article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in four dimensions , the einstein tensor is the only conserved symmetric tensor that depends on the metric and its derivatives up to second order . however for spacetimes possessing more than four dimensions , as assumed in both string theory and brane world cosmology ,
this is not the case . in string theory ,
extra dimensions are a theoretical necessity since superstring theory requires a ten - dimensional spacetime to be consistent from the quantum point of view , while in brane world cosmology matter and gauge interactions are localized on a 3-brane , embedded into a higher dimensional spacetime in which gravity propagates throughout the whole of spacetime .
the most natural extension of general relativity in higher dimensional spacetimes with the assumption of einstein that the left hand side of the field equations is the most general symmetric conserved tensor containing no more than two derivatives of the metric is lovelock theory .
lovelock @xcite found the most general symmetric conserved tensor satisfying this property .
the resultant tensor is nonlinear in the riemann tensor and differs from the einstein tensor only if the spacetime has more than 4 dimensions .
since the lovelock tensor contains metric derivatives no higher second order , the quantization of the linearized lovelock theory is ghost - free @xcite .
the concepts of action and energy - momentum play central roles in gravity .
however there is no good local notion of energy for a gravitating system .
quasilocal definitions of energy and conserved quantities for einstein gravity @xcite define a stress energy tensor on the boundary of some region within the spacetime through the use of the well - defined gravitational action of einstein gravity with the surface term of gibbons and hawking @xcite .
our first aim in this paper is to generalize the definition of the quasilocal stress energy tensor for computing the conserved quantities of a solution of third order lovelock gravity with zero curvature boundary .
the first step is to find the surface terms for the action of third order lovelock gravity that make the action well - defined .
these surface terms were introduced by myers in terms of differential forms @xcite .
the explicit form of the surface terms for second order lovelock gravity has been written in ref .
@xcite . here
, we write down the tensorial form of the surface term for the third order lovelock gravity , and then obtain the stress energy tensor via the quasilocal formalism .
of course , as in the case of einstein gravity , the action and conserved quantities diverge when the boundary goes to infinity .
we will also introduce a counterterm to deal with these divergences .
this is quite straightforward for the cases we consider in which the boundary is flat .
this is because all curvature invariants are zero except for a constant , and so the only possible boundary counterterm is one proportional to the volume of the boundary regardless of the number of dimensions .
the coefficient of this volume counterterm is the same for solutions with flat or curved boundary .
the issue of determination of boundary counterterms with their coefficients for higher - order lovelock theories is at this point an open question .
since the lovelock lagrangian appears in the low energy limit of string theory , there has in recent years been a renewed interest in lovelock gravity .
in particular , exact static spherically symmetric black hole solutions of the gauss - bonnet gravity ( quadratic in the riemann tensor ) have been found in ref .
@xcite , and of the maxwell - gauss - bonnet and born - infeld - gauss - bonnet models in ref .
the thermodynamics of the uncharged static spherically black hole solutions has been considered in @xcite , of solutions with nontrivial topology and asymptotically de sitter in @xcite and of charged solutions in @xcite .
very recently nut charged black hole solutions of gauss - bonnet gravity and gauss - bonnet - maxwell gravity were obtained @xcite .
all of these known solutions in gauss - bonnet gravity are static .
not long ago one of us introduced two new classes of rotating solutions of second order lovelock gravity and investigated their thermodynamics @xcite , and made the first attempt for finding exact solutions in third order lovelock gravity with the quartic terms @xcite .
our second aim in this paper is to obtain rotating asymptotically anti de sitter ( ads ) black holes of third order lovelock gravity and investigate their thermodynamics . apart from their possible relevance to string theory
, it is of general interest to explore black holes in generalized gravity theories in order to discover which properties are peculiar to einstein s gravity , and which are robust features of all generally covariant theories of gravity .
the outline of our paper is as follows .
we give a brief review of the field equations of third order lovelock gravity and the counterterm method for calculating conserved quantities in sec .
[ fiel ] . in sec .
[ rot ] we introduce the @xmath3-dimensional solutions with a complete set of rotational parameters and investigate their properties . in sec .
[ therm ] we obtain mass , angular momentum , entropy , temperature , charge , and electric potential of the @xmath3-dimensional black hole solutions and show that these quantities satisfy the first law of thermodynamics .
we also perform a local stability analysis of the black holes in the canonical and grand canonical ensembles .
we finish our paper with some concluding remarks .
the action of third order lovelock gravity in the presence of electromagnetic field may be written as @xmath4 where @xmath5 is the cosmological constant , @xmath6 and @xmath7 are gauss - bonnet and third order lovelock coefficients , @xmath8 is electromagnetic tensor field and @xmath9 is the vector potential .
the first term is the cosmological term , the second term , @xmath10 , is the einstein term , the third term is the gauss - bonnet lagrangian given as @xmath11 and the last term is the third order lovelock term @xmath12 from a geometric point of view the combination of these terms in seven and eight dimensions is the most general lagrangian that yields second order field equations , as in the four - dimensional case for which the einstein - hilbert action is the most general lagrangian producing second order field equations , or the five- and six - dimensional cases , for which the einstein - gauss - bonnet lagrangian is the most general one fulfilling this criterion . since the third lovelock term in eq .
( [ act1 ] ) is an euler density in six dimensions and has no contribution to the field equations in six or less dimensional spacetimes , we therefore consider @xmath3-dimensional spacetimes with @xmath13 .
varying the action with respect to the metric tensor @xmath14 and electromagnetic tensor field @xmath15 the equations of gravitation and electromagnetic fields are obtained as : @xmath16 where @xmath17 is the energy - momentum tensor of electromagnetic field , @xmath18 is the einstein tensor , and @xmath19 and @xmath20 are the second and third order lovelock tensors given as @xcite : @xmath21 @xmath22 the einstein - hilbert action ( with @xmath23 ) does not have a well - defined variational principle , since one encounters a total derivative that produces a surface integral involving the derivative of @xmath24 normal to the boundary .
these normal derivative terms do not vanish by themselves , but are canceled by the variation of the gibbons - hawking surface term @xcite @xmath25 the main difference between higher derivative gravity and einstein gravity is that the surface term that renders the variational principle well - behaved is much more complicated .
however , the surface terms that make the variational principle well - defined are known for the case of gauss - bonnet gravity@xcite to be @xmath26 , where @xmath27 is @xmath28 and where @xmath29 is induced metric on the boundary , @xmath30 is trace of extrinsic curvature of boundary , @xmath31 is the @xmath32-dimensional einstein tensor of the metric @xmath33 and @xmath34 is the trace of @xmath35 for the case of third order lovelock gravity , the surface term that makes the variational principle well defined is @xmath36 , where @xmath37 is @xmath38 in eq .
( [ ib3 ] ) @xmath39 is the second order lovelock tensor ( [ love2 ] ) for the boundary metric @xmath33 , and @xmath40 is the trace of @xmath41k_{ab } \notag \\ & & -(4k^{3}-12kk_{ed}k^{ed}+8k_{de}k_{f}^{e}k^{fd})k_{ac}k_{b}^{c}-24kk_{ac}k^{cd}k_{de}k_{b}^{e } \notag \\ & & + ( 12k^{2}-12k_{ef}k^{ef})k_{ac}k^{cd}k_{db}+24k_{ac}k^{cd}k_{de}k^{ef}k_{bf}\ } \label{pab}\end{aligned}\ ] ] in general @xmath42 is divergent when evaluated on solutions , as is the hamiltonian and other associated conserved quantities @xcite .
one way of eliminating these divergences is through the use of background subtraction @xcite , in which the boundary surface is embedded in another ( background ) spacetime , and all quasilocal quantities are computed with respect to this background , incorporated into the theory by adding to the action the extrinsic curvature of the embedded surface .
such a procedure causes the resulting physical quantities to depend on the choice of reference background ; furthermore , it is not possible in general to embed the boundary surface into a background spacetime . for asymptotically ads solutions , one can instead deal with these divergences via the counterterm method inspired by ads / cft correspondence @xcite .
this conjecture , which relates the low energy limit of string theory in asymptotically anti de - sitter spacetime and the quantum field theory on its boundary , has attracted a great deal of attention in recent years .
the equivalence between the two formulations means that , at least in principle , one can obtain complete information on one side of the duality by performing computation on the other side .
a dictionary translating between different quantities in the bulk gravity theory and their counterparts on the boundary has emerged , including the partition functions of both theories . in the present context
this conjecture furnishes a means for calculating the action and conserved quantities intrinsically without reliance on any reference spacetime @xcite by adding additional terms on the boundary that are curvature invariants of the induced metric .
although there may exist a very large number of possible invariants one could add in a given dimension , only a finite number of them are nonvanishing as the boundary is taken to infinity .
its many applications include computations of conserved quantities for black holes with rotation , nut charge , various topologies , rotating black strings with zero curvature horizons and rotating higher genus black branes @xcite .
although the counterterm method applies for the case of a specially infinite boundary , it was also employed for the computation of the conserved and thermodynamic quantities in the case of a finite boundary @xcite .
extensions to de sitter spacetime and asymptotically flat spacetimes have also been proposed @xcite .
all of the work mentioned in the previous paragraph was limited to einstein gravity . here
we apply the counterterm method to the case of the solutions of the field equations of third order lovelock gravity . at any given dimension
there are only finitely many counterterms that one can write down that do not vanish at infinity .
this does not depend upon what the bulk theory is i.e. whether or not it is einstein , gauss - bonnet , 3rd order lovelock , etc .
indeed , for asymptotically ( a)ds solutions , the boundary counterterms that cancel divergences in einstein gravity should also cancel divergences in 2nd and 3rd order lovelock gravity
. the coefficients will be different and depends on @xmath5 and lovelock coefficients as we will see this for the volume term in the flat boundary case below .
of course these coefficients should reduce to those in einstein gravity as one may expect . unfortunately we do not have a rotating solution to either gauss - bonnet or 3rd - order lovelock gravity that does not have a flat boundary at infinity .
consequently we restrict our considerations to counterterms for the flat - boundary case , i.e. @xmath43 , for which there exists only one boundary counterterm @xmath44 where @xmath45 is a scale length factor that depends on @xmath46 , @xmath6 and @xmath47 , that must reduce to @xmath46 as @xmath6 and @xmath48 go to zero . having the total finite action @xmath49
, one can use the quasilocal definition @xcite to construct a divergence free stress - energy tensor . for the case of manifolds with zero curvature boundary
the finite stress energy tensor is @xmath50 the first three terms in eq .
( [ stres ] ) result from the variation of the surface action ( [ ib1])-([pab ] ) with respect to @xmath51 , and the last term is the counterterm that is the variation of @xmath52 with respect to @xmath51 . to compute the conserved charges of the spacetime
, we choose a spacelike surface @xmath53 in @xmath54 with metric @xmath55 , and write the boundary metric in adm form : @xmath56 where the coordinates @xmath57 are the angular variables parameterizing the hypersurface of constant @xmath1 around the origin , and @xmath58 and @xmath59 are the lapse and shift functions respectively . when there is a killing vector field @xmath60 on the boundary , then the quasilocal conserved quantities associated with the stress tensors of eq .
( [ stres ] ) can be written as @xmath61 where @xmath62 is the determinant of the metric @xmath55 , and @xmath63 is the timelike unit normal vector to the boundary @xmath64**. * * for boundaries with timelike ( @xmath65 ) and rotational ( @xmath66 ) killing vector fields , one obtains the quasilocal mass and angular momentum @xmath67 provided the surface @xmath53 contains the orbits of @xmath68 .
these quantities are , respectively , the conserved mass and angular momentum of the system enclosed by the boundary @xmath53 . note that they will both depend on the location of the boundary @xmath53 in the spacetime , although each is independent of the particular choice of foliation @xmath69 within the surface @xmath70 .
as stated before , the third order lovelock term in eq .
( [ act1 ] ) is an euler density in six dimensions and has no contribution to the field equations in spacetimes of dimension six or less .
taking @xmath13 , we obtain the @xmath3-dimensional solutions of third order lovelock gravity with nonvanishing electromagnetic field with @xmath71 rotation parameters and investigate their properties .
the rotation group in @xmath2 dimensions is @xmath72 and therefore the number of independent rotation parameters is @xmath73 $ ] , where @xmath74 $ ] is the integer part of @xmath75 .
the metric of an @xmath3-dimensional asymptotically ads rotating solution with @xmath76 $ ] rotation parameters whose constant @xmath77 hypersurface has zero curvature may be written as @xcite @xmath78 where @xmath79 , the angular coordinates are in the range @xmath80 and @xmath81 is the euclidean metric on the @xmath82-dimensional submanifold with volume @xmath83 . using eq .
( [ emeq ] ) , one can show that the vector potential can be written as @xmath84 where @xmath85 is an arbitrary real constant which is related to the charge of the solution . to find the function @xmath86
, one may use any components of eq .
( [ geq ] ) .
the simplest equation is the @xmath87 component of these equations which can be written as @xmath88 f^{\prime } + \lambda r^{6 } \notag \\ & & + 360(_{\phantom{n}{6}}^{n-1})\alpha _ { 3}f^{3}-12(_{\phantom{n}{4}% } ^{n-1})\alpha _ { 2}r^{2}f^{2}+(_{\phantom{n}{2}}^{n-1})r^{4}f =- q^{2}r^{8 - 2n } \label{eqf}\end{aligned}\ ] ] where prime denotes the derivative with respect to @xmath1 .
we present the full solution to this equation in the appendix . here , for simplicity , we consider the solutions of eq .
( [ eqf ] ) for a restricted version of the @xmath89 s given as @xmath90 equation ( [ eqf ] ) with condition ( [ con ] ) has one real and two complex solutions that are the complex conjugate of each other .
the real solution of eq .
( [ eqf ] ) with condition ( [ con ] ) is @xmath91 where @xmath92 is the mass parameter .
although the other components of the field equation ( [ geq ] ) are more complicated , one can check that the metric ( [ met2 ] ) satisfies all the components eq .
( [ geq ] ) provided above @xmath93 is given by ( [ frs ] ) .
unlike the solution in gauss - bonnet gravity , which has two branches , the solution ( [ frs ] ) has only one branch . indeed , eq . ( [ eqf ] ) with the above @xmath94 s has the real solution ( [ frs ] ) and two complex solutions that are complex conjugates of each other .
this feature is the same as the asymptotically ads solution of ref .
@xcite , which has a unique anti de sitter vacuum .
these solutions are asymptotically ads or ds for negative or positive values of @xmath5 respectively . in this paper
we are interested in the case of asymptotically ads solutions , and therefore we put @xmath95 . one can show that the kretschmann scalar @xmath96 diverges at @xmath97 , and therefore there is a curvature singularity located at @xmath97 . seeking possible black hole solutions , we turn to looking for the existence of horizons . as in the case of rotating black hole solutions of einstein gravity , the above metric given by eqs .
( [ met2 ] ) and ( [ fr ] ) has both killing and event horizons .
the killing horizon is a null surface whose null generators are tangent to a killing field . the proof that a stationary black hole event horizon must be a killing horizon in the four - dimensional einstein gravity @xcite can not obviously be generalized to higher order gravity .
however the result is true for all known static solutions .
although our solution is not static , the killing vector @xmath98 is the null generator of the event horizon , where @xmath71 denotes the number of rotation parameters .
the event horizon is defined by the solution of @xmath99 . for the case of uncharged solutions ,
there exists only one event horizon located at @xmath100 this feature is different from the case of uncharged spherically symmetric solutions of third order lovelock gravity with curved horizon @xcite , for which one may have uncharged black holes with two horizons , extreme ones , or a naked singularity .
as we demonstrate in the appendix , the general uncharged solution does not have two horizons**. * * the charged solution presents a black hole solution with two inner and outer horizons , provided the mass parameter @xmath92 is greater than , @xmath101 , an extreme black hole for @xmath102 and a naked singularity otherwise , where @xmath103 is @xmath104 * * * * the general charged solution also has only two horizons , as shown in the appendix .
one can obtain the temperature and angular momentum of the event horizon by analytic continuation of the metric . setting @xmath105 and @xmath106
yields the euclidean section of ( [ met2 ] ) , whose regularity at @xmath107 requires that we should identify @xmath108 and @xmath109 , where @xmath110 and @xmath111 s are the inverse hawking temperature and the angular velocities of the outer event horizon .
one obtains @xmath112 next , we calculate the electric charge of the solutions . to determine the electric field we should consider the projections of the electromagnetic field tensor on special hypersurfaces .
the normal to such hypersurfaces is @xmath113 and the electric field is @xmath114 .
then the electric charge per unit volume @xmath115 can be found by calculating the flux of the electric field at infinity , yielding @xmath116 the electric potential @xmath117 , measured at infinity with respect to the horizon , is defined by @xcite @xmath118 where @xmath119 is the null generator of the horizon given by eq .
( [ kil ] ) .
we find @xmath120 conserved quantities associated with the spacetime described by ( [ met2 ] ) can be obtained via the counterterm method .
using eqs .
( [ mtot ] ) and ( [ jtot ] ) ,
the mass and angular momentum will be finite provided @xmath121 where we note that @xmath45 reduces to @xmath46 as @xmath94 goes to zero . the mass and angular momentum per unit volume @xmath115 can then be obtained through the use of eqs .
( [ mtot ] ) and ( [ jtot ] ) .
we find @xmath122 m , \label{mtot1 } \\
j_{i } & = & \frac{1}{16\pi } n\xi ma_{i } \label{jtot1}\end{aligned}\ ] ] black hole entropy typically satisfies the so - called area law , which states that the entropy of a black hole equals one - quarter of the area of its horizon @xcite .
this near - universal law applies to all kinds of black holes and black strings in einstein gravity @xcite .
however in higher derivative gravity the area law is not satisfied in general @xcite .
it is known in lovelock gravity that @xcite @xmath123}k\alpha _ { k}\int d^{n-1}x\sqrt{\tilde{g}}% \tilde{\mathcal{l}}_{k-1 } \label{enta}\ ] ] where the integration is done on the @xmath124-dimensional spacelike hypersurface of killing horizon , @xmath125 is the induced metric on it , @xmath126 is the determinant of @xmath125 and @xmath127 is the @xmath71th order lovelock lagrangian of @xmath128 . for the topological class of black holes we are considering ,
the horizon curvature is zero .
consequently the area law holds . denoting the volume of the hypersurface boundary at constant @xmath0 and @xmath1 by @xmath129
, we obtain @xmath130 for the entropy per unit volume @xmath115 .
the entropy can also be obtained through the use of gibbs - duhem relation @xmath131 where @xmath132 is the finite total action evaluated on the classical solution , and @xmath133 and @xmath134 are the conserved charges and their associate chemical potentials respectively . for simplicity , we consider the uncharged solutions , for which @xmath135 and @xmath136 .
using eqs .
( [ act1 ] ) , ( [ ib1])-([ib3 ] ) and ( [ ict ] ) , the finite total action per unit volume @xmath115 can be calculated as @xmath137 now using gibbs - duhem relation ( [ gibs ] ) and eqs .
( [ mtot1 ] ) , ( [ jtot1 ] ) and [ acttot ] ) one may confirm that the entropy per unit volume obeys the area law of eq .
( [ ent1 ] ) .
we first obtain the mass as a function of the extensive quantities @xmath138 , @xmath139 and @xmath140 . using the expression for the entropy , the mass , the angular momenta , and the charge given in eqs .
( [ bet1 ] ) , ( [ ch ] ) , ( [ mtot1 ] ) and ( [ jtot1 ] ) , and the fact that @xmath141 , one can obtain a smarr - type formula as @xmath142 where @xmath143 and @xmath144 is the positive real root of the following equation @xmath145}\right\ } ^{(2n-2)}=0 .
\label{zsmar}\ ] ] one may then regard the parameters @xmath138 , @xmath146 and @xmath140 as a complete set of extensive parameters for the mass @xmath147 and define the intensive parameters conjugate to @xmath138 , @xmath148 and @xmath140 .
these quantities are the temperature , the angular velocities , and the electric potential @xmath149 it is a matter of straightforward calculation to show that the intensive quantities calculated by eqs .
( [ smar])-([dsmar ] ) are the same as those found earlier in this section .
thus , the thermodynamic quantities calculated in this section satisfy the first law of thermodynamics , @xmath150 the stability of a thermodynamic system with respect to the small variations of the thermodynamic coordinates , is usually performed by analyzing the behavior of the entropy @xmath151 near equilibrium .
the local stability in any ensemble requires that @xmath151 be a concave function of its extensive variables or that its legendre transformation is a convex function of the intensive variables**. * * the stability can also be studied by the behavior of the energy @xmath152 which should be a convex function of its extensive variable .
thus , the local stability can in principle be carried out by finding the determinant of the hessian matrix of @xmath152 with respect to its extensive variables @xmath153 , @xmath154 $ ] @xcite . in our case
the entropy @xmath138 is a function of the mass , the angular momenta , and the charge .
the number of thermodynamic variables depends on the ensemble that is used . in the canonical ensemble , the charge and the angular momenta
are fixed parameters , and therefore the positivity of the heat capacity @xmath155 is sufficient to ensure local stability .
the heat capacity @xmath156 at constant charge and angular momenta is @xmath157[% ( n-1)(n-2)r_{+}^{2(n-1)}+2q^{2}l^{2}]\{n(n-1)r_{+}^{2(n-1)}-2q^{2}l^{2}\ } \label{cap}\ ] ] where @xmath158 is @xmath159 - 4(n-1)q^{2}l^{2}r_{+}^{(2n-2)}[(3n-6)\xi ^{2}-n^{2}+3 ] \notag \\ & & + n(n-1)^{2}(n-2)r_{+}^{(4n-4)}[(n+2)\xi ^{2}-(n+1 ) ] \label{cap2}\end{aligned}\ ] ] figure [ figure1 ] shows the behavior of the heat capacity as a function of the charge parameter .
we see that @xmath160 is positive in various dimensions and goes to zero as @xmath85 approaches its extreme value .
thus , the @xmath3-dimensional asymptotically ads charged rotating black brane is locally stable in the canonical ensemble . in the grand - canonical ensemble , after some algebraic manipulation , we obtain @xmath161l^{2}\xi ^{6}r_{+}^{3n-4}}\left [ \frac{% n(n-1)r_{+}^{2(n-1)}+2(2n-3)l^{2}q^{2}}{(n-1)(n-2)r_{+}^{2(n-1)}+2l^{2}q^{2}}% \right ] .
\label{dhes}\ ] ] as one can see from eq .
( [ dhes ] ) , @xmath162 is positive over all phase space .
hence the @xmath3-dimensional asymptotically ads charged rotating black brane in third order lovelock gravity is locally stable in the grand - canonical ensemble .
in this paper , first , we introduced the surface terms for the third order lovelock gravity which make the action well - defined .
this is achieved by generalizing the gibbons - hawking surface term for einstein gravity or generalizing the surface terms of gauss - bonnet gravity .
we generalized the stress energy momentum tensor of brown and york in einstein gravity for the third order lovelock gravity for spacetimes with zero curvature at the boundary .
as in the case of einstein gravity , @xmath163 , @xmath164 , @xmath165 and @xmath37 of eqs .
( [ act1 ] ) , ( [ ib1 ] ) , ( [ ib2 ] ) and ( [ ib3 ] ) are divergent when evaluated on the solutions , as is the hamiltonian and other associated conserved quantities .
we also introduced a counterterm dependent only on the boundary volume , which removed the divergences of the action and conserved quantities of this solution of third order lovelock gravity .
we also found a new class of rotating solutions , whose hypersurfaces of constant @xmath0 and @xmath1 have zero curvature , in third order lovelock gravity in the presence of cosmological constant and electromagnetic field . these solutions are asymptotically ads or ds for @xmath166 or @xmath167 respectively .
we obtained solutions for special values of @xmath6 and @xmath48 given in eq .
( [ con ] ) with negative cosmological constant . in the absence of an electromagnetic field ,
these solutions present black branes with one event horizon .
the charged solutions may be interpreted as black brane solutions with two inner and outer event horizons for @xmath168 , extreme black holes for @xmath169 or naked singularity for @xmath170 , where @xmath103 is given in eq .
( [ mext ] ) .
we found that the killing vectors are the null generators of the event horizon , and therefore the event horizon is a killing horizon for the stationary solution of the third order lovelock gravity explored in this paper .
we computed physical properties of the brane such as the temperature , the angular velocity , the entropy , the electric charge and potential .
finally , we obtained a smarr - type formula for the mass of the black brane solution as a function of the entropy , the charge and the angular momenta of the black brane and investigated the first law of thermodynamics .
we found that the conserved and thermodynamics quantities satisfy the first law of thermodynamics .
we also studied the phase behavior of the @xmath3-dimensional charged rotating black branes in third order lovelock gravity and showed that there is no hawking - page phase transition in spite of the charge and angular momenta of the branes .
indeed , we calculated the heat capacity and the determinant of the hessian matrix of the mass with respect to @xmath138 , @xmath146 and @xmath140 of the black branes and found that they are positive for all the phase space , which means that the brane is stable for all the allowed values of the metric parameters discussed in sec .
[ therm ] .
this phase behavior is commensurate with the fact that there is no hawking - page transition for a black object whose horizon is diffeomorphic to @xmath171 and therefore the system is always in the high temperature phase @xcite .
acknowledgments this work has been supported in part by research institute for astrophysics and astronomy of maragha , iran and also by the natural sciences and engineering research council of canada .
r. mann is grateful to the kavili institute for theoretical physics where part of this work was carried out .
the general solution in @xmath2 dimensions is @xmath172 where @xmath173 and @xmath174 the above @xmath86 reduces to the solution ( [ frs ] ) when @xmath175 .. at large @xmath1 the function @xmath71 approaches a constant , and the spacetime is asymptotically de sitter or anti de sitter depending on the overall sign of @xmath176 where we have assumed @xmath177 without loss of generality .
first , we investigate the conditions of the reality of @xmath86 . in order to have real @xmath178 the expression @xmath179
should be positive .
this occurs for @xmath180 , but for negative @xmath181 , this holds if @xmath182 .
the analysis of this case proceeds as follows .
1 ) if @xmath183 , then @xmath184 is zero some where and the condition @xmath185 violated , and therefore the function @xmath86 is complex near the root(s ) of @xmath186 .
2 ) for @xmath187 the condition @xmath185 holds provided @xmath188 note that in this case @xmath189 should be larger than @xmath190 . seeking possible black hole solutions , we turn to looking for the existence of horizons .
the roots of the metric function @xmath86 are located at @xmath191 which reduces to the value of @xmath192 when @xmath193 as in eq .
( [ con ] ) .
note that @xmath194 is real if @xmath195 .
thus , there is no black hole solution for @xmath196**. * * for the case of uncharged solutions , there exists only one event horizon located at @xmath197 provided @xmath62 is positive . if @xmath62 is negative or zero then there will be no roots ( corresponding to a naked singularity ) .
note that @xmath198 when @xmath193 and therefore the above equation reduces to eq .
( [ rhq0 ] ) .
again this feature is different from the case of uncharged spherically symmetric solutions of third order lovelock gravity with curved horizon @xcite , for which one may have uncharged black holes with two horizons , extreme ones , or a naked singularity .
the charged solution presents a black hole solution with two inner and outer horizons , provided the mass parameter @xmath92 is greater than @xmath101 , an extreme black hole for @xmath102 and a naked singularity otherwise , where @xmath103 is @xmath199 again , one may note that for @xmath193 ( @xmath200 ) eq .
( [ mextgen ] ) reduce to @xmath101 given in eq .
( [ mext ] ) .
a. strominger , j. high energy phys .
* 10 * , 034 ( 2001 ) ; * 11 * , 049 ( 2001 ) ; v. balasubramanian , p. horova , and d. minic , _ ibid . _ * 05 * , 043 ( 2001 ) ; e. witten , hep - th/0106109 ; d. klemm , nucl . phys . *
b625 * , 295 ( 2002 ) ; v. balasubramanian , j. deboer , and d. minic , phys .
rev d * 65 * , 123508 ( 2002 ) ; s. nojiri and s. d. odintsov , phys .
b * 519 * , 145 ( 2001 ) ; j. high energy phys . * 12 * , 033 ( 2001 ) ; phys .
b * 523 * , 165 ( 2001 ) ; * 528 * , 169 ( 2002 ) ; r. g. cai , phys .
b * 525 * , 331 ( 2002 ) ; nucl .
* b628 * , 375 ( 2002 ) ; r. g. cai , y. s. myung , and y. z. zhang , phys .
d * 65 * , 084019 ( 2002 ) ; r. bousso , a. maloney , and a. strominger , _ ibid .
_ * 65 * , 104039 ( 2002 ) ; a. m. ghezelbash and r. b. mann , j. high energy phys . *
01 * , 005 ( 2002 ) ; m. h. dehghani , phys .
d * 65 * , 104003 ( 2002 ) .
t. jacobson and r. c. myers , phys .
lett . * 70 * , 3684 ( 1993 ) ; r. m. wald , phys .
d * 48 * , r3427 , ( 1993 ) ; m. visser , _ ibid_. * 48 * , 5697 ( 1993 ) ; t. jacobson , g. kang and r. c. myers , _ ibid_. * 49 * , 6587,(1994 ) ; v. iyer and r. m. wald , _ ibid_. * 50 * , 846 ( 1994 ) . | we generalize the quasilocal definition of the stress energy tensor of einstein gravity to the case of third order lovelock gravity , by introducing the surface terms that make the action well - defined .
we also introduce the boundary counterterm that removes the divergences of the action and the conserved quantities of the solutions of third order lovelock gravity with zero curvature boundary at constant @xmath0 and @xmath1 .
then , we compute the charged rotating solutions of this theory in @xmath2 dimensions with a complete set of allowed rotation parameters .
these charged rotating solutions present black hole solutions with two inner and outer event horizons , extreme black holes or naked singularities provided the parameters of the solutions are chosen suitable .
we compute temperature , entropy , charge , electric potential , mass and angular momenta of the black hole solutions , and find that these quantities satisfy the first law of thermodynamics .
we find a smarr - type formula and perform a stability analysis by computing the heat capacity and the determinant of hessian matrix of mass with respect to its thermodynamic variables in both the canonical and the grand - canonical ensembles , and show that the system is thermally stable .
this is commensurate with the fact that there is no hawking - page phase transition for black objects with zero curvature horizon . |
it is now generally accepted that the galactic bulge is a tri - axial , bar - like structure .
observational evidence for a bar has arisen from several sources , such as the study of gas kinematics ( e.g. @xcite ) , surface brightness ( e.g. @xcite ) , star counts ( e.g. @xcite ) and microlensing ( e.g. @xcite ) ; see @xcite for a review .
observational data have been used to constrain dynamical models of the galaxy .
@xcite used the cobe - dirbe multi - wavelength observations of the galactic centre @xcite to constrain several analytic bar models . @xcite
used optical observations of red clump giant ( rcg ) stars to constrain theoretical bar models .
similarly , @xcite and @xcite traced the bulge rcg population in the infrared .
this work uses a sample of stars 30 times larger than that of @xcite , with a greater number of fields distributed across a larger area of the galactic bulge , thus allowing finer constraints to be placed on the bar parameters than those determined by @xcite .
our current understanding of the galactic bar is that it is orientated at about @xmath6 to the sun galactic centre line - of - sight , with the near end in the first galactic longitude quadrant .
the bar length is around 3.1 3.5 with axis ratio approximately @xmath7 @xcite .
the above bar parameters are generally accepted as a working model , however they are not well determined .
our understanding of the complete structure of the inner galactic regions is similarly incomplete .
for example , recent infra - red star counts collected by the spitzer space telescope for galactic longitudes @xmath8 = @xmath9 @xmath10 are best explained assuming a long thin bar oriented at an angle of @xmath11 to the sun galactic centre line @xcite while most previous studies ( performed at @xmath12 ) prefer a short bar with an opening angle of @xmath13 .
recently , @xcite report that nir observations of rcgs support the hypothesis that a long thin bar oriented at @xmath14 co - exists with a distinct short tri - axial bulge structure oriented at @xmath15 .
in addition , there may be some fine features , such as a ring in the galactic bulge @xcite , or a secondary bar @xcite , that are not yet firmly established .
it is therefore crucial to obtain as many constraints as possible in order to better understand the structure of the inner galaxy . in this paper
we present an analysis of rcg stars observed in the galactic bulge fields during the second phase of the ogle microlensing project @xcite .
these stars are bright and they are approximately standard candles , hence their magnitudes can be taken as an approximate measure of their distances .
number counts in 34 central bulge fields with @xmath16 and @xmath17 are used to constrain analytic tri - axial bar models , and thereby obtain estimates on bar parameters .
we repeat the analysis with 44 fields with @xmath18 .
we find the fitted bar parameters support the general orientation and shape of the bar reported by other groups .
this paper is organised as follows : in section 2 we describe the ogle microlensing experiment and photometry catalogue and we illustrate how rcg stars can be used as approximate distance indicators ; in section 3 we detail how rcgs in the ogle - ii proper motion catalogue are selected ; in section 4 we compute the distance modulus to the red clump in 45 ogle - ii fields and thereby trace the central mass density of the galaxy ; in section 5 we describe how rcg star count histograms for each field can be used to constrain analytic bar models of the inner galaxy ; our results and their comparison to previous works is given in section 6 and in section 7 we discuss the implications and limitations of these results .
the ogle @xcite and moa @xcite microlensing collaborations currently make routine observations of crowded stellar fields towards the galactic bulge , and issue alerts when a microlensing event is detected . a result of this intense monitoring is the creation of massive photometry databases for stars in the galactic bulge fields .
such databases are extremely useful for kinematic and population studies of the central regions of the galaxy .
@xcite obtained the proper motions for millions of stars in the ogle - ii database for a large area of the sky .
[ fig : fields ] shows the ogle - ii fields towards the galactic bulge . in this paper
we focus on the population of red clump giant stars at the galactic centre .
red clump giants are metal - rich horizontal branch stars ( * ? ? ?
* and references therein ) .
theoretically , one expects their magnitudes to have ( small ) variations with metallicity , age and initial stellar mass @xcite .
empirically they appear to be reasonable standard candles in the @xmath19-band with little dependence on metallicities @xcite .
@xcite used rcg stars in 12 fields ( see fig . [
fig : fields ] ) observed during the first phase of the ogle microlensing experiment , ogle - i , to constrain several analytic models of the galactic bar density distribution .
@xcite , @xcite and @xcite similarly used ir observations of rcgs to trace the bulge stellar density .
we follow similar procedures to extract rcg stars from the ogle - ii galactic bulge fields and to constrain analytic models .
we compute the reddening - independent magnitude for all stars in each of the 45 ogle - ii fields : @xmath20 where @xmath21 and @xmath22 are the extinctions in the @xmath19 and @xmath23 bands determined by @xcite .
we select stars which have @xmath24 , where @xmath25 is a constant chosen for each field that excludes the main - sequence dwarf stars , and @xmath26 , which corresponds to the magnitude of rcg stars closer than 15 kpc and @xmath27 respectively , with @xmath28 . ] .
[ fig : cmd ] shows the sample of stars selected from the @xmath29 cmd for ogle - ii field 1 .
reddening - independent magnitude vs colour diagram for ogle - ii field 1 .
the red clump is clearly visible .
stars are selected ( black dots ) using the criteria @xmath24 , where @xmath25 is a constant chosen for each field , and @xmath26 ( solid lines , see text ) . ]
the selected stars are then collected in @xmath30 magnitude bins , see fig .
[ fig : hist ] . a function comprised of quadratic and gaussian components
is used to model this number count histogram in each of the ogle - ii fields : @xmath31 \label{eq : fitfun}\ ] ] where is the spread of the red clump giant magnitudes , is the mean apparent magnitude of the red clump , and @xmath32 , @xmath33 and @xmath34 are coefficients for the gaussian and quadratic components respectively .
these six parameters are allowed to vary for each of the ogle - ii fields and the best - fitting values obtained by minimising @xmath35^{2}$ ] where the sum is taken over all @xmath36 histogram bins which cover the range @xmath37 .
the error on histogram number counts is assumed to be @xmath38 , i.e. poissonian .
number count histogram for selected stars in ogle - ii field 1 .
the heavy solid line is the best - fitting model of the form given by equation ( [ eq : fitfun ] ) .
the last histogram bin is generally not included in the fitting due to incompleteness effects near the limiting magnitude of @xmath26 . ] the errors on the mean magnitude and distribution width of the red clump stars , and respectively , are determined using a maximum - likelihood analysis ( see e.g. @xcite ) .
we determine the distance modulus to the red clump in each field as : @xmath39 where @xmath40 is the distance to the red clump measured in parsecs , is the fitted peak reddening - independent magnitude of the red clump , @xmath41 is the mean value of @xmath42 for each field .
@xmath43 @xcite and @xmath44 @xcite are the mean absolute magnitude and colour of the population of local red clump giant stars .
we assume that the properties of the local population of red clump giant stars are the same as the galactic bulge population , however it is likely that population effects are significant .
we discuss the effects of red clump population effects in section [ sec : popeffects ] .
table [ tab : distmod ] lists the fitted parameters @xmath45 and @xmath46 , along with the mean value of @xmath41 for each field .
the distance modulus , @xmath47 , computed from equation ( [ eq : mu ] ) showed that there was a significant offset : the fitted mean red clump giant magnitudes were uniformly too faint compared to that expected of typical local rcg stars at 8 , resulting in overly large distance moduli .
@xcite found that the ogle - ii rcgs are approximately 0.3 mag fainter than expected when assuming that the population of rcgs in the bulge is the same as local .
for this reason , we apply an offset of 0.3 mag to the distance moduli computed above .
the shifted distance moduli @xmath48 are given in table [ tab : distmod ] .
the true line - of - sight dispersion , , can be approximated by @xmath49 where is the gaussian dispersion fitted using equation ( [ eq : fitfun ] ) , @xmath50 is the intrinsic dispersion of the rcg luminosity function and @xmath51 is an estimate of the photometric errors @xcite .
we use @xmath52 ( see section [ sec : data ] ) and @xmath53 , along with the tabulated values of to determine .
[ fig : distmod ] shows the mean distance to the red clump stars in each of the 45 ogle - ii fields listed in table [ tab : distmod ] .
the fields with galactic longitude @xmath16 show clear evidence of a bar , with a major axis oriented at @xmath54 to the sun - galactic centre line - of - sight . for fields with @xmath55 and @xmath56 the mean position of the red clump stars do not continue to trace the major axis of a linear structure .
@xcite find similar evidence that the position of the red clump stars is not predicted by a bar for @xmath57 , suggesting that this is a detection of the end of the bar , or the beginning of a ring - like structure .
we investigate these possibilities in section [ sec : ring ] .
the uncertainties on the mean position of the red clump are large ; the largest term in the error expression for arises from the relatively large uncertainty in the intrinsic colour of the rcgs .
the true line - of - sight dispersions , , are consistent with a wide range of bar position angles , but the mean position of the red clump in each direction strongly suggest a bar oriented along a direction consistent with that determined by the previous work referred to in section [ sec : intro ] .
ccccccccc field & @xmath8 & @xmath33 & & & @xmath41 & @xmath58 & & @xmath59 + structure of the inner region of the milky way as traced by red clump giant stars extracted from the ogle - ii microlensing survey data .
the mean position of the red clump for 45 ogle - ii fields is indicated by black dots with errors shown by black lines .
the grey lines along each line of sight indicate the 1-@xmath60 spread in distances obtained from fitting equation ( [ eq : fitfun ] ) to the red clump data in each field and correcting for intrinsic red clump luminosity dispersion and photometric errors ; see text .
a position angle of 25is shown by the thin solid line .
lines - of - sight at @xmath61 and @xmath62 are indicated along the top axis . ]
@xcite found that the extinction - corrected magnitudes of rcg stars in the ogle - ii bulge fields were 0.3 mag higher than that of a rcg with an intrinsic magnitude equal to those of local rcgs , placed at a distance of 8 .
@xcite notes that the cause of this offset is uncertain but may be resolved when detailed v - band ogle - ii photometry of rr lyrae stars will allow improved extinction zero - point estimations for all fields .
there is evidence that the properties of bulge rcgs are different to local rcg stars , in particular age and metallicity , resulting in a different average absolute magnitude ( see e.g. @xcite ) .
this could in part explain the observed offset of 0.3 mag between local and bulge rcg stars .
the distance modulus plotted in fig .
[ fig : distmod ] is as for equation ( [ eq : mu ] ) : @xmath63 where @xmath64 mag is the offset between the observed bulge rcg population and a fiducial local rcg placed at 8 . including the difference in intrinsic magnitude between local and bulge rcg populations we have : @xmath65 where is the intrinsic rcg magnitude difference between local and bulge populations and @xmath66 is a magnitude offset arising from effects other than population differences .
the theoretical population models of @xcite estimate @xmath67 .
the remaining magnitude offset @xmath68 can be accounted for by decreasing the assumed value of the galactocentric distance from 8 to 7.3 .
this assumption of is in agreement with the value of @xmath69 ( stat ) @xmath70 ( sys ) determined by @xcite , and that of @xmath71 determined by @xcite .
we note that as long as the stellar population is uniform in the galactic bulge ( an assumption largely consistent with the data ) , then this unknown offset only affects the zero point ( and hence the distance to the galactic centre ) , and the absolute bar scale lengths .
however the ratios between bar scale lengths are very robust , which we demonstrate in section [ sec : changer0 ] where we determine the bar parameters for several values of .
a metallicity gradient across the galactic bulge would have an effect on the intrinsic colour and absolute magnitude of bulge rcg stars as a function of galactic longitude .
@xcite find that there is a connection between metallicity gradient and structural features in a sample of 6 barred galaxies . @xcite measured metallicities for k giants in two fields at 1.5 and 1.7 from the galactic centre .
the average metallicity was [ fe / h ] = @xmath72 , lower than that of k giants in baade s window .
however , no such metallicity gradient was reported for galactocentric distance range 500 pc 3.5 by @xcite .
similarly , determine that there is no metallicity gradient within angles @xmath73 to @xmath74 of the galactic centre ( corresponding to 300 800 pc from the gc assuming a sun - gc distance of 8 ) .
we continue to investigate the structure of the galactic bar by fitting analytic models of the stellar density to the observed rcg data in the ogle - ii fields .
following @xcite we use the analytic models of @xcite to fit the observed data .
three model families ( gaussian , exponential and power - law ) are used : @xmath75 \rho_{\rm e1 } & = \rho_{0 } \exp(-r_{\rm e } ) \\
\rho_{\rm e2 } & = \rho_{0 } \exp(-r ) \label{eq : e2 } \\
\rho_{\rm e3 } & = \rho_{0 } k_{0}(r_{\rm s } ) \\[20pt ] \rho_{\rm p1 } & = \rho_{0 } \left(\frac{1}{1+r}\right)^{4}\\ \rho_{\rm p2 } & = \rho_{0 } \left(\frac{1}{r(1+r)^{3}}\right ) \\
\rho_{\rm p3 } & = \rho_{0 } \left(\frac{1}{1+r^{2}}\right)^{2 } \label{eq : p3 } \\[10pt ] \notag\end{aligned}\ ] ] where @xmath76 is the modified bessel function of the second kind and @xmath77^{1/2 } \\
r_{\rm e } & \equiv \left[\frac{|x|}{x_{0 } } + \frac{|y|}{y_{0 } } + \frac{|z|}{z_{0 } } \right ] \\
r_{\rm s } & \equiv \left[\left[\left(\frac{x}{x_{0}}\right)^{2 } + \left(\frac{y}{y_{0}}\right)^{2}\right]^{2 } + \left(\frac{z}{z_{0}}\right)^{4 } \right]^{1/4}\end{aligned}\ ] ] the co - ordinate system has the origin at the galactic centre , with the @xmath78 plane defining the mid - plane of the galaxy and the @xmath79 direction parallel to the direction of the galactic poles .
the @xmath80 direction defines the semi - major axis of the bar .
the functions are rotated by an angle @xmath81 around the @xmath79-axis .
an angle of @xmath82 corresponds to the major axis of the bar pointing towards the sun .
the functions can also be rotated by an angle @xmath83 around the @xmath84 axis , corresponding to the sun s position away from the mid - plane of the galaxy .
we aim to fit the observed number count histograms for each field as a function of magnitude .
given a magnitude , the number of stars with this magnitude is @xcite : @xmath85 where the integration is taken over distance @xmath86 .
we perform the integration over this range of @xmath87 as we do not expect the tri - axial bulge density structure to exceed these limits .
the constant @xmath88 is dependent on the solid angle subtended around each line of sight .
the luminosity @xmath89 is given by @xmath90 and @xmath91 is a constant .
the luminosity function @xmath92 is @xmath93\ ] ] where @xmath94 is the luminosity of the red clump and @xmath95 is the intrinsic spread in red clump giant luminosity and is held constant .
there are ten parameters to be determined in the above equations : the three bar scale lengths @xmath96 , @xmath97 , @xmath3 ; the bar orientation and tilt angles @xmath81 and @xmath83 ; the luminosity function parameters @xmath98 , @xmath99 , @xmath100 and @xmath94 and the density function parameter @xmath101 .
there is evidence that the centroid of the bar is offset from the centre of mass of the galaxy , a feature commonly observed in external galaxies ( @xcite and references therein ) .
we include an additional parameter in the modelling process , @xmath102 , which allows for this possible centroid offset .
theoretical number counts are computed for the @xmath103 fields at longitudes @xmath104 as @xmath105 where the offset parameter @xmath102 is determined over all fields for a given density model .
we apply an exponential cut - off to the density functions similar to that of @xcite : @xmath106 where @xmath107 is in kpc , @xmath108 and @xmath109 is a cut - off radius .
the theoretical constraint on the maximum radius of stable stellar orbits is the co - rotation radius , @xmath109 .
we adopt the co - rotation value of @xmath110 determined by @xcite in equation ( [ eq : cutoff ] ) .
we later repeat the modelling process without this theoretical cut - off , see section [ sec : changer0 ] .
@xcite included a further fitting parameter to account for a possible metallicity gradient across the bulge , but found that this did not significantly affect the bar parameters .
the discussion in section [ sec : popeffects ] suggests that there is no appreciable metallicity gradient over the bulge region investigated here .
we therefore do not include an additional model parameter corresponding to a metallicity gradient in the model fitting analysis .
the model fitting was performed using a standard non - linear neadler - mead minimisation algorithm . for each of the nine density profile models we minimise @xmath111 : @xmath112 where the summations are taken over each of the 26 histogram bins in @xmath113 ogle - ii fields
, @xmath114 is the observed histogram data for field @xmath25 and @xmath115 is the model number count histogram from equation ( [ eq : numcount ] ) .
the error in @xmath116 was taken to be @xmath117 .
the 11 parameters @xmath96 , @xmath97 , @xmath3 , @xmath81 , @xmath83 , @xmath118 , @xmath98 , @xmath99 , @xmath100 , @xmath94 and @xmath102 were fitted for each of the nine density profiles given in equations .
( [ eq : g1 ] [ eq : p3 ] ) for the 34 ogle - ii fields which have @xmath16 . a naive interpretation of fig .
[ fig : distmod ] is that fitting all fields including those with @xmath55 and @xmath56 with a single bar - like structure would not be successful and indeed initial modelling using data from all fields resulted in bar angles of @xmath119 . as seen in fig .
