
j .- u . nabi would like to acknowledge the support provided by twas and the local hospitality provided by the egyptian center for theoretical physics ( ectp ) , cairo - egypt , where this project was partially completed .
nabi also wishes to acknowledge the support of research grant provided by the higher education commission , pakistan , through hec project no .
20 - 1283 .
bethe , g. e. brown , j. applegate , and j. m. lattimer , nucl . phys . *
a324 * , 487 ( 1979 ) .
nabi , ph.d .
thesis , heidelberg university germany , 1999 .
d. prialnik , _ an introduction to the theory of stellar structure and evolution _
( cambridge university press , cambridge , 2000 ) . k. kotake , k. sato and k. takahashi , rep .
prog . phys . *
69 * , 971 ( 2006 ) .
t. rnnqvist et al . , nucl . phys . *
a563 * , 225 ( 1993 ) .
m. b. aufderheide , s. d. bloom , g. j. mathews and d. a. resler , phys .
c * 53 * , 3139 ( 1996 ) .
g. m. fuller , w. a. fowler , m. j. newman , astrophys .
ser * 42 * , 447 ( 1980 ) ; * 48 * , 279 ( 1982 ) ; astrophys .
j. * 252 * , 715 ( 1982 ) ; * 293 * , 1 ( 1985 ) .
j. a. halbleib and r. a. sorensen , nucl . phys .
* a98 * , 542 ( 1967 ) . s. g. nilsson , mat .
selsk * 29 * , 16 ( 1955 ) .
j. krumlinde and p. mller p , nucl . phys . *
a417 * , 419 ( 1984 ) .
k. muto , e. bender , t. oda and h. v. klapdor , z. phys .
a * 341 * , 407 ( 1992 ) .
m. b. aufderheide , nucl . phys . *
a526 * , 161 ( 1991 ) .
nabi , h. v. klapdor - kleingrothaus , at .
data nucl .
data tables * 88 * , 237 ( 2004 ) .
d. m. brink , rep .
* 21 * , 144 ( 1958 ) .
nabi , m. sajjad , phys .
c * 77 * , 055802 ( 2008 ) .
nabi , eur .
j. * a40 * , 223 ( 2009 ) .
nabi , phys .
scripta * 81 * , 025901 ( 2010 ) .
nabi , int .
e * 19 * , 1 ( 2010 ) .
nabi , adv .
81 * , 025901 ( 2010 ) .
k. langanke and g. martnez - pinedo , nucl .
a673 * , 481 ( 2000 ) .
nabi , astrophys .
space sci . * 331 * , 537 ( 2011 ) .
g. p. yost _
( particle data group ) , phys . lett . *
b204 * , 1 ( 1988 ) .
k. muto , e. bender , and h. v. klapdor , z. phys .
a * 333 * , 125 ( 1989 ) .
i. stetcu and c. w. johnson , phys .
c * 69 * , 024311 ( 2004 ) .
nabi , h. v. klapdor - kleingrothaus , at .
data nucl .
data tables * 71 * , 149 ( 1999 ) .
s. raman , c. h. malarkey , w. t. milner , c. w. nestor jr and p. h. stelson , at .
data nucl .
data tables * 36 * , 1 ( 1987 ) .
p. mller and j. r. nix , at .
data nucl .
data tables * 26 * , 165 ( 1981 ) . g. audi , a. h. wapstra and c. thibault c , nucl . phys . * a729 * , 337 ( 2003 ) .
a. a. dzhioev _
_ e__t al . ,
phys . rev .
c * 81 * , 015804 ( 2010 ) .
nabi , phys .
c * 78 * , 045801 ( 2008 ) .
fe as function of stellar temperatures and densities .
t@xmath64 gives the stellar temperature in units of @xmath47 k. in the legend , @xmath65 gives the log to base 10 of stellar density in units of @xmath66.,width=377 ] .calculated electron and positron capture rates on @xmath1fe on a fine - grid temperature - density scale .
the calculated rates are tabulated in logarithmic ( to base 10 ) scale . in the table , -100 means that the rate is smaller than 10@xmath67 . for units
see text.[tab.1 ] [ cols="^,^,^,^,^,^,^,^,^ " , ] | the acquisition of precise and reliable nuclear data is a prerequisite to success for stellar evolution and nucleosynthesis studies .
core - collapse simulators find it challenging to generate an explosion from the collapse of the core of massive stars .
it is believed that a better understanding of the microphysics of core - collapse can lead to successful results .
the weak interaction processes are able to trigger the collapse and control the lepton - to - baryon ratio ( @xmath0 ) of the core material .
it is suggested that the temporal variation of @xmath0 within the core of a massive star has a pivotal role to play in the stellar evolution and a fine - tuning of this parameter at various stages of presupernova evolution is the key to generate an explosion . during the presupernova evolution of massive stars , isotopes of iron ,
mainly @xmath1fe , are considered to be key players in controlling @xmath0 ratio via electron capture on these nuclide .
recently an improved microscopic calculation of weak interaction mediated rates for iron isotopes was introduced using the proton - neutron quasiparticle random phase approximation ( pn - qrpa ) theory .
the pn - qrpa theory allows a microscopic _ state - by - state _ calculation of stellar capture rates which greatly increases the reliability of calculated rates .
the results were suggestive of some fine - tuning of the @xmath0 ratio during various phases of stellar evolution . here
we present for the first time the fine - grid calculation of the electron and positron capture rates on @xmath1fe .
core - collapse simulators may find this calculation suitable for interpolation purposes and for necessary incorporation in the stellar evolution codes . |
we observed all the second - to - second trades and instant messages of all 66 stock day traders in a typical trading firm from 9/07 - 02/09 ( see material and methods ) .
these day traders traded only stocks and made @xmath0 million trades .
@xmath1 of their trades are live , non - computerized trades ( computerized trades were omitted from the analysis ) .
day traders typically do not hold stocks for more than a day .
they typically enter and fully exit all their positions daily , which creates a standardized measure of performance : whether the trader made or lost money at the end of each day . on average ,
these traders make money on just @xmath2 of their trades . despite sitting in the same firm , these traders generally trade independently rather than in teams .
this is because they typically trade different stocks .
one trader might trade high tech , one trader health care , another trader autos , and so on .
trading different stocks helps them diversify the firm s holdings , exploit their specialized sector knowledge , and avoid trading against each other .
this means that traders have little incentive to simply mimic each other s trades or trading behavior .
nevertheless , despite trading different stocks , traders do process common market information .
common information includes federal reserve announcements , new job figures , housing market change , speculations about bankruptcies , or other global socio - economic data that traders attempt to disambiguate by exchanging information with others as they endeavor to discover the right time to trade @xcite .
a key form of information exchange here , and increasingly in other human complex systems , is instant messaging @xcite .
instant messages among traders and their network are based on elective relationships .
each trader has the autonomy to communicate with persons of their choice .
hence , they are hierarchical ties in which orders are dictated from managers to subordinates . within the communications
, information is both professional and personal .
typical information includes interpretations of market news , expectations of where the market might be moving , rumors , and forms of personal information commonly exchanged among business friends @xcite .
our extensive field research and interviews with traders at the firm confirmed that the content of these traders messages included information consistent with the content found in other research @xcite .
all the traders in the firm exchanged instant messages throughout the day with their network .
instant messages were sent and received from their terminals or mobile devices . by federal law
all instant messages tied to trading go through the firm s capture system .
the importance of instant messages to these traders is instantiated by the intensity its use .
we analyzed the full population of @xmath3 million instant messages that our traders exchanged with their network of contacts in the industry .
to measure the synchronous trading of each trader with other traders , we defined a measure that quantifies the extent to which an individual s specific selection of time to trade is the same as the selection of other traders . to compare the synchronous trading among all traders across our observation period
, we quantified the degree to which the number of traders @xmath4 trading within the same time windows as trader @xmath5 in day @xmath6 compares to the same value when randomizing just the trades of trader @xmath5 ( fig .
1 ) . specifically , this randomization ensures normalizing individual activity , while keeping the trading structure and information heterogeneities of each specific trading day constant ( e.g. , number of traders , total number and timing of interactions , and number of interactions per trader ) .
mathematically , we defined synchronous trading as @xmath7 , where @xmath4 is the observed number of simultaneous traders and @xmath8 and @xmath9 are the average and standard deviation of simultaneous traders across an ensemble of random replicates within which the trades of trader @xmath5 were randomly shuffled . the greater the degree of a trader @xmath5 s daily @xmath10 , the greater her synchronous trading can be , and vice versa . additionally , we defined the advanced trading @xmath11 and delayed trading @xmath12 in a similar fashion as the synchronous trading at zero - time lag @xmath10 , but quantified the number of traders trading one window late and one window early , respectively .
we examined multiple time windows and reported the full analysis for 1-second windows .
this window size was chosen for the main results for several reasons .
first , the 1-second level of resolution comports well with the frenetic information environment and fast reaction time dynamics of modern markets @xcite .
the time scale in which information heterogeneities exist has increasingly shortened with the growth of computerized trading , which now accounts for between @xmath13 to @xmath14 of the trading volume on financial exchanges @xcite . in computerized trading , preset algorithms trade very large volumes of shares in hundreds of a second , which means that traders must react to market opportunities that appear and disappear on a second - by - second scale .
also consistent with the view that information moves at high speeds in modern markets and that traders react at that level , we found that the traders in this firm do display a propensity to trade on 1-second time scales .
the average empirical interval of consecutive 1-second trades is @xmath15 with a standard deviation of @xmath16 .
the maximum interval was @xmath17 and it occurred only twice in the data . similarly , reaction time research has found that human reaction occurs in less than 1-second time frames @xcite .
second , we chose 1-second intervals for synchronous trading because it is the finest , most conservative time resolution in our data .
larger than 1-second intervals require a priori knowledge to find the appropriate balance between a window large enough to encapsulate changes in slow , non - computerized information heterogeneities and yet not so large that unrelated activities appear synchronized because they occur in a large window . working empirically to estimate this balance , we tested larger than 1-second intervals .
we found that our results exist for intervals up to 15 seconds .
this window size seems to be a realistic limit for slower moving types of information and suggests that synchronicity is associated with individual thresholds that range across different information heterogeneities in this complex system @xcite .
our examination of the existence of synchronous trading revealed three interesting findings .
first , figure 2a pools all of our data and shows the probability density of synchronous trading at zero - time lag @xmath10 , advanced trading @xmath11 and delayed trading @xmath12 .
we observed that synchronous trading is significantly different ( @xmath18 using kolmogorov - smirnov test ) to advanced and delayed trading .
values greater than @xmath19 comprise @xmath20 of the entire data and are due to two traders , who could have a better access to information or better reaction times@xcite .
we conservatively omitted these outliers from our statistical analyses and found that our results did not change , confirming that synchronous trading is a special characteristic of collective behavior @xcite .
this also suggests that timing is a key factor driving the decision of traders , and reminds us about the high - frequency changes in the market @xcite .
second , figure 2b shows that the average synchronous trading @xmath21 increases with the market s daily uncertainty ( @xmath22 using markov randomizations ) , as given by the standard market volatility index , the vix @xcite .
this finding supports the idea that collective behavior is associated with uncertainty as in the case of biological systems @xcite .
the greater the level of uncertainty faced by individuals , the more likely is a collective behavior such as schooling or flocking to arise .
these findings suggest a parallel in human systems . as the level of uncertainty in the market increases ,
the more likely is synchronous behavior to occur . under high - uncertainty days
, the average synchronous trading can increase to @xmath23 , i.e. the average synchronous trading of all traders is almost two standard deviations higher than the expected by chance .
third , we found evidence that synchronous trading does not appear to be due to coordination .
unlike coordinated behavior , where pairs or sets of actors consistently align their behavior , synchronous activity commonly displays the opposite pattern @xcite .
we found that @xmath24 of all pairwise correlations between activity patterns of two different traders are non - significant @xmath25 .
this is probably because no two individuals consistently follow the same strategic behavior and the same two actors are not always correct in their assessments about the market .
similarly , if synchronous trading was driven by coordination , we should observe simultaneous trades of predominantly the same stock @xcite .
however , we find that 96% of our simultaneous trades are of different stocks . moreover ,
the trades are of different types : 60% of the simultaneous trades involve both buying and selling activities .
individual daily performance @xmath26 can be assessed by whether the trader @xmath5 loses or makes money at the end of the day @xmath6 . moreover , because the amount of money made by a trader at day s end depends on various factors , such as market volatility , number of stocks traded , and size of trades , a simple binary outcome variable appropriately standardizes their performance by considering whether the trader lost ( @xmath27 ) or made money ( @xmath28 ) .
this is coded as @xmath29 and @xmath30 , respectively .
we quantified the relationship between a trader s synchronous trading @xmath10 and performance @xmath26 with a logistic regression of the form @xmath31 ( see materials and methods ) .
we found that synchronous trading @xmath10 was significantly ( @xmath18 ) and positively associated with a trader s performance ( fig .
3 ) . using the same logistic analysis
, we compared advanced and delayed trading with end - of - day performance @xmath26 .
the results indicate that both advanced and delayed trading are statistically unrelated ( @xmath25 ) to end - of - day performance .
this reveals that synchronous trading , though arising without apparent coordination , indicates a uniquely beneficial time to trade that neither advanced nor delayed trading can reveal .
an important proposed contribution of our work is to identify those factors associated with the level of collective human behavior . in biological systems , local communication channels
have been identified as a correlate of the rise of synchronous behavior @xcite . following this line of reasoning
, we found distinctive associations between traders instant messaging patterns and synchronous trading .
first , we found that instant messaging volume is associated with trading volume throughout the day , suggesting a close connection between the two . pooling all
traders i m and trade activity over our observation period , figure 4a shows that ims have a significant correlation ( @xmath32 ) with trades over the day . on average ,
i m and trade activity rise rapidly after the 9:30am opening bell , peak at 10 am , decline at lunch time , uptick again from 1 - 3 pm , and finally decline precipitously at the 4 pm closing bell .
second , research has shown that collective synchronous activity can arise when a coupling mechanism delays or pauses the timing of individual activities @xcite .
for example , cicadas have been found to have an internal clock that stimulates chirping .
this clock would induce a cicada to regularly chirp whether or not it was exposed to the chirps of other cicadas .
synchrnous chirping arises because the internal and individual chirp activity is delayed by the chirp of another cicada , coupling the timing of the internal chirp with the collective chirps of other cicadas @xcite . to determine
if instant messages can play a coupling role , we observed whether they were associated with the rise of synchronous trading .
as noted above , instant messages play the important function of transferring information that helps traders disambiguate market information , and , because a trader can not trade and i m at the same time , instant messaging necessarily can delay his or her trades .
this would suggest that if we were to compare the observed intensity of synchronous trading of a trader relative to the synchronous trading expected by chance , as we did above , but this time we randomized the trader s trades across all 1-second time windows of the day except those 1-second time windows where there was an instant message , we would expect the intensity of synchronous trading to increase in the presence of non - random instant messages . to quantify the im - trade coupling , we compared the degree to which synchronous trading changes when randomizing a single trader s trades in any second in which an i m was not sent or received .
methodologically , we quantified the im - trade coupling by @xmath33 , where @xmath34 is the synchronous trading calculated by randomly shuffling over the seconds with no i m activity .
consistent with our conjectures , we found a positive and significant association ( @xmath32 using markov randomizations ) between the im - trade coupling @xmath35 and the synchronous trading @xmath10 of each trader ( fig .
4b , materials and methods ) .
this reveals that the instant messaging patterns of traders is associated to their trades such that the observed level of synchronous trading increases as the communication pattern is increasingly different from what would be expected by chance .
the more non - random the instant messaging pattern , the greater the synchronous trading .
this suggests that the local communication patterns of individuals have an important association with the rise of simultaneous activity , which in turn is associated to their performance .
if one assumes that ims are used to corroborate the meaning of the market throughout the day , then our findings suggest that the increasingly structured communication strategically aims to help each individual trader make a decision about when to trade .
synchronicity is a pervasive and mysterious drive in nature @xcite . in animal ,
biological , and physical systems synchronicity reduces uncertainty , as when school of anchovies evade predation , neurons co - fire to process complex information , or perturbations reduce noise in physical systems .
synchronicity also apparently arises from local interactions without the aid of centralized leadership .
this suggests that while more research needs to be done on synchronicity s functional role in complex human systems , it may furnish a functional alternative to leadership in rich informational environments .
we examined the association between synchronicity and performance in a complex system where performance increases with uncertainty reduction . examining a typical proprietary trading firm wherein
the traders individually race to be the first to disambiguate a constant stream of uncertain market information in an effort to make profitable trades , we found that when a stock trader in that firm trades at the same time as other traders in the firm , his or her financial performance is significantly increased .
we also found a coupling mechanism ; we found that traders instant message communication patterns are positively associated with the rise of sync .
building on synchronicity principles found in other complex systems , we speculate that the mechanisms underlying these empirical patterns involve rapid information aggregation through instant messaging networks . because separate traders in the firm have different instant message contacts in the market , each trader samples , separate local inferences about the eventual meaning of market information .
when these diverse points of view converge , the traders trade in synchronicity such that the synchronous timing of trades reflects a point of crowd wisdom despite no conscious intention to do so on the part of any individual trader .
these mechanisms suggest that synchronicity in human systems reflect some of the same principles found in animal systems , namely that synchronicity appears to arise with attention to local information rather than centralized leadership . in the human system
we examine , and in human systems where quick response behavior is likely a mix of being reactive and thoughtful about the information presented , we also purport that the rapid aggregation of local information from diverse points of view plays a role in the performance benefits of the synchronicity we observe .
if one assumes that each of the traders have their own expertise , training , and assumptions that go into deriving inferences from the market information they process , it suggests that actions consistent with the corroboration of diverse viewpoints are likely to be a better approximation of the true meaning of information than singular or myopic points of view .
the powerful information processing capacities of humans in complex systems may furnish unique opportunities to apply the ideas developed here about human synchronicity to other contexts .
we would speculate that in many increasingly rich information environments there are benefits to understanding synchronicity .
for example , currently in the domain of intelligence and national security , many security officers face a frenetic pace of information not unlike the traders we studied .
they too receive constant feeds of information videos , text , voice , blogs , rss newsfeeds , and tweets and are in constant communication with their own instant message network throughout the day .
moreover , like traders they are also racing to disambiguate news .
disambiguating information quickly means a potential pre - empt of an attack whereas advanced or delayed disambiguation can mean jumping the gun " or waiting too long respectively @xcite .
disease control agencies around the world all monitor large amounts of time sensitive data in attempts to identify possible outbreaks . in both situations , and more generally , in situations where information overloads might overwhelm individual decision making ability and information
is time sensitive , creating systems that can capture moments of synchronicity may help identify whether an action is functional or not . using observational data to describe this phenomenon provided a rich mix of real data but we are unable to completely test these mechanisms .
future work might devise experiments , perhaps in one of the mock trading labs now in existence in universities , by manipulating the content and rate of change of market news , providing access vs. no access to instant message communication networks , and changing the information sampled from the instant message networks from myopic to diverse .
future research might also begin to examine the potential dysfunctions of synchronicity in human complex systems . under what conditions does collective genius turn into mob madness ?
another direction for future research is to explore the differences between synchronicity and other collective behavior mechanisms .
we observed all the 66 day traders at an anonymous trading firm from 9/26/2007 - 02/20/2009 .
day traders keep short - term positions and do not hold inventories of stocks ; they enter and exit positions each day , normally between 9.30am-4:00pm .
our traders are point - and - clickers" they make trades in real - time in 98% of the time ( the 1.2% of the trades done algorithmically were omitted and did not affect the results ) .
40 - 70% of the trading on nyse is point - and - click .
we observed these traders trading approximately 4500 different stocks over various exchanges , which suggests that they sample a large part of the market .
as in most trading firms , traders do not trade everyday of every week for various reasons . similarly , in this firm , no more than 22 traders were at their desks on any one day we analyzed all the @xmath0 million intraday stock trades of these day traders and their @xmath3 million instant messages exchanged across their networks .
the performance data was calculated by the firm using standard industry metrics .
traders can not trade via ims .
to check the robustness of the association between synchronous trading and individual performance to other potential influences such as number of trades and daily uncertainty , we performed the same analysis with terms for number of trades , an interaction term for number of trades and @xmath10 , and we also controlled for market volatility ( i.e. vix ) .
the extended model has the form @xmath36 . additionally , to control for unobservable factors of each particular trader , we used fixed effects ( dummy variables for each trader ) in the logistic regression . under all circumstances ,
synchronous trading was significantly and positively associated with individual performance .
to check the robustness of the association between im - trade coupling and synchronous trading to other potential influences such as number of ims and daily uncertainty , we performed a linear regression with terms for number of ims , an interaction term for number of ims and @xmath35 , and we also controlled for market volatility ( i.e. vix ) .
the extended model has the form @xmath37 . additionally , to control for unobservable factors of each particular trader , we used fixed effects ( dummy variables for each trader ) in the regression . under all circumstances ,
im - trade coupling was significantly and positively associated with synchronous trading .
we would like to thank alex arenas , jordi bascompte , jordi duch , william kath , tae - hyun kim , eduardo lpez , hani mahmassani , dean malmgren , alejandro morales gallardo , mason porter , mark rivera , daniel stouffer , felix reed - tsochas , marta sales - pardo , uri wilensky , and the members of the nico weekly seminar series for useful discussions that led to the improvement of this work .
we also thank the kellogg school of management , northwestern university , the northwestern institute on complex systems ( nico ) for financial support .
research was also sponsored by the army research laboratory and was accomplished under cooperative agreement number w911nf-09 - 2 - 0053 .
the views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies , either expressed or implied , of the army research laboratory or the u.s .
government is authorized to reproduce and distribute reprints for government purposes notwithstanding any copyright notation here on .
steven r , galloway t , berka c , behneman a ( 2010 ) a neurophysiologic approach for studying team cognition . _
proceedings of the interservice / industry training , simulation , and education conference ( i / itsec ) _ : 1 - 8 . | successful animal systems often manage risk through synchronous behavior that spontaneously arises without leadership . in critical human systems facing risk , such as financial markets or military operations , our understanding of the benefits associated to synchronicity is nascent but promising .
building on previous work illuminating commonalities between ecological and human systems , we compare the activity patterns of individual financial traders with the simultaneous activity of other traders an individual and spontaneous characteristic we call synchronous trading .
additionally , we examine the association of synchronous trading with individual performance and communication patterns . analyzing empirical data on day traders second - to - second trading and instant messaging , we find that the higher the traders synchronous trading , the less likely they lose money at the end of the day .
we also find that the daily instant messaging patterns of traders are closely associated with their level of synchronous trading .
this suggests that synchronicity and vanguard technology may help cope with risky decisions in complex systems and furnish new prospects for achieving collective and individual goals .
ynchronous behavior has been found to enhance individual and group performance across a variety of domains even though the individuals might make no conscious effort to coordinate their behavior @xcite .
similarly , in systems of collaboration and competition , synchronous behavior can elude simple associations with individual benefits @xcite .
cicadas that chirp simultaneously with others find the best balance between risk and reward @xcite .
cicadas that chirp in advance or in delay of the full chorus relish the best chances of finding a mate but may suffer the greatest risk of being spotted by a predator @xcite .
congruently timed humans actions have been surmised to provide potential benefits revealed by the meaningful coincidence of synchronous behavior @xcite .
for instance , it has been found that simultaneous discoveries , or the times when multiple individuals arrive at a similar conclusion simultaneously , is collective evidence that a solution is valid @xcite . in this paper
, we studied the association between individual performance and the simultaneous activity patterns followed by independent decision makers under risk .
these conditions exist in many high - frequency decision contexts but are uniquely well documented in financial systems @xcite , where continuous change in information creates recurring uncertainty about when to trade and the second - to - second actions of financial traders are recorded @xcite .
reducing the risk of losing money is the essence of trading @xcite . over time
the risks of trading can decrease as information is disambiguated .
however , as this happens , the increasing certainty of information is incorporated into low - return prices .
thus , racing to be the first to discover the right time to trade is the critical problem to be solved @xcite . by analogy , this optimal timing may represent the mating sweet spot observed for cicadas .
chirp in the sweet - spot and the chance of mating / returns is relatively high and predation / losses is relatively low .
this suggests that as separate traders disambiguate their local view of news , they can spontaneously and simultaneously react as a group , without intention to coordinate , producing a synchronous behavior that might reveal the right time to trade in the market . here
, we tested whether traders performance is relatively better when trading simultaneously with other traders an individual and spontaneous characteristic we call synchronous trading .
additionally , traders need to assess whether information is positive or negative for a stock , the potential magnitude of the information s impact , and the degree to which the information is already reflected in the price @xcite .
social corroboration is key to making these assessments @xcite .
it reduces cognitive overload and ambiguity when diverse views converge on a common interpretation @xcite and typically takes place among persons tied through social network relations @xcite .
we tested whether daily instant messaging patterns of traders are associated with the rise of synchronous trading .
we believe our results can have broad implications for understanding fast collective action solutions to decision making under uncertainty . |
ultrasonic ( us ) backscattered echoes from blood contain frequency - dependent information that can be used to obtain quantitative parameters reflecting the aggregation state of red blood cells ( rbcs ) .
recently , two parameters describing rbc aggregation , the packing factor and mean fractal aggregate diameter , were extracted from the structure factor size estimator ( sfse).@xmath5 the sfse is a second - order data reduction model based on the structure factor and adapted to a dense medium such as blood .
this approach is based on the analysis of the backscattered power spectrum that contains information about the size , spatial organization , concentration and mechanical properties of scatterers ( i.e. rbcs ) .
the difficulty to apply the sfse _ in vivo _ is that the spectral content of backscattered echoes is also affected by attenuation caused by intervening tissue layers between the probe and the blood flow .
more generally , ultrasound scatterer size estimation techniques for tissue characterization ( such as liver , kidney , prostate or breast ) are facing similar challenges.@xmath6 to evaluate correctly microstructural parameters , it is thus of major interest to take into account tissue attenuation effects .
a few groups@xmath7 developed measurement techniques to evaluate the frequency - dependent attenuation in order to compensate _ a posteriori _ the backscattered power spectrum .
the goal of this letter is to further develop this strategy for _ in vivo _ measures of rbc scatterer sizes .
we propose to determine simultaneously blood structural parameters and total attenuation by using an optimization method , termed the structure factor size and attenuation estimator ( sfsae ) .
this method consists in fitting the spectrum of the backscattered radio - frequency ( rf ) echoes from blood to an estimated spectrum by a modified sfse model .
this approach is similar to that presented by bigelow _ _
et al.__@xmath8 , who estimated the effective radius of tissue microstructure and total attenuation from simulated backscattered signals . herein , _ in vitro _
experimental evaluation of the sfsae is performed .
porcine rbcs were sheared in a couette flow system , and ultrasonic rf echoes were obtained using a 25 mhz center - frequency transducer . since skin is one of the most attenuating tissue layers during _ in vivo _ scanning , three skin - mimicking phantoms with different attenuation coefficients were successively introduced between the transducer and the blood flow .
this study shows the sfsae ability to evaluate three parameters ( the packing factor , mean fractal aggregate diameter and total attenuation ) .
blood can be seen as a very dense suspension of red cells .
these rbcs can not be treated as independent scatterers since particle interactions ( collision , attraction , deformation , flow dependent motions ) are strong .
the theoretical model of ultrasound backscattering by blood that we developped@xmath5 is based on the particle approach@xmath9 , which consists of summing contributions from individual rbcs and modeling the rbc interaction by a particle pair - correlation function .
assuming that the born approximation is valid ( weak scattering ) , the model proposed in ref .
1 can be modified to predict the theoretical backscatter coefficient from blood : @xmath10 where @xmath3 is the wave vector , @xmath11 is the number density of rbcs in blood estimated by measuring the hematocrit @xmath12 by microcentrifugation ( @xmath13 , where @xmath14 is the volume of a rbc ) , @xmath15 is the backscattering cross section of a single rbc , @xmath16 is the structure factor describing the spatial organization of rbcs , and @xmath17 is the frequency - dependent attenuation function .
the backscattering cross - section @xmath15 of a weak scattering particle small compared to the wavelength ( rayleigh scatterer ) can be determined analytically as follows : @xmath18 , where @xmath19 is the variation of impedance between the rbc and its suspending medium ( i.e. the plasma ) .
the structure factor @xmath16 is by definition the fourier transform of the pair - correlation function@xmath20 @xmath21 and can be approximated by its second - order taylor expansion@xmath5 in @xmath3 as @xmath22 in this expression , @xmath23 represents the probability of finding two particles separated by a distance @xmath24 .
@xmath25 is the low - frequency limit of the structure factor ( @xmath26 ) called the packing factor.@xmath27
@xmath4 is the radius of 3d rbc aggregates assumed to be isotropic .
we introduce @xmath28 as the isotropic diameter of an aggregate ( expressed in number of rbcs ) with @xmath29 the radius of one rbc sphere - shaped model of volume @xmath14 . the attenuation function @xmath17 is given by : @xmath30 , where @xmath31 is the frequency and @xmath32 is the attenuation coefficient ( in db / mhz ) defined by :
@xmath33 , where @xmath34 and @xmath35 are respectively the intervening tissue layer attenuations ( in db / cm / mhz ) and thicknesses . according to the above equation , we thus assume , as a first approximation , that the attenuation increases linearly with the frequency : @xmath36 .
the measured backscatter coefficient reported in this study was computed as @xmath37 in eq .
( 3 ) , the mean backscattered power spectrum @xmath38 was obtained by averaging the power spectra of 400 backscattered echoes from blood .
the mean power spectrum @xmath39 was obtained from a reference sample of non - aggregated rbcs at a low hematocrit of 6@xmath0 ( i.e. rayleigh scatterers).@xmath40 in this case , 400 echoes were also averaged .
the backscatter coefficient of this reference sample @xmath41 was estimated using the `` rayleigh estimation '' approach used by yu and cloutier@xmath5 , which theoretical value is given by eq .
( 13 ) in ref . 8 ( three dimensional perkus yevick packing factor for cylinders ) .
this reference sample was used to compensate the backscattered power spectrum @xmath38 for the electromechanical system response , and the depth - dependent diffraction and focusing effects caused by the us beam .
the packing factor @xmath25 , aggregate diameter @xmath42 and total attenuation along the propagation path @xmath32 were determined by matching the measured @xmath43 given by eq .
( [ eq : bscmeas ] ) with the theoretical @xmath44 given by eq .