[ fig : distmod ] this result is consistent with the mean position angle of all fields , but clearly does not describe the bar correctly .
we therefore begin modelling the bar using data from the central 34 fields which have @xmath16 . in section [ sec : ring ]
we proceed by including the data from fields with @xmath55 and @xmath56 in the modelling process and in section [ sec : changer0 ] we consider the effect of changing .
the range of field latitudes is small and therefore the information available in the latitude direction from the total dataset unlikely to be able to constrain @xmath83 strongly .
initial modelling runs held @xmath83 constant at @xmath120 .
once a minimum was determined for each model holding @xmath121 , this parameter was allowed to vary with the ten other parameters .
figure [ fig : matrixe2 ] shows the number count histograms of observed red clump giants in the 34 ogle - ii fields with @xmath16 , along with the best - fitting model ( equation ( [ eq : numcount ] ) ) using the e2 density profile given in equation ( [ eq : e2 ] ) as an example [ eq : p3 ] ) appear in the on - line supplementary material .
each set of axes is arranged roughly in order of decreasing galactic longitude .
the magnitude of the red clump peak increases with decreasing longitude , consistent with a bar potential .
the observed number count histograms are reasonably well - fitted by all models in most fields .
the most obvious exceptions are fields 6 , 7 , 14 , 15 and 43
. these fields are at the most extreme latitudes represented in the data . in the case of fields 14 and 15
the algorithm fails to fit satisfactorily both observed histograms for a given model . upon closer inspection of the observed histogram data for these two fields ,
it is noted that the total number of stars in the field 15 histogram is significantly less than that for field 14 .
the models preferred by all fields clearly can not reproduce the decline in total star count between fields 14 and 15 . upon inspection of the ccd pixel position of stars for field 15
it was found that there are no stars recorded in a region along one edge of the field .
the lack of stars in this strip is due to missing @xmath23-band data for these stars .
similarly , there is a lack of stars in field 15 in the lower right corner , due to heavy dust extinction in this area .
all models predict the peak of the red clump to be at a lower magnitude than that observed for field 43 .
a similar offset is seen in the other high latitude fields 14 and 15 .
these fields are all at latitude @xmath122 and the observed offset in all three fields is likely to be related to this common position in latitude .
the cause of this effect is currently unknown . the observed offset between the maximum density position predicted by the tri - axial bar models and that observed for some fields may imply some asymmetry of the bar in the latitude direction , as suggested by @xcite and references therein .
fields 6 and 7 , both at @xmath123 , are also poorly fitted by all models .
the histograms of observed red clump stars in these fields show a curious double peak .
clearly the density functions equations ( [ eq : g1 ] [ eq : p3 ] ) are inadequate for describing such a feature
. this double - peaked structure may be the result of another population of stars lying along the line of sight , at a distance different to the main bulge population .
the best - fitting model curves typically have a peak at magnitudes corresponding to the highest peak in the observed histogram , thereby suggesting that the second population , if real , and composed of a significant fraction of rcgs , exists between us and the bulge population .
a population of asymptotic giant branch ( agb ) stars @xcite may also be the cause of the secondary peak in the number count histograms for these and similar fields ( e.g. field 10 , fig . [
fig : histmatrixwide_g ] ) .
closer investigation of the stellar characteristics and kinematics of stars in these fields may help in describing the origin of the second number count peak .
the best - fitting parameter values for all the nine density models are presented in table [ tab : paramvals ] .
two sets of parameter values are given , one where the tilt angle , @xmath83 , was held at @xmath124 , and the other where @xmath83 was allowed to vary .
the values of are well in excess of the number of degrees of freedom : @xmath125 , and can therefore not be used to set reliable errors on the parameters values in the usual way .
it is likely that the errors on the histogram data are underestimated , resulting in extreme values of .
we assumed that the errors on the number count data would be poissonian and we have not added any error in quadrature that might arise from any systematic effects . furthermore ,
as mentioned above , the extreme latitude fields 6 , 7 , 14 , 15 and 43 are poorly fit by every model .
the cause for this is unknown , however the effect is to add a relatively high contribution to the total value of compared to other fields .
the values of can only be used to differentiate between the various models in a qualitative manner . in terms of relative performance ,
model e3 best reproduces the observed number count histograms for the fields tested here , and model g2 provides the worst fit to the data .
cccccccc & & & & axis ratios + model & @xmath83 & @xmath126 & @xmath96 & @xmath97 & @xmath3 & & @xmath127 + the position angle of the bar semi - major axis with respect to the sun - galactic centre line - of - sight is @xmath126 and is related to the rotation angle @xmath81 by @xmath128 .
we find @xmath129 @xmath130 .
the addition of @xmath83 as another variable parameter does not result in a significant improvement in .
the lack of information in the latitude direction means that the data have little power in constraining parameters such as @xmath83 .
the absolute values of the scale lengths in table [ tab : paramvals ] can not be directly compared between models .
the axis ratios @xmath131 and @xmath132 can be compared however .
[ fig : axisratios ] shows the axis ratios @xmath131 and @xmath132 for the nine models tested .
the ratio of the major bar axis scale length to the minor bar axis in the plane of the galaxy , @xmath131 , has values @xmath133
@xmath134 , with the exception of that for model e1 , for which @xmath135 .
@xcite found @xmath136 @xmath137 , with the exception of model e1 for which @xmath138 .
it is interesting to note that our values of @xmath131 have a similar range to that of @xcite , when we exclude the outlying result from the same model ( e1 ) .
our results do suggest a slimmer bar , i.e. higher values of @xmath131 compared to @xcite .
both these results and the results of @xcite are consistent with the value of @xmath139 reported by @xcite .
the ratio of the major bar axis scale length to the vertical axis scale length @xmath132 was found to be @xmath140
@xmath141 , with the same exception of model e1 , which has an outlying value of @xmath142 . again , the range of ratio values is comparable to that of @xcite who found @xmath143
@xmath144 , with the exception of model e1 again , which had @xmath145 .
scale length ratios @xmath131 ( squares ) and @xmath132 ( triangles ) for all models .
solid and open symbols show the ratios for the best - fitting models using the 34 fields with @xmath16 , and all 44 fields ( see section [ sec : ring ] ) respectively . ]
the mean axis ratios are @xmath127 are @xmath146 : @xmath147 : @xmath148 . excluding the outlying results from the e1 model , the ratios are @xmath146 : @xmath149 : @xmath150 .
these results suggest a bar more prolate than the general working model with @xmath7 @xcite .
[ fig : distmod ] shows the mean position of the bulge red clump stars . at longitudes @xmath16 the bulge
red clump stars follow the main axis of the bar . at greater angular distances
, the mean positions of the red clump stars are clearly separated from the main bar axis .
@xcite find similar evidence and suggest that this could indicate either the end of the bar or the existence of a ring - like structure .
we used the ogle - ii data to investigate these possibilities .
the ogle - ii data from fields with longitude @xmath151 and @xmath152 were excluded from the original bar modelling based on the findings illustrated by fig .
[ fig : distmod ] , and because initial modelling efforts using all data from all fields simultaneously failed to produce satisfactory results .
using the best - fitting model ( e3 ) determined using the central ogle - ii fields , fig .
[ fig : histmatrixwide ] shows the predicted number count densities for fields with @xmath151 and @xmath152 , overlaid on the observed number count histograms .
it is clear from the left hand column of fig .
[ fig : histmatrixwide ] that the mean position of the red clump in fields with @xmath152 are significantly removed from that predicted by the e3 linear bar model .
however , the right hand column of fig .
[ fig : histmatrixwide ] shows that the observed position of the red clump is in rough agreement with that predicted from the model for fields with @xmath151 .
we test this further , by taking lines of sight through the best - fitting e3 model and computing the number count profile using equation ( [ eq : numcount ] ) .
the mean magnitude of the red clump is converted to a distance for each line of sight .
[ fig : midplanescan ] shows the density contours of the analytic e3 model overlaid with the mean position of the red clump determined in this way .
mean position of the red clump determined using the best - fitting analytic bar model ( e3 ) .
gray lines indicate 11 lines - of - sight through the e3 model , with @xmath153 and @xmath154 and contours indicating the density profile of the model .
the dashed line shows the orientation of the bar major axis , with the sun positioned at @xmath155 .
number count profiles are generated using equation ( [ eq : numcount ] ) .
the mean red clump magnitude for each line - of - sight was converted to a distance , and plotted as solid circles .
black circles indicate the mean position of the red clump when the exponential cut - off , equation ( [ eq : cutoff ] ) is applied .
grey circles indicate the mean rcg position when this cut - off is not imposed .
the observed features in fig . [
fig : distmod ] at longitudes @xmath156 are more clearly reproduced when the cut - off is applied to the analytic model . ]
the features of fig .
[ fig : distmod ] at longitudes @xmath156 are qualitatively reproduced in fig .
[ fig : midplanescan ] .
this suggests that the observed data are likely to be consistent with a single bar - like structure , rather than requiring additional structure elements .
we note however , that the abrupt departure of the location of the density peak in fig .
[ fig : distmod ] at longitudes @xmath157 is best reproduced in fig .
[ fig : midplanescan ] by the analytical density model when the exponential cut - off of equation ( [ eq : cutoff ] ) is imposed .
fields with @xmath151 and @xmath152 should be included in the modelling of the bar , as they are likely to provide further constraints on the final model . initial attempts to model the bar using all fields failed because the fitting algorithm found a model with a bar angle of @xmath158 , consistent with the observed data , but not consistent with the current understanding of the bar .
[ fig : midplanescan ] shows that the morphology indicated by the observed data in fig .
[ fig : distmod ] can be explained using a bar structure oriented at @xmath159 to the sun - gc line - of - sight .
@xcite also note that the line - of - sight density for tri - axial bulges reaches a maximum where the line - of - sight is tangential to the ellipsoidal density contours .
these authors also quantify the difference between the positions of maximum density and the intersection of the line - of - sight with the major axis of the bar .
the fitting procedure for all nine tri - axial models was repeated , including now the 10 ogle - ii fields with @xmath151 or @xmath152 .
figure [ fig : histmatrixwide_g ] shows the best - fitting number count profiles for the ten non - central fields for the ` g ' type of analytic tri - axial model .
the best - fitting number count profiles for fields with @xmath16 are not significantly different to those shown in figure [ fig : matrixe2 ] .
the main features of the observed number count profiles are reproduced by the best - fitting analytic models .
there are however instances where details of the observed number counts are not reproduced .
magnitude bins with @xmath160 are poorly fitted by the g - type models in positive longitude fields .
two of the e - type models ( e2 , e3 ) similarly fail to trace these data .
the number count profile for model e1 shows a flattened peak , resulting from the pronounced box - like nature of the density profile .
all the p - type models appear to reproduce the data at these magnitudes with similar profiles .
all models predict a rcg peak at magnitudes greater than that observed in fields 28 and 29 .
this offset between predicted and observed rcg peak location is also seen in fields 26 and 27 but to a lesser degree .
the best - fitting parameters are shown in table [ tab : paramvalswide ] .
the model which results in the lowest value of is still e3 .
model g2 also remains the worst - fitting model to the data .
cccccccc & & & & axis ratios + model & @xmath83 & @xmath126 & @xmath96 & @xmath97 & @xmath3 & & @xmath127 + we now consider the change to the best - fitting model parameters when data from all fields are used in the analysis .
the position angle of the bar semi - major axis with respect to the sun - galactic centre line - of - sight is now @xmath161 @xmath1 .
the bar scale lengths @xmath96 , @xmath97 and @xmath3 also changed ; fig .
[ fig : paramchange ] shows the ratio of the best - fitting scale lengths determined using all 44 fields to those found using only the central 34 fields . on average , upon including the data from wide longitude fields , the semi - major axis scale length , @xmath96 , decreased by @xmath162 ; the semi - minor axis scale length , @xmath97 , increased by @xmath163 and the vertical scale length , @xmath3 , remained essentially unchanged .
the relatively large change in the semi - minor axis scale length is intuitively understandable , as a bar position angle of @xmath164 with respect to the sun - gc line - of - sight mean the semi - minor axis direction has a large vector component in the galactic longitude direction .
the additional constraints of the data from fields at extended galactic longitudes can therefore have a pronounced effect on the fitted values of @xmath97 .
similarly , we expect some weak correlation between the semi - major and semi - minor axis scale lengths , evidenced by the slight decrease on average for the @xmath96 values upon adding the data from wide longitude fields .
it is unsurprising that the values of @xmath3 are unchanged , as adding data from the wide longitude fields has little power in further constraining model elements in the direction perpendicular to the galactic midplane .
ratio of bar scale lengths from best - fitting models using data from all 44 ogle - ii fields to those using data from the central 34 ogle - ii fields . ] as above , we consider the ratio of the axis scale lengths for each model ( see fig . [
fig : axisratios ] ) .
the ratio @xmath131 for all nine models using all field data has values 2.7 3.0 .
this range is now completely consistent with the range 2.5 3.3 reported by @xcite .
the ratio @xmath132 now lies in the range 3.6 3.9 , with the exception of that for model g1 which has @xmath165 .
the range @xmath166 3.6 3.9 is narrower than that found previously , 3.4 4.2 , yet still higher on average than 2.8 3.8 reported by @xcite .
the scale length ratios are now @xmath4 , which compared to those determined using only the central ogle - ii fields ( @xmath167 ) are closer to the working model proposed by @xcite which has @xmath5 .
there are features in the number count histograms that are not faithfully reproduced by the analytic tri - axial bar models for wide longitude fields .
specifically , the predicted number count dispersions around the rcg peak magnitudes for fields with @xmath168 are too small compared to that observed , see fig .
[ fig : histmatrixwide_g ] .
while the location of the observed maximum line - of - sight density can be reproduced using just a bar , it is possible that the reason why the analytic bar models underestimate the observed number count dispersions for fields with @xmath168 is because these models do not include elements which correspond to extended aggregations of stars at or near the ends of the bar .
clumps of stars at the ends of the bar would increase the line - of - sight density dispersion at wide longitudes .
such extra aggregations of stars have been observed in galaxies with boxy or peanut - shaped bulges @xcite .
their origin may be due to the superposition of members of the @xmath169 family of orbits @xcite many of which have loops near the end of the bar .
alternatively , the aggregations of stars near the ends of the bar may be the edge - on projection of an inner ring ( @xcite and references therein ) .
the presence of another density structure such as a long thin bar oriented at @xmath170 @xcite might also contribute to the relatively large line - of - sight density dispersion .
the effects of such a structure would be most pronounced at wide longitudes .
the presence of a spiral arm might similarly contribute to the large line - of - sight density dispersion for these fields .
we consider in this section the effect on the fitted model parameters of changing the galactocentric distance @xmath171 .
it was noted in section [ sec : popeffects ] that there is an offset between the peak magnitude of rcgs observed in the ogle - ii data and that expected for a rcg , with absolute magnitude similar to that of local rcgs , placed at a distance of 8.0 .
the observed rcgs are systematically 0.3 mag fainter than the fiducial local rcg at 8.0 .
there are two possible reasons for this magnitude offset .
firstly , the adopted galactocentric distance of @xmath172 may be incorrect ; secondly , there may be population variations between local and bulge rcgs , resulting in different intrinsic magnitudes in the two populations . in section [ sec : popeffects ]
we noted that the offset predicted from the theoretical population models of @xcite estimate @xmath67 .
we postulated that if the remaining magnitude offset is completely accounted for by a change in the galactocentric distance , this would mean that @xmath173 .
conversely , if we adopted the value of @xmath174 determined by @xcite , the magnitude offset would become @xmath175 mag , where including the population effect predicted by @xcite would leave an unaccounted - for offset of @xmath176 mag . in this section we present the results of further modelling where we apply magnitude offsets which correspond to different values of @xmath171 . in section [ sec :
results ] , we note that for the ogle - ii fields 6 , 7 , 14 , 15 and 43 , which have latitudes removed from the majority of ogle - ii fields , there are systematic offsets between the rcg peak in observed number count histograms and the predictions of all nine tri - axial models tested . for this reason
, we limited the fields used to those which have galactic latitude @xmath177 .
we also noted in section [ sec : results ] that due to the lack of field coverage in the latitude direction the data are not effective at constraining the bar tilt angle @xmath83 .
for this reason , @xmath83 was held constant at @xmath178 in the following modelling analysis . during the previous modelling efforts of section [ sec : results ]
it was also found that applying the exponential cut - off of equation ( [ eq : cutoff ] ) had little effect on determining the best - fitting bar parameters .
this cut - off was not applied in the following analysis .
the best - fitting bar parameters were determined for the nine tri - axial bar models for the ogle - ii fields occupying the latitude strip @xmath177 using magnitude offsets corresponding to @xmath171 values of 8.0 , 7.6 and 7.3 .
the resulting best - fitting parameters are listed in table [ tab : changer0 ] .
cccccccc & & & axis ratios + model & @xmath171 ( ) & @xmath81 ( ) & @xmath96 & @xmath97 & @xmath3 & & @xmath127 + from the results in table [ tab : changer0 ] we see that the fitted bar opening angle @xmath81 increases for all models as @xmath171 decreases . the fitted scale lengths @xmath96 , @xmath97 and @xmath3 all decrease linearly as @xmath171 decreases .
the values of in table [ tab : changer0 ] indicate that a smaller value of @xmath171 is favoured by all models .
the ratio between the scale heights remains remarkably constant with varying @xmath171 , indicating that while the orientation of the bar changes slightly with varying @xmath171 , the overall shape of the bar does not change .
the mean values of the scale length ratios for all models , and all values of @xmath171 are @xmath179 .
this result is very close to that determined using data from all 44 ogle - ii fields , see section [ sec : results ] above .
red clump giant stars in the ogle - ii microlensing survey catalogue can be used as tracers of the bulge density over a large region towards the galactic centre .
nine analytic tri - axial bar models were initially fitted to the number count histograms of red clump stars observed in 34 ogle - ii fields which have @xmath16 .
the models all have the major axis of the bar oriented at @xmath180
@xmath130 to the sun - galactic centre line - of - sight .
this orientation is in agreement with the results of @xcite and @xcite which give a bar angle of @xmath180
@xmath181 , and is marginally in agreement with the value of @xmath182 from @xcite .
@xcite obtain best - fitting non - parametric models to the cobe / dirbe l band map of the inner galaxy with bar angles of @xmath180 @xmath183 .
we find the ratio of the bar major axis scale length to minor axis scale length in the plane of the galaxy to be @xmath184 @xmath134
, higher than the value of 2.0 2.4 reported by @xcite , but consistent with the upper end of the range 2.5 3.3 found by @xcite .
the ratio of bar major axis scale length to vertical axis scale length was found to be @xmath185 4.2 , again higher on average than that reported by @xcite who found @xmath186 3.8 , and higher than the value of @xmath187 generally adopted @xcite .
the working model proposed by @xcite gives the scale length ratios as @xmath5 .
our results suggest a more prolate model with @xmath167 .
the observed separation of the mean position of red clump giants from the bar major axis at @xmath156 was shown to be a geometric effect , rather than evidence of a more complicated structure such as a ring .
the observed data from these fields were used to further constrain the bar models .
the resulting bar position angles was found to be @xmath0 @xmath1 .
this narrower range is consistent with the several values of the bar position angle found by previous studies .
the bar scale length ratios were determined to be @xmath4 , which are closer to the working model proposed by @xcite than those values found using data from only the central 34 ogle - ii fields .
reasons for the difference between the bar axis ratios determined here and the general working model of @xcite may include rcg population effects noted above in section [ sec : popeffects ] .
the intrinsic luminosity of bulge rcg stars was assumed to be the same as the local population , however it was found that an offset of @xmath188 mag had to be applied to the computed distance moduli in order to obtain results consistent with standard bar models .
@xcite found the mean magnitude of observed bulge clump stars is found to be 0.3 mag fainter to that expected assuming ( i ) the properties of bulge rcgs are the same as the local population and ( ii ) the distance to the galactic centre is 8 .
a possible implication is that the adopted distance to the galactic centre of 8 may be incorrect .
the bar modelling procedure was repeated for three values of the sun - galactic centre distance @xmath171 : 7.3 , 7.6 and 8.0 , using data from ogle - ii fields which have @xmath177 . some fields with latitudes outside this range
were found to have number count histograms which were not reproducible by any linear tri - axial model of the bar tested in this work , and were therefore excluded .
in addition , most of the ogle - ii fields latitudes @xmath189 were excluded .
the low amount of information in the latitude direction means that the data have little leverage in determining some of the model parameters . without more data in these regions ,
model fitting algorithms may not be able to refine some model parameters such as the tilt angle @xmath83 .
the three values of @xmath171 used are the ` default ' value of 8.0 ; 7.6 , as determined by @xcite ; and 7.3 , which corresponds to the value of @xmath171 consistent with the observed mean rcg magnitudes in the ogle - ii bulge fields assuming the population effects of @xcite . the fitted scale lengths @xmath96 , @xmath97 and @xmath3 were found to increase linearly with increasing values of @xmath171 , while the bar opening angle @xmath81 decreased slightly with increasing @xmath171 .
the shape of the bar , as quantified by the ratio of axis scale lengths , was found to be insensitive to different values of @xmath171 .
the mean ratio of scale lengths over all models and values of @xmath171 was found to be @xmath190 , which is slightly closer to the working model of @xcite for the galactic bar than that determined using all 44 ogle - ii fields .
the goodness - of - fit measure decreased for all models as the value of @xmath171 was lowered .
improved modelling for the galactic bar may be possible through the addition of further elements to the methods presented here .
the possibility of a metallicity gradient across the bulge was not accounted for in this work . including spiral terms ( see e.g. @xcite ) in the analytic density profiles
may similarly result in a closer reproduction of the observed number count profiles .
the observed maximum line - of - sight density can be reproduced using just a bar , without requiring additional structure elements . however , the finer details of the number count histograms , especially at wide longitudes , were not reproduced by the tri - axial bar models used here .
the predicted number count dispersions around the peak rcg magnitude were underestimated by the analytical models , particularly for fields with @xmath168 .
it is possible that these finer features can be reproduced using models which include extra stellar agglomerations at the ends of the bar .
the additional stellar densities could arise from specific stellar orbits aligned with the bar , or due to the projection effect of an inner ring .
a long thin bar as postulated by @xcite might similarly increase the line - of - sight density dispersion , particularly for wide longitude fields , as might the presence of a spiral arm . modelling the bar using non - parametric methods
( see e.g. @xcite ) may provide valuable insight into the existence and nature of such additional features .
some preliminary work applying these methods has begun .
data from current and future surveys of the galactic bulge region will be useful for refining the constraints on the bar parameters .
the third evolution of the ogle microlensing experiment , ogle - iii , is currently in progress , covering a larger region of the central galactic region than ogle - ii .
infra - red observations of the bulge have the advantage of lower extinction effects due to dust , compared to optical observations .
current catalogues which would be suitable for investigating the structure of the galactic bar include the point source catalogue from the 2mass all sky data release @xcite , the galactic plane survey from the ukirt infra - red deep sky survey @xcite and data from the isogal , spitzer / glimpse @xcite and akari @xcite space telescope surveys . proposed infra - red surveys towards the galactic centre include the galactic bar infra - red time - domain ( gabarit ) survey which aims to monitor the galactic bar region in the k band for microlensing events ( kerins 2006 , private communication ) .
the analysis of red clump giant star counts in these surveys may result in tighter constraints on the properties of the bar , and can be combined with constrains on stellar kinematics from proper motion surveys @xcite in order to develop dynamical models of the inner galaxy .
we thank d. faria , .
wyrzykowski , r. james and w. evans for helpful discussions .
njr acknowledges financial support by a pparc pdra fellowship and the lkbf .
njr thanks the kapteyn astronomical institute , rug , for visitor support .
this work was partially supported by the european community s sixth framework marie curie research training network programme , contract no .
mrtn - ct-2004 - 505183 ` angles ' . | red clump giant stars can be used as distance indicators to trace the mass distribution of the galactic bar .
we use rcg stars from 44 bulge fields from the ogle - ii microlensing collaboration database to constrain analytic tri - axial models for the galactic bar .
we find the bar major axis is oriented at an angle of @xmath0
@xmath1 to the sun - galactic centre line - of - sight .
the ratio of semi - major and semi - minor bar axis scale lengths in the galactic plane @xmath2 , and vertical bar scale length @xmath3 , is @xmath4 , suggesting a slightly more prolate bar structure than the working model of @xcite which gives the scale length ratios as @xmath5 .
galaxy : bulge - galaxy : centre - galaxy : structure |
this work was supported by the national science foundation through nsf grant phy-9503384 , and by a nasa theory grant .
m. rugers and c. j. hogan , report astro - ph/9603084 , 1996 ( unpublished ) ; m. rugers and c. j. hogan , astrophys . j. lett . * 459 * , l1 ( 1996 ) ; e. j. wampler _
et al . _ ,
( to be published ) ; r. f. carswell _ et al .
_ , mon . not
r. astron .
soc . * 278 * , 506 ( 1996 ) ; a. songaila , l. l. cowie , c. j. hogan , and m. rugers , nature * 368 * , 599 ( 1994 ) ; r. f. carswell _ et al .
_ , mon . not
. soc . * 268 * , l1 ( 1994 ) .
p. i. krastev , s. t. petcov , and l. qiuyu , iassns - ast 96/11 , hep - ph/9602033 , 1996 ( unpublished ) ; s. m. bilenky and c. giunti , z. phys .
c * 68 * , 495 ( 1995 ) ; p. i. krastev and s. t. petcov , nucl . phys .
b * 449 * , 605 ( 1995 ) .
w. kwong and s. p. rosen , phys .
lett . * 68 * , 748 ( 1992 ) .
g. m. fuller , j. r. primack , and y .- z .
qian , phys .
d * 52 * , 1288 ( 1995 ) ; j. j. gomez - cadenas and m. c. gonzales - garcia , z. phys .
( to be published ) ; s. goswami , cupp-95/4 , hep - ph/9507212 ( unpublished ) ; e. ma and p. roy , phys .
d * 52 * , r4780 ( 1995 ) ; e. ma and j. pantaleone , phys .
d * 52 * , r3763 ( 1995 ) ; r. foot and r. r. volkas , phys .
d * 52 * , 6595 ( 1995 ) ; z. g. berezhianai and r. n. mohapatra , phys .
d * 52 * , 6607 ( 1995 ) ; e. j. chun , a. s. joshipura , and a. y. smirnov , phys .
b * 357 * , 608 ( 1995 ) .
there is a small but nontrivial discrepancy between our results and the those of shi _ et al .
_ @xcite . for @xmath62 ,
the atmospheric neutrino solution would imply @xmath63 according to shi _
@xcite , while we obtain @xmath64 .
the difference is due to our use of the @xmath25 correction to @xmath2 , and the fact that we numerically integrate the altered n@xmath14p reactions in our bbn calculation .
thus the constraints on the @xmath17 atmospheric neutrino solution are not as relaxed as might have been expected for d / h @xmath53 . | studies of limits on active - sterile neutrino mixing derived from big bang nucleosynthesis considerations are extended to consider the dependance of these constraints on the primordial deuterium abundance .
this study is motivated by recent measurements of d / h in quasar absorption systems , which at present yield discordant results .
limits on active - sterile mixing are somewhat relaxed for high d / h . for low d / h ( @xmath0 ) , no active - sterile neutrino mixing is allowed by currently popular upper limits on the primordial @xmath1he abundance @xmath2 .
for such low primordial d / h values , the observational inference of active - sterile neutrino mixing by upcoming solar neutrino experiments would imply that @xmath2 has been systematically underestimated , unless there is new physics not included in standard bbn .
upper limits on the abundance of @xmath1he produced in big bang nucleosynthesis ( bbn ) have been used to limit mixing between active ( @xmath3 , @xmath4 , or @xmath5 ) and sterile ( @xmath6 , no standard model interactions ) neutrinos @xcite . in this paper , we point out and discuss how these constraints are dependant on the adopted primordial deuterium abundance .
previous limits on sterile neutrino mixing have assumed a value for the lower bound on the baryon - to - photon ratio @xmath7 derived from interstellar medium and solar system measurements of deuterium ( d ) and @xmath8he , and models of chemical and galactic evolution .
recent measurements of d / h in quasar absorption systems ( qas ) have yielded discordant values of this ratio , some higher than previously derived ranges @xcite , and some lower @xcite .
several factors make an investigation of the primordial d / h dependance of bbn constraints on active - sterile neutrino mixing timely : the discordant qas measurements of d / h ; the fact that future solar neutrino experiments may be able to distinguish and identify @xmath9 mixing @xcite ; and the use of sterile neutrinos in schemes for neutrino masses and mixings that explain all available data @xcite . as is well known ( _ e.g. _ , ref
@xcite ) , the abundance of @xmath1he produced by bbn is essentially determined by the ratio of neutron to proton number densities ( n / p ) at `` weak freeze - out '' ( wfo ) .
wfo occurs when the reactions that interchange neutrons and protons proceed too slowly relative to the expansion rate of the universe to keep n / p at its equilibrium value of n / p @xmath10 . here
@xmath11 mev is the neutron - proton mass difference , and @xmath12 is the photon temperature . mixing between active and sterile neutrinos increases ( n / p)@xmath13 , and therefore the primordial @xmath1he mass fraction @xmath2 , in two ways .
first , active - sterile neutrino mixing effectively brings more degrees of freedom into thermal contact , increasing the energy density and hence the expansion rate of the universe .
second , active - sterile mixing especially @xmath9 mixing depletes the electron neutrino and antineutrino populations , reducing the rates of the n @xmath14 p interconversion reactions .
both of these effects cause n / p to freeze out at a lower temperature . using a neutrino ensemble evolution formalism @xcite that includes both neutrino oscillations ( with matter effects ) and neutrino collisions , previous authors
@xcite have produced exclusion plots in the @xmath15-@xmath16 plane for both @xmath9 and @xmath17 mixing @xcite . here
@xmath15 and @xmath16 are the difference of the squares of the neutrino vacuum mass eigenvalues and a measure of the vacuum mixing angle , respectively , associated with two - flavor neutrino mixing .
these studies showed that for @xmath18 , both the @xmath17 solution to the atmospheric neutrino problem @xcite and the @xmath9 large - angle msw solution to the solar neutrino problem @xcite are excluded for @xmath19 . in our study of the primordial d / h dependance of bbn constraints on active - sterile neutrino mixing ,
we have employed the same neutrino evolution formalism @xcite as previous authors .
we have neglected any net lepton number contributed by the neutrinos .
( the recently reported effect of active - sterile neutrino mixing _ generating _ net lepton number does not occur in the regions of parameter space we consider here @xcite . ) in this case , the neutrino and antineutrino sectors evolve identically .
a fermi - dirac momentum distribution for all neutrinos is assumed , but allowance is made for non - equilibrium number densities .
the differential equations in the formalism yield @xmath20 , and @xmath12 as functions of time . here
@xmath21 denotes the fraction of a full fermionic degree of freedom contributed by neutrino species @xmath22 , which we shall hereafter call the `` number density parameter '' of neutrino species @xmath22 . in the equations below we will take @xmath23 , since we are working under the assumption that the net lepton number contributed by the neutrinos is negligible . in our bbn computation
we have employed the kawano @xcite update of the wagoner @xcite code , with the latest world average neutron lifetime , @xmath24 s @xcite ; the reaction rates of ref .
@xcite ; and a correction of @xmath25 to @xmath2 due to finite nucleon mass and timestep - dependant effects @xcite .
we have altered the kawano code to use the ` temperature series ' of neutrino number density parameters , @xmath26 , to compute the energy density contributed by neutrinos and the n @xmath14 p interconversion rates .
the neutrino energy density is @xmath27 the n @xmath14 p rates are @xmath28 @xmath29 @xmath30 @xmath31 @xmath32 @xmath33 in these expressions @xmath34 , where @xmath35 and @xmath36 are the total electron ( or positron ) energy and rest mass , respectively ; @xmath37 ; @xmath38 , where @xmath39 is the appropriate neutrino temperature ; @xmath40 ; and @xmath41 is a constant obtained by solving the equation @xmath42 for @xmath41 , where @xmath43 is the experimentally measured neutron lifetime .
the lower limit on @xmath7 obtained from a standard bbn calculation with @xmath44 is not appropriate for bbn with active - sterile mixing .
this is because the lower limit on @xmath7 depends on the expansion rate , often codified as an effective number of neutrino generations @xmath45 @xcite . since active - sterile mixing increases @xmath45 ( at least for the range of parameter space we consider here @xcite ) ,
it affects the lower bound on @xmath7 .
therefore , we will plot our results as a function of the primordial d / h value the experimentally determined quantity rather than as a function of @xmath7 .
these considerations are most important for the @xmath17 atmospheric neutrino mixing solution , and much less important ( nearly negligible ) for the @xmath9 small - angle msw solution to the solar neutrino problem . in fig .
1 , a representative @xmath17 atmospheric neutrino mixing solution ( @xmath46 ev@xmath47 , @xmath48 @xcite ) is _ assumed _ , and the resulting bbn @xmath1he yield is plotted as a function of the bbn d / h yield .
the value of @xmath7 ( given as @xmath49 ) at various values of d / h is also indicated on the figure . for a given value of d / h
, the implied abundance of @xmath1he can be interpreted as the observational upper limit required to constrain the solution .
alternatively , a _ detection _ of these neutrino mixing parameters by , for example , future atmospheric neutrino experiments would yield an independent determination of the primordial @xmath1he abundance , so long as d / h were known from qas studies .
this could be very interesting , given the recent emphasis on the systematic uncertainties in the determination of the primordial @xmath1he abundance as derived from helium recombination lines in extragalactic hii regions @xcite .
the conclusions reached from fig . 1 are essentially the same over the range of @xmath15 ( @xmath50 ev@xmath47 ) for the proposed @xmath17 mixing explanation of the atmospheric neutrino problem .
fig .
2 is similar to fig . 1 , but with a representative @xmath9 small - angle msw solution to the solar neutrino problem assumed ( @xmath51 ev@xmath47 , @xmath52 @xcite ) .
these mixing parameters have only a very small effect on the bbn @xmath1he yield . for the most recent data from solar neutrino experiments and the standard solar model @xcite ,
there is no @xmath9 large - angle msw solution to the solar neutrino problem @xcite .
some of the qas data suggest d / h @xmath53 @xcite .
figs .
1 - 2 show the range d / h @xmath54 , as determined in ref .
@xcite .
this range of d / h implies a lower bound on @xmath7 that is significantly lower than that used in previous studies .
since a lower @xmath7 implies a lower @xmath1he yield , high d / h relaxes constraints on any effect that increases the expansion rate , including mixing with sterile species @xcite .
fig .
1 shows , however , that for d / h @xmath53 , the @xmath17 atmospheric neutrino solution is still somewhat constrained @xcite if current observational inferences @xcite of primordial @xmath1he are correct : @xmath55 , where the first error is statistical and the second systematic .
of course , if this @xmath17 atmospheric neutrino mixing solution were inferred from atmospheric neutrino experiments , and qas studies confirm d / h @xmath53 , the implied @xmath1he abundance of @xmath56 would be significantly higher than the central value of @xmath57 cited above .
the @xmath9 small - angle msw solar neutrino solution is allowed for d / h @xmath53 .
fig .
2 shows that an observational upper bound of @xmath58 would be required to restrict this small angle solution if such a high d / h is indeed the primordial value .
other very high quality qas data arguably better @xcite for the determination of d / h than that used in ref .
@xcite suggest d / h @xmath0 @xcite .
this value of d / h is incompatible with standard bbn with @xmath44 @xcite for current observational inferences of the primordial @xmath1he abundances @xcite , and any mixing with sterile neutrinos would only exacerbate the problem .
as mentioned previously , however , it has been argued that @xmath2 has been systematically underestimated , and a more appropriate upper limit on @xmath2 may actually be @xmath59 @xcite .
it is unlikely that the systematic error in @xmath2 is enough to allow the @xmath17 atmospheric neutrino solution for d / h @xmath0 .
however , observation of the @xmath9 small - angle msw solar neutrino solution , together with a solid determination of d / h @xmath0 , would require that @xmath2 has been systematically and significantly underestimated by about @xmath60 ( see fig .
2 ) , unless there is non - standard physics during the bbn epoch @xcite .
this is a somewhat trivial point , since the mixing parameters of the small - angle @xmath9 msw solution produce only very slightly more @xmath1he than the standard bbn picture with @xmath44 , for which the `` crisis '' at low d / h is well - known @xcite .
useful constraints on the @xmath9 small - angle msw solar neutrino solution would require very precise observational knowledge of @xmath7 and @xmath2 .
this may , however , still be interesting in view of the fact that future solar neutrino experiments may be able to distinguish the sterile neutrino oscillation - based solution from other solutions @xcite . also , many models that seek to satisfy all available constraints on neutrino properties employ the @xmath9 small - angle msw solar neutrino solution @xcite ( but see ref .
@xcite ) . |
the discovery of neutrino masses and mixings through solar @xcite , atmospheric @xcite , and reactor @xcite neutrino oscillation experiments has provided new clues to solving the mysteries of the standard model .
flavor mixing in the lepton - sector , together with the well - known mixing in the quark - sector , may lead to a new understanding of what ` flavor ' is , why there are three generations of fermions , and where the quark- and lepton - mass hierarchies come from .
cp violation in the neutrino - sector could potentially be large enough to account for the matter - antimatter asymmetry in the universe @xcite . due to these possibilities ,
several long - baseline ( lbl ) neutrino oscillation experiments are in progress @xcite with many more being proposed @xcite for the purpose of better determining the neutrino masses and mixing parameters . since the neutrino beams of lbl experiments necessarily traverse the earth , the understanding of matter effects @xcite is crucial in extracting the masses and mixing parameters in vacuum from the measured oscillation probabilities in matter .
the calculation of matter effects requires the diagonalization of the effective hamiltonian in matter , which is an energy- and matter - density - dependent @xmath0 matrix .
this can be done numerically on a computer with ease , but the dependence of the oscillation probabilities on the vacuum parameters is not transparent @xcite .
it is also possible to write down exact analytical expressions for the effective parameters in matter in terms of those in vacuum @xcite , but the expressions are too complicated to be illuminating .
approximate expressions for the oscillation probabilities have also been worked out , but were limited in their range of applicability @xcite , or still too complicated to be of practical use @xcite . in this paper
, we derive simple approximate expressions for the effective mass - squared - differences and mixing angles in matter , in terms of the corresponding parameters in vacuum .
they are simple enough to be used by hand , yet accurate enough in a wide energy range .
this paper is organized as follows . in section ii , we review the formalism of neutrino oscillations to fix our notation , and list the current experimental bounds on the mixing angles and mass - squared differences . in section iii ,
we derive our approximate expressions for the effective mixing angles and mass - squared differences in matter , for both neutrinos and anti - neutrinos . in section
iv , we provide sample calculations to illustrate the accuracy of our approximation . in section
v , we present simple analytical approximations for the oscillation probabilities and demonstrate their utility in understanding which parameters are best constrained at what baseline lengths .
section vi concludes .
assuming three - generation neutrino mixing , the flavor eigenstates @xmath1 @xmath2 are related to the three mass eigenstates @xmath3 @xmath4 via the maki - nakagawa - sakata ( mns ) matrix : @xcite @xmath5 a popular parametrization is given by @xcite @xmath6 with @xmath7 \left [ \begin{array}{ccc } c_{13 } & 0 & s_{13 } e^{-i\delta } \\ 0 & 1 & 0 \\ -s_{13 } e^{i\delta } & 0 & c_{13 }
\end{array } \right ] \left [ \begin{array}{ccc } c_{12 } & s_{12 } & 0 \\ -s_{12 } & c_{12 } & 0 \\ 0 & 0 & 1 \end{array } \right ] \cr & = & \left [ \begin{array}{ccc } c_{12}c_{13 } & s_{12}c_{13 } & s_{13 } e^{-i\delta } \\ -s_{12}c_{23 } - c_{12}s_{13}s_{23}e^{i\delta } & \phantom{-}c_{12}c_{23 } - s_{12}s_{13}s_{23}e^{i\delta } & c_{13}s_{23 } \\
\phantom{-}s_{12}s_{23 } - c_{12}s_{13}c_{23}e^{i\delta } & -c_{12}s_{23 } - s_{12}s_{13}c_{23}e^{i\delta } & c_{13}c_{23 } \end{array } \right ] \;,\cr & & \cr \mathcal{p } & = & \mathrm{diag}(e^{i\alpha_1/2},e^{i\alpha_2/2},1 ) \ ; , \label{upparam}\end{aligned}\ ] ] where @xmath8 , and @xmath9 .
without loss of generality , we can adopt the convention @xmath10 , @xmath11 @xcite .
of the six parameters in this expression and the three masses , which add up to a total of nine parameters , neutrino@xmath12neutrino oscillations are only sensitive to six : * the three mixing angles : @xmath13 , @xmath14 , @xmath15 , * two mass - squared differences : @xmath16 , @xmath17 , where @xmath18 , and * the cp - violating phase : @xmath19 . the majorana phases , @xmath20 and @xmath21 , only appear in lepton - number violating processes such as neutrinoless double beta decay , and can not be determined via neutrino@xmath12neutrino oscillations .
the absolute scale of the neutrino masses also remain undetermined since neutrino oscillation is an interference effect .
if a neutrino of flavor @xmath22 is created at @xmath23 with energy @xmath24 , then the state of the neutrino at @xmath23 is @xmath25 at @xmath26 , the same state is @xmath27 using @xmath28 we find @xmath29 therefore , the amplitude of observing the neutrino of flavor @xmath30 at @xmath26 is given by ( dropping the irrelevant overall phase ) @xmath31 \left[\sum_{i=1}^3 \exp\left(-i\dfrac{m_i^2}{2e}l\right)\ , ( v_{\mathrm{mns}})^*_{\alpha i}{\left| \nu_i \right\rangle}\right ] \cr & = & \sum_{j=1}^3 ( v_{\mathrm{mns}})_{\beta j}\exp\left(-i\dfrac{m_j^2}{2e}l\right)\ , ( v_{\mathrm{mns}})^*_{\alpha j}\;,\end{aligned}\ ] ] and the probability of oscillation from @xmath1 to @xmath32 with neutrino energy @xmath24 and baseline @xmath33 is given by @xmath34 where @xmath35 since @xmath36 only two of the three @xmath37 s in eq .
( [ palphatobeta ] ) are independent . eliminating @xmath38 from eq .
( [ palphatobeta ] ) , we obtain @xmath39 for the @xmath40 case , and @xmath41 for the @xmath42 case , where @xmath43 is the jarskog invariant : @xmath44 in the parametrization given in eq .
( [ upparam ] ) , we have @xmath45 with @xmath46 the oscillation probabilities for the anti - neutrinos are obtained by replacing @xmath47 with its complex conjugate , which only amounts to flipping the sign of @xmath19 in the parametrization of eq .
( [ upparam ] ) .
it is clear from eq .
( [ palphatoalpha ] ) that @xmath48 , which is to be expected from the cpt theorem . for flavor changing oscillations , only the jarskog term in eq .
( [ palphatonotalpha ] ) changes sign .
when @xmath49 , the expressions given above can be expanded in @xmath50 to yield @xmath51 where @xmath52 neglecting terms of order @xmath53 and higher , we obtain the following simplified expressions for a few specific processes : @xmath54 where @xmath55 we have made the identifications @xmath56 where @xmath57 and @xmath58 are the mixing angles extracted from atmospheric @xcite and reactor @xcite neutrino oscillation experiments , respectively , based on two - flavor oscillation analyses . the current experimental bounds from atmospheric neutrinos at the 90% confidence level are @xmath59 only the absolute value of @xmath17 is known since @xmath60 at leading order in @xmath50 .
the 90% confidence limit on @xmath61 from the chooz experiment , which measured @xmath62 , depends on the not - yet - well - known value of @xmath63 : @xmath64 when @xmath65 , @xmath66 simplifies to @xmath67 since @xmath68 is known to be small from chooz , we can identify @xmath13 with the 2-flavor solar mixing angle @xmath69 .
the current experimental bounds at the 90% confidence level from solar neutrinos are @xcite @xmath70 the sign of @xmath71 is known to be positive since @xmath72 is required for the msw effect @xcite to work .
.comparison of the size of the ratio @xmath73 and the 90% confidence limit on @xmath15 . [ cols="^,^,^",options="header " , ]
we would like to thank masafumi koike and masako saito for helpful communications .
takeuchi would like to thank the hospitality of the particle theory group at ochanomizu women s university , where a major portion of this work was carried out during the summer of 2005 .
this research was supported in part by the u.s .
department of energy , grant de
fg0592er40709 , task a ( t.t . ) .
m. b. smy _ et al . _
[ super - kamiokande collaboration ] , phys .
d * 69 * , 011104 ( 2004 ) [ arxiv : hep - ex/0309011 ] . s. n. ahmed _ et al . _ [ sno collaboration ] , phys .
lett . * 92 * , 181301 ( 2004 ) [ arxiv : nucl - ex/0309004 ] . t. araki _ et al . _
[ kamland collaboration ] , phys .
lett . * 94 * , 081801 ( 2005 ) [ arxiv : hep - ex/0406035 ] .
y. ashie _ et al . _
[ super - kamiokande collaboration ] , phys .
* 93 * , 101801 ( 2004 ) [ arxiv : hep - ex/0404034 ] ; y. ashie _ et al . _
[ super - kamiokande collaboration ] , phys .
d * 71 * , 112005 ( 2005 ) [ arxiv : hep - ex/0501064 ] .
m. apollonio _ et al . _
[ chooz collaboration ] , phys .
b * 466 * , 415 ( 1999 ) [ arxiv : hep - ex/9907037 ] ; m. apollonio _ et al . _
[ chooz collaboration ] , eur .
j. c * 27 * , 331 ( 2003 ) [ arxiv : hep - ex/0301017 ] .
m. fukugita and t. yanagida , phys . lett .
b * 174 * , 45 ( 1986 ) .
m. h. ahn _
[ k2k collaboration ] , phys .
* 93 * , 051801 ( 2004 ) [ arxiv : hep - ex/0402017 ] , e. aliu _ et al .
_ [ k2k collaboration ] , phys . rev .
lett . * 94 * , 081802 ( 2005 ) [ arxiv : hep - ex/0411038 ] , c. mariani [ k2k collaboration ] , arxiv : hep - ex/0505019 .
g. rosa [ opera collaboration ] , nucl .
suppl . * 145 * , 98 ( 2005 ) , h. pessard [ opera collaboration ] , arxiv : hep - ex/0504033 . g. s. tzanakos [ minos collaboration ] , aip conf .
proc . * 721 * , 179 ( 2004 ) .
d. ayres _
[ nova collaboration ] , arxiv : hep - ex/0210005 , d. s. ayres _
[ nova collaboration ] , arxiv : hep - ex/0503053 , y. itow _ et al .
_ , arxiv : hep - ex/0106019 ; y. hayato [ t2k collaboration ] , nucl .
* 143 * , 269 ( 2005 ) , k. kaneyuki [ t2k collaboration ] , nucl .
phys . proc .
suppl . * 145 * , 178 ( 2005 ) , y. hayato [ t2k collaboration ] , nucl . phys . proc .
suppl . * 147 * , 9 ( 2005 ) .
see also the jhf neutrino working group s home page , http://neutrino.kek.jp / jhfnu/. m. v. diwan _ et al .
_ , phys .
d * 68 * , 012002 ( 2003 ) [ arxiv : hep - ph/0303081 ] .
n. okamura , m. aoki , k. hagiwara , y. hayato , t. kobayashi , t. nakaya and k. nishikawa , arxiv : hep - ph/0104220 ; m. aoki , k. hagiwara , y. hayato , t. kobayashi , t. nakaya , k. nishikawa and n. okamura , phys .
d * 67 * , 093004 ( 2003 ) [ arxiv : hep - ph/0112338 ]
. h. s. chen _ et al .
_ [ vlbl study group h2b-1 ] , arxiv : hep - ph/0104266 , y. f. wang , k. whisnant , z. h. xiong , j. m. yang and b. l. young [ vlbl study group h2b-4 ] , phys . rev .
d * 65 * , 073021 ( 2002 ) [ arxiv : hep - ph/0111317 ] . h. minakata and h. nunokawa , phys .
b * 413 * , 369 ( 1997 ) [ arxiv : hep - ph/9706281 ] ; v. barger , d. marfatia and k. whisnant , phys .
b * 560 * , 75 ( 2003 ) [ arxiv : hep - ph/0210428 ] .
m. ishitsuka , t. kajita , h. minakata and h. nunokawa , phys .
d * 72 * , 033003 ( 2005 ) [ arxiv : hep - ph/0504026 ] .
k. hagiwara , n. okamura and k. i. senda , arxiv : hep - ph/0504061 . o. mena requejo , s. palomares - ruiz and s. pascoli , phys . rev .
d * 72 * , 053002 ( 2005 ) [ arxiv : hep - ph/0504015 ] ; o. mena and s. parke , phys . rev .
d * 72 * , 053003 ( 2005 ) [ arxiv : hep - ph/0505202 ] ; o. mena , s. palomares - ruiz and s. pascoli , arxiv : hep - ph/0510182 .
l. wolfenstein , phys .
d * 17 * , 2369 ( 1978 ) ; r. r. lewis , phys . rev .
d * 21 * , 663 ( 1980 ) ; v. d. barger , k. whisnant , s. pakvasa and r. j. n. phillips , phys .
d * 22 * , 2718 ( 1980 ) ; s. p. mikheyev and a. yu . smirnov , yad .