( [ eq : bscwiths ] ) . for this purpose , we searched values of @xmath25 , @xmath42 and @xmath32 minimizing the cost function @xmath45=@xmath46 . in all studied cases , the cost function seemed to have a unique global minimum , as was observed by plotting the cost function surface @xmath47 with varying values of @xmath32 .
an example is given in fig .
us measurements were performed in a couette flow system to produce a linear blood velocity gradient at a given shear rate ( see figure 1 in ref .
the system consisted of a rotating inner cylinder with a diameter of 160 mm surrounded by a fixed concentric cylinder of diameter 164 mm .
a 60 ml blood sample was sheared in the 2 mm annular space between the two coaxial cylinders . the us scanner ( vevo 770 , visualsonics ,
toronto , canada ) equipped with the rmv 710 probe was used in m - mode for this study .
the single - element focused circular transducer had a center frequency of 25 mhz , a diameter of 7.1 mm and a focal depth of 15 mm .
we operated at a sampling frequency of 250 mhz with 8 bits resolution ( gagescope , model 8500cs , montreal , canada ) .
the probe was mounted in the side wall of the fixed outer cylinder and was positioned to have its focal zone at the center of both cylinders . to ensure ultrasonic coupling ,
the hole within the outer stationary cylinder ( containing the probe ) was filled with a liquid agar gel based mixture .
when solidified , this gel was cut to match the curvature of the cylinder to avoid any flow disturbance .
the gel was a mixture of distilled water , 3@xmath0 ( w / w ) agar powder ( a9799 , sigma chemical , saint - louis , mo ) , 8@xmath0 ( w / w ) glycerol and a specific concentration of 50 @xmath48 cellulose scattering particles ( s5504 sigmacell , sigma chemical , saint - louis , mo ) that determined the attenuation coefficient .
four experiments were performed with four mixtures having sigmacell ( sc ) concentrations varying from 0@xmath0 to 0.75@xmath0 ( w / w ) .
the 0@xmath0 concentration constituted the non - attenuating gel and the three other mixtures mimicked skin attenuations .
the attenuation coefficients of the reference ( 0@xmath0 ) and three skin - mimicking phantoms @xmath49 were determined by using a standard substitution method .
two transducers with center frequencies of 25 mhz ( vevo 770 , visualsonics , toronto , canada ) and 20 mhz ( v317-sm panametrics , waltham , ma ) were aligned coaxially facing each other for transmission measurements .
transmitted signals were recorded both with and without the agar gel sample in the acoustic path .
the attenuation coefficient was then estimated using a log spectral difference technique.@xmath50 for a given concentration of sc , measurements were obtained from two different sample thicknesses , and for each , four regions were scanned for averaging purpose .
values obtained were 0.007 @xmath51 0.002 , 0.101 @xmath51 0.028 , 0.208 @xmath51 0.029 and 0.317 @xmath51 0.039 db / cm / mhz for sc concentrations of 0 , 0.25 , 0.50 and 0.75@xmath0 , respectively .
the thickness of the skin - mimicking phantoms @xmath52 being fixed to 1 cm , their attenuation coefficients were thus in the same range as the human dermis ( 0.21 db / mhz at 14 - 50 mhz considering a 1 mm dermis thickness@xmath53 ) .
fresh porcine whole blood was obtained from a local slaughter house , centrifuged and the plasma and buffy coat were removed .
two blood samples were then prepared : ( i ) a h6 reference sample , which was a 6@xmath0 hematocrit non - aggregating rbcs resuspended in saline solution ; and ( ii ) a 40@xmath0 hematocrit t40 test sample , which consisted of rbcs resuspended in plasma to promote aggregation .
the h6 sample was sheared at 50 s@xmath54 and coupled with the 0@xmath0 sc concentration agar gel .
echoes were selected with a rectangular window of length 0.8 mm at four depths every 0.2 mm ( i.e. with 75@xmath0 overlap between windows ) . for each depth , the power spectra of the backscattered echoes were averaged over 400 acquisitions to provide @xmath39 .
then , the h6 sample was removed and the t40 blood was introduced in the couette device . in the first 30 s ,
a shear rate of 500 s@xmath54 was applied to disrupt rbc aggregates .
the shear rate was then reduced to residual values of 2 , 5 , 10 , 20 and 30 s@xmath54 for 90 s. after that , for each shear rate , acquisitions of 400 rf lines were performed for 80 s. echoes were windowed as for the h6 sample at the same depths and their power spectra were averaged to obtain @xmath38 .
this protocol was repeated four times with the four agar - based phantoms .
the experiment with the 0@xmath0 sc phantom was realized in order to have reference results on packing factors @xmath55 and aggregate diameters @xmath56 obtained from the classical sfse.@xmath5 these parameters were assumed to be true values of packing factors and aggregate diameters for all shear rates , and will be compared in the next section with packing factors and diameters estimated by the sfsae and by the sfse when skin - mimicking phantoms are used .
it is important to emphasize the fact that the h6 reference sample was also measured with the 0@xmath0 sc phantom .
the phantom attenuation , although small with no sc , therefore affected equivalently both spectra @xmath38 and @xmath39 in eq .
the resulting measured backscatter coefficient @xmath41 was thus not biased by attenuation .
the terminology no attenuation " was used for this experiment in the following .
figure 2a reports results on @xmath55 and @xmath56 for the sfse in the case of no attenuation .
typical results of the sfsae minimization procedure for the different agar phantoms at a shear rate of 5 s@xmath54 are given in fig .
all results on @xmath25 , @xmath42 and @xmath32 from the sfsae are summarized in fig . 3 for all residual shear rates .
in this figure , the relative errors for each parameter correspond to : @xmath57 , @xmath58 and @xmath59 , with @xmath60 measured in transmissions .
more specifically , @xmath60 corresponds to @xmath61 , where @xmath62 is the skin - mimicking phantom attenuation estimated in transmission , and @xmath63 is the blood attenuation taken equal to 0.022 db / mhz@xmath5 for all shear rates . to underline the necessity to take into account the attenuation ,
parameters @xmath64 and @xmath65 were evaluated with the sfse without attenuation - compensation when skin - mimicking phantoms were used . because of the frequency - dependent distortion produced by the attenuating medium , large relative errors can be seen in fig .
4a for both parameters . however , by compensating the backscatter coefficients in the sfse with the value measured in transmission ( section 3.2 ) , relative errors in fig .
4b are largely reduced to values comparable to those estimated with the sfsae ( see fig .
3b ) . the sfsae ( fig .
3 ) gave quantitatively satisfactory estimates of @xmath25 , @xmath42 and @xmath32 with relative errors below 22@xmath0 , for shear rates between 5 and 20 s@xmath54 .
the sfse with attenuation - compensation ( fig .
4b ) gave estimates of @xmath66 and @xmath67 with relative errors below 12@xmath0 for shear rates between 2 and 10 s@xmath54 , and below 28@xmath0 for the shear rate of 20 s@xmath54 .
however , for the sfsae , the average estimates for the shear rate of 2 s@xmath54 were less accurate ( relative errors below 57@xmath0 for @xmath25 and below 30@xmath0 for @xmath32 ) .
the estimation of @xmath42 was satisfactory at that shear rate ( relative errors below 14@xmath0 ) .
the worse results of @xmath25 , @xmath42 and @xmath32 were obtained at 30 s@xmath54 for the highest attenuation . the apparent limit of applicability of the sfsae method for shear rates of 2 and 30 s@xmath54 may be explained by considering the following . at 2 s@xmath54 , for the frequency bandwidth considered ( 9 - 30 mhz ) , the sfse and consequently the sfsae seem to reach their limit of applicability for large aggregate sizes ( typically @xmath68 17.5 in fig . 2a , i.e. @xmath69 = 4.8 ) .
this limit is illustrated by the bad fit of the sfse model in fig .
2a at 2 s@xmath54 .
the bad estimations of the sfsae at 30 s@xmath54 are explained by the fact that the aggregate diameters were estimated to zero and attenuations were overestimated . at this high shear rate , rbc aggregation
is partially inhibited and the signal - to - noise ratio ( snr ) of our measurements was reduced ( @xmath70 -4 db between 20 and 30 s@xmath54 for all phantoms ) . the accuracy of the estimates was thus degraded with increasing attenuations , as can be seen from the large relative errors at the highest attenuation with the sfsae but also with the sfse with attenuation - compensation ( @xmath66 and @xmath67 ) . to conclude , the sfsae performed well for @xmath71 2.7 ( i.e. @xmath72 10 at 5 s@xmath54 ) and under the condition that the snr is sufficiently good ( snr @xmath1 28 db corresponding to the snr at 30 s@xmath54 for the 0.25@xmath0 sc ) .
although the sfsae gave less accurate estimates for 2 and 30 s@xmath54 , the estimated parameter values presented in fig .
3a show that the sfsae gave qualitatively satisfactory estimates for the three sc skin - mimicking phantoms at all shear rates , since the estimates of @xmath25 and @xmath42 versus shear rates had the same behaviors as @xmath55 and @xmath56 .
the performance of the new sfsae was assessed with experimental measurements on blood in a couette flow device .
the accuracy of the estimates obtained with the sfsae was not as satisfactory as those obtained with the sfse with attenuation - compensation ( i.e when _ a priori _ are known about the attenuation ) .
nevertheless , the sfsae has the major advantage to be easily applicable _ in vivo _ because of the simultaneous estimation of the blood structural properties and total attenuation ( contrary to the sfse attenuation - compensation method , needing the attenuation and thickness of the tissue intervening layers to be known ) .
this work thus confirms the _ in vivo _ applicability of rbc aggregate size and structure estimations .
complementary studies are nevertheless required to determine the validity domain of the sfsae according to @xmath69 and attenuation .
this work was supported by the canadian institutes of health research ( grants @xmath73mop-84358 and cmi-72323 ) , by the heart and stroke foundation of canada ( grant @xmath73pg-05 - 0313 ) , and by the national institutes of health of usa ( grant @xmath73ro1hl078655 ) .
dr cloutier is recipient of a national scientist award of the fonds de la recherche en sant du qubec .
we are also thankful to dr f. destrempes for his helpful discussion on the optimization tool .
* @xmath5f .
t. h. yu and g. cloutier , experimental ultrasound characterization of red blood cell aggregation using the structure factor size estimator " , j. acoust . soc . am . *
122 * , 645 - 656 ( 2007 ) .
* @xmath74v .
roberjot , s. l. bridal , p. laugier , and g. berger , absolute backscatter coefficient over a wide range of frequencies in a tissue - mimicking phantom containing two populations of scatterers " , ieee trans . ultras .
, ferroelect .
contr . * 43 * , 970 - 978 ( 1996 ) . *
@xmath8 t .
a. bigelow , m. l. oelze , and w. d. obrien , estimation of total attenuation and scatterer size from backscatter ultrasound waveforms " , j. acoust .
117 * , 1431 - 1439 ( 2005 ) . *
he and j. f. greenleaf , application of stochastic analysis to ultrasonic echoes - estimation of attenuation and tissue heterogeneity from peaks of echo envelope " , j. acoust .
79 * , 526 - 534 ( 1986 ) .
* @xmath76b .
j. oosterveld , j. m. thijssen , p. c. hartman , r. l. romijn , and g. j. e. rosenbusch , ultrasound attenuation and texture analysis of diffuse liver disease : methods and preliminary results " , phys .
biol * 36 * , 1039 - 1064 ( 1991 ) .
* @xmath77l .
y. l. mo and r. s. c. cobbold , theoretical models of ultrasonic scattering in blood " , in _ ultrasonic scattering in biological tissues _ , edited by k. k. shung and g. a. thieme ( crc , boca raton , fl , 1993 ) , chap . 5 , pp .
125 - 170 .
* @xmath20v .
twersky , low - frequency scattering by correlated distributions of randomly oriented particles " , j. acoust .
81 * , 1609 - 1618 ( 1987 ) .
* @xmath78k .
k. shung , on the ultrasound scattering from blood as a function of hematocrit " , ieee trans .
, ferroelect .
, freq . contr . *
su-26 * , 327 - 331 ( 1982 ) . * @xmath40s .
h. wang and k. k. shung , an approach for measuring ultrasonic backscattering from biological tissues with focused transducers " , ieee trans . biomed .
* 44 * , 549 - 554 ( 1997 ) .
* @xmath50 r. kuc and m. schwartz , estimating the acoustic attenuation coefficient slope for liver from reflected ultrasound signals " , ieee trans .
sonics ultrasonics * su-26 * , 353 - 362 ( 1979 ) .
* @xmath79 l .- c .
nguyen , f. yu , and guy cloutier , _ in vitro _
study of frequency - dependent blood echogenicity under pulsatile flow " , proc .- ieee ultrason .
2007 , in press . *
i. raju and m. a. srinivasan , high - frequency ultrasonic attenuation and backscatter coefficients of _ in vivo _ normal human dermis and subcutaneous fat " , ultrasound in med .
* 27 * , 1543 - 1556 ( 2001 ) .
* figure 1 .
( color online ) ( a ) typical aspect of the logarithm of the cost function @xmath45 for a fixed value of @xmath32 . the logarithm is shown here in order to enhance the visual contrast .
this cost function has one minimum denoted ( @xmath80 ) that depends on @xmath32 .
( b ) typical aspect of the function @xmath81 for varying values of @xmath32 ( @xmath82 and @xmath83 being calculated for each @xmath32 ) .
this cost function has a single minimum .
* figure 2 .
( color online ) ( a ) backscatter coefficients for blood sheared at different residual shear rates and measured with the 0@xmath0 sc concentration phantom ( no attenuation ) , and corresponding fitting with the classical sfse with no compensation for attenuation .
( b ) backscatter coefficients for blood sheared at 5 s@xmath54 and measured with each of the four phantoms .
the corresponding fitted models are the sfse for the 0@xmath0 sc phantom , and the sfsae for the three other skin - mimicking phantoms ( 0.25 , 0.5 and 0.75@xmath0 sc ) .
* figure 3 .
( color online ) ( a ) values of @xmath25 , @xmath42 and @xmath32 ( in db / mhz ) for different residual shear rates estimated by the classical sfse for the 0@xmath0 sc concentration and by the sfsae for the three skin - mimicking phantoms .
( b ) corresponding relative errors .
* figure 4 .
( color online ) relative errors of the packing factor and aggregate diameter for the three skin - mimicking phantoms obtained with the sfse ( a ) with no compensation for attenuation ( @xmath64 and @xmath65 ) , and ( b ) with attenuation - compensation using the attenuation values estimated in transmission ( @xmath66 and @xmath67 ) .
parameters @xmath64 and @xmath66 and similarly @xmath65 and @xmath67 are compared with @xmath55 and @xmath56 , respectively . | the analysis of the ultrasonic frequency - dependent backscatter coefficient of aggregating red blood cells reveals information about blood structural properties .
the difficulty to apply this technique _ in vivo _ is due to the frequency - dependent attenuation caused by intervening tissue layers that distorts the spectral content of backscattering properties from blood microstructures .
an optimization method is proposed to simultaneously estimate tissue attenuation and blood structure factor . in an _ in vitro _
experiment , the method gave satisfactory estimates with relative errors below 22@xmath0 for attenuations between 0.101 and 0.317 db / cm / mhz , signal - to - noise ratios@xmath128 db and @xmath22.7 ( @xmath3 being the wave number and @xmath4 the aggregate radius ) .
.5 cm pacs numbers : 43.80.qf , 43.80.cs , 43.35.bf , 43.35.yb |
a mapping @xmath0 , on a domain @xmath1 has finite distortion if the following conditions are fulfilled : 1 .
@xmath2 , 2 .
@xmath3 , 3 .
there exists a measurable @xmath4 , so that for almost every @xmath5 we have @xmath6 where @xmath7 is the operator norm .
if @xmath8 almost everywhere , we say that @xmath9 is @xmath10-quasiregular .
if @xmath11 and @xmath12 , we recover complex analytic functions .
see @xcite , @xcite and @xcite for the theory of quasiregular mappings , and @xcite , @xcite for the theory of mappings of finite distortion .
let @xmath13 be a bounded subdomain of @xmath14 with a distinguished point @xmath15 .
we will consider the collection of mappings of finite distortion @xmath0 satisfying @xmath16 and @xmath17 where @xmath18 denotes the lebesgue @xmath19-measure of @xmath13 .
moreover , we assume that there are constants @xmath20 such that @xmath21 let @xmath22 denote the class of mappings satisfying and .
the class of @xmath10-quasiregular mappings with will be denoted by @xmath23 . if @xmath24 is the class of analytic functions on the unit disc @xmath25 with @xmath26 and @xmath27 in ( * ? ? ?
* corollary 1 ) chang and marshall proved the following sharp extension of an earlier result by beurling @xcite , that is @xmath28 this result is sharp , since for beurling functions @xmath29 , @xmath30 , @xmath31 one can show that @xmath32 for every @xmath33 .
after this many natural generalizations of have been proved .
first , marshall @xcite gave another proof for using different techniques than in @xcite .
poggi - corradini @xcite was able to shorten the proof of marshall @xcite using so called egg - yolk principle " .
poggi - corradini and rajala proved the egg - yolk principle " for quasiregular mappings in space and with this they showed in ( * ? ? ?
* theorem 1.1 ) that for @xmath34 we have @xmath35 where the constant @xmath36 is sharp in dimension @xmath11 . here
@xmath37 and @xmath38 denote the unit ball and its boundary , respectively .
note that mappings @xmath39 can be extended to @xmath40 using the fact that @xmath9 has radial limits a.e . on @xmath38 .
recently it was shown by kkinen and rajala @xcite that extends to the class @xmath41 in the form @xmath42 where @xmath43 .
this result is sharp in terms of the exponent @xmath44 but the sharp constant @xmath45 is not known even in dimension @xmath11 , see discussion in ( * ? ? ?
* section 4 ) .
the purpose of this paper is to prove and for more general domains than the unit ball .
we will consider domains @xmath13 that are uniform , the justification for this restriction is given in section 3 .
the boundary of a uniform domain is known to have hausdorff dimension strictly less than @xmath19 but for technical reasons we will assume more regularity than this . if @xmath46 is a positive borel measure we say that @xmath47 is upper @xmath48-ahfors regular with respect to @xmath46 if there exist constants @xmath49 and @xmath50 such that for any @xmath5 @xmath51 our main theorems are the following : [ thm : regularity of boundary extension for qr ] let @xmath52 , @xmath34 , be a @xmath53-uniform domain with center @xmath54 and @xmath46 any positive borel measure satisfying .
then there exists a constant @xmath55 such that @xmath56 where @xmath57 is the extension of @xmath9 to @xmath47 given in definition [ def : natural boundary extension ] .
[ thm : regularity of boundary extension ] let @xmath52 , @xmath34 , be a @xmath53-uniform domain with center @xmath54 and @xmath46 any positive borel measure satisfying . then there exists a constant @xmath58 such that @xmath59 where @xmath57 is the extension of @xmath9 to @xmath47 given in definition [ def : natural boundary extension ] .
these results are sharp with respect to the exponent inside the exponential map since @xmath37 is a particular example of a uniform domain .
the sharp constants @xmath60 and @xmath45 remain open , we would need new techniques for this problem .
the general idea behind the proof is similar to the one used in the previous papers @xcite , essentially relying on a version of a theorem of beurling @xcite originally stated for analytic functions . allowing more general domains produces a two - fold problem : first we have to find a way to extend our mapping to the boundary and also we need a new version of the beurling theorem
. it could be possible to deduce theorem [ thm : regularity of boundary extension ] from earlier works by cianchi @xcite and @xcite as follows : one first uses the orlicz - sobolev extension in uniform domains for each of the component functions @xmath61 , @xmath62 , to obtain a global orlicz - sobolev function @xmath63 ; then we can choose a quasicontinuous representative of @xmath64 , which means outside a set of small capacity , @xmath64 is continuous .
this in return gives us a natural way to extend our original function @xmath61 to the boundary of @xmath13 as @xmath65 . now extending results for orlicz - sobolev functions in @xcite to similar ones given for lorentz - sobolev functions in ( * ? ? ?
* theorems 2.4 and 2.5 ) and using above arguments one can show that theorem [ thm : regularity of boundary extension ] is valid for each component function @xmath66 .
so theorem [ thm : regularity of boundary extension ] follows once we set @xmath67 . in this way
, one could prove theorem [ thm : regularity of boundary extension ] with a constant @xmath60 independent of @xmath68 and @xmath69 and thus can not obtain the sharp constant since the sharp constant depends on the distortion @xmath10 ; see @xcite .
we believe that the techniques used here can be further extended to more irregular domain than uniform domains since most of our results can be generalized to domains satisfying _ quasihyperbolic boundary condition _
@xcite ) .
this paper is organized as follows .
section 2 fixes the notation . in section 3
we will show that @xmath9 can be extended to the boundary of a uniform domain along john - curves and that this extension is well - defined with a small exceptional set on @xmath47 in terms of hausdorff gauges .
section 4 contains the main ingredients of the proofs of theorems [ thm : regularity of boundary extension for qr ] and [ thm : regularity of boundary extension ] , which is the capacity estimate given in theorem [ thm : main thm ] .
finally in section 5 we will prove our main theorems .
a curve @xmath70 in @xmath13 is a continuous mapping @xmath71\to\omega$ ] .
the trace @xmath72)$ ] of a curve @xmath70 is denoted by @xmath73 .
a curve @xmath70 is said to connect points @xmath74 if @xmath75 and @xmath76 and points @xmath5 , @xmath77 if @xmath76 , @xmath78)\subset\omega$ ] and @xmath79 we also use the notation @xmath80 to denote the ( euclidean ) length of a curve @xmath70 . the quasihyperbolic metric @xmath81 in a domain @xmath82 is defined as @xmath83 where the infimum is taken over all rectifiable curves @xmath70 in @xmath13 which join @xmath84 to @xmath85 .
this metric was introduced by gehring and palka in @xcite .
a curve @xmath70 connecting @xmath84 to @xmath85 for which @xmath86 is called a quasihyperbolic geodesic .
quasihyperbolic geodesics joining any two points of a proper subdomain of @xmath14 always exists ; see ( * ? ? ?
* lemma 1 ) . given two points @xmath87 , we denote by @xmath88 $ ] any quasihyperbolic geodesic that joins @xmath84 and @xmath85 .
let @xmath1 be a domain .
we will denote by @xmath89 a whitney type decomposition of @xmath13 , where @xmath90 is a collection of closed cubes @xmath91 with pairwise disjoint interiors and having edges parallel to coordinate axis , such that @xmath92 moreover , there exists @xmath93 such that the cubes dilated by factor @xmath94 have uniformly bounded overlap , the diameters of @xmath95 take values in @xmath96 and satisfy condition @xmath97 we will also use the notation @xmath98
in the previous papers @xcite , @xcite , @xcite and @xcite the boundary extension of @xmath99 was defined via the radial limits of @xmath9 that exist at almost every point in @xmath38 by absolute continuity on lines , and . in this section
we prove similar results for more general class of domains , namely for uniform domains .
when considering general bounded domains @xmath13 there might be big parts of the boundary @xmath100 that are not accessible by a rectifiable curve inside @xmath13 , see figure [ fig : topocomb ] .
to avoid such situations , we restrict our considerations to john domains . a bounded domain @xmath52 is a @xmath53-john domain with distinguished point @xmath15 ( center ) if every point @xmath5 can be joined to @xmath54 by a @xmath53-john curve .
recall also that a rectifiable curve @xmath101\to\omega$ ] parametrized by arc length is called a @xmath53-john curve joining @xmath84 to @xmath54 if it satisfies @xmath102 , @xmath103 and @xmath104 for every @xmath105 $ ] .
it follows from the lower semicontinuity of @xmath80 and ascoli - arzela theorem that each boundary point can be connected to @xmath54 by a curve satisfying .
thus every point on the boundary is accessible , even in a non - tangential manner .
notice that typically there will be uncountably many @xmath53-john curves between @xmath106 and @xmath77 . for the rest of this subsection the standing assumptions
are : let @xmath13 be a @xmath107-john domain with center @xmath54 , @xmath90 a whitney decomposition of @xmath13 and @xmath108 .
if @xmath109 then define @xmath110 as the collection of all @xmath53-john curves connecting @xmath111 to @xmath54 and @xmath112 for @xmath95 and @xmath113 , we define the @xmath53-shadow of @xmath114 on @xmath115 by @xmath116 when @xmath117 , we write @xmath118 instead of @xmath119 .
we will omit the sub / superscripts @xmath53 in the above notation if @xmath120 .
the following two lemmas are well - known and the proofs can be found , for instance , in ( * ? ? ?
* section 4 ) .
[ shadowproperties ] let @xmath95 and @xmath109 .
then @xmath118 is closed and there exists a constant @xmath121 such that @xmath122 also there exists a constant @xmath123 such that for any @xmath124 @xmath125 [ johnmeas ] assume that @xmath126 is a borel measure on @xmath47 and @xmath127 measurable .
then there exists a constant @xmath123 such that , for each @xmath124 we have @xmath128 our aim is to prove that mappings in the classes @xmath129 and @xmath130 have limits along @xmath107-john curves emanating from the center @xmath54
. this will be true for big pieces of the boundary in terms of hausdorff dimension .
for this we introduce a notion of discrete length of a curve : assume that @xmath109 and @xmath131 . given a continuous mapping @xmath0 we define the discrete length of @xmath132 by @xmath133=\sum_{\substack{q\in\mathscr{w}\\ q\cap \gamma\neq\emptyset } } \operatorname{diam}{f(q)}.\ ] ] it follows from the definition and lemma [ shadowproperties ] that if @xmath134<\infty$ ] then the limit @xmath135 exists .
we also need oscillation estimates for mappings of finite distortion @xmath0 .
the proof of the next lemma is contained in @xcite , see also @xcite .
[ oscilla ] let @xmath0 be a mapping of finite distortion and assume @xmath136 .
then we have the following : a. if @xmath9 is quasiregular , then @xmath137 for every @xmath138 for which @xmath139 . here
constant @xmath50 depends only on @xmath19 and @xmath10 .
b. if @xmath9 has exponentially integrable distortion , then @xmath140 for every @xmath138 for which @xmath139 .
here constant @xmath50 depends only on @xmath19 , @xmath68 and @xmath141 .
now we are ready to state a theorem which tells that a mapping of finite distortion can be extended to @xmath100 along john curves with a small exceptional set .
[ thm : extension ] assume that @xmath0 is a mapping of finite distortion with @xmath142 .
let @xmath143 be the set of points @xmath144 for which there exists a curve @xmath145 so that @xmath9 does not have a limit along @xmath70 .
then we have the following : a. if @xmath9 is quasiregular and @xmath146 is a doubling gauge function satisfying @xmath147 then @xmath148 . b. if @xmath9 has exponentially integrable distortion and @xmath146 is a doubling gauge function satisfying @xmath149 then @xmath148 .
we will follow the proof of theorem 1.1 in @xcite and actually prove that @xmath150 , where @xmath151 is the set of points in @xmath152 for which there is a curve @xmath153 so that @xmath132 has infinite discrete length . note that @xmath154 . on the contrary ,
assume that @xmath155
. then @xmath156 , where @xmath157 is the set of points in @xmath109 for which there exist @xmath153 so that @xmath158\geq k.$ ] then by frostman s lemma there exists a borel measure @xmath126 supported in @xmath157 so that @xmath159 for every @xmath160 and @xmath161 by the definition of @xmath157 @xmath162\,d\mu(w)\\ & \leq\int_{a_k}\sum_{\substack{q\in\mathscr{w}\\ q\cap\gamma_w\neq\emptyset}}\operatorname{diam}{f(q)}\,d\mu(w).\end{aligned}\ ] ] from now on , we assume that @xmath9 has exponentially integrable distortion .
let @xmath93 be the constant in the definition of whitney decomposition @xmath90 for which the cubes @xmath163 dilated by a factor @xmath94 have uniformly bounded overlap .
if @xmath164 then set @xmath165 , where @xmath166 is the center of @xmath114 and @xmath167 .
using lemma [ oscilla ] for balls @xmath168 we have the following chain of inequalities : @xmath169 above we also used the fact that @xmath170 for @xmath171 and also the property that @xmath172 s have uniformly bounded overlap .
the left hand side is approximated using lemmas [ shadowproperties ] and [ johnmeas ] together with the doubling property of gauge function @xmath146 @xmath173 putting these estimates together gives @xmath174 since @xmath146 satisfies we know that the sum on the right hand side is finite ( and independent of @xmath175 ) and thus @xmath176 for every @xmath177 and thus @xmath178 as @xmath179 , which is a contradiction .
thus we have @xmath150 and the claim follows . for quasiregular @xmath9 we do not get the term @xmath180 in and thus a gauge satisfying gives the desired conclusion .
as a corollary of theorem [ thm : extension ] we get that the set @xmath143 has hausdorff dimension zero as well as the set @xmath151 .
moreover , for a quasiregular @xmath9 we have @xmath181 for every @xmath182 , where @xmath183 .
also , if @xmath9 has exponentially integrable distortion , then @xmath181 for every @xmath184 . in the quasiregular case
this means that @xmath143 is of conformal capacity zero in terms of gauge functions , see @xcite .
assume that @xmath13 is a @xmath107-john domain .
by the previous section we have a natural candidate for the extension of @xmath185 to @xmath47 , that is @xmath186 so that @xmath187 where @xmath188 is a @xmath107-john - curve connecting @xmath54 to @xmath189 .
the problem with this extension is that it might not be well - defined even for a smooth mapping @xmath190 .
let @xmath191 , where @xmath192 is the ( euclidean ) line segment connecting @xmath193 to @xmath194 .
then @xmath195 is a @xmath53-john domain with @xmath196 .
set @xmath197 and @xmath198 .
there is a smooth mapping @xmath199 such that @xmath200 and @xmath201 and thus the extension @xmath57 is not well - defined for any point in @xmath202 , where @xmath203 . in order to exclude such domains we have to impose more conditions on @xmath13 .
examples like this do not work if we strengthen our assumption from john condition to @xmath13 being uniform .
a bounded domain @xmath52 is a @xmath53-uniform domain if there exists a constant @xmath204 such that each pair of points @xmath205 can be joined by a @xmath53-uniform curve @xmath70 in @xmath13 , namely a curve @xmath70 with the following two properties : @xmath206 @xmath207 these domains first appeared in the work of gehring and osgood @xcite .
later these domains were used in many different contexts such as sobolev extension domains and uniformizing gromov hyperbolic spaces , for these see @xcite . it should be noticed that quasihyperbolic geodesics between two points in a @xmath53-uniform domain are @xmath208-uniform curves with @xmath209 @xcite .
we say that @xmath13 is a john ( or uniform ) domain if it is @xmath53-john ( or @xmath53-uniform ) for some @xmath204 .
we may assign for each @xmath53-uniform domain a center point @xmath54 . indeed ,
if @xmath13 is a @xmath53-uniform domain , then we may fix a point @xmath15 such that @xmath13 is a @xmath208-john domain with center @xmath54 .
moreover , the john constant @xmath208 depends only on @xmath53 @xcite . since we are studying boundary extension of mappings , it is convenient to introduce the following concept .
[ def : uniform domain with center ] we say that @xmath13 is a @xmath107-uniform domain with center @xmath54 if @xmath13 is a uniform domain and as a john domain , it is @xmath107-john with center @xmath54 . as was observed in section [ subsec : exist limits ] , for a given point @xmath210 in a john domain @xmath13 , there are typically several john curves connecting @xmath84 to @xmath54
. thus it might happen that @xmath185 has different limits at @xmath144 along different john curves in @xmath211 ; see for instance figure [ fig : counterjohn ] . in the rest of this section
, we will show that if @xmath13 is @xmath107-uniform domain with center @xmath54 , then mappings in @xmath22 can be extended to @xmath47 in a unique way along curves in @xmath211 .