* 42 * , 1441 ( 1985 ) , [ sov .
j. nucl .
phys . * 42 * , 913 ( 1986 ) ] ; s. p. mikheyev and a. yu .
smirnov , nuovo cimento * c9 * , 17 ( 1986 ) . p. lipari , phys .
d * 61 * , 113004 ( 2000 ) [ arxiv : hep - ph/9903481 ] ; v. d. barger , s. geer , r. raja and k. whisnant , phys .
b * 485 * , 379 ( 2000 ) [ arxiv : hep - ph/0004208 ] . k. kimura , a. takamura and h. yokomakura , j. phys .
g * 29 * , 1839 ( 2003 ) ; k. kimura , a. takamura and h. yokomakura , phys .
d * 66 * , 073005 ( 2002 ) [ arxiv : hep - ph/0205295 ] ; k. kimura , a. takamura and h. yokomakura , phys .
b * 537 * , 86 ( 2002 ) [ arxiv : hep - ph/0203099 ] .
j. arafune , m. koike and j. sato , phys . rev .
d * 56 * , 3093 ( 1997 ) [ erratum - ibid .
d * 60 * , 119905 ( 1999 ) ] [ arxiv : hep - ph/9703351 ] .
m. freund , phys .
d * 64 * , 053003 ( 2001 ) [ arxiv : hep - ph/0103300 ] .
z. maki , m. nakagawa and s. sakata , prog .
* 28 * , 870 ( 1962 ) .
`` neutrino mass , mixing , and flavor change '' , b. kayser , in the _ review of particle physics _
lett . b**592 * * ( 2004 ) 1 .
f. ardellier _ et al . _
[ double - chooz collaboration ] , arxiv : hep - ex/0405032 ; f. dalnoki - veress [ double - chooz collaboration ] , arxiv : hep - ex/0406070 . f. suekane [ kaska collaboration ] , arxiv : hep - ex/0407016 . v. barger , d. marfatia and k. whisnant , phys . rev .
d * 65 * , 073023 ( 2002 ) [ arxiv : hep - ph/0112119 ] .
m. aoki , k. hagiwara and n. okamura , phys .
b * 554 * , 121 ( 2003 ) [ arxiv : hep - ph/0208223 ] .
k. hagiwara , nucl .
* 137 * , 84 ( 2004 ) [ arxiv : hep - ph/0410229 ] .
h. minakata and h. nunokawa , jhep * 0110 * , 001 ( 2001 ) [ arxiv : hep - ph/0108085 ] .
m. honda , n. okamura , and t. takeuchi , in preparation . | we present simple analytical approximations to matter - effect corrected effective neutrino mixing - angles and effective mass - squared - differences .
the expressions clarify the dependence of oscillation probabilities in matter to the mixing angles and mass - squared - differences in vacuum , and are useful for analyzing long - baseline neutrino oscillation experiments . |
the centimeter continuum radiation from classic hii regions is dominated by strong free - free emission from the extended ionized gas present there . however , when observed with the high angular resolution provided by an interferometer , the extended emission is filtered out and one starts to detect compact sub - arcsecond sources of various natures ( see garay et al .
1987 ; churchwell et al .
1987 ; felli et al . 1993 ; zapata et al .
2004 for the case of orion a ) .
the brightest of these sources are the hypercompact ( hc ) hii regions , that trace the ionized gas produced by young ob stars still embedded in dense molecular gas ( e.g. kurtz 2005 ; lizano 2008 ) .
the externally ionized globules are also sources of free - free radiation and result from the interaction of the uv photons of the ob stars in the region with remaining blobs of neutral gas existing inside the hii region ( e.g. garay et al .
1987 ) . the proplyds ( odell et al . 1993 ) are similar to the externally ionized globules , but in this case the object being ionized is a protoplanetary disk surrounding a young star .
the last two known types of free - free emitters are the jets emanating from accreting protostars ( anglada 1996 ; eisloffel et al . 2000 ) and the spherical ionized winds produced by massive stars ( e.g. bieging et al . 1989 ; contreras et al .
there are also two types of sources with non - thermal radio continuum emission .
young low - mass stars can have strong magnetospheric activity and emit detectable gyrosynchrotron radiation ( andr et al .
1988 ) . finally , there is also strong evidence that systems of massive binary stars can produce synchrotron radiation in the region where their winds collide ( e.g. pittard et al .
2006 ; ortiz - len et al .
2011 ) . in table 1
we present a summary of the characteristics of these different types of compact radio sources .
in this paper we present a sensitive , high angular resolution study made with the jansky very large array ( jvla ) of the national radio astronomy observatory ( nrao ) toward the classic hii region m17 ( the omega nebula , ngc 6618 , w38 , s45 ) . in 2 we present the observations , while in 3 we list and briefly discuss the compact radio sources detected . in
4 we use our data to present an explanation for the spectral index of order 1 observed in the hypercompact hii region m17 uc1 . in 5 we discuss the time variable sources in our sample , while in 6 we concentrate on cen 1a and 1b , the members of the massive binary system that ionizes most of m17 . in 7 we try to model the time - variable emission of cen 1a in terms of a thermal model , concluding that this is not feasible .
finally , 8 presents a brief discussion on some of the other individual sources and in 9 we summarize our conclusions .
the jvla observations were made at 4.96 , 8.46 and 22.46 ghz using two intermediate frequency ( if ) bandwidths of 128 mhz each , separated by 128 mhz , and containing both circular polarizations .
each if was split into 64 channels of 2 mhz each . for the continuum images we averaged the central 54 channels , rejecting five channels at each end of the bandwidth .
we observed in three epochs during the year 2011 : june 16 ( 2011.46 ) and 27 ( 2011.49 ) , and july 28 ( 2011.57 ) .
this cadence was adopted with the purpose of searching for variability at 4.96 and 8.46 ghz in the timescales of about 10 , 30 , and 40 days .
the observations at 22.46 ghz were made only in the second epoch ( 2011 june 27 ) with the purpose of determining better spectral indices .
in addition to these timescales , we searched for rapid variability ( i.e. in timescales of minutes ) during the two hours of each individual run .
at all 3 epochs , the jvla was in the highest angular resolution a configuration .
the data reduction was made using the software package aips of nrao , following the recommendations for jvla data given in appendix e of its cookbook ( that can be found in http://www.aips.nrao.edu/cook.html ) .
the amplitude calibrator was j1331 + 3030 , with adopted flux densities of 7.392 , 5.205 and 2.519 jy at 4.96 , 8.46 and 22.46 ghz , respectively .
the phase calibrator was j1832 - 1035 , with the bootstrapped flux densities given in table 2 .
the data were self - calibrated in amplitude and phase and we made images using only visibilities with baselines larger than 50 k@xmath1 , suppressing structures larger than @xmath2 .
the search for variability within each individual run was made without self - calibration , since this procedure tends to smooth out amplitude variations within the time interval where it is applied . at 4.96 and 8.46
ghz we analyzed regions of @xmath3 in solid angle , while at 22.46 ghz the region analyzed was @xmath4 . at 22.46 ghz
our 4-@xmath5 sensitivity at the center of the field was @xmath00.4 mjy beam@xmath6 and the only source detected was m17 uc1 , a well - known hypercompact hii region embedded in the molecular cloud adjacent to the sw of the m17 hii region ( felli et al .
1980 ) . at 4.96 and/or 8.46 ghz we detected a total of 38 compact sources , in addition to m17 uc1 .
these sources are listed in table 3 .
the rms noise of the images is not constant across all the solid angle analyzed for two reasons .
first , the primary beam response decreases as we move away from the center of the field ( e.g. cotton & perley 2010 ) .
the correction for this effect increases both signal and noise .
in addition , in this region there is an arc - shaped ionization front to the east and northeast of m17 uc1 that makes it noisier than expected ( felli et al .
1984 ; johnson et al . 1998 ) .
this structure most probably traces the interaction of the ionizing photons of ngc 6618 , the cluster ionizing m17 , with the molecular clump where m17 uc1 is embedded . assuming a typical noise of 0.05 mjy beam@xmath6 at the center of the field for the 8.46 ghz observations and following anglada et al .
( 1998 ) , we expect to detect only about one background source above 4-@xmath5 over all the solid angle analyzed .
we can then conclude that practically all the detected sources are related to m17 .
in table 3 we list the 38 detected compact radio sources , giving their positions , flux densities at 8.46 and 4.96 ghz , the resulting spectral index , and their angular size .
most of the sources are unresolved ( @xmath7 ) . in table 4
we list the 19 compact radio sources with reported counterparts at other wavelengths ( in all cases , stars detected previously at optical , near - infrared , or x - rays ) .
a counterpart was considered to be associated with the jvla source if its position coincided within @xmath8 of the radio position .
we tried to find systematic differences between the radio sources with and without counterparts by plotting them in a spectral index versus flux density diagram ( see fig .
1 ) . this diagram does not segregate the two classes of source and we tentatively suggest that the radio sources without counterparts are similar to those that have a counterpart , but are very heavily obscured which makes detection difficult at wavelengths other than radio .
this bright source has an angular size of @xmath9 , that at a distance of 1.98 kpc ( xu et al . 2011 ) implies a diameter of @xmath00.004 pc , falling in the category of the hypercompact hii regions ( kurtz 2005 ) .
its morphology is cometary . for this source
we obtain total flux densities of 44 , 107 , and 194 mjy at 4.96 , 8.46 , and 22.46 ghz , respectively .
these flux densities indicate a spectral index of @xmath10 ( @xmath11 ) , consistent with the value of @xmath12 typically found in hypercompact hii regions ( hofner et al .
1996 ; kurtz 2002 ; sewilo et al .
the reason for this intermediate spectral index remains unknown .
homogeneous hii regions are expected to show @xmath13 when optically thick ( low frequencies ) and @xmath14 when optically thin ( high frequencies ) . in these sources
the transition from optically - thick to optically - thin ocurrs in a small frequency range .
the indices of @xmath12 observed in hypercompact hii regions can not be explained as the result of observing this transition region since they extend over at least an order of magnitude in frequency .
an alternative explanation is given by the hierarchical clumping model of ignace & churchwell ( 2004 ) , where it is proposed that the radio continuum energy spectrum comes from an ensemble of spherical clumps . in this model , for example , the spectral index of @xmath15 should be detected all across the face of the nebula .
taking advantage of the excellent quality of our 8.46 ghz data , we derived the spectral index for m17 uc1 as a function of position
. our 4.96 ghz data does not have enough angular resolution to clearly resolve the source and our 22.46 ghz data is centered on the ngc 6618 cluster and m17 uc1 is located too far from the phase center and the integration is relatively short ( a total of 20 minutes on - source ) .
we then used for the comparison with our 8.46 ghz jvla data , archive vla data taken at 22.49 ghz in 1982 march 20 under project fell . for these archive vla observations the amplitude calibrator was 1331 + 305 , with an adopted flux density of 2.52 jy and the phase calibrator was 1730@xmath16130 , with a bootstrapped flux density of [email protected] jy . to obtain a reliable astrometry ,
the position of 1730@xmath16130 was updated to the most recently improved value provided by the jvla calibrator manual . both the 8.46 and 22.46 ghz observations were made in the a configuration . to obtain comparable ( u
, v ) coverage at both frequencies we did not include the two inner antennas at each arm for the 8.46 ghz observations and the three outer antennas at each arm for the 22.46 ghz observations .
the resulting ( u , v ) coverages are shown in figure 2 .
the images at both frequencies were restored using the synthesized beam of the 8.46 ghz observations . in figure 3
we show the contour images at 8.46 and 22.49 ghz as well as the derived spectral index from these two images .
the spectral index image shows that there is a smooth gradient from values of @xmath18 ( optically - thick emission ) near the head of the cometary nebula to values of @xmath19 ( optically - thin emission ) near the tail .
this result suggests that , at least in this hypercompact hii region the intermediate value of @xmath01 for the spectral index comes from a gradient in optical depth across the face of the nebula .
we speculate that if this is the explanation for the spectral index found in other hypercompact hii regions , they should show a cometary morphology when observed with sufficiently high angular resolution . for this effect to be observable at frequencies similar to those used here , we need objects with large average emission measures , of order @xmath20 @xmath21 pc , so that they can present significant optical depths at frequencies of order 15 ghz .
of the 38 compact sources detected , two of them show large variations between the three observing runs . the first time variable source is cen 1a , whose flux densities are listed in table 5 .
images at the three epochs are shown in fig .
4 . this source was not detected in the first two epochs but in the last epoch it became quite bright , with a flux density about an order of magnitude larger than the 4-@xmath5 upper limits obtained for the first two epochs .
the other variable source was the b1 star ngc 6618 b 159 , whose flux densities are given in table 6 .
in addition to possible long term variability between runs , we also searched for short timescale variability within the individual runs .
our procedure to search for such variability was the following .
first , images of the source of interest were made with the aips task imagr for each observation epoch . with the aips task jmfit ,
the precise position of the primary component was determined .
the @xmath22 data were recentered at the position of the main component using the aips task uvfix . with these new @xmath22 data ,
the real and imaginary parts of the interferometer data were plotted as a function of time , averaged over the @xmath22 plane .
the real part gives us information on the flux density of the source and the imaginary part on its position and symmetry .
we then averaged over time in bins of approximately 10 minutes .
a detailed description of this technique is given by neria et al .
the flux density as a function of time for the two time variable sources during a given run is shown in figures 5 ( cen 1a ) and 6 ( ngc 6618 b 159 ) .
no fast variability ( hours or less ) was detected for these sources . finally , of the seven sources detected by rodrguez et al .
( 2009 ) , only one is not detected by us .
this source is named vla 2 by rodrguez et al .
( 2009 ) and is located at the position @xmath23 .
rodrguez et al .
( 2009 ) reported an 8.46 ghz flux density of [email protected] mjy , while from the images made from the concatenated ( u , v ) data we set a 4-@xmath5 upper limit of 0.07 mjy .
as noted before , also cen 1a is time - variable .
the remaining five sources detected by rodrguez et al .
( 2009 ) were detected by us at similar flux densities .
cen 1a and cen 1b are a massive binary system ( with angular separation of @xmath24 ; chini et al .
the o4 v stars that form this binary system are the main ionizing sources of the m17 h ii region ( e. g. hoffmeister et al .
2008 ) and the brightest x - ray objects in the zone ( broos et al .
cen 1a and cen 1b were first detected as radio sources by rodrguez et al .
our data provides new information on these radio sources .
while cen 1a is strongly time - variable ( see table 4 ) , cen 1b appears to be a steady source .
the flux densities found by us for 2011 june - july are consistent with the flux densities reported by rodrguez et al .
( 2009 ) for 1988 october and 2000 december .
the available information suggests that cen 1b is constant over timescales of decades .
its spectral index , as derived from the 2011 june - july observations is @xmath25 , suggesting optically - thin synchrotron emission . in constrast ,
cen 1a is strongly variable .
as noted above , it was detected only in the last epoch of our observations , showing a spectral index of @xmath26 . as was pointed out by rodrguez et al .
( 2009 ) , the flux density expected from a stellar wind in a single star scenario is too low ( @xmath0 0.1 mjy at 8.46 ghz ) to account for the relatively large flux density detected ( @xmath02.6 mjy at 8.46 ghz ) .
this fact and the strong variability observed suggest a binary origin for the source , with the emission arising from a wind - wind colliding region ( wcr ) between the stars forming the binary system . in this binary scenario ,
the variability is thought to be the result of an eccentric orbit and/or a high inclination angle of the orbital plane . in this context , the spectral index observed could be compatible with both optically - thin thermal emission ( pittard 2010 ) and non - thermal ( synchrotron ) emission ( contreras et al .
1997 ) , arising both from the wcr .
its flat value is in agreement with synchrotron emission being affected by free - free absorption from the unshocked winds , turning the negative value expected from intrinsic optically - thin synchrotron emission into the flat spectrum observed for cen 1a .
the thermal contribution from a wcr is expected to increase the flux density only a factor of a few with respect to that from the single stellar wind ( stevens 1995 , pittard 2010 ) .
on the other hand , the non - thermal emission has been observed to increase the flux by more than one order of magnitude with respect to the minimum states of the radio light curves of colliding wind binaries ( cwb ; e.g. hd 168112 ; blomme et al .
thus , the strong increase in the flux observed in cen 1a suggests a non - thermal origin for the emission detected
. however , given that the spectral index is consistent with optically - thin free - free emission , we have considered thermal emission models , but can not find any that are reasonable and match the observations .
we conclude that the observed radio emission from cen 1a most probably has a non - termal nature and that the observed flat spectrum is consistente with a combination of synchrotron emission and free - free absorption , as noted before .
non - thermal emission in massive stars is usually attributed to them being binary systems , with the emission coming from the wind collision region .
alternatively , the radio emission could be coming from fermi acceleration in wind shocks of a single star due to the radiation - driven instability ( lucy & white 1980 ; white & chen 1994 ) .
this system provides the opportunity to study simultaneously two stars , a variable one ( cen 1a ) and a steady one ( cen 1b ) and to try to understand what is the reason for their different behaviours .
this radio source is within a few arcsec of the bright infrared object kw ( kleinmann & wright 1973 ; povich et al . 2009 ) and it is probably associated with it .
this source is remarkable because it is clearly detected at 8.46 ghz , but not at 4.96 ghz ( see figure 7 ) , resulting in a spectral index of @xmath27 .
this spectral index is consistent with optically - thick free - free emission , optically - thick synchrotron or even with optically thin dust emission .
the source is arc - shaped ( see figure 7 ) and could be tracing the edge of a heated dust globule .
there are no known counterparts to this source in the simbad database .
we also searched unsuccessfully for a counterpart in the glimpse catalog of candidate young stellar objects with a high probability of association with the m17 complex ( povich et al . 2009 ) .
its unusual radio spectrum deserves further study .
we presented sensitive , high angular resolution observations made with the jansky very large array ( jvla ) at 4.96 and 8.46 ghz toward the hii region m17 .
our main conclusions are listed in what follows .
we detected 38 compact radio sources , practically all associated with the region .
only 19 of these sources have counterparts , in all cases stars previously detected at infrared , optical , or x - rays .
we argue that the radio sources without counterparts are similar to those with counterparts , but that they are heavily obscured , which makes detection difficult at wavelengths other than radio .
we studied the spectral index of the hypercompact hii region m17 uc1 , finding that its spectral index of value @xmath01 is due to a gradient in optical depth across its face .
we speculate that this type of gradient could explain similar spectral indices detected in other hypercompact hii regions . in this case
, all these sources should show a cometary morphology .
we discussed the nature of the radio emission from the massive binary system cen 1a and 1b , concluding that both are most probably non - thermal emitters , although the first one is highly time variable and the second one is steady . non - thermal emission in massive stars is usually attributed to them being binary systems , with the emission coming from the wind collision region .
this system provides the opportunity to study simultaneously two stars , a variable one ( cen 1a ) and a steady one ( cen 1b ) and to try to understand the reasons for their different behaviours .
, a.c.r . , and l.f.r .
acknowledge the financial support of dgapa , unam and conacyt , mxico .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
( 8.46 ghz ) and 0.45 mjy beam@xmath6 ( 22.49 ghz ) , the respective rms noise of each image .
( bottom ) color image of the spectral index as a function of position , as derived from the 8.46 and 22.49 ghz images .
the color bar on top of this image gives the coding for the spectral index .
the half power contour of the beam is @xmath28 with @xmath29 and it is shown in the bottom left corner of the images . [ spxk],scaledwidth=100.0% ] 3 , 3 , 6 , 10 , 15 , 20 and 30 times 72 @xmath30jy beam@xmath6 , the average rms noise of the images .
the half power contour of the synthesized beams is shown in the bottom left corner of each panel .
( right column ) same as in the left column , but for the frequency of 8.46 ghz .
contours are given times 57 @xmath30jy beam@xmath6 , the average rms noise of the images .
[ cen1ab],scaledwidth=100.0% ] mjy .
the imaginary component is consistent with zero , indicating that the source is symmetric about the phase center ( the origin of the visibility plane ) and has no significant structure on these spatial scales .
[ fig_timex],scaledwidth=100.0% ] mjy .
the imaginary component is consistent with zero , indicating that the source is symmetric about the phase center ( the origin of the visibility plane ) and has no significant structure on these spatial scales .
[ fig_timexb159],scaledwidth=100.0% ] , 3 , 4 , 5 , and 6 times 30@xmath30jy beam@xmath6 .
the source is clearly detected at 8.46 ghz , but not at 4.96 ghz .
the half power contour of the synthesized beam is shown in the bottom left corner of the images and is @xmath31 with @xmath32 for the 4.96 ghz image and @xmath33 with @xmath34 for the 8.46 ghz image .
[ jvla35all],scaledwidth=50.0% ] ccccccc class of & emission & spectral & & time & & + source & mechanism & index & polarization & variability & morphology & excitation + hc hii region & free - free & @xmath01 & no & no & various & internal + ionized globule & free - free & @[email protected] & no & no & cometary & external + proplyds & free - free & @[email protected] & no & no & cometary & external + jet & free - free & @xmath00.6 & no & yes & elongated & internal + spherical wind & free - free & @xmath00.6 & no & no & unresolved & internal + low - mass protostars & gyrosynchrotron & @xmath162 + 2 & circular & yes & unresolved & internal + massive binary stars & synchrotron & @[email protected] & linear & yes & cometary & internal + cccc frequency & 2011 jun 16 & 2011 jun 27 & 2011 jul 28 + ( ghz ) & ( jy ) & ( jy ) & ( jy ) + 4.96 & [email protected] & [email protected] & [email protected] + 8.46 & [email protected] & [email protected] & [email protected] + 22.46 & & [email protected] & + cccccccc 1 & 18 20 13.335 & -16 12 50.69 & & @[email protected] & [email protected] & @[email protected]@xmath170.5 & u + 2 & 18 20 17.575 & -16 12 21.52 & & [email protected] & [email protected] & @[email protected] & @xmath36 + 3 & 18 20 19.475 & -16 13 29.87 & & [email protected] & @[email protected] & @xmath37 + [email protected] & u + 4 & 18 20 21.435 & -16 12 06.81 & & [email protected] & [email protected] & @[email protected] & u + 5 & 18 20 21.637 & -16 11 17.91 & & [email protected] & [email protected] & + [email protected] & u + 6 & 18 20 22.369 & -16 12 04.95 & & [email protected] & [email protected] & + [email protected] & u + 7 & 18 20 25.317 & -16 11 48.74 & & [email protected] & [email protected] & + [email protected] & @xmath38 + 8 & 18 20 25.496 & -16 10 53.77 & & [email protected] & [email protected] & + [email protected] & u + 9 & 18 20 25.297 & -16 11 39.83 & & [email protected] & [email protected] & + [email protected] & @xmath39 + 10 & 18 20 26.239 & -16 11 15.85 & & [email protected] & [email protected] & + [email protected] & u + 11 & 18 20 26.745 & -16 07 21.82 & & [email protected] & [email protected] & + [email protected] & @xmath36 + 12 & 18 20 27.377 & -16 11 38.59 & & [email protected] & [email protected] & @[email protected] & u + 13 & 18 20 27.855 & -16 09 55.60 & & [email protected] & [email protected] & @[email protected] & u + 14 & 18 20 28.022 & -16 11 27.00 & & [email protected] & [email protected] & + [email protected] & u + 15 & 18 20 28.233 & -16 13 17.83 & & [email protected] & [email protected] & @[email protected] & u + 16 & 18 20 28.283 & -16 11 30.53 & & [email protected] & [email protected] & @[email protected] & u + 17 & 18 20 28.384 & -16 10 14.13 & & [email protected] & [email protected] & + [email protected] & u + 18 & 18 20 29.118 & -16 08 58.61 & & [email protected] & [email protected] & + [email protected] & @xmath39 + 19 & 18 20 29.436 & -16 10 49.84 & & [email protected] & [email protected] & + [email protected] & u + 20 & 18 20 29.811 & -16 10 45.58 & & [email protected] & [email protected] & @[email protected] & u + 21@xmath40 & 18 20 29.897 & -16 10 44.45 & & [email protected] & [email protected] & + [email protected] & u + 22 & 18 20 30.003 & -16 10 35.26 & & [email protected] & [email protected] & + [email protected] & u + 23 & 18 20 30.131 & -16 10 39.45 & & [email protected] & [email protected] & @[email protected] & u + 24 & 18 20 30.323 & -16 10 50.65 & & [email protected] & [email protected] & @[email protected] & u + 25 & 18 20 30.342 & -16 11 43.67 & & [email protected] & @[email protected] & @xmath37 + [email protected] & u + 26 & 18 20 30.415 & -16 11 03.63 & & [email protected] & [email protected] & + [email protected] & u + 27 & 18 20 30.442 & -16 10 53.08 & & [email protected] & [email protected] & @[email protected] & u + 28 & 18 20 30.577 & -16 11 04.13 & & [email protected] & [email protected] & + [email protected] & u + 29 & 18 20 30.632 & -16 10 28.51 & & [email protected] & [email protected] & + [email protected] & u + 30 & 18 20 30.777 & -16 10 59.35 & & [email protected] & [email protected] & + [email protected] & u + 31@xmath40 & 18 20 31.112 & -16 09 29.83 & & [email protected] & [email protected] & + [email protected] & u + 32 & 18 20 31.523 & -16 10 32.56 & & [email protected] & [email protected] & + [email protected] & u + 33 & 18 20 32.981 & -16 11 15.55 & & [email protected] & [email protected] & + [email protected] & u + 34 & 18 20 33.060 & -16 11 21.59 & & [email protected] & [email protected] & @[email protected] & u + 35 & 18 20 33.090 & -16 09 40.85 & & [email protected] & @[email protected] & @[email protected] & @xmath41 + 36 & 18 20 33.504 & -16 10 44.57 & & [email protected] & [email protected] & + [email protected] & u + 37 & 18 20 33.609 & -16 10 21.66 & & [email protected] & [email protected] & @[email protected] & u + 38 & 18 20 33.767 & -16 10 46.39 & & [email protected] & [email protected] & @[email protected] & u + cccc jvla & & + number & counterpart@xmath42 & @xmath43(2000 ) &
@xmath44(2000 ) + 3 & 2mass j18201947 - 1613298 & 18 20 19.472 & -16 13 29.89 + 4 & cl * ngc 6618 b 335 & 18 20 21.435 & -16 12 06.85 + 5 & cl * ngc 6618 b 331 & 18 20 21.640 & -16 11 18.00 + 8 & cl * ngc 6618 b 267 & 18 20 25.500 & -16 10 53.77 + 13 & cl * ngc 6618 b 228 & 18 20 27.852 & -16 09 56.04 + 15 & [ bft2007 ] 471 & 18 20 28.260 & -16 13 17.91 + 16 & cl * ngc 6618 b 222 & 18 20 28.300 & -16 11 31.19 + 19 & cxou j182029.4 - 161050 & 18 20 29.439 & -16
10 50.22 + 20 & cen 1b = [ pcb2009 ] 2 & 18 20 29.811 & -16 10 45.67 + 21 & cen 1a = [ pcb2009 ] 1 & 18 20 29.900 & -16 10 44.40 + 22 & cxou j182030.0 - 161034 & 18 20 30.030 & -16 10 34.81 + 23 & cxou j182030.0 - 161039 & 18 20 30.080 & -16 10 39.97 + 25 &
cl * ngc 6618 b 179 & 18 20 30.346 & -16 11 43.69 + 27 & cl * ngc 6618 b 174 & 18 20 30.448 & -16 10 53.09 + 28 & 2mass j18203057 - 1611040 & 18 20 30.578 & -16 11 04.10 + 29 & sstglmc g015.0618 - 00.6890 & 18 20 30.649 & -16 10 28.25 + 31 & cl * ngc 6618 b 159 & 18 20 31.115 & -16 09 29.88 + 32 & 2mass j18203152 - 1610326 & 18 20 31.523 & -16 10 32.61 + 34 & cl * ngc 6618 b 137 & 18 20 33.061 & -16 11 21.56 + ccccc & frequency & 2011 jun 16 & 2011 jun 27 & 2011 jul 28 + source & ( ghz ) & ( mjy ) & ( mjy ) & ( mjy ) + cen 1a & 4.96 & @[email protected] & @[email protected] & [email protected] + cen 1a & 8.46 & @[email protected] & @[email protected] & [email protected] + cen 1a & 22.46 & & @[email protected] & + & & & & + cen 1b & 4.96 & [email protected] & [email protected] & [email protected] + cen 1b & 8.46 & [email protected] & [email protected] & [email protected] + cen 1b & 22.46 & & @[email protected] & + cccc frequency & 2011 jun 16 & 2011 jun 27 & 2011 jul 28 + ( ghz ) & ( mjy ) & ( mjy ) & ( mjy ) + 4.96 & [email protected] & [email protected] & @[email protected] + 8.46 & [email protected] & [email protected] & @[email protected] + 22.46 & & @[email protected] & + | the classic hii region m17 is one of the best studied across the electromagnetic spectrum .
we present sensitive , high angular resolution observations made with the jansky very large array ( jvla ) at 4.96 and 8.46 ghz that reveal the presence of 38 compact radio sources , in addition to the well known hypercompact cometary hii region m17 uc1 . for this last source we find that its spectral index of value @xmath01 is due to a gradient in opacity across its face . of
the 38 compact radio sources detected , 19 have stellar counterparts detected in the infrared , optical , or x - rays .
finally , we discuss the nature of the radio emission from the massive binary system cen 1a and 1b , concluding that both are most probably non - thermal emitters , although the first is strongly time variable and the second is steady . |
the transition - metal perovskite oxides @xmath2o@xmath1 ( with @xmath3@xmath4 y , la , or other trivalent rare - earth ion , and @xmath5@xmath4 ti or v ) are regarded as some of the key materials for understanding the strong coupling among spin , orbital , and lattice degrees of freedom in correlated electron systems.@xcite according to the electronic structure calculations in the local - density approximation ( lda ) , all these compounds can be classified as `` @xmath0 systems '' , as all of them have a common transition - metal @xmath0-band , located near the fermi level , which is well separated from the oxygen-@xmath6 band and a hybrid transition - metal @xmath7 and either y(@xmath8 ) or la(@xmath9 ) band , located correspondingly in the lower- and upper - part of the spectrum ( fig .
[ fig.dossummary ] ) .
the number of electrons that are donated by each ti and v site into the @xmath0-band is correspondingly one and two .
these electrons are subjected to the strong intraatomic coulomb repulsion , which is not properly treated by lda and requires some considerable improvement of this approximation , which currently processes in the direction of merging lda with various model approaches for the strongly - correlated systems.@xcite nevertheless , lda continues play an important role for these systems as it naturally incorporates into the model analysis the effects of the lattice distortion , and does it formally without any adjustable parameters .
although the origin of the lattice distortion in the @xmath0 perovskite oxides is not fully understood , is is definitely strong and exhibits an appreciable material - dependence , which can be seen even visually in fig .
[ fig.structure ] .
the interplay of this lattice distortion with the coulomb correlations seems to be the key factor for understanding the large variation of the magnetic properties among the @xmath0 perovskite oxides .
the difference exists not only between ti- and v - based compounds , but also within each group of formally isovalent materials , depending on whether it is composed of the y or la atoms .
the latter seems to be a clear experimental manifestation of the distortion effect , which is related with the difference of the ionic radii of y and la .
all together this leads to the famous phase diagram of the distorted @xmath0 perovskite oxides , where each material exhibits quite a distinct magnetic behavior : ytio@xmath1 is a ferromagnet;@xcite latio@xmath1 is a three - dimensional ( g - type ) antiferromagnet;@xcite yvo@xmath1 has the low - temperature g - type antiferromagnetic ( afm ) phase , which at around @xmath10 k transforms into a chain - like ( c - type ) antiferromagnetic phase;@xcite and lavo@xmath1 is the c - type antiferromagnet.@xcite on the theoretical side , the large variety of these magnetic phases has been intensively studied using model approaches ( refs . ) as well as the first - principles electronic structure calculations ( refs . ) .
the problem is still far from being understood , and remains to be the subject of numerous contradictions and debates .
surprisingly that at present there is no clear consensus not only between model and first - principles electronic structure communities , but also between researchers working in each of these groups . presumably , the most striking example is latio@xmath1 , where in order to explain the experimentally observed g - type afm ground state , two different models , which practically exclude each other , have been proposed .
one is the model of orbital liquid , _ which implies the degeneracy of the atomic @xmath0 levels in the crystalline environment_.@xcite another model is based on the theory of crystal - field ( cf ) splitting , _ which lifts the orbital degeneracy _ and leads to the one particular type the orbital ordering compatible with the g - type antiferromagnetism.@xcite the situation in the area of first - principles electronic structure calculations is controversial as well .
although majority of the researchers now agree that in order to describe properly the electronic structure of @xmath0 perovskite oxides , one should go beyond the conventional lda and incorporate the effects of intraatomic coulomb correlations , this merging is typically done in a semi - empirical way , as it relies on a certain number of adjustable parameters , postulates , and the form of the basis functions used for the implementation of various corrections on the top of lda.@xcite there are also certain differences regarding both the definition and the approximations used for the cf splitting in the electronic structure calculations , which will be considered in details in sec .
[ sec : kinetic ] .
since the magnetic properties of @xmath0 perovskite oxides are extremely sensitive to all such details , it is not surprising that there is a substantial variation in the results of first - principles calculations , which sometimes yield even qualitatively different conclusions about the cf splitting and the magnetic structure of the distorted @xmath0 perovskite oxides.@xcite these discrepancies put forward a severe demand on the creation of a really parameter - free scheme of electronic structure calculations for the strongly - correlated systems .
therefore , the main motivation of the present work is twofold .
+ ( i ) in our previous work ( ref . )
we have proposed a method of construction of the effective hubbard - type model for the electronic states near the fermi level on the basis of first - principles electronic structure calculations . in the present work
we apply this strategy to the @xmath0 states of the distorted perovskite oxides .
namely , we will derive the parameters of the hubbard hamiltonian for the @xmath0 bands and solve this hamiltonian using several different techniques , including the hartree - fock ( hf ) approximation , the perturbation theory for the correlation energy , and the the theory of superexchange interactions taking into account the effects of the multiplet structure of the atomic states .
of course , our method is based on a number of approximation , which have been introduced in ref . and will be briefly discussed in sec .
[ sec : model ] .
however , we would like to emphasize from the very beginning that our policy here is _ not to use any adjustable parameters _ apart from the approximations considered in ref . .
thus , we believe that it poses a severe test for the proposed method , and the obtained results should clearly demonstrate that our general strategy , which can be expressed by the formula _ first - principles electronic structure calculations _ @xmath11 _ construction of the model hamiltonian _
@xmath11 _ solution of the model hamiltonian_,@xcite is indeed very promising .
for example , at the hf level , using relatively simple model hamiltonian , which is limited exclusively by the @xmath0 bands , we will be able to reproduce the main results of all - electron calculations.@xcite furthermore , due to the simplicity of the model hamiltonian we can easily go beyond the hf approximation and include the correlation effects .
+ ( ii ) why do we need to convert the results of first - principles electronic structure calculations into a model ?
apart from the purely computational reason , related with the reduced dimensionality of the hilbert space for the solution of the many - electron problem,@xcite the story of distorted @xmath0 perovskite oxides clearly shows that the model consideration has yet another advantage , which is typically not sufficiently appreciated in the computational community .
it is true that the field of first - principles electronic structure calculations is currently on the rise , and the calculation of the basic properties for many materials will soon become a matter of routine . however , the methods of electronic structure calculations are based on some approximations , the limitations of which should be clearly understood .
furthermore , like the experiment data , the results of first - principles electronic structure calculations will always require some interpretation , which would transform the world of numbers and trends into a `` parallel world '' of rationalized model categories capturing the essence of the electronic structure calculations .
the understanding of the results of calculations in terms of these categories opens a way , on the one hand , to the material engineering of compounds with a desired set of properties , and , on the other way , to `` engineering '' of the new methods of electronic structure calculations in the direction of elucidation and overcoming the existing approximations . in this work
we will illustrate how the results of first - principles calculations for the distorted perovskite oxides can be interpreted in terms of a limited number of model parameters , such as the crystal - field splitting , transfer integrals , and the intraatomic coulomb interactions , which can be regarded as the basic operating blocks for understanding the properties of these materials as well as the limitation of approximations existing in the methods of electronic structure calculations .
particularly , we will explicitly show that the atomic - spheres - approximation ( asa ) , which was employed in the series of publications ( refs . ) , is not enough as it neglects the nonsphericity of the madelung potential .
the latter plays an important role and in many cases predetermines the character of the magnetic ordering in the distorted @xmath0 perovskite oxides .
we will also show that once the parameters of coulomb interactions are determined from the first principles , the commonly used mean - field hf approximation does not necessary guaranty the right answer for the magnetic properties of @xmath0 perovskite oxides .
however , we will argue that this is a normal situation , and in the majority of cases , a better agreement with the experimental data can be obtained by systematically including the correlation effects beyond the hf approximation . in this sense ,
our strategy is completely different from conventional lda@xmath12@xmath13 calculations , where the on - site coulomb interaction @xmath13 is typically treated as an adjustable parameter ( e.g. , refs . ) . by changing @xmath13
, one can certainly get a better numerical agreement with some experimental data already at the hf level .
however , one should clearly understand that such an empirical treatment actually disguises the actual role played by the correlation effects in the narrow - band compounds .
the rest of the paper is organized as follows . in sec .
[ sec : structure ] we will briefly remind the main details of the crystal and magnetic structure of the distorted perovskite oxides .
the procedure of constructing the model hamiltonian as well as the results of calculations of the cf splitting , transfer integrals , and on - site coulomb interactions for the isolated @xmath0 band will be briefly explained in sec .
[ sec : model ] .
particularly , in sec . [
sec : kinetic ] we will discuss highly controversial situation around the values of the cf splitting extracted from electronic structure calculations,@xcite and argue that the main difference is caused by two factors : ( i ) certain arbitrariness with the choice of the wannier functions for the @xmath0 bands of the distorted perovskite oxides ; ( ii ) additional approximations used for the nonspherical part of the crystalline potential inside atomic spheres .
the methods of solution of the model hamiltonian will be described sec .
[ sec : method ] , and the results of calculations will be presented in sec .
[ sec : results ] . finally , in sec .
[ sec : summary ] we will summarize the main results of our work .
the distorted perovskite oxides contain four formula units in the primitive cell . the transition - metal ( @xmath5 ) atoms are located at @xmath14 ( site 1 ) , @xmath15 ( site 2 ) , @xmath16 ( site 3 ) , and @xmath17 ( site 4 ) , in terms of three primitive translations : @xmath18 , @xmath19 , and @xmath20 ( see fig . [ fig.structure ] ) .
the distortion can be either orthorhombic or monoclinic .
the space group of the orthorhombic phase is @xmath21 ( in the schnflies notations or @xmath22 in the hermann - maguin notations , no . 62 in the international tables ) .
in this case all @xmath5-sites are equivalent and can be transformed to each other using symmetry operations of the @xmath21 group .
the monoclinic phase has the space group @xmath23 ( @xmath24 , no . 14 in the international tables).@xcite in this case , there are two nonequivalent pairs of @xmath5-sites : ( 1,2 ) and ( 3,4 ) .
each pair is allocated within the same @xmath25-plane , so that the atoms can be transformed to each other using symmetry operations of the @xmath23 group . however , there is no symmetry operation , which connects the atoms belonging to different @xmath25-planes .
we use the experimental lattice parameters and the atomic positions reported in ref . for ytio@xmath1 ( the data corresponds to the temperature t@xmath4@xmath26 k ) , in ref . for latio@xmath1 ( t@xmath4@xmath27 k ) , in ref . for yvo@xmath1 ( t@xmath4@xmath28 k and @xmath29 k , for the orthorhombic and monoclinic phase ,
respectively ) , and in ref . for lavo@xmath1 ( t@xmath4@xmath30 k ) .
there are five possible magnetic structure , which can be obtained by associating with each transition - metal site either positive ( @xmath31 ) or negative ( @xmath32 ) direction of spin , without enlarging the unit cell .
therefore , each magnetic structure can be denoted by means of four vectors associated with the transition - metal sites ( 1 2 3 4 ) .
they are + 1 .
( @xmath31@xmath31@xmath31@xmath31 ) , which is called the ferromagnetic ( f ) phase ; + 2 .
( @xmath31@xmath31@xmath32@xmath32 ) , the layered ( a - type ) antiferromagnetic phase ; + 3 .
( @xmath31@xmath32@xmath31@xmath32 ) , the chainlike ( c - type ) antiferromagnetic phase ; + 4 .
( @xmath31@xmath32@xmath32@xmath31 ) , the totally antiferromagnetic ( g - type ) phase ; + 5 .
( @xmath32@xmath31@xmath31@xmath31 ) , the spin - flip phase . in the monoclinic structure , there are two different spin - flip phases : ( @xmath32@xmath31@xmath31@xmath31 ) and ( @xmath31@xmath31@xmath32@xmath31 ) , which will be denoted as flip - i and flip - ii , respectively .
+ similar classification can be used the orbital ordering .
typically , two orthogonal orbitals at the neighboring transition - metal sites are said to be ordered antiferromagnetically , although such a definition is not unique.@xcite
our first goal is the construction of the effective multi - orbital hubbard model for the isolated @xmath0 bands : @xmath33 where @xmath34 ( @xmath35 ) creates ( annihilates ) an electron in the wannier orbital @xmath36 of the site @xmath37 , and @xmath38 is a joint index , incorporating all remaining ( spin and orbital ) degrees of freedom .
the matrix @xmath39@xmath40@xmath41 parameterizes the kinetic energy of electrons , where the site - diagonal part ( @xmath37@xmath4@xmath42 ) describes the local level - splitting , caused by the crystal field and ( or ) the spin - orbit interaction , and the off - diagonal part ( @xmath37@xmath43@xmath42 ) stands for the transfer integrals .
@xmath44@xmath45@xmath46 are the matrix elements of _ screened _ coulomb interaction @xmath47@xmath45@xmath48 , which are supposed to be diagonal with respect to the site indices . the parameters of the hubbard hamiltonian ( [ eqn : hmanybody ] ) can be derived `` from the first principles '' , starting from the electronic structure in lda .
this procedure has been already discussed in details in ref . .
here we only remind the main ideas and present the results for the distorted perovskite compounds .
all calculations have been performed using linear muffin - tin - orbital ( lmto ) method in the atomic - spheres - approximation.@xcite we have also considered the additional correction to the crystal - field splitting , coming from the nonsphericity of electron - ion interactions , beyond conventional asa.@xcite the kinetic - energy part of the hubbard hamiltonian was constructed using the downfolding method.@xcite it yields certain set of parameters @xmath49 .
the wannier functions @xmath50 for the the isolated @xmath0 bands can be formally reconstructed from @xmath49 using the definition @xmath51@xmath4@xmath52 , where @xmath53 is the lda hamiltonian in asa.@xcite the ( characteristic ) example of such wannier functions constructed for latio@xmath1 is shown in fig .
[ fig.latio3wf ] , and their extension in the real space is illustrated in fig .
[ fig.latio3wfextension ] .
the functions are well localized : about 80 - 85% of their total weight is concentrated at the central ti site , 5 - 9 % belong to neighboring oxygen sites , and about 10 % is distributed over la , ti , and o sites located in next coordination spheres .
another measure of localization of the wannier functions is the expectation value of the square of the position operator : @xmath54@xmath4@xmath55@xmath45@xmath56,@xcite which yields @xmath54@xmath4 @xmath57 , @xmath58 , and @xmath59 @xmath60 for @xmath38@xmath4 1 , 2 , and 3 , respectively .
the wannier functions for latio@xmath1 are less localized in comparison with the more distorted ytio@xmath1 , where @xmath61 is of the order of @xmath62-@xmath63 @xmath60.@xcite however , this is to be expected . the parameters @xmath49 include all kinds of hybridization ( or covalent mixing ) effects between transition - metal @xmath0 and other atomic orbitals . however , there are other effects , which are not yet included in @xmath49 .
they come from the nonsphericity ( n - s ) of the madelung potential for the electron - ion interactions , and contribute to the cf splitting .
the proper correction to @xmath49 can be easily calculated in the basis of wannier functions @xmath50 : @xmath64 where @xmath65 is the total charge associated with the site @xmath42 ( namely , the nuclear charge minus the screening electronic charge encircled by the atomic sphere ) , and @xmath66 is the position of the electron in the sphere @xmath37 . the main idea behind this treatment of the cf splitting is based on certain hierarchy of interactions in solids .
it implies that the strongest interaction , which leads to the energetic separation of the @xmath0 band from other bands ( fig .
[ fig.dossummary ] ) , is the covalent mixing .
for example , in many transition - metal oxides this interaction is mainly responsible for the famous splitting between the transition - metal @xmath0 and @xmath7 states.@xcite the nonsphericity of the madelung potential is considerably weaker than this splitting .
however , it can be comparable with the effects of covalent mixing in the narrow @xmath0 band .
therefore , the basic idea is to treat this nonsphericity as a pseudo - perturbation,@xcite and calculate the matrix elements of the madelung potential in the basis of wannier functions constructed for spherically averaged asa potential .
the same strategy can be used for the spin - orbit ( s - o ) interaction , which yields the following correction to the kinetic part of the model hamiltonian : @xmath67 ( @xmath68 being the lda potential and @xmath69 being the vector of pauli matrices ) .
one of the most controversial issues , which is actively discussed in the literature , is the values and the directions of the cf splitting in the distorted @xmath0 perovskite oxides .
therefore , we would like to discuss this problem more in details . basically , there are two sources of discrepancies , which largely affect the conclusions about the orbital ordering and the magnetic ground state : + 1 .
( the origin of the cf splitting : the nonsphericity of the madelung potential versus the hybridization effects ) the importance of nonsphericity of the madelung potential has been emphasized by several authors . the original idea is due to mochizuki and imada , who considered the @xmath0-level splitting in the ti - compounds associated with the displacements of y and la sites.@xcite it was paraphrased by cwik _ et al._,@xcite who suggested the main effect comes from the deformation of the tio@xmath70 octahedra .
a more general picture has been considered by schmitz _ et al._,@xcite who summed up all contributions in the madelung potential .
a weak point of all these approaches is the approximate treatment of the hybridization effects , which largely relied on the model parameters .
moreover , the result depends crucially on the value of the dielectric constant , which is treated as an adjustable parameter . on the other hand , the parameters of the model hamiltonians extracted from the first - principles electronic structure calculations using either downfolding ( refs . , , and ) or wannier function ( ref . )
methods automatically include all effects of the covalent mixing . in this sense , these are more rigorous techniques .
however , all these calculations were supplemented with the additional atomic - spheres - approximation and neglected the nonspherical part of the madelung potential . the madelung term has been also neglected in our previous work ( ref . ) . as we shall see in secs .
[ sec : yto ] and [ sec : lto ] , it will definitely revise several statements of ref . .
however , the final conclusion about the magnetic ground state of ytio@xmath1 and latio@xmath1 remains valid .
+ 2 . ( the nonuniqueness of the wannier functions )
different calculations yield different parameters of the cf splitting .
for example , for latio@xmath1 different authors reported the following parameters of the cf splitting ( between lowest and highest @xmath0-levels ) : 93 mev ( ref . ) , 200 mev ( ref . ) , and 270 mev ( ref . ) .
there is a particularly bad custom to criticize ref .
, @xcite which reports the smallest value , even with certain hints at the accuracy of calculations.@xcite it is also premature to think that the small cf splitting will inevitably lead to the orbital liquid scenario,@xcite because the transfer interactions are also strongly affected by the lattice distortion , which makes a big difference from the idealized cubic perovskites.@xcite first , we would like to consider the second part of problem and argue that different values of the cf splitting are most likely related with the different choice of the wannier functions , which by no means is unique .
this is _ not _ a problem of accuracy of calculations . in the downfolding method employed in refs . , , and , all basis states were divided in two groups : the `` @xmath0 '' part @xmath71 , and the `` rest '' of the basis functions @xmath72 . the effective hamiltonian is constructed by eliminating the `` rest '' part.@xcite a similar idea ( although formulated in slightly different way ) is employed in projections scheme for the construction of the wannier functions,@xcite where @xmath71 play a role of trial orbitals .
the basic difficulty here is that , in the distorted perovskites , the set of atomic `` @xmath0 '' orbitals can not be defined in an unique fashion : since the local symmetry is not cubic , the abbreviations like `` @xmath0 '' and `` @xmath7 '' will always reflect some bad quantum numbers for the state , which are mixed by the crystal field and/or the transfer interactions .
in numerical calculations , the set of `` @xmath0 '' orbitals is always specified in some local coordinate frame , and the choice of this frame appeared to be different in different calculations .
for example , pavarini _ et al . _
( refs . and ) and streltsov _ et al .
_ ( ref . ) selected their local coordinate frames from some geometrical considerations .
a completely different strategy was pursued by the present author in ref . , where the atomic `` @xmath0 '' orbitals were determined from the diagonalization of the local density matrix constructed from the @xmath0 bands and projected onto all five @xmath73-orbitals of the transition - metal sites .
then , we are ready to argue that different choice of the local coordinate frame naturally explains the difference in the parameters of the cf splitting reported by different authors . for these purposes we consider two different setups in the downfolding scheme , and consider latio@xmath1 as an example .
the first scheme is absolutely identical to that proposed in ref . , where the atomic `` @xmath0 '' orbitals have been defined as three most populated orbitals obtained from the diagonalization of the density matrix for the @xmath0 bands . in the second scheme ,
we first construct a more general @xmath74@xmath75@xmath74 tight - binding hamiltonian , comprising the ti(@xmath73 ) and la(@xmath9 ) states , and reproducing the behavior of @xmath74 overlapping ti(@xmath73)-la(@xmath9 ) bands .
other orbitals have been eliminated using the downfolding method .
then , we diagonalize the site - diagonal part of this hamiltonian and define three lowest eigenstates at the ti - site as the atomic `` @xmath0 '' orbitals .
after that we eliminate the rest of the ti(@xmath73 ) and la(@xmath9 ) states using the downfolding method and obtain the minimal @xmath76@xmath75@xmath76 hamiltonian for the @xmath0 bands .
both downfolding schemes are nearly perfect and well reproduce the behavior of the @xmath0 bands in the reciprocal space ( fig .
[ fig.latio3twoschemes ] ) .
however , they yield very different parameters after the fourier transformation to the real space . for example , the splitting of atomic @xmath0 levels ( in mev ) obtained in the schemes i and ii is ( @xmath45@xmath77,@xmath78,@xmath79 ) and ( @xmath45@xmath80,@xmath81,@xmath82 ) , respectively . moreover , the eigenvectors corresponding to the lowest `` @xmath0 '' levels appear to be also different . in the orthorhombic coordinate frame , specified by the vectors @xmath18 , @xmath19 , and @xmath20 ,
the eigenvectors have the following form ( referred to the site 1 ) : @xmath83@xmath4@xmath84@xmath45@xmath85@xmath45@xmath86@xmath45@xmath87@xmath12@xmath88@xmath45@xmath89 and @xmath90@xmath4@xmath91@xmath45@xmath92@xmath12@xmath93@xmath12@xmath94@xmath12@xmath95@xmath45@xmath89 , for the scheme i and ii , respectively .
thus , the small value of the cf splitting reported in ref . is related with the particular choice of the wannier functions ( or the parameters of the downfolding scheme ) .
had we changed our definition of the local coordinate frame , our conclusion would have been also different , and we could easily obtain the cf splitting of the order of @xmath96 mev ( and even larger ) . then , it is of course right to ask which scheme is better ? in principle , the physics should not depend on the choice of the wannier functions , and as long as we are dealing only with the kinetic - energy part of the model hamiltonian , both schemes are totally equivalent as they equally well reproduce the behavior of the @xmath0 bands .
however , what we want to do next is to combine this kinetic - energy part with the coulomb interactions , and _ to use only the site - diagonal part of these interactions_. this is of course an approximation , and in order to justify it one should guarantee that the wannier functions , which are used as the basis for the matrix elements of the coulomb interactions , were sufficiently well localized in the real space , so that all inter - site matrix elements could be neglected .
the degree of localization of the wannier orbitals is related with the spread of transfer integrals .
loosely speaking , in order to contribute to the transfer integral between @xmath97-@xmath98 neighbors , the wannier function should have a finite weight at this neighbor .
the distance - dependence of transfer integrals calculated in two different scheme is shown in fig .
[ fig.latio3twoschemes ] .
one can clearly see that the transfer integrals obtained in the scheme i are indeed well localized and basically restricted by the nearest neighbors .
the transfer integrals obtained in the scheme ii are less localized and spread far beyond the nearest neighbors .
therefore , the wannier functions , corresponding to the scheme i , should be more localized .
this is not surprising , because the scheme i guarantee that the density matrix ( or the integrated density of states ) at the transition - metal site is already well described by the central parts ( or `` heads '' ) of the wannier functions , given by the atomic orbitals @xmath71.@xcite therefore , the tails of the wannier functions , coming from the neighboring sites , should be small and cancel each other . in the scheme
ii , the local density matrix is composed by both `` heads '' of the wannier functions as well as their tails coming from the neighboring sites .
intuitively , this means that the tail part of the wannier functions is larger for the scheme ii and these functions are less localized .
thus , we believe that the scheme i is more suitable for the purposes of our work and we apply it to all perovskite compounds .
the transfer integrals , obtained in such a way , are indeed well localized and restricted by the nearest neighbors ( fig .
[ fig.avrtransfer ] ) . the next important contribution to
the cf splitting comes from the nonsphericity of the madelung potential .
this effect is _ comparable _ with the cf splitting of the covalent type .
we have also found that there are several different contributions to the cf splitting , which tend to cancel each other .
for example , the cf splitting of the covalent type in ytio@xmath1 and latio@xmath1 is largely compensated by the nonsphericity of the madelung potential coming from the region encircling the the neighboring oxygen sites and corresponding to the cut - off radius @xmath99@xmath100@xmath101 in fig .
[ fig.cfconvergence].@xcite the next important contribution comes from the y / la and ti sites , located in the next coordination spheres ( @xmath99@xmath100@xmath78 ) . in some compounds , even longer - range interactions spreading up to @xmath99@xmath100@xmath30
can have a relatively large weight in the sum ( [ eqn : cfei ] ) .
the final scheme of the `` @xmath0''-level splitting , which takes into account all these effects , is summarized in fig .
[ fig.cfsummary ] .
the splitting is not particularly large .
nevertheless , as we shall see below , it appears to be sufficient in order to explain the experimentally observed orbital ordering and the magnetic ground state _ in all considered compounds , except latio@xmath1_. the madelung potential changes not only the magnitude of the crystal - field splitting , but also the _ type _ of the atomic `` @xmath0''-orbitals , which are split off by the distortion ( table [ tab : cforbitals ] ) .
.lowest ( for ytio@xmath1 and latio@xmath1 ) and highest ( for yvo@xmath1 and lavo@xmath1 ) atomic orbital obtained from the diagonalization of the site - diagonal part of the model hamiltonian in the downfolding method ( denoted as the `` covalent part '' ) and and after including the nonsphericity of the madelung potential ( denoted as `` total '' ) .
the symbols ` o ' and ` m ' stand for orthorhombic and monoclinic phases , respectively .
the order of the basis orbitals is @xmath102 , @xmath103 , @xmath104 , @xmath105 , and @xmath106@xmath45@xmath107 , in the orthorhombic coordinate frame .
the positions of the transition - metal sites are shown in fig .
[ fig.structure ] . [ cols="^,^,^,^,^",options="header " , ] the magnetic ground state of ytio@xmath1 is g - a - f , in agreement with the neutron scattering data.@xcite the ferromagnetic moment is aligned along the @xmath20-axis , and the canting angle is relatively small .
the absolute values of the spin and orbital magnetic moments are different from those reported in our previous work.@xcite the difference is related with the additional transformation to the lmto basis , which allows to calculate the local magnetic moments directly at the transition - metal sites , whereas in ref .
they have been calculated in the wannier basis , which had a substantial weight at the oxygen sites .
this comparison clearly shows that the covalent effects play a very important role and allow to explain a substantial reduction of the local magnetic moments at the transition - metal sites . as it was already discussed in sec .
[ sec : yto ] , the correlation effects in ytio@xmath1 favor the afm coupling and systematically lower the energies of all afm states relative to the ferromagnetic one ( see table [ tab : yto ] ) . after including the spin - orbit interaction ,
this will lead to the additional spin canting away from the collinear ferromagnetic arrangement .
for example , in the framework of superexchange approach we have obtained the following values of the spin and orbital magnetic moments ( in @xmath108 , referred to the site 1 ) : @xmath109@xmath4@xmath110 and @xmath111@xmath4@xmath112 .
it readily explains the experimental values reported for the f and g components of the magnetic moments , correspondingly along the @xmath20 and @xmath18-directions,@xcite while the a - type afm component along the @xmath19-direction is somewhat larger in our calculations .
the reason is not completely clear .
the orbital magnetic moments in ytio@xmath1 are strongly quenched by the crystal field .
the orthorhombic yvo@xmath1 has nearly - collinear magnetic structure , where the g - type afm moment is aligned along the @xmath20-axis .
the total magnetic moment @xmath113@xmath12@xmath114@xmath4@xmath115 parallel to the @xmath20-axis is in excellent agreement with the experimental value of @xmath116 reported by blake _ et al . _ and corresponding to t@xmath4@xmath28 k.@xcite ulrich _ et al .
_ reported somewhat larger value of @xmath117 , also oriented along @xmath20 .
the weak ferromagnetic component along the @xmath18-direction has been also observed experimentally.@xcite details of magnetic ordering in the monoclinic phase of yvo@xmath1 are somewhat controversial .
blake _ et al .
_ reported the c - type afm ordering for both @xmath19- and @xmath20-components of the magnetic moments in the orthorhombic @xmath22 notations , which correspond to the @xmath18- and @xmath20-directions in the monoclinic @xmath24 notations.@xcite this is totally consistent with our finding .
the quantitative difference can be naturally explained by the finite temperature effects in the intermediate phase.@xcite furthermore , we predict the a - type _ antiferrimagnetic _
ordering for the remaining @xmath19-component ( in the @xmath24 notations ) , which implies nearly antiferromagnetic alignment of the @xmath19-projections of the magnetic moments in the adjacent @xmath25-planes .
however , since the sites @xmath118 and @xmath119 are not fully equivalent in the monoclinic structure , the @xmath19-projections do not compensate each other , and the system exposes the net magnetic moment , which can couple to the magnetic field.@xcite the _ antiferrimagnetic _ ordering along the @xmath19-axis is also consistent with the temperature - induced magnetization reversal behavior of yvo@xmath1 observed by ren,@xcite and could be a natural explanation for this effect .
more generally , it is right to say that the magnetic coupling along the @xmath20-direction of the monoclinic phase is always either ferr__i__magnetic or antiferr__i__magnetic , because the sites @xmath118 and @xmath119 are _ not equivalent_. the c - a - c magnetic ground state appears to be different from the magnetic structure reported by ulrich _ et al .
_ , who have observed the c - type afm ordering for the @xmath18- and @xmath19-components , and the g - type afm ordering for the remaining @xmath20-component ( apparently , in the @xmath22 notations).@xcite one possible explanation for this difference could be the coexistence of several magnetic states in a narrow energy range .
such a behavior has been indeed observed in our hf calculations : in addition to the c - a - c state we could obtain another self - consistent solution corresponding to the a - c - f phase with a slightly higher energy ( about @xmath120 mev per one formula unit , which is of the order of the magnetocrystalline anisotropy energy ) .
the new phase has the following magnetic moments ( in @xmath108 , where the first and second lines corresponds to the sites @xmath118 and @xmath119 , respectively ) : @xmath121 although the exact form of the magnetic state a - c - f is still different from the observation by ulrich _ et al._,@xcite we can speculate that they probably used a different experimental setup which yielded the realization of another magnetic phase , which was different from the finding by blake _ et al . _
( ref . ) and ren _ et al . _ ( ref . ) .
at least , the c - type afm component parallel to the @xmath25-plane is qualitatively consistent with the form of the a - c - f phase obtain in our hf conclusions , and to certain extent the results of hf calculations can be further modified by the correlation effects . in the case of
lavo@xmath1 we could find three stable solutions : a - c - a , c - a - c , and c - f - g .
the first two are similar to the a - c - f and c - a - c states , emerging in monoclinic yvo@xmath1 .
the only difference is that the a - c - a state has lower energy and , therefore , is realized as the ground state in the case of lavo@xmath1 . like in yvo@xmath1
, the a - type state corresponds to the _ antiferrimagnetic _ ordering .
therefore , lavo@xmath1 is expected to have the net magnetic moment in the @xmath122-plane .
the third ( c - f - g ) solution has nearly collinear spin structure : @xmath123 it corresponds to the stable g - type afm phase emerging without the spin - orbit interaction ( see table [ tab : yvoo ] ) .
nevertheless , the a - c - a state appears to be well separated from the c - a - c and c - f - g states , correspondingly by @xmath124 and @xmath125 mev per one formula unit .
the quenching of the orbital magnetic moments in two different sublattices of monoclinic yvo@xmath1 and lavo@xmath1 nicely correlates with the values of the cf splitting ( fig .
[ fig.cfsummary ] ) , where larger cf splitting at the site 3 results in smaller orbital magnetization .
it is probably meaningless to discuss the relativistic effects in latio@xmath1 , where we could not reproduce the correct magnetic ground state .
we could agree with the criticism risen by haverkort _ et al . _
( ref . ) that our cf splitting alone does not explain details of their spin - resolved photoemission spectra ( actually , our value of the parameter @xmath126 , obtained in the hf calculations after the transformation to the lmto basis and radial integration over the ti - sphere is about @xmath45@xmath127 , which exceeds the experimental value by factor two ) .
however , it does not make a sense to present as an alternative the results of calculations yielding the same a - type afm ground state,@xcite which totally agrees with our finding and ( unfortunately ) disagrees with the experiment , even though these calculations yield somewhat larger value of the cf splitting for latio@xmath1 . as it has been already discussed in sec .
[ sec : kinetic ] , different values of the cf splitting obtained by different authors are most probably related with the nonunique choice of the wannier functions for the @xmath0 bands of the distorted perovskite oxides .
one can formally adjust the theoretical cf splitting in order to meet certain demands of some particular class of the experimental data .
however , will it solve a more fundamental problem of the magnetic ground state of latio@xmath1 ?
in addition , we show in table [ tab : hfgs ] the values of the band gap obtained in the hf calculations for the magnetic ground state . for ytio@xmath1 , yvo@xmath1 , and
lavo@xmath1 there is a good agreement with the experimental optical data.@xcite however , for latio@xmath1 the experimental gap is much smaller ( @xmath100@xmath128 ev).@xcite this may indicate again at the particular importance of correlation effects in latio@xmath1 .
the main purpose of this work was to make a bridge between first - principles electronic structure calculations and model approaches for the strongly - correlated systems , and illustrate how it works for the series of distorted @xmath0 perovskite oxides .
the whole plan included three major steps : _ first - principles electronic structure calculations _ @xmath11 _ construction of the model hamiltonian _ for the isolated @xmath0 bands near the fermi level @xmath11 _ solution of this model hamiltonian _ using several different approaches .
the choice of the distorted @xmath0 perovskite oxides was motivated by the fact that they represent a good example of systems where it is practically impossible to construct a relevant model without the impact from the first - principles calculations : simply , the lattice distortion is too complex and there are too many model parameters , which can not be fixed in unbiased way . in this sense , we strongly believe that any theoretical model for such complex oxide materials should be based on the results of first - principles electronic structure calculations
. otherwise it could be just an abstract mathematical construction deprived of clear physical grounds .
the present work clearly demonstrates that nowadays the idea of constructing the parameter - free model hamiltonians for the strongly - correlated systems is quite feasible : all model parameters in our work , including the intraatomic coulomb interactions , have been derived from the first - principles calculations using the method proposed in ref . .
apart from the approximations inherent to the method , the procedure of constructing the model hamiltonian was totally parameter - free , and our analysis _ did not rely on the use of any adjustable parameters_. therefore , it is remarkable that using such a parameter - free approach we could propose a consistent explanation for a number of puzzling properties of the distorted @xmath0 perovskite oxides .
these first results are really encouraging and we would like to hope that the method proposed in ref . can be successfully applied in the future for the analysis of electronic and magnetic properties of other narrow - band compounds . at the mean - field hartree - fock level , the results of all - electron full - potential lda@xmath12@xmath13 calculations ( refs . )
can be successfully reproduced in our model approach for the isolated @xmath0 bands .
we could easily rationalize the main results of these relatively heavy calculations and elucidate the main microscopic interactions responsible for the formation of different magnetic structures in the case of ytio@xmath1 , yvo@xmath1 , and lavo@xmath1 .
we argue that the nonsphericity of the madelung potential should be an indispensable ingredient of electronic structure calculations , and the results of commonly used atomic - spheres - approximation should be corrected in order to include these effects .
the crystal distortion plays an important role in the physics of @xmath0 perovskite oxides . at least for ytio@xmath1 , yvo@xmath1 , and lavo@xmath1 ,
the knowledge of the experimental lattice parameters and the atomic positions greatly helps in explaining the magnetic properties of these compounds . of course , some questions still remain .
particularly , what is the origin of this distortion ?
why is it so different for different compounds , that is finally manifested in the formation of completely different magnetic structures ?
very similar arguments have been used for lamno@xmath1 , which is a characteristic example of the jahn - teller distorted @xmath7 perovskite oxides .
particularly , it was argued that the experimental distortion not only stabilize the a - type afm ground state of lamno@xmath1,@xcite but is also responsible for the opening of the band gap.@xcite however , the direction of the jahn - teller distortion in lamno@xmath1 can be naturally understood in terms of anharmonicity of the electron - lattice interaction.@xcite in this sense , lamno@xmath1 is an easy example .
then , is it possible to rationalize the behavior of @xmath0 perovskite oxides in a similar way and come up with some suitable electron - lattice model , which would explain not only the direction of the lattice distortion in each particular compound , but also the difference between these compounds ? finally , we emphasize the importance of correlation effects in the @xmath0 band of distorted perovskite oxides .
although the mean - field hartree - fock approach provides a satisfactory description for the magnetic properties of ytio@xmath1 , yvo@xmath1 , and lavo@xmath1 , the inclusion of the correlation effects systematically improves the agreement with the experimental data for all three compounds .
definitely , latio@xmath1 is an exceptional example for which we could not obtain the correct g - type afm ground state neither at the level of hartree - fock approach nor after including some correlation effects , though in a very approximate form .
however , we expect that the situation may still change by systematically employing more rigorous theories for the correlation effects .
the approximations considered in the present work were simply not enough in the case of latio@xmath1 .
v. i. anisimov , a. i. poteryaev , m. a. korotin , a. o. anokhin , and g. kotliar , j. phys
. condens .
* 9 * , 7359 ( 1997 ) ; s. y. savrasov , g. kotliar , and e. abrahams , nature * 410 * , 793 ( 2001 ) ; k. held , g. keller , v. eyert , d. vollhardt , and v. i. anisimov , phys .
lett . * 86 * , 5345 ( 2001 ) ; a. i. lichtenstein and m. i. katsnelson , phys .
b * 57 * , 6884 ( 1998 ) .
for example , from the viewpoint of superexchange interactions , it is more natural to define the `` orthogonal '' orbitals as the ones which are decoupled via the matrix of transfer integrals between neighboring transition - metal sites : @xmath129@xmath4@xmath130 , which is not always the same as the regular orthogonality condition.@xcite o. k. andersen , phys . rev .
b * 12 * , 3060 ( 1975 ) ; o. gunnarsson , o. jepsen , and o. k. andersen , _ ibid .
_ * 27 * , 7144 ( 1983 ) ; h. l. skriver , _ the lmto method _ , ( springer - verlag , berlin , 1984 ) ; o. k. andersen , z. pawlowska , and o. jepsen , phys .
b * 34 * , 5253 ( 1986 ) .
the wannir functions have been calculated using the method proposed in ref . .
in these calculations , the wannier functions for the @xmath0 bands have been orthogonalized to the o(@xmath6 ) , ti(@xmath7 ) , and y(@xmath8 ) bands , in the case of ytio@xmath1 ; o(@xmath6 ) , ti(@xmath7 ) , and la(@xmath131 ) bands , in the case of latio@xmath1 ; o(@xmath6 ) , v(@xmath7 ) , and y(@xmath8 ) bands , in the case of yvo@xmath1 ; and o(@xmath6 ) , v(@xmath7 ) , and la(@xmath9 ) bands , in the case of lavo@xmath1 .
we have confirmed that such a accuracy is sufficient for calculations of the crystal - field splitting caused by the nonsphericity of the madelung potential ( [ eqn : cfei ] ) .
m. w. haverkort , z. hu , a. tanaka , g. ghiringhelli , h. roth , m. cwik , t. lorenz , c. schssler - langeheine , s. v. streltsov , a. s. mylnikova , v. i. anisimov , c. de nadai , n. b. brookes , h. h. hsieh , h .- j .
lin , c. t. chen , t. mizokawa , y. taguchi , y. tokura , d. i. khomskii , and l. h. tjeng , phys .
lett . * 94 * , 056401 ( 2005 ) . note that in order to improve asa for the loosely packed atomic structures , several empty spheres have been added in the process of lmto calculations .
a typical example for ytio@xmath1 is given in ref . .
these spheres are negatively charged ( @xmath132@xmath133@xmath130 ) and some of them are located in the nearest neighborhood to the transition - metal sites . the contribution of these spheres to the madelung potential makes some difference from the conventional form of the cf splitting expected from the the point - charge model .
o. gunnarsson , o. k. andersen , o. jepsen , and j. zaanen , phys .
b * 39 * , 1708 ( 1989 ) ; o. gunnarsson , a. v. postnikov , and o. k. andersen , _ ibid .
_ * 40 * , 10407 ( 1989 ) ; v. i. anisimov and o. gunnarsson , _ ibid . _
* 43 * , 7570 ( 1991 ) ; i. v. solovyev and p. h. dederichs , _ ibid . _ * 49 * , 6736 ( 1994 ) .
j. kanamori , prog .
phys . * 30 * , 275 ( 1963 ) .
note that the definition of the kanamori parameters @xmath134 and @xmath135 is different from the parameters @xmath13 and @xmath136 , which are typically used in lda@xmath12@xmath13 calculations.@xcite nevertheless , there is a simple relation connecting these two sets of parameters : @xmath134@xmath4@xmath13@xmath12@xmath137 and @xmath135@xmath138@xmath139 .
a. i. liechtenstein , m. i. katsnelson , v. p. antropov , and v. a. gubanov , j. magn .
magn . matter .
* 67 * , 65 ( 1987 ) ; i. v. solovyev , in _ recent research developments in magnetism and magnetic materials _
( transworld research network , india , 2003 ) , vol . 1 ,
253 - 294 .
note that @xmath140@xmath12@xmath141@xmath142@xmath143@xmath12@xmath144 .
therefore , according to eq .
( [ eqn : esuperexchange ] ) , two separate excitations into the singlet ( s ) and triplet ( t ) states are more favorable energetically rather than the excitation into the hartree - fock state with an averaged energy .
note , however , that the lda@xmath12@xmath13 calculations reported in refs . and for yvo@xmath1 overestimate the total energy difference between different magnetic states , apparently due to the double - counting of the nonsphericity effects . by default , in the full - potential implementation of the lda@xmath12@xmath13 method , the nonsphericity of the on - site coulomb and exchange - correlation potential is taken into account twice : in the lda - part and in the correction for the on - site coulomb interactions .
therefore , the lda - part should be additionally averaged inside some atomic sphere and we belive that it is more physical to treat this part in asa.@xcite this problem mainly concerns with the @xmath145 compounds , which are subjected to the hund s rule effects coming from the nonspherical coulomb interactions between two @xmath0 electrons .
a more general problem is related with the technical implementations of the lda@xmath12@xmath13 method ( e.g. the choice of the parameter @xmath13 and the basis orbitals for the correction ) , which probably explain a small difference existing between our results and ref . for ytio@xmath1 .
@xmath146@xmath4@xmath147@xmath45@xmath148@xmath12@xmath149@xmath45@xmath150 and @xmath151@xmath4@xmath147@xmath45@xmath148@xmath45@xmath149@xmath12@xmath150 , where @xmath152 is the total energy either in the pure hf approximation or after including the correlation effects in the second order of perturbation theory .
note that our scheme of calculation of the exchange parameters is based on the infinitesimal rotations of the spin magnetic moments , which implies that the spins are classical .
therefore , the total energy of the heisenberg model has the form ( [ eqn : eheisenberg ] ) , where @xmath153 are related with the regular heisenberg parameters @xmath154 as @xmath153@xmath4@xmath155 .
therefore , to be consistent we our definition of the exchange parameters , we use the classical expression also for the transition temperature and replace @xmath156@xmath12@xmath157 by @xmath158 . in order to use the quantum - mechanical definition for the transition temperature ,
one should do a proper mapping of the total - energy change onto the heisenberg model , which is much more complex , because in quantum case , the total energy remains proportional to @xmath158 only for the ferromagnetically coupled spins .
the antiferromagnetic arangement yields an additional prefactor @xmath38 , which is specific to the geometry of the system and bounded by the following conditions : @xmath159@xmath12@xmath160@xmath161@xmath38@xmath161@xmath118 .
the latter are far from restrictive when the spin is small .
therefore , the quantum effects , which are especially important for the spin @xmath162 , will modify not only the expression for the transition temperature , but also the definition of the parameters of the heisenberg model themselves .
d. treves , phys .
rev . * 125 * , 1843 ( 1962 ) ; a. s. moskvin and e. v. sinitsyn , sov .
solid state * 14 * , 2198 ( 1973 ) ; t. yamaguchi , j. chem .
. solids * 35 * , 479 ( 1974 ) ; i. v. solovyev , n. hamada , and k. terakura , physica b * 237 - 238 * , 44 ( 1997 ) .
t. arima , y. tokura , and j. b. torrance , phys .
b * 48 * , 17006 ( 1993 ) ; y. okimoto , t. katsufuji , y. okada , t. arima , and y. tokura , phys . rev .
b * 51 * , 9581 ( 1995 ) ; s. miyasaka , y. okimoto , and y. tokura , j. phys .
* 71 * , 2086 ( 2002 ) .
n. hamada , h. sawada , and k. terakura ( 1995 ) : in _ spectroscopy of mott insulators and correlated metals _ , edited by a.
fujimori and y. tokura ( springer , berlin , 1995 ) ; w. e. pickett and d. j. singh , phys .
rev . b * 53 * , 1146 ( 1996 ) ; i. v. solovyev and k. terakura , in _ electronic structure and magnetism of complex materials _ , ed .
by d. j. singh and d. a. papaconstantopoulos ( springer , berlin , 2003 ) . | several puzzling aspects of interplay of the experimental lattice distortion and the the magnetic properties of four narrow @xmath0-band perovskite oxides ( ytio@xmath1 , latio@xmath1 , yvo@xmath1 , and lavo@xmath1 ) are clarified using results of first - principles electronic structure calculations .
first , we derive parameters of the effective hubbard - type hamiltonian for the isolated @xmath0 bands using newly developed downfolding method for the kinetic - energy part and a hybrid approach , based on the combination of the random - phase approximation and the constraint local - density approximation , for the screened coulomb interaction part .
apart form the above - mentioned approximation , the procedure of constructing the model hamiltonian is totally parameter - free .
the results are discussed in terms of the wannier functions localized around transition - metal sites .
the obtained hamiltonian was solved using a number of techniques , including the mean - field hartree - fock ( hf ) approximation , the second - order perturbation theory for the correlation energy , and a variational superexchange theory , which takes into account the multiplet structure of the atomic states .
we argue that the crystal distortion has a profound effect not only on the values of the crystal - field ( cf ) splitting , but also on the behavior of transfer integrals and even the screened coulomb interactions .
even though the cf splitting is not particularly large to quench the orbital degrees of freedom , the crystal distortion imposes a severe constraint on the form of the possible orbital states , which favor the formation of the experimentally observed magnetic structures in ytio@xmath1 , yvo@xmath1 , and lavo@xmath1 even at the level of mean - field hf approximation .
it is remarkable that for all three compounds , the main results of all - electron calculations can be successfully reproduced in our minimal model derived for the isolated @xmath0 bands .
we confirm that such an agreement is only possible when the nonsphericity of the madelung potential is explicitly included into the model . beyond the hf approximation ,
the correlations effects systematically improve the agreement with the experimental data . using the same type of approximations we could not reproduce the correct magnetic ground state of latio@xmath1 .
however , we expect that the situation may change by systematically improving the level of approximations for dealing with the correlation effects . |
all astronomers recognize that spectroscopy offers a wealth of information that can help characterize the properties of the observing target . in the context of stellar astrophysics
, spectroscopy plays many fundamental roles .
the relative strengths and widths of stellar absorption lines provide access to physical properties like effective temperature ( @xmath2 ) and surface gravity ( @xmath3 ) , enabling model comparisons in the hertzsprung - russell diagram to estimate the masses and ages so crucial to understanding stellar evolution , as well as individual elemental abundances or the collective metallicity " ( typically parameterized as @xmath4}$ ] ) , facilitating studies of the chemical hallmarks of different stellar populations . with sufficient resolution
, a spectrum also conveys information about rotation ( @xmath5 ) and kinematics ( e.g. , association with a cluster or companion through the radial velocity , @xmath6 ) .
while many fields benefit from such spectroscopic measurements , they are of acute interest to the exoplanet community .
there , all estimates of the planet properties are made _ relative _ to the host properties ( e.g. , the mass function and planet - to - host radius _ ratio _ are constrained with the radial velocity or transit techniques , respectively ) . moreover ,
essential clues to the planet formation process are encapsulated in the dependences of planet frequency on host mass ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and metallicity ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . the robust and quantitative extraction of physical ( or empirical ) parameters from an observed spectrum can be an extraordinary challenge .
stellar models serve as comparative benchmarks to associate observed spectral features with the parameters of interest . generating a synthetic model spectrum involves a complex numerical treatment of the stellar structure and radiative transfer through the atmosphere ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
detailed models calibrated to individual stars are important , but rare ( e.g. , the sun , vega ) ; therefore , these stellar models are relatively untested in large swaths of parameter - space . moreover , they necessarily include simplifications to treat complicated physical processes ( e.g. , convection ) or computational limitations ( e.g. , boundary conditions ) , and often must rely on incomplete or inaccurate atomic and molecular information ( e.g. , opacities ) . in principle , the models could be improved with appropriate reference to spectroscopic data . nevertheless , they are remarkably successful in reproducing many diagnostic spectral features .
there are various well - tested approaches being used in stellar astrophysics to compare these models with observed spectra and thereby infer basic parameters .
perhaps the most common is a straightforward empirical technique that relies on distilling an information - rich subset of the data , usually in the form of spectral line equivalent widths and/or local continuum shapes
. a combined sequence of the ratios of these quantities can be especially sensitive to a given model parameter ( e.g. , moog ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . this
indexing " approach has the advantage of being trivially fast .
but , each condensed relationship is only informative over a limited swath of parameter - space , and it potentially masks degeneracies that are encoded in the spectral line shapes .
another standard approach exploits the cross - correlation of an observed spectrum with a suite of model templates to optimize a set of parameters , usually with some weighting applied to specific spectral regions ( e.g. , spc ; * ? ? ?
* ) . in this case
, the speed advantage is maintained ( perhaps enhanced ) and more data content is used ( particularly in the spectral dimension ) , thereby achieving higher precision even for data with comparatively low signal - to - noise .
the disadvantage is that the model quality and parameter inferences are assessed in a heuristic ( rather than probabilistic ) sense , making it difficult to quantify uncertainty in the stellar parameters .
a more direct method employs a pixel - by - pixel comparison between model and data .
this has the benefits of increased parametric flexibility ( e.g. , one can fit for arbitrary abundances or structures ) and a proper inference framework ( usually a least - squares approach , although increasingly in a bayesian format ; * ? ? ?
* ; * ? ? ?
ultimately , rather than pre - computing a library of sythetic spectra , one would like to incorporate the spectral synthesis back - end ( e.g. , sme ; @xcite ) directly into the likelihood function , bypassing any interpolation when assessing the fit of stellar parameters in - between grid points in the library .
unfortunately , this is not yet computationally feasible beyond a limited wavelength range . in this article
, we construct a flexible forward - modeling method for the general spectroscopic inference problem in a bayesian framework , building on the best aspects of the latter two approaches highlighted above .
the key developments in this design include a spectral emulator to address the difficult task of interpolation in coarsely sampled synthetic spectral libraries and a non - trivial covariance matrix parameterized by both global ( stationary ) and local ( non - stationary ) gaussian process kernels .
when combined with an appropriately sophisticated set of quantitative metrics for the relevant physical parameters , this method will efficiently propagate systematic uncertainties into the parameter inferences .
ultimately , this approach could be employed to leverage spectroscopic data as a reference for improving the models .
a complete overview of the methodology behind this approach is provided in section [ sec : method ] . some tests and example applications ( for a high resolution optical spectrum of an f star , and a medium - resolution near - infrared spectrum of a mid - m star )
are described in section [ sec : examples ] .
finally , a discussion of its potential utility , especially the possibility of extending it to develop data - driven spectral models , is provided in section [ sec : discussion ] .
this section describes a generative bayesian modeling framework that confronts some of the key technical obstacles in the spectroscopic inference problem . the goal is to conservatively extract the maximal amount of information about a prescribed ( and usually degenerate ) parameter set by forward - modeling an observed spectrum , while also recognizing and explicitly accounting for the covariances ( and potentially biases ) introduced by pathologically imperfect models .
the method is modular , and therefore can easily incorporate additional physical or nuisance parameters as desired without sacrificing an accurate reflection of the limitations in the data .
the specific applications discussed here are related to the spectra of individual stars , but the methodology is generic ( and could be used for the composite spectra of unresolved stellar clusters , galaxies , etc . ) .
figure [ fig : flowchart ] serves as a graphical guide to the mechanics of this modeling framework , and the remainder of this section .
first , a model spectrum is generated for a given set of physical parameters ( section [ subsec : synthetic ] ; appendix [ sec : appendix ] ) , and then post - processed to mimic reality using a set of observational and practical nuisance parameters ( section [ subsec : postprocess ] ) .
next , a direct , pixel - by - pixel comparison between the data and model spectra is made with a prescribed likelihood function and a parametric treatment of the covariances between pixel residuals ( section [ subsec : likelihood ] ) .
that process is iterated using markov chain monte carlo ( mcmc ) simulations in a multi - stage gibbs sampler to numerically explore the posterior probability density of the model conditioned on the data , and thereby to determine constraints on the parameters of interest ( section [ subsec : mcmc ] ) . along the way
, these procedures are illustrated with observations of the high resolution optical spectrum from a nearby f star .
that specific application , along with some alternative demonstrations of the method , are discussed in more detail in section [ sec : examples ] .
there are many approaches for generating a model spectrum , @xmath7 , for a specific set of parameters , @xmath8}\}$ ] . in the most direct case of spectral synthesis ,
a model atmosphere structure is assembled and simulations of energy transport through it are conducted with a radiative transfer code ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . however , in general this approach is often computationally prohibitive for most iterative methods of probabilistic inference .
one partial compromise is to interpolate over a library of atmosphere structures that were pre - computed for a discrete set of parameter values , @xmath9 , for some arbitrary @xmath10 .
then , perform a radiative transfer calculation with that interpolated atmosphere to synthesize @xmath11 ( e.g. , sme ; * ? ? ?
a more common variant is to interpolate over a pre - synthesized library of model spectra , @xmath12 ( e.g. , * ? ? ?
* ; * ? ? ?
although the former approach is preferable , the computational cost of repeated spectral synthesis is enough to make a detailed exploration of parameter space less appealing ( although see section [ sec : discussion ] ) .
although the framework we are advocating is applicable for _ any _ back - end " that generates a model spectrum , it is illustrated here using the latter approach with the @xcite phoenix library . in practice , this reliance on spectral interpolation within a model library requires a sophisticated treatment of associated uncertainties .
the key problems are that the spectra themselves do not vary in a straightforward way as a function of @xmath10 ( especially within spectral lines ) , and that the typical model library is only sparsely sampled in @xmath10 . because of these issues , standard interpolation methods necessarily result in some information loss .
the practical consequence is that the inferred posteriors on the model parameters are often sharply peaked near a grid point in the library , @xmath9 , potentially biasing the results and artificially shrinking the inferred parameter uncertainties ( e.g. , @xcite ) . to mitigate these effects ,
we develop a spectral emulator " that smoothly interpolates in a sparse model library and records a covariance term to be used in the likelihood calculation that accounts for the associated uncertainties .
the emulator is described in detail in appendix [ sec : appendix ] .
we first decompose the model library into a representative set of eigenspectra using a principal component analysis . at each gridpoint in the library
, the corresponding spectrum can be reconstructed with a linear combination of these eigenspectra .
the weights associated with each eigenspectrum contribution vary smoothly as a function of the parameters , and so are used to train a gaussian process to interpolate the weights associated with any arbitrary @xmath10 . in this way , the emulator delivers a probability distribution that represents the range of possible interpolated spectra . by then marginalizing over this distribution
, we can modify the likelihood function to propagate the associated interpolation uncertainty . in the remainder of this section , the details of generating the reconstructed ( interpolated ) spectrum are not especially relevant ( see appendix [ sec : appendix ] ) .
typically , the raw " interpolated model spectrum @xmath13 that was generated above is highly over - sampled , and does not account for several additional observational and instrumental effects that become important in comparisons with real data .
therefore , a certain amount of post - processing is required before assessing the model quality .
we treat that post - processing in two stages . the first stage deals with an additional set of extrinsic " parameters , @xmath14 , that incorporate some dynamical considerations as well as observational effects related to geometry and the relative location of the target .
the second stage employs a suite of nuisance parameters , @xmath15 , designed to forward model some imperfections in the data calibration .
we can further divide @xmath14 into those parameters that impact the model primarily in the spectral or flux dimensions . for the former ,
we consider three kernels that contribute to the line - of - sight velocity distribution function .
the first , @xmath16 , treats the instrumental spectral broadening . for illustrative purposes , we assume @xmath16 is a gaussian with a mean of zero and a constant width @xmath17 at all @xmath18 , although more sophisticated forms could be adopted .
the second , @xmath19 , characterizes the broadening induced by stellar rotation , parameterized by @xmath20 as described by ( * ? ? ?
* his eq . 18.14 ) , the rotation velocity at the stellar equator projected on the line of sight ( where @xmath21 is the inclination of the stellar rotation axis ) . and
the third , @xmath22 , incorporates the radial velocity through a doppler shift .
the model spectrum is modified by the parameters @xmath23 $ ] through these kernels , using a convolution in velocity - space , @xmath24 and then re - sampled onto the discrete wavelengths corresponding to each data pixel , @xmath25 where the @xmath26 symbol denotes a re - sampling operator that maps the model spectrum onto the @xmath27-element model vector @xmath28 ( @xmath27 is the number of pixels in the spectrum ) .
figure [ fig : broadening ] shows a ( condensed ) graphical representation of these post - processing steps . at this stage ,
the model is further modified in the flux dimension .
a typical synthetic spectrum is computed as the flux that would be measured _ at the stellar surface _ , and so needs to be diluted by the subtended solid angle , @xmath29 , where @xmath30 is the stellar radius and @xmath31 is the distance .
an additional wavelength - dependent scaling factor is applied to account for interstellar extinction , assuming some previously - derived extinction law @xmath32 ( e.g. , * ? ? ?
* ) that is parameterized by @xmath33 .
the parameters @xmath34 $ ] are then applied as @xmath35 with simplified notation such that @xmath36 $ ] , where @xmath37 $ ] .
some spectral libraries provide spectra as with peak fluxes normalized to a constant value , in that case , @xmath38 will simply serve as an arbitrary scaling parameter .
the procedure so far is composed of straightforward operations demanded by practical astronomical and computing issues . if the data were _ perfectly _ calibrated , we could proceed to a likelihood calculation that makes a direct comparison with @xmath39 .
however , the calibration of the continuum shape for data with reasonably large spectral range is often not good enough to do this .
a common example of this imperfect calibration can be readily seen when comparing the overlaps between spectral orders from echelle observations .
even if such imperfections ( e.g. , in the flat field or blaze corrections , or perhaps more likely in the flux calibration process ) induce only minor , low - level deviations in the continuum shape , they can add up to a significant contribution in the likelihood function and thereby potentially bias the results . the traditional approach to dealing with this issue has been avoidance ; a low - order polynomial or spline function is matched ( separately ) to the model and the data and then divided off to normalize the spectra . while this is straightforward to do for earlier type stars , it only masks the problem .
this normalization procedure disposes of _ useful _ physical information content available in the continuum shape , and can be considerably uncertain in cases where the spectral line density is high ( e.g. , for cooler stellar photospheres ) .
moreover , it can not propagate the uncertainty inherent in deriving the normalization functions into a proper inference framework . instead
, we employ a more rigorous approach that forward - models the calibration imperfections with a set of nuisance parameters that modify the shape of the model spectrum . by later marginalizing over these nuisance parameters ,
we properly account for any uncertainties that these kinds of calibration imperfections induce on the stellar parameters of interest while also maintaining the useful information in the continuum shape . in practice , this is achieved by distorting segments of the model with polynomials , @xmath40 ( e.g. , * ? ? ?
* ; * ? ? ?
figure [ fig : chebyshev ] demonstrates how these nuisance parameters are applied to the model . for @xmath41 spectral orders ,
each denoted with index @xmath42 , the model spectrum can be decomposed as @xmath43 where @xmath44 is an @xmath45 degree chebyshev function .
the @xmath46 coefficients are considered a set of nuisance parameters , @xmath47 $ ] .
judicious priors can ensure that the real spectral features ( e.g. , molecular bands ) are not treated as residual calibration artifacts . the lowest - degree ( scaling )
coefficient , @xmath48 , is degenerate with the solid angle , @xmath38 .
therefore , we enforce an additional constraint that the mean of the polynomial is unity . for data with a single spectral order ,
this means simply setting @xmath49 . in the multiple order case ,
we assign @xmath49 in an arbitrary order as an anchor , but permit the @xmath48 in other orders to be different .
the fit of the model spectrum is assessed by comparing to the data with a pixel - by - pixel likelihood calculation .
if we denote the data spectrum as @xmath50 , then a corresponding residual spectrum ( an @xmath27-element vector ) can be defined for any input parameter set , @xmath51 to quantify the probability of the data conditioned on the model , we adopt a standard multi - dimensional gaussian likelihood function @xmath52^{1/2 } } \exp\left ( -\frac{1}{2 } { \mathsf{r}}^{\mathsf{t}}{\mathsf{c}}^{-1 } { \mathsf{r}}\right ) \label{eqn : likelihood}\ ] ] that penalizes models which yield larger residuals and explicitly allows for covariances in the residual spectrum through the @xmath53 matrix @xmath54 . for practical reasons ,
the log - likelihood is used as the quality metric , where @xmath55 the covariance matrix @xmath54 characterizes both the measurement uncertainty ( @xmath56 ; noise " ) in each pixel and the covariance between pixels .
when using a spectral emulator to interpolate model spectra , @xmath54 will be the sum of the covariance matrix described here and the emulator matrix derived in appendix [ sec : appendix ] ( eq .
[ eqn : modc ] ) . in the special case where each pixel is independent , the covariance matrix is diagonal , @xmath57 , where @xmath58 is the uncertainty in pixel @xmath21 and @xmath59 is the kronecker delta function , and eq .
[ eqn : lnlikelihood ] reduces to the familiar @xmath60 the sum of the square of the residuals weighted by their inverse variances .
however , that simplification rarely applies in practice .
a more complex covariance matrix is required , so that additional off - diagonal terms can be used to explicitly characterize ( 1 ) pixel - to - pixel covariances imposed by the discrete over - sampling of the line - spread function , and ( 2 ) highly correlated residuals as manifestations of systematic imperfections in the model library .
the following sections describe how these issues are addressed by constructing a more sophisticated @xmath54 .
astronomical spectrographs are designed to have the detector over - sample the instrumental line - spread function with at least a few pixels .
therefore , adjacent pixels never record independent samples of the true spectrum . in that case , a difference between an observed and modeled spectral feature creates a correlated residual that spans multiple pixels .
this can be demonstrated clearly in the autocorrelation of @xmath61 : a slight model mismatch will produce correlated residuals over a characteristic scale similar to the instrumental or rotation broadening kernel width ( whichever is larger ) .
figure [ fig : class0 ] shows an example of these correlated residuals in real data ; a significant autocorrelation signal is seen on an @xmath628 pixel scale , corresponding to the 6.8 km s@xmath63 fwhm of @xmath16 .
it is important to distinguish here between noise " and the fit residuals .
noise introduced to the spectrograph by astrophysical or instrumental effects is generally uncorrelated with wavelength .
the arrival and propagation of each photon through the instrument and into the detector can be considered an independent event .
in essence , the noise itself is not correlated , but the fit residuals likely are . however , from a mathematical perspective the correlated residuals can be treated in the same way as correlated noise , by constructing a non - trivial covariance matrix with off - diagonal terms . in practice , this is achieved by parameterizing @xmath54 with a kernel that describes the covariance between any pair of pixels , indexed @xmath64 , representing wavelengths @xmath65 and @xmath66 . for a well - designed spectrograph and sufficiently accurate model , this _ global _
( i.e. , present throughout the spectrum ) covariance should have a relatively low amplitude and small correlation length . to describe that structure , we use a stationary covariance kernel ( or radial basis function ) with an amplitude that depends only on the velocity separation between two pixels , @xmath67 where @xmath68 is the speed of light .
this kernel is used to characterize the covariance between pixel residuals , @xmath69 a variety of useful kernels have been developed in the field of gaussian processes to parameterize such a covariant structure ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , and are seeing increased use in many areas of astrophysics ( for some specific examples in stellar and planetary applications , see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
after some experimentation , we adopted the matrn kernel with @xmath70 because it performed well at reproducing the appearance of realistic residuals for this specific problem . in this case , @xmath71 with @xmath72 $ ] , an amplitude ( @xmath73 ) and a scale ( @xmath74 ) .
the @xmath75 are termed _
hyperparameters _ here ; because a gaussian process describes a population of functions generated by random draws from a probability distribution set by a mean vector and a covariance matrix , the kernel parameters are naturally part of a hierarchical model . in this specific case ,
the functions described by these hyperparameters represent many realizations of covariant residuals from a spectral fit .
figure [ fig : matrix ] shows an example of the gaussian process kernel and the covariant residuals that can be generated from it . to ensure that @xmath54 remains a relatively sparse matrix ( for computational expediency ) , we employ a hann window function @xmath76 to taper the kernel .
the truncation distance @xmath77 can be set to a multiple of the scale ( we set @xmath78 ) .
in addition to the global covariance structure , there can be local regions of highly correlated residuals .
these patches of large @xmath61 are usually produced by pathologically incorrect spectral features in the model , due to systematic imperfections like missing opacity sources or poorly constrained atomic / molecular data ( e.g. , oscillator strengths ) .
some representative examples are shown in figure [ fig : badlines ] .
to parameterize such regions in @xmath54 , we introduce a sequence of non - stationary kernels that explicitly depend on the actual wavelength values of a pair of pixels ( @xmath65 and @xmath66 ) , and not simply their separation ( @xmath79 ) . assuming that these local residual features are primarily due to discrepancies in the spectral line depth ( rather than the line shape or central wavelength ) , a simple gaussian is a reasonable residual model . in that case
, the pixel residuals of the @xmath80-th such local feature could be described as @xmath81\ ] ] with peak amplitude @xmath82 , central wavelength @xmath83 , and width @xmath84 .
we assume that the amplitude of this gaussian feature is drawn from a normal distribution @xmath85 with mean 0 and variance @xmath86 .
the pixels in this gaussian - shaped residual are correlated because each pixel shares a common random scale factor ( @xmath82 ) .
then , the covariance of any two pixels in this region is given by eq .
[ eqn : expectation ] , where the expectation value is taken with respect to the probability distribution in eq .
[ eqn : amplitude ] @xmath87 a_k \exp \left [ - \frac{r^2(\lambda_j , \mu_k)}{2 \sigma_k^2 } \right ] \right \rangle \nonumber \\ & = & \langle a_k^2 \rangle \exp \left [ - \frac{r^2(\lambda_i , \mu_k ) + r^2(\lambda_j , \mu_k)}{2 \sigma_k^2 } \right ] \nonumber \\ & = & a_k^2 \exp \left [ - \ , \frac{r^2(\lambda_i , \mu_k ) + r^2(\lambda_j , \mu_k)}{2 \sigma_k^2}\right ] .