[ thm : uniqueness of limit in lipschitz domains ] let @xmath52 be a @xmath107-uniform domain with center @xmath54 , @xmath185 and @xmath146 a doubling gauge function satisfying
. then @xmath9 has a unique limit along curves @xmath145 for @xmath212-a.e .
@xmath213 , i.e. if @xmath214 so that @xmath215 then @xmath216 .
we need the following two lemmas to prove theorem [ thm : uniqueness of limit in lipschitz domains ] .
[ lemma : product tends to zero ] let @xmath13 be a @xmath107-john domain with center @xmath54 and @xmath108 .
if @xmath9 has exponentially integrable distortion and @xmath217 is the set of points @xmath109 for which there exists a curve @xmath218 so that @xmath219 then @xmath220 , where @xmath146 is as in . above
, @xmath221 is defined in the following way : for a fixed @xmath218 connecting @xmath54 to @xmath222 , we denote by @xmath223 those cubes that intersect @xmath70 and labeled in the order from @xmath54 to @xmath111 , and set @xmath224 notice that if @xmath225 , then @xmath226 let @xmath157 be the collection of points @xmath109 for which there exists @xmath227 so that labeling as in the lemma we have @xmath228 then @xmath229 and thus it is enough to show that @xmath150 .
towards a contradiction assume the opposite @xmath155 and denote by @xmath126 the frostman measure whose support lies in @xmath157 , which satisfies @xmath230 and @xmath231 for every ball @xmath232 .
then we may follow the proof of theorem [ thm : extension ] to obtain @xmath233 with a universal constant @xmath50 .
thus we have a contradiction and the lemma is proved .
[ lemma : chaining1 ] let @xmath1 be a @xmath107-uniform domain with center @xmath54 and @xmath234 a whitney decomposition of @xmath13 .
then there exists a constant @xmath235 such that for any @xmath236 and pair @xmath237 with @xmath238 , @xmath239 and @xmath240 there exists a chain of whitney cubes @xmath241 connecting points @xmath84 and @xmath242 such that the number of cubes is uniformly bounded with respect to @xmath48 , i.e. @xmath243 , where @xmath235 depends only on @xmath19 , @xmath107 and @xmath208 but not on @xmath48 . moreover
if @xmath244 for some @xmath245 then @xmath246 for any @xmath247 from this collection with @xmath248 .
let @xmath236 and fix @xmath237 with @xmath238 , @xmath239 and @xmath240 .
connect @xmath84 to @xmath242 by a quasihyperbolic geodesic @xmath249_k$ ] and enumerate the cubes @xmath250 , that intersect @xmath249_k$ ] , in a such way that @xmath251 , @xmath252 and @xmath253 for every @xmath254 .
by @xcite we know that the number of cubes in this chain is comparable to @xmath255 and moreover , since @xmath13 is uniform , by ( * ? ? ?
* lemma 2.13 ) we have a constant @xmath256 such that @xmath257 the first part of the lemma follows .
next we prove the latter part .
since @xmath244 for some @xmath245 there exists a curve @xmath258 such that @xmath259 .
let @xmath260 .
let @xmath247 be any other whitney cubes from the collection and fix @xmath261 .
let @xmath262 denote the piecewise linear curve from @xmath85 to @xmath263 s.t .
we define curve @xmath265 as follows .
we first travel from @xmath111 along @xmath70 to @xmath85 and denote this by @xmath266 .
next , follow @xmath262 from @xmath85 to @xmath263 and denote this by @xmath267 , then go back from @xmath263 to @xmath85 along @xmath262 and denote this by @xmath268 .
finally , follow @xmath70 from @xmath85 to @xmath54 and denote this by @xmath269 . in this way we obtained curve @xmath270 connecting @xmath111 to @xmath54 . to illustrate that @xmath265 is a @xmath271-john curve with @xmath248 , we assume that @xmath272\to \omega$ ] is parametrized by arc length and we have to verify with @xmath271 .
for @xmath273 $ ] we have @xmath274 since @xmath258 .
next assume that @xmath275 $ ] .
first note that all the whitney cubes in the collection have uniformly comparable size since the number @xmath276 is bounded from above uniformly by @xmath235 and adjacent whitney cubes have comparable size and thus we may estimate @xmath277 using the above estimates , we have @xmath278 last , assume that @xmath279 $ ] .
now we have @xmath280 , which imply @xmath281 and thus @xmath282 let @xmath217 denote the set defined in lemma [ lemma : product tends to zero ] .
we know that @xmath220 .
let @xmath283 and @xmath284 be such that @xmath285 we will show that @xmath9 has the same limit along any curve @xmath286 .
fix such a curve @xmath287 .
given @xmath288 , let @xmath289 $ ] be such that @xmath290 and @xmath291 . since @xmath292 are both @xmath107-john curves this implies that @xmath293 we now apply lemma [ lemma : chaining1 ] with @xmath294 and for points @xmath295 and @xmath296 to find a finite sequence of whitney cubes @xmath297 with @xmath243 .
choose @xmath298 so that @xmath299 and @xmath300
. then by lemma [ oscilla ] we have @xmath301 by lemma [ lemma : chaining1 ] there is a constant @xmath302 s.t .
@xmath303 for every @xmath304 .
note that @xmath276 does not depend on @xmath305 .
it follows from lemma [ lemma : product tends to zero ] that we may choose @xmath288 so small that @xmath306 this together with shows that @xmath307 as @xmath308 . from this
we may conclude that @xmath309 and the proof is complete .
based on theorem [ thm : uniqueness of limit in lipschitz domains ] , we introduce the following definition .
[ def : natural boundary extension ] let @xmath13 be a @xmath107-uniform domain with center @xmath54 and @xmath310 .
we denote by @xmath311 the unique limit of @xmath9 at @xmath312 along @xmath107-john curves .
notice that by theorem [ thm : extension ] and theorem [ thm : uniqueness of limit in lipschitz domains ] , @xmath57 is well - defined @xmath212-a.e . in @xmath313 .
note that the unit ball @xmath314 is a 1-uniform domain with center at the origin . according to definition [ def : natural boundary extension ]
, @xmath315 is the unique limit of @xmath9 along the unique 1-john curve , which is just the line segment @xmath316 , thus in this case our definition [ def : natural boundary extension ] coincides with the radial extension .
in this section , we prove the following weighted capacity estimates in john domains . the techniques used here are rather standard and
have been used effectively in @xcite . on the other hand
, they give a new proof of the key level set estimate , namely theorem 1.2 in @xcite , where @xmath13 is the unit ball @xmath317 .
recall that for disjoint compact sets @xmath115 and @xmath318 in @xmath319 , we denote by @xmath320 the @xmath321-weighted capacity of the pair @xmath322 : @xmath323 where the infimum is taken over all continuous functions @xmath324 which satisfy @xmath325 for @xmath326 and @xmath327 for @xmath328 . [ thm : main thm ] let @xmath1 be a @xmath107-john domain with diameter one .
let @xmath329 be a compact set satisfying @xmath330 with @xmath331 .
then @xmath332 where @xmath333 is the central whitney cube containing the john center @xmath54 . moreover , if @xmath334 $ ] be a weight which is exponentially integrable , then @xmath335 the constant @xmath235 in and depends only on the data associated to @xmath13 and @xmath321 .
we only prove the weighted estimate , since the proof of the unweighted case is similar ( and simpler ) .
let @xmath336 be a test function for the @xmath337-weighted @xmath19-capacity of the pair @xmath338 .
for each @xmath326 , we may fix a @xmath107-john curve @xmath70 joining @xmath84 to @xmath54 in @xmath13 and define @xmath339 to be the collection of whitney cubes that intersect @xmath70 .
we define a subcollection @xmath340 as follows : @xmath341 consists of a chain of whitney cubes , which continue along @xmath339 until it reaches the last cube @xmath342 for which @xmath343 . since adjacent whitney cubes @xmath114 and @xmath344 have @xmath345
, we must have @xmath346 for all @xmath347 .
we first assume that for every @xmath326 we have @xmath348 for @xmath349 .
we claim that @xmath350 if @xmath326 and @xmath351 then @xmath352 and thus by lebesgue differentiation theorem we have @xmath353 . the standard chaining argument involving poincar inequality
( see for instance @xcite ) gives us the estimate @xmath354 note that for each @xmath91 , by hlder s inequality , @xmath355 notice that the function @xmath356 is convex and we may use jensen s inequality to deduce that @xmath357 where @xmath358 . plugging the above estimate in , we obtain @xmath359 by our assumption @xmath330 and thus we can choose a frostman measure @xmath126 on the compact set @xmath115 with growth function @xmath360 i.e. @xmath126 is a borel measure on @xmath14 which satisfies the growth condition @xmath361 for every ball @xmath232
. integrating over the set @xmath115 with respect to the frostman measure @xmath126 , we see that @xmath362 we now interchange the order of summation and integration to deduce that @xmath363 applying hlder s inequality to the right hand side leads to the following upper bound for @xmath364 @xmath365 the sum in the previous estimate can be dealt with in the following way : choose @xmath366 with @xmath367 so that @xmath346 implies that @xmath368 for some @xmath369 .
we claim that @xmath370 the growth condition on @xmath126 and lemmas [ shadowproperties ] and [ johnmeas ] imply that @xmath371 inequality follows from this since the sum on the right hand side converges .
putting together all the estimates above yields @xmath372 it follows that @xmath373 to get from here we just note that by concavity of @xmath146 @xmath374 and thus we have @xmath375 we are left with the case that there is @xmath326 so that @xmath376
. in this case , we have @xmath377 and @xmath376 .
by following the quasihyerbolic geodesic from @xmath378 to @xmath54 , we can find a finite chain of whitney cubes @xmath379 which joins @xmath333 to @xmath380 , namely , cubes @xmath381 such that @xmath382 is the central cube , @xmath383 .
moreover , @xmath384 where the last inequality follows from the fact that the length of a quasihyperbolic geodesic is comparable with the number of whitney cubes it intersects . then repeating the chaining argument for finite chain of cubes @xmath385 gives the estimate @xmath386 from which we obtain @xmath387 since @xmath388 . in theorem
[ thm : main thm ] we have assumed that @xmath13 is a john domain .
however , the proof works for more general domains .
more precisely , it hold if @xmath13 satisfies the quasihyperbolic boundary condition @xcite .
the only difference is that in theorem [ thm : extension ] , one has to consider limits along , instead of john curves , quasihyperbolic geodesics to the boundary .
[ rmk : on weighted capacity in quasihyperbolic boundary condition ] note that , in theorem [ thm : main thm ] , we have assumed that @xmath389 .
this assumption was only used to ensure that @xmath390 is positive . for domains with different diameter
, one just needs to replace the right - hand side of with @xmath391 where the constant @xmath235 in front now depends additionally on @xmath13 as well .
similar changes applies to .
another observation one has to notice is that if we replace the central whitney cube @xmath333 by a ball @xmath392 in the capacity estimates in theorem [ thm : main thm ] , then the estimates and become @xmath393 and @xmath394 respectively , where the constant @xmath235 in both and now depends additionally on @xmath13 and @xmath395 . indeed ,
notice that if @xmath396 , then @xmath397 .
so we may repeat the proof of theorem [ thm : main thm ] by considering two cases : * for every @xmath326 , @xmath398 for @xmath349 ; * there exists a point @xmath326 so that @xmath399 .
the proof of the first case proceeds identically as before since @xmath400 . in the second case , since @xmath399 and @xmath401
, we may proceed again as in the proof of theorem [ thm : main thm ] using the same finite chaining argument .
we are ready to prove the main results .
the proof is similar than the one given in ( * ? ? ?
* section 2 ) .
let @xmath13 be a @xmath107 uniform domain with center @xmath54 .
we will first prove the case @xmath185 , the proof for @xmath402 is analogous .
assume that @xmath46 is a positive borel measure satifying for @xmath49 .
since @xmath47 is compact it follows from that @xmath403 furthermore , together with the finiteness of @xmath404 imply that for any borel set @xmath127 we have @xmath405 we want to show that there are constants @xmath406 and @xmath407 independent of @xmath9 such that @xmath408 where @xmath57 is the boundary extension given in definition [ def : natural boundary extension ] . by fubini s theorem @xmath409 where @xmath410 .
thus by it is enough to bound @xmath411 define @xmath412 and @xmath413 since the functions in @xmath22 are equicontinuous , we can find @xmath414 such that @xmath415 for any @xmath416 and @xmath417 . here
@xmath395 depends only on @xmath418 .
let @xmath419 denote the collection of curves @xmath71\to\mathbb{r}^n$ ] that connect points in @xmath420 to points in @xmath421 .
then by ( * ? ? ?
* proposition 10.2 ) @xmath422 similarly as in @xcite using the test function @xmath423 when @xmath424 and @xmath425 otherwise for the weighted modulus we have the following upper bound @xmath426 from theorem [ thm : main thm ] we have a lower bound @xmath427 see also remark [ rmk : on weighted capacity in quasihyperbolic boundary condition ] .
these together with the estimate @xmath428 following from hlder s inequality , see @xcite , imply that @xmath429 from which we conclude that for @xmath430 we have @xmath431 where the upper bound does not depend on @xmath9 .
the authors would like to thank academy professor pekka koskela and professor kai rajala for many helpful discussions on the topic .
we are very grateful to the anonymous referee for many useful comments that improve our exposition . | in this paper , we consider mappings on uniform domains with exponentially integrable distortion whose jacobian determinants are integrable .
we show that such mappings can be extended to the boundary and moreover these extensions are exponentially integrable with quantitative bounds .
this extends previous results of chang and marshall @xcite on analytic functions , poggi - corradini and rajala @xcite and kkinen and rajala @xcite on mappings of bounded and finite distortion . |
the redshifted neutral hydrogen ( hi ) 21-cm radiation is present as a background radiation in all low - frequency radio observations below @xmath1 . a statistical detection of the random fluctuations in this background radiation present an unique technique for probing the large scale structures in the universe during the post - reionization era @xmath2 ( bharadwaj , nath & sethi 2001 ; bharadwaj & sethi 2001 ) .
observations of high redshift quasars show that the diffuse inter - galactic medium ( igm ) is highly ionized at redshifts @xmath3 , and the bulk of the hi in the post - reionization era resides in discrete , dense hi clouds with column densities greater than @xmath4 ( wolfe et al .
1995 ; lanzetta , wolfe & turnshek 1995 ; storrie - lombardi , mcmahon & irwin 1996 ; proux et al .
several measurements of @xmath5 and @xmath6 have been carried out both at low and high redshifts . at low redshifts ( @xmath7 and lower ) we have measurements of @xmath5 from hi galaxy surveys ( zwaan et al .
2005 ; martin et al . 2010
; delhaize et al . 2013 ) and hi stacking ( lah et al . 2007 ; rhee et al . 2013 , 2016 ) .
observations of the damped ly@xmath8 ( dla ) systems provide us measurements of @xmath6 for both low redshifts ( @xmath7 , rao , turnshek & nestor 2006 ; meiring et al .
2011 ) and high redshifts(@xmath9 , prochaska & wolfe 2009 ; noterdaeme et al .
these observations tell us that the hi content of the universe remains almost constant @xmath10 over the entire redshift range @xmath11 ( lanzetta et al .
1995 ; storrie - lombardi et al . 1996
; rao & turnshek 2000 ; proux et al .
2003 ) . unlike traditional galaxy redshift surveys which need to identify individual galaxies in order to map out the large scale structures in the universe
, the proposed 21-cm surveys do not need to identify the individual hi sources .
the collective redshifted 21-cm radiation from the individual hi clouds appears as a diffuse background radiation and the source clustering is imprinted as the fluctuations in this background radiation .
future redshifted 21-cm observations are currently perceived to be a very promising probe of the formation and evolution of large - scale structures in the post - reionization era ( bharadwaj & pandey 2003 ; bharadwaj & ali 2005 ; wyithe & loeb 2008 ) .
such observations hold the potential of measuring the baryon acoustic oscillation ( bao ) which is a very sensitive probe of dark energy ( wyithe , loeb & geil 2008 ; chang et al . 2008 ; seo , dodelson & marriner 2010 ; masui , mcdonald & pen 2010 ) .
further , a measurement of just the 21-cm power spectrum can also be used to constrain cosmological parameters ( visbal , loeb & wyithe 2009 ; bharadwaj , sethi & saini 2009 ) .
the giant meterwave radio telescope ( gmrt , swarup et al . 1991 ) is sensitive to the cosmological hi signal from a range of redshifts in the post - reionization era ( bharadwaj & pandey 2003 , bharadwaj & ali 2005 ) and ghosh et al .
( 2011a , b ) have carried out preliminary observations towards detecting this signal from @xmath12 .
the upgraded ugmrt is expected to have a larger bandwidth for which the prospects of a detection are presented in chatterjee et al .
the canadian hydrogen intensity mapping experiment ( chime , newburgh et al .
2014 , bandura et al .
2014 ) aims to measure the bao in the redshift range @xmath13 .
the future tianlai ( chen 2012 , 2015 ) and ska1-mid ( bull et al .
2015 ) also aim to measure the redshifted hi 21-cm signal from the post - reionization era .
the ooty radio telescope ( ort ; swarup et al .
1971 ) is currently being upgraded ( prasad & subrahmanya 2011 , subrahmanya et al .
2017b ) to a linear radio - interferometric array , the ooty wide field array ( owfa ; subrahmanya , manoharan & chengalur 2017a ) .
the owfa operates at a nominal frequency of @xmath14 which corresponds to the redshifted 21-cm signal from a redshift @xmath15 .
measuring the redshifted hi 21-cm power spectrum is one of the major science goals of owfa ( subrahmanya et al .
theoretical estimates ( bharadwaj , sarkar & ali 2015 ) predict that it should be possible to measure the amplitude of the 21-cm power spectrum with @xmath16 of observations using owfa . a more recent study
( sarkar , bharadwaj & ali 2017 ) indicate that it should be possible to measure the 21-cm power spectrum in several different @xmath17 bins in the range @xmath18 with @xmath19 of observations .
the complex visibilities are the primary quantities measured by any radio - interferometric array like owfa .
it is important and interesting to directly quantify the hi signal in terms of the expected contribution to the measured visibilities .
ali & bharadwaj ( 2014 ) present detailed predictions for the correlations between the visibilities measured at different baselines and frequency channels expected from the statistical hi signal , various astrophysical foregrounds and the system noise .
similar predictions for the hi signal incorporating non - linear effects are also presented in a more recent paper ( gehlot & bagla 2017 ) .
these theoretical estimates predict that the visibilities measured at owfa will be dominated by astrophysical foregrounds which are expected to be several orders of magnitude larger than the hi signal .
foreground removal ( e.g. ghosh et al .
2011b ) is therefore an important issue for detecting the redshifted 21-cm signal .
the astrophysical foregrounds are all expected to have a smooth frequency dependence in contrast to the hi signal , and most foreground removal techniques rely on this to distinguish between the foregrounds and the hi signal . however ,
the chromatic response of the telescope introduces frequency structures in the foregrounds , and it is necessary to model these and account for these in any foreground removal technique . here , simulations play a crucial role in modelling the impact of various complicated effects like the telescope s chromatic response , calibration ( marthi & chengalur 2014 ) etc .
simulations which incorporate the expected foregrounds , hi signal and also various instrumental and post - processing effects are essential for testing and validating any foreground removal and power spectrum estimation technique .
marthi ( 2017 ) has developed a programmable emulator for owfa , this provides a platform for simulating the visibilities that will be measured at owfa .
the foreground modelling and predictions are presented in marthi et al .
( 2017 ) . in this paper
we focus on modelling the hi signal contribution to the visibilities that will be measured at owfa .
bagla , khandai & datta ( 2010 ) have proposed a semi - numerical technique to simulate the redshifted hi signal .
this uses n - body simulations to predict the locations and masses of dark matter halos , and relies on a prescription to populate the halos with hi .
the same technique has been utilized in subsequent work by khandai et al.(2011 ) , guha sarkar et al.(2012 ) , villaescusa - navarro et al .
( 2014 ) and sarkar , bharadwaj & anathpindika ( 2016 ) .
it may be noted that this technique requires the n - body simulation to have a rather high spatial resolution @xmath20 ( comoving ) in order to resolve the halos which host the hi . in comparison ,
an angular extent of @xmath21 and a frequency spread of @xmath22 respectively corresponds to comoving distances of @xmath23 in the transverse direction and @xmath24 along the line of sight at @xmath25 .
considering the fact that owfa s field of view ( fov ) subtends several degrees on the sky and the bandwidth covers several tens of @xmath26 in frequency , it is computationally prohibitive to implement this technique to directly simulate the signal and calculate the visibilities expected at a wide field array like owfa . in this paper
we present two different methods to simulate the hi signal and calculate the visibilities expected in observations with owfa . in the first method we use the results of the semi - numerical simulations ( e.g. bagla , khandai & datta 2010 ) to construct an input model for the expected 21-cm power spectrum .
this is used to directly simulate different random realizations of the brightness temperature fluctuations and calculate the visibilities.the hi signal here is assumed to be a gaussian random field .
this method models the hi signal entirely as a diffuse radiation , and is completely oblivious to the discrete nature of the clouds which host the hi .
as noted above , the discrete clouds which host the hi each subtends an angle ( @xmath27 or smaller ) that is much smaller than the angular resolution ( @xmath28 ) of owfa .
the velocity structure @xmath29 of the hi inside the individual clouds , is however well within reach of owfa s frequency resolution ( @xmath30 or smaller ) , and this is expected to have an impact on the observed hi signal .
we have also used a second prescription , which is based on earlier work by bharadwaj & srikant ( 2004 ) , uses a cosmological n - body simulation to generate the matter distribution at the desired redshift . here
we identify each simulation particle with a source that hosts the hi and use this to calculate the expected visibilities .
this technique provides us with the freedom to implement any desired line profile for the hi emission from the individual sources .
the simulations presented in this paper are preliminary , thermal noise and complicated systematics have not been included here . a brief outline of the paper follows . in section 2
we present the model 21-cm power spectrum that was used to simulate the hi signal , and section 3 provides a brief overview of owfa .
the first ( direct ) simulation technique is presented in section 4 , which also contains the results from this approach .
the second ( particle based ) technique , along with the results , are presented in section 5 .
the results from both the simulation techniques are discussed and summarized in section 6 .
we have used the fitting formula of eisenstein & hu(1998 ) for the @xmath31-cdm transfer function to generate the initial , linear matter power spectrum .
the cosmological parameter values used are as given in ade et al.(2014):@xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 .
we model the 3d spatial power spectrum @xmath38 of the redshifted 21-cm brightness temperature fluctuations at @xmath39 as @xmath40 where , @xmath41 and @xmath42 is the power spectrum of the hi density fluctuations , all other symbols have their usual meaning ( bharadwaj & ali 2005 ) . as mentioned earlier ,
dla observations indicate @xmath43 .
this implies that the mass averaged neutral hydrogen fraction has a value @xmath44 which we have used in our work . assuming that the hi gas traces the underlying matter distribution with a linear bias @xmath45
, we relate @xmath46 to the underlying matter power spectrum @xmath47 as @xmath48 where @xmath49 , and @xmath50 is the component of the comoving wave vector @xmath51 along the line of sight .
the term @xmath52 arises due to of the effect of hi peculiar velocities ( bharadwaj , nath & sethi 2001 ; bharadwaj & ali 2005 ) , and @xmath53 is the linear redshift distortion parameter , where @xmath54 is the dimensionless linear growth rate .
we currently do not have any observational constraint on @xmath45 for high redshifts .
however , the results from the different semi - numerical simulations , and also an interpolated compilation of @xmath45 values ( padmanabhan , roy choudhury & refregier 2015 ) , indicate that it is reasonably well justified to assume a constant hi bias @xmath55 at wave numbers @xmath56 at the redshift @xmath57 .
we have used this value and @xmath58 for our entire analysis , this corresponds to @xmath59 ( bharadwaj , sarkar & ali 2015 ) .
in this paper we consider the so called `` phase - i '' of owfa ( e.g. ali & bharadwaj 2014 , subrahmanya , manoharan & chengalur 2017a ) where @xmath60 successive dipoles are combined to form a single antenna which gives rise to a @xmath61 element interferometer .
each antenna element has a rectangular aperture @xmath62 of dimensions @xmath63 along e - w and @xmath64 along n - s respectively .
this gives a field - of - view ( fov)of @xmath65 ( e - w ) @xmath66 ( n - s ) . the 40 antenna elements are equally spaced in a linear array with a separation of @xmath67 along the n - s direction .
the telescope is expected to have a frequency bandwidth @xmath68 .
we assume that the whole bandwidth is divided in @xmath69 frequency channels of @xmath70 channel width .
note that the values of @xmath71 , @xmath72 and @xmath73 assumed here are only representative values , and the actual values in the final implementation of the telescope may be somewhat different . the complex visibilities @xmath74 measured by owfa will be recorded for every independent pair of antennas for every available frequency channel . for any pair of antennas , the baseline @xmath75 quantifies the antenna separation in units of the observing wavelength . for observation towards the celestial equator ,
the baselines available at owfa are @xmath76 where @xmath77 is the total number of antennas for phase i. there is considerable redundancy in the owfa baseline distribution in that there are many different antenna pairs which correspond to the same baseline .
any baseline @xmath78 occurs @xmath79 times in the array . the baseline @xmath78 corresponding to a fixed antenna separation also changes with the observing wavelength @xmath80 .
we have used the flat sky approximation for our calculations throughout this paper . here
@xmath81 is a two dimensional vector in the plane of the sky with the origin at the centre of the fov .
the visibilities @xmath74 are the fourier transform of the product of the specific intensity distribution on the sky @xmath82 and the telescope s primary beam pattern @xmath83 .
we calculate the visibilities using @xmath84 where @xmath85 is the planck function in the raleigh - jeans limit which is valid at the frequency range of our interest and @xmath86 is the brightness temperature distribution on the sky .
the primary beam pattern for the rectangular aperture of the antenna elements ( ali & bharadwaj 2014 ) can be expressed as @xmath87 where @xmath88 and @xmath89 are respectively the @xmath90 and @xmath91 components of @xmath81 in a cartesian coordinate system that is aligned with the plane of the aperture and has the @xmath91 axis along the n - s direction .
we assume this pattern in our simulations , although we note that in practice tapering of the illumination will cause the owfa element pattern to have lower side lobes and somewhat broader main lobes .
we have decomposed the brightness temperature @xmath86 as the sum of an uniform average background @xmath92 and another component @xmath93 which fluctuates with direction on the sky @xmath94 in the flat sky approximation , the visibilities measured at owfa do not respond to the uniform background @xmath92 ( ali & bharadwaj 2014 ) , and it is adequate to only consider the fluctuations @xmath95 in eq .
( [ eq : vis1 ] ) to calculate the visibilities .
in this section we present simulations where we use an input model for 3d spatial power spectrum @xmath38 ( eq .
[ eq : temp_pow ] ) to directly generate @xmath93(eq .
[ eq : fluc ] ) on the sky and calculate the expected visibilities(eq .
[ eq : vis1 ] ) .
the @xmath96 owfa observations correspond to hi at @xmath57 which is at a comoving distance of @xmath97 .
this sets the conversion scale from angular separation to comoving distance in the transverse direction .
we further have @xmath98 which sets the conversion scale from frequency separation to comoving distance in the radial direction .
the owfa phase i fov of @xmath99 corresponds to @xmath100 in the two transverse directions and the bandwidth of @xmath101 corresponds to @xmath102 in the radial direction .
the simulations were carried out using a @xmath103 cubic grid of spacing @xmath104 covering a comoving volume @xmath105 .
we have chosen the grid spacing @xmath106 so that it exactly matches the channel width @xmath107 , where the spectral channels are of width @xmath108 .
we have considered a @xmath109 ^ 3 $ ] grid which corresponds to a comoving volume of @xmath110 ^ 3$].the considered volume is kept few times larger to incorporate the side lobe contributions ( figure [ fig:2 ] ) , that comes from the radiation received in the side lobes of the antenna pattern .
we use the input power spectrum @xmath111 to generate the fourier components of the brightness temperature fluctuations @xmath112 \ , .
\label{eq : den_k}\ ] ] on a grid of wave vectors @xmath113 corresponding to the simulation volume . here
@xmath114 and @xmath115 are two real valued independent gaussian random variable of unit variance which satisfy @xmath116 the fourier transform of @xmath117 yields a single realization of the 21-cm brightness temperature fluctuations @xmath118 on the simulation grid .
these fluctuations are , by construction , a gaussian random field with the input 21-cm power spectrum .
we use different sets of the random variables @xmath114 and @xmath115 to generate different statistically independent realizations of @xmath118 .
note that the adopted @xmath38 include the effect of redshift space distortion along the line of sight .
figure [ fig:1 ] shows a comparison of the input model power spectrum with the power spectrum estimated from a single realization of the simulation .
we find that the simulated power spectrum is in good agreement with he input model .
the simulation volume is aligned with the @xmath119 axis along the line of sight .
this corresponds to a frequency width of @xmath120 along the @xmath119 axis .
we have cut the box into 9 equal segments along the line of sight to produce 9 independent realizations each corresponding to a bandwidth of @xmath121 .
the grid index , measured from the further boundary and increasing towards to observer along the line of sight was directly converted to channel number @xmath122 whereby @xmath123 . the two transverse directions were restricted to @xmath124 and @xmath125 along the @xmath90 and @xmath91 axes respectively , and these were converted to angles relative to the center ( @xmath88 , @xmath89 ) = ( @xmath126 , @xmath127 ) .
the extent along the @xmath90 and @xmath91 axes were chosen so as to contain approximately the first three nulls of the beam pattern @xmath128 ( bottom panel of figure [ fig:2 ] ) along each direction .
the cut out region of the simulation box procedure provides us with @xmath129 the brightness temperature fluctuation on the sky ( upper panel of figure [ fig:2 ] ) at different frequency channels @xmath130 .
we calculate the visibilities using a discretized version of eq .
( [ eq : vis1 ] ) whereby @xmath131 where @xmath132 is the angular resolution of the simulation grid , the sum is over all the grid points ( labelled using @xmath133 ) corresponding to the fixed frequency channel @xmath130 , and @xmath134 is the @xmath91 component of the vector @xmath135 corresponding to the grid point @xmath133 .
the baseline and frequency channel configuration of owfa phase i probes the @xmath136 range @xmath137 to @xmath138 , and the @xmath50 range @xmath139 to @xmath140 ( bharadwaj et al .