\label{eqn : kregion}\end{aligned}\ ] ] the full local covariance kernel covering all of the possible gaussian residuals is composed of a linear combination of kernels , @xmath88 with a corresponding set of hyperparameters @xmath89 $ ] .
note that we again taper the kernels with hann windows ( eq . [ eqn : hann ] ) to ensure a sparse covariance matrix ; in this case , the truncation distance @xmath77 can be set to some multiple of the width parameter ( e.g. , @xmath90 ) . in effect , these kernels systematically down - weight the influence of strong residuals in the likelihood calculation , mitigating any potential bias they might induce on inferences of the interesting parameters ( @xmath91 ) .
similar in spirit to robust linear regression and bad data " mixture models @xcite , these kernels provide a means for ( correlated ) outlier rejection that preserves the integrity of the probabilistic framework ( as opposed to the common manual or threshold - based techniques of masking or clipping ) . in principle , the concept of these local kernels can be extended to account for more complex residual structures .
for example , late - type stars with imperfectly modeled molecular bandheads may produce a complicated pattern of positive and negative residuals or a pronounced mismatch over a relatively large spectral scale .
this phenomenologically different local covariance behavior can still be treated in this framework if an appropriate kernel morphology is adopted .
we can now compute the covariance matrix employed in the likelihood calculation ( eq . [ eqn : lnlikelihood ] ) as the linear combination of the trivial pixel - by - pixel noise matrix and the global and local kernels discussed above , @xmath92 with hyperparameters @xmath93 $ ] .
the factor @xmath94 is a parameter that scales up the poisson noise in each pixel by a constant factor to account for additional detector or data reduction uncertainties ( e.g. , read noise , uncertainties in the spectral extraction procedure , etc . ) ; typically @xmath95 for well - calibrated optical spectra .
if there are @xmath96 local covariance patches ( see section [ subsec : mcmc ] on how this is determined ) , then there are @xmath97 elements in the set of covariance hyperparameters , @xmath98 .
figure [ fig : matrix ] provides a graphical illustration of how the kernels that comprise the covariance matrix are able to reproduce the structure present in a typical residual spectrum .
the bayesian framework of this inference approach permits us to specify prior knowledge about the model parameters , @xmath99 .
as will be discussed further in sections [ sec : examples ] and [ sec : discussion ] , in most cases it is necessary to utilize some independent information ( e.g. , from asteroseismology constraints or stellar evolution models ) as a prior on the surface gravity .
but otherwise we generally recommend a conservative assignment of uniform priors , such that @xmath100 is flat over the spectral library grid ( and zero elsewhere ) and @xmath101 is flat for physically meaningful values ( e.g. @xmath102 , @xmath103 , and @xmath104 )
. for ( early type ) stars with a clear continuum , it makes sense to assume flat priors on the polynomial parameters @xmath105 .
however , information about the calibration accuracy ( e.g. , from comparisons of multiple calibration sources in the same observation sequence ) can be encoded into a simple prior on the chebyshev coefficients ; for example , gaussian priors with widths that represent the fractional variance between different derived calibration functions would be reasonable . for ( late type )
stars with a poorly defined continuum , some judicious tapering of the priors ( such that small coefficients at high @xmath106 are preferred ) may be required to ensure that broad spectral features are not absorbed into the polynomial ( see section [ sec : examples ] ) . in general , uniform ( non - negative ) priors
are recommended for the global kernel hyperparameters . for the local kernels , we typically adopt uniform priors for the amplitudes and means \{@xmath107 , @xmath83 } , but construct a logistic prior for the widths \{@xmath84 } that is flat below the width of the line - of - sight velocity distribution function ( defined as the convolution of three broadening kernels in eq .
[ eqn : broadening ] ) , @xmath108 , and smoothly tapers to zero at larger values : @xmath109 such a prior formulation prevents local kernels from diffusing to large @xmath84 and low @xmath107 , since that kind of behavior is better treated by the global kernel .
when modeling real data , there is no _ a priori _ information about the locations @xmath110 of the local kernels ; they are instantiated as needed ( see section [ subsec : mcmc ] ) .
however , using the knowledge gained from previous inferences of similar targets , one could instead start by instantiating kernels at the outset with priors on @xmath110 where there are known to be systematic issues with the synthetic spectra .
the inference framework developed here has a natural blocked structure between the collections of interesting " parameters , @xmath111 $ ] , the nuisance parameters @xmath105 , and the covariance hyperparameters @xmath98 .
the conditional dependencies of these parameters are shown graphically in figure [ fig : pgm ] as a directed acyclic graph @xcite . to explore the posterior distribution , @xmath112 for this type of structure
, it is convenient to employ markov chain monte carlo ( mcmc ) simulations with a blocked gibbs sampler coupled to the metropolis - hastings algorithm .
this procedure works by sampling in a subset of parameters ( with metropolis - hastings proposals ) conditioned on the current ( fixed ) values of the other parameters . after each iteration
, the gibbs sampler updates the sampled parameters and then cycles through all the ( previously fixed ) different parameter subsets in the same way ( for a more mathematical description , see chapter 11 of @xcite ) . a step - by - step prescription follows , where the @xmath113 iteration of the gibbs sampler is indexed with a superscript : + ( 1 ) initialize the parameters .
one might set @xmath114 based on estimates in the literature or scaling behaviors , and make simple assumptions about @xmath115 .
here , we set the chebyshev coefficients ( @xmath115 ) so that the polynomials are constant ( @xmath116 and @xmath117 , @xmath118 @xmath42 ) and assume only the trivial noise spectrum ( and spectral emulator kernel ; see appendix [ sec : appendix ] ) contributes to the @xmath54 ( i.e. , @xmath119 ) .
+ ( 2a ) start the @xmath113 iteration of the gibbs sampler .
for each iteration of the metropolis - hastings algorithm , sample in @xmath91 to evaluate the posterior ( eq . [ eqn : post ] ) following the framework laid out in sections [ subsec : likelihood ] and [ subsec : priors ] .
this represents a slice " through the posterior space conditioned on the other parameters being held fixed ( @xmath120 and @xmath121 ) .
then update @xmath122 .
+ ( 2b ) for each spectral order , sample in the polynomial parameters @xmath105 and covariance hyperparameters @xmath98 , conditioned on the other parameters being held fixed @xmath123 . then update @xmath124 and @xmath125 .
+ ( 3 ) repeat step ( 2 ) for 20,000 samples .
+ ( 4 ) repeat the procedure in steps ( 1)(3 ) with different initializations , storing the samples for each markov chain .
after removing the burn - in samples for each chain , we compute the gelman - rubin convergence diagnostic , @xmath126 ( * ? ? ?
* their eq .
if @xmath127 , we can be reasonably sure that all of the chains have converged to the posterior distribution .
+ in step ( 2b ) , local covariance kernels are instantiated according to the following procedure .
first , an average " residual spectrum is generated by combining @xmath62500 residual spectra that were stored during a burn - in period using only the global kernels ( prior to this storage , the markov chain is thinned to account for autocorrelation of the posterior samples ) .
this average spectrum is then iteratively examined for deviations outside a critical threshold .
when a large residual is identified , a local kernel is introduced with a mean ( @xmath83 ) at its location . after some experimentation with different threshold criteria , we chose to instantiate when the local residual is @xmath1284@xmath129 the standard deviation in the average residual spectrum .
alternative schemes , such as re - evaluating the kernel locations with each iteration of the gibbs sampler , yield similar results ; however , the adopted approach consistently converges with minimal computational overhead .
once all the local kernels are instantiated , the gibbs sampler is run for another period of burn - in . ; in effect , the model will act as if the kernel were deleted automatically . ]
this entire procedure can be a significant computational challenge .
a typical spectrum has @xmath130 , and therefore the many evaluations of the matrix product @xmath131 in the likelihood calculation can be numerically expensive .
however , because @xmath54 is a symmetric , positive semi - definite matrix , we can employ cholesky factorization to optimize the evaluation of the matrix product and avoid the direct calculation of the matrix inversion ( @xmath132 ) .
for multi - order echelle spectra or multiple spectra of the same target ( perhaps taken with different instruments ) , the nuisance parameters for each segment of the spectrum are independent .
this means that the computationally intensive steps of generating a model spectrum for a specific wavelength range and evaluating the likelihood can be parallelized .
the only segment of the code that needs to be synchronized is the mcmc proposal of stellar parameters , which are shared between all chunks of the spectrum .
the massive parallelization of this algorithm on a computer cluster therefore enables the simultaneous inference of interesting parameters over wide spectral ranges at high resolution , or from multiple datasets . to sample the posteriors in this mode
, we extend the metropolis - hastings sampler included in the emcee package @xcite to function within a parallelized blocked gibbs sampler .
the time required to thoroughly explore the posterior depends on both the data volume and the desired precision on the inference of the covariance hyperparameters .
if only the stellar parameters @xmath91 are of interest , one can first optimize the kernel parameters and then proceed with them fixed , since the stellar parameter posteriors are relatively insensitive to the precise value of the kernel parameters ( once near their optimal value ) .
a fit of an @xmath133 spectrum with @xmath12830 echelle orders takes @xmath622 hours ( parallelized on a cluster ) .
if the full posteriors for the nuisance parameters are desired , the computation might take an order of magnitude longer .
in this section , we illustrate how the modeling framework operates for two real datasets .
the first is an elaboration of the example shown throughout section [ sec : method ] , using a high resolution optical spectrum of the f5 star ( and transiting exoplanet host ) wasp-14 @xcite .
the second uses a medium resolution near - infrared spectrum of the m5 dwarf gliese 51 ( hereafter gl 51 ) , observed as part of the nasa / irtf library of spectral standards @xcite . in both cases , we sequentially build up the complexity of the modeling framework to demonstrate how each of the components described in section [ sec : method ] affects the posteriors on the parameters of interest ( @xmath10 ) .
we adopt the recent incarnation of the phoenix library @xcite for the models , although comment on systematic differences between libraries in section [ sec : systematics ] .
a high resolution ( @xmath134 ) optical spectrum of wasp-14 was obtained on 2009 june 14 using the tillinghast reflector echelle spectrograph ( tres ; * ? ? ?
* ) on the fred lawrence whipple observatory 1.5 m telescope .
tres delivers an echelle spectrum with 51 orders that cover the full optical range , from 38609100 .
the data were reduced and calibrated using standard techniques in the tres pipeline ( cf .
, @xcite ; see @xcite for more specific details ) . at 5100 , the s / n is @xmath62150 per resolution element .
following @xcite , we focus here on the central three tres orders , covering @xmath625100 - 5400 .
we start with a standard " inference , using the most commonly employed likelihood function ( i.e. , @xmath135 , with a trivial covariance matrix using only the poisson uncertainties ) .
interpolation in the model library is performed with a basic tri - linear algorithm ( in this specific case , @xmath136 has only three dimensions ) . to avoid a prominent systematic ( see section [ sec : systematics ] ) , we fix the surface gravity to @xmath137 ( with a @xmath138-function prior ) .
this independent prior information comes from the combination of a constraint on the mean stellar density based on exoplanet transit depth measurements and a comparison of optical photometry with stellar models in the color - magnitude diagram @xcite .
the resulting marginal posteriors on @xmath139 and [ fe / h ] , listed in table [ table : tests ] and shown in figure [ fig : phoenix_posterior ] , are remarkably narrow unbelievably so , given how subtly the spectrum changes over such small parameter deviations .
for the second test , we increase the complexity of the covariance matrix by introducing the global kernel treatment discussed in section [ subsec : global_covariance ] .
we find non - negligible amplitudes and correlation lengths for these kernels , as would be expected for a typical correlated residual spectrum .
with respect to the standard inference , the uncertainty associated with @xmath2 has increased by roughly a factor of three , but the @xmath4}$ ] posterior is only marginally broadened ( by @xmath6250% ) . upon closer inspection of the latter ,
it becomes clear that the posterior has an artificially sharp peak located at a grid point in the model library ( @xmath4}= -0.5 $ ] ) .
this ` noding ' is an artifact of naive interpolation over a sparsely - sampled dimension in the library grid ; when the uncertainty in the interpolation itself constitutes a significant fraction of the total error budget , the fit will be driven toward grid points ( where the interpolation error is naturally minimized ; see also @xcite ) . to mitigate this behavior , we need to employ an interpolation scheme that properly incorporates this kind of uncertainty .
clc|r@ @xmath140 lr@ @xmath140 l[!bh ] ( 1 ) & linear & trivial & 6280 & 5 & @xmath1410.471 & 0.004 + ( 2 ) & linear & + @xmath142 & 6297 & 16 & @xmath1410.500 & 0.006 + ( 3 ) & emulator & + @xmath142 & 6281 & 26 & @xmath1410.482 & 0.012 + ( 4 ) & emulator & + @xmath143 & 6301 & 29 & @xmath1410.431 & 0.012 therefore , in a third test we implement the bayesian emulator described in appendix [ sec : appendix ] to propagate uncertainty in the interpolation .
this procedure successfully avoids the ` noding ' behavior in @xmath4}$ ] , and inflates the associated uncertainty by a factor of 2.5 compared to the standard " inference approach .
the uncertainty on @xmath2 is now 5@xmath129 larger than in the original test . finally ,
in a fourth test we fold in the methodology for the local covariance kernels described in section [ subsec : local_covariance ] .
this has little effect on the widths of the parameter posteriors ( @xmath14410% increase ) , but does shift their peaks to slightly higher values in both @xmath2 and @xmath4}$ ] .
we suspect this is likely driven by a bias in the inference of @xmath4}$ ] , produced because the phoenix models tend to have more ` outlier ' spectral lines with _ over - predicted _ line depths . without the local covariance kernels to downweight these outliers , the models tend toward lower metallicity to account for them .
but when the local kernels are included , this bias is reduced and a more appropriate higher @xmath4}$ ] value is inferred .
figure [ fig : phoenix_residuals ] demonstrates how well the modeling framework can match the character of the residual spectrum when employing the sophisticated covariance matrix ( test 4 ) advocated here .
a moderate resolution ( @xmath145 ) near - infrared spectrum of gl 51 was obtained on 2000 nov 6 using the spex instrument @xcite on the 2.3 m nasa infrared telescope facility ( irtf ) .
spex is a cross - dispersed echelle spectrograph that covers the red - optical to thermal - infrared spectrum ( 0.75.5@xmath146 m ) in two settings .
these data were obtained as part of the irtf spectral standard library project @xcite , and were processed through the well - vetted spextool reduction pipeline @xcite to deliver a fully calibrated spectrum . at 2.1@xmath146 m , the s / n is @xmath62400 per resolution element .
clc|r@ @xmath140 lr@ @xmath140 l[!bh ] ( 1 ) & linear & trivial & 3256 & 3 & 0.89 & 0.01 + ( 2 ) & linear & + @xmath142 & 3022 & 35 & 0.00 & 0.03 + ( 3 ) & emulator & + @xmath142 & 3230 & 30 & 0.27 & 0.03 + ( 4 ) & emulator & + @xmath143 & 3180 & 35 & 0.28 & 0.04 modeling late - type stellar atmosphere structures and their spectra is considerably more complex than for sun - like stars , due to lingering uncertainties in the atmosphere physics and molecular opacities . especially confounding is the presence of complex condensates ( clouds ) at the coolest temperatures @xcite , making it considerably more challenging to determine ( sub- ) stellar properties @xcite .
various approaches have been taken to infer the key parameters in the face of these difficulties , including iteratively masking regions with poor spectral agreement ( e.g. , * ? ? ?
astutely , @xcite note that such a scheme may exclude regions of the spectrum that contain intrinsically useful information for discriminating between physical properties , and that a more sophisticated approach would weight each spectral region based on its consistency with the data .
the modeling framework that we have constructed here does exactly that . as another demonstration of this framework , we carried out the sequence of tests outlined in the previous section for the @xmath1-band portion of the spex spectrum of gl 51 . following the analysis of similar data for this star by @xcite
, we fix the surface gravity to @xmath147 based on a comparison with standard stellar evolution models .
the test results are listed in table [ table : tests_gl51 ] ; the posteriors for @xmath2 and @xmath4}$ ] are shown together in figure [ fig : gl51_posterior ] .
like wasp-14 , we find that a more appropriate treatment of the covariance matrix results in a substantial broadening of the parameter posteriors ; the uncertainties on @xmath2 and @xmath4}$ ] are inflated by a factor of @xmath6210 and 4 , respectively .
however , in this case the parameter values ( posterior peaks ) also exhibit substantial movement along the sequence of tests .
the underlying cause of this behavior lies with the and resonance line depths , which are systematically under - predicted in the phoenix library ( even for high metallicities ; see also @xcite ) .
@xcite suggest that these discrepancies may be the consequence of inaccurate atomic data ( oscillator strengths and/or opacities ) .
in the first test with a trivial covariance matrix , these ` outlier ' lines drive the model to favor a very high @xmath4}$ ] .
but when we consider the more sophisticated versions of @xmath54 that employ gaussian processes to treat correlated residuals , the contribution of these features to the likelihood calculation is reduced , and therefore @xmath4}$ ] returns to a more appropriate range . because this portion of the spectrum has only these two outlier features , their influence can be mitigated either with a larger global covariance kernel amplitude ( @xmath73 ) , or with a smaller @xmath73 but significant contributions from local covariance kernels ( which explains why there is little difference between the posteriors in the third and fourth tests in the sequence ) . for reference ,
figure [ fig : gl51_residuals ] compares the observations with draws from the posterior distribution for the advocated modeling approach ( corresponding to the fourth test ) .
the methodology behind the likelihood calculations we have developed could prove especially useful for spectroscopic inferences of the parameters of cool stars like gl 51 , where substantial uncertainties in their more complex atmospheres will naturally produce systematic deviations between models and data .
however , many of those discrepancies will be manifested in molecular features , which likely result in considerably more complex residual structures than noted here ( e.g. , the tio bands in the red - optical ; see * ? ? ?
* their fig . 9 ) .
the overall framework we have employed should still function , although more appropriate local covariance kernels may need to be developed to capture the different nature of these outliers .
for example , one might employ hybrid kernels ( like the product of a truncated exponential and a matrn kernel ) or empirically - motivated parametric shapes ( e.g. , a saw - tooth pattern ) to provide a better representation than a simple gaussian feature .
the results of the sequence of tests in the previous two sections illustrate some key issues in the spectroscopic inference of stellar parameters .
first , the residual spectra derived from ( typically ) imperfect models exhibit correlated structure ( e.g. , see fig . [
fig : class0 ] ) that can not be explained well with a trivial ( diagonal ) covariance matrix .
if that naive assumption is made ( as is usually the case ) , the resulting posteriors are unrealistically narrow and may end up being biased ( particularly for @xmath4}$ ] or for cases influenced by prominent ` outlier ' lines ) .
this issue of implausibly small formal uncertainties has long been recognized in the stellar spectroscopy community .
the standard solution has been to add ( in quadrature ) a ` floor ' contribution , imposed independently on each parameter and meant to be representative of the systematics ( e.g. , see @xcite or @xcite for clear and open discussions of this approach ) .
the key problems with this tactic are that these systematics are in reality degenerate ( and so should not be applied independently ) and that they dominate the uncertainty budget , but are in a large sense arbitrary they are not self - consistently derived in the likelihood framework .
our goal here has been to treat one aspect of this systematic uncertainty budget internal to the forward - modeling process , by employing a non - trivial covariance matrix that accounts for generic issues in the pixel - by - pixel inference problem . given the results above , we have demonstrated that this procedure successfully accounts for a substantial fraction of the ( empirically motivated ) _ ad hoc _ systematic ` floor ' contribution typically adopted in inference studies . however , although a likelihood function that can properly account for the character of the residuals is important , it does not by itself treat _ all _ of the important kinds of systematics in the general spectroscopic inference problem . in future work that can build on the flexible likelihood formalism we have advocated here ,
there are three other important sources of systematic uncertainty that should be considered : ( 1 ) data calibration ; ( 2 ) optimized parameter sensitivity ; and ( 3 ) model assumptions , or flexibility .
we discuss each of these issues briefly , with attention paid to potential remedies that fit within the likelihood framework developed here .
perhaps the most familiar source of systematics lies with issues in the data calibration . in the idealized case of perfect calibration
, the physical parameters inferred from different observations of the same ( static ) source should be indistinguishable . but given the complexity of a detailed spectroscopic calibration , that is not typically the case in practice . the common approach to quantify the systematic uncertainties contributed by calibration issues
is to compare the inferences made using different spectra ( e.g. , from different instruments and/or observations ) .
the final parameter values are usually presented as an average of these separate inferences , with the uncertainties inflated by adding in quadrature some parameter - independent terms that account for their dispersion .
the more appropriate way of combining these inferences is to model the individual spectra simultaneously in a hierarchical framework like the one discussed in section [ sec : method ] : in that way , the dispersion is appropriately propagated into the parameter uncertainties while any intrinsic degeneracies are preserved ( which is not possible in the standard ` weighted average ' approach ) .
ultimately , one could also introduce some empirically - motivated nuisance parameters that are capable of forward - modeling imperfections in the data calibration , similar to the approach adopted in section [ subsec : postprocess ] ( e.g. , see fig . [
fig : chebyshev ] ) .
another important source of systematic _ bias _ comes from the fact that certain physical parameters have only a relatively subtle effect on the spectrum .
stellar spectroscopists are familiar with this being an issue when inferring the surface gravity , @xmath3 , since it is primarily manifested as low - level modifications to the wings of certain spectral lines like mg b and in the equivalent widths of lines from singly - ionized elements like and .
when modeling data with a large spectral range , the effects of varying @xmath3 are small compared to the residuals introduced by the many other model imperfections .
consequently , the surface gravity will not be constrained well , and inferences on @xmath3 ( and therefore other degenerate parameters ) can be substantially biased . as an example , when fitting the wasp-14 data in section [ subsec : wasp ] without prior information on the surface gravity , we find a shift of @xmath620.9 dex to lower @xmath3 ( and accompanying shifts in @xmath2 and @xmath4}$ ] ) .
if we instead use a customized version of the @xcite models designed to more accurately reproduce this part of the optical spectrum for sun - like stars ( as employed by spc ; @xcite ) , there is still a 0.2 dex shift compared to the independent , accurate constraints from the transiting planet @xcite
. similar work with larger samples indicate a typical scatter in the @xmath3 values inferred solely from spectra relative to independent , accurate constraints from other data ( @xmath620.5 dex ; @xcite ) .
there are two commonly utilized , and not mutually exclusive , approaches to mitigating this kind of bias .
first is the judicious use of a prior , based on either independent and accurate measurements ( e.g. , from asteroseismology , dynamical masses and distances , etc . ) or stellar evolution models ( as is demonstrated here ) . of course , such information is unfortunately not always readily available for the target of interest .
a second approach is to severely limit the spectral range of the data being modeled , focusing primarily on those spectral features especially sensitive to the parameter of interest . but that carries its own risk , since the models derived from the inferred posteriors might well be wildly inconsistent with the rest of the spectrum .
recently , @xcite proposed a sophisticated , iterative approach that apparently resolves this issue , employing a sequence of conditional inferences based on sets of specific spectral features that are especially sensitive to individual physical parameters .
this seems like a promising component for future inclusion in the likelihood framework we have developed in section [ sec : method ] . finally , and perhaps most significant , there are also sources of systematic bias and uncertainty introduced by limitations in the synthetic stellar models themselves .
different models make varied assumptions in their treatments of the atmosphere structures , boundary conditions ( e.g. , convection ) , fundamental atomic / molecular data ( e.g. , opacities ) , and radiative transfer . taken together , these variations produce notably different model spectra for the same values of the physical parameters . as a benchmark for estimating the scope of this source of bias ,
we re - performed the inference described in the fourth test of section [ subsec : wasp ] , but using the customized @xcite model library instead of the @xcite library .
the resulting inferences for @xmath2 and @xmath4}$ ] are in excellent agreement with those derived by @xcite using the spc method , but are shifted by 150k ( higher ) and 0.15dex ( higher ) , respectively , compared to the phoenix results . while the relevant physics included in these models is very similar for these temperatures and the inferred stellar parameters are similar in an absolute sense , it is still striking that the systematic shift between models is several times larger than the statistical uncertainties derived from our likelihood function . at this point
, there is little to be done to rectify these model - dependent differences ; in the future , one hopes that the model inputs can be refined based on feedback from the data ( see sect . [ sec : discussion ] ) .
any inferences of physical parameters should only be considered in the context of the assumed models . aside from these different assumptions and inputs ,
the limited _ flexibility _ of these models certainly also contributes to the systematic uncertainty budget , and is possibly also a source of systematic bias .
model spectral libraries typically have neglected dimensions in parameter - space that , if made available , would be expected to broaden and possibly shift the posteriors for the primary physical parameters .
one typical example lies with element - specific abundance patterns , often distilled to the enhancement of @xmath148-elements ( i.e. , [ @xmath148/fe ] ) . if the target star has a non - zero [ @xmath148/fe ] ( an enhancement or deficit relative to the solar ratios ) , but is fit with a single , global metallicity pattern , it is not clear that the sophisticated covariance formalism developed here would be capable of appropriately capturing such residual behavior .
another prominent example of an important hidden parameter dimension is the microturbulence , which for some spectral types and spectral resolution may impact the spectrum in a similar way as the surface gravity ( and may be partly responsible for the @xmath3 bias discussed above ; * ? ? ?
* ) . to mitigate the resulting deficiencies in precision ( and potentially accuracy ) on the inference of other parameters , we would ideally employ libraries or modeling front - ends that can incorporate some flexibility in these hidden ( i.e. , ignored ) dimensions of parameter - space ( e.g. , individual elemental or group - based abundance patterns , microturbulence , etc . )
astronomers exploit spectroscopy to retrieve physical information about their targets .
ideally , such inferences are made with the maximal precision afforded by the measurement noise , and accurately reflect the uncertainties with minimal systematic bias .
but in practice , the spectral models used as references are never perfect representations . even modest mismatches between data and model can propagate substantial systematic uncertainty into the inference problem . in high - sensitivity applications ( e.g. , stellar and exoplanetary astrophysics ) , ignoring these systematics
can give a false sense of both precision and accuracy in the inferences of key parameters .
typically , the more egregious of these imperfections are mitigated " by dismissal ( explicitly not considering a subset of the data ; e.g. , masking , clipping ) .
rarely , they are confronted directly with painstaking , computationally expensive fine - tuning of more general ( nuisance ) parameters in the model ( e.g. , oscillator strengths , opacities ) , albeit only over a very limited spectral range and region of physical parameter - space .
we have presented an alternative approach to dealing with this fundamental issue , grounded in a generative bayesian framework .
the method advocated here constructs a sophisticated likelihood function , employing a non - trivial covariance matrix to treat the correlated pixel - to - pixel residuals generated from intrinsically imperfect models . that matrix is composed of a linear combination of _ global _ ( stationary ) and _ local _ ( non - stationary ) gaussian process kernels , which parameterize an overall mild covariance structure as well as small patches of highly discrepant outlier features .
in the context of a given model parameterization ( i.e. , synthetic spectral library , or a more complex and flexible model generator ) , the framework we have developed provides a better inference than the standard @xmath149 ( or cross - correlation ) comparison .
we have built up a series of tests that demonstrates how the emulator , global kernels , and local kernels affect the nature of the inference on the stellar parameters . to demonstrate how the framework is used , we determined the surface parameters of main - sequence stars with mid - f and mid - m spectral types from high - s / n optical and near - infrared data , with reference to pre - computed model libraries ( sect .
[ sec : examples ] ) .
the source code developed here is open and freely available for use : see http://iancze.github.io/starfish .
the novelty of employing this kind of likelihood function in the spectroscopic inference problem is that the treatment of data model mismatches ( in essence , the fit quality ) is explicitly built into the forward - modeling framework .
this offers the unique advantage that discrepant spectral features ( outliers ) , which may contain substantial ( even crucial ) information about the parameters of interest , can still effectively propagate their useful information content into the posteriors with a weighting that is determined self - consistently . from a practical standpoint ,
this means that a larger spectral range can be used and model imperfections can be downweighted by the usage of covariance kernels .
the global covariance framework provides more appropriate estimates of the posterior probability distribution functions ( i.e. , the precision or uncertainty estimates ) for the model parameters . the automated identification and disciplined downweighting of problematic outlier " spectral lines ( those that can not be reproduced with any combination of the model parameters ) with local covariance kernels can prevent them from overly influencing ( and possibly biasing , especially in cases with few spectral features available ) the inferences . in many cases ,
the underlying physical problem lies with incorrect ( or inaccurate ) atomic and/or opacity data used in the models . in this sense
, the posteriors of the hyperparameters of the local covariance kernels can actually indicate in what sense and scale these inputs need to be modified to better reproduce observational reality .
the approach we describe is generally applicable to any spectroscopic inference problem ( e.g. , population synthesis in unresolved star clusters or galaxies , physical / chemical models of emission line spectra in star - forming regions , etc . ) .
moreover , it has the flexibility to incorporate additional information ( as priors ) or parametric complexity ( if desired ) , and could be deployed as a substitute for a simplistic @xmath149 metric in already - established tools ( e.g. , sme )
. another potential application might be in the estimation of radial velocities using traditional doppler - tracking pipelines for exoplanet or binary star research .
poorly modeled micro - tellurics can lead to incorrect measurements of radial velocities for certain contaminated chunks of the spectrum , causing them to give unrealistically precise but biased velocity measurements .
a flexible noise model would broaden the posteriors on these points and allow them to be combined into a more accurate systemic velocity .
ultimately , the benefits of employing covariance kernels to accommodate imperfect models could be extended well beyond modeling the spectra of individual targets . in principle , the approach we have described here can be used to systematically discover and quantify imperfections in spectral models and eventually to build data - driven improvements of those models that are more appropriate for spectroscopic inference . by fitting many stellar spectra with the same family of models
, we can catalog the covariant structure of the fit residuals especially the parameters of the local covariance kernels to collate quantitative information about where and how the models tend to deviate from observational reality . that information can be passed to the spectral synthesis community , in some cases enabling modifications that will improve the quality of the spectral models . on a large enough scale
, this feedback between observers and modelers could be used to refine inputs like atomic and molecular data ( oscillator strengths , opacities ) , elemental abundance patterns , and perhaps the stellar atmosphere structures .
if one has access to the radiative synthesis process that generates the model spectra , there are many possible means to improve their quality .
in particular , a process of history matching can be used to rule out regions of parameter space where the models do not fit well ( e.g. , for a use in galaxy formation simulations , see @xcite ) .
for example , if one had full control over the radiative synthesis code , stellar structure code , and atomic line database , one could improve the performance of the spectral emulator by ruling out regions of parameter space for these separate components that are inconsistent with a collection of observed spectra , such as a set of standard stars spanning the full range of spectral classifications . in a similar vein
, we could also simultaneously use several synthetic spectral libraries to infer the stellar parameters while also identifying discrepant regions of the spectrum .
a treatment using multiple synthetic libraries would likely reveal interesting correlations between model discrepancies , such as a specific signature among many lines ( e.g. deviations in spectral line shape that can not be explained by variations in @xmath150 ) .
conversely , if a discrepant feature is seen for all models , it could be due to either an anomaly with the given star ( e.g. , a chromospheric line due to activity or perhaps an intervening interstellar absorption line ) or a correlated difficulty among all models ( e.g. , an incorrect atomic constant ) .
alternatively , this kind of feedback could be used to make data - driven modifications to the already existing models , creating a new semi - empirical model library .
this could be accomplished by linking the parameters of the covariance kernels while fitting many stars of similar spectral type in a hierarchical bayesian model , which would add confidence to the assessment that certain spectral features are _
systematic _ outliers and offer general quantitative guidance on how to weight them in the likelihood calculation . rather than simply assembling an empirical spectral library using only observations , this combined machine - learning approach would naturally provide a physical anchoring for the key physical parameters , since they are reflected in the spectra based on the physical assumptions in the original models .
this kind of large - scale analysis holds great promise in the ( ongoing ) era of large , homogeneous high resolution spectroscopic datasets ( e.g. , like those being collected in programs like the apogee and hermes surveys ; @xcite ) , since they provide enormous leverage for identifying and improving the underlying model systematics .
+ the authors would like to acknowledge the following people for many extraordinarily helpful discussions and key insights : daniel foreman - mackey , guillermo torres , david latham , lars buchhave , john johnson and the exolab , daniel eisenstein , rebekah dawson , tom loredo and the exostat group , allyson bieryla , and maxwell moe .
two anonymous reviewers provided encouraging and very useful comments on the manuscript draft that greatly improved its clarity and focus .
ic is supported by the nsf graduate fellowship and the smithsonian institution .
is supported at harvard by nsf grant ast-1211196 .
this research made extensive use of astropy @xcite and the julia language @xcite .
the spectral emulator is designed to serve as an improved interpolator for the synthetic spectral library . rather than a ( tri-)linear interpolator , which would deliver a single spectrum for a given @xmath151 , the spectral emulator delivers a probability distribution of possible interpolated spectra .
in this manner , it is possible to incorporate realistic uncertainties about the interpolation process into the actual likelihood calculation . in the limit of moderate to high signal - to - noise spectra
, these interpolation uncertainties can have a significant effect on the posterior distribution of @xmath151 .
a schematic of the emulator is shown in figure [ fig : flowchart_appendix ] , which is a continuation of figure [ fig : flowchart ] .
briefly , the emulator consists of a set of eigenspectra , representing the synthetic spectral library , that can be summed together with different weights to reproduce any spectrum originally in the library . to produce spectra that have @xmath151 in between @xmath152 ,
the weights are modeled with a smooth gaussian process ( gp ) .
this gp delivers a probability distribution over interpolated spectra , which can then be incorporated into the covariance matrix introduced in section [ subsec : likelihood ] .
here we describe the design and construction of our spectral emulator .
model library spectra are stored as ( 1-dimensional ) arrays of fluxes , sampled on high resolution wavelength grids . in the case of interest here ,
the sets of model parameters @xmath153}\}]$ ] define the dimensions of the library grid . the full spectral library , @xmath154 , is therefore encapsulated in a 4-dimensional array .
the libraries used here have grid spacings of 0.5dex in @xmath3 and 0.5 dex in @xmath4}$ ] ; the @xcite library steps by 250k in @xmath139 , but the phoenix library has finer coverage in 100k increments .
the first step in designing a spectral emulator is to break down the library into an appropriate basis @xcite .
we chose the principal component basis to decompose the library into a set of eigenspectra " , following the techniques of @xcite . prior to this decomposition ,
we isolate a subset of the library ( containing @xmath155 spectra ) with parameter values that will be most relevant to the target being considered ( e.g. , for gl 51 , this means considering only effective temperatures below @xmath623800k ) .
we then standardize these spectra by subtracting off their mean spectrum and then whiten " them by dividing off the standard deviation spectrum measured in each pixel across the grid .
the mean spectrum is @xmath156 and the standard deviation spectrum is @xmath157 ^ 2 } , \ ] ] where @xmath158 denotes the full collection of the @xmath155 sets of stellar parameters under consideration in the library and @xmath159 denotes a single set of those parameters drawn from this collection .
both @xmath160 and @xmath161 are vectors with length @xmath162 , the same size as a raw synthetic spectrum ( @xmath163 ) . in effect
, all library spectra are standardized by subtracting the mean spectrum and dividing by the standard deviation spectrum @xmath164 the eigenspectra are computed from this standardized grid using principal component analysis ( pca ; * ? ? ?
each eigenspectrum is a vector with length @xmath162 , denoted as @xmath165 , where @xmath80 is the principal component index @xmath166 .
we decided to truncate our basis to the first @xmath167 eigenspectra , where @xmath167 is decided by the minimum number of eigenspectra required to reproduce any spectrum in the grid to better than 2% accuracy for all pixels ( the typical error for any given pixel is generally much smaller than this , @xmath168 ) .
as an example , the eigenspectra basis computed for gl 51 using the phoenix library is shown in the top panel of figure [ fig : pca_reconstruct ] . using the principal component basis
, we can lossily reconstruct any spectrum from the library with a linear combination of the eigenspectra @xmath169 where @xmath170 is the weight of the @xmath171 eigenspectrum .
these weights are 3-dimensional scalar functions that depend on the stellar parameters @xmath151 . any given weight , which is generally a smooth function of the stellar parameters ( see the left panel of figure [ fig : gp_interp ] ) , can be determined at any grid point in the library by taking the dot product of the standardized synthetic spectrum with the eigenspectrum @xmath172 to simplify notation , we can write the collection of eigenspectra weights in a length-@xmath167 column vector @xmath173 and horizontally concatenate the eigenspectra into a matrix with @xmath27 rows and @xmath167 columns @xmath174 then , we can rewrite eq .
( [ eqn : reconstruct ] ) as @xmath175 where @xmath176 represents the element - wise multiplication of two vectors . to recapitulate
, the framework described above can be used to decompose the synthetic spectra in a model library into a principal component basis , allowing us to reconstruct any spectrum in the library as a ( weighted ) linear combination of @xmath167 eigenspectra .
the weights corresponding to each eigenspectrum are moderately - smooth scalar functions of the three stellar parameters , @xmath10 .
therefore , to create a spectrum corresponding to an arbitrary set of these parameters that is not represented in the spectral library , we must interpolate the weights to this new set . in practice
, it may be possible to use a traditional scheme like spline interpolation to do this directly .
however , we found that with sensitive spectra ( e.g. , for gl 51 the s / n is @xmath128400 ) , the uncertainty in the interpolated representation of the spectrum can constitute a significant portion of the total uncertainty budget .
this , combined with the under - sampling of the synthetic grid can cause artificial noding " of the posterior near grid points in the synthetic library , because the interpolated spectrum is not as good as the raw spectrum at the grid point . even explicitly accounting for interpolation error by doing
drop - out " interpolation tests and empirically propagating it forward does not relieve this noding issue .
so instead , we address this problem by employing a gaussian process to model the interpolation of the eigenspectra weights over @xmath10 , thereby encapsulating the range of possible interpolated spectra .
each weight is modeled by a gaussian process for each eigenspectrum . for a single eigenspectrum with index @xmath80
, we denote the collection of @xmath177 evaluated for all the spectra in the library as a length @xmath155 vector @xmath178 .
the gaussian process treats @xmath178 as a collection of random variables drawn from a joint multi - variate gaussian distribution @xcite , @xmath179 with @xmath180 denoting the covariances .
the kernel that describes the covariance matrix for this distribution is assumed to be a 3-dimensional squared exponential , @xmath181 \nonumber \\ & & \times \exp \left [ -\frac{(\log g_i - \log g_j)^2}{2 \ , \ell_{{\log g}}^2 } \right ] \\ & & \times \exp \left [ -\frac { ( { [ { \rm fe}/{\rm h}]}_i - { [ { \rm fe}/{\rm h}]}_j)^2}{2 \ ,
\ell_{{[{\rm fe}/{\rm h}]}}^2 } \right ] \nonumber , \label{eqn : emulator_kernel}\end{aligned}\ ] ] with hyperparameters @xmath182 , @xmath183 , @xmath184 , @xmath185}}$ ] } representing an amplitude and length scale for each dimension of @xmath10 .
unlike the matrn kernel used in section [ sec : method ] ( which produces a more structured behavior reminiscent of the spectral residuals ) , this squared exponential kernel has a smooth functional form that is more appropriate to represent the behavior of the eigenspectra weights across the library grid , as demonstrated in figure [ fig : gp_interp ] .
the @xmath186-dimensional covariance matrix is @xmath187 the evaluation of the covariance kernel for all pairings of stellar parameters at library gridpoints .
once the gaussian processes for each @xmath80 are specified , we can construct the joint distribution .
@xmath188 we use @xmath189 to denote the concatenation of @xmath178 vectors into a single length @xmath190 vector , and @xmath191 as the @xmath192 covariance matrix , @xmath193 although we could optimize the hyperparameters of each gaussian process independently based upon how well it reproduces the collection of weights for that eigenspectrum , ideally we would like to optimize the hyperparameters according to a metric that describes how well the emulator actually reproduces the original library of synthetic spectra .
following @xcite , we write down a likelihood function describing how well the reconstructed spectra match the entirety of the original synthetic grid @xmath194 \label{eqn : em_data_likelihood}\end{gathered}\ ] ] here , @xmath195 represents a length @xmath196 vector that is the collection of all of the synthetic flux vectors concatenated end to end . the precision of the eigenspectra basis representation , or the statistical error in the ability of the emulator to reproduce the known eigenspectra
is represented by @xmath197 . because we have truncated the eigenspectra basis to only @xmath167 components , where @xmath198 is much smaller than the number of raw spectra in the library
, the emulator will not be able to reproduce the synthetic spectra perfectly . by including this nugget " term in the emulator , we are also forward propagating the interpolation uncertainty for @xmath151 near or at values of @xmath152 .
we specify a broad @xmath199 function prior on @xmath197 because we expect it to be well constrained by the data .
@xmath200 where shape @xmath201 and rate @xmath202 . to facilitate the manipulation of eqn [ eqn : em_data_likelihood ]
, we create a large @xmath203 matrix that contains the all of the eigenspectra @xmath204\ ] ] where @xmath205 is the kronecker product .
this operation creates a matrix , which , when multiplied by the vector @xmath189 , enables ( lossy ) reconstruction of the entire synthetic library @xmath206 up to truncation error in the eigenspectrum basis ( @xmath197 ) . for a given @xmath197 ,
the maximum likelihood estimate for eqn [ eqn : em_data_likelihood ] is @xmath207 .
using @xmath208 , we can factorize eqn [ eqn : em_data_likelihood ] into @xmath209\\ \times \lambda_\xi^{m ( n_\textrm{pix } - m ) /2 } \exp \left [ -\frac{\lambda_\xi}{2 } { \cal f}^{\mathsf{t}}\left ( i - \phi(\phi^{\mathsf{t}}\phi)^{-1 } \phi^t \right ) { \cal f } \right ] \\\end{gathered}\ ] ] now , only the middle line of this distribution depends on @xmath208 , so we can reformulate this equation into a dimensionality reduced likelihood function and absorb the other terms into a modified prior on @xmath197 .
@xmath210\end{gathered}\ ] ] to summarize , we have reduced the dimensionality of the distribution from @xmath211 to @xmath212 although we introduced the likelihood function in eqn [ eqn : em_data_likelihood ] , we have yet to include the gaussian processes or the dependence on the emulator parameters @xmath213 .
we do this by multiplying eqn [ eqn : dimensionality_reduced ] with our prior distribution on the weights ( eqn [ eqn : weight_prior ] ) , @xmath214 and integrate out the dependence on @xmath189 .
we perform this integral using eqn a.7 of @xcite for the product of two gaussians , which yields @xmath215 \end{gathered}\ ] ] the dimensionality reduction operation changes the priors on @xmath197 ( eqn [ eqn : gamma_priors ] ) to @xmath216 @xmath217 to complete the posterior distribution for the emulator , we specify @xmath199 function priors on the gaussian process length scale kernel parameters @xmath213 .
typically , these priors are broad and peak at lengths corresponding to a few times the spacing between grid points , which helps the gaussian process converge to the desired emulation behavior .
the full posterior distribution is given by @xmath218 where the prior is given by @xmath219 } } , b_{{[{\rm fe}/{\rm h}]}}).\end{gathered}\ ] ] now that we have fully specified a posterior probability distribution , we can sample it and find the joint posteriors for the parameters @xmath197 and the @xmath213 for all @xmath80 simultaneously .
once we have identified the best - fit parameters for the emulator , we fix these parameters for the remainder of the spectral fitting . now , the emulator is fully specified and can be used to predict the values of the weights at any arbitrary set of stellar parameters @xmath151 by considering them drawn from the joint distribution @xmath220 where @xmath221 is an augmented covariance matrix that includes the point @xmath151 . to simplify notation
, we let @xmath222\ ] ] with this notation , the @xmath223 matrix @xmath224 is the region of the covariance matrix that describes the relations between the set of parameters in the grid , @xmath225 .
the @xmath226 matrix @xmath227 ( and its transpose @xmath228 ) describe the relations between the set of parameters in the grid and the newly chosen parameters to interpolate at @xmath151 .
the structure of @xmath227 is set by evaluating @xmath229 ( eqn [ eqn : emulator_kernel ] ) across a series of rows of @xmath225 like in @xmath191 , for @xmath230 , and across @xmath167 columns of @xmath151 .
@xmath231 is a @xmath232 diagonal matrix that represents @xmath229 evaluated at the zero - spacing parameter pair ( @xmath233 ) , @xmath230 .
then , to predict a vector of weights at the new location , we use the conditional probability @xmath234 where @xmath235 @xmath236 these equations are also commonly referred to as kriging equations @xcite . though the notation is complex , the interpretation is straightforward : the probability distribution of a set of eigenspectra weights @xmath237 is a @xmath167-dimensional gaussian distribution whose mean and covariance are a function of @xmath151 , conditional upon the ( fixed ) values of @xmath208 and the squared exponential hyperparameters ( an example for a single @xmath170 is shown in figure [ fig : gp_interp ] , right panel ) . if we desired actual values of the interpolated weights , for example to reconstruct a model spectrum , we could simply draw a gaussian random variable @xmath237 from the probability distribution in eq .
( [ eqn : weight_conditional ] ) .
however , because we now know the probability distribution of the weight as a function of @xmath151 , we can rewrite our data likelihood function ( eq . [ eqn : lnlikelihood ] ) in such a way that it is possible to analytically marginalize over all possible values of @xmath237 , and thus all probable spectral interpolations .
up until this point , we have described the reconstruction of a spectrum as a linear combination of the eigenspectra that characterize the synthetic library ( figure [ fig : pca_reconstruct ] ) .
but in practice , that reconstructed spectrum must be further post - processed as detailed in section [ subsec : postprocess ] . fortunately , because convolution is a linear operation , we can first post - process the raw eigenspectra according to @xmath14 , and then represent the reconstructed spectrum as a linear combination of these modified eigenspectra without loss of information . unfortunately , the doppler shift and resampling operations are not linear operations , and there will be some loss of information when trying to approximate them in this manner .
however , we find that in practice when the synthetic spectra are oversampled relative to the instrument resolution by a reasonable factor , the flux error due to resampling is smaller than 0.2% across all pixels , and thus any effect of that information loss is negligible . for notational compactness , we let @xmath238 , @xmath239 , and @xmath240 represent the post - processed eigenspectra , with an implied dependence on the current values of the extrinsic observational parameters ( @xmath241 ) and the polynomial nuisance parameters @xmath242 . now , the model spectrum is a function of the vector of eigenspectra weights @xmath243 where @xmath244 because the gaussian process describes a probability distribution of the weights , we now have a distribution of possible ( interpolated ) models and the likelihood function ( eq . [ eqn : likelihood ] ) is specified conditionally on the weights , @xmath245 the final task of designing the spectral emulator is to combine this data likelihood function with the posterior predictive distribution of the eigenspectra weights ( eq . [ eqn : weight_conditional ] ) and then marginalize over the weights @xmath246 such that we are left with a modified posterior distribution of the data that incorporates the range of probable interpolation values for the model . to perform this multidimensional integral
, we use a convenient lemma found in ( * ? ? ?
* their appendix a ) : if the probability distributions of @xmath237 and @xmath247 are specified conditionally as in eq .
[ eqn : weight_conditional ] and [ eqn : likelihood_conditional ] , respectively , then the marginal distribution ( eq . [ eqn : marginal ] ) is @xmath248 where the dependence on the model parameters is now made explicit .
we can couch this modified likelihood function in the form of eqn [ eqn : lnlikelihood ] by rewriting @xmath249 @xmath250 @xmath251 where @xmath252 can be thought of as the mean model spectrum " given the model parameters , and the covariance matrix has been modified to account for the various probable manifestations of the model spectrum about that mean spectrum .
, a. s. , reyl , c. , schultheis , m. , & allard , f. 2010 , in sf2a-2010 : proceedings of the annual meeting of the french society of astronomy and astrophysics , ed . s. boissier , m. heydari - malayeri , r. samadi , & d. valls - gabaud , 275 , d. b. , de silva , g. , freeman , k. , bland - hawthorn , j. , & hermes team .
2012 , in astronomical society of the pacific conference series , vol .