2015 ) . for our simulations we have the corresponding values @xmath141 ,
@xmath142 , @xmath143 and @xmath140 .
as can be seen our simulations adequately cover the entire @xmath17 range that will be probed by owfa phase i. the redshifted 21-cm brightness temperature fluctuations are , by construction , a gaussian random field .
the visibilities obtained from the simulations also are random , and each realization of the simulation will yield a different set of values for the visibilities . it is necessary to consider the statistical properties of the simulated visibilities in order to validate the simulations . in this paper
we consider the correlations between the visibilities at two different baselines @xmath78 and @xmath144 and two different frequencies @xmath130 and @xmath145 @xmath146 the angular brackets @xmath147 here denote an ensemble average over different realizations of the simulation .
the hi contribution to the two - visibility correlation is directly related to the power spectrum @xmath38 ( bharadwaj & sethi 2001 ; bharadwaj & ali 2005 ) .
for owfa we have ( ali & bharadwaj 2014 ) , @xmath148 \ , \label{eq : viscor2}\end{aligned}\ ] ] where @xmath149 is the fourier transform of the beam pattern @xmath83 and @xmath51 refers to a comoving wave vector with radial and transverse components @xmath50 and @xmath150 respectively . in evaluating this expression it has been assumed that for all the terms in the _ r.h.s . _
the variation with @xmath151 is much slower in comparison with the variation with @xmath152 . as a consequence
we have held the values of @xmath153,@xmath154 , @xmath155 and @xmath78 fixed at that corresponding to the central frequency .
earlier work ( ali & bharadwaj 2014 ) has shown that for owfa the visibility correlation @xmath156 has non - zero values only for the situation when the two baselines are the same ( @xmath157 ) or if they are adjacent @xmath158 , the correlation is zero in all other situations .
further , the correlation is approximately four times smaller if we consider adjacent baselines instead of the same baseline ( bharadwaj et al .
2015 ) . in this paper
we have restricted the analysis to the situation where @xmath159 and we have considered @xmath160 with @xmath161 to validate the simulations .
we have used @xmath162 independent realizations of the simulated visibilities to evaluate the two - visibility correlation @xmath163 ( eq . [ eq : viscor1 ] ) .
the results thus obtained were compared with the linear theory predictions ( eq . [ eq : viscor2 ] ) in order to validate the simulations .
we first consider the visibility correlation @xmath164 which corresponds to the situation where @xmath165 . in this case
we have @xmath166 in eq .
( [ eq : viscor2 ] ) for which the visibility correlation @xmath167 is maximum
. figure [ fig:3 ] shows a comparison between the linear theory predictions ( line ) and the simulations ( points ) .
the @xmath168 error bars on simulated data points are estimated from the @xmath162 independent realizations of the simulated visibilities .
we see that the simulations are in very good agreement with the linear theory predictions .
we next consider how the visibility correlation @xmath163 varies with respect to @xmath169 .
it is convenient to use the frequency decorrelation function @xmath170 ( datta , roy choudhury & bharadwaj 2007 ) which is defined as @xmath171 this quantifies how rapidly the hi signal decorrelates as we increase the frequency separation @xmath169 .
the correlation is maximum for @xmath172 where @xmath173 , and the linear theory analytic calculations ( ali & bharadwaj 2014 ) predict the correlation to fall as @xmath169 is increased .
figure [ fig:4 ] shows the variation of the decorrelation function as a function of @xmath169 for different values of @xmath174 .
we observe that the simulations ( dashed lines ) are in good agreement with the linear theory predictions ( solid lines ) for all values of @xmath174 .
the hi signal decorrelates slowly with increasing @xmath152 at the short baseline ( e.g. @xmath175 for @xmath176 ) in comparison to the long baselines ( e.g. @xmath177 for @xmath178 ) where the hi signal decorrelates relatively rapidly .
we see that for @xmath179 we have @xmath180 at @xmath181 beyond which the value of @xmath182 falls further . in comparison , for @xmath183 we have @xmath180 at a smaller frequency separation of @xmath184 . in summary here , we note that our results show that the simulations faithfully reproduce both the @xmath174 and the @xmath152 dependence of the visibility correlation @xmath167 expected from the linear theory predictions ( eq .
[ eq : viscor2 ] ) thereby validating the simulations .
the simulations presented in the previous section treat the hi signal entirely as a diffuse radiation field and are completely oblivious to the discrete nature of the astrophysical sources which host the hi . while each discrete source subtends an angle that is much smaller than the angular resolution of owfa , the velocity structure of the hi inside the individual sources is well within reach of owfa s frequency resolution and this is expected to have an impact on the observed hi signal.in this section we outline a particle based simulation which allows the flexibility of introducing any desired line profile for the radiation from the individual sources .
the simulation technique here follows an earlier work ( bharadwaj & srikant 2004 ) .
the simulation uses a cosmological particle mesh ( pm ) n - body code to generate the dark matter particle distribution at @xmath185 .
the simulations were done using a @xmath186 ^ 3 $ ] grid which corresponds to a comoving volume of @xmath187 ^ 3 $ ] with @xmath188 ^ 3 $ ] particles in it .
we have chosen the grid spacing @xmath189 so that it exactly matches the frequency channel width @xmath190 .
we also incorporate the redshift space distortion effect from the velocity information of the dark matter particles .
figure [ fig:5 ] shows the ratio of the power spectrum of the density fluctuations in the dark matter distribution from n - body simulations and the power spectrum obtained from linear theory at @xmath57 .
we find that the power spectrum obtained from the n - body simulation starts showing deviations from the linear power spectrum at fourier modes @xmath191 and the deviation increases as we move on to higher @xmath17 modes .
these deviations are a consequence of the non - linear evolution of the density fluctuations at small scales .
this is an additional feature where the present technique improves upon the method presented in the previous section .
we remind the reader that the previous method was entirely based on the predictions of linear theory .
we have cut the simulation box into 6 equal segments along the line of sight to produce 6 independent realizations each corresponding to @xmath121 along the line of sight .
the simulation boxes were further cut to have dimensions @xmath192 and @xmath193 respectively along the @xmath90 and @xmath91 axes , and the particle distribution in the resulting boxes was used to calculate the hi signal .
we assume that the hydrogen has a mean neutral fraction @xmath194 and the total hi content is equally distributed amongst all the n - body particle , whereby distribution follows the underlying particle density @xmath195 . incorporating this
, we model the redshifted 21-cm brightness temperature distribution in the simulation box as @xmath196 where the factor @xmath197 introduces the linear bias , @xmath198 is the mean particle density , and @xmath199 where @xmath200 refers to the positions of the individual particles in redshift space .
it is important to note that the impact of the discrete nature of the hi sources depends on @xmath198 the number density of hi clouds .
the actual number density of the hi cloud around @xmath57 is not known , and in this work we have arbitrarily set @xmath201 which corresponds to the particle number density in the simulation . in the last section we described how we map the brightness temperature distribution @xmath202 to a brightness temperature distribution @xmath86 in angle and frequency in order to calculate the visibilities ( eq . [ eq : vis1 ] ) .
the position of each simulation particle @xmath200 also undergoes a similar mapping . as mentioned earlier ,
the two transverse dimensions are mapped to angles ( @xmath203 , @xmath204 ) = ( @xmath205 , @xmath206 ) and the radial direction is mapped to frequency ( @xmath207 ) where all the positions and angles are all with reference to the box center . here
, the hi radiation from a particular source at a radial distance of @xmath208 will be received at a single frequency @xmath209 which is determined by the cosmology , and the particular source will not contribute to any of the other frequencies . in reality , the hi emission emanating from any astrophysical source will have a spread in frequencies due to a variety of reasons including the thermal line broadening and the bulk motions of the hi gas inside the source .
we include this in the simulation by introducing a line profile @xmath210 associated with each source .
the line profile determines the frequency spread of the redshifted hi 21-cm emission from a particular source , and it is normalized as @xmath211 while the line profile may , in principle , differ from source to source we have here assumed the same line profile for all the sources in order to keep the analysis simple .
it may further be noted that the line profile here has been defined with respect to the observed frequency in the rest frame of the telescope . using eq ( [ eq : b1 ] ) in eq .
( [ eq : vis1 ] ) and incorporating the line profile , we obtain a relation which allows us to calculate the owfa visibilities corresponding to the simulated particle distribution @xmath212 while we have the freedom of implementing any desired line profile for the hi sources , for the purpose of this paper we have chosen a very simple line profile @xmath213 here we have considered three values of @xmath214 . for @xmath215 every source here is assigned a flat line profile of width @xmath216 , which is also the frequency resolution assumed for owfa , this corresponds to a velocity spread of @xmath217 at the rest frame of the source , which is a reasonable value for the hi velocities inside the individual sources expected at this redshift . for @xmath218 the assigned flat line profiles are of broader width @xmath219 and @xmath220 , which respectively correspond to twice and thrice the frequency resolution of owfa .
figure [ fig:6 ] shows the visibility correlation @xmath221 determined from the simulations .
we have run @xmath162 independent realizations of the simulation for which the mean and the standard deviation are respectively shown with points and the corresponding error bars .
we find that for all values of @xmath222 the value of @xmath223 obtained from the simulations is approximately twice the linear prediction for the shortest baseline @xmath174 with @xmath179 .
for @xmath215 the values obtained from the simulations are considerably larger than the linear theory prediction ( solid line ) for all values of @xmath174 .
the simulated values are roughly a factor of @xmath224 times larger than the linear prediction in the baseline range @xmath225 which corresponds to @xmath226 , the deviation between the simulations and the linear theory predictions is found to increase at longer baselines @xmath227 . barring the shortest baseline
, the values of @xmath228 progressively decrease as @xmath222 is increased to 2 and 3 . for @xmath229
the results are still in excess of the linear predictions .
for @xmath230 the results match with the linear theory predictions for @xmath231 whereas the results are below the linear theory predictions for larger baselines .
the deviations from linear theory seen for @xmath232 and @xmath233 can be attributed to a combination of the effect of the finite line width ( eq . [ eq : b3 ] ) and the non - linear evolution of the particle distribution in the n - body simulations ( figure [ fig:5 ] ) .
however , we expect the non - linear evolution to be more significant at the longer baselines , and it is difficult to use these two effects to explain why the shortest baseline shows a larger deviation compared to the other adjacent longer baselines for all values of @xmath222 . the earlier work ( ali & bharadwaj 2014 ) has shown that the owfa visibilities only respond to the brightness temperature fluctuations @xmath234 , and are insensitive to the mean component @xmath92 .
this however had assumed the flat sky approximation , and the sky was assumed to have an infinite extent .
this however does not hold for the simulations which also assume the flat sky approximation but have a finite sky extent .
the @xmath235 primary beam pattern ( eq . [ eq : pbeam ] ) is now truncated after a finite extent and as a consequence the visibilities all pick up a contribution from @xmath92 .
this contribution could be relatively large for the short baselines , however it is expected to die down at the longer baselines .
we can explain the large deviation at the shortest baseline by attributing it to the mean component @xmath92 that is picked up due to the finite sky coverage .
we note that this problem does not arise in the previous section where the mean component is not included in the simulations . in order to verify our understanding of the origin of the large deviation at the shortest baseline ,
we have considered a situation where the brightness temperature distribution ( eq . [ eq : fluc ] ) has only a contribution from @xmath92 and the fluctuations @xmath236 have been set to zero .
these simulations were run for @xmath237 only . considering the ratio @xmath238 ( figure [ fig:7a ] ) we find that this falls by approximately three orders of magnitude as we go from @xmath179 to @xmath239 .
the ratio falls even further if we consider longer baselines @xmath240 .
this supports our interpretation that in our simulations the shortest baseline picks up a significant contribution from the mean brightness temperature , however this contamination is not significant at the longer baselines .
figure [ fig:7 ] shows the variation of @xmath241 with @xmath152 for three different baselines .
notice that @xmath241 obtained from the simulations at the shortest baseline saturates at a relatively large value @xmath242 at frequency separations @xmath243 in contrast to the linear predictions which are close to zero in this range .
a similar discrepancy is not seen at the longer baselines where both the simulations and the linear prediction are close to zero for large @xmath152 .
this discrepancy at the shortest baseline also can be explained by the fact that the shortest baseline picks up a relatively large contribution from @xmath92 due to the limited angular extent of the simulations .
this problem does not seem to be significant at the longer baselines , and we have dropped the shortest baseline from the subsequent discussion of this section . for these simulations we find that it is more convenient to directly study @xmath244 instead of considering @xmath241 to analyze how the visibility correlation behaves as a function of @xmath169 .
figures [ fig : a ] and [ fig : b ] show @xmath244 as a function of @xmath169 for the two values @xmath245 and @xmath246 respectively . we have already seen that for @xmath215 the values of @xmath247 from the simulations are @xmath248 times larger than the linear prediction in the range @xmath249 ( figure [ fig:6 ] ) , and this discrepancy increases at longer baselines @xmath227 . here
we find that for @xmath237 , @xmath247 from the simulations are @xmath250 times larger than the linear prediction at @xmath245 ( figures [ fig : a ] ) .
the discrepancy in @xmath247 is reduced for @xmath229 and the results are roughly consistent with linear theory for @xmath251 .
we further see that the discrepancies are restricted to @xmath165 only for all values of @xmath222 .
the simulations are consistent with the linear predictions for larger values of @xmath152 , except possibly for a slight excess seen at @xmath252 for @xmath229 .
the discrepancy at @xmath165 arises from the line profile @xmath253 .
the hi emission from any individual source in the simulation will be spread across a finite frequency width introducing an excess correlation within the line width ( bharadwaj & srikant 2004 ) . for @xmath215
the line width is the same as the channel width @xmath73 which is why the discrepancy is only limited to @xmath165 and does not extend to larger frequency separations for @xmath237 . for @xmath254
the line width is doubled and its height is halved .
this explains why the deviation from linear theory in @xmath255 drops relative to @xmath215 .
this is also possibly responsible for the small excess that we see at @xmath256 for @xmath254 . for @xmath257
the line width is thrice that for @xmath215 whereas its height is three times smaller .
there is no noticeable excess correlation seen for @xmath257 because of the reduced line height , and the results are consistent with linear theory for all values of @xmath152 .
we next consider a larger baseline @xmath258 for which @xmath244 is shown in figure [ fig : b ] . for @xmath237
we find that the value obtained from the simulations is @xmath250 times larger than the linear prediction for @xmath165 and the simulations are consistent with the linear predictions for all larger values of @xmath152 .
for @xmath254 the discrepancy from the linear predictions reduces in magnitude but it now extends to @xmath259 and @xmath216 . for @xmath257
the result from the simulation falls below the linear prediction at @xmath165 while we have an excess at @xmath260 and @xmath219 . in summary
we find that the magnitude and frequency extent of the deviations from the linear predictions is strongly correlated with the value of @xmath222 which decides the frequency width and height of the emission line from the individual sources .
this leads to a picture where the deviations from linear theory seen in figures [ fig:6 ] , [ fig:7],[fig : a ] and [ fig : b ] can be largely attributed to the effect of the line profile of the individual sources . in the present analysis we are unable to identify any noticeable signature of he non - linear evolution of the underlying dark matter distribution .
finally we note that the good agreement between the simulations and the linear predictions found at the long baselines for large frequency separations confirms the fact that the longer baselines do not pick up a significant contribution from @xmath92 .
in this paper we present two different methods for simulating the redshifted hi 21-cm visibility signal expected in observations with phase i of owfa .
the first method directly generates random realizations of the brightness temperature fluctuations @xmath93 corresponding to the input model power spectrum @xmath38 at @xmath261 ( eq . [ eq : temp_pow ] ) .
these simulations treat the hi signal entirely as a diffuse radiation and ignores the fact that the hi actually resides in discrete clouds .
further , it also assumes that the redshifted 21-cm radiation is a gaussian random field whose properties are completely described by the power spectrum @xmath38 calculated using linear perturbation theory .
the second method uses a cosmological n - body simulation to generate a @xmath25 particle distribution .
we identify these particles as the discrete sources that host the hi .
this method allows us the freedom to assign any desired line profile to the hi emission from the individual sources .
it also captures the non - linear evolution of the density field . for the present work we have used a simple line profile where each source is assigned an uniform line width .
the simulations were done for three different line widths @xmath262 which respectively correspond to @xmath263 in our model for the line profile .
the simulated visibilities , in both cases , are a random signal . we have considered the two - visibility correlation @xmath244 to analyze the results of the simulations and compare them with the predictions of linear theory .
as expected , we find that the first method faithfully reproduces the predictions of linear theory . for the second method
, we find that the visibilities at the shortest baseline pick up a relatively large contribution from @xmath92 possibly due to the limited sky coverage of the simulations .
the visibilities at the longer baselines are not significantly contaminated by a contribution from @xmath92 .
we consider two long baselines , @xmath264 respectively in order to analyze the results of the second method .
we find that the results from the simulations show discrepancies from the linear theory predictions .
these discrepancies are restricted to @xmath152 values which are comparable to or smaller than the frequency width of the line profile , and the simulations are consistent with linear theory at larger @xmath152 .
the magnitude of the discrepancy depends on the width of the line profile , and for @xmath215 we find an excess over the linear predictions at @xmath265 .
the magnitude of this excess comes down when @xmath222 is changed from @xmath266 to @xmath267 .
we interpret this as arising from the fact that height of the line profile is halved when @xmath222 is changed from @xmath266 to @xmath267 .
we find that the simulations fall below the linear theory prediction at @xmath165 for @xmath257 . in this case
the increased line width possibly smears out some of the correlation , and the line height is also further reduced .
in conclusion we note that simulated data generated using the methods presented in this paper will play an important role in validating the owfa data analysis pipeline .
simulated data will also play an important role in correctly interpreting the observed signal once a detection is made . in this work
we find that it is important to take into account the discrete nature of the coulds .
the line profile of the individual clouds play an imprtant role in shaping the statistical properties of the redshifted 21-cm signal at small frequency separations which are comparable to the line width of the emission from the individual clouds .
we note that similar results have been reported in an earlier paper ( bharadwaj & srikant 2004 ) . for the present work we have used a rather simplistic line profile , we plan to incorporate more realistic models for the line profile in future work .
further , the impact of the discrete nature of the hi sources depends on @xmath198 the number density of clouds which has been set arbitrarily to @xmath201 which corresponds to the particle number density in the simulation .
we plan to consider more realistic mass distribution schemes in future work .
the current work reveals that the flat - sky paradigm adopted here fails to correctly model the visibilities at the shortest baseline , and it is necessary to implement the spherical sky in order to correctly handle this .
this is also expected to be important for owfa phase ii which will have a much larger field of view .
we plan to also address this in subsequent work .
the authors would like to thank jayaram chengalur for the useful suggestions and discussions .
sc acknowledges the university grant commission , india for providing financial support through senior research fellowship .
sc would also like to thank anjan kumar sarkar and debanjan sarkar for their help and useful discussions . | the upcoming ooty wide field array ( owfa ) will operate at @xmath0 which corresponds to the redshifted 21-cm signal from neutral hydrogen ( hi ) at z = 3.35 .
we present two different prescriptions to simulate this signal and calculate the visibilities expected in radio - interferometric observations with owfa . in the first method we use an input model for the expected 21-cm power spectrum to directly simulate different random realizations of the brightness temperature fluctuations and calculate the visibilities .
this method , which models the hi signal entirely as a diffuse radiation , is completely oblivious to the discrete nature of the astrophysical sources which host the hi . while each discrete source subtends an angle that is much smaller than the angular resolution of owfa , the velocity structure of the hi inside the individual sources is well within reach of owfa s frequency resolution and this is expected to have an impact on the observed hi signal .
the second prescription is based on cosmological n - body simulations . here
we identify each simulation particle with a source that hosts the hi , and we have the freedom to implement any desired line profile for the hi emission from the individual sources . implementing a simple model for the line profile , we have generated several random realizations of the complex visibilities .
correlations between the visibilities measured at different baselines and channels provides an unique method to quantify the statistical properties of the signal .
we have used this to quantify the results of our simulations , and explore the relation between the expected visibility correlations and the underlying hi power spectrum .
cosmology : large scale structure of universe - inter - galactic medium - diffuse radiation |
in @xcite the authors associated grbner toric degenerations @xmath0 of @xmath1 to each trivalent tree @xmath2 with @xmath3 leaves .
these degenerations induce toric degenerations @xmath4 of @xmath5 , the space of @xmath3 ordered , weighted ( by @xmath6 ) points on the projective line .
we denote the corresponding toric fibers by @xmath7 and @xmath8 respectively .
our goal in this paper is to give a geometric ( euclidean polygon ) description of the toric fibers as stratified symplectic spaces ( see @xcite for this notion ) and describe the action of the compact part of the torus as `` bendings of polygons . ''
we start by identifying the grassmannian @xmath1 with the moduli space of `` imploded spin - framed '' @xmath3-gons in @xmath9 .
we define the space of _ imploded framed vectors _ , which is topologically the cone @xmath10 of @xmath11 , as the space @xmath12 modulo the equivalence relation @xmath13 for all @xmath14 , where @xmath15 is the first standard basis vector of @xmath9 .
we will see later that this equivalence relation is `` implosion '' in the sense of @xcite .
we call the equivalence class of @xmath16 an `` imploded framed vector '' .
note that the isotropy @xmath17 of @xmath18 in @xmath11 has a natural right action on the space of imploded framed vectors .
now fix a covering homomorphism @xmath19 such that the cartan subgroup of diagonal matrices @xmath20 in @xmath21 maps onto @xmath17 .
we define the space of imploded _
spin_-framed vectors , which is topologically the cone @xmath22 , as the space @xmath23 modulo the equivalence relation @xmath13 for all @xmath24 .
we call the equivalence class of @xmath16 an `` imploded spin - framed vector '' .
an _ imploded spin - framed @xmath3-gon _ is an @xmath3-tuple @xmath25 of imploded spin - framed vectors such that @xmath26 .
let @xmath27 denote the space of imploded spin - framed @xmath3-gons .
there is an action of @xmath21 on @xmath27 given by @xmath28 we let @xmath29 denote the quotient space .
note that since we may scale the edges of an @xmath3-gon the space @xmath29 is a cone with vertex the zero @xmath3-gon ( all the edges are the zero vector in @xmath9 ) .
finally , note that there is a natural right action of @xmath30 on @xmath27 , which rotates frames but fixes the vectors : @xmath31 we will see later that @xmath29 occurs naturally in equivariant symplectic geometry , @xcite : it is the symplectic quotient by the left diagonal action of @xmath21 on the right imploded cotangent bundle of @xmath32 .
the space @xmath27 is the zero level set of the momentum map for the left diagonal action on the right imploded product of cotangent bundles .
hence we find that the space @xmath29 has a ( residual ) right action of an @xmath3torus @xmath30 , the maximal torus in @xmath32 which rotates the imploded spin - frames .
let @xmath33 be the quotient of the subspace of imploded spin - framed @xmath3-gons of perimeter @xmath34 by the action of the diagonal embedded circle in @xmath30 . in what follows @xmath35 denotes the affine cone over the grassmannian @xmath1 for the plcker embedding .
thus @xmath35 is the subcone of @xmath36 consisting of the decomposable bivectors ( the zero locus of the plcker equations ) .
the reason why we consider @xmath33 here is the following theorem ( proved in [ hjconstruction ] ) which gives a polygonal interpretation of @xmath1 .
it is the starting point of our work .
[ polygongrass ] 1 .
@xmath29 and @xmath35 are homeomorphic .
2 . this homeomorphism induces a homeomorphism between @xmath33 and @xmath37 let @xmath38 denote a fixed convex planar @xmath3-gon . throughout the paper we will use the symbol @xmath2 to denote either a triangulation of @xmath38 or its dual trivalent tree . accordingly
we fix a triangulation @xmath2 of @xmath38 .
the points of @xmath7 are `` imploded spin - framed @xmath3gons '' with a fixed perimeter in @xmath9 modulo an equivalence relation called @xmath2-congruence and denoted @xmath39 that depends on the triangulation @xmath2 . now that we have a polygonal interpretation of @xmath1 we will impose the equivalence relation of @xmath2-congruence ( to be described below ) on our space of framed polygons and obtain a polygonal interpretation of @xmath7 corresponding to the triangulation ( tree ) @xmath2 .
we now describe the equivalence relation of @xmath2-congruence . here
we will discuss only the case of the _ the standard triangulation _
@xmath40 , that is the triangulation of @xmath38 given by drawing the diagonals from the first vertex to the remaining nonadjacent vertices . the dual tree to the standard triangulation
will be called the caterpillar or fan ( see figure [ fig : caterpillartree7leaves ] ) .
however we will state our theorems in the generality in which they are proved in the paper namely for all trivalent trees @xmath2 with @xmath3 leaves .
the reader is urged to refer to the pictures below to understand the following description .
the equivalence relation for the standard triangulation ( in the case of @xmath3-gon linkages ) described below was first introduced in @xcite .
we have extended their definition to all triangulations and to @xmath3-gons equipped with imploded spin - frames .
label the diagonals of the triangulation counterclockwise by @xmath34 through @xmath41 .
for each @xmath42 let @xmath43}$ ] denote the subspace of @xmath27 where the diagonals corresponding to the elements in @xmath44 are zero and all other diagonals are nonzero .
let @xmath45 be a point in @xmath43}$ ] .
suppose that @xmath46 .
since @xmath47 diagonals are zero there will be @xmath48 sums of the form @xmath49 that are zero , and the @xmath3-gon underlying @xmath50 will be the wedge of @xmath48 closed subpolygons corresponding to the @xmath48 closed subpolygons in the collapsed reference polygon .
thus we can divide @xmath50 into @xmath48 imploded spin - framed closed subpolygons ( in terms of formulas we can break up the above @xmath3-tuple into @xmath48 sub @xmath51-tuples of edges for @xmath52 ) .
we may act on each imploded spin - framed subpolygon ( @xmath51-tuple ) by a copy of @xmath21 .
we pass to the quotient @xmath53 by dividing out @xmath43}$ ] by the resulting action of @xmath54 . by definition @xmath2-congruence
is the equivalence relation induced by the above quotient operation , see figure [ fig : treeequivalence ] .
we let @xmath55 denote the quotient space @xmath56 .
thus @xmath55 is decomposed into the pieces @xmath57 } = \widetilde{p}_n(\mathrm{su}(2))^{[s]}/\sim_{\tree_0}$ ] .
we will call the resulting decomposition in the special case of @xmath40 the kamiyama - yoshida decomposition ( or ky - decomposition ) . in this paper
we use a weakened definition of the term _ decomposition _ ; we will refer to decompositions of spaces where the pieces are products of spaces with isolated singularities .
the equivalence relation of @xmath2-congruence can be defined analogously for any triangulation of @xmath38 ( equivalently any trivalent tree with @xmath3-leaves ) and induces an equivalence relation on @xmath33 ( and many other spaces associated to spaces of @xmath3-gons in @xmath9 , for example , the space of @xmath3-gons itself or the space of @xmath3-gon linkages ) .
we will use the symbol @xmath39 to denote all such equivalence relations . we will use @xmath58 to denote the quotient of @xmath33 by the equivalence relation @xmath39 .
we can now state our first main result ( this is proved in [ homeomorphismvtop ] ) .
[ firstmaintheorem ] 1 .
the toric fiber of the toric degeneration of @xmath35 corresponding to the trivalent tree @xmath2 is homeomorphic to @xmath59 .
2 . the toric fiber of the toric degeneration of @xmath1 corresponding to the trivalent tree @xmath2 is homeomorphic to @xmath58 .
the quotient map from @xmath33 to @xmath58 given by passing to @xmath2-congruence classes maps the generic fiber of the toric degeneration onto the special ( toric ) fiber .
we next describe the space @xmath60 that will be proved later to be homeomorphic to the toric fiber @xmath61 .
again we will restrict ourselves to the standard triangulation in our description . starting with the space @xmath62 of closed @xmath3-gon linkages with side - lengths @xmath6 ,
we define @xmath63 this is the construction of @xcite - see below for some pictures .
[ secondmaintheorem ] the toric fiber @xmath64 of the toric degeneration of @xmath5 corresponding to the trivalent tree @xmath2 is homeomorphic to @xmath65 . the quotient map from @xmath5 to @xmath65 maps the generic fiber of the toric degeneration onto the special ( toric ) fiber .
the result in the previous theorem was conjectured by philip foth and yi hu in @xcite .
the motivation for the above construction becomes more clear once we introduce the bending flows of @xcite and @xcite .
the lengths of the @xmath41 diagonals created are continuous functions on @xmath5 and are smooth where they are not zero .
they give rise to hamiltonian flows which were called _ bending flows _ in @xcite . the bending flow associated to a given diagonal has the following description .
the part of the @xmath3-gon to one side of the diagonal does not move , while the other part rotates around the diagonal at constant speed .
the lengths of the diagonals are action variables which generate the bending flows , the conjugate angle variables are the dihedral angles between the fixed and moving parts .
however _ the bending flow along the @xmath47-th diagonal is not defined at those @xmath3-gons where the @xmath47-th diagonal is zero_. if the bending flows are everywhere defined ( for example if one of the side - lengths is much larger than the rest ) then we may apply the theorem of delzant , @xcite to conclude that @xmath5 is toric .
however for many @xmath6 ( including the case of regular @xmath3 -gons ) the bending flows are not everywhere defined .
the point of @xcite was to make the bending flows well - defined by dividing out the subspaces of @xmath67 where a collection of @xmath47 diagonals vanish by @xmath68 .
we illustrate their construction with two examples .
first , let @xmath69 so @xmath5 is the space of regular hexagons with side - lengths all equal to @xmath34 .
let @xmath70 be the subspace of @xmath5 where the middle ( second ) diagonal vanishes .
thus @xmath70 is the space of `` bowties '' ( see figure [ fig : bowties ] ) modulo the diagonal action of @xmath11 on the two equilateral triangles .
we can no longer bend on the second diagonal because we have no axis of rotation . passing to @xmath2-congruence classes
collapses the space @xmath70 to a point by dividing by the action of @xmath71 .
bending along the second diagonal fixes this point by definition . for our second example , we consider the space of regular octagons with all side - lengths equal to @xmath34 and the subspace @xmath72 where the middle diagonal vanishes .
thus @xmath72 is the space of wedges of rhombi modulo the diagonal action of @xmath11 on the two rhombi , see figure [ fig : tworhombi ] . passing to @xmath2-congruence classes
amounts to dividing by the action of @xmath71 on the two rhombi .
we obtain the action of the bending flows on @xmath55 as follows . in this case
we will be given a lift of the one - parameter group bending along a diagonal to @xmath21 .
the one parameter group acts through its quotient in @xmath11 by bending along the diagonal .
if an edge moves under this bending then the imploded spin - frame is moved by the one - parameter group in @xmath21 in the same way .
hence , one part of the imploded spin - framed polygon is fixed and the other moves by a `` rigid motion '' i.e .
all the spin - framed edges of the second part are moved by the same one - parameter group in @xmath21 .
the bendings give rise to an action of an @xmath41 torus @xmath73 on @xmath59 .
there are also edge - rotations that apply a one - parameter group to the imploded spin - frame but do not move the edge .
this action on frames is the action of the torus @xmath30 coming from the theory of the imploded cotangent bundle of @xmath32 .
this action is not faithful , the diagonal subtorus acts trivially .
the bendings together with the edge rotations give rise to an action of a compact @xmath74 torus @xmath75 .
the main step in proving theorems [ firstmaintheorem ] and [ secondmaintheorem ] is to produce a space @xmath76 that `` interpolates '' between @xmath55 and @xmath77 .
our construction of @xmath76 was motivated by the construction of the toric degeneration of @xmath21-character varieties of fundamental groups of surfaces given by hurtubise and jeffrey in @xcite .
the connection is that the space @xmath5 can be interpreted as the ( relative ) character variety of the fundamental group of the @xmath3-punctured two - sphere with values in the translation subgroup of the euclidean group @xmath78 - a small loop around the @xmath47-th puncture maps to translations by the @xmath47-th edge of the polygon ( considered as a vector in @xmath9 ) .
take the triangulated model ( convex planar ) @xmath3-gon @xmath38 and break it apart into @xmath79 triangles @xmath80 .
equivalently we break apart the dual tree @xmath2 into a forest @xmath81 consisting of @xmath79 tripods .
attach to each of the @xmath82 edges of the @xmath79 triangles ( or each edge of the forest @xmath81 ) a copy of @xmath83 .
now right - implode each copy of @xmath83 so that a copy of @xmath84 , the imploded cotangent bundle of @xmath21 , is attached to each edge . since @xmath85 admits an action of the circle ( from the right ) and @xmath21 from the left the resulting space admits an action of a torus @xmath86 of dimension @xmath82 and a commuting action of @xmath87 .
the torus has a product decomposition @xmath88 where @xmath89 is the product of factors corresponding to the @xmath3 edges of the polygon @xmath38 and @xmath90 is the product of factors corresponding to the @xmath41 diagonals of @xmath38 .
note that each diagonal of @xmath38 occurs in two triangles so corresponds to two edges of @xmath81 .
hence each diagonal gives rise to a two - torus @xmath91 in @xmath92 which we will refer to as the two - torus corresponding to that diagonal .
define a subtorus @xmath93 of @xmath90 of dimension @xmath41 by taking an antidiagonal embedding of @xmath94 in each two - torus corresponding to a diagonal .
now take the symplectic quotient ( at level zero ) of ( the product of ) the three copies of @xmath95 associated to the three sides of each triangle by @xmath21 acting diagonally .
for each triangle ( or each tripod ) we obtain a resulting copy of @xmath96 .
the resulting product @xmath97 has an induced action of the torus @xmath86 .
glue the copies @xmath98 associated to the triangles together along the edges of the triangles associated to diagonals by taking the symplectic quotient at level zero by the torus @xmath99 described above .
each of the two previous symplectic quotients has a corresponding git quotient . taking both git quotients
we obtain a space which is an affine torus quotient of affine space .
hence the combined symplectic quotient is the space underlying the affine toric variety @xmath100 let @xmath101 be the element in complexified torus @xmath102 of @xmath86 such that all the edge components coincide with @xmath103 and all the components corresponding to diagonals are @xmath34 .
then theorem [ firstmaintheorem ] follows by putting together items 2 and 4 in the next theorem , which is proved in [ homeomorphismvtop ] .
[ firstauxiliarytheorem ] 1 . the toric varieties @xmath104 and @xmath105 are isomorphic as affine toric varieties . 2 .
the grading action of @xmath106 on @xmath104 corresponds to the action of @xmath107 on @xmath108 ( this is well - defined ) .
consequently @xmath109 is projectively isomorphic to the quotient of @xmath105 by this @xmath110 action - we will denote this quotient by @xmath111 .
3 . there is a homeomorphism ( that creates imploded spin - frames along the diagonals of the triangulation ) @xmath112 4 .
the homeomorphism @xmath113 induces a homeomorphism from @xmath58 to the projective toric variety @xmath111 .
the quotient @xmath114 admits a residual action by the quotient torus @xmath115 .
this quotient torus contains a factor that can be identified with @xmath116 .
the toric fiber @xmath117 is obtained from @xmath76 by taking the symplectic quotient by @xmath116 at level @xmath6 .
theorem [ secondmaintheorem ] follows from the two statements of the following theorem .
[ secondauxiliarytheorem ] 1 .
the toric variety @xmath118 is isomorphic to the toric variety @xmath8 .
2 . for each @xmath6 the homeomorphism
@xmath113 induces a homeomorphism @xmath119 note that the toric varieties @xmath120 resp . @xmath121
mediate between the kamiyama - yoshida spaces @xmath59 resp . @xmath122 and the toric varieties @xmath7 resp .
@xmath123 .
it remains to place the edge rotations and bending flows in their proper context ( in terms of symplectic and algebraic geometry ) .
we do this with the following theorems .
we note that since the spaces @xmath105 and @xmath124 are quotients of the affine space @xmath97 by tori they inherit symplectic stratifications from the orbit type stratification of @xmath97 according to @xcite .
we identify the edge flows and bending flows with the action of the maximal compact subgroups ( compact torus ) of the complex tori that act holomorphically with open orbits on the toric varieties @xmath7 and @xmath123 .
this is accomplished by the following theorem , which is proved by theorems [ noncompact ] and [ noncompact2 ] . 1 .
the action of the edge rotations corresponds under @xmath113 to the residual action of @xmath125 .
the action of the bending flows corresponds under @xmath113 ( resp .
@xmath126 ) to the residual action of @xmath127 . we may now give the edge rotations and the bending flows on @xmath59 and @xmath65 a natural hamiltonian interpretation .
this is proved by the last theorem along with proposition [ pullback ] and theorem [ diagonalpullback ] .
[ stratification ] 1 .
the edge rotations on @xmath59 are the stratified symplectic hamiltonian flows ( in the sense of @xcite ) associated to the lengths of edges in @xmath2-congruence classes of spin - framed polygons .
2 . the bending flows on @xmath59 ( resp .
@xmath65 ) are the stratified hamiltonian flows associated to the lengths of the diagonals of @xmath2-congruence classes of spin - framed polygons ( resp .
polygonal linkages ) .
* acknowledgements . *
we thank bill goldman , henry king and reyer sjamaar for useful conversations .
the authors would also like to thank andrew snowden and ravi vakil , a number of ideas from the collaboration @xcite have reappeared in this paper .
we thank philip foth and yi hu for posing the problem solved in this paper .
they first observed that the toric degenerations of flag varieties constructed by @xcite could be descended to the associated weight varieties .
we especially thank philip foth for pointing out that the construction of @xcite gives one toric degeneration of @xmath5 for each triangulation of the model convex @xmath3-gon .
this led us to consider triangulations other than the standard one .
we thank bernd sturmfels for telling us about @xcite and @xcite which led us to understand from the point of view of tree metrics why there was one toric degeneration for each triangulation .
we thank allen knutson for pointing out that toric degenerations of grassmannians were first constructed by sturmfels in @xcite .
finally we should emphasize that the notion of @xmath2-congruence is based on the work of kamiyama and yoshida @xcite and that the notion of bending flows is based on @xcite and @xcite .
throughout this paper the term @xmath3-gon will mean a closed @xmath3-gon in @xmath9 modulo translations .
more precisely an @xmath3-gon @xmath128 will be an @xmath3-tuple @xmath129 of vectors in @xmath9 satisfying the closing condition @xmath130 we will say the @xmath131 is the @xmath47-th edge of @xmath128 .
we will say two @xmath3-gons @xmath128 and @xmath132 are congruent if there exists a rotation @xmath133 such that @xmath134 we will let @xmath135 denote the space of closed @xmath3-gons in @xmath9 and @xmath136 denote the quotient space of @xmath3-gons modulo congruence .
now let @xmath137 be an @xmath3-tuple of nonegative real numbers .
we will say an @xmath3-gon @xmath128 is an @xmath3-gon _ linkage _ with side - lengths @xmath6 if the @xmath47-th edge of @xmath128 has length @xmath138 .
we will say an @xmath3-gon or @xmath3-gon linkage is _ degenerate _ if it is contained in a line .
we define the configuration space @xmath139 to be the set of @xmath3gon linkages with side - lengths @xmath6 .
we will define the moduli space @xmath5 of @xmath3-gon linkages to be the quotient of the configuration space by @xmath11 .
the space @xmath5 is a complex analytic space ( see @xcite with isolated singularities at the degenerate @xmath3-gon linkages .
recall that we have defined a reference convex planar @xmath3-gon @xmath38 .
let @xmath140 be an ordered pair of nonconsecutive vertices of @xmath38 .
for any @xmath3-gon @xmath141 we have corresponding vertices @xmath142 and @xmath143 defined up to simultaneous translation .
the vector in @xmath9 pointing from @xmath142 to @xmath143 will be called a _ diagonal _ of @xmath128 .
we let @xmath144 be the length of this diagonal . in @xcite and
@xcite the authors described the hamiltonian flow corresponding to @xmath144 - see the introduction . in @xcite these flows
were called _
bending flows_. furthermore it was proved in @xcite and @xcite if two such diagonals do not intersect then the corresponding bending flows commute .
since each triangulation @xmath2 of @xmath38 contains @xmath145 nonintersecting diagonals it follows that each one has an integrable system on @xmath5 for each such @xmath2 .
unfortunately these flows are not everywhere defined .
the bending flow corresponding to @xmath146 is not well - defined for those @xmath128 where @xmath144 is zero and the hamiltonian @xmath146 is not differentiable at such @xmath128 .
in this section we will construct the space @xmath29 of imploded spin - framed @xmath3gons in @xmath9 modulo @xmath21 and prove that this space is isomorphic to @xmath35 as a symplectic manifold and as a complex projective variety .
we will first construct @xmath29 using the extension and implosion technique of @xcite without reference to @xmath3gons in @xmath9 , then relate the result to @xmath1 .
then we will show that a point in @xmath29 can be interpreted as a euclidean @xmath3-gon equipped with an imploded spin - frame .
in this subsection we will define the imploded extended moduli space @xmath147 of @xmath3-gons in @xmath148 for a general semisimple lie group @xmath149 .
we have included this subsection to make the connection with @xcite . throughout
we assume that @xmath149 is semisimple . to motivate the definition of the next subsection we briefly recall two definitions .
an @xmath3-gon in @xmath148 is an @xmath3-tuple of vectors @xmath150 satisfying the closing condition@xmath151 we define the moduli space of @xmath3-gons to be the set of all @xmath3-gons modulo the diagonal coadjoint action of @xmath149 .
now choose @xmath3 coadjoint orbits @xmath152 .
we define an @xmath3-gon _ linkage _ to be an @xmath3-gon such that @xmath153 .
we define the moduli space of @xmath3-gon linkages to be the set of all @xmath3-gon linkages modulo the diagonal coadjoint action of @xmath149 .
we leave the proof of the following lemma to the reader .
the moduli space of @xmath3-gon linkages is the symplectic quotient@xmath154 [ moduliofconnections ] the moduli space of @xmath3-gons is in fact a moduli space of flat connections modulo gauge transformations ( equivalently a character variety ) .
the moduli space of @xmath3-gon linkages is a moduli space of flat connections with the conjugacy classes of holonomies fixed in advance ( a relative character variety ) . in this case
the flat connections are over an @xmath3-fold punctured @xmath155-sphere and the structure group is the cotangent bundle @xmath156 .
the holonomy around each puncture is a `` translation '' , i.e. a group element of the form @xmath157 . to see this connection in more detail the reader
is referred to 5 of @xcite .
we will not need this connection in what follows .
the following definition is motivated by the definition of the extended moduli spaces of flat connections of @xcite .
we define the extended moduli space @xmath158 to be the symplectic quotient of @xmath159 by the left diagonal action of @xmath149 : @xmath160 the space @xmath158 has a @xmath161 action coming from right multiplication on @xmath162 .
we take each @xmath162 to be identified with @xmath163 using the left - invariant trivialization .
@xmath164 the momentum mapping associated to the diagonal left action on @xmath159 ( identified with @xmath165 using the left - invariant trivialization ) is @xmath166 the expression on the right above clearly corresponds to the 0-momentum level of @xmath167 which is @xmath158 .
the reason for the term _ extended _ moduli space of @xmath3-gons is that this space is obtained from the moduli space of @xmath3-gons by adding the frames @xmath168 .
note that the moduli space of @xmath3-gons is embedded in the extended moduli space as the subspace corresponding to @xmath169 .
if we wish to fix the conjugacy classes of the second components we will call the above the extended moduli space of @xmath3-gon _
linkages_. choose a maximal torus @xmath170 and a ( closed ) weyl chamber @xmath171 contained in the lie algebra @xmath172 of @xmath173 .
note that the action of @xmath161 on @xmath158 by right multiplication induces an action by the torus @xmath174 on @xmath158 .
we now obtain the imploded extended moduli space @xmath147 by _ imploding _ , following @xcite , the extended moduli space @xmath158 .
[ definitionofimplodedextended moduli space @xmath147 ] @xmath176 following @xcite we will use @xmath177 to denote the imploded cotangent bundle @xmath178 for the benefit of the reader we will recall the definition of @xmath177 .
we have @xmath179 here @xmath180 is the momentum map for the action of @xmath149 by right translation and we have identified @xmath172 ( resp .
@xmath171 ) with @xmath181 ( resp . a dual weyl chamber ) using the killing form .
the equivalence relation @xmath182 is described as follows .
let @xmath183 be a open face of @xmath171 and let @xmath184 be a generic element of @xmath183 .
let @xmath185 be the subgroup of @xmath149 which is the derived subgroup of the stabilizer of @xmath184 under the adjoint representation .
we define @xmath186 and @xmath187 in @xmath188 to be equivalent if @xmath189 and @xmath190 lie in the same face @xmath183 of @xmath171 and @xmath186 and @xmath187 are in the same orbit under @xmath185 .
thus we divide out the inverse images of faces by different subgroups of @xmath149 . in what follows we will use the symbol @xmath182 to denote this equivalence relation ( assuming the group @xmath149 , the torus @xmath191 and the chamber @xmath171 are understood ) .
we define the space @xmath192 by @xmath193 noting that right implosion commutes with left symplectic quotient we have @xmath194 , \ldots , [ g_n,\alpha_n ] ) \mid \alpha_1 , \ldots ,
\alpha_n \in \delta,\sum_{i=1}^n ad_\mathfrak{g_i}(\alpha_i ) = 0\big\}.\ ] ] in the above @xmath195 $ ] denotes the equivalence class in @xmath162 relative to the equivalence relation @xmath182 above .
we will sometimes use @xmath192 to denote the product @xmath196 . in what follows
we let @xmath197 denote the element of the complexification maximal torus @xmath198 . in this subsection
we will prove theorem [ polygongrass ] by calculating @xmath29 , the imploded extended moduli space for the group @xmath21 and its quotient @xmath33 .
we will in fact need a slightly more precise version than that stated in the introduction .
[ polygongrasspreciseversion ] 1 .
there exists a homeomorphism @xmath199 .
the homeomorphism @xmath200 intertwines the natural action of the maximal torus @xmath201 of @xmath202 with ( the inverse of ) that of @xmath203 acting on imploded frames . 3 . the homeomorphism @xmath200 intertwines the grading circle ( resp .
@xmath204 ) actions on @xmath35 with the actions of @xmath205 ( resp .
@xmath206 ) 4 .
the homeomorphism @xmath200 induces a homeomorphism between @xmath1 and @xmath207 in order to prove the theorem we first need to compute @xmath208 . in this section
we will review the formula of @xcite for the ( right ) imploded cotangent bundle @xmath208 .
let @xmath20 be the maximal torus of @xmath21 consisting of the diagonal matrices .
let @xmath172 be the lie algebra of @xmath20 and let @xmath171 be the positive weyl chamber in @xmath172
( so @xmath171 is a ray in the one - dimensional vector space @xmath172 ) .
we will take as basis for @xmath172 the coroot @xmath209 ( multiplied by @xmath47 ) , that is @xmath210 then @xmath209 may be identified with a basis vector over @xmath211 for the cocharacter lattice @xmath212 .
it is tangent at the identity to a unique cocharacter .
let @xmath213 be the fundamental weight of @xmath214 thus @xmath215 then @xmath213 may be identifed with a character of @xmath191 ( it is the derivative at the identity of a unique character ) which is a basis for the character lattice of @xmath20 .
let @xmath216 be the momentum map for the action of @xmath21 on @xmath217 induced by right multiplication .
specializing the definition of the imploded cotangent bundle to @xmath218 we find that the ( right ) implosion of @xmath217 is the set of equivalence classes @xmath219 where two points @xmath220 are equivalent if @xmath221 it follows from the general theory of @xcite that @xmath208 has an induced structure of a stratified symplectic space in the sense of @xcite with induced isometric actions of @xmath21 induced by the left action of @xmath21 on @xmath217 and @xmath20 induced by the right action of @xmath20 on @xmath83 .
however the exceptional feature of this special case is that @xmath208 is a _ manifold_. the multiplicative group @xmath222 acts on @xmath208 by the formula @xmath223 = [ g , \mu \lambda \varpi_1]\ ] ] whereas the actions of @xmath224 and @xmath225 described above are given by @xmath226 = [ g_0 g , \lambda \varpi_1 ] \
\text { and } t \cdot [ g , \lambda \varpi_1 ] = [ gt , \lambda \varpi_1].\ ] ] the product group @xmath227 fixes the point @xmath228 $ ] and acts transitively on the complement of @xmath228 $ ] in @xmath208 .
the reader will verify , see example 4.7 of @xcite , that the symplectic form @xmath229 on @xmath208 is homogeneous of degree one under @xmath222 .
the following lemma is extracted from example 4.7 of @xcite .
since this lemma is central to what follows we will go into more detail for its proof than is in @xcite .
[ implodedcotangentbundle ] the map @xmath230 given by @xmath231 ) = \sqrt{2\lambda } g \begin{pmatrix}1\\0\end{pmatrix } \ \text{and } \ \phi([e,0 ] ) = \begin{pmatrix}0\\0\end{pmatrix}\ ] ] induces a symplectomorphism of stratified symplectic spaces onto @xmath95 where @xmath95 is given its standard symplectic structure . under this isomorphism
the action of @xmath21 on @xmath208 goes to the standard action on @xmath95 and the action of @xmath191 on @xmath208 goes to the action of @xmath191 given by @xmath232 .
the inverse to @xmath233 is given ( on nonzero elements of @xmath95 ) by @xmath234,\ ] ] where @xmath235 it follows that @xmath233 is a homeomorphism .
it remains to check that it is a symplectomorphism .
noting that @xmath233 is homogeneous under @xmath222 of degree one - half and the standard symplectic form @xmath236 is homogeneous of degree two it follows ( from the transitivity of the action of @xmath227 ) that the symplectic form @xmath237 is a constant multiple of @xmath229 .
thus is suffices to prove that the two above forms both take value @xmath34 on the ordered pair of tangent vectors @xmath238 .
we leave this to the reader .
the fact that we have inverted the usual action of the circle on the complex plane will have the effect of changing the signs of the torus hamiltonians that occur in the rest of this paper ( from _ minus _ one - half a sum of squares of norms of complex numbers to _ plus _ one - half of the corresponding sum ) .
taking into account the above remark we obtain a formula for the momentum map for the action of @xmath20 on @xmath208 which will be critical in what follows .
now we have [ momentummap ] the momentum map @xmath239 for the action of @xmath20 on @xmath240 is given by @xmath241 consequently , the hamiltonian @xmath242 for the fundamental vector field on @xmath243 induced by @xmath209 is given by @xmath244 we can now prove theorem [ polygongrasspreciseversion ] .
recall we have introduced the abbreviation @xmath245 for the product @xmath246 .
we will use @xmath247 to denote an element of @xmath245 .
it will be convenient to regard @xmath247 as a function from the edges of the reference polygon @xmath38 into @xmath248 .
we note that there is a map from @xmath245 to not necessarily closed @xmath3-gons in @xmath9 , the map that scales the first element of the frame by @xmath249 and forgets the other two elements of the frame this sends each equivalence class @xmath250 $ ] to @xmath251 for @xmath252 . by the previous lemmas we may represent an element of @xmath245 by the @xmath155 by @xmath3 matrix @xmath253 . the left diagonal action of the group @xmath21 on @xmath245 is then represented by the diagonal action of @xmath21 on the columns of @xmath254 .
we let @xmath255 $ ] denote the orbit equivalence class of @xmath254 for this action .
we let @xmath201 be the compact @xmath3 torus of diagonal elements in in @xmath256 .
then @xmath201 acts by scaling the columns of the matrices @xmath254 .
this action coincides with the inverse of the action of the @xmath3-torus @xmath30 coming from the theory of the imploded cotangent bundle . by this
we mean that if an element in @xmath201 acts by scaling a column by @xmath249 then the corresponding element acts by scaling the same column by @xmath257
. the first statement of theorem [ polygongrasspreciseversion ] will be a consequence of the following two lemmas .
let @xmath258 be the space of @xmath155 by @xmath3 complex matrices @xmath254 as above .
in what follows we will let @xmath259 denote the first row and @xmath260 the second row of the above matrix @xmath254 .
we give @xmath258 the hermitian structure given by @xmath261 we will use @xmath180 to denote the momentum map for the action of @xmath21 on the left of @xmath254 . [ levelzero]@xmath262 hence @xmath263 @xmath264 the two rows of @xmath254 have the same length and are orthogonal .
first note that the momentum map @xmath265 for the action of @xmath21 on @xmath95 is given by@xmath266 [ affinegrass ] @xmath267 the map @xmath268 given by @xmath269 maps onto the decomposable vectors , descends to the quotient by @xmath21 , and induces the required isomorphism . in what follows it will be important to understand the above result in terms of the git quotient of @xmath258 by @xmath214 acting on the left .
since we are taking a quotient of affine space by a reductive group the geometric invariant theory quotient of @xmath258 by @xmath270 coincides with the symplectic quotient by @xmath21 and consequently is @xmath35 , @xcite , see also @xcite , theorem 4.2 .
note that since @xmath21 is simple the question of normalizing the momentum map does not arise .
it will be important to understand this in terms of the subring @xmath271^{\mathrm{sl}(2 , \mathbb{c})}$ ] of invariant polynomials on @xmath258 . in what follows let @xmath272 or @xmath273 $ ] be the determinant of the @xmath155 by @xmath155 submatrix of @xmath254 given by taking columns @xmath47 and @xmath274 of @xmath254 ( the @xmath275-th plcker coordinate or bracket ) .
the following result on one of the basic results in invariant theory , see @xcite , chapter 2 .
[ pnggit ] @xmath271^{\mathrm{sl}(2 , \mathbb{c})}$ ] is generated by the @xmath276 subject to the plcker relations .
but the above ring is the homogeneous coordinate ring of @xmath1 .
this gives the required invariant - theoretic proof of theorem [ polygongrass ] .
in fact we could do _ all _ of the previous analysis in terms of invariant theory by using example 6.12 of @xcite to replace the imploded cotangent bundle @xmath277 by the quotient @xmath278 where @xmath279 is the subgroup of @xmath214 of strictly upper - triangular matrices .
we leave the details to the reader .
we have now proved the first statement in theorem [ polygongrasspreciseversion ] for both symplectic and git quotients .
we note the consequence ( since every element of @xmath96 is decomposable ) @xmath280 it is clear that @xmath200 is equivariant as claimed and the second statement follows .
since the grading action of @xmath103 on @xmath281 is the action that scales each plcker coordinate by @xmath249 the third statement is also clear .
it remains to prove the fourth statement of theorem [ polygongrasspreciseversion ] .
first recall the standard @xmath202 invariant positive definite hermitian form @xmath282 on @xmath36 is given on the bivector @xmath283 by @xmath284 we can obtain @xmath1 as a symplectic manifold the cone @xmath35 of decomposable bivectors by taking the subset of decomposable bivectors of norm squared @xmath285 then dividing by the action of the circle by scalar multiplication ( symplectic quotient by @xmath94 of level @xmath286 ) .
thus we can find a manifold diffeomorphic to @xmath1 by computing the pull - backs of the function @xmath287 and scalar multiplication of bivectors by the circle under the map @xmath288 .
we represent elements of @xmath29 by left @xmath21 orbit - equivalence classes @xmath255 $ ] of @xmath155 by @xmath3 matrices @xmath254 such that the rows @xmath259 and @xmath260 of @xmath254 are orthogonal and of the same length .
@xmath289 we have @xmath290 but @xmath291 and @xmath292 .
hence @xmath293 now we ask the reader to look ahead and see that in subsection [ frommatricestopolygons ] we define a map @xmath294 from @xmath29 to the space of @xmath3-gons in @xmath9 .
furthermore @xmath295)$ ] is the @xmath3-gon underlying the imploded spin - framed @xmath3-gon represented by @xmath255 $ ] .
accordingly we obtain the following corollary of the above lemma .
@xmath296 is four times the square of the perimeter of the @xmath3-gon in @xmath9 underlying the imploded spin - framed @xmath3-gon represented by @xmath255 $ ] .
now observe that scaling the bivector @xmath297 by @xmath298 corresponds to scaling the class of the matrix @xmath255 $ ] by the square - root of @xmath298 , note that the choice square - root is irrelevant since @xmath299 = [ a].\ ] ] but scaling the columns of @xmath254 corresponds to rotating the corresponding imploded spin - frames .
thus fixing the perimeter of the underlying @xmath3-gon to be one corresponds to taking the symplectic quotient of the cone of decomposable bivectors by @xmath94 at level @xmath155 .
and we have now completed the proof of the second part of theorem [ polygongrass ] .
we have identified the momentum map for the action of @xmath21 on @xmath301 .
the torus @xmath201 acts on @xmath301 by scaling the columns .
we leave the proof of the following lemma to the reader .
let @xmath302 be the momentum map for the above action of @xmath201 .
@xmath303 the action of @xmath201 descends to the quotient @xmath29 with the same momentum map ( only now @xmath254 must satisfy @xmath304 ) .
thus we have a formula for the momentum map for the action of @xmath201 on the grassmannian in terms of special representative matrices @xmath254 ( of @xmath21-momentum level zero ) .
it will be important to extend this formula to all @xmath305 of rank two .
we do this by relating the above formula to the usual formula for momentum map of the action of @xmath201 on a general two plane in @xmath306 represented by a general rank 2 matrix @xmath305 .
this momentum map is described in proposition 2.1 of @xcite .
let @xmath254 be a @xmath155 by @xmath3 matrix of rank two and @xmath255 $ ] denote the two plane in @xmath306 spanned by its columns .
then the @xmath47-th component of the momentum map of @xmath201 is given by@xmath307 ) = \frac{1}{\|a\|^2}\sum_{j , j \neq i } \|z_{ij}(a ) \|^2.\ ] ] here @xmath308 denotes the sum of the squares of the norms of the plcker coordinates of @xmath255 $ ] .
the required extension formula is them implied by [ equalityofhamiltonians ] suppose that @xmath263 .
let @xmath309 be the @xmath47-th column of @xmath254 .
then @xmath310 it suffices to prove the formula in the special case that @xmath311 .
if the first row of @xmath254 is zero then both sides of the equation are zero .
so assume the first row is not zero .
apply @xmath312 to @xmath254 so that the first row of @xmath313 is of the form @xmath314 where @xmath315 .
let @xmath316 be the @xmath47-th row of @xmath313 .
note that @xmath317 is still zero , hence the length of the second column of @xmath313 is equal to @xmath318 .
also @xmath319 .
now compute the left - hand side of the equation .
we have @xmath320 and hence ( noting that @xmath321 ) we have@xmath322 note that the sum on the right - hand side is the length squared of the second column of @xmath313 .
the lemma follows .
let @xmath323 be the cone defined by@xmath324 then @xmath325 is the cone on the hypersimplex @xmath326 of @xcite 2.2 .
note that @xmath327 is the set of nonnegative real numbers satisfying the usual triangle inequalities .
we can now apply lemma [ equalityofhamiltonians ] and the results of 2.2 of @xcite to deduce [ triangleinequalities]@xmath328 we give another proof of this theorem in terms of the side - lengths inequalities for euclidean @xmath3-gons in the next subsection .
the point of this subsection is to show that a point of @xmath29 may be interpreted as a euclidean @xmath3gon equipped with an imploded spin - frame .
the critical role in this interpretation is played by a map @xmath329 .
the map @xmath183 is the hopf map . in this section
we define a map @xmath183 from the imploded cotangent bundle @xmath331 to @xmath9 .
this map will give the critical connection between the previous constructions and polygonal linkages in @xmath9 .
we first define @xmath332 by@xmath333 ) = ad^\ast(g)(\lambda \varpi_1),\ ] ] where @xmath213 denotes the fundamental weight of @xmath21 .
we leave the following lemma to the reader .
[ circlebundle ] @xmath334 factors through the action of @xmath20 .
the map @xmath334 restricted to the complement of the point @xmath228 $ ] is a principal @xmath20 bundle ( the hopf bundle ) . define the normal frame bundle of @xmath335 to be the set of triples @xmath336 where @xmath337 is a nonzero vector in @xmath9 and the pair @xmath338 is a properly oriented frame for the normal plane to @xmath337 .
the projection to @xmath337 is a principal circle bundle .
the normal frame bundle is homotopy equivalent to @xmath11 and admits a unique nontrivial two - fold cover which again is a principal circle bundle over @xmath339 which we call the bundle of normal spin - frames for @xmath339 . in order to prove proposition [ framedngons ] below
we will need the following lemma .
we leave its proof to the reader . in what follows
we identify the coadjoint action of @xmath21 with its action ( through its quotient @xmath11 ) on @xmath9 .
[ framedvectors ] 1 . the map @xmath334 restricted to the subset of @xmath243 with @xmath340 is the spin - frame bundle of @xmath339 .
the action of @xmath20 on @xmath243 preserves the set where @xmath340 and induces the circle action on the normal spin - frames .
we have that @xmath243 is symplectomorphic to @xmath95 and @xmath341 is isomorphic as a lie algebra to @xmath9 , and so the map @xmath334 induces a map from @xmath95 to @xmath9 . the induced map is the map @xmath183 .
we now give details .
identify @xmath341 with the traceless hermitian matrices @xmath342 via the pairing @xmath343 given by @xmath344 in this identification we have @xmath345 we recall there is a lie algebra isomomorphism @xmath312 from @xmath9 to the traceless skew - hermitian matrices given by given by @xmath346 the isomorphism @xmath347 dual to @xmath312 from the traceless hermitian matrices @xmath342 to @xmath9 is given by @xmath348 note that in particular , @xmath349 .
we now define the map @xmath183 .
define @xmath350 as follows .
recall @xmath351 then define @xmath352 where the superscript zero denotes traceless projection .
we define @xmath183 by@xmath354 the map @xmath355 is the momentum map for the action of @xmath21 on @xmath95 after we identify @xmath342 with the dual of the lie algebra of @xmath21 using the imaginary part of the trace .
accordingly we will use the notations @xmath355 and @xmath356 interchangeably . in the next lemma
we note an important equivariance property of @xmath183 .
let @xmath357 be the double cover .
we leave its proof to the reader .