458 , galactic archaeology : near - field cosmology and the formation of the milky way , ed .
w. aoki , m. ishigaki , t. suda , t. tsujimoto , & n. arimoto , 421 | we present a modular , extensible likelihood framework for spectroscopic inference based on synthetic model spectra .
the subtraction of an imperfect model from a continuously sampled spectrum introduces covariance between adjacent datapoints ( pixels ) into the residual spectrum . for the high signal - to - noise data with large spectral range that is commonly employed in stellar astrophysics , that covariant structure can lead to dramatically underestimated parameter uncertainties ( and , in some cases , biases ) .
we construct a likelihood function that accounts for the structure of the covariance matrix , utilizing the machinery of gaussian process kernels .
this framework specifically address the common problem of mismatches in model spectral line strengths ( with respect to data ) due to intrinsic model imperfections ( e.g. , in the atomic / molecular databases or opacity prescriptions ) by developing a novel local covariance kernel formalism that identifies and self - consistently downweights pathological spectral line outliers . " by fitting many spectra in a hierarchical manner , these local kernels provide a mechanism to learn about and build data - driven corrections to synthetic spectral libraries .
an open - source software implementation of this approach is available at http://iancze.github.io/starfish , including a sophisticated probabilistic scheme for spectral interpolation when using model libraries that are sparsely sampled in the stellar parameters .
we demonstrate some salient features of the framework by fitting the high resolution @xmath0-band spectrum of wasp-14 , an f5 dwarf with a transiting exoplanet , and the moderate resolution @xmath1-band spectrum of gliese 51 , an m5 field dwarf . |
under the action of gravity a colloidal suspension sediments to form a stratified fluid .
the equilibrium density profile of the colloidal particles results from the balance between the gravitational force and thermodynamic forces as derived from the free energy of the system .
the density profiles usually exhibits a dense layer of colloidal particles at the bottom of the container above which a light cloud of colloidal particles floats . in this last regime ,
the density of particles is small enough to treat the fluid as an ideal gas . under the reasonable assumption that density gradients can be neglected , the equilibrium colloidal density obey the well known barometric law : @xmath0 here
, @xmath1 denotes the density profile of the colloidal particles , @xmath2 is the altitude and @xmath3 is the gravitational length where @xmath4 is the inverse temperature , @xmath5 is the buoyant mass of a colloidal particle and @xmath6 the intensity of the gravitational field .
this exponential law is of practical interest since it gives a prescription for the measurement of the buoyant mass @xmath5 of the particles .
however a recent experimental study of the sedimentation profiles of strongly de - ionized charged colloidal suspensions @xcite lead the authors to challenge the validity of this barometric law .
an exponential behaviour was indeed observed in the asymptotic regime , but the measured gravitational length @xmath7 could differ significantly from the expected one ( a factor of two ) .
@xmath7 was found to systematically overestimate the actual value @xmath8 , with the result that the buoyant mass measured within these experiments is systematically reduced compared to the known buoyant mass of the particles .
some theoretical efforts have been made to study this problem .
first biben and hansen @xcite solved numerically the problem in a mean field approach , but unfortunately due to numerical difficulties the samples height considered where of the order of the micron while in the experiment the samples height are of the order of the centimeter . as a consequence , the dilute region at high altitude could not be studied in this approach .
nevertheless the numerical results show a positive charge density at the bottom of the container and a negative charge at the top while the bulk of the container is neutral .
this result show that a non - zero electric field exists in the bulk of the container and acts against gravity for the colloids .
more recently one of the authors studied a two - dimensional solvable model for this problem @xcite .
this model is not very realistic ( the valency of the colloids was @xmath9 and there was no added salt ) but has the nice feature of being exactly solvable analytically .
it confirmed the condenser effect noticed for small height containers in ref .
@xcite . for large height containers it showed a new interesting phenomenon :
while there is still a positive charge density at the bottom of the container , the negative charge density is not any more at the top of the container floating but at some altitude .
interestingly , the analytical expression for the density profiles in the asymptotic regime predicts a decay in @xmath10 for the colloidal density . besides the @xmath11 factor that can not be explained by a mean field approach ,
no mass reduction is predicted by this model .
however one should be cautious when comparing two - dimensional systems to the three dimensional case because the density in not relevant in two - dimensional coulomb systems : no matter how small the density is the system is always coupled , the ideal gas regime is never attained .
for this reason a decay of the density similar to the one of an ideal gas is in itself surprising in two dimensions . lately
new results based on an approximate version of the model introduced in reference @xcite lead the authors of these studies @xcite to conclude that the mean - field approach was indeed able to predict a mass reduction in the asymptotic regime . here
we present some new results about this problem treated under the poisson - boltzmann approximation , and show that it is indeed not the case .
let us consider some colloidal particles ( for example some latex spheres ) in a solution with some amount of added salt .
in a polar solvent like water the colloids release some counterions and therefore acquire a surface electric charge @xmath12 ( @xmath13 is a entire number usually positive and @xmath14 is the charge of the electron ) .
we consider that the colloidal sample is monodisperse , all colloids have the same valency @xmath13 , and that the counterions and the salt cations are both monovalent and therefore we shall not make any distinction between cations coming from the colloids and salt cations .
we then consider a three - component system composed of colloidal particles with electric charge @xmath12 and mass @xmath5 , counterions with charge @xmath14 and coions with charge @xmath15 .
we shall neglect the masses of the counterions and coions when compared with the mass of the colloids .
the solvent shall be considered in a primitive model representation as a continuous medium of relative dielectric permittivity @xmath16 ( for water at room temperature @xmath17 ) .
the system is in a container of height @xmath18 , the bottom of the container is at @xmath19 altitude .
we consider that the system is invariant in the horizontal directions .
the density profiles of each species are denoted by @xmath1 , @xmath20 and @xmath21 ( @xmath2 is the vertical coordinate ) for the colloids , the cations and the anions respectively at equilibrium .
let us define the electric charge density ( in units of @xmath22 ) @xmath23 and the electric potential @xmath24 , solution of the poisson equation @xmath25 it is instructive to recall that the poisson - boltzmann equation can be derived from the minimization of the free energy density functional @xmath26&= & \sum_{i\in\{\mathrm{col},+,-\}}\int_0^h k_{\mathrm{b}}t \rho_i(z ) \left[\ln(\lambda_i^3\rho_i(z))-1\right]\,dz \nonumber\\ & & + \int_0^h mgz\rhocol(z)\,dz + \frac{1}{2}\int_0^h e\rho(z)\phi(z)\end{aligned}\ ] ] where @xmath27 is the de broglie wavelength of species @xmath28 .
minimization of the grand potential with respect to the densities : @xmath29 , where @xmath30 is the chemical potential of species @xmath28 , yields [ boltzmann ]
@xmath31 we shall work in the canonical ensemble , the prefactors @xmath32 which depend on the chemical potentials @xmath30 are determined by the normalizing conditions @xmath33 where @xmath34 is the total number of particles per unit area of species @xmath28 .
the system is globally neutral so we have @xmath35 .
let us introduce the following notations : @xmath3 is the gravitational length of the particles , @xmath36 is the bjerrum length , @xmath37 is the dimensionless electric potential and @xmath38 .
@xmath39 are the debye lengths associated to the counterions and the coions and @xmath40 is the debye length associated to the colloidal particles . for a quantity @xmath41 depending on the altitude , let us define its mean value @xmath42 . with these notations equations ( [ poisson ] ) and ( [ boltzmann ] ) yield the modified poisson - boltzmann equation @xmath43 from eq .
( [ poisson - boltzmann ] ) it is clear that the problem has the following scale invariance : if @xmath44 is a solution of ( [ poisson - boltzmann ] ) then @xmath45 is a solution of the problem with the rescaled lengths @xmath46 and @xmath47 .
the advantage of the density functional formulation of the problem is that it allows for systematic corrections to the poisson - boltzmann approximation .
for instance , one may be interested in the effect of the finite size of the macroions .
let @xmath48 be the diameter of the colloids , @xmath49 the volume fraction of colloids and @xmath50 the effective densities of the microions .
then , the finite size of the colloids can be accounted in a local density approximation ( lda ) by adding to the free energy density functional ( [ freeenergy ] ) the free energy excess term @xmath51 where @xmath52 is the excess free energy of a hard sphere fluid derived by the carnahan starling equation of state @xcite @xmath53 and by replacing in ( [ freeenergy ] ) the entropy term of the microions by @xmath54 minimization of this new free energy functional gives the modified version of eqs .
( [ boltzmann ] ) [ boltzmann - mod ] @xmath55 \\ \rho_+(z)&=&\rho_+^0 ( 1-\etacol(z))\exp(-\beta e \phi(z))\\ \rho_-(z)&=&\rho_-^0 ( 1-\etacol(z))\exp(\beta e \phi(z))\end{aligned}\ ] ] there are several length scales in this problem : the gravitational length of the colloids @xmath8 , the debye or screening length , the height @xmath18 of the container and eventually the hard core diameter of the particules . in a realistic case @xmath18 is of the order of the centimeter , @xmath8 of the order of @xmath56 mm , the screening length of the order of 10 nm .
we are faced to a practical numerical problem , when we will transpose the problem to a lattice , the lattice spacing should be smaller than all the physical lengths , but since @xmath18 is much larger than the others lengths , the number of sites in the lattice should be very high ( of order @xmath57 ) . a possible approach to deal with
this problem is to study small containers as in ref .
@xcite . in this paper
we want to study the case of high containers so we will consider very deionized systems in which the screening length is of the order of 0.1 mm , much larger than in usual physical cases , the other lengths taking `` physical '' values .
that way the necessary number of points in the lattice will be reasonable ( a few hundreds ) . also , since the screening length is so large , the hard core of the macroions will not change the results drastically from the case of point particles so we will concentrate from now on on the poisson - boltzmann problem for point particles ( eq . [ poisson - boltzmann ] ) .
equation ( [ poisson - boltzmann ] ) is solved numerically by an iterative method @xcite . using the green function of the one - dimensional laplacian @xmath58 the poisson equation ( [ poisson ] )
can be written as @xmath59 starting with an arbitrary electric potential , one can compute the corresponding density profiles using eqs .
( [ boltzmann ] ) and derive a new electric potential using eq .
( [ poisson - integrale ] ) , then reiterate the process until a stationary solution is attained .
in practice instead of using the new potential directly for the next iteration a mixing of the old and new densities is used . as stated before we have to consider very deionized systems
in which the bjerrum length @xmath60 is smaller than @xmath61 ( the physical value of @xmath60 for water at room temperature is 7 ) . figure [ fig : tous ] shows the density profiles of each species , the charge density , the electric potential and the electric field profile of a typical sample with the following parameters : @xmath62 , @xmath63 mm , @xmath64 mm , @xmath65 , a salt concentration @xmath66
mmol / l and a mean colloidal volume fraction @xmath67 ( we consider that the particles have a hard core diameter @xmath68 nm to express the colloidal density as a volume fraction in order to use units familiar with the experiments but we do not account for hard core effects in the poisson
boltzmann equation ) .
the log plot of colloidal density profiles is similar to the experimental ones @xcite . in the bottom
there is a slow decay whereas at high altitudes there is a faster barometric decay . since we did not take into account the hard core of the particles in the theory we do not find the discontinuity in the density profiles near the bottom of the sample observed in the experiments @xcite , due to the phase transition of the colloids from an amorph solid to a fluid . at very low altitudes ( near the bottom , very high volume fractions ) the poisson
boltzmann theory is not valid .
the charge density profile confirms the results of ref .
@xcite , that there is a strong accumulation of positive charges at the bottom of the container while there is a cloud of negative charge density floating at some altitude @xmath69 .
there are clearly two neutral regions in the container : one at low altitude between the positive charge density at the bottom and the negative cloud , in which a non - vanishing electric field exists and a second neutral region at high altitude , over the negative cloud .
the electric field in the lower region acts against gravity for the colloids therefore , as seen in the log plot of the colloids density profile , the decay is much slower than the one for an ideal neutral gas .
numeric results for other series of samples suggest that this electric field is proportional to @xmath70 .
in the upper region the colloidal density drops exponentially as @xmath71 since the electric potential is almost constant and the electric field vanishes .
since the different densities vary with the altitude we can define a local screening length which depends on the altitude by @xmath72 the two regions of the sediment are caracterized by a very different behavior of this local screening length . in the lower region
the colloidal density is so high that @xmath73 . in that region
the colloids dominate the screening length . on the other hand in the upper region
the colloidal density is very small and salt now controles the screening length which is then constant since at high altitudes the cations and anions densities are almost constant and equal to the mean salt concentration as seen in figure [ fig : tous ] .
it is interesting to notice that electric charges accumulate in the intermediate region around @xmath69 where there is a change of regime , in agreement with macroscopic electrostatics principles .
the preceding remark allows us to understand how the physical parameters ( mean volume fraction of colloids , mass of the colloids , amount of added salt ) will modify the altitude @xmath69 which separates the two regions .
for example if we add more salt , @xmath69 will diminish since we reach sooner the regime where @xmath74 .
we have computed the density profiles in several other cases changing the values of the parameters in order to find the depency of @xmath69 in these parameters .
our numerical results suggest that @xmath75 with @xmath76 and @xmath77 .
the preceding equation can be written in a more attracting way , introducing the screening length associated to the salt @xmath78 and the effective screening length associated to the colloids @xmath79 , as @xmath80 with @xmath81 .
we do not consider here boundary effects : this equation is only valid if @xmath69 is smaller than @xmath18 .
the finite height @xmath18 of the container will have the effect to `` push '' the negative cloud downwards if the parameters are such that @xmath69 approaches the top of the container .
the same holds for the bottom of the container if @xmath69 is too small .
another quantity of interest is the size @xmath82 of the negative cloud , defined as the mid - height width of the negative peak in the charge density profile ( see figure [ fig : tous ] ) .
since we know that at @xmath69 altitude , @xmath83 is of the same order of magnitude as @xmath84 , the screening length at that altitude is proportional to @xmath85 . from basic electrostatics
we know that the system will only tolerate charges over a length of order of the screening length , we deduce that @xmath82 is proportional to @xmath85 .
in fact the numerical results suggest also a linear dependency of @xmath82 in @xmath8 : @xmath86 with @xmath87 , @xmath88 , and the same restrictions concerning boundary effects as for the equation for @xmath69 .
as we mentioned before , at high altitudes ( larger than @xmath69 ) the electric potential is almost constant and the electric field vanishes . from this
it is clear that the colloidal density will decay as @xmath71 and there is no apparent reduced mass .
nevertheless let us notice that in the regime where the electric potential is almost constant in our calculations the corresponding colloidal volume fraction is smaller than @xmath89 .
such volume fractions can not be measured experimentally . in practice the optical methods used in @xcite allow to measure only volume fractions larger than @xmath90 .
a possible explanation to the apparent mass observed in the experiments is that for volume fractions higher than @xmath90 the asymptotic regime where the electric field vanish have not been reached yet : there is a residual electric field responsible of the observed reduced mass . in order to test this hypothesis we made a log plot of several colloidal volume fraction profiles restricting the plot to volume fractions higher than @xmath90 ( figure [ fig : pentes ] ) .
we computed the slope of the wing of the colloidal density to find an effective gravitational length @xmath7 which is higher than the actual gravitational length @xmath8 as observed in the experiments .
futhermore when we plot the colloidal volume fraction profile and the corresponding electric field profile together ( figure [ fig : logeta - et - e ] ) we notice that for volume fractions higher than @xmath90 the electric field is not zero . the different plots in figure [ fig : pentes ] where obtained using different salt concentrations , so the sediment height ( which is proportional to @xmath69 ) varies .
in this case we found that the apparent mass is a decreasing function of the height of the sediment , in agreement with the experiments .
however the sediment height can be changed by changing other parameters like the mean colloidal density or their valency @xmath13 .
computing the apparent gravitational length @xmath7 as defined before for other series of samples , we found that the apparent gravitational length @xmath7 does not depend on @xmath13 or the mean colloidal density . in our model
the ratio @xmath91 is only a function of the salt density .
figure ( [ fig : lgstar - lambdasel ] ) shows the ratio @xmath91 as a function of salt screening length @xmath85 .
as mentionned in the introduction the model presented above has motivated several studies both numerically @xcite and analytically @xcite .
the purpose of this section is to compare our numerical results with the most achieved version of the theory presented in reference @xcite .
this theoretical approach is based on a constrained minimization of the free energy functional ( [ freeenergy ] ) assuming an exponential _ ansatz _ for the density profiles : [ hartmut ] @xmath92 with this parametrization @xmath93 is the ratio of the reduced mass @xmath94 by the buoyant mass @xmath5 of a colloidal particle , @xmath95 denotes the fixed salt concentration , and @xmath96 is the fixed overall colloidal density per unit area , i.e. @xmath97 . the system considered in @xcite is semi - infinite , @xmath19 corresponds to the bottom of the sample and @xmath98 . @xmath99 and @xmath100 are the two variational dimensionless parameters of the theory , and the equilibrium density profiles @xmath1 and @xmath21 correspond to the values of @xmath99 and @xmath100 that minimize the free energy functional ( [ freeenergy ] ) .
following reference @xcite , the minimization conditions express : [ hartmut_min ] @xmath101 where @xmath102 is the coupling parameter ( @xmath60 is the bjerrum length introduced previously ) , and @xmath103 is the relative amount of added salt .
function @xmath104 is defined by @xmath105 .
although equations ( [ hartmut_min ] ) require a numerical treatment , it is possible to extract asymptotic expressions when the coupling parameter @xmath106 is vanishingly small ( strong gravitational coupling regime ) or large compared to unity ( strong coulomb coupling regime ) .
such an analysis is presented in reference @xcite , and we only reproduce here the main features . when gravitational coupling is strong , @xmath107 , the reduced mass is given by @xmath108 ( forall values of the salinity @xmath109 ) and
therefore no mass reduction is observed in this regime ( in agreement with the numerical results presented in reference @xcite ) . on the contrary , in strong coulomb coupling regimes , @xmath110 ,
quite a large mass reduction is predicted , even in low salinity regimes @xmath111 ( in such a situation the mass reduction is given by @xmath112/(z+1 ) + o(1/\gamma)$ ] ) .
our numerical results based on a free minimization of the functional ( [ freeenergy ] ) show that it is indeed not the case , even though we observe nice exponential asymptotic behaviours at high altitudes . to emphasize this point we present in table [ table : res ] data obtained mostly in the strong coulomb coupling regime .
excepted for the first value @xmath113 , which corresponds to the opposite strong gravitational coupling regime , the agreement between the theory presented in reference @xcite and the present free minimization is very poor .
the effective mass predicted by the free minimization procedure corresponds to the actual mass within less than 0.1% in most situations , whereas the theory predicts effective masses that are only 2% of the actual mass ! to our opinion , the failure of the parametrization presented in reference @xcite is not due to the exponential _ ansatz _ itself , but to the constraint of global charge neutrality applied to the asymptotic regime .
the theory presented in reference @xcite assumes that the profiles are exponential from the bottom of the sample to the top , as a result the free energy functional ( [ freeenergy ] ) has to be minimized with the constraint of global charge neutrality .
however , the actual situation is quite different . if we refer to the experimental work done by piazza et al .
@xcite , the exponential regime is reached only above a macroscopic layer of strongly interacting colloidal particles .
data presented in the previous section ( see e.g. figure [ fig : tous ] ) resulting from a free minimization of the functional also exhibit a dense macroscopic layer of colloidal particles in the bottom of the cell , and these profiles can not be simply represented by a single exponential .
this feature can be incorporated to the model suggested by lwen , by splitting the cell into two parts .
the upper part of the cell ( above a given altitude `` @xmath114 '' ) corresponds to the asymptotic region where the profiles can be accurately represented by an exponential , whereas below @xmath114 the profiles are more complicated .
as we can see , @xmath114 is defined by the condition that the profiles are exponential above it .
there is then no upper bound on the value of @xmath114 and the asymptotic profiles should not depend on its precise value . as a result @xmath114
can be chosen arbitrarily large . as a consequence ,
the part of the fluid located below @xmath114 can be considered as a reservoir fixing the chemical potential of the ionic species @xmath115 and @xmath116 ( @xmath117 is irrelevant since the local density @xmath20 is held fixed , and is thus not a variational parameter ) .
although the full system must be charge neutral , the asymptotic part above @xmath114 has no reason to be neutral .
we are then lead to minimize the free energy of the upper part of the cell in the grand - canonical ensemble . assuming that parametrization ( [ hartmut ] ) is valid above @xmath114 the minimization equation associated to the colloidal particles reads : @xmath118\over\partial a } = \mu_{\mathrm{col}}{\partial n_{\mathrm{col}}\over\partial a}\ ] ] where @xmath119 $ ] is now the free energy functional above @xmath114 : @xmath120&= & \sum_{i\in\{\mathrm{col},+,-\}}\int_{z_o}^{+\infty } k_{\mathrm{b}}t \rho_i(z ) \left[\ln(\lambda_i^3\rho_i(z))-1\right]\,dz \nonumber\\ & & + \int_{z_o}^{+\infty } mgz\rhocol(z)\,dz + \frac{1}{2}\int_{z_o}^{+\infty } e\rho(z)\phi(z)\end{aligned}\ ] ] and @xmath121 is the number of colloidal particles above @xmath114 per unit area . after some algebra , this minimization equation can be written on the form : @xmath122 = 0\end{aligned}\ ] ] where @xmath123 .
we can easily check that when @xmath124 we recover the first equation of condition ( [ hartmut_min ] ) . as @xmath125 can be chosen arbitrarily large in our model
, we easily see that this equation implies @xmath126 ( no mass reduction ) and : @xmath127 this new version of the theory is consistent with our numerical results predicting no mass reduction .
a free minimization of the poisson boltzmann theory used in references @xcite have been performed in this article which lead us to conclude that this simple mean field theory does not predict any mass reduction contrarily to previous approximate minimization of the same functional .
these new results are fully consistent with the analytical results obtained in a two - dimensional case by tllez @xcite . in particular
, we observe the same condenser effect between the bottom of the container and the top of the dense region , resulting from a competition between electroneutrality and entropy of the microions .
data plotted in figure [ fig : pentes ] give a possible explanation for the experimental results obtained by piazza et al . @xcite .
although in the asymptotic regime we observe no mass reduction , this regime is attained for very low values of the colloidal packing fractions , below the experimental resolution ( @xmath90 ) in some situations . as a result
, the residual electrostatic field can affect the profiles resulting in an apparent effective mass . | we investigate the sedimentation equilibrium of a charge stabilized colloidal suspension in the regime of low ionic strength . we analyze the asymptotic behaviour of the density profiles on the basis of a simple poisson boltzmann theory and
show that the effective mass we can deduce from the barometric law corresponds to the actual mass of the colloidal particles , contrary to previous studies .
# 1 |
protoplanetary disks are believed to be the birthplaces of planets ( e.g. , * ? ? ?
* ) ; hence , understanding the evolution of these disks guides our understanding of the process of planet formation .
disks which have substantial infrared excesses but reduced fluxes at wavelengths @xmath920 @xmath10 m , i.e. , transitional disks @xcite , could be related to the early phases of planet formation ( see a recent review of * ? ? ? * ) and are therefore particularly important for understanding how , where , and when planets form . for many transitional disks , partial inner holes or partial gaps
have been directly resolved by interferometry at ( sub)millimeter wavelengths ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and imaging at near - infrared wavelengths @xcite .
numerous mechanisms have been proposed to explain the clearing of gaps in transitional disks , including grain growth ( e.g. , * ? ? ?
* ) , photoevaporation ( e.g. , * ? ? ? * ) , and gravitational interactions with orbiting planets ( e.g. , * ? ? ?
* ; * ? ? ?
two possible methods to distinguish the disk - planet interactions from other aforementioned proposed gap - clearing mechanisms could be the detection of ( 1 ) a planetary companion in the inner hole / gap region ( e.g. , * ? ? ? * ) or ( 2 ) a ring - like gap between optically thick inner and outer disks ( i.e. , pre - transitional disk ; * ? ? ?
* ) because dynamical formation of wide gaps could be the only surviving mechanism for wide gapped disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ) one good candidate to investigate the inner hole / gap region at tens au in the disk in pre - transitional disks is the weak - lined t tauri star pds 70 ( k5 type ; 0.82@xmath11 ; @xmath1210 myr ; * ? ? ? * ; * ? ? ?
* ; * ? ? ?
a scattered light disk with a radius at 14 to 140 au was detected by @xmath13-band imaging @xcite .
the possible presence of inner and outer disks with different temperatures were suggested by @xcite and @xcite , which may imply that pds 70 is a pre - transitional disk object . in this @xmath14 , we present high resolution imaging of pds 70 with subaru / hiciao and gemini / nici .
observations of pds 70 were conducted with hiciao @xcite on the subaru telescope in polarized differential imaging ( pdi ) mode , combined with angular differential imaging ( adi ) mode @xcite , on 2012 february 28 ut .
@xmath0-band linear polarization images were taken under the program seeds ( strategic explorations of exoplanets and disks with subaru ; * ? ? ?
* ) . in pdi+adi , we employed a double wollaston prism to split incident light into four images , each has 5@xmath6 by 5@xmath6 field of view with a pixel scale of 9.5 mas / pixel , to make the saturated radius as small as possible .
polarization was measured by rotating the half waveplate to four angular positions ( in the order of 0@xmath15 , 45@xmath15 , 22.5@xmath15 , and 67.5@xmath15 ) .
we obtained 22 full waveplate rotation cycles , taking a 15-s exposure per waveplate position .
the total integration time of the @xmath1 image was 660 s. even with a large airmass of @xmath72 during observations of pds 70 , the adaptive optics system ( ao188 ; * ? ? ? * ) provided a stellar psf of pds 70 with fwhm of 0.@xmath61 in the @xmath0 band .
the angle of the total field rotation was @xmath713@xmath15 .
the polarimetric data were reduced with the same procedure as for @xcite using iraf . since the strehl ratio of the stellar psf of pds 70 was @xmath70.26 , a stellar halo remained in the image , i.e , the psf convolved by seeing was not perfectly corrected by the ao and partially remains .
this halo appears to be polarized along the minor axis of disks since the polarization of forward and back scattering is smaller due to the deviation from 90@xmath15 in the scattering angle , and then the stellar halo has a significant net - polarization . in other words , since the flux of a psf convolved by seeing contains contributions from both the central star and disk , the seeing - psf might be polarized along the minor axis of a disk . to correct for this @xmath16polarized halo@xmath17
, we first derived the net - polarizarion of pds 70 with aperture polarimetry .
we then made a model @xmath16polarized halo@xmath17 of pds 70 by multiplying the added image of @xmath18- and @xmath19-rays by the derived polarization of @xmath20 with @xmath21 .
the errors were calculated with the photometric errors , and are thus a lower limit . finally , we subtracted the model @xmath16polarized halo@xmath17 and obtained final stokes @xmath22 and @xmath23 images .
figure [ f1 ] demonstrates the subtraction of the @xmath16polarized halo@xmath17 .
we carried out adi observations of pds 70 in the @xmath3 band on 31 march 2012 , using the near - infrared coronagraphic imager ( nici ) and the 85-element ao system @xcite mounted on the 8.1 m gemini south telescope at cerro pachon , chili .
nici utilizes a 1024 @xmath24 1024 aladdin insb array with an plate scale of 18 mas / pix and field of view of 18 @xmath24 18 arcsec .
we obtained 145 frames for pds 70 with 0.76-s @xmath24 25 coadd for @xmath3 band .
although the sky was clear for the full night , the seeing was variable , therefore a subset of 95 frames were combined to produce the final image .
the total exposure time was 1805 sec and the spatial resolution was achieved 0.@xmath611 .
the angle of the total field rotation is @xmath7100@xmath15 .
the data reduction was performed with the loci algorithm @xcite in annular regions of 1000 @xmath24 fwhm ( @xmath25 1000 ) with reference images selected from frames with at least 0.5 @xmath24 fwhm field rotation ( @xmath26 0.5 ) .
a large optimization area diminishes the impact of the disk on the optimization process @xcite .
the ratio of radial and azimuthal width of the optimization area ( @xmath27 ) is 1 .
figure [ f2 ] shows @xmath0-band pi images and the @xmath3-band loci image of pds 70 assuming a distance of 140 pc , along with radial surface brightness profiles .
we find a clear elliptical ring in the @xmath0-band , which has not been reported in previous high - resolution imaging @xcite .
a partial elliptical disk is observed in the @xmath3-band , due to the inevitable loss of flux in the process of loci ; hence , we derive radial profiles of suface brightness based on the @xmath0-band pi image only and companion mass limits from the @xmath3-band loci image only .
we consider that the ellipse shape is due to the system s inclination , and show the results of fitting an ellipse to these data in table [ table1 ] .
the position angle of the major axis and the inclination of the disk are similar with those of @xmath28-band imaging ( pa @xmath29 and @xmath30 ; * ? ? ?
we measured an offset of 44 @xmath31 3 mas ( @xmath76 au ) at pa=87.9@xmath15 between the geometric center of the disk and the central star .
the positional accuracy of the central star is 1.5 mas ( 0.2 au ) .
the direction of this offset is roughly consistent with that of the minor axis of 68.6@xmath15 , and the sign of this offset indicates that the southwest side is inclined toward us ( i.e. the near side ) ( see the model image in @xcite ) .
this geometry is also consistent with the facts that ( 1 ) the northeast side of the disk is wider due to the back illumination of the wall @xcite , and ( 2 ) the southwest side is brighter than the northwest side due to forward scattering @xcite .
assuming the cavity edges correspond to the peak pi of the disk , a radius of the cavity is measured as @xmath770 au . the outer radius of the disk is measured to be @xmath7140 au in figure [ f2](d ) , and corresponds to the location at which our sensitivity is no longer sufficient to detect extended emission .
we also found a flux deficit of @xmath1 in the direction of the minor axis .
since the observed scattering angle at the minor axis deviates from 90@xmath15 , the polarization fraction along the minor axis is lower .
the @xmath1 at the minor axis is therefore lower than at the major axis , and such hole - like structures are similar to those discussed in @xcite .
we checked the proper motion of the companion cadidate to the north of pds 70 reported by @xcite .
pds 70 has a proper motion of @xmath32 mas / yr @xcite ; the separation between pds 70 and the companion cadidate should increase if the companion cadidate is a background star . since the separation in hiciao and nici images are 324.13 @xmath31 0.15 au and 324.44 @xmath31 0.10 au , respectively , the separation in @xcite is inferred to be @xmath33 au .
our estimation has a good agreement with the actual observed separation of 301.75 @xmath31 0.06 au in @xcite , and therefore , we concluded that the companion cadidate is a background star . since the follow - up @xmath3-band observations with gemini / nici failed to detect any significant signals of point - like sources in the gap we put constraints of upper limits for companion(s ) .
figure [ f2](f ) shows the detectable masses of companions at 5@xmath34 .
the loci parameter of the optimization area is 250 @xmath24 fwhm , which is different from that described in sec .
[ loci ] . for that
, we first applied a median filter with 0.@xmath611 width to the image , and then calculated the standard deviation as a noise level in concentric annuli along the major axis of the disk .
the mass was calculated by assuming the cond evolutionary model @xcite , a distance of 140 pc , and an age of 10 myr @xcite .
we took into account the flux loss due to the partial self - subtraction by testing how point sources are affected by loci .
the detectable mass limit is tens @xmath8 , therefore , stellar companions down to masses associated with massive brown dwarfs are excluded within the gap .
although the sed fitting for pds 70 has been performed in previous studies @xcite , both the availability of new archival photometric data ( table [ table2 ] ) and our imaging results motivates us to revisit the sed of the system using monte carlo radiative transfer ( mcrt ) .
note that though pds 70 has been observed by spitzer irs , the object was mispointed by @xmath72.3@xmath6 with components along both the spatial and dispersion axes , therefore , the irs spectrum of pds 70 is not used in this work . * setup for modeling . * the method for our mcrt simulations is described in @xcite and ( * ? ? ? * in prep . ) . in a subsequent paper ii (
* in prep . )
, we will perform a detailed radiative transfer modeling of both the sed and the seeds imagery , and present a fiducial disk+cavity model which reproduces both observations well .
we will also explore the parameter space around the fiducial model in paper ii , and provide a full discussion on the constraints on the various model parameters there . in this letter
, we only briefly describe the fiducial model and its resulting sed .
we note that this model is not a _
best _ fitting model in an absolute sense , since a full @xmath35 fitting of the observations with all the free parameters in a protoplanetary disk is essentially impossible @xcite .
however the constraints on many parameters such as the cavity size , depletion , and the surface density of the inner disk are reasonably tight , as will be shown in paper ii .
our model contains a cavity 70 au in radius .
the surface density both inside and outside the cavity decreases with radius as @xmath36 , where @xmath37 au , while @xmath38 inside the cavity is reduced to @xmath39 the extrapolated value from the outer disk , with @xmath40 being the depletion factor .
the temperature structure of the disk is determined from the radiative transfer calculations .
the inner edge of the disk is self - consistently determined at the dust sublimation temperature ( @xmath41 k ) .
we ignore accretion in the model , as suggested by its nature of being a weak line t tauri star @xcite .
we use a pre - main sequence star of spectral type k5 , radius 1.39 @xmath42 , mass 0.82 @xmath43 , and temperature 4500 k for the central source , as suggested by @xcite and @xcite .
the disk has a gaussian density distribution in the vertical direction , with scale heights @xmath44 as input parameters .
two disk components are included : a thick disk with small grains ( sub - micron - sized ) , representing the pristine ism - like dust ; and a thin disk with large grains ( up to @xmath7mm - sized ) and 20% of the scale height of the small grains , as the result of grain growth and settling @xcite .
most of the dust mass ( 0.967 ) is in the settled disk , and the total dust - to - gas ratio is assumed to be 1/100 .
the sed is produced assuming a disk inclination angle @xmath45 .
the other parameters are summarized in table [ table1 ] .
* sed fitting . *
figure [ f3 ] shows the good agreement between our model sed and available photometric data . as we will show in paper ii , the thermal emission from the cavity wall at @xmath770 au peaks at @xmath46 m .
the wall emission is a major signature of ( pre-)transitional disk sed ( e.g. , * ? ? ?
the surface density of small dust is @xmath47 g @xmath48 at 0.1 au , and the opacity of small dust is @xmath49@xmath50 g@xmath51 at @xmath52 m ( roughly the peak of the stellar radiation ) .
this makes the inner disk optically thick in the vertical direction .
since we assume a surface density profile decreasing with increasing radius , when moving out the cavity gradually becomes vertically optically thin , and the transition happens at @xmath53 au if only taking into account the small dust ( the existence of big dust inside the cavity is poorly determined from sed and scattered light image ) . as we will discuss in detail in paper ii , this optically thick inner disk is needed to explain the @xmath72 - 40 @xmath10 m sed , as models with optically thin inner disk fail to reproduce the sed .
following the convention in the literature @xcite , we classify pds 70 as a pre - transitional disk object .
grain growth is one mechanism which can potentially form disk gaps ( e.g. , * ? ? ?
it is capable of reproducing the sed of transitional disks but not observed millimeter - wavelength images .
however , the sharpness of pds 70 s @xmath0-band gap edge suggests grain growth is unlikely to be the primary reason , since the models generally predict smooth gap edge features ( e.g. , * ? ? ?
photoevaporation is another possible mechanism .
a photoevaporative wind can prevent outer disk material from feeding the inner disk ; without this supply , the inner disk material will rapidly accrete onto the central star , creating a cavity inside out @xcite .
although a pre - transitional disk - like structure could be produced for a very short period ( see figure 1 in * ? ? ?
* ) , it is quite unlikely to observe such a snap - shot by coincidence , so that it may be difficult to explain the existence of the optically thick inner disk inferred from the sed .
dynamical interactions with ( sub)stellar companions ( e.g. , * ? ? ?
* ) or orbiting planets ( e.g. , * ? ? ?
* ; * ? ? ?
* ) are also potential mechanisms , as several binary systems ( e.g. , coku tau 4 ; * ? ? ?
* ) are known to have a cavity in their circumbinary disk , and simulations show that multiple planets can open wide gaps @xcite , furthermore , the survival of an optically thick inner disk within a few au in @xcite mimics the inner disk of pds 70 .
our @xmath3-band observation has only ruled out companions with a mass of @xmath5450@xmath8 in the gap region .
therefore , dynamical interactions with low mass brown dawarfs or giant planets may be the origin of the gap .
to put a futher constraint of an upper limit mass , future observations should be pursued , such as the aperture - masking interferometric observations ( e.g. , * ? ? ?
* ) , which has a better contrast than loci imaging .
since pds 70 harbors the pre - transitional disk with one of the largest gaps , this object may be one of the best candidate for future high - contrast planet imagers .
we are grateful to an anonymous referee for providing many useful comments leading to an improved paper .
we appreciate support from the subaru telescope and the gemini south telescope staff , especially from jennie berghuis and dr .
tom hayward .
this work is partly supported by a grant - in - aid for science research in a priority area from mext , by the mitsubishi foundation , and by the u.s .
national science foundation under awards no .
1009203 and 1009314 .
cc + diameter of the major axis ( au ) & 137.0 @xmath31 1.1 + diameter of the minor axis ( au ) & 88.6 @xmath31 0.5 + position angle of the major axis ( @xmath15 ) & 158.6 @xmath31 0.7 + inclination by ellipse fitting ( @xmath15 ) & 49.7 @xmath31 0.3 + geometric center ( au : au ) & ( @xmath55 @xmath31 0.3 : 0.2 @xmath31 0.4 ) + + @xmath56 & 0.003 + @xmath57 & 0.967 + @xmath58(100au ) & 2 + @xmath59 & 1.2 + @xmath60(100au ) & 2 + @xmath61 & 1.2 + @xmath62 & @xmath63 + @xmath64(100au ) & 10 + @xmath65 & 1.2 + @xmath66(100au ) & 10 + @xmath67 & 1.2 + @xmath68 & @xmath63 + lccc wavelength & @xmath69 ( mjy ) & note & plotted color + & & & in fig .
[ f3 ] + @xmath23 & 9.7 & @xcite & cyan + @xmath70 & 41.8 & @xcite & cyan + @xmath71 & 99.2 & @xcite & cyan + @xmath72 & 160.9 & @xcite & cyan + @xmath73 & 216.3 & @xcite & cyan + 2mass ( @xmath74 ) & 311.9 @xmath31 6.9 & @xcite & magenta + 2mass ( @xmath0 ) & 342.7 @xmath31 12.6 & @xcite & magenta + 2mass ( @xmath13 ) & 275.6 @xmath31 5.8 & @xcite & magenta + wise ( 3.4 @xmath10 m ) & 188.1 @xmath31 4.0 & @xcite & green + wise ( 4.6 @xmath10 m ) & 142.0 @xmath31 2.6 & @xcite & green + wise ( 12 @xmath10 m ) & 153.9 @xmath31 2.3 & @xcite & green + wise ( 22 @xmath10 m ) & 341.8 @xmath31 0.7 & @xcite & green + akari ( 9 @xmath10 m ) & 201.2 @xmath31 25.8 & vizier ii/297 & black + akari ( 18 @xmath10 m ) & 209.8 @xmath31 13.4 & vizier ii/297 & black + akari ( 90 @xmath10 m ) & 851.1 @xmath31 62.6 & vizier ii/298 & black + iras ( 12 @xmath10 m ) & 251 @xmath31 25.1 & @xcite & blue + iras ( 25 @xmath10 m ) & 348 @xmath31 27.8 & @xcite & blue + iras ( 60 @xmath10 m ) & 884 @xmath31 61.9 & @xcite & blue + mips ( 24 @xmath10 m ) & 349.7 @xmath31 7.0 & spitzer heritage archive & red + mips ( 70 @xmath10 m ) & 1049.9 @xmath31 19.9 & spitzer heritage archive & red + mips ( 160 @xmath10 m ) & 873.0
@xmath31 36.2 & spitzer heritage archive & red + andrews , s. m. et al .
2011 , , 732 , 42 alexander , r. d. , clarke , c. j. , & pringle , j. e. 2006 , mnras , 369 , 229 artymowicz , p. & lubow , s. h. 1994 , , 421 , 651 baraffe , i. chabrier , g. , barman , t. s. , allard , f. , & hauschildt , p. h. 2003 , , 402 , 701 bessell , m. s. , castelli , f. , & plez , b. 1998 , a&a , 333 , 231 birnstiel , t. , andrews , s. m. , & ercolano , b. 2012 , arxiv:1206.5802 buenzli , e. , thalmann , c. , & vigan , a. et al .
2010 , , 524 , l1 clarke , c. , gendrin , a. , & sotomayor , m. 2001 , mnras , 328 , 485 chun , m. et al .
2008 , proc .
spie , 7015 , 49 cohen , m. , wheaton , w. a. , & megeath , s. t. 2003 , aj , 126 , 1090 cutri , r. m. et al .
2003 , 2mass all - sky catalog of point sources , vizier online data catalog ii/246 cutri , r. m. et al .
2003 , wise all - sky data release , vizier online data catalog ii/311 dong , r. et al .
2012 , , 750 , 161 dong , r. et al .
2012 , in prep .
dullemond , c. p. & dominik , c. 2004 , , 417 , 159 dullemond , c. p. & dominik , c. 2004 , , 421 , 1075 dullemond , c. p. & dominik , c. 2005 , , 434 , 971 espaillat , c. , calvet , n. , & dalessio , p. et al .
2007 , , 670 , l135 fukagawa , m. , tamura , m. , itoh , y. , kudo , t. , imaeda , y. , oasa , y. , hayashi , s. s. , & hayashi , m. 2006 , , 636 , l153 gregorio - hetem , j. , lepine , j. r. d. , quast , g. r. , torres , c. a. o. , & de la reza , r. 1992 , , 103 , 549 gregorio - hetem , j. , & hetem , a. 2002 , mnras , 336 , 197 hashimoto , j. et al .
2011 , , 729 , l17 hayano , y. et al .
2004 , , 5490 , 1088 hayashi , c. , nakazawa , k. , & nakagawa , y. 1985 , in protostars and planets ii , ed .
d. c. black , & m. s. matthews ( tucson , az : univ .
arizona press ) , 1100 ireland , m. j. & kraus , a. l. 2008 , , 678 , l59 jarrett , t. h. et al .
2011 , , 735 , 112 kraus , a. l. & ireland , m. j. 2012 , , 745 , 5 lafrenire , d. , marois , c. , doyon , r. , nadeau , d. , & artigau , e. 2007 , , 660 , 770 marois , c. , lafrenire , d. , doyon , r. , macintosh , b. , & nadeau , d. 2006 , , 641 , 556 mathis , j. s. 1990 , ara&a , 28 , 37 mathews , g. s. , williams , j. p. , & mnard , f. 2012 , , 753 , 59 metchev , s. a. , hillenbrand , l. a. , & meyer , m. r. 2004 , , 600 , 435 moshir , m. 1989 , iras faint source catalog , vizier online data catalog ii/156a muto , t. et al .
2012 , , 748 , l22 nakajima , t. et al .
1989 , , 97 , 1510 papaloizou , j. c. b. , nelson , r. p. , kley , w. , masset , f. s. , & artymowicz , p. 2007
, in protostars and planets v , ed .
b. reipurth , d. jewitt , & k. keil ( tucson : univ .
arizona press ) , 655 perrin , m. d. , schneider , g. , duchene , g. , pinte , c. , grady , c. a. , wisniewski , j. p. , & hines , d. c. 2009 , , 707 , l132 pitu , v. , dutrey , a. , guilloteau , s. , chapillon , e. , & pety , j. 2006 , , 460 , l43 riaud , p. , boccaletti , a. , baudrand , j. , & rouan , d. 2006 , , 458 , 317 roeser , s. , demleitner , m. , & schilbach , e. 2010 , , 139 , 2440 strom , k. m. , strom , s. e. , edwards , s. , cabrit , s. , & skrutskie , m. f. 1989 , , 97 , 1451 tamura , m. et al .
2006 , , 6269 , 28 tamura , m. 2009 , in proc .
`` exoplanetsand disks : their formation and diversity '' ( 2009 march9 - 12 ) , eds .
usuda , t. , ishii , m. and tamura , m. ( national astronomical observatory of japan ) , p. 11 - 16
thalmann , c. et al .
2010 , , 718 , l87 williams , j. p. & cieza , l. a. 2011 , ara&a , 49 , 67 whitney , b. a. et al .
2012 , in prep .
zhu , z. , nelson , r. p. , hartmann , l. , espaillat , c. , & calvet , n. 2011 , , 729 , 47 | we present high resolution @xmath0-band polarized intensity ( @xmath1 ; fwhm = @xmath2 : 14 au ) and @xmath3-band imaging data ( fwhm @xmath4 : 15 au ) of the circumstellar disk around the weak - lined t tauri star pds 70 in centaurus at a radial distance of 28 au ( @xmath5 ) up to 210 au ( 1.@xmath65 ) . in both images ,
a giant inner gap is clearly resolved for the first time , and the radius of the gap is @xmath770 au .
our data show that the geometric center of the disk shifts by @xmath76 au toward the minor axis .
we confirm that the brown dwarf companion candidate to the north of pds 70 is a background star based on its proper motion . as a result of sed fitting by monte carlo radiative transfer modeling
, we infer the existence of an optically thick inner disk at a few au . combining our observations and modeling ,
we classify the disk of pds 70 as a pre - transitional disk .
furthermore , based on the analysis of @xmath3-band imaging data , we put an upper limit mass of companions at @xmath730 to @xmath750@xmath8 within the gap . taking account of the presence of the large and sharp gap
, we suggest that the gap could be formed by dynamical interactions of sub - stellar companions or multiple unseen giant planets in the gap . |
nature may be viewed as a collection of _ complex systems _ @xcite .
consequently , a principal question is how these systems , which typically consist of many simple interacting units , develop qualitatively new and high - level kinds of organization .
this is the problem of connecting the microscopics of constituents with the coherent structures that characterize organisms and communities .
it may often be assumed that the fundamental laws of physics , such as hamilton and maxwell equations , are individual properties of the units .
still , it is only very rare that the origin and form of natural phenomena can be inferred from basic laws .
what is the relevance of fundamental physics to predict the weather , to design new materials and drugs or to understand the origin of life ?
it is remarkable that statistical physics recently addressed the problem of connecting emergent behavior to the constituents properties in a more indirect manner , too .
that is , main concepts in the theory of phase transitions , such as correlations , criticality , scale invariance and self - similarity that characterize the global behavior of the simplest model cases happen to be ubiquitous in nature .
this brings many interesting , high - level phenomena to the attention of physicists , and the study of ( nonequilibrium ) phase transitions has consequently been animated @xcite . as a matter of fact ,
an important observation in nature is that the complex systems of interest are often open , and out of a thermodynamic equilibrium state .
their simplest condition is that of a nonequilibrium steady state .
that is , a constant flux of some quantity ( matter , energy , ... ) is typically involved and the state is , in general , not determined solely by external constraints , but depends upon their _ history _ as well .
under such a nonequilibrium condition , as the control parameters temperature or potential gradients , or reactant feed rates , for instance are varied , the steady state may become unstable and be replaced by another ( or , perhaps , by a periodic or chaotic state ) .
nonequilibrium instabilities are attended by ordering phenomena so analogous to those of equilibrium statistical mechanics that one may speak of _ nonequilibrium phase transitions .