[ equivariance ] let @xmath358 .
then@xmath359 we next have we have a commutative diagram @xmath360 first observe that @xmath361 and @xmath362 are homogeneous of degree one with respect to the @xmath222 actions on their domain and range ( note that @xmath233 is homogeneous of degree @xmath363 and @xmath183 is homogeneous of degree two ) . also both intertwine the @xmath21 action on @xmath364 with its action through the double cover on @xmath9 . since the action of @xmath227 on @xmath330 is transitive away from @xmath228 $ ] it suffices to prove the two above maps coincide at the point @xmath365 $ ]
but it is immediate that both maps take the value @xmath366 at this point .
[ spinf ] we will need the following calculation .
@xmath367 ) = \lambda f ( ad_g^*(\varpi_1))\ ] ] in particular this implies that @xmath368)\| = \frac{1}{2}\lambda$ ] .
the following lemma is a direct calculation .
[ factoroftwo ] the formula for @xmath369 in the usual coordinates is@xmath370 consequently , the euclidean length of the vector @xmath371 is given by@xmath372 [ spinlength ] note that the length of @xmath183 is related to the hamiltonians @xmath242 for the infinitesimal action of @xmath172 by@xmath373 also , by the above remark if @xmath374 ) = ( z , w)$ ] then @xmath375 later we will need the following determination of the fibers of @xmath183 .
[ phasefactor ] @xmath376 the implication @xmath377 is immediate .
we prove the reverse implication .
thus we are assuming the equations @xmath378 square each side of the first equation .
take four times the norm squared of each side of the second equation and adding the resulting equation to the new first equation to obtain@xmath379 and consequently@xmath380 now it is an elementary version of the first fundamental theorem of invariant theory that if we are given two ordered pairs of vectors in the plane so that the lengths of corresponding vectors are equal and the symplectic inner products between the two vectors in each pair coincide then there is an element in @xmath381 that carries one ordered pair to the other ( this is one definition of the oriented angle between two vectors ) . we now construct the required map to euclidean @xmath3-gons .
we define a map @xmath382 by defining @xmath383 to be the orbit of @xmath384 under the diagonal action of @xmath11 . here
@xmath309 is the @xmath47-th column of @xmath254 .
let @xmath385 denote the space of closed @xmath3-gons in @xmath9 .
[ framedngons ] 1 .
@xmath294 induces a homeomorphism from @xmath386 onto @xmath387 .
the fiber of @xmath294 over a euclidean @xmath3-gon is naturally homeomorphic to the set of imploded spin framings of the edges of that @xmath3-gon .
the side - lengths of @xmath383 are related to the norm squared of the columns of @xmath254 by @xmath388 here @xmath389 is the @xmath47-th edge of any @xmath3-gon in the congruence class represented by @xmath383 .
we first prove that if @xmath254 is in the zero level set of @xmath180 then @xmath383 is a _ closed _ @xmath3-gon .
recall @xmath390 be the @xmath47-th column of @xmath254 .
then we have@xmath391 hence the sum @xmath392 of the edges @xmath383 is given by@xmath393 now it is a formula in elementary matrix multiplication that we have@xmath394 but by lemma [ levelzero ] we find that @xmath395 is scalar whence@xmath396 and@xmath397 we next prove that @xmath294 is onto .
it is clear that @xmath294 maps @xmath398 onto @xmath399 ( this may be proved one column at a time ) .
let @xmath400 be the edges of a closed @xmath3-gon in @xmath9 .
choose @xmath305 such that @xmath401 .
but by the above@xmath402 hence @xmath396 whence @xmath395 is scalar and @xmath263 .
we now prove that @xmath294 is injective .
suppose there exists @xmath358 such that @xmath403
. then @xmath404 .
hence by lemma [ phasefactor ] we have @xmath405 for some @xmath406 .
the second statement follows because the action of @xmath201 corresponds to a transitive action on the set of imploded spin framings . indeed using the identification above between between @xmath9 and @xmath407 we
may replace @xmath294 by the map @xmath408 given by@xmath409,\cdots,[g_n,\lambda_n \varpi_i])= ( ad^{\ast}g_1(\lambda_1 \varpi_1 ) , \cdots , ad^{\ast}g_n(\lambda_n \varpi_1)).\ ] ] if no @xmath410 is zero by lemma [ framedvectors ] the ( classes of ) the @xmath3-tuple @xmath411,\cdots , [ g_n,\lambda_n \varpi_1]$ ] represent the imploded spin - frames over the @xmath3 vectors in the image .
then by lemma [ circlebundle ] and lemma [ framedvectors ] two such @xmath3tuples correspond to frames over the same image @xmath3-gon if and only the two @xmath3tuples are related by right multiplication by @xmath412 with @xmath413 fixing @xmath213 for all @xmath47 . hence if no @xmath410 is zero then the two @xmath3tuples are related as above if and only if they are in the same @xmath201 orbit by definition of the @xmath201 action . since @xmath201 acts transitively on the spin - frames over a given @xmath3-gon we see that in this case the fiber of @xmath414 is the set of spin - frames over the image @xmath3-gon and the second statement is proved .
if some subset of the @xmath410 s is zero then we replace the corresponding @xmath415 s by the identity ( this does not change the imploded frame ) and procede as above with the remaining components .
the last statement in the theorem is lemma [ factoroftwo ] .
let @xmath416 be the map that assigns to a closed @xmath3-gon the lengths of its sides .
it is standard ( see for example the introduction of @xcite ) that the image of @xmath417 is the polyhedral cone @xmath325 .
we can now give another proof of proposition [ triangleinequalities ] based on euclidean geometry .
we restate it for the convenience of the reader .
@xmath418 the proposition follows from the commutative diagram@xmath419 we now compute @xmath421 using the fact that @xmath11 is covered by @xmath21 .
let the semi simple lie group @xmath422 be a quotient of the semi simple lie group @xmath149 by a finite group @xmath423 @xmath424 in example 4.7 of @xcite , guillemin - jeffrey - sjamaar , show that this quotient gives a description nomeomorphism @xmath425 .
the following lemma is left to the reader to prove .
[ quotientimplosion ] the homeomorphism @xmath233 induces a homeomorphism from @xmath426 to @xmath427 .
[ compare ] the double cover @xmath428 induces a homeomorphism from @xmath429 to @xmath421 .
here h is the finite 2-group @xmath430 and @xmath431 is the center of @xmath21 .
we have the following analogue of theorem [ framedngons ] above .
we leave its proof to the reader .
we now have orthogonal ( imploded ) framings instead of imploded spin framings .
we note that the map @xmath294 above descends to a map @xmath432 .
[ orthogonalframedngons ] 1 .
@xmath433 induces a homeomorphism from @xmath434 onto the polygon space @xmath387 .
the fiber of @xmath433 over a euclidean @xmath3-gon of the induced map from @xmath29 onto @xmath387 is naturally homeomorphic to the set of imploded orthogonal framings of the edges of that @xmath3-gon .
suppose one has a planar regular @xmath3-gon subdivided into triangles .
we call this a triangulation of the @xmath3-gon .
the dual graph is a tree with @xmath3 leaves and @xmath79 internal trivalent nodes ( see figure [ fig : triangulated_hexagon ] ) .
we will see below how the tree @xmath2 determines a grbner degeneration of the grassmannian @xmath1 to a toric variety .
these toric degenerations first appeared in @xcite .
recall that the standard coordinate ring @xmath435 $ ] ( for the plcker embedding ) of the grassmannian is generated by @xmath436 , for @xmath437 , subject to the quadric relations , @xmath438 for @xmath439 .
these relations generate the plcker ideal @xmath440 . for each pair of indices
@xmath441 let @xmath442 denote the length of the unique path in @xmath2 joining leaf @xmath47 to leaf @xmath274 .
for example , in figure [ fig : triangulated_hexagon ] we have @xmath443 .
to any monomial @xmath444 we assign a weight @xmath445 let @xmath446 denote the initial ideal with respect to the weighting @xmath447 .
it is a standard result in the theory of grbner degenerations that one has a flat degeneration of @xmath448 = \c[\{z_{i , j}\}_{i < j } ] / i_{2,n}$ ] to @xmath449 / i_{2,n}^\tree$ ] .
below we outline how this works using the rees algebra .
the weight @xmath447 induces an ( increasing ) filtration on the ring of the grassmannian .
let @xmath450 be the vector subspace of @xmath285 spanned by monomials of weight at most @xmath451 .
then , for any elements @xmath452 and @xmath453 , the product @xmath454 belongs to @xmath455 .
let @xmath456 denote the associated graded ring @xmath457 , where @xmath458 .
the rees algebra @xmath459 is given by , @xmath460 , where @xmath461 is an indeterminant .
then ( cf .
@xcite ) @xmath459 is flat over @xmath462 $ ] , @xmath463 } \c[t , t^{-1 } ] \cong r[t , t^{-1}]$ ] , and @xmath463 } ( \c[t]/(t ) ) \cong r^\tree$ ] .
since the ring @xmath285 is already graded , there is a natural grading of the rees algebra @xmath459 , where the degree of the indeterminant variable @xmath461 is defined to be zero , and in general , the degree of a product @xmath464 , where @xmath465 , is equal to @xmath466 .
now , the specializations @xmath463 } \c[t]/(t - a)$ ] ( at @xmath467 ) are all graded algebras , where @xmath468 for @xmath469 .
in particular the associated graded algebra @xmath456 is bigraded ; one grading comes from the grading of @xmath285 , and the other from the filtration of @xmath285 .
we have that @xmath470 has a flat morphism to the affine line @xmath471)$ ] with all fibers projective varieties , with general fiber isomorphic to @xmath1 and special fiber @xmath472 at @xmath473 .
we shall reserve a special name for the case that the triangulation is such that all diagonals contain the initial vertex , forming a fan .
the associated tree has the appearance of a caterpillar and was called such in @xcite .
we shall call this triangulation the `` fan '' and the associated degeneration the lg degeneration , since the special fiber in this case agrees with the special fiber of the degeneration of @xmath1 given by lakshmibai and gonciulea in @xcite .
now we will see why the special fiber @xmath474 } \c[t]/(t)$ ] is toric .
this was also proved in @xcite where these degenerations were first discovered .
first we establish a basis for @xmath448 $ ] as a complex vector space .
suppose that the vertices @xmath475 of a multigraph ( multiple edges allowed between two vertices ) are drawn on the unit circle , in cyclic clockwise order , and all edges are drawn as straight line segments ( a chord of the circle ) .
then , if no two edges cross ( it is okay for an edge to have multiplicity greater than one ) we call the graph a kempe graph , in honor of the work of a. kempe in 1894 ( see @xcite ) . for an example see figure [ fig : kempe_graph ] .
in fact two edges @xmath275 and @xmath476 cross exactly when @xmath477 , assuming that @xmath478 .
we assign a monomial @xmath479 to each graph @xmath149 , given by @xmath480 where @xmath481 is the multi - set of edges of @xmath149 . for a graph @xmath149 ,
let @xmath482 be the number of edges in @xmath149 , which is one half of the total sum of valencies of the all the vertices .
note that @xmath483 .
the proof of the proposition and theorem below appears in @xcite .
the monomials @xmath479 as @xmath149 runs over the set of all kempe graphs with vertex set @xmath484 form a @xmath485-basis for @xmath448 $ ] .
furthermore , the @xmath479 for which @xmath486 and @xmath149 is a kempe graph form a basis for the @xmath487-th graded piece of @xmath285 .
[ leadingterm ] suppose that @xmath488 and @xmath489 are kempe graphs .
suppose that @xmath490 then , there is a unique kempe graph @xmath491 such that * @xmath492 , * @xmath493 , * for all @xmath149 , if @xmath494 and @xmath495 then @xmath496 .
suppose that @xmath497 is an edge of @xmath488 and @xmath498 is an edge of @xmath489 such that @xmath497 and @xmath498 cross . without loss of generality we can assume that @xmath499 .
we have the plcker relation @xmath500 , and the latter two terms ( viewed as graphs having two edges ) are kempe graphs since neither pair of edges cross .
denote @xmath501 as the graph with edge set @xmath502 .
let @xmath503 with the two edges @xmath497 and @xmath498 removed . we have , @xmath504 we will show that one of the two monomials on the right hand side of the above equation has the same @xmath2weight ( @xmath447 ) as does @xmath505 and the other term has strictly smaller weight .
after sufficiently many plcker relations as above are applied we get @xmath505 expressed as a unique integral combination of kempe graphs with a unique term of maximal weight equal to the weight of @xmath505 .
let @xmath506 denote the shortest path in @xmath2 joining @xmath47 to @xmath274 .
note that the paths @xmath507 and @xmath508 must intersect one another since they cross when drawn as straight line segments in the graph @xmath501 .
now consider the two pairs of paths in @xmath2 : * @xmath509 and @xmath510 .
* @xmath511 and @xmath512 .
exactly one of the above two above pairs of paths cover precisely the same set of edges in @xmath2 as does the crossing pair ( @xmath507,@xmath508 .
say for example it is the pair ( @xmath509 , @xmath510 ) . then @xmath513 .
furthermore the two paths @xmath509 and @xmath510 must meet one another within the tree @xmath2 along some internal edges @xmath514 ( although they are non - crossing when drawn as straight line segments ) .
the edges @xmath514 are also the edges common to the two paths @xmath507 and @xmath508 .
the third pair of paths ( @xmath511 , @xmath512 ) covers the same set of edges as do the first two pairs of paths , excepting the edges @xmath514 above .
in particular we have that @xmath515 let @xmath516 denote the kempe graph @xmath491 from the above theorem .
let @xmath517 denote the graded commutative semigroup of kempe graphs @xmath149 with binary operation @xmath518 .
the grading is given by @xmath519 .
( to theorem [ leadingterm ] ) the special fiber at @xmath473 of the degeneration , @xmath520 } \c[t]/(t)$ ] , is isomorphic to the semigroup algebra @xmath521 $ ] . given any graph @xmath149 , there is an associated weighting @xmath522 of the tree @xmath2 where the weight assigned to an edge @xmath523 of @xmath2 is equal to the number of edges @xmath275 in @xmath149 such that the path @xmath506 in @xmath2 joining @xmath47 to @xmath274 passes through @xmath523 .
we denote this weight by @xmath524 .
we now determine the image of the map @xmath525 .
[ admissibleweighting ] given a weighting @xmath526 of nonnegative integers to the edges of @xmath2 , then @xmath527 for some graph @xmath149 iff for each triple @xmath528 of edges meeting at a common ( internal ) vertex of @xmath2 the sum @xmath529 is even , and @xmath530 satisfy the triangle inequalities .
let a @xmath149path mean a path in the tree @xmath2 joining vertices @xmath47 to @xmath274 where @xmath275 is an edge of @xmath149 . thus @xmath149-paths are in bijection with edges of @xmath149 , and are meant to be counted with multiplicity .
now fix an internal vertex @xmath531 of @xmath2 with neighboring vertices @xmath532 with connecting respective edges @xmath528 . for each @xmath441 with @xmath533 ,
let @xmath534 be the number of @xmath149-paths passing through each of @xmath142 and @xmath143 .
thus * @xmath535
. * @xmath536 .
* @xmath537 .
now solving for the @xmath534 we have : * @xmath538 .
* @xmath539 . * @xmath540 .
since the @xmath534 are non - negative we have that @xmath541 , @xmath542 , and @xmath543 satisfy the triangle inequalities . since the @xmath534 are integers we have that @xmath544 is even .
conversely , given a weighting of @xmath2 satisfying the above conditions , we will get non - negative integers @xmath545 at each internal vertex @xmath337 .
we may thus define a graph @xmath546 on the three neighboring vertices @xmath547 by setting the multiplicity of edge @xmath275 to be @xmath548 .
if @xmath549 is a neighboring internal vertex , say for example sharing edge @xmath550 with @xmath337 , then we have @xmath551 .
hence we may glue together @xmath552 and @xmath553 to form a kempe graph @xmath554 on the neighboring vertices of @xmath337 and @xmath549 by adjoining the edges whose respective paths go through the edge @xmath555 of @xmath2 .
note that the weighting of the edges of the tree pertaining to @xmath554 is equal to @xmath526 .
continuing in this way , we may form a kempe graph @xmath149 on the leaves of @xmath2 such that @xmath527 , by gluing together all the @xmath546 for @xmath337 and internal vertex of @xmath2 . we shall call a weighting @xmath526 or @xmath2 an admissible weighting if @xmath527 for some graph @xmath149 . for each admissible weighting @xmath526 of @xmath2
, there exists a unique kempe graph @xmath149 such that @xmath527 . furthermore , if @xmath488 and @xmath489 are kempe graphs , then @xmath556 .
certainly the paths in the tripods of the proof of proposition [ admissibleweighting ] may be drawn so that they are non - crossing .
such a non - crossing tripod graph is unique . in the process of gluing all these tripod graphs together to form a graph @xmath149 such that @xmath557 , it is clear there is only one way to join the paths so that no two paths are non - crossing .
now by straightening these non - crossing paths into line segments , we see that they remain non - crossing and so form a kempe graph .
a quick inspection of the proof of theorem [ leadingterm ] reveals that the graph @xmath516 has weight equal to the sum @xmath558 .
note that @xmath559 .
we define the degree of an admissible weighting @xmath560 to be one -half the sum of the weights of the leaf edges ( a leaf edge is an edge incident to a leaf ) .
since each edge of the graph @xmath149 contributes a weight of one to two leaf edges we have @xmath561 define @xmath562 to be the graded semigroup of admissible weightings of the edges of @xmath2 .
[ gradedsemigroupiso ] the map that associates to an element of @xmath563 the induced weighting of the edges of @xmath2 induces an isomorphism @xmath564 . here
we briefly review the construction of the git quotients of the grassmannian by the maximal torus @xmath191 in @xmath566 . the git quotient ( or technically , the subspace of closed points in the git quotient ) is homeomorphic to a configuration space @xmath565 of polygonal linkages in @xmath9 with prescribed integral side lengths @xmath567 .
see @xcite , theorem 2.3 , or @xcite for more details .
let @xmath191 be the torus of diagonal matrices in @xmath566 .
there is a natural action of @xmath191 on the grassmannian @xmath1 .
the git quotient @xmath568 depends upon the choice of a @xmath191-linearized line bundle of @xmath1 .
let @xmath569 be the line bundle associated to the plcker embedding of @xmath1 .
let @xmath570 be the parabolic subgroup fixing the point @xmath571 .
it is the subgroup of matrices @xmath572_{1 \leq i , j \leq n}$ ] of determinant one , such that @xmath573 for @xmath574 and @xmath575 .
we identify @xmath1 with the homogeneous space @xmath576 , by identifying @xmath577 with @xmath578 .
let @xmath579 be the second fundamental weight of @xmath580 .
associated to @xmath581 is a character @xmath582 given by @xmath583_{1 \leq i , j \leq
n } ) = \det \begin{pmatrix } a_{11 } & a_{12 } \\
a_{21 } & a_{22 } \end{pmatrix}$ ] . following the borel - weil construction
, we may take the total space of @xmath569 to be the product @xmath584 modulo the equivalence relation @xmath585 , for all @xmath586 , @xmath587 , and @xmath588 .
we denote the equivalence class of @xmath589 as above by @xmath590 $ ] .
the bundle map @xmath591 is given by @xmath590 \mapsto gp$ ] .
the plucker coordinate ring of @xmath1 is now @xmath592 , and it is generated in degree one by the brackets @xmath273 $ ] which are associated to global sections @xmath593 of @xmath569 , by @xmath594 $ ] , where @xmath595_{1 \leq k , l \leq n}$ ] .
we suppose that @xmath596 is an even integer .
the line bundle @xmath597 of @xmath569 may be identified with the product @xmath584 modulo the relation @xmath598 .
we define an action of @xmath191 on @xmath597 via the character @xmath599 of @xmath191 , given by @xmath600 .
we define the action of @xmath191 on @xmath597 by @xmath601 = [ \mathbf{t } \cdot g , \chi_\br(\mathbf{t})z],\ ] ] where @xmath602 .
we call this the @xmath6linearization of @xmath597
. the space @xmath565 is homeomorphic to @xmath603 where @xmath604 indicates the @xmath191-invariant global sections of @xmath605 .
the ring @xmath606 is naturally graded by @xmath279 .
the invariant sections of @xmath607 are spanned by monomials @xmath608 over all graphs @xmath149 having multi - degree @xmath609 , i.e. the valency of vertex @xmath47 is @xmath610 for each @xmath47 .
the degree of @xmath479 is then @xmath279 , if @xmath149 has multidegree @xmath611 .
restricting to torus invariants is an exact functor so the flat degenerations of the grassmannian described above restrict to flat degenerations of @xmath612 .
furthermore , the special fiber of this restricted degeneration is toric , since it is the @xmath191-quotient of a toric variety .
therefore we obtain a flat degeneration of @xmath612 to a toric variety @xmath613 for each triangulation @xmath2 of the model @xmath3-gon .
the associated semigroup @xmath614 is the set of kempe graphs having valency a multiple of @xmath6 ; it is a sub - semigroup of @xmath517 , however the grading of @xmath614 is not the same as that of @xmath517 .
instead the degree of a kempe graph @xmath615 of multidegree @xmath611 is @xmath279 , rather than @xmath616 as it would have in @xmath517 .
we have that @xmath617),\ ] ] with @xmath614 graded as mentioned above .
define the graded semigroup @xmath618 to be the graded subsemigroup of @xmath619 with leaf - edge weights that are integral multiples of @xmath6 .
the admissible weightings of the tree @xmath2 relating to the sub - semigroup @xmath614 must satisfy that the weighting of the outer edges @xmath620 ( the edge @xmath131 is adjacent to leaf @xmath47 ) is some multiple @xmath279 of @xmath6 , meaning that @xmath621 for each @xmath47 , and the multiple @xmath279 is the degree of the weighting .
thus we have [ gradedsubsemigroupiso ] the isomorphism @xmath622 of proposition [ gradedsemigroupiso ] induces an isomorphism @xmath623 .
in this section we will define an equivalence relation on polygons and polygonal linkages which depends on the choice of a trivalent tree @xmath2 .
first we will collect some results about trivalent trees which will be useful in what follows .
let @xmath2 be a trivalent tree with @xmath3 leaves which we assume is dual to a triangulation of @xmath38 .
we will say a vertex is _ internal _ if it is not a leaf .
the triangles in the triangulation of @xmath38 correspond to the internal vertices of @xmath2 .
we will say an edge is a _ leaf edge _ or an _ outer edge _ if it is incident to a leaf .
thus the leaf edges are dual to the edges of @xmath38 .
an edge of @xmath2 which does not border a leaf will be called an _ inner edge_. thus the inner edges of @xmath2 are dual to the diagonals of the triangulation of @xmath38 .
we say two leaves of @xmath2 are a matched pair of leaves if they have a common neighbor .
the following technical lemma will be very useful for giving inductive proofs .
[ trivalentinduction ] for any trivalent tree @xmath2 it is possible to find a sequence of subtrees @xmath624 such that * the tree @xmath40 is a tripod . *
the tree @xmath625 can be identified with @xmath626 joined with a tripod along some @xmath131 . *
each internal edge of @xmath2 appears as some @xmath131 .
let @xmath627 be a trivalent tree .
let @xmath628 be the ( not necessarily trivalent ) subtree of @xmath627 with vertex set the internal vertices of @xmath627 and edge set equal to all edges of @xmath627 not connected to leaves .
it is easy to see that @xmath628 is also a tree and therefore has a leaf .
let @xmath629 be the vertex corresponding to this leaf , then @xmath3 is trivalent by definition and therefore must be connected to two leaves in @xmath2 .
this shows that @xmath2 must have a vertex @xmath337 connected to two leaf edges , @xmath337 is connected to a third edge , @xmath523 , by splitting @xmath2 along @xmath523 we obtain a tripod and a new trivalent tree .
the three items above now follow by induction . for each tree
@xmath2 we choose once and for all an ordering on the trivalent vertices of @xmath2 as follows .
pick any matched pair of leaves and label their trivalent vertex @xmath630 with @xmath3 the number of leaves of @xmath2 . then , repeat this proceedure for the tree given by removing @xmath630 and its matched pair .
note that this ordering is not unique .
we now discuss the decompositions of @xmath2 obtained by splitting certain internal edges .
we first describe the decompositions of @xmath38 which will induce the required decompositions of @xmath2 . for a diagonal @xmath631 in the triangulation @xmath2
, we define the operation of splitting @xmath38 along @xmath631 by removing a small tubular neighborhood of @xmath631 in @xmath38 .
this results in a disjoint union of two triangulated polytongs @xmath632 .
we may generalize this operation by performing the same operation for any subset @xmath44 of the set of diagonals of @xmath38 to obtain a union of triangulated polygons @xmath633 .
the dual graph to @xmath633 is a forest @xmath634 that may be obtained by removing a small open interval from each edge corresponding to a diagonal in @xmath44 . if we choose @xmath44 to be the entire set of diagonals we obtain the decomposition @xmath635 of @xmath38 into triangles and @xmath2 into the forest @xmath81 of trinodes
. it will be important in what follows that we have the quotient map @xmath636 that glues together the pairs of vertices that are the boundaries of the open intervals removed from @xmath2 .
we will say two vertices of @xmath634 are equivalent if they have the same image under @xmath637 and two edges of @xmath634 are equivalent if they contain equivalent vertices .
there are two types of leaf edges of @xmath634 . the first type correspond to edges of @xmath38 ( leaf edges of @xmath2 ) and the second correspond to ( split ) diagonals of @xmath38 .
the @xmath3 edges of @xmath634 corresponding to leaf edges of @xmath2 will from now on be referred to as distinguished edges , we denote this set @xmath638 . for a connected component @xmath639 of @xmath634
we let @xmath640 .
note that these components uniquely partition the set @xmath641 .
fix a triangulation of the model n - gon with dual tree @xmath2 .
recall @xmath385 , the space of n - gons in @xmath9 .
a polygon @xmath642 comes with a fixed ordering on its edges , these ordered edges are in bijection with the leaves of the tree @xmath2 .
a set of diagonals @xmath44 corresponds to a set of internal edges of @xmath2 .
we know that such a set defines a unique partition of the edges of @xmath643 , @xmath644 given by grouping together the distinguished edges of @xmath634 that lie in a component tree of the forest @xmath634 . for @xmath645 , if all diagonals in the set @xmath44 are zero then the edge sets @xmath646 define closed polygons .
split the polygon @xmath38 along @xmath44 to obtain a union of polygons .
choose a polygon @xmath647 in the union .
then @xmath647 is dual to a component @xmath309 of the forest @xmath634 .
the edges of the polygon @xmath647 are either edges of the orginal original polygon ( so distinguished leaves of the tree @xmath309 ) and hence edges of the euclidean polygon @xmath128 or diagonals of @xmath44 and hence zero diagonals of the euclidean polygon @xmath128 .
we assume that we have chosen @xmath47 consistently with the division of the distinguished edges above whence the set of distinguished edges of @xmath309 equals the set @xmath648 . now
since @xmath647 closes up the sum of all the vectors in @xmath9 associated to the edges of @xmath647 is zero . but this sum is the sum of the vectors in @xmath9 associated to the distinguished edges of @xmath309 ( that is the elements in @xmath648 ) and a set of vectors all of which are zero
. the following groups will be useful in defining structures on spaces of polygons .
let @xmath649 be the group of maps from the set @xmath638 into a group @xmath149 .
define @xmath650 to be the subgroup of @xmath649 of maps which are constant along the distinguished edges of each component @xmath639 of @xmath634 notice that @xmath650 naturally splits as a product of @xmath149 over the components of @xmath634 .
let @xmath651 be the component corresponding to the component @xmath639 .
the sets @xmath44 of zero diagonals define a filtration of the space @xmath385 where the subspace @xmath652 in the filtration is defined to be the set of all points @xmath643 such that the diagonals in @xmath44 have zero length .
this in turn defines a decomposition of @xmath385 into subspaces . @xmath653 } = \mathrm{pol } _
n(\r^3)^s \setminus \cup_{s \subset j } \mathrm{pol } _ n(\r^3)^j\ ] ] the subspace @xmath654}$ ] is the collection of all points in @xmath385 such that exactly the diagonals in @xmath44 have zero length .
this also induces a filtration on @xmath5 .
define an action of @xmath655 on @xmath654}$ ] by letting @xmath656 act diagonally on the edges in @xmath646 .
the equivalence relation given by @xmath655 orbit type on @xmath654}$ ] fit together to give an equivalence relation @xmath657 , which we call @xmath2-congruence , on @xmath658 .
we may describe the above equivalence relation geometrically as follows .
a polygon @xmath659}$ ] is a wedge of @xmath660 closed polygons @xmath661 wedged together at certain vertices of the polygon @xmath128 .
although each @xmath661 may contain several wedge points , since the vertices of @xmath128 are ordered there will be a first wedge point @xmath142 .
then we apply a rotation @xmath415 about @xmath142 to each @xmath661 for @xmath662 and identify points in the resulting orbit of @xmath655 .
@xmath2-congruence for @xmath385 induces an equivalence relation on the space of @xmath3-gon linkages @xmath663 , which we also call @xmath2-congruence .
define @xmath664 kamiyama and yoshida studied the space @xmath65 for the special case when @xmath2 was the _ fan _ tree .
note that @xmath65 inherits a filtration by the subspaces @xmath665 .
let @xmath666 } = ( v_{\br}^{\tree})^{s } \setminus \cup_{s \subset j } ( v_{\br}^{\tree})^j$ ] .
the spaces @xmath666}$ ] define a decomposition of @xmath65 .
we can say more about the pieces of this decomposition .
@xmath666}$ ] is canonically homeomorphic to @xmath667 where @xmath668 is the subvector of linkage lengths corresponding to the elements in @xmath669 , and @xmath670 is the dense open subset of @xmath671 corresponding to polygons with no zero diagonals .
let us first describe a map @xmath672 .
the diagonals @xmath44 define a partition of edges sets @xmath648 for each @xmath673 . by the above proposition
, @xmath648 corresponds to a closed polygon .
so we may send a member of the equivalence class of @xmath128 to the product of the equivalence classes defining these closed polygons in @xmath674 , with the appropriate @xmath668 .
this map is certainly onto , and by the definition of @xmath675 , it factors through the relation @xmath676 .
hence , we get a 1 - 1 and onto continuous function @xmath677 , which is a homeomorphism by the compactness of @xmath674 .
we may restrict this map to the spaces @xmath666}$ ] which define the induced decomposition of @xmath65 . recall that these are the polygons with exactly the @xmath44 diagonals zero .
we therefore obtain that @xmath666}$ ] is homeomorphic to @xmath678 , since the fibers of @xmath679 are sometimes odd - dimensional , the quotient map @xmath334 can not be algebraic even when @xmath6 is integral . in this section
we introduce a generalization of the @xmath2-congruence relation by lifting @xmath2-congruence from @xmath385 to @xmath29 . in section 3
we constructed a map @xmath680 . pulling back the decomposition of @xmath387 into @xmath2- congruence classes by @xmath294
produces a decomposition of @xmath29 by the spaces we will denote @xmath681 } = f_n^{-1}(\mathrm{pol } _ n(\r^3)^{[s]})$ ] .
[ p_nstrat ] elements of @xmath681}$ ] are the spin - framed polygons in @xmath29 such that the following equation holds , @xmath682 where the sum is over all @xmath683\in \mathbf{e}_j(\mathbf{e})$ ] , the edges in the @xmath274-th partition defined by @xmath44 , for each @xmath274 .
this follows from theorem [ framedngons ] and remark [ spinf ] . by the previous lemma
we have a decomposition of @xmath684}$ ] into imploded spin - framed polygons @xmath685 we define an action of @xmath686 on @xmath681}$ ] by letting @xmath687 act diagonally on the framed edges associated to the component @xmath639 of @xmath634 .
the @xmath686-orbits on @xmath681}$ ] fit together to give an equivalence relation @xmath657 , which we again call @xmath2-congruence , on @xmath29 .
@xmath688 note that this defines a decomposition on @xmath59 into subspaces @xmath689 } = p_n(\mathrm{su}(2))^{[s]}/\mathrm{su}(2)^{d(s)}$ ] .
we have seen that @xmath29 carries an action of @xmath30 given by the following formula .
let @xmath690 , @xmath691 , \ldots , [ g_nt_n , \lambda_n\varpi_1).\ ] ] in particular , note that the action of the diagonal @xmath692 element in @xmath30 is trivial , because this corresponds to acting on the left by the nontrivial central @xmath21 element .
since @xmath413 fixes @xmath213 for @xmath252 this action does not change the image of a point under @xmath294 , hence it fixes each piece of the decomposition .
also this action commutes with the @xmath686-action on the piece @xmath681}$ ] , so the action of @xmath30 must descend to @xmath59 .
in this section we will construct the affine toric variety @xmath693 of imploded triangulated @xmath21framed @xmath3gons ( without imposing the condition that the side - lengths are @xmath6 ) and its projective quotient @xmath694 .
we have tried to follow the notation of @xcite when possible . in @xcite
the superscript @xmath695 on @xmath635 refers to a `` pair of pants '' decomposition @xmath695 of the surface . for us
superscript @xmath2 on @xmath105 refers to the triangulation of the @xmath3-gon .
the connection is the following . under the correspondence between moduli spaces of @xmath3-gons and character varieties briefly explained in remark [ moduliofconnections ] ( and explained in detail in 5 of @xcite ) the standard triangulation corresponds to the following `` pair of pants '' decomposition of the @xmath3 times punctured two - sphere .
represent the sphere as the complex plane with a point at infinity .
take the punctures to be the points @xmath696 on the real line .
draw small circles around the punctures .
now draw n-3 more circles with centers on the @xmath186-axis , so that the first circle contains the small circles around @xmath34 and @xmath155 , the next circle contains the circle just drawn and the small circle around 3 and the last circle contains all the previous circles except the small circles around @xmath697 and @xmath3 .
the graph dual to the pair of pants decomposition is the tree @xmath2 that is dual to the triangulation .
furthermore the decomposed tree @xmath81 is dual to the pair - of - pants decomposition of the @xmath3-punctured sphere obtained by cutting the sphere apart along the above @xmath698 circles - we might say that @xmath81 is the pair - of - pants decomposition of @xmath2 .
using this correspondence the reader should be able to relate what follows with @xcite for the case of the @xmath3-fold punctured sphere
. it will be important in what follows that we have earlier defined the quotient map @xmath699 that glues together the pairs of vertices that are the boundaries of the open intervals removed from @xmath2 .
we define @xmath700 to be the product of @xmath208 over the edges of @xmath81 , hence an element @xmath701 is a map from the @xmath82 edges of @xmath81 into @xmath208 or equivalently an assignment of an element of @xmath208 to each each of the @xmath82 edges of the triangles @xmath702 in the triangulation of @xmath38 .
it will be important later to note that there is a forgetful map that restricts @xmath703 to the distinguished edges of @xmath81 to obtain an element @xmath247 of @xmath245
. it will be convenient to represent the resulting product @xmath704 by a @xmath155 by @xmath705 matrix . to do this
we will linearly order the @xmath705 edges by first ordering the @xmath79 triangles ( tripods ) and then ordering the @xmath706 edges for each triangle ( tripod ) .
now use the lexicographic ordering on pairs @xmath707 ( @xmath708 ) .
the point is that in the matrix @xmath709 below the @xmath710 s belonging to the same triangle ( tripod ) appear consecutively .
let @xmath711 denote the @xmath274-th edge of the @xmath47-th tripod .
we will use the following notation for the entries in the matrix @xmath709 to indicate that each successive group of @xmath706 columns of @xmath709 belongs to a triangle in the triangulation .
@xmath712 in what follows we will abbreviate @xmath713 to @xmath700 .
the action of the group @xmath715 on @xmath713 is then represented by acting on the columns of @xmath709 .
we will represent elements @xmath716 of @xmath715 as @xmath705-tuples @xmath717 we let @xmath718 denote the `` diagonal '' subgroup of @xmath715 defined by the condition that for each triangle @xmath719 ( tripod @xmath720 ) we have @xmath721 we will regard @xmath718 as the space of mappings @xmath722 from the tripods in @xmath81 to @xmath21 .
we then define the space @xmath714 as the symplectic quotient @xmath723 thus @xmath714 is obtained by taking the symplectic quotient of each triple of edges belonging to one of the triangles @xmath724 , by the group @xmath718 .
the resulting space @xmath725 is clearly the product of @xmath79 copies of @xmath726 .
+ we will often denote an element of @xmath714 by @xmath727,[t_2],\cdots,[t_{n-2}])$ ] where @xmath728 $ ] , the @xmath47-th triangle denotes the equivalence class of the matrix @xmath729 such that the momentum image of @xmath730 under the momentum map for @xmath21 is zero ( equivalently the rows are orthogonal with the same length ) . in equation ( [ p3 ] )
we have seen that @xmath731 since the space of imploded triangles is the product of @xmath79 copies of @xmath732 we see that @xmath714 is the affine space obtained by taking the product of @xmath79 copies of @xmath733 .
it has an action of a @xmath705 torus @xmath86 induced by the right actions of the torus @xmath734 on the @xmath705 copies of @xmath208 . in terms of our matrices
@xmath709 this amounts to scaling the columns of @xmath709 ( right multiplication of @xmath709 by @xmath86 ) .
however note that the entry of an element of @xmath86 corresponding to the edge @xmath523 of @xmath81 scales the column of @xmath709 corresponding to @xmath523 by its _
inverse_. we will use the notation @xmath735 to indicate the complexification of @xmath86 .
similarly for any compact group @xmath149 that appears in this paper , the notation @xmath736 indicates the complexification of @xmath149 .
let @xmath116 , @xmath737 , @xmath738 be as in the introduction .
now we glue the diagonals of @xmath38 back together by taking the symplectic quotient by @xmath93 at level @xmath739 .
@xmath740 the spaces @xmath742 and @xmath105 all have analogues when @xmath743 is replaced by @xmath11 which are quotients by a finite group of the corresponding spaces for @xmath21 .
there also analogues of the tori @xmath744 and @xmath93 for @xmath11 which we will denote @xmath745 and @xmath746 respectively which are quotients of the corresponding tori for @xmath21 .
we leave the details to the reader .
since affine git quotients coincide with symplectic quotients , see @xcite and @xcite , theorem 4.2 , we may also obtain @xmath105 as the git quotient of @xmath97 by the complex torus @xmath747 .
for each triangle @xmath748 ( tripod @xmath749 ) we have a corresponding @xmath96 with plcker coordinates @xmath750 .
the coordinate ring of the affine variety @xmath105 will be the ring of invariants @xmath751^{\underline{\t}^-_d}$ ] .
this ring of invariants will be spanned by the ring of invariant _ monomials _ which we now determine .
there is a technical problem here .
we need to know that we have chosen the correctly normalized momentum map for the action @xmath752 .
but by theorem [ twistandshift ] the correct normalization is the one that is homogeneous linear in the squares of the norms of the coordinates which is the one we have used here . as a geometric quotient of affine space by a torus
the space @xmath105 is an affine toric variety .
we now analyze the affine coordinate ring of @xmath105 . in what follows we label the leaf of the tripod @xmath720 incident to the edge @xmath754 by @xmath487
- we will do this only when the tripod of which @xmath487 is a vertex is clearly indicated .
we leave the proof of the following lemma to the reader .
the monomial @xmath755 is @xmath756invariant if and only if the exponents @xmath757 satisfy the system of equations @xmath758 before stating a corollary of the lemma we need a definition .
let @xmath753 be the subset of @xmath759 satisfying the equations in the above lemma .
clearly @xmath753 is a semigroup under addition .
let @xmath760 $ ] denote the associated semigroup algebra .
the affine coordinate ring of @xmath105 is isomorphic to the semigroup ring @xmath760 $ ] .
now we will relate the semigroup @xmath753 and the monomials of the lemma to graphs on vertices of the decomposed tree @xmath81 .
monomials in the plcker coordinates @xmath761 for @xmath762 fixed correspond to graphs on the vertices @xmath763 of the tripod @xmath762 as follows .
we associate to the exponent @xmath764 of @xmath761 the graph consisting of @xmath764 arcs joining the vertex @xmath47 of @xmath762 to the vertex @xmath274 .
each triple @xmath765 for @xmath762 fixed determines a graph on the leaves of the tripod @xmath762 .
thus each element @xmath766 corresponds to a collection of @xmath79 graphs , each on 3 vertices , one for each tripod in @xmath81 .
also each such element @xmath767 corresponds to a product of monomials attached to tripods .
this leads to a bijective correspondence between monomials and graphs .
the condition @xmath768 specifies that the number of arcs associated to the identified edges @xmath769 of the tripod @xmath720 , and @xmath770 of the tripod @xmath771 must agree .
later , this will allow us to glue these `` local '' graphs to obtain a `` global '' graph on @xmath3 vertices .
in the next section we will associate a weighting of @xmath2 to the @xmath772 .
we define @xmath111 , a projectivization of @xmath105 , which will be isomorphic to @xmath773 .
the coordinate ring of the affine toric variety @xmath105 is the semigroup algebra @xmath760 $ ] .
hence to define the @xmath774 action required for projectivization , it suffices to define a grading on @xmath753 .
first , for @xmath775 we define the associated weight of the edge @xmath776 of the tripod @xmath762 in @xmath81 .
let @xmath777 .
we define the degree of an element @xmath766 by @xmath778 where the sum is over all @xmath779 which represent leaf edges of @xmath2 .
we will see below that the under the isomorphism from the semigroup @xmath780 to the semigroups of admissible weightings @xmath781 the number @xmath782 will be the weight assigned to the edge @xmath776 of the tree @xmath2 .
thus the degree of @xmath767 is then half the sum of the weights of the leaf edges .
give @xmath753 the grading defined above , then we define @xmath783)\ ] ] note that as defined @xmath111 is a git @xmath774-quotient of @xmath105 .
in this section we will prove the first statement of theorem [ firstauxiliarytheorem ] namely that the toric variety @xmath111 constructed in the previous section is the isomorphic to @xmath785 by proving proposition [ threeisomorphicsemigroups ] below .
recall that in section 4 it was shown that the two semigroups @xmath786 and @xmath619 are isomorphic .
@xmath786 is the semigroup of kempe graphs on @xmath787 vertices , with the product @xmath788 .
@xmath619 is the semigroup of admissible weightings on @xmath2 under addition .
recall that an admissible weighting is an integer weight satisfying the triangle inequalities about each internal vertex of @xmath2 , along with the condition that the sum of the weights about each internal node be even .
it was also shown in section 4 that the semigroup algebra of @xmath789 is isomorphic to the coordinate ring of @xmath7 .
[ twoisomorphicsemigroups ] the graded semigroups @xmath619 and @xmath753 are isomorphic . in all that follows we deal with tripods ,
so we give the unique leaf edge incident to the leaf labelled @xmath47 the label @xmath47 , and vice - versa . with this in mind
, @xmath790 can be thought of as giving the number of arcs in a graph on the leaves of a tripod @xmath791 between the @xmath47-th and @xmath274-th leaves .
the number @xmath792 is a natural number assigned to the @xmath47-th edge of a tripod @xmath791 obtained by counting the number of arcs which have unique path in @xmath791 containing the @xmath47-th edge .
the elements of both @xmath753 and @xmath619 both associate triples of integers to each tripod of @xmath793 with certain `` gluing conditions '' .
let us consider a single tripod @xmath794 .
a triple of numbers @xmath795 is an admissible weighting of @xmath791 if and only if there are integers @xmath790 , @xmath796 such that @xmath797 . to see this simply note that the equations @xmath798 have natural solutions if and only if @xmath799 is admissible .
therefore we may define a map from @xmath619 to @xmath753 by solving for @xmath764 , with an obvious inverse given by solving for the weighting on the edge @xmath779 . to see that these maps are well defined , note that the gluing condition @xmath800 is exactly equivalent to the weights on @xmath801 and @xmath802 being equal when these tuples represent the same diagonal in @xmath2 . since both semigroup operations are defined by adding integers , and since both @xmath803 and its inverse are linear functions over each trinode , these maps are semigroup isomorphisms .
finally , note that the grading on @xmath753 was chosen specifically to match the grading on @xmath786 , we leave direct verification of this to the reader .
the semigroup @xmath786 is generated as a graded semigroup by the elements corresponding to graphs with exactly one edge .
the element corresponding to the graph with one edge , between the @xmath47 and @xmath274 vertices corresponds to the plcker coordinate @xmath276 . recall that the unique path in the tree @xmath2 joining the leafs @xmath47 and @xmath274 has been denoted @xmath804 .
the path @xmath804 gives rise to a sequence of edges of the forest @xmath81 ( where we pass from one tripod to the next by passing from an edge @xmath769 to an equivalent edge @xmath770 .
we let @xmath805 be the corresponding product of plcker coordinates @xmath806 .
thus @xmath807 where the @xmath808 and @xmath809 are the edges in the unique path defined by the path @xmath506 corresponding to @xmath276 in @xmath2 . note that @xmath805 is a @xmath756-invariant monomial in the homogeneous coordinate ring of @xmath97 - moreover it is a generator of the ring of invariants .
the isomorphism of toric rings @xmath803 from the homogeneous coordinate ring of @xmath785 to @xmath810 is then given on generators by @xmath811 it is important to see that the degree assigned to @xmath812 by the isomorphism @xmath803 ( namely _ one _ ) is different from that given by counting the @xmath813 in the product formula for @xmath814 .
the latter count is in fact the speyer - sturmfels weight @xmath815 of the plcker coordinate @xmath276 .
we verify the second statement of theorem [ firstauxiliarytheorem ] .
we have seen that the action of @xmath816 that gives the grading scales each @xmath814 by @xmath249 .
clearly this action is induced by the action on the matrix @xmath709 that scales each row corresponding to a leaf edge by @xmath817 , in other words by the action of the element @xmath818 as claimed in the second statement of theorem [ firstauxiliarytheorem ] .
we conclude by explaining why the actions of @xmath819 and @xmath820 coincide .
it suffices to prove that @xmath821 acts trivially .
but since the operation that scales all rows of @xmath709 by @xmath822 is induced by an element of @xmath718 it acts trivially on @xmath823 and hence on @xmath105 . but also the operation of scaling all the rows of @xmath709 that belong to nonleaf edges of @xmath81 is induced by an element of @xmath824 hence this operation too is trivial on @xmath105 . but
@xmath821 is the composition of the two operations just proved to be trivial .
we now give the proof of theorem [ firstmaintheorem ] .
we first note that we have identified an ordered subset of edges of @xmath81 with the leaf edges of @xmath2 ( equivalently the edges of @xmath38 ) .
this identification gives an isomorphism @xmath826 .
we will prove the following .
[ noncompact ] the spaces @xmath827 and @xmath828 are equivariantly homeomorphic with respect to @xmath829 .
we will use the symbol @xmath830 to denote the above equivariant homeomorphism and @xmath831 to denote its inverse . by the results of subsection [ firsttorusquotient ] it suffices to construct a @xmath829-equivariant homeomorphism ( and its inverse ) from @xmath827 to the above symplectic quotient which we will continue to denote @xmath828 .
we first construct the @xmath829-equivariant map @xmath832 .
there is a simple idea behind this map .
we have an inclusion of the leaf edges of @xmath2 into the edges of @xmath81 .
now an element of @xmath833 is a map @xmath703 from the edges of @xmath81 into @xmath95 .
the map @xmath831 is induced by the map @xmath834 that restricts @xmath703 to the leaf edges of @xmath2
. however we need to verify that the image of an element of momentum level zero for @xmath835 has @xmath21momentum level zero , and that the induced map of zero momentum levels descends to the required quotients . for the rest of this discussion ,
let @xmath836 be an element of @xmath837 .
each @xmath838 is an imploded spin framing of the edges of the triangle @xmath720 .
this means that @xmath839\ ] ] such that @xmath840 we will henceforth denote @xmath841 , thus the symbol @xmath719 stands for a triangle together with an imploded spin - frame on its edges .
let @xmath843 .
note that if @xmath844 fixes @xmath213 by conjugation , then @xmath461 is diagonal and @xmath845
. furthermore , @xmath846 . let @xmath847,[t_2],\ldots,[t_{n-2}]$ ] be an element of @xmath828 .
[ rhoflip ] suppose that the @xmath487-th edge of @xmath720 is identified with the @xmath848-th edge of @xmath771 in @xmath2 and that the edge @xmath849 comes after the edge @xmath769 in the above ordering of edges of @xmath81 .
then using the left actions of @xmath21 we may arrange that @xmath850 suppose we have @xmath851 = [ ( a_1 , \ell_1 \varpi_1),(a_2 , \ell_2\varpi_1 ) , ( h , d \varpi_1)],\ ] ] if the diagonal defined by @xmath852 and @xmath853 has length @xmath739 there is nothing to prove since any two frames are equivalent .
suppose then that this diagonal is not @xmath739 . then by applying @xmath854 to @xmath855
, we get @xmath856,\ ] ] which is equivalent to @xmath855 .
hence the consecutive diagonal frames ( the third frame of @xmath857 and the first frame of @xmath858 ) are now related by right multiplication by @xmath842 .
we shall henceforth choose representatives in @xmath833 for elements of @xmath828 so that nonzero frames associated to equivalent edges satisfy this `` @xmath842 flip condition '' .
if the frame associated to one edge of a pair of equivalent edges is zero then we require that the frame associated to the other edge of the pair is also zero .
we say such elements of @xmath105 are _
normalized_. note that the definition of normalized depends on the ordering of the triangles @xmath719 .
it is important to note that if an element @xmath859,[t_2],\cdots,[t_{n-2}])$ ] is normalized then so is @xmath860 for any @xmath861 .
this is because @xmath862 .
we leave the details to the reader .
thus we may speak of a @xmath738-equivalence class as being @xmath863
. we will define @xmath864 on such normalized elements .
[ norm ] for any @xmath865 there exist @xmath866 and a normalized @xmath867 such that @xmath868 let @xmath869 be connected , and let @xmath870 be a tripod which shares exactly one edge , @xmath631 with @xmath871 .
let @xmath779 and @xmath872 be the edges of @xmath81 which map to @xmath631 .
suppose @xmath873 is normalized at all diagonals corresponding to internal edges of @xmath871 .
let @xmath866 be the element which is @xmath874 on the @xmath875-th factor , and the identity elsewhere
. then @xmath876 is normalized at all diagonals corresponding to internal edges of @xmath877 . by lemma [ trivalentinduction ]
we can always find a sequence of connected trees @xmath878 with @xmath879 a tripod such that each internal edge of @xmath2 appears as an edge shared by the images of @xmath880 and @xmath625 in @xmath2 for some @xmath47 .
the lemma now follows by induction .
given a normalized element @xmath881 define @xmath882 to be the element in @xmath245 given by projecting on the components @xmath883 such that @xmath884 maps to a leaf edge in @xmath2 under @xmath885 .
if @xmath886 and is normalized then @xmath887 , that is the polygon in @xmath9 associated to @xmath882 closes up .
first observe that @xmath888 , now because each triangle closes up the sum of @xmath889 over all edges of @xmath81 is zero .
but by the observation just above the sum over pairs of equivalent edges of @xmath81 cancel .
hence the sum over leaf edges of @xmath2 is zero as required .
[ wdphi ] @xmath890 induces a well - defined map on the quotient @xmath891 and a well - defined map @xmath892 to define @xmath893 we need only show that for any pair of normalized @xmath894 and @xmath895 in @xmath896 and @xmath897 such that @xmath898 we have @xmath899 in @xmath59 .
the assumption @xmath898 is equivalent to @xmath900 for each @xmath901 and @xmath902 .
hence @xmath903 , and in particular a ( pair of ) diagonals of @xmath894 vanishes if and only if the same diagonals vanish for @xmath895 that is @xmath904 hence the forests @xmath905 and @xmath906 coincide
. we must show that the images of @xmath907 and @xmath908 in @xmath59 coincide , which amounts to proving that the @xmath722 is constant on any connected component in the forest @xmath905 ( recall that we think of @xmath722 as a function from the trivalent vertices of @xmath2 to @xmath21 ) .
clearly it suffices to prove that if @xmath711 and @xmath909 are equivalent then @xmath910 suppose the imploded spin framings on the two edges are respectively @xmath911 and @xmath912 .
we have equations @xmath913 @xmath914 and because we have assumed both @xmath894 and @xmath895 are normalized we have @xmath915 @xmath916 we can rearrange these equations to get @xmath917 we cancel @xmath918 to obtain @xmath919 .
but from above @xmath920 this proves the first statement of the lemma .
it is clear that @xmath893 is constant on @xmath738 orbits , so it descends to give @xmath864 .
this shows that @xmath864 is well - defined and continuous because of the continuity of @xmath921 @xmath922 is proper ( hence closed ) .
suppose @xmath923 is a compact subset of @xmath29 .
then the lengths of the columns of any matrix @xmath254 ( so the edge - lengths of the corresponding @xmath3-gon ) representing an element of @xmath923 are uniformly bounded by a constant @xmath639 .
we may reinterpret the above uniform bound as a bound on all edge lengths of all triangles in the given triangulation of @xmath38 that are also edges of @xmath38 .
but we may assume all our matrices @xmath254 are in the zero level sets of the momentum maps for @xmath718 and @xmath93 .
it follows by lemma [ trivalentinduction ] and the triangle inequalities that the lengths of all columns of any matrix @xmath709 representing any element in @xmath105 in the inverse image of @xmath923 are bounded by @xmath924 where @xmath925 is a positive integer depending only on the tree @xmath2 .
then @xmath926 is contained in the image of a subset of @xmath833 homeomorphic to the product of @xmath82 copies of the ball of radius @xmath924 in @xmath95 we now construct the map @xmath927 inverse to @xmath864 .
we first define @xmath928 on @xmath245 then verify that the resulting map descends to @xmath59 . we will need to be able to add a diagonal frame on a triangle with two already framed edges .
[ framing ] suppose that two sides of an imploded spin - framed triangle @xmath929 , @xmath930 are given .
then we can find @xmath931 and @xmath932 so that @xmath929 , @xmath930 @xmath933 is an imploded spin - framed triangle .
precisely we may solve @xmath934 moreover @xmath935 is uniquely determined by @xmath936 and @xmath937 and if @xmath938 any two choices of @xmath939 for given @xmath940 and @xmath941 are related by right multiplication of @xmath20 .
if both @xmath936 and @xmath937 are zero then we take @xmath935 to be zero and @xmath942 .
suppose then that exactly one of @xmath936 and @xmath937 is nonzero . without loss of generality ,
assume @xmath936 is nonzero . put @xmath943 and choose @xmath944 .
hence we have @xmath945 note that the map @xmath946 given by @xmath947 is the hopf fibration and consequently we may solve the equation @xmath948 locally in a neighborhood of @xmath940 , in particular we can solve this equation in a neighborhood of @xmath949 of @xmath950 .
the number @xmath951 is uniquely determined but @xmath939 is defined only up to right multiplication by an element of @xmath952 .
now assume that both @xmath936 and @xmath937 are nonzero .
first assume that the sum @xmath953 is zero , then we are forced to take @xmath954 and we may choose any framing data we wish , so we choose @xmath942 .
assume then that @xmath955 is nonzero .
we choose @xmath931 and @xmath935 so that @xmath956 again , @xmath938 is uniquely determined , and @xmath939 is defined up to right multiplication by an element of @xmath952 .
we define the required framed triangle @xmath191 by @xmath957 we make use of lemma [ framing ] to extend any framing of the edges of @xmath38 to an element of @xmath833 .
the resulting object will be a normalized element @xmath958 .
[ extend ] suppose we are given an imploded spin framing @xmath247 of the edges of the model convex @xmath3-gon @xmath38 which is of momentum level zero for the action of @xmath21 .
then we may extend the framing by choosing imploded spin - frames on the diagonals and an enumeration of the triangles of @xmath38 so that the resulting element @xmath959 is 1 .
normalized 2 . of momentum level zero for @xmath718 3 . of momentum level zero for @xmath960
moreover any two such extensions are equivalent under the action of the torus @xmath93 .
we prove the existence of the framing by induction on @xmath3 .
we take as base case @xmath961 , here there is nothing to prove .
let @xmath38 be a model convex @xmath3-gon with an imploded spin framing on its edges .
define @xmath962 .
choose a triangle @xmath191 in the triangulation of @xmath38 which shares two edges with @xmath38 .
hence two sides of @xmath191 are framed .
let @xmath523 be the remaining side ( a diagonal of @xmath38 ) .
suppose @xmath523 has length @xmath249 .
define @xmath963 .
apply lemma [ framing ] to frame the third side @xmath523 such that the resulting framing is of momentum zero level for @xmath21 .
split @xmath191 off from @xmath38 to obtain an @xmath697gon @xmath964 .
suppose the framing on @xmath523 is @xmath965 $ ]
. give the edge @xmath966 of @xmath964 which is not yet framed the framing @xmath967 $ ] .
we obtain the existence part of the lemma by induction .
now we prove uniqueness .
suppose that @xmath968 is another extension of @xmath247 , and that @xmath968 assigns the frame @xmath969 $ ] to @xmath523 with @xmath970 .
then again by lemma [ framing ] either @xmath971 or there exists an element @xmath225 such that @xmath972 . in case
@xmath340 the frame on @xmath966 is necessarily @xmath973 = [ g \varrho t^{-1 } , \lambda \varpi_1]$ ] .
thus the new frames on the pair of equivalent edges of @xmath81 are @xmath974 $ ] followed by @xmath975 $ ] .
note that the framing of @xmath964 obtained by restriction of @xmath968 satisfies the three properties in the statement above , and also it is an extension of the framing of its boundary induced by @xmath968 .
hence for the case @xmath340 the induction step of the uniqueness part of the lemma is completed . in case @xmath971
then we may represent both frames by @xmath976 $ ] .
but since @xmath971 we have ( by definition of implosion ) for any @xmath977 @xmath978 = [ t , \lambda \varpi_1 ] = [ t^{-1 } , \lambda \varpi_1].\ ] ] this completes the induction step in the uniqueness part of the lemma . by the above lemma we obtain a well - defined map @xmath979 we prove this map descends to @xmath59 .
we let @xmath980 denote the map obtained from @xmath928 by postcomposing it with projection to @xmath105 .
let us note the following obvious refinement of lemma [ extend ] .
suppose @xmath981 is a subpolygon of @xmath38 which meets @xmath38 along diagonals of length zero in the realization of @xmath38 given by the framing of the edges .
then given any extension of the framing of @xmath38 to all diagonals of @xmath981 we may find an extension of the framing of @xmath38 to all diagonals of @xmath38 agreeing with the given one on @xmath981 .
the map @xmath980 descends to @xmath59 .
we first verify that @xmath980 descends to @xmath29 .
in fact we will show it is equivariant under @xmath21 where the action of @xmath21 on @xmath833 is by the diagonal in @xmath718 .
let @xmath982 .
let @xmath247 be a framing of the edges of @xmath38 and @xmath703 be the extension of lemma [ extend ] .
apply lemma [ extend ] to extend @xmath983 to an element @xmath968 of @xmath700 so that this extension also satisfies the three properties in the statement of lemma [ extend ] .
we wish to prove that the resulting element of @xmath833 is in the @xmath93-orbit of @xmath984 .
but since any two extensions satisfying the three properties of lemma [ extend ] are equivalent under @xmath93 it suffices to prove that the image of _ any one _ of the above extensions is equal to @xmath985 .
but there is an extension that is obviously equivalent to @xmath985 namely @xmath985 itself .
it remains to check that @xmath980 descends to @xmath59 .
let @xmath247 and @xmath703 be as above .
let @xmath986 be another framing of the edges of @xmath38 in the same @xmath2-congruence class as @xmath247 .
we have to prove that @xmath987 by transitivity of the @xmath2-congruence relation it suffices to prove this for @xmath986 .
let @xmath988 be a component of @xmath905 .
we may assume that @xmath988 contains leaf edges .
apply an element @xmath982 to the frames of the leaf edges of @xmath989 to obtain a new framing @xmath986 of the edges of @xmath38 .
we now prove @xmath990 for this choice of @xmath986 .
apply lemma [ extend ] to @xmath986 to obtain an element @xmath968 of @xmath833 such that the three properties in lemma [ extend ] are satisfied .
we must prove that @xmath703 and @xmath968 have the same image in @xmath105 .
as before in the paragraph above it suffices to find one extension @xmath968 of @xmath986 which has the same image as @xmath703 .
we now construct such an extension @xmath991 the subtree @xmath988 corresponds to a subpolygon @xmath992 of @xmath38 such that the edges of @xmath992 that correspond to diagonals of @xmath38 have zero length .
define @xmath991 by defining the frame on an edge @xmath523 of @xmath81 that is in @xmath988 to be @xmath347 applied to the value of @xmath703 on @xmath523 and if @xmath523 is not in @xmath988 then define the value of @xmath991 on that edge to be the value of @xmath703 on that edge .
we leave to the reader that @xmath991 satisfies the three properties of lemma [ extend ] .
now we construct an element @xmath993 such that @xmath994 recall that an element of @xmath718 may be regarded as a function from the trivalent nodes of @xmath2 to @xmath21 .
we define @xmath995 to be the function that is @xmath996 if the trivalent node is not in @xmath988 , and is @xmath347 if the trinode is in @xmath988 .
since an edge @xmath523 of @xmath2 is in @xmath988 if and only if the trivalent node incident to @xmath523 we do indeed have @xmath994 because @xmath113 is the inverse of a closed map we have @xmath113 is continuous .
finally we have @xmath831 and @xmath830 are inverses to one another it is obvious that @xmath997 .
this is because neither of the two map changes the framing of the edges of @xmath38 .
now let @xmath703 be a normalized element of @xmath835-momentum level zero in @xmath833 .
let @xmath50 be the restriction of @xmath703 to the edges of @xmath38 .
thus @xmath703 is an extension of @xmath50 satisfying the three properties of lemma [ extend ] . but
@xmath998 also has restriction to @xmath38 given by @xmath50 and also satisfies the three properties of lemma [ extend ] .
hence @xmath703 and @xmath998 are in the same @xmath93-orbit and hence their images in @xmath105 coincide .
it is clear from the definition of @xmath831 that this map intertwines the actions of @xmath30 and @xmath999 .
this proves theorem [ noncompact ] . in the previous subsection we constructed a homeomorphism @xmath1000 with inverse @xmath864 .
the torus @xmath86 acts on @xmath105 such that the hamiltonian for the circle factor corresponding to the edge @xmath1001 of @xmath81 is given by @xmath1002 . in
spin - framed coordinates this is @xmath1003 .
we will compute the pullback of these functions to @xmath59 via @xmath113 .
let @xmath1004 by the composition of the projection on the @xmath779-th factor of @xmath833 with the function @xmath355 from subsection [ frommatricestopolygons ] .
let @xmath1005 be the composition of the projection on the @xmath523-th factor of @xmath245 with the function @xmath355 from subsection [ frommatricestopolygons ] .
for a distinguished edge @xmath779 of @xmath81 let @xmath1006 be associated edge of the model convex planar @xmath3gon .
then we have the following identity @xmath1007 in what follows we will need to compute the pull - back of @xmath1008 to @xmath59 for those distinguished edges @xmath1001 . let @xmath347 be the map from subsection [ frommatricestopolygons ] , then we have @xmath1009 for any normalized @xmath1010 which maps to @xmath247 .