_ these are ubiquitous in physics and biology , and have also been described in the social sciences @xcite .
the simplest examples of nonequilibrium phase transitions occur in lattice models
. the analysis of more realistic situations is presently confronted , among other problems , with the lack of a general formalism , analogous to equilibrium statistical mechanics .
that is , nonequilibrium dynamics is not derivable from an energy function .
one must actually find time - independent solutions of master or kinetic equations , which is a formidable task in practice .
therefore , general theoretical approaches are scarce .
it is true that , for cases in which fluctuations are of minor significance , a macroscopic description , i.e. , a set of partial differential equations is often preferable to a lattice model , for instance , in predicting a nonequilibrium phase diagram . however , such macroscopic descriptions imply mean - field behavior , while lattice models exhibit a range of critical phenomena and other details which are at least as interesting as in equilibrium @xcite .
the lack of theory also explains that most interesting information has been gained by means of computer simulations of the lattice models .
as a first example of a complex lattice system that exhibits nonequilibrium phase transitions , let us consider an artificial neural network that was introduced and studied before @xcite .
this consists of a set of @xmath0 binary _ neurons , _
@xmath1 evolving in time by stochastic equations , @xmath2 \nonumber \\ + ( 1-p){\displaystyle\sum\limits_{\mathbf{x , y}}}{\displaystyle\sum\limits_{j\mathbf{^ { \prime}}_{xy}}}[-\varpi(j_{xy}\rightarrow j\mathbf{^{\prime}}_{xy})p_{t}(\mathbf{s } , \mathbf{j } ) + \varpi(j\mathbf{^{\prime}}_{xy}\rightarrow j_{xy})p_{t}(\mathbf{s},\mathbf{j}^{xy } ) ] \label{e25}\end{aligned}\ ] ] here @xmath3 is the configuration of synaptic intensities , and @xmath4 @xmath5 stands for @xmath6 @xmath7 after
the change @xmath8 @xmath9 @xmath10 .
the function @xmath11 is taken independent of the current @xmath12 and @xmath13 where @xmath14 is a local field . for @xmath15 ( [ e25 ] )
reduces to the familiar hopfield model in which the neurons evolve in the presence of a set of ( frozen ) synaptic intensities .
it is assumed that these in some way contain information from a set of @xmath16 stored patterns , @xmath17 e.g. , the hebb choice @xmath18 after appropriate normalization . under such conditions
, the model asymptotically tends to the equilibrium state for temperature @xmath19 and energy function , @xmath20 this state sometimes corresponds to a configuration closely resembling one of the stored patterns ; the system is therefore said to exhibit _ associative memory_. however , this simple case is not sufficiently efficient for applications ; e.g. , errors when recovering a given pattern are large for most values of @xmath21 @xmath16 and @xmath22 and the steady state may not be `` pure '' but correspond to a mixture of two or more stored patterns . for @xmath23 equation ( [ e25 ] )
transforms @xcite into @xmath24 , \ ] ] where the transition probability per unit time is the superposition @xmath25 .\ ] ] for appropriate choices of this superposition , i.e. , of functions @xmath26 and @xmath27 this system behaves qualitatively differents from the hopfield case .
that is , it can be shown analytically in some cases and , more generally , by computer simulations that a second - order ( equilibrium ) phase transition for @xmath28 transforms for @xmath29 into a first - order ( nonequilibrium ) phase transition .
this has some dramatic consequences concerning the recognition of a given pattern out of a deteriorated image of it .
in particular , for a wide and practically interesting range of @xmath21 @xmath16 and @xmath22 mixture states do not occur and the recovery process happens to be rather robust and accurate @xcite . this study induced us to investigate a cellular automaton version of the original model .
firstly , the function @xmath30 is properly determined ; this choice importantly affects in practice some of the system properties , e.g. , the nature of its phase transitions .
the simulation then proceeds by choosing at random any of the stored patterns , say @xmath31 and updating all the neurons in the lattice assuming the set @xmath32 i.e. , the synaptic intensities corresponding to the selected pattern .
next , this step is repeated again and again .
the draw is performed in such a way that the time average for each local @xmath33 gives hebb s rule , @xmath34 ( or , alternatively , any other learning rule one may use in the system definition ) .
the preliminary results that are available at the time of this writing reveal that this case exhibits a very robust and efficient process of pattern recognition . for most parameter values , starting from a perturbed pattern , the system rapidly transforms that configuration into the stored pattern that is closest to it .
figure 1 shows the time evolution of the overlap between the actual state of the system and one of the stored patterns at indicated temperature .
this is in units of the respective model critical temperature , either @xmath35 and @xmath36 for our system and for the hopfield case , respectively . _
type 1 _ here refers to the case in which the system stores @xmath37 patterns whose sites are generated completely at random , so that each site is independent of the others ; _ type 2 _ is for @xmath38 stored patterns generated using the logistic map in the chaotic region of its parameter space , so that some correlation exits between sites . in both cases ,
our algorithm rapidly detects the pattern which is closest to the initial state .
this behavior holds essentially for other values of the parameters .
the reason behind the _ good _ properties of our system seems to be that , for appropriate dynamics , the actual state only evolves noticeably at steps in which synapses correspond to the selected pattern @xcite .
this behavior opens the model to a wide range of possible applications .
the driven lattice gas ( dlg ) is a @xmath39dimensional ( @xmath40 in the following ) lattice gas at temperature @xmath19 in which transitions in ( against ) one of the principal lattice directions say @xmath41 to be referred to as the _ field direction _ are favored ( unfavored ) . for periodic boundary conditions ,
this induces a net current of particles along the field direction . at high
@xmath22 the system is in a disordered state while , for half - filled lattices ( the only case of interest in this paper ) , there is a second - order ( nonequilibrium ) critical point , below which the dlg segregates showing anisotropic , stripe - like configurations parallel to the field @xcite .
establishing the universality class of the dlg is a main issue not only concerning a better understanding of these model properties but also much more generally , in relation to the theory of nonequilibrium phase transitions and critical phenomena .
in fact , the dlg is recognized as one of the more intriguing model examples of nonequilibrium phenomena .
recent field theory @xcite motivated performing new and extensive computer simulations .
these focused on the case of an `` infinite '' drive ( particles along the field direction are not allowed to go backwards ) for both _ large _ squares and rectangular @xmath42 lattices of different , appropriate sizes .
it has thus been demonstrated numerically that the dlg belongs to the same universality class as a lattice gas under a randomly fluctuating field,@xcite so that ( using the renormalization group jargon ) the particle current which does not occur in the latter is not a _ relevant _ feature of the dlg .
main critical exponents follow for both cases as @xmath43 @xmath44 and @xmath45 these important results ( that have more recently been confirmed independently @xcite ) can be interpreted within the context of the existing field theory @xcite .
it should be noted however , that the present form of this theory does not fit well the results of further numerical investigation of the same model .
that is , studying small values of the field suggests that the dlg has no relation to the equilibrium lattice gas so that one can not go perturbatively from the latter to the case of small fields .
there is some indication that the dlg belongs to the same universality class with @xmath46 for any value of the field as long as the configurations are stripped @xcite . if this is confirmed , the chances are that the dlg will again attract considerable attention during the coming years . in fact , in addition to the above issue on criticality , there are further interesting questions concerning this model .
one is the nature of its kinetic behavior as a configuration evolves from a disordered state to the stripped one .
extending the arguments first checked for systems that evolve towards equilibrium,@xcite one should perhaps expect _ self - similarity _ with time of the structure function .
that is , as the system undergoes nucleation and then cluster coagulation according to a sort of spinodal decomposition , even though this is strongly anisotropic and will eventually lead to a nonequilibrium steady state , it seems reasonable to assume the existence of a unique relevant length .
one should expect this to be the length that characterizes the ( transverse ) clustering process in the system , say @xmath47 .
consequently , quantities changing with time should not depend explicitly on the time variable but only through @xmath48 for example , the structure function @xmath49 is expected to depict a scaled form @xmath50 independent of @xmath51 when plotted accordingly , this has recently been confirmed using the width of the stripes as the relevant length , as shown in fig .
2 . moreover , we found that @xmath53 for large values of @xmath54 , which is the generalization of porod s law to dlg .
this power law behavior , as compared to the @xmath55 tail observed in bidimensional equilibrium binary mixtures , reflects the fact that coarsening in dlg is effectively a unidimensional process which takes place in the direction perpendicular to the field .
it has also been shown that @xmath56 in general , as in standard spinodal decomposition,@xcite though the mechanisms leading to this behavior seem to be different .
more specifically , we found that @xmath57 for long enough times , where @xmath58 and @xmath59 are constants which depend on temperature , while for intermediate times we observe @xmath60 ( see fig .
the analysis of all the above results on the dynamics of dlg allows us to conclude that the particle current does not play any important role in the late stage coarsening of dlg , the anisotropy being the relevant ingredient present in this process .
* acknowledgments . *
the unpublished work partially reported in this paper was performed in collaboration with a. achahbar , e. albano , p.l .
garrido , m.a .
muoz and j.j .
torres ; a proper account of it is given elsewhere @xcite .
we also acknowledge financial support under the spanish projects dgeseic pb97 - 0842 and sepcyt bfm2001 - 2841 . | we comment on some recent , yet unpublished results concerning instabilities in complex systems and their applications .
in particular , we briefly describe main observations during extensive computer simulations of two lattice nonequilibrium models .
one exhibits robust and efficient processes of pattern recognition under synaptic coherent activity ; the second example exhibits interesting critical behavior and simulates nucleation and spinodal decomposition processes in driven fluids . _
key words _ : neural networks , driven lattice gas , spinodal decomposition , nonequilibrium phase transitions , celular automata . _
pacs _ : 05.50.+q , 05.70.fh , 05-70.ln , 64.60.qb , 75.40.mg |
in a recent paper kitano @xcite gives a formula for the reidemeister torsion of the 3-manifold obtained by a dehn surgery on the figure eight knot . in this paper
we generalize his result to all twist knots .
specifically , we will compute the reidemeister torsion of the complement of a twist knot in @xmath0 and that of the 3-manifold obtained by a dehn surgery on a twist knot .
let @xmath1 be the link in figure 1 , where @xmath2 denote the numbers of half twists in the boxes .
positive ( resp .
negative ) numbers correspond to right - handed ( resp .
left - handed ) twists .
note that @xmath1 is a knot if and only if @xmath3 is even .
the knot @xmath4 , where @xmath5 , is known as a twist knot . for more information on @xmath1 , see @xcite . in this paper
we fix @xmath6 .
let @xmath7 be the complement of @xmath8 in @xmath0 .
the fundamental group of @xmath7 has a presentation @xmath9 where @xmath10 are meridians and @xmath11 .
a representation @xmath12 is called nonabelian if the image of @xmath13 is a nonabelian subgroup of @xmath14 .
suppose @xmath15 is a nonabelian representation .
up to conjugation , we may assume that @xmath16 \quad \text{and } \quad \rho(b ) = \left [ \begin{array}{cc } s & 0 \\ -u & s^{-1 } \end{array } \right]\ ] ] where @xmath17 is a root of the riley polynomial @xmath18 , see @xcite .
let @xmath19 and @xmath20 .
let @xmath21 be the chebychev polynomials of the second kind defined by @xmath22 , @xmath23 and @xmath24 for all integers @xmath25 .
[ main1 ] suppose @xmath15 is a nonabelian representation .
if @xmath26 then the reidemeister torsion of @xmath7 is given by @xmath27 now let @xmath28 be the 3-manifold obtained by a @xmath29-surgery on the twist knot @xmath8 .
the fundamental group @xmath30 has a presentation @xmath31 where @xmath32 is the canonical longitude corresponding to the meridian @xmath33 .
[ main2 ] suppose @xmath15 is a nonabelian representation which extends to a representation @xmath34 .
if @xmath35 then the reidemeister torsion of @xmath28 is given by @xmath36 \(1 ) one can see that the expression @xmath37 is actually a polynomial in @xmath38 .
\(2 ) theorem [ main2 ] generalizes the formula for the reidemeister torsion of the 3-manifold obtained by a @xmath29-surgery on the figure eight knot by kitano @xcite .
\(1 ) if @xmath39 , then @xmath40 is the trefoil knot . in this case
the riley polynomial is @xmath41 , and hence @xmath42 \(2 ) if @xmath43 , then @xmath44 is the figure eight knot . in this case
the riley polynomial is @xmath45 , and hence @xmath46 the paper is organized as follows . in section [ section - chev ]
we review the chebyshev polynomials of the second kind and their properties . in section
[ nab ] we give a formula for the riley polynomial of a twist knot , and compute the trace of a canonical longitude . in section
[ section - r ] we review the reidemeister torsion of a knot complement and its computation using fox s free calculus .
we prove theorems [ main1 ] and [ main2 ] in section [ section - proof ] .
recall that @xmath21 are the chebychev polynomials defined by @xmath22 , @xmath23 and @xmath24 for all integers @xmath25 .
the following lemma is elementary .
[ chev ] one has @xmath47 let @xmath48 .
[ p_k ] one has @xmath49 we have @xmath50 the lemma follows .
[ p^2_k ] one has @xmath51 let @xmath52 we have @xmath53 since @xmath54 , we obtain @xmath55 for all integers @xmath25 .
hence @xmath56 .
[ formulas ] suppose @xmath57 \in sl_2(\bc)$ ] .
then @xmath58 , \label{power}\\ \sum_{i=0}^k v^i & = & \left [ \begin{array}{cc } p_{k}(t ) - d p_{k-1}(t ) & b p_{k-1}(t)\\ c p_{k-1}(t ) & p_{k}(t ) - a p_{k-1}(t ) \end{array } \right ] , \label{sum - power}\end{aligned}\ ] ] where @xmath59 .
moreover , one has @xmath60
since @xmath61 , by the cayley - hamilton theorem we have @xmath62 .
this implies that @xmath63 for all integers @xmath25 .
hence , by induction on @xmath25 , one can show that @xmath64 . since @xmath65 $ ] , follows . since @xmath66
, follows directly from . by lemma [ p^2_k ]
we have @xmath67 then follows from lemma [ p_k ] .
in this section we give a formula for the riley polynomial of a twist knot .
this formula was already obtained in @xcite .
we also compute the trace of a canonical longitude .
recall that @xmath68 and @xmath69 .
the fundamental group of @xmath7 has a presentation @xmath70 where @xmath10 are meridians and @xmath11 .
suppose @xmath15 is a nonabelian representation .
up to conjugation , we may assume that @xmath16 \quad \text{and } \quad \rho(b ) = \left [ \begin{array}{cc } s & 0 \\ -u & s^{-1 } \end{array } \right]\ ] ] where @xmath17 is a root of the riley polynomial @xmath18 .
we now compute @xmath18 . since @xmath71,\ ] ] by lemma [ formulas ] we have @xmath72,\ ] ] where @xmath73 .
hence , by a direct computation we have @xmath74\ ] ] where @xmath75 it is known that the canonical longitude corresponding to the meridian @xmath33 is @xmath76 , where @xmath77 is the word in the letters @xmath10 obtained by writing @xmath78 in the reversed order .
we now compute its trace .
this computation will be used in the proof of theorem [ main2 ] .
[ s^2 ] one has @xmath79 since @xmath17 is a root of the riley polynomial @xmath18 , we have @xmath80 .
lemma [ chev ] then implies that @xmath81 by replacing @xmath82 into the first factor of the above expression , we obtain the desired equality . [ longitude ] one
has @xmath83 by lemma [ formulas ] we have @xmath84.\ ] ] similarly , @xmath85.\ ] ] hence , by a direct calculation we have @xmath86 the lemma then follows from lemma [ s^2 ] .
in this section we briefly review the reidemeister torsion of a knot complement and its computation using fox s free calculus . for more details on the reidemeister torsion ,
see @xcite .
let @xmath87 be a chain complex of finite dimensional vector spaces over @xmath88 : @xmath89 such that for each @xmath90 the followings hold * the homology group @xmath91 is trivial , and * a preferred basis @xmath92 of @xmath93 is given .
let @xmath94 be the image of @xmath95 . for each @xmath96 choose a basis @xmath97 of @xmath98 .
the short exact sequence of @xmath88-vector spaces @xmath99 implies that a new basis of @xmath93 can be obtained by taking the union of the vectors of @xmath97 and some lifts @xmath100 of the vectors @xmath101 .
define @xmath102 $ ] to be the determinant of the matrix expressing @xmath103 in the basis @xmath92 .
note that this scalar does not depend on the choice of the lift @xmath100 of @xmath101 .
the _ torsion _ of @xmath87 is defined to be @xmath104^{(-1)^{i+1 } } \ \in \bc\setminus\{0\}.\ ] ] once a preferred basis of @xmath87 is given , @xmath105 is independent of the choice of @xmath106 .
let @xmath28 be a finite cw - complex and @xmath34 a representation .
denote by @xmath107 the universal covering of @xmath28 .
the fundamental group @xmath30 acts on @xmath107 as deck transformations .
then the chain complex @xmath108 has the structure of a chain complex of left @xmath109$]-modules .
let @xmath110 be the 2-dimensional vector space @xmath111 with the canonical basis @xmath112 . using the representation @xmath13
, @xmath110 has the structure of a right @xmath109$]-module which we denote by @xmath113 .
define the chain complex @xmath114 to be @xmath115 } v_{\rho}$ ] , and choose a preferred basis of @xmath114 as follows .
let @xmath116 be the set of @xmath96-cells of @xmath28 , and choose a lift @xmath117 of each cell .
then @xmath118 is chosen to be the preferred basis of @xmath119 .
a representation @xmath13 is called _ acyclic _ if all the homology groups @xmath120 are trivial .
the reidemeister torsion @xmath121 is defined as follows : @xmath122 let @xmath123 be a knot in @xmath0 and @xmath124 its complement .
we choose a wirtinger presentation for the fundamental group of @xmath124 : @xmath125 let @xmath126 be a representation .
this map induces a ring homomorphism @xmath127\to m_2(\bc)$ ] , where @xmath128 $ ] is the group ring of @xmath129 and @xmath130 is the matrix algebra of degree @xmath131 over @xmath132 .
consider the @xmath133 matrix @xmath134 whose @xmath135-component is the @xmath136 matrix @xmath137 where @xmath138 denotes the fox calculus .
for @xmath139 , denote by @xmath140 the @xmath141 matrix obtained from @xmath134 by removing the @xmath142th column .
we regard @xmath140 as a @xmath143 matrix with coefficients in @xmath88 .
then johnson showed the following .
@xcite [ johnson ] let @xmath126 be a representation such that @xmath144 .
then the reidemeister torsion of @xmath124 is given by @xmath145
suppose @xmath15 is a nonabelian representation which extends to a representation @xmath34 . recall that @xmath32 is the canonical longitude corresponding to the meridian @xmath33 .
if @xmath173 , then by @xcite ( see also @xcite ) the reidemeister torsion of @xmath28 is given by @xmath174 | we compute the reidemeister torsion of the complement of a twist knot in @xmath0 and that of the 3-manifold obtained by a dehn surgery on a twist knot . |
the ising model in @xmath0 dimensions is perhaps the most studied statistical mechanical model and has a special place in the theory of critical phenomena since the groundbreaking work of onsager @xcite .
its scaling limit at or near the critical point is recognized to give rise to euclidean ( quantum ) field theories . in particular , at the critical point , the lattice magnetization field should converge , in the scaling limit , to a euclidean random field @xmath1 corresponding to the simplest reflection - positive conformal field theory @xcite . as such , there have been a variety of representations in terms of free fermion fields @xcite and explicit formulas for correlation functions ( see , e.g. , @xcite and references therein ) . in @xcite , c.m .
newman and the present author introduced a representation of @xmath1 in terms of random geometric objects associated with schramm - loewner evolutions ( sles ) @xcite ( see also @xcite ) and conformal loop ensembles ( cles ) @xcite namely , a gas ( or random process ) of continuum loops and associated clusters and ( renormalized ) area measures .
the purpose of the present paper is twofold , as we now explain .
first of all , we provide a detailed proof of the existence of subsequential limits of the lattice magnetization field as a square integrable random variable and a random generalized function ( theorem [ thm : subseqential - limits ] ) following the ideas presented in @xcite .
we also introduce a cutoff field whose scaling limit admits a geometric representation in terms of rescaled counting measures associated to critical fk clusters , and show that it converges to the magnetization field as the cutoff is sent to zero ( theorem [ thm : cutoff - removal ] ) .
secondly , we describe work in progress @xcite of the author with c. garban and c.m .
newman aimed at establishing uniqueness of the scaling limit of the lattice magnetization and conformal covariance properties for the limiting magnetization field .
we also explain how the existence and conformal covariance properties of the magnetization field should imply the convergence , in the scaling limit , of a version of the model with a vanishing ( in the limit ) external magnetic field to a field theory with exponential decay of correlations , and how they can be used to determine the free energy density of the model up to a constant ( equation ( [ eq : free - energy ] ) ) .
we consider the standard ising model on the square lattice @xmath2 with ( formal ) _ hamiltonian _ @xmath3 where the first sum is over nearest - neighbor pairs in @xmath2 , the spin variables @xmath4 are @xmath5-valued and the external field @xmath6 is in @xmath7 . for a bounded @xmath8 , the _ gibbs distribution _ is given by @xmath9 , where @xmath10 is the hamiltonian ( [ eq : hamiltonian ] ) with sums restricted to sites in @xmath11 , @xmath12 is the _ inverse temperature _ , and the _ partition function _
@xmath13 is the appropriate normalization needed to obtain a probability distribution .
we are mostly interested in the model with zero ( or vanishing ) external field , and at the critical inverse temperature , @xmath14 . for all @xmath15 ,
the model has a unique _ infinite - volume gibbs distribution _ for any value of the external field @xmath6 , obtained as a weak limit of the gibbs distribution for bounded @xmath11 by letting @xmath16 . for any value of @xmath15 and of @xmath6 , expectation with respect to the unique infinite - volume gibbs distribution
will be denoted by @xmath17 . at the _ critical point _
, that is when @xmath18 and @xmath19 , expectation will be denoted by @xmath20 . by translation invariance ,
the _ two - point correlation _ @xmath21 is a function only of @xmath22 , which at the critical point we denote by @xmath23 .
we want to study the random field associated with the spins on the rescaled lattice @xmath24 in the scaling limit @xmath25 .
more precisely , for functions @xmath26 of bounded support on @xmath27 , we define for the critical model @xmath28 dz = \theta_a \sum_{z \in a \ , { \mathbb z}^2 } f(z ) s_{z / a } \ , , \ ] ] with scale factor @xmath29 where @xmath30 ^ 2 \cap a \ , { \mathbb z}^2 $ ] and @xmath31 ^ 2 \cap { \mathbb z}^2 $ ] .
the block magnetization , @xmath32 ^ 2})$ ] , where @xmath33 denotes the indicator function , is a rescaled sum of identically distributed , _ dependent _ random variables . in the high temperature case , @xmath34 , and with zero external field ,
@xmath19 , the dependence is sufficiently weak for the block magnetization to converge , as @xmath35 , to a mean - zero , gaussian random variable ( see , e.g. , @xcite and references therein ) . in that case , the appropriate scaling factor @xmath36 is of order @xmath37 , and the field converges to gaussian white noise as @xmath35 ( see , e.g. , @xcite ) . in the critical case ,
however , correlations are much stronger and extend to all length scales , so that one does not expect a gaussian limit .
a proof of this will be presented elsewhere @xcite ; in this paper we are concerned with the existence of subsequential limits for the lattice magnetization field , and their geometric representation in terms of area measures of critical fk clusters .
the fk representation of the ising model with zero external field , @xmath19 , is based on the @xmath38 random - cluster measure @xmath39 ( see @xcite for more on the random - cluster model and its connection to the ising model ) .
a spin configuration distributed according to the unique infinite - volume gibbs distribution with @xmath19 and inverse temperature @xmath15 can be obtained in the following way .
take a random - cluster ( fk ) bond configuration on the square lattice distributed according to @xmath39 with @xmath40 , and let @xmath41 denote the corresponding collection of fk clusters , where a cluster is a maximal set of sites of the square lattice connected via bonds of the fk bond configuration ( see figure [ fig : loops ] ) .
one may regard the index @xmath42 as taking values in the natural numbers , but it s better to think of it as a dummy countable index without any prescribed ordering , like one has for a poisson point process .
let @xmath43 be ( @xmath44)-valued , i.i.d .
, symmetric random variables , and assign @xmath45 for all @xmath46 ; then the collection @xmath47 of spin variables is distributed according to the unique infinite volume gibbs distribution with @xmath19 and inverse temperature @xmath48 .
when @xmath18 , we will use the notation @xmath49 , and @xmath50 for expectation with respect to @xmath51 . , black horizontal and vertical edges represent fk bonds .
the fk clusters are highlighted by lighter ( green ) loops on the medial lattice.,width=340 ] a useful property of the fk representation is that , when @xmath19 , the ising two - point function can be written as @xmath52 as an immediate consequence , we have @xmath53 = e_c \left [ \sum_i | \hat{\cal c}^a_i restriction of the rescaled cluster @xmath54 in @xmath24 to @xmath55 ^ 2 $ ] , and @xmath56 is the number of ( @xmath57)-sites in @xmath58 .
( note that @xmath58 need not be connected . ) using the fk representation , we can write ( [ eq : lat - field1 ] ) as @xmath59 where @xmath60 and the @xmath61 s , as before , are @xmath5-valued , symmetric random variables independent of each other and everything else .
we can now easily see that @xmath36 was chosen so that the second moment of the block magnetization @xmath62 , defined earlier , is exactly one : @xmath63 ^ 2 } ) \right]^2 \right\rangle_c = e_c\left[\sum_i \left(\mu^a_i({\bf 1}_{[0,1]^2})\right)^2\right ] = \theta^2_a e_c \left [ \sum_i |\hat{\cal c}^a_i|^2 \right ] = 1 \ , .\ ] ] we can associate in a unique way to each rescaled counting measure @xmath64 the interface @xmath65 in the medial lattice between the corresponding ( rescaled ) fk cluster @xmath66 and the surrounding fk clusters .
since all fk clusters are almost surely finite at the critical point ( @xmath67 ) , such interfaces form closed curves , or loops , which separate the corresponding clusters @xmath66 from infinity ( see fig .
[ fig : loops ] ) .
there are two types of loops : ( 1 ) those with sites of @xmath57 immediately on their inside and ( 2 ) those with sites of @xmath68 immediately on their outside .
we denote by @xmath69 the ( random ) collection of all loops of the first type associated with the fk clusters @xmath70 .
each realization of @xmath69 can be seen as an element in a space of collections of loops with the aizenman - burchard metric @xcite .
( the latter is the induced hausdorff metric on collections of curves associated to the metric on curves given by the infimum over monotone reparametrizations of the supremum norm . )
it follows from @xcite and the rsw - type bounds of @xcite ( see section 5.3 there ) that , as @xmath35 , @xmath69 has subsequential limits in distribution to random collections of loops in the aizenman - burchard metric . in the scaling limit
, one gets collections of nested loops that can touch ( themselves and each other ) , but never cross . in order to study the magnetization field ,
we introduce some more notation .
let @xmath71 denote the space of continuous functions on @xmath27 with compact support , endowed with the metric of uniform convergence .
let @xmath72 denote the space of probability distributions on @xmath7 ( with the borel @xmath73-algebra ) with finite second moment , endowed with the wasserstein ( or minimal @xmath74 ) metric @xmath75 \right)^{1/2 } \
, , \ ] ] where @xmath76 and @xmath77 are coupled random variables with respective distributions @xmath78 and @xmath79 , @xmath80 denotes expectation with respect to the coupling , and the infimum is taken over all such couplings ( see , e.g. , @xcite and references therein ) .
convergence in the wasserstein metric @xmath81 is equivalent to convergence in distribution plus convergence of the second moment . for brevity , we will write @xmath82 and @xmath83 , instead of @xmath71 and @xmath72 , unless we wish to emphasize the role of the metrics .
we further denote by @xmath84 the space of infinitely differentiable functions on @xmath27 with compact support , equipped with the topology of uniform convergence of all derivatives , and by @xmath85 its topological dual , i.e. , the space of all generalized functions .
the next theorem shows that the lattice magnetization field has subsequential scaling limits in terms of continuous functionals , in a distributional sense using the wasserstein metric @xmath81 , and in the sense of generalized functions by an application of the bochner - minlos theorem .
( we remark that the last statement of theorem [ thm : subseqential - limits ] is not optimal in the sense that similar conclusions should apply to a larger class of functions than @xmath86 . )
[ thm : subseqential - limits ] for any sequence @xmath87 , there exists a subsequence @xmath88 such that , for all @xmath89 , the distribution @xmath90 of @xmath91 converges in the wasserstein metric ( [ eq : w - distance ] ) , as @xmath92 , to a limit @xmath93 such that the map @xmath94 is continuous .
furthermore , for every subsequential limit @xmath95 , there exists a random generalized function @xmath96 with characteristic function @xmath97 .
theorem [ thm : subseqential - limits ] represents the starting point of a joint project with c. garban and c.m .
newman aimed at establishing uniqueness of the scaling limit of the lattice magnetization field and its conformal covariant properties .
one not only expects a unique scaling limit for the lattice magnetization field , but based on the representation ( [ eq : lattice - field ] ) , one would like to write the limiting field @xmath1 as @xmath98 where the @xmath99 s are the putative scaling limits of the @xmath100 s that appear in ( [ eq : lattice - field ] ) . indeed , in the scaling limit , one should obtain a collection @xmath101 of mutually orthogonal , finite measures supported on the scaling limit of the critical fk clusters .
however , due to scale invariance , @xmath102 should contain ( countably ) infinitely many elements , and the scaling covariance expected for the @xmath103 s suggests that the collection @xmath104 is in general not absolutely summable .
what meaning , if any , can we then attribute to the sum in ( [ eq : geom - rep ] ) ? to help answer that question , we introduce the @xmath105-cutoff lattice magnetization field @xmath106 where the elements of the collection @xmath107 of all rescaled ( random ) measures that are involved in ( [ eq : cutoff - lattice - field ] ) are those associated to rescaled fk clusters @xmath66 that intersect the support of @xmath26 and whose corresponding loops @xmath65 have diameter @xmath108 . once again
, one would like to write the scaling limit of the cutoff field as @xmath109 " . in this case however , the sum would be unambiguous because it would contain only a _ finite _ number of terms . a proof of the latter fact follows from prop .
[ prop : crossprob ] in section [ sec : beyond ] . combined with ( [ eq : second - moment ] ) and prop .
[ prop : small - clusters ] in section [ sec : proofs ] , prop .
[ prop : crossprob ] implies that the collection of @xmath100 s corresponding to macroscopic fk clusters has nontrivial subsequential scaling limits .
indeed , it is clear from equation ( [ eq : second - moment ] ) that no @xmath110 ^ 2})$ ] can diverge as @xmath111 .
in addition , prop . [
prop : small - clusters ] says that `` small '' fk clusters do not contribute to the magnetization in the scaling limit and thus , by prop .
[ prop : crossprob ] , the number of fk clusters which contribute significantly to @xmath112 ^ 2})$ ] remains _ bounded _ as @xmath35 .
since @xmath113 for all @xmath37 , this implies that not all @xmath114 ^ 2})$ ] s can converge to 0 as @xmath35 .
[ prop : tau ] ensures that the same conclusions hold not only for the collection of @xmath100 s with @xmath115 ^ 2}$ ] , but for other functions as well . the result below shows that , in the scaling limit , one recovers the `` full '' magnetization field from the cutoff one by letting the cutoff go to zero .
[ thm : cutoff - removal ] for any sequence @xmath87 , there exists a subsequence @xmath116 such that , for all @xmath89 and all @xmath117 , the distributions of @xmath91 and @xmath118 converge in the wasserstein metric ( [ eq : w - distance ] ) as @xmath92 .
moreover , if @xmath119 and @xmath120 denote the respective limits , @xmath120 converges to @xmath119 in the wasserstein metric ( [ eq : w - distance ] ) as @xmath121 . in view of theorem [ thm : cutoff - removal ]
, one can interpret the sum in equation ( [ eq : geom - rep ] ) as a shorthand for the limit of the cutoff field as the cutoff is removed .
combined with the fact that the collection of @xmath100 s has nontrivial subsequential scaling limits , as explained above , theorems [ thm : subseqential - limits ] and [ thm : cutoff - removal ] partly establish the geometric representation proposed in @xcite . in order to establish the existence of a unique scaling limit for the collection of @xmath64 s as measures , and to obtain their conformal covariance properties and those of the limiting magnetization field @xmath1 ,
more work is needed .
this is discussed in the next section .
the lattice magnetization field is expected to have a unique scaling limit @xmath1 with the property of transforming covariantly under conformal transformations , i.e. , if @xmath122 is a conformal map , @xmath123 where @xmath124 is the ising magnetization exponent .
( with an abuse of notation , we identify @xmath27 and the complex plane @xmath125 . )
it is natural to attempt to prove such results using announced results for fk percolation ( see @xcite ) which identify the scaling limit of the fk cluster boundaries ( see fig .
[ fig : loops ] ) with sle - type random fractal curves whose distribution is invariant under conformal transformations . in order to exploit such results , one can use techniques developed in @xcite to study the scaling limit of bernoulli and dynamical percolation in two dimensions . roughly speaking ,
the idea is to prove that the scaling limit of the ensemble @xmath107 of rescaled counting measures associated to the fk clusters is a measurable function of the collection of limiting ( macroscopic ) loops between fk clusters .
to illustrate the idea , we take a small detour and discuss briefly the scaling limit of bernoulli percolation , focusing on site percolation on the triangular lattice .
the full " scaling limit of percolation , comprising all interface loops separating macroscopic clusters , was obtained by camia and newman in @xcite and shown to be a ( nested ) conformal loop ensemble ( cle ) in @xcite . in @xcite camia
, fontes and newman proposed to construct the near / off - critical scaling limit of percolation , with density of open sites @xmath126 ( where @xmath127 is a parameter , @xmath37 the lattice spacing , and @xmath128 the percolation correlation length exponent ) , from the critical one `` augmented '' by a `` poissonian cloud '' of marks on the double points of the limiting loops ( i.e. , where a loop touches itself or where two different loops touch each other ) . back on the lattice
, the marked points would correspond to `` pivotal '' sites that switch state when the density of open sites is changed from @xmath129 to @xmath130 , causing a macroscopic change in connectivity .
( the last sentence should be interpreted in the context of the canonical coupling of percolation models at different densities of open sites . in this coupling , a percolation model with density @xmath130 of open sites
is obtained by assigning independent , uniform random variables @xmath131 $ ] to the sites @xmath132 of the lattice , and declaring open all sites with @xmath133 , and closed all other sites . )
a key step in the implementation of this idea is the construction of the intensity measure of the poisson process of marks .
since the points to be marked are double points , it was argued in @xcite that the intensity measure should arise as the scaling limit of the appropriately rescaled counting measure of @xmath105-macroscopically pivotal sites on the lattice with spacing @xmath37 , where an @xmath105-macroscopically pivotal site @xmath132 has four neighbors which are the starting points of four alternating paths , two made of ( nearest - neighbor ) open sites and two of closed ones , reaching a distance @xmath105 away from @xmath132 .
the occurrence of an @xmath105-macroscopically pivotal site @xmath132 in a percolation configuration is called a _
four - arm event_. the scaling limit of the counting measure of @xmath105-macroscopically pivotal sites was obtained by garban , pete and schramm @xcite ( see also @xcite ) and used by the same authors , in the spirit of the program proposed by camia , fontes and newman , to construct the near / off - critical scaling limit of percolation . in particular , garban , pete and schramm
@xcite consider the joint distribution of the collection of interface loops and the ( random ) counting measure of @xmath105-macroscopically pivotal sites , @xmath134 , and show that it converges to the law of some random variable @xmath135 , where @xmath136 is the collection of limiting loops and @xmath137 is a random borel measure .
moreover , they show that @xmath137 is a measurable function of @xmath138 .
this last observation is in fact crucial , since the known uniqueness of the scaling limit of the interface loops implies the uniqueness of @xmath137 .
in addition , one can deduce how @xmath137 changes under conformal transformations from the knowledge of how @xmath136 changes under those same transformations .
the latter can be deduced for the collection @xmath136 from the fact that it is a nested cle whose loops are sle - type curves .
heuristically , one can convince oneself that it is reasonable to expect that @xmath137 be a measurable function of @xmath136 by noticing that knowing the macroscopic loops should be sufficient to give a good estimate of the number of macroscopically pivotal sites . for a discussion on how to turn this observation into a proof ,
the reader is referred to sect .
4.3 of @xcite , where complete proofs of the results mentioned in the previous paragraph can also be found . in sect .
5 of @xcite , the authors discuss how to obtain similar results for rescaled counting measures of other special sites .
in particular , they show how to obtain what they call the `` cluster '' or `` area '' measure , which counts the number of open sites contained in clusters of diameter larger than some cutoff @xmath139 .
the occurrence of such a site @xmath132 corresponds to the event that there is a path of ( nearest - neighbor ) open sites starting at @xmath132 and reaching a distance @xmath105 away from @xmath132 .
such an event is called a _
one - arm event _ , and we will call @xmath132 a _ one - arm site_. the proof in this case is in fact simpler because the event is simpler , involving only one path . at this point
the reader should note that the area measures @xmath64 introduced in the previous section in connection with the magnetization field also count one - arm sites , with the only difference that the relevant one - arm events are now in the context of fk bond percolation .
fk percolation is more difficult to analyze than bernoulli percolation , due to the dependencies in the distribution of fk configurations ( as opposed to the product measure corresponding to bernoulli percolation ) .
however , it seems that one can successfully adapt the techniques of @xcite , at least for the case of one - arm sites which is relevant for the magnetization . as a consequence ,
thanks to the results announced in @xcite , one should obtain uniqueness of the limiting ensemble @xmath101 of area measures for the fk clusters and of the magnetization field @xmath1 , as well as a proof of ( [ eq : conf - cov ] ) and of the fact that , for any conformal map @xmath122 , @xmath140 is equidistributed with @xmath141 .
because of the latter property , we call the putative collection of measures @xmath142 , obtained as the scaling limit of the collection of rescaled counting measures @xmath107 , a _ conformal measure ensemble_.
the uniqueness and conformal covariance properties of @xmath1 play an important role in the analysis of the _ near - critical _ scaling limit ( called _ off - critical _ in the physics literature ) with a vanishing ( in the limit ) external field ( at the critical inverse temperature @xmath143 ) .
more precisely , consider an ising model on @xmath24 with ( formal ) hamiltonian ( [ eq : hamiltonian ] ) and external field @xmath144 inside the square @xmath145 ^ 2 $ ] , and zero outside it .
we call @xmath146 the _ renormalized external field _ and note that the term @xmath147 ^ 2 } s_{z / a}\ ] ] in the hamiltonian implies that the gibbs distribution of this particular ising model is given by @xmath148 ^ 2 } s_{z / a } \right ) d\nu^a = \frac{1}{z_{h , l}^a } \exp\left(h m^a_l \right ) d\nu^a \ , , \ ] ] where @xmath149 is the gibbs distribution corresponding to zero external field , @xmath150 is the appropriate normalization factor , and @xmath151 denotes the block magnetization inside @xmath145 ^ 2 $ ] . as a consequence , in the scaling limit ( @xmath35 ) one would obtain a distribution @xmath152 such that @xmath153 ^ 2})\right ) d\nu^0 \ , , \ ] ] where @xmath154 ^ 2 } ) d\nu^0 $ ] and @xmath155 is the limiting distribution corresponding to zero external field .
the question is now whether @xmath152 converges to some @xmath156 as @xmath157 , and whether @xmath156 corresponds to the physically correct near / off - critical scaling limit .
heuristically , the correct normalization to obtain a nontrivial near / off - critical scaling limit is such that the correlation length @xmath158 remains bounded away from zero and infinity . scaling theory implies that @xmath159 for small external field @xmath6 .
this gives @xmath160 , which coincides with the normalization needed to obtain a nontrivial magnetization field ( given by @xmath36 ) , as can be seen from ( [ eq : theta - first ] ) and the asymptotic behavior of @xmath161 . with this in mind
, we consider an ising model on @xmath24 with an external field @xmath162 inside @xmath163 and @xmath164 outside , for some large @xmath165 .
using the two - dimensional ising critical exponent @xmath166 for the magnetization ( i.e. , @xmath167 for small @xmath6 , where @xmath168 denotes the spin at the origin ) , and denoting by @xmath169 the sum over @xmath132 in @xmath170 , we can write the block magnetization in the unit square as @xmath171 since the result is finite , this rough computation suggests a positive answer to the previous question . indeed , using the convergence of the lattice magnetization field to the continuum one and scaling properties of the critical fk clusters
, it appears possible to show @xcite that , as @xmath172 , @xmath152 has a unique weak limit , denoted by @xmath156 , and that @xmath156 represents the scaling limit of the ising model on @xmath24 with external field @xmath144 on the whole plane .
the idea behind a proof of this makes use of the well - known `` ghost spin '' representation of the ising model with an external field , in which an additional site with spin that agrees with the external field is added and connected to all the sites of the square lattice .
the external field term in the hamiltonian can then be written ( formally ) as @xmath173 , where the ghost spin @xmath174 is equal to the sign of the external field @xmath6 .
one can describe the ising model with an external field using the fk representation on the new graph comprising the square lattice and the additional site carrying the ghost spin .
note however that the density of fk bonds incident on the site carrying the ghost spin is not given by @xmath175 , as for the other bonds , but by @xmath176 .
the following key observation is an easy consequence of standard properties of fk percolation . if a subset @xmath11 of the square lattice is surrounded by a circuit @xmath177 of fk bonds that belong to a cluster which also contains the site carrying the ghost spin , the fk and spin configurations in @xmath11 are independent of the fk and spin configurations outside the circuit @xmath177 .
the rsw - type bounds proved in @xcite , together with the fkg inequality @xcite and scaling properties of the fk clusters and their area measures , imply that the probability to find such a circuit @xmath177 surrounding any bounded subset @xmath11 is one .
this shows that the @xmath152-probability of any event that depends only on the restriction of the spin configuration to a finite subset @xmath11 of the square lattice has a limit as @xmath172 .
consequently , the distribution @xmath152 has a weak limit @xmath156 as @xmath157 .
it is interesting to note that the argument alluded to above also shows that @xmath156 is locally absolutely continuous with respect to the zero - field measure @xmath155 .
this is in contrast to the situation in two - dimensional percolation , where the critical and near - critical measures are mutually singular @xcite .
it should be noted , however , that the ising analogue of that type of percolation near - critical scaling limit is to set @xmath19 and let @xmath178 , rather than set @xmath179 and let @xmath180 .
one expects the near / off - critical field to be `` massive '' in the sense that correlations under @xmath156 should decay exponentially at large distances . to understand why this should be the case ,
it is again useful to resort to the ghost spin representation discussed earlier .
remember that the ising two - point function can be expressed in terms of connectivity properties of the fk clusters ( see the discussion about the fk representation preceding equation ( [ eq : theta ] ) ) . because of that , exponential decay of correlations is equivalent to the statement that , if two sites of the square lattice , @xmath132 and @xmath181 , belong to the same fk cluster @xmath182 , the probability that @xmath182 does not contain the site carrying the ghost spin decays exponentially in the distance between @xmath132 and @xmath181 .
but the scaling law for the area measures , @xmath183 for all @xmath184 , suggests that a macroscopic fk cluster of diameter at least @xmath185 ( that is , of order @xmath186 in units of the lattice spacing @xmath37 ) should contain at least @xmath187 sites , precisely enough to compensate for the small intensity of the external field @xmath160 , which determines the probability of a cluster to contain the site carrying the ghost spin via the density , @xmath176 , of fk bonds connected to that site .
the exponential decay of correlations can be used to show the existence of the _ free energy density _
@xmath188 at the critical ( inverse ) temperature , defined by @xmath189 ^ 2 } ) ) d\nu^0\right ) } \ , , \ ] ] provided that the limit exists .
( because of symmetry , it suffices to consider positive external fields , @xmath190 . ) for the nearest - neighbor lattice ising model , following a standard argument ( see for instance @xcite , lecture 8) , one can show the existence of the free energy by partitioning @xmath145 ^ 2 $ ] into equal squares of fixed size and writing the hamiltonian as a sum of terms of two types : those corresponding to the interactions between spins inside a square , and the boundary terms that account for the interactions between different squares .
the contribution of the latter terms to the free energy vanishes in the limit @xmath157 because the boundary terms grow only linearly in @xmath165 , implying the existence of the limit defining the free energy . in our situation ,
the above argument is not immediately applicable because we have already taken the scaling limit and are now dealing with a continuum model .
we can however try to mimic that argument . for that purpose ,
we introduce the functions @xmath191 ^ 2 } ) ) d\nu^t\right ) } \ , , \ ] ] where @xmath192 denotes the near / off - critical magnetization field with renormalized external field @xmath193 .
we now write @xmath194 ^ 2 } ) = \sum_k \phi^0_t(\text{square}_k)$ ] , where @xmath195 denotes the @xmath196th element in a set of equal squares of fixed size that partition @xmath197 ^ 2 $ ] .
although the random variables @xmath198 are clearly not independent , the exponential decay of correlations under @xmath199 implies that they are only weakly correlated when the squares are far apart , suggesting a finite limit for @xmath200 as @xmath201 .
one can indeed show that the exponential decay of the covariance between different squares implies that @xmath202 .
the fkg inequality easily implies that @xmath203 for @xmath204 , and that @xmath205 and @xmath200 are increasing in @xmath206 .
therefore , one can conclude the existence of a finite limit for @xmath205 as @xmath207 . comparing the definitions of @xmath205 and @xmath188 ,
this strongly suggests ( and can be used to prove ) the existence of the limit defining @xmath188 .
integrating ( [ eq : conf - cov ] ) , one can check that @xmath208 ^ 2 } ) \stackrel{dist.}{= } \alpha^{15/8 } \ , \phi^0({\bf 1}_{[-l , l]^2 } ) \ , , \ ] ] consistent with the scaling law for area measures .
if the limit defining the free energy density exists ( and is unique ) , the above observation implies that @xmath209 , which means that the free energy density must take the form @xmath210 for some constant @xmath211 .
an immediate consequence of ( [ eq : free - energy ] ) would be the determination of the tail behavior of the block magnetization : @xmath212 ^ 2})>x ) \sim \exp{(-c_2 \ , x^{16 } ) } \text { \,\,\,\,\,\,\
, for } x>0 \text { and some constant } c_2>0.\ ] ] this result would follow from the methods of @xcite ( see , in particular , theorem 1.4 and corollary 2.6 there for one - sided bounds of the same type under similar conditions ) and it would show , incidentally , that the scaling limit magnetization field is not gaussian .
in this section , we briefly discuss the applicability of the approach presented in @xcite and in this paper to higher dimensions , @xmath213 , and to @xmath214-state potts models with @xmath215 .
although the @xmath0 scaling limit ising magnetization field @xmath1 should transform covariantly under conformal transformations and have close connections to the schramm - loewner evolution ( sle ) , no conformal machinery seems necessary to establish the existence of subsequential scaling limits in terms of area measures of critical fk clusters . a main ingredient used in this paper is prop . [ prop : small - clusters ] , which essentially says that `` small '' fk clusters do not contribute to the magnetization in the scaling limit .
this follows from the behavior of the two - point function at long distance ( prop .
[ prop : tau ] ) . inspecting the proof , it is easy to check that , in order for prop .
[ prop : small - clusters ] to hold in dimension @xmath216 , @xmath23 should behave at long distance like @xmath217 with @xmath218 ( see @xcite ) .
such a decay for @xmath161 should be valid for all @xmath216 .
( in particular , @xmath219 should be 0 above four dimensions , a result which has been proved when the number of dimensions is sufficiently high @xcite . )
however , there is a significant difference between dimensions below and above @xmath220 , where 4 is the upper - critical dimension for the ising model . as we mentioned earlier , for @xmath0 the number of terms in the sum that defines the cutoff field ( [ eq : cutoff - lattice - field ] ) remains a.s .
finite in the scaling limit .
this is due to the following result , whose proof is postponed to the next section .
[ prop : crossprob ] for @xmath221 , let @xmath222 denote the number of distinct clusters @xmath223 that include sites in both @xmath224 and @xmath225 . for any @xmath226 , there exists @xmath227 such that for all @xmath221 and all small @xmath228 and any @xmath229 , @xmath230 it follows that for any bounded @xmath231 and @xmath139 , the number of distinct clusters @xmath223 of diameter @xmath232 touching @xmath233 is bounded in probability as @xmath35 . the analogue of prop .
[ prop : crossprob ] is expected to fail above the upper - critical dimension @xmath220 ( see appendix a of @xcite ) .
when it fails , there can be infinitely many fk clusters with diameter greater than @xmath105 in a bounded region and so prop .
[ prop : small - clusters ] would not preclude @xmath1 from being a gaussian ( free ) field .
but it appears that at least for @xmath234 , both the analogue of prop .
[ prop : crossprob ] and a representation of @xmath235 as a sum of finite measures with random signs ought to be valid .
an analogous representation for the scaling limit magnetization fields of @xmath214-state potts models also ought to be valid , at least for values of @xmath214 such that for a given @xmath236 , the phase transition at @xmath143 is second order .
( this was pointed out to the authors of @xcite by j. cardy . )
the phase transition is believed to be first order for integer @xmath237 when @xmath238 and for @xmath239 when @xmath240 ( see @xcite ) ; this leaves , besides the ising case , @xmath240 and @xmath241 and @xmath242 .
we denote the states or colors of the @xmath214-state potts model by @xmath243 , and recall that in the fk representation on the lattice , all sites in an fk cluster have the same color while the different clusters are colored independently with each color equally likely . in the scaling limit , there would be finite measures @xmath244 , and the magnetization field in the color-@xmath196 direction would be @xmath245 with the @xmath246 s taking the value @xmath247 with probability @xmath248 ( for the color @xmath196 ) and the value @xmath249 with probability @xmath250 ( for any other color ) . for a fixed @xmath196
the @xmath246 s would be independent as @xmath251 varies , but for a fixed @xmath251 they would be _ dependent _ as @xmath196 varies because @xmath252 .
the proofs of prop .
[ prop : crossprob ] and prop . [ prop : small - clusters ] below follow @xcite ; we include them here for completeness .