[ pullback ] for @xmath1011 we have @xmath1012 where @xmath1013 maps to @xmath254 .
the proposition follows from @xmath1014 note that @xmath1015 is the length of the @xmath1016 edge of @xmath383 .
we also need to compute the @xmath113 pullbacks of the hamiltonian functions @xmath1017 for @xmath1018 an internal edge of @xmath2 , this will be important when we wish to indentify the hamiltonian flows of these pullbacks with the bending flows .
note that @xmath1018 corresponds to a diagonal @xmath1019 in a model @xmath3-gon @xmath38 .
let @xmath869 be a connected subtree such that every leaf of @xmath871 except one , say @xmath1020 , is a leaf of @xmath2 .
then for any element @xmath1021 we can compute @xmath1022 for @xmath703 a normalized element of @xmath1023 which maps to @xmath247 . first note
that we compute @xmath1024 by taking half the length of the @xmath1020 edge for any normalized element @xmath1025 which maps to @xmath247 , this is equal to @xmath1026 . by the normalized condition and
the closing condition imposed on @xmath703 , the leaf edge of @xmath871 form a closed polygon , which implies that @xmath1027 .
the previous proposition allows us to conclude the following theorem .
[ diagonalpullback ] let @xmath1028 be the associated diagonal in @xmath383 .
@xmath1029 we have the following equation @xmath1030 the right hand side of this equation is equal to @xmath1031 this last expression is in turn equal to @xmath1032 .
this , along with proposition [ pullback ] , proves theorem [ stratification ] . in this subsection
we will descend the homeomorphisms of affine varieties @xmath864 and @xmath113 to their projective quotients completing the proof of theorem [ firstauxiliarytheorem ] and hence the proof of theorem [ firstmaintheorem ] . recall that we defined elements @xmath1033 and @xmath1034 for @xmath1035 .
we note @xmath1036 since @xmath830 is equivariant we have @xmath1037 the map @xmath830 induces a homeomorphism from @xmath58 to the symplectic quotient of @xmath105 by the grading circle action .
it suffices to prove that the pull - back by @xmath113 of the grading circle action on @xmath105 is the grading circle action on @xmath59 .
this follows by replacing @xmath249 by @xmath1038 in equation ( [ degreeequivariance ] ) .
we complete the proof of theorems [ firstauxiliarytheorem ] and [ firstmaintheorem ] by proving the symplectic quotient of @xmath1039 by the grading circle action coincides with the git quotient of @xmath105 by the grading @xmath204-action linearized by acting by the identity on the fiber of the trivial complex line bundle over @xmath105 .
we are forced to give an indirect argument because @xmath1039 is not smooth and there do not seem to be theorems asserting the isomorphism of symplectic and git quotients of nonsmooth spaces .
we first note that the above git quotient is canonically isomorphic to the git quotient of @xmath97 by the product group @xmath1040 where the first factor acts by the grading action linearized by the identity action on the fiber @xmath485 of the trivial complex line bundle over @xmath97 .
this follows because we can take the ring of invariants for a product group acting on a ring @xmath285 by first taking the invariants of one factor and then taking the invariants of the resulting ring by the second factor .
the corresponding quotient by stages for symplectic quotients is also true but slightly harder .
it is proved in theorem 4.1 of @xcite .
thus it remains to prove that the symplectic quotient of @xmath97 by the product of the grading @xmath94-action and @xmath93 is isomorphic to the git quotient of @xmath97 by the product of the grading @xmath204 action and the complexified torus @xmath1041 where the first factor acts by the grading action and the product is linearized by acting by the identity map applied to the projection on the first factor .
this follows immediately from theorem [ twistandshift ] once we establish that the momentum map for the product of the grading @xmath94-action and @xmath93 is proper .
we now prove that the momentum map for the product of the grading circle action and @xmath93 is proper .
in fact we prove a different result that turns out to be equivalent to the one we need here because we will need this different result below .
[ propernesstheorem ] the momentum mapping @xmath1042 for the action of @xmath1043 is proper .
earlier we identified @xmath1044 with the space of framed triangles @xmath732 .
let @xmath1045 , \ldots)$ ] be an element of @xmath1046 . under this identification
we have that @xmath1047 , \ldots)$ ] is the vector of elements @xmath1048 where @xmath711 is a distinguished edge of @xmath81 .
similarly @xmath1049 , \ldots)$ ] is the vector of elements @xmath1050 where @xmath711 and @xmath1051 represent the same internal edge of @xmath2 . in order to show that @xmath1052 is a proper map
, we must show that if all @xmath1048 are bounded for @xmath711 distinguished , and all differences @xmath1050 are bounded for all @xmath711 and @xmath1051 which represent the same internal edge of @xmath2 , then all @xmath1048 are bounded .
let @xmath1053 , [ g_2 , \lambda_2\varpi_1 ] , [ g_3 , \lambda_3\varpi_1])$ ] be an element of @xmath732 . since @xmath1054 the @xmath410 are side - lengths of a closed triangle , hence if two of the three are bounded , so is the third .
furthermore , if @xmath711 and @xmath1051 represent the same internal edge of @xmath2 , and @xmath1048 and the difference @xmath1050 are bounded , then @xmath1055 is bounded .
the proposition now follows from trivalent induction , see lemma [ trivalentinduction ] .
the momentum mapping @xmath1042 for the action of @xmath1056 is proper . here
@xmath94 acts by the grading circle action .
the hamiltonian for the grading circle action is the half the sum of the hamiltonians for the circle @xmath116 factors of @xmath116 .
but note that each of the summands is bounded if and only if the sum is ( since they have the same sign )
. there is one more technical point .
we claim that the git quotient coincides with the symplectic quotient at momentum level @xmath1057 .
we will prove first that the level for the @xmath1058 is zero .
but this follows because the torus @xmath1058 acts linearly on @xmath1059 and trivially on the fiber whereas the grading circle action is twisted by the identity action on the fiber .
in this section we will prove theorem [ secondmaintheorem ] by proving theorem [ secondauxiliarytheorem ] . in this subsection
we prove the first statement of theorem [ secondauxiliarytheorem ] . by definition
the toric variety @xmath1060 is obtained as the git quotient of @xmath693 by @xmath1061 using the linearization given by the character @xmath1062 here @xmath1063 denotes the element of @xmath1061 corresponding to @xmath1064 .
we recall that the graded ring of @xmath64 is the semigroup ring @xmath1065 $ ] where @xmath1066 is the graded subsemigroup of @xmath786 defined by taking graphs with valence @xmath1067 for positive integers @xmath487 .
define the subsemigroup @xmath1068 of @xmath753 to be the inverse image of @xmath618 under the isomorphism from @xmath753 to @xmath619 under the isomorphism of proposition [ twoisomorphicsemigroups ] .
the @xmath756-invariant monomial @xmath1069 is @xmath1061invariant for the twist @xmath1070 if and only if the exponents @xmath1071 satisfy the system of equations @xmath1072 by proposition [ twoisomorphicsemigroups ] we may rewrite the above equation as a condition on the leaf weights @xmath1073 the lemma follows from the formula for @xmath1063 acting on @xmath347 , namely @xmath1074 } f(z).\ ] ] here @xmath1075 is the coordinate of @xmath1061 corresponding to the leaf edge @xmath754 and @xmath1076 .
the set of @xmath1061-invariant monomials in @xmath761 _ for the twist by @xmath1070 _ is the subset of invariants _ of degree @xmath1077 _ ( the @xmath1077-th graded piece of the associate semigroup of lattice points ) . as an immediate consequence we have [ rthsemigroup ]
the toric variety @xmath124 is the projective toric variety associated to the graded semigroup @xmath1068 . now by proposition [ gradedsubsemigroupiso ]
we have an isomorphism of graded semigroups @xmath1078 .
since by definition we have an isomorphism of graded semigroups @xmath1079 we obtain the required isomorphism of graded semigroups @xmath1080 and we have proved the desired isomorphism of projective toric varieties . in this section
we will prove the second statement of theorem [ secondauxiliarytheorem ] .
@xmath65 is homeomorphic to the symplectic quotient of @xmath828 by @xmath116 at level @xmath6
. recall that in proposition [ pullback ] we proved the following formula .
let @xmath1081 then @xmath1082 where @xmath1016 is the leaf edge of @xmath2 corresponding to the leaf edge @xmath1001 of the decomposed tree @xmath81 .
thus @xmath113 induces a homeomorphism between the @xmath2-congruence classes of imploded spin - framed @xmath3-gons with side - lengths @xmath6 and the @xmath6-th level set for the momentum map of @xmath116 .
but also @xmath113 is @xmath829-equivariant , hence the above bijection descends to a homeomorphism from the @xmath30-quotient of the imploded spin - framed @xmath3-gons to the @xmath116 symplectic quotient at level @xmath6 .
it remains to prove that the symplectic quotient of @xmath105 at level @xmath6 coincides with the git quotient for the action on the trivial bundle using the twist by @xmath1083 .
let @xmath569 be the trivial complex line bundle over @xmath1059 .
then @xmath569 descends to the trivial complex line bundle @xmath1084 over @xmath828 .
the torus @xmath1061 acts on @xmath1084 by twisting by the character @xmath1083 .
we need to prove that the git quotient of @xmath828 by @xmath1061 with linearization the above action on @xmath1084 is homeomorphic to the symplectic quotient by the maximal compact subgroup of @xmath1061 at level @xmath6 .
the argument is almost the same as that of subsection [ quotientscoincide ] .
in particular we use reduction in stages to reduce to the corresponding problem for the quotients of @xmath1059 by the product torus @xmath1085 . here
we twist the action by the character that is @xmath1083 on @xmath1061 and trivial on @xmath824 .
since the momentum map for the action by @xmath1085 is proper by proposition [ propernesstheorem ] the equality of quotients follows by theorem [ twistandshift ] .
the momentum level for the product group is then @xmath1086 .
now we will relate the action of @xmath1087 on @xmath105 and @xmath1088 to bending flows .
in this subsection we will define a complement @xmath1089 to @xmath93 in @xmath90
. it will be more convenient to work with this complement rather than the quotient @xmath1090 .
we recall that an element of @xmath86 corresponds to a function @xmath347 from the edges of @xmath81 to the circle .
the subgroup @xmath90 corresponds to the functions with value the identity on the distinguished edges of @xmath81 .
the nondistinguished edges of @xmath1091 occur in equivalent pairs .
the subtorus @xmath93 of @xmath90 consists of those @xmath347 whose values on equivalent pairs of nondistinguished edges are inverse to each other . in order to construct the complement
@xmath1089 we need to choose an edge from each distinguished pair . to do this in a systematic way we use the ordering of the trinodes of @xmath81 induced by the trivalent induction construction , see lemma [ trivalentinduction ] .
suppose the edges @xmath779 and @xmath872 are equivalent .
the two edges occur in different trinodes .
we relabel the edges by @xmath1092 and @xmath1093 by labeling the edge that comes in the first trinode by ( the superscript ) plus and the edge that comes in a later trinode by minus .
thus every nondistinguished edge is either a plus edge or a minus edge .
we now define the subtorus @xmath1089 as the subtorus of @xmath90 consisting of those functions that take value the identity on all nondistinguished minus edges .
it is clear that @xmath1089 is the required complement .
it is also clear that @xmath1089 is a complement to @xmath1043 in @xmath86 .
we point out here that the choice of where to place the identity in the definition of this complement is irrelevant with respect to the hamiltonian functions - see below .
we now prove the following theorem .
[ noncompact2 ] 1 .
the homeomorphism @xmath830 intertwines the bending flows on @xmath59 with the action of @xmath1094 on @xmath105 .
2 . the homeomorphism @xmath1095 intertwines the bending flows on @xmath1096 with the action of @xmath1094 on @xmath124 .
let @xmath1097 and @xmath1098 .
let @xmath631 be a diagonal of the triangulated @xmath3-gon @xmath38 and suppose the triangles @xmath719 and @xmath1099 share the diagonal @xmath631 .
suppose @xmath631 divides @xmath38 into two polygons @xmath992 and @xmath1100 with @xmath1101 and @xmath1102 .
let @xmath762 and @xmath875 be the trinodes associated to @xmath191 and @xmath1103 respectively , and let @xmath1104 $ ] and @xmath1105 $ ] be spin - framed representatives of the edges of @xmath703 corresponding to the diagonal @xmath631 .
suppose @xmath1106 .
since the representatives are normalized we have @xmath1107 let @xmath461 be the @xmath1108-th component of @xmath1109 ( the @xmath1110-th component is @xmath34 ) . under the action of @xmath1109 the edge @xmath1104 $ ] becomes @xmath1111 $ ] and all other components are unchanged .
thus the resulting element of @xmath833 is no longer normalized . to normalize this element
we multiply all the frames on the edges and diagonals of @xmath1100 by @xmath1112 .
note that the image of @xmath1113 . is a rotation about the oriented line in @xmath9 corresponding to @xmath1114 which is the diagonal corresponding to @xmath631 of the euclidean @xmath3-gon @xmath128 underlying the imploded framed @xmath3-gon @xmath703 .
thus the euclidean @xmath3-gon underlying the imploded framed @xmath3-gon @xmath703 is bent along the diagonal @xmath631 .
furthermore the frames of all the edges and diagonals of @xmath1100 are transformed by applying the element of the one - parameter group @xmath1112 in @xmath21 which covers rotation along the diagonal . applying @xmath864 we forget the frames along the diagonals but the frames on the edges transform in the same way as before .
this amounts to bending the framed @xmath3-gon @xmath1115 along the diagonal @xmath631 .
the first statement implies the second statement because the bending flows on @xmath65 are descended from those on @xmath1116 , and the action of @xmath1089 on @xmath124 is descended from the action on @xmath1117 it remains to prove that the action of @xmath1089 is hamiltonian with the given hamiltonians .
we claim that @xmath1094 preserves the orbit - type stratification .
since @xmath1118 is abelian , the isotropy subgroup @xmath1119 of @xmath1120 is equal to the isotropy group @xmath1121 of @xmath1122 , for any @xmath1123 ; in particular this is true for @xmath1124 .
although the space @xmath1088 is possibly singular , we can work in a given symplectic stratum .
thus it makes sense to say that @xmath1094 acts in a hamiltonian fashion ( on each individual stratum ) , and we may identify the torus action on the whole space by the hamiltonians of the @xmath94 factors .
these hamiltonians are smooth in the sense of @xcite , since these functions are obtained by restricting @xmath1125 invariant continuous functions on @xmath1126 to individual strata .
each @xmath1127 is indexed by an edge of @xmath81 .
given an internal edge @xmath1128 , the factor @xmath1129 of @xmath1094 has hamiltonian function @xmath1130 on any given orbit - type stratum .
the @xmath1131 corresponds to the diagonal @xmath1132 of the associated @xmath3-gon in @xmath9 . by lemma [ diagonalpullback ] and lemma [ spinlength ] we have @xmath1133 , and
so @xmath1134 is _ one - half _ that of the hamiltonian for the @xmath1128-th factor @xmath1135 of @xmath1094 . in what follows
we will need the following elementary lemma .
suppose @xmath1136 is a hamiltonian action of the circle on a stratified symplectic space @xmath1137 .
suppose the action is generated by the hamiltonian potential @xmath347 and that the element @xmath1138 acts trivially .
let @xmath1139 be the induced action of the quotient circle @xmath1140 .
then the hamiltonian potential for the @xmath1139 action is @xmath1141 .
we have @xmath1142 the bending flow torus @xmath1143 acting on @xmath1088 , is such that the @xmath1144-th factor has hamiltonian @xmath1145 .
therefore , the @xmath1094 action factors through the action of @xmath73 via a two to one cover on each component of @xmath1094 .
thus we may conclude that @xmath73 and @xmath1146 coincide .
let @xmath1147 be the image of @xmath1094 in @xmath1148 under the surjection @xmath1149 .
the action of @xmath1146 on @xmath1088 coincides with the bending flows .
precisely the action is hamiltonian and the circle factor corresponding to @xmath1144 has hamiltonian potential @xmath1134 .
let @xmath1150 be the quotient map .
let @xmath1128 be any internal edge of @xmath2 .
the action of @xmath1151 on @xmath1088 factors through the action of @xmath1152 , through the map @xmath1153 . consequently by the lemma above the hamiltonian function of @xmath1151 is twice that of the hamiltonian of @xmath1154 .
hence the hamiltonian for @xmath1154 is @xmath1134 , as it is for @xmath1155 .
since the hamiltonian functions of @xmath1156 and @xmath73 coincide , their actions must coincide on @xmath1088 .
the results in this appendix were obtained with the aid of w. goldman . the goal of this appendix is to prove theorem [ twistandshift ] below - we match the level for the symplectic quotient with the twist used to define the linearization in forming the git quotient . in four places in the paper , namely subsections [ firsttorusquotient ] , [ secondtorusquotient ] , [ thirdtorusquotient ] and subsubsection [ fourthtorusquotient ] we applied the results of @xcite to deduce that for a torus acting on affine space linearized by acting on the trivial line bundle by a character @xmath1157
the git quotient is homeomorphic to the symplectic quotient at level the derivative of @xmath1157 at the identity provided the momentum map was proper .
however there is a technical problem about the normalization of the momentum map chosen for the action of the torus ( there is an indeterminancy of an additive constant vector ) the correct normalization of the momentum map must depend on the action of the torus on the total space of the line bundle .
it is given in formula ( 2.3 ) , page 116 of @xcite which we now state for the convenience of the reader .
let @xmath1158 be a hermitian line bundle @xmath569 .
let @xmath149 be a compact group acting on @xmath1159 by automorphisms of the hermitian structure .
let @xmath1160 and @xmath1161 .
in what follows @xmath1162 denotes the vector field on @xmath1159 induced by @xmath1163 , @xmath1164 denotes the horizontal lift of the vector field @xmath1165 induced on @xmath1166 and @xmath1167 denotes the canonical vertical vector field ( induced by the @xmath1168 action ) .
the connection and curvature forms take values in the lie algebra @xmath1169 of @xmath1168 .
we can now state the formula from @xcite : @xmath1170 we can check the conventions involved in equation ( [ normalization ] ) by applying the connection form @xmath1171 to both sides to obtain @xmath1172 applying @xmath631 to each side and cartan s formula we find that @xmath1173 is a hamiltonian potential for @xmath1174 if and only if the symplectic form @xmath229 and the connection form @xmath1171 are related by @xmath1175 suppose now we twist the action of the torus @xmath149 on the total space of the line bundle by scaling each fiber by a fixed character @xmath1157 .
this changes the invariant sections and hence changes the git quotient .
note that the differential of @xmath1157 at the identity of @xmath149 is an element @xmath1176 of @xmath148 .
thus we could change the momentum map @xmath265 by adding @xmath1176 and obtain a new momentum map .
the following lemma is an immediate consequence of equation ( [ normalization ] ) .
we now restrict to the case of a torus @xmath191 acting linearly on a symplectic vector space @xmath1177 .
we assume we have chosen a @xmath191-invariant complex structure @xmath1178 on @xmath1179 so that @xmath229 is of type @xmath1180 for @xmath1178 ( this means @xmath1178 is an isometry of @xmath229 ) and the symmetric form @xmath1181 given by @xmath1182 is positive definite .
we let @xmath260 be the subspace of @xmath1183 of type @xmath1057 vectors , that is @xmath1184 .
we define a positive - definite hermitian form @xmath1185 on @xmath260 by @xmath1186 we will abbreviate @xmath1187 to @xmath1188 in what follows .
we define a symplectic form @xmath254 on @xmath260 by @xmath1189 .
we note that the map @xmath1190 is a symplectomorphism from @xmath1191 to @xmath1177 .
suppose that we have chosen an @xmath1185-orthonormal basis for @xmath260 so we have identified @xmath1192 we let @xmath1193 be the compact torus consisting of the diagonal matrices with unit length elements on the diagonal and @xmath1194 be the complexification of @xmath1195 .
we let @xmath1196 be the dual of the lie algebra of @xmath1195 .
we define @xmath1197 by @xmath1198 then @xmath1199 is a momentum map for the hamiltonian action of @xmath1195 on @xmath260 .
we note that all possible momentum maps are obtained from @xmath1199 by adding a vector @xmath1200 .
thus @xmath1199 is the unique momentum map vanishing at the origin of @xmath260 .
let @xmath1201 be total space of the trivial line bundle @xmath569 over @xmath260 .
we first describe the hermitian holomorphic structure on @xmath569 .
we give @xmath569 a holomorphic structure by requiring that the nowhere vanishing section @xmath1202 of @xmath1159 defined by @xmath1203 is holomorphic .
thus if @xmath1204 is an open subset of @xmath260 and @xmath392 is a local section over @xmath1204 then @xmath392 is holomorphic if and only if the function @xmath347 on @xmath1204 defined by @xmath1205 is holomorphic .
we define a hermitian structure on @xmath569 by defining @xmath1206 hence the section @xmath1207 given by @xmath1208 has unit length at every point .
we note that @xmath1209 in agreement with @xcite , pg .
let @xmath1210 denote the principal circle bundle of unit length vectors in @xmath1159 .
our goal is to describe the lift of the action of @xmath1195 to @xmath1159 so that the normalized momentum map corresponding to this lifted action is @xmath1199 .
we note there is a distinguished lift given by @xmath1211 by definition this lift leaves invariant the holomorphic section @xmath1202 but it also leaves invariant the unit length section @xmath1207 because the function @xmath1212 is invariant under @xmath202 and hence under @xmath1195 . hence @xmath1195 leaves fixed the hermitian structure on @xmath1159 and @xmath1194 leaves the holomorphic section @xmath1202 fixed .
we will call such lifts _ untwisted _ and we will say the linearization consisting of the trivial bundle together with the previous action is untwisted . we will now prove we will compute @xmath1199 in the trivialization of @xmath1210 given by @xmath1207 . we note that since @xmath1207 is invariant under @xmath1195 the untwisted lift of the compact torus above ( relative to the trivialization by @xmath1202 ) remains untwisted in the trivialization by @xmath1207 .
we let @xmath200 denote the coordinate in the fiber circle of @xmath1210 so @xmath1213 . hence if @xmath1214 , then @xmath1215 are coordinates in @xmath1210 . by the formula for @xmath1171
we see that the horizontal lift of @xmath1224 is @xmath1225 and the horizontal lift of @xmath1226 is @xmath1227 . since the operation of taking horizontal lifts is linear over the functions the corollary follows .
now suppose @xmath191 is a compact torus with complexification @xmath1228 and @xmath191 acts on @xmath260 through a representation @xmath1229 .
assume further that we linearize the action of @xmath1228 on @xmath260 by the untwisted linearization .
it is clear from equation ( [ normalization ] ) that we obtain the _ normalized _ momentum map @xmath239 for @xmath191 by restricting the _ normalized _ momentum map for @xmath1195 .
more precisely let @xmath1230 be the induced map on dual spaces .
then we have @xmath1231 thus the normalized momentum map corresponding to the untwisted linearization of the linear action of @xmath191 is homogeneous linear in the squares of the @xmath1232 s .
we now obtain the desired result in this appendix by applying lemma [ twistshift ] and theorem 2.18 ( page 122 ) of @xcite .
[ twistandshift ] let @xmath1228 be a complex torus with maximal compact subtorus @xmath191 and let @xmath1157 be a character of @xmath86 . then the git quotient for a linear action of @xmath1228 on a complex vector space @xmath260 with linearization given by @xmath1233 and the action @xmath1234 is homeomorphic to the symplectic quotient by @xmath191 obtained using the momentum map @xmath1235 if the momentum map for the action of @xmath191 is proper . here
@xmath1236 is the momentum map that vanishes at the origin of @xmath260 .
g. schwarz , _ the topology of algebraic quotients _
, topological methods in algebraic transformation groups ( ed .
hanspeter kraft , ted petrie , gerald w. schwarz ) , birkhauser , progress in mathematics * 80 * ( 1989 ) , 135 - 151 . | speyer and sturmfels @xcite associated grbner toric degenerations @xmath0 of @xmath1 to each trivalent tree @xmath2 with @xmath3 leaves .
these degenerations induce toric degenerations @xmath4 of @xmath5 , the space of @xmath3 ordered , weighted ( by @xmath6 ) points on the projective line .
our goal in this paper is to give a geometric ( euclidean polygon ) description of the toric fibers as stratified symplectic spaces and describe the action of the compact part of the torus as `` bendings of polygons . ''
we prove the conjecture of foth and hu @xcite that the toric fibers are homeomorphic to the spaces defined by kamiyama and yoshida @xcite . |
recent observations indicate that @xmath0 of the energy density of the universe exists in the form of stable non
baryonic , non luminos ( dark ) matter .
supersymmetric ( susy ) models provide a good candidate for the cold dark matter component of the universe . if r
parity is conserved , the lightest neutralino is absolutely stable and can play the role of dark matter .
however the minimal supersymmetric standard model ( mssm ) suffers from the so - called @xmath1problem .
the mssm superpotential contains only one bilinear term @xmath2 .
the parameter @xmath1 is expected to be of the order of @xmath3 or gut scale . at the same time
the correct pattern of electroweak ( ew ) symmetry breaking requires @xmath4 .
the @xmath1problem can be solved within the minimal non minimal supersymmetric standard model ( mnssm ) .
the superpotential of the mnssm can be written as @xmath5 at the ew scale the singlet superfield @xmath6 gets non - zero vacuum expectation value ( vev ) @xmath7 and an effective @xmath1-term ( @xmath8 ) is generated .
thus in the mnssm the @xmath1 problem is solved in a similar way to the next to minimal supersymmetric standard model ( nmssm ) , but without the accompanying problems of singlet tadpoles or domain walls .
the neutralino sector of the mnssm is formed by the superpartners of the neutral gauge bosons ( @xmath9 ) and neutral higgsino fields ( @xmath10,@xmath11,@xmath12 ) . in the field basis @xmath13 the neutralino mass matrix takes a form @xmath14 0 & m_2 & m_z c_w c_{\beta } & -m_z c_w s_{\beta } & 0 \\[2 mm ] -m_z s_w c_{\beta } & m_z c_w c_{\beta } & 0 & -\mu_{eff } & -\frac{\lambda v}{\sqrt{2 } } s_{\beta } \\[2 mm ] m_z s_w s_{\beta } & -m_z c_w s_{\beta } & -\mu_{eff } & 0 & -\frac{\lambda v}{\sqrt{2 } } c_{\beta } \\[2 mm ] 0 & 0 & -\frac{\lambda v}{\sqrt{2 } } s_{\beta } & -\frac{\lambda v}{\sqrt{2 } } c_{\beta } & 0 \end{array } \right)\,,\ ] ] where @xmath15 , @xmath16 , @xmath17 , and @xmath18 . here @xmath19 and @xmath20 , where @xmath21 and @xmath22 are the vevs of @xmath23 and @xmath24 , respectively . the neutralino spectrum in the mnssm may be parametrised in terms of @xmath25 in the minimal sugra inspired susy gut s @xmath26 .
the validity of perturbation theory up to the gut scale implies that @xmath27 .
when @xmath28 is small the non observation of higgs bosons at lep rules out @xmath29 .
lep searches for susy particles also set lower bounds on @xmath30 . in order to find theoretical bounds on the neutralino masses
@xmath31 it is convenient to consider the matrix @xmath32 . in the basis @xmath33 the bottom - right @xmath34 block of
@xmath35 takes the form @xmath36 where @xmath37 , @xmath38 .
since the minimal eigenvalue of any hermitian matrix is less than its smallest diagonal element at least one neutralino in the mnssm is always light , i.e. @xmath39 . therefore in contrast with the mssm the lightest neutralino in the mnssm remains light even when the susy breaking scale tends to infinity .
the mass of the lightest neutralino must be also smaller than the minimal eigenvalue @xmath40 of the submatrix ( [ 4 ] ) , i.e. @xmath41\,.\ ] ] the value of @xmath42 decreases with increasing @xmath43 . taking into account the restrictions on @xmath43 and @xmath44 we find that @xmath45 @xcite .
the obtained theoretical restriction on @xmath46 allows to discriminate the mnssm from other susy models .
the mass of the lightest neutralino can be computed numerically by solving the characteristic equation @xmath47 + m_z^2\biggl(\tilde{m}-\kappa\biggr)\biggl(\kappa^2+\mu_{eff}\sin 2\beta \kappa- \nu^2\biggr)=0\ , , \end{array}\ ] ] where @xmath48 . because in the mnssm @xmath46 is normally much smaller than the masses of the other neutralinos one can ignore @xmath49 , @xmath50 and @xmath51 terms in the characteristic equation so that it reduces to @xmath52 where @xmath53 @xmath54 then the mass of the lightest neutralino can be approximated by @xcite @xmath55 as follows from fig .
12 the approximate solution describes the numerical one with relatively high accuracy . with increasing @xmath56 and @xmath57 the lightest neutralino mass decreases .
if either @xmath43 or @xmath58 then @xmath59 according to eq.([11 ] ) @xmath60 is inversely proportional to @xmath56 .
it decreases when @xmath61 grows and vanishes when @xmath28 tends to zero .
when @xmath62 the mass of the lightest neutralino is proportional to @xmath63 .
when @xmath64 the lightest neutralino is predominantly singlino that makes its direct observation at future colliders rather challenging . | we argue that the allowed range of the mass of the lightest neutralino in the mnssm is limited .
we establish the theoretical upper bound on the lightest neutralino mass and obtain an approximate solution for this mass . |
Subsets and Splits