+ * proof of prop .
[ prop : crossprob ] .
* we define a dual fk model by inserting a bond in the dual lattice , @xmath253 , whenever the corresponding dual edge is not crossed by a bond of the fk configuration on the original lattice , @xmath254 .
the proof is by induction on @xmath196 . for @xmath255 ,
the result follows from rsw - type bounds ( theorem 1 of @xcite see @xcite for the original rsw ) since @xmath256 is equivalent to the _ absence _ of a circuit of dual fk bonds ( i.e. , bonds of the dual fk model ) in the @xmath257-annulus about @xmath258 . by self - duality at the critical point
, this event has the same probability as the absence of a circuit of fk bonds in the original fk model , which in turn is bounded away from one as @xmath35 , by rsw .
now suppose @xmath259 .
then one may do an exploration of the @xmath223 s that touch @xmath260 until @xmath261 are found that reach @xmath262 , making sure that all cluster explorations have been fully completed without obtaining information about the outside of the clusters . at that point , the complement @xmath233 of some random finite @xmath263 remains to be explored and the conditional random - cluster ( fk ) distribution in @xmath233 is @xmath264 with a _
free _ boundary condition on the boundary ( or boundaries ) between @xmath233 and @xmath265 . by rsw , the @xmath266-probability of a crossing by a sequence of fk bonds in @xmath233 of the @xmath257-annulus is bounded above by the original @xmath267 .
thus we have the last claim of the proposition follows from ( [ eq : crossprob ] ) because one may choose @xmath269 ^ 2)$ ] points @xmath270 in @xmath27 so that any @xmath223 of diameter @xmath232 touching @xmath11 will be counted in @xmath271 for at least one @xmath270 .
+ the next proposition corresponds to hypothesis 1.1 of @xcite ( with the exponent @xmath272 there taken to be @xmath124 ) , where it is shown how , for the critical two - dimensional ising model , the hypothesis follows from rsw - type bounds for fk percolation .
such bounds have recently been proved in @xcite .
( a derivation of similar bounds , sufficient to verify hypothesis 1.1 , is also contained in @xcite , but it relies on the convergence of spin - cluster interfaces to @xmath273 , a result that should follow from smirnov s work but has not been proved yet . ) * proof . * using prop .
[ prop : tau ] , we can compare @xmath283 for small @xmath284 as @xmath285 to @xmath286 by using the second inequality of ( [ eq : connectivity ] ) to compare each @xmath287 to the @xmath288 s with @xmath289 in the unit length square centered on @xmath290 ( so that we may take @xmath290 as @xmath291 ) .
since there are approximately @xmath292 such @xmath258 sites , we have that @xmath293 using this lower bound ( with @xmath294 and @xmath295 ) and ( [ eq : theta ] ) , and letting @xmath233 denote the support of @xmath26 and @xmath296 , we have that @xmath297 & \leq & \left(\sup_{x \in d}|f(x)|\right)^2 \limsup_{a\to0 } \theta_a^{2 } \ , e_c\left [ \sum_{i:\text{diam}(\gamma^a_i)\leq\varepsilon } |{\cal c}^a_i \cap d|^2 \right ] \nonumber \\ & \leq & \left(\sup_{x \in
d}|f(x)|\right)^2 \limsup_{a\to0 } \frac{\sum_{z , w \in d_a , ||z - w|| \leq \varepsilon } \tau_c(w / a - z / a)}{\sum_{x , y \in \lambda_{1/a } } \tau_c(y - x ) } \nonumber \\ & \leq & \left(\sup_{x \in d}|f(x)|\right)^2 \limsup_{a\to0 } \frac{k ' ( 1/a)^2 \sum_{z ' \in \lambda_{\varepsilon /a } } \tau_c(z')}{k '' ( 1/a)^2 \sum_{z \in \lambda_{1/(2a ) } } \tau_c(z ) } \nonumber \\ & = & k ' '' \varepsilon^{7/4 } \ , .
\nonumber \,\,\ , \fbox{}\end{aligned}\ ] ] we are now ready to prove the two theorems .
+ * proof of theorem [ thm : subseqential - limits ] .
* let @xmath233 denote the support of @xmath26 ; in view of ( [ eq : second - moment ] ) and ( [ eq : connectivity ] ) ( compare the proof of prop .
[ prop : small - clusters ] ) , @xmath298 ^ 2 \right\rangle_c & = & \limsup_{a \to 0 } e_c\left [ \sum_i ( \mu^a_i(f))^2 \right ] \\ &
\leq & \left(\sup_{x \in d}|f(x)|\right)^2 \limsup_{a \to 0 } \theta_a^2 e_c \left[\sum_i |{\cal c}^a_i \cap d|^2 \right ] < \infty \ ] ] and thus @xmath299 has subsequential limits in distribution as @xmath35 . boundedness of the second moment of @xmath299 and classic ising model results ( see , e.g. , @xcite and references therein ) imply that the fourth moment of @xmath299 remains bounded as @xmath300 . as a consequence ( see , e.g. , problem 14 in section 8.3 of @xcite , p. 164 ) , any subsequential limit of @xmath299 has a finite second moment which is the limit of the second moment of @xmath299 .
thus , the distribution of @xmath299 has subsequential limits in the wasserstein metric ( [ eq : w - distance ] ) as @xmath35 .
since the euclidean distance makes @xmath301 ^ 2 $ ] a compact metric space , the space @xmath302 ^ 2)$ ] of continuous , real - valued functions on @xmath301 ^ 2 $ ] with the supremum norm is separable .
every subspace of a separable metric space is separable , thus the space @xmath303 ^ 2)$ ] of continuous functions with compact support contained in @xmath301 ^ 2 $ ] with the supremum norm is also separable .
any topological space which is the union of a countable number of separable subspaces is separable , which implies that @xmath304 ^ 2)$ ] is separable .
let @xmath305 denote a countable , dense subset of @xmath306 ; it is clear from the above discussion that we can choose @xmath307 , where @xmath308 is a countable , dense subset of @xmath303 ^ 2)$ ] . by a standard diagonalization argument , for every sequence @xmath87
, there exists a subsequence @xmath88 such that , for all @xmath309 , the distribution @xmath310 of @xmath311 has a limit @xmath312 in the wasserstein metric @xmath81 as @xmath313 . by inspection of the definition of @xmath81
, we have the following straightforward inequalities : @xmath314 now consider a function @xmath26 in @xmath82 but not in @xmath305 .
since @xmath26 has compact support , @xmath315 ^ 2)$ ] for some @xmath316 .
if @xmath317 , the positivity of @xmath318 for all @xmath319 ( or the independence of the @xmath61 s in the fk representation ) implies that @xmath320 ^ 2})\right)^2 \right ] \
, , \ ] ] and equation ( [ eq : second - moment ] ) and the first inequality of ( [ eq : connectivity ] ) imply that @xmath321 ^ 2})\right)^2\right]$ ] is bounded as @xmath35 . for @xmath322 and @xmath196 sufficiently large , this leads to @xmath323 ^ 2})\right)^2 \right ] \right)^{1/2 } \ , . \ ] ] ( the 3 in the last term is arbitrary , any number greater that 2 would do , provided that @xmath322 and @xmath196 are sufficiently large . )
if @xmath317 , as @xmath324 , @xmath325 converges to @xmath326 in the wasserstein metric @xmath81 and so the right hand side of the above upper bound for @xmath327 can be made arbitrarily small by first choosing @xmath328 appropriately , and then taking @xmath322 and @xmath196 sufficiently large .
this shows that @xmath90 is a cauchy sequence in @xmath72 .
since @xmath72 is complete , as @xmath92 , @xmath90 converges in the wasserstein metric @xmath81 to a probability distribution @xmath93 .
the continuity of @xmath329 is a consequence of the following inequalities , valid for every @xmath196 , @xmath330 ^ 2 } ) \right]^2 \right\rangle_c^{1/2 } + w_2(p^k_g , p^0_g ) \
, , \end{aligned}\ ] ] where @xmath316 is chosen so large that @xmath331 ^ 2)$ ] .
this implies @xmath332 ^ 2 } ) \right]^2 \right\rangle_c^{1/2 } \,\ ] ] and the conclusion .
we now prove the last statement of the theorem .
since @xmath86 is a nuclear space , we can apply the bochner - minlos theorem ( see for example @xcite , theorem 3.4.2 , p. 52a proof can be found in @xcite ) . in order to do so
, we define @xmath333 and check the following conditions ( where 0 here denotes both the number 0 and the 0 element of @xmath86 ) : the first condition is clear from the definition of @xmath341 since @xmath119 is concentrated at the point @xmath342 when @xmath343 . to establish the second condition ,
let @xmath344 and note that @xmath345 along a converging subsequence , @xmath346 converges to @xmath347 , yielding the desired inequality , @xmath348 .
the remaining step is to establish the continuity of @xmath341 .
first note that convergence in the topology of @xmath86 implies uniform convergence . with this in mind
, the continuity of @xmath341 follows immediately from the continuity of @xmath95 proved earlier , which in particular implies that , if @xmath26 converges to @xmath328 uniformly , the characteristic function of @xmath119 converges pointwise to that of @xmath326 , and so @xmath349 converges to @xmath350 . in conclusion , by an application of the bochner - minlos theorem
, there exists a random , continuous , linear functional @xmath351 with characteristic function @xmath341 . + * proof of theorem [ thm : cutoff - removal ] .
* we first note that the proof of theorem [ thm : subseqential - limits ] works also with @xmath299 replaced by @xmath352 for any @xmath139 , implying in particular convergence of the @xmath105-cutoff field in the wasserstein metric along subsequences of @xmath35 .
this , combined with a standard diagonalization argument , implies that for any sequence @xmath87 , there exists a subsequence @xmath88 such that the distributions of @xmath91 and @xmath118 converge in the wasserstein metric @xmath81 as @xmath92 for all @xmath89 and all @xmath336 .
let @xmath119 and @xmath120 denote the respective limits , and let @xmath90 denote the distribution of @xmath91 and @xmath353 the distribution of @xmath118 . by inspection of the definition of @xmath81 and the positivity of @xmath318 for all @xmath319 ( or the independence of the @xmath61 s in the fk representation )
, we have the following inequalities : @xmath354 \right)^{1/2 } + w_2(p^k_{f , m},p^0_{f , m } ) \ , . \ ] ] the proof of the theorem is concluded by letting first @xmath355 and then @xmath356 , and using the convergence of @xmath90 to @xmath119 and of @xmath353 to @xmath120 in the wasserstein metric @xmath81 , as well as prop . [ prop : small - clusters ] . +
* acknowledgments . *
the author thanks christophe garban and charles m. newman for their suggestions and ellen saada for her patient encouragement during the preparation of this paper .
he also thanks c.m .
newman for many useful comments and discussions , wouter kager for providing fig .
[ fig : loops ] , and an anonymous referee for a careful reading of the manuscript and several useful suggestions .
f. camia , l.r .
fontes and c.m .
newman , the scaling limit geometry of near - critical 2d percolation , _ j. stat .
_ * 125 * , 11551171 ( 2006 ) .
f. camia , l.r .
fontes and c.m .
newman , two - dimensional scaling limits via marked nonsimple loops , _ bull .
* 37 * , 537559 ( 2006 ) .
f. camia , c. garban and c.m .
newman , in preparation .
f. camia and c.m .
newman , two - dimensional critical percolation : the full scaling limit , _ comm .
phys . _ * 268 * , 138 ( 2006 ) .
f. camia and c.m .
newman , sle@xmath357 and cle@xmath357 from critical percolation , in _ probability , geometry and integrable systems _ ( m. pinsky , b. birnir , eds . ) , pp .
103130 , cambridge univ . press , cambridge ( 2008 ) .
j. cardy , conformal field theory and statistical mechanics , in _ exact methods in low - dimensional statistical physics and quantum computing _ , lecture notes of the les houches summer school : volume 89 , july 2008 , ( j. jacobsen , s. ouvry , v. pasquier , d. serban , l. cugliandolo , eds . ) , oxford univ .
press , oxford ( 2010 ) .
j. cardy , sle for theoretical physicists , _
annals of physics _ * 318 * , 81118 ( 2005 ) .
seymour and d.j.a .
welsh , percolation probabilities on the square lattice , in _ advances in graph theory _ ( b. bollobs , ed . ) , annals of discrete mathematics * 3 * , north - holland , amsterdam , pp .
227245 ( 1978 ) .
s. sheffield , exploration trees and conformal loop ensembles , _ duke math .
j. _ * 147 * , 79129 ( 2009 ) .
s. sheffield and w. werner , conformal loop ensembles : construction via loop - soups , arxiv:1006.2373v2 [ math.pr ] ( 2010 ) .
s. smirnov , conformal invariance in random cluster models .
i. holomorphic fermions in the ising model , _ ann .
_ * 172 * , 1435 - 1467 ( 2010 ) .
w. werner , random planar curves and schramm - loewner evolutions , in _ lectures on probability theory and statistics _ , lecture notes in math .
1840 , pp .
107195 , springer , berlin ( 2004 ) . | we study the continuum scaling limit of the critical ising magnetization in two dimensions .
we prove the existence of subsequential limits , discuss connections with the scaling limit of critical fk clusters , and describe work in progress of the author with c. garban and c.m .
newman .
* keywords : * continuum scaling limit , critical and near / off - critical ising model , euclidean field theory , fk clusters . *
ams 2000 subject classification : * 82b27 , 60k35 , 82b43 , 60d05 . |
galactic black hole candidates ( bhc ) are predominantly found in two states : the hard state , in which the x - ray spectrum is a comptonization spectrum emerging from a hot electron cloud with a typical electron temperature of @xmath0150kev @xcite , and the soft state , in which the x - ray spectrum is thermal with a characteristic temperature of @xmath1kev to which a steep power - law is added ( * ? ? ?
* ; * ? ? ?
* and references therein ) .
the x - ray states are also known to be associated with radio states ( see , e.g. , * ? ? ? * for a review ) .
the canonical bhc stays predominantly in the hard state , and only occasionally transits into the soft state for a few months ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* see also fig .
[ fig : softlag ] ) .
in addition to spectral analysis , timing analysis can also yield insight into the radiation mechanisms in bhcs . apart from power spectrum analysis @xcite ,
higher order statistics like the frequency - dependent coherence function and time lags have proven to be useful in evaluating accretion models @xcite .
although very different in their specific physical assumptions , most models for the hard and the soft state assume that the observed time lags are an indicator of some characteristic geometrical size of the compton corona . in the case of ,
the apparent decrease of the x - ray lags during the 1996 soft state transition of the source was considered evidence that the size of the comptonizing corona during the soft state is smaller than during the hard state @xcite .
first comparisons of transition and soft state lags to hard state lags , however , indicated that the physical interpretation has to be more complex @xcite .
in 1998 we initiated a monitoring campaign of with rxte to systematically study long term variations of the hard state properties over a period of several years .
previous results have been presented elsewhere @xcite . in 1998 weekly pointings of @xmath03ks duration and in 1999 two - weekly pointings of @xmath010ks duration were performed . in this contribution
we present results of the analysis of these observations .
together with archival observations from 1996 , we consider 103 observations .
we concentrate on the data from the proportional counter array ( pca ; * ? ? ?
* ) , and use the rxte all sky monitor ( asm ; * ? ? ?
* ) to place our observations in the context of cyg x-1 s long term evolution .
the data were reduced using the procedures described in detail elsewhere @xcite .
we mainly show results obtained from lightcurves with a resolution of 4 or 16ms observed in two energy bands @xmath24kev and @xmath0813kev .
the chosen energy bands are known to be well suited for time lag studies of this source @xcite .
the exact bands used to compare the 1996 data to the monitoring data were the closest matches possible for the available pca modes .
furthermore , we describe the results of the spectral analysis of standard pca spectra in the 220kev range .
we present the long term evolution of the averaged time lags , the frequency dependent time lag spectra and power spectra , and the spectral evolution during the flare of 1998 july .
finally , we discuss these results in the light of current accretion models .
rxte asm count rate ( line ) and mean time lag ( symbols ) between the energy band below 4kev and the energy band from 8 to 13kev in the frequency range from 3.2 to 10hz for the 1998/1999 monitoring campaign as well as for the rxte data obtained during the 1996 soft state . during hard state phases
there is a high correlation between the soft x - ray luminosity and the time lag.,scaledwidth=75.0% ] comparison between the power spectral densities ( 813kev ) and time lags for the 1996 state transition , a characteristic 1996 hard state observation , and the 1998 july failed state transition ( subfigures a and c ) .
subfigures b and d compare the psds and lags between the hard and the soft state .
, scaledwidth=70.0% ] comparison between the power spectral densities ( 813kev ) and time lags for the 1996 state transition , a characteristic 1996 hard state observation , and the 1998 july failed state transition ( subfigures a and c ) .
subfigures b and d compare the psds and lags between the hard and the soft state .
, scaledwidth=70.0% ] in fig .
[ fig : softlag ] , we display the daily averaged rxte asm count rate superposed on the mean time lag between the soft and the hard energy bands in the frequency band from 3.2 to 10hz .
we have found that the x - ray time lag is most variable in this frequency band @xcite . in the lower panel ,
the 1998/1999 monitoring is displayed .
it shows that in the hard state there is a clear correlation between the soft x - ray luminosity and the mean time lag ( pearson s correlation coefficient of 0.78 ) .
this is especially true during `` flaring '' events ( e.g. , in 1998 july , see below ) . in the light of current accretion models ,
however , we would expect the lag to decrease as the source softens ( with the corona becoming smaller ) .
this was thought to happen during the 1996 full state transition , where the lag was seen to be considerably smaller in the soft state than immediately before and after @xcite .
considering the hard state monitoring observations as well , however , we see a different picture ( upper panel of fig . [
fig : softlag ] ) : while the absolute values of the lags are comparable in the hard state and in the soft state , they are significantly longer during the transitions .
in fact , the frequency dependence of the lag is very similar for the soft and hard states ( fig .
[ fig : flare]d ) .
it is interesting to note that for the hard energy band ( 813kev ) and the frequency range considered here , the timing behavior as characterized by the power spectral density ( psd ) also looks similar ( fig .
[ fig : flare]b ) . a qualitative model for explaining this psd phenomenology in terms of transfer of variability structures through the disk / corona system has recently been presented by @xcite .
previous analyses of two hard state observations already suggested that the soft and hard state lags might not be so different as previously thought @xcite .
our numerous hard state monitoring observations now clearly indicate that _ the x - ray lag spectrum of is rather independent of the spectral state_. the _ enhanced lags are associated with transition or flaring intervals _ , and not with the state of the source itself .
of special interest are the episodes of enhanced asm count rates in the hard state , which , as we have seen , also show larger mean time lags than the `` canonical '' hard state . figs .
[ fig : flare]a and c display the psd and the time lag spectrum for an observation performed during the 1998 july flare and compares them to a typical hard state observation as well as to the transition phase of the 1996 state transition .
compared to the typical hard state lag spectrum the appearance of an additional component in the 1 to 10hz range can be seen during the 1998 july flare .
this component becomes more prominent and `` flattens '' during the hard to soft transition in 1996 may , with the lag being longer by almost a factor of 10 at 6hz as compared to 1996 december .
as has been reported by @xcite , the psds of the state transition show an additional steep component at lower frequencies , a shift of the break frequency to higher frequencies , and additional `` qpo - like '' features .
the psd of the 1998 flare is intermediate between the hard state and the state transition , i.e. , the break frequency has moved from @xmath3hz to @xmath4hz .
furthermore , a quasi - periodic oscillation appears at @xmath5hz , i.e. , in the frequency range that exhibits the enhanced lag ( the qpo might also be present in the hard state psd at @xmath6hz ) . since the flaring episode displays all the features of the real state transition , presumably at the onset of their appearance
, we identify it as a `` failed state transition '' .
this interpretation as a failed state transition is confirmed by the spectral analysis : outside of the `` flaring episode '' the hard state x - ray spectrum as seen by the pca can be described by the canonical power - law ( photon index @xmath7 ) , reflected off neutral material with a covering angle of @xmath8 ( the inclination has been fixed at @xmath9 ) .
an additional broad ( @xmath10kev ) fe line at 6.4kev with an equivalent width of 150ev is required ( @xmath11 ; note that the @xmath12% systematics of @xcite have been applied ) . during the flare the spectrum softens to @xmath13 and the covering factor increases to @xmath14 . at low energies
an additional soft spectral component appears ( modeling this component as a multi - temperature disk black body results in @xmath15kev .
the soft component contributes @xmath07% of the flux at 3kev .
thus , although the power law component still dominates the higher energy spectrum , there is clear evidence for a strong soft component , such that the 1998 july event should be associated with a failed state transition .
although the flaring episodes introduce considerable long term variability into the hard state properties , the timing characteristics can be reproduced to great detail in observations separated by more than a year ( fig .
[ fig : hard ] ) : the time lag spectra and the psds of the 1996 december and the 1998 february observation are identical .
this includes the `` shelves '' in the lags and the `` wiggles '' in the psds that might indicate that multiple processes are responsible for the observed timing behavior ( see also * ? ? ?
comparison of the psds and time lags for two hard state observations separated by 14 months .
both indicators for the temporal variability are virtually indistinguishable.,title="fig:",scaledwidth=38.0% ] comparison of the psds and time lags for two hard state observations separated by 14 months .
both indicators for the temporal variability are virtually indistinguishable.,title="fig:",scaledwidth=38.0% ]
since comptonization models generally give a satisfactory explanation for the x - ray spectrum , most models for the generation of the x - ray time lags assume that the lags are at least partially produced by scattering of seed photons ( which might have an intrinsic time lag ) in a compton corona ( * ? ? ?
* and references therein ) .
comptonization can not be the only cause for the time lags , however , as the magnitude of the lag implies very large compton coronae ( @xmath16300 gravitational radii * ? ? ?
* ) , in conflict with the assumption of most comptonization models . furthermore , the energy dependence of the lags is also in disagreement with comptonization @xcite .
we note , however , that the coherence function in the frequency regime considered here is not equal to unity , but slightly smaller .
this might be an indication that ( part of ) the time lag is produced by non - linear processes in the accretion disk , and not by the time delay due to the compton scatterings in the corona ( churazov , 2000 , priv .
comm . ) .
nevertheless , although all of these arguments point against comptonization as the sole cause for the x - ray lags , one is tempted to assume that the magnitude of the lag is at least somehow proportional to the characteristic size of the x - ray emitting region . accepting this paradigm ,
however , implies that the size of the x - ray emitting corona during the soft and hard states is the same as the lags during these states are identical .
this is puzzling , given that the luminosity of the corona , and thus the release mechanism for the potential energy of the accreted material , is so different between the hard and the soft state . during state transitions , larger x - ray lags
are observed in the 110hz band .
these larger lags might be related to the radio outbursts seen during the transitions @xcite .
galactic bhcs in the hard state emit optically thick radio emission , while the soft state is radio quiet . during transitions
, the radio emission tends to be optically thin and highly variable ( e.g. , gx 339@xmath174 , @xcite , but also cyg x-1 itself ; see @xcite for a review ) .
for cyg x-1 , there exists a good correlation between the radio and the soft x - ray flux @xcite , and state transitions are correlated to radio outbursts ( * ? ? ? * see also @xcite )
. such outburst are usually associated with the ejection of synchrotron radiation emitting outflows ( see * ? ? ?
* ; * ? ? ?
* for the case of gx 339@xmath174 ) .
we suggest that the radio outflows and the larger lags might be related @xcite .
one possible source for the electrons seen in the radio outflow might be the accretion disk corona . during state transitions , when the accretion disk is thought to reconfigure , the corona might be ejected and then observed as the radio outflow . in this picture ,
soft x - rays from the accretion disk might get compton upscattered in the base of this outflow . as the base is much larger than the common accretion disk corona
, one would expect larger x - ray lags .
we note that such a picture , where part of the observed lags is due to the presence of a jet , has been previously suggested by van paradijs ( 1999 , priv .
, see also @xcite ) .
this work has been supported by dfg grant sta 173/22 , by dlr grant 50 oo 9605 , by nasa grants nag5 - 3072 , nag5 - 3225 , and nag5 - 7265 , and by travel grants from the daad and the nsf .
corbel , s. , r. fender , t. tzioumis , m. a. nowak , v. mcintyre , et al . , coupling of the x - ray and radio emission in the black hole candidate and compact jet source gx 339@xmath174 .
_ a&a _ , * 359 * , 251268 , 2000 .
cui , w. , temporal x - ray properties of galactic black hole candidates : observational data and theoretical models . in , eds .
poutanen , j. , & svensson , r. , astron .
pacific , conf .
161 , p. 102
, 1999 .
jahoda , k. , j. h. swank , a. b. giles , m. j. stark , t. strohmayer , et al . , in orbit performance and calibration of the rossi x - ray timing explorer ( rxte ) proportional counter array ( pca ) . in , ed .
siegmund , o. h. , proc .
spie , no .
2808 , pp .
5970 , bellingham , wa .
spie , 1997 .
poutanen , j. , accretion disc - corona models and x/@xmath18-ray spectra of accreting black holes . in , eds .
abramowicz , m. a. , bjrnsson , g. , & pringle , j. e. , p. 100
, cambridge .
cambridge univ . press , 1998 .
wilms , j. , m. a. nowak , j. b. dove , r. p. fender , and t. di matteo , low luminosity states of the black hole candidate gx 339@xmath174 .
i. asca and simultaneous radio / rxte observations .
_ apj _ , * 522 * , 460475 , 1999 .
zhang , s. n. , i. f. mirabel , b. a. harmon , r. a. kroeger , l. f. rodriguez , et al . , galactic black hole binaries : multifrequency connections . in , eds .
dermer , c. d. , strickman , m. s. , & kurfess , j. d. , aip conf .
410 , pp . 141162 , woodbury .
aip , 1997 . | we present temporal and spectral results from monitoring with the rossi x - ray timing explorer ( rxte ) in 1998 and 1999 .
we concentrate on the long term evolution of the hard state timing properties , comparing it to the 1996 soft state evolution .
this leads to the following results : 1 .
the hard and soft state time lag spectra are very similar , 2 . during state transitions , the lags in the 110hz range increase by more than an order of magnitude , 3 . in the hard state itself , flaring events can be seen the temporal and spectral evolution during the flare of 1998 july identifies it as a `` failed state transition '' . during ( failed ) state transitions , the time lag spectra and the power spectra change predominantly in the 110hz range .
we suggest that this additional variability is produced in ejected coronal material disrupting the synchrotron radiation emitting outflows present in the hard state . |
the geodesic flow on a riemannian manifold represents the extremals of the least action principle , namely it is determined by the motion of a certain physical system in the manifold .
it is known that the geodesic equations are second order non - linear differential equations and they usually appear in the form of euler - lagrange equations of motion .
magnetic curves generalize geodesics . in physics ,
such a curve represents a trajectory of a charged particle moving on the manifold under the action of the magnetic field .
let @xmath0 be an @xmath1-dimensional riemannian manifold .
a _ magnetic field _ is a closed 2-form @xmath2 on @xmath3 and the _ lorentz force _ of a magnetic field @xmath2 on @xmath0 is an @xmath4 tensor field @xmath5 given by @xmath6 the _ magnetic trajectories _ of @xmath2 are curves @xmath7 on @xmath3 that satisfy the _ lorentz equation _ ( sometimes called the _ newton equation _ ) @xmath8 lorentz equation generalizes the equation satisfied by the geodesics of @xmath3 , namely @xmath9 therefore , from the point of view of the dynamical systems , a geodesic corresponds to a trajectory of a particle without an action of a magnetic field , while a magnetic trajectory is _ a flowline of the dynamical system _ , associated with the magnetic field .
since the lorentz force is skew symmetric we have @xmath10 so the magnetic curves ( trajectories ) have constant speed @xmath11 .
when the magnetic curve @xmath12 is arc length parametrized @xmath13 , it is called a _
normal magnetic curve_. recall that a vector field @xmath14 on @xmath3 is _ killing _ if and only if it satisfies the killing equation : @xmath15 for every vector fields @xmath16 on @xmath3 , where @xmath17 is the levi civita connection on @xmath3 .
a typical example of uniform magnetic fields is obtained by multiplying the volume form on a riemannian surface by a scalar @xmath18 ( usually called _ strength _ ) .
when the surface is of constant gaussian curvature @xmath19 , trajectories of such magnetic fields are well known .
more precisely , on the sphere @xmath20 , @xmath21 , trajectories are small ( euclidean ) circles of radius @xmath22 , on the euclidean plane they are circles and the period of motion equals to @xmath23 , while , on a hyperbolic plane @xmath24 , @xmath21 , trajectories can be either closed curves ( when @xmath25 ) , or open curves . moreover , when @xmath26 normal trajectories are horocycles ( see e.g. ) .
this problem was extended also for different ambient spaces .
for example , if the ambient is a complex space form , khler magnetic fields are studied ( see @xcite ) , in particular , explicit trajectories for khler magnetic fields are found in the complex projective space @xmath27 @xcite .
khler magnetic fields appear in theoretical and mathematical physics , varying from quantum field theory and string theory to general relativity .
if the ambient is a contact manifold , the fundamental 2-form defines the so - called _ contact magnetic field_. interesting results are obtained when the manifold is sasakian , namely the angle between the velocity of a normal magnetic curve and the reeb vector field is constant ( see @xcite ) .
moreover , explicit description for normal flowlines of the contact magnetic field on a 3-dimensional sasakian manifold is known @xcite . in the case of a 3-dimensional riemannian manifold @xmath28
, 2-forms and vector fields may be identified via the hodge star operator @xmath29 and the volume form @xmath30 of the manifold .
thus , magnetic fields mean divergence free vector fields ( see e.g. @xcite ) .
in particular , killing vector fields define an important class of magnetic fields , called _ killing magnetic fields_. it is known that geodesics can be defined as extremal curves for the action energy functional . a variational approach to describe killing magnetic flows in spaces of constant curvature is given in @xcite .
note that , one can define on @xmath3 the _ cross product _ of two vector fields @xmath31 as follows @xmath32 if @xmath14 is a killing vector field on @xmath3 , let @xmath33 be the corresponding killing magnetic field .
by @xmath34 we denote the inner product .
then , the lorentz force of @xmath35 is ( see @xcite ) @xmath36 consequently , the lorentz force equation ( [ lorentzeq ] ) can be written as @xmath37 in what follows we consider the 3-dimensional euclidian space @xmath38 , endowed with the usual scalar product @xmath39 .
the fundamental solutions of are @xmath40 and they give a basis of killing vector fields on @xmath38 . here
@xmath41 denote the global coordinates on @xmath38 and @xmath42 is regarded as a vector space .
the easiest example is to consider the killing vector field @xmath43 .
( similar discussions can be made for @xmath44 and @xmath45 , respectively . )
its trajectories are helices with axis @xmath46 , namely @xmath47 , where @xmath48 and @xmath49 .
an interesting fact is that lancret curves ( i.e. general helices ) in @xmath38 are characterized by the following property ( in our framework ) : they are magnetic trajectories associated with magnetic fields parallel to their axis . a similar result , relating killing magnetic fields and lancret curves is provided on the 3-sphere ( see e.g. @xcite ) .
theorems of lancret for general helices in 3-dimensional real space forms are presented in @xcite . in this paper
we consider the following magnetic field @xmath50 in @xmath38 , determined by the killing vector field @xmath51 .
the other two rotational vector fields @xmath52 and @xmath53 give rise to analogue classifications for corresponding magnetic trajectories .
the aim of this note is to find all magnetic curves corresponding to @xmath35 .
the main result we obtain is the following : * theorem .
* _ the magnetic trajectories of the killing magnetic field @xmath35 are : ( a ) planar curves situated in a vertical strip ; ( b ) circular helices and ( c ) curve parametrized by _ @xmath54 _ where @xmath55 and @xmath56 satisfy _
@xmath57 _ and @xmath58 is a polynomial of degree @xmath59 . _ we are able to obtain explicit solutions in case ( c ) and we represent some examples by using numerical approximations for some integrals .
recall , for later use , some basic facts on _ normal elliptic integral of the first kind _
( see for example @xcite ) : @xmath60 where @xmath61 and @xmath62 .
the angle @xmath63 is called _ jacobi amplitude _ and the function @xmath64 in known as _
jacobi elliptic sine_. the number @xmath65 is called _ modulus _ and for applications to engineering and physics it belongs to @xmath66 .
let us consider the killing vector field @xmath67 on @xmath68 , which defines the magnetic field @xmath50 .
the lorentz force @xmath69 acts on the vector space @xmath70 as follows : @xmath71 for the euclidian space @xmath38 the lorentz force equation becomes @xmath72 where the curve @xmath73\longrightarrow{\mathbb{e}}^3 $ ] , @xmath74 is parametrized by arc length , namely @xmath75 and at the moment @xmath76 it passes through the point @xmath77 , with the velocity @xmath78 , such that @xmath79 * proof of the theorem . * our aim is to determine the magnetic curves of @xmath35 .
the equation ( [ magn ] ) yields the following ordinary differential equations system @xmath80 in order to solve it , note that from the first two equations we get a prime integral @xmath81 while from the third equation we obtain @xmath82 notice that @xmath83 can not vanish identically ( on a subinterval of @xmath84 ) .
indeed , if @xmath85 then @xmath86 , @xmath87 and @xmath88 with @xmath89 .
hence , @xmath90 , @xmath91 and combining with we get a contradiction .
it follows that one can not have horizontal magnetic curves corresponding to @xmath14 . in the sequel it is more convenient to consider cylindrical coordinates @xmath92 on @xmath93 .
thus , for our curve we have @xmath94 where @xmath95 , @xmath96 .
* case i. * first we study the general case , when @xmath83 is not constant ( equivalently @xmath55 is not constant ) . the relations ( [ xy ] ) and ( [ zp ] ) lead to @xmath97 @xmath98 where we put @xmath99 and @xmath100 .
the arc length parametrization condition ( [ arc ] ) , together with , becomes @xmath101 multiplying by @xmath102 , using and denoting @xmath103 by @xmath104 , for all @xmath105 , one gets @xmath106 we start to study the above differential equation for some particular values of the constants @xmath107 and @xmath108 .
if @xmath109 i.e. @xmath110 it follows that the angle @xmath56 is constant , @xmath111 , so the magnetic trajectory is a planar curve , with @xmath112 more precisely , the curve lies in the plane @xmath113 .
the initial conditions expressed in cylindrical coordinates , may be written as @xmath114 and the condition @xmath110 becomes @xmath115 , @xmath116 , for a certain @xmath117 .
it follows that @xmath118 @xmath119 since @xmath120 it follows that @xmath121 .
let us solve the equation ( [ f ] ) , for three particular situations arising from the initial conditions : 1 . if @xmath122 and @xmath123 , then the equation ( [ f ] ) takes the form @xmath124 and it has solution if and only if @xmath125 , i.e. @xmath126 $ ] , so the magnetic curve @xmath7 lies inside a cylinder .
in fact , being a planar curve , @xmath7 stays in a vertical strip centered in @xmath127 and of width @xmath128 .
+ we have @xmath129 and we consider only the _ plus _ sign ( the other situation may be treated in similar way ) .
supposing @xmath130 , we have that @xmath131 and the integral @xmath132 are strictly increasing functions .
thus , the equation @xmath133 has a unique solution in the interval @xmath134 , namely @xmath135 , where @xmath136 is the inverse function of @xmath137 .
consequently , @xmath138 .
in fact @xmath136 may be expressed in terms of the elliptic functions .
more precisely , @xmath139 where @xmath140 is determined by @xmath141 . + summarizing , the magnetic curve is given by @xmath142 + in order to draw a picture of our curve , one can use matlab to compute the parametrization . the idea is to calculate the integrals numerically , as riemann sums .
see appendix .
[ ii ] 2 .
if @xmath143 , then the equation ( [ f ] ) becomes @xmath144 from which we have that @xmath145 equivalently @xmath146 , so the magnetic curve @xmath7 stays inside a cylinder of radius 2 .
in fact , being planar , the curve lies in a vertical strip centered on @xmath147-axis .
the equation can be written in the form @xmath148 taking the _ plus _ sign , one gets the solution @xmath149 where @xmath150 .
hence @xmath151 and the magnetic curve is parametrized by @xmath152 we draw a picture of this ( planar ) curve . [
fig : sol_03 ] , @xmath153,width=283 ] let us finalize the examination of the equation for @xmath122 .
the polynomial @xmath154 has three real solutions , namely @xmath155 , @xmath156 and @xmath157 .
if @xmath131 is a solution for , then @xmath158 should be negative .
recall that @xmath121 .
we have 1 . if @xmath159 , then @xmath160 .
it follows that @xmath161 and the discussion is similar as in case @xmath123 .
more precisely we have @xmath162 where @xmath140 is defined by @xmath163 .
2 . if @xmath164 , then @xmath165 .
it follows that @xmath166 .
thus , the curve @xmath7 lies between two cylinders since @xmath167 . as before ,
the curve is situated in a union of two vertical strips .
again , the discussion is similar as in case @xmath123 . in terms of elliptic functions , we may write @xmath168 where @xmath140 is defined by @xmath169 . + in order to visualize an example , consider the following initial conditions : @xmath170 , @xmath171 , @xmath172 and @xmath173 , @xmath174 , @xmath175 ( this yields @xmath122 and @xmath176 ) .
+ we will use again matlab to compute the integrals ( numerically ) and to draw the picture .
[ fig : exb_pg5 ] , @xmath176,width=283 ] return to for @xmath177 and notice that the equation @xmath178 has the discriminant @xmath179\ ] ] and the following situations appear : * the equation has three distinct solutions iff @xmath180 * the polynomial @xmath58 has multiple roots iff @xmath181 . *
the polynomial @xmath58 has one real root and two complex conjugate roots iff @xmath182 .
a detailed analysis of the above situations , lead us to conclude , after taking into account classical vite s formulas , that the equation ( [ f ] ) has solutions if and only if @xmath183 . indeed , if @xmath182 , let @xmath184 and @xmath185 be the complex solutions of , and @xmath186 its real solution .
then , the ode can be rewritten as @xmath187 where @xmath188 denotes the real part of the complex number @xmath189 . from the third vite s formula
we conclude that @xmath186 should be negative , and consequently , the previous equality can not occur . on the other hand , if @xmath181 , analyzing the coefficients one can not have a triple root ( since @xmath190 .
hence , let @xmath191 be the double root , and let @xmath192 be the third one . with a similar argument as above , @xmath186 is negative and the ode becomes @xmath193 again
, this equality can not hold .
it follows that @xmath194 should be ( strictly ) positive .
let @xmath189 , @xmath186 , @xmath195 be the three distinct solutions of .
the third vite s formula yields @xmath196 , and hence * either @xmath197 are all negative , * or two of them ,
@xmath189 and @xmath186 , are positive and the third one , @xmath198 , is negative . in case a ) the ode : @xmath199 has no solution .
this happens if and only if @xmath200 and @xmath177 ( for the proof use the second and the third vite s formulas ) together with @xmath183 ( for example if @xmath201 and @xmath202 ) . in case
b ) , equivalently to @xmath183 , @xmath121 and @xmath203 , the equation has a solution in the interval defined by the positive solutions @xmath204 of .
since the function @xmath205 is strictly increasing , @xmath133 has a unique solution @xmath131 , denoted by @xmath206 .
thus we have @xmath207 , and @xmath208 . in this case , the magnetic curve @xmath7 is given by @xmath209 @xmath210 we may express @xmath55 in terms of elliptic functions , namely @xmath211 where @xmath212 , @xmath213 , and @xmath214 . in the appendix we will draw a picture ( using the same technique in matlab as before ) corresponding to the following data : @xmath215 , @xmath216 , for which we have @xmath217 , @xmath218 and @xmath219 .
[ ex_abc ] the situation @xmath220 can be treated in similar way .
finally , notice that for @xmath221 we get @xmath222 and this case was discussed above .
* case ii .
* now , let us study the remaining case when @xmath223 .
we immediately have that @xmath224 and from ( [ zp ] ) we obtain @xmath225 this means that the magnetic trajectory @xmath7 lies on the circular cylinder of radius @xmath226 .
two subcases must be discussed : @xmath227 and @xmath228 .
ii.1 : : in the case when @xmath227 the magnetic curve is given by @xmath229 y(t)=y_0\cos(\sqrt{-w_0}~t)+\frac{v_0}{\sqrt{-w_0}}\sin(\sqrt{-w_0}~t)\\[2 mm ] z(t)=z_0+tw_0 .
\end{cases}\ ] ] this curve is a helix around the above cylinder .
+ at this point , we have to find which are the initial conditions leading this situation . to do this , using and
, we should have the following relations @xmath230 where @xmath231 and @xmath232 have the usual meaning and @xmath233 .
ii.2 : : if @xmath228 , the ode system has the following solution @xmath234 but in this case the condition is satisfied if and
only if @xmath235 and @xmath236 . this situation can not occur .
as we have already said in introduction , the best known example of magnetic fields in the euclidean space @xmath38 is furnished by the 2-form @xmath237 , corresponding to the killing vector field @xmath238 . in this section
we consider the killing magnetic field @xmath239 , determined by the killing vector field @xmath240 on @xmath38 , where @xmath241 is an arbitrary constant .
we briefly describe its magnetic curves .
the action of the lorentz force @xmath242 on the vector space @xmath70 is given by : @xmath243 solving the lorentz force equation @xmath244 , we obtain the family of magnetic curves @xmath245 , parametrized by @xmath246 write the first frnet equation @xmath247 where @xmath248 is the curvature and @xmath249 is the normal of the curve . using the equation we obtain that the square of the curvature is @xmath250 moreover , classical computations give the torsion @xmath251 notice that even both the curvature @xmath248 and the torsion @xmath252 depend on the strength @xmath18 , the ratio @xmath253 does not .
we conclude with some comments : 1 .
if @xmath254 the curvature is @xmath255 and the torsion is @xmath256 so the magnetic line is a ( planar ) circle .
2 . if @xmath257 then @xmath258 , so the magnetic curves are vertical lines . 3 . in other cases
the magnetic curves are circular helices .
in this section we present a matlab program in order to compute , by numerical approximation of the involved integrals , the parametrization of magnetic curve obtained in case
i ( i ) from page . since the curve is planar we consider @xmath259 .
* acknowledgements . *
the first authors is a postdoctoral researcher in the framework of the program posdru 89/1.5/s/49944 , al .
i. cuza university of iai , romania .
the second author is supported by a fulbright grant no .
498 at the michigan state university , usa . | we explicitly determine all magnetic curves corresponding to the killing magnetic fields on the 3-dimensional euclidean space . * keywords and phrases . * killing magnetic field , lorentz force , magnetic curve . * 2010 msc : * 53a04 , 65d17 |
this paper is devoted to nonlinear schrdinger equations ( nls ) of the form @xmath1 here , @xmath2 is a complex valued function , @xmath3 is a possibly rough / discontinuous potential and @xmath4 is a smooth function ( in terms of the density @xmath5 ) that describes the nonlinearity .
a common example is the cubic nonlinearity given by @xmath6 , for @xmath7 , for which the equation is known as the gross - pitaevskii equation modeling for instance the dynamics of bose - einstein condensates in a potential trap @xcite . in this paper
we study galerkin approximations of the nls using a finite element space discretization to account for missing regularity due to a possibly discontinuous potential and we use a crank - nicolson time discretization to conserve two important invariants of the nls , namely the mass and the energy
. we aim at deriving rate - explicit a priori error estimates and the influence of rough potentials on these rates .
the list of references to numerical approaches for solving the nls ( both time - dependent and stationary ) is long and includes @xcite and the references therein .
a priori error estimates for @xmath8 finite element approximations for the nls have been studied in @xcite , where an implicit euler discretization is considered in @xcite , a mass conservative one - stage gauss - legendre implicit runge - kutta scheme is analyzed in @xcite , mass conservative linearly implicit two - step finite element methods are treated in @xcite and higher order ( dg and cg ) time - discretizations are considered in @xcite ( however these higher order schemes lack conservation properties ) . the only scheme that is both mass and energy conservative at the same time is the modified crank - nicolson scheme analyzed by sanz - serna @xcite and akrivis et al .
@xcite , which is also the approach that we shall follow in this contribution .
the analysis of this modified crank - nicolson scheme is devoted to optimal @xmath9-error estimates for sufficiently smooth solutions in both papers @xcite and @xcite .
sanz - serna treats the one - dimensional case @xmath10 and periodic boundary conditions and akrivis et al .
consider @xmath11 and homogeneous dirichlet boundary conditions .
although the modified crank - nicolson scheme is implicit , in both works , optimal error estimates require a constraint on the coupling between the time step @xmath12 and the mesh size @xmath13 . in @xcite
the constraint reads @xmath14 whereas a relaxed constraint of the form @xmath15 is required in @xcite .
the results are related to the case of the earlier mentioned cubic nonlinearity of the form @xmath6 and a potential is not taken into account .
the present paper generalizes the results of akrivis et al .
@xcite to the case of a broader class of nonlinearities and , more importantly , accounts for potential terms in the nls .
if the potential is sufficiently smooth , even the previous constraints on the time step can be removed without affecting the optimal convergence rates . to the best of our knowledge , the only other paper that includes potential terms in a finite element based nls discretization is @xcite which uses a one - stage gauss - legendre implicit runge - kutta scheme that is not energy - conserving .
while these results essentially require continuous potentials , many physically relevant potentials are discontinuous and very rough .
typical examples are disorder potentials @xcite or potentials representing quantum arrays in the context josephson oscillations @xcite . as the main result of the paper
, we will also prove convergence in the presence of such potentials with convergence rates .
the rates are smaller than the optimal ones for smooth solutions and a coupling condition between the discretization parameters shows up again .
while the sharpness of these results remains open , we shall stress that we are not aware of a proof of convergence of any discretization ( finite elements , finite differences , spectral methods , etc . ) of the nls in the presence of purely @xmath16-potentials and that we close this gap with our paper .
the structure of this article is as follows .
section [ sec : problem ] introduces the model problem and its discretization .
the main results and the underlying assumptions are stated in section [ sec : main - results ] .
sections [ s : errorsemi][s : errorfull ] are devoted to the proof of these results .
we present numerical results in section [ sec : numexp ] .
some supplementary material regarding the feasibility of our assumptions is provided as appendix [ appendix - b ] .
let @xmath17 ( for @xmath18 ) be a convex bounded polyhedron that defines the computational domain .
we consider a real - valued nonnegative disorder potential @xmath19 . besides being bounded
, @xmath20 can be arbitrarily rough . given such @xmath20 , some finite time @xmath21 and some initial data @xmath22 , we seek a wave function @xmath23,h^1_0({\mathcal{d}}))$ ] with @xmath24,h^{-1}({\mathcal{d}}))$ ] such that @xmath25 and @xmath26 for all @xmath27 and almost every @xmath28 $ ] .
note that any such solution automatically fulfills @xmath29,l^2({\mathcal{d}}))$ ] so that @xmath30 makes sense .
the nonlinearity in the problem is described by a smooth ( real - valued ) function @xmath4 with @xmath31 and the growth condition @xmath32 observe that this implies by sobolev embeddings that @xmath33 is finite for any @xmath34 .
we define @xmath35 then , for any @xmath36 , the ( non - negative ) energy is given by @xmath37 [ prop - exist - and - unique ] there exists at least one solution to problem .
for a corresponding result we refer to ( * ? ? ?
* proposition 3.2.5 , remark 3.2.7 , theorem 3.3.5 and corollary 3.4.2 ) . however , uniqueness is only known in exceptional cases
. if @xmath38 and @xmath39 the solution is unique locally in time , i.e. , on a subinterval @xmath40 ( cf .
* theorem 3.6.1 ) ) . for further settings that guarantee uniqueness ,
see ( * ? ? ?
* corollary 3.6.2 , remark 3.6.3 and remark 3.6.4 ) .
@xmath41 _ temporal discretization .
_ we consider a time interval @xmath42 $ ] and a corresponding family of admissible partitions .
a partition @xmath43 is admissible if the @xmath44th time interval is given by @xmath45 $ ] and if @xmath46 .
furthermore , we assume that the family of partitions is quasi - uniform , i.e. if @xmath47 denotes the @xmath44th time step size and if the maximum is denoted by @xmath48 , then there exists a ( discretization independent ) constant @xmath49 such that @xmath50 for all partitions from the family .
@xmath41 _ spatial discretization .
_ for the space discretization we consider a finite dimensional subspace @xmath51 of @xmath52 that is parametrized by a mesh size parameter @xmath13 .
we make two basic assumptions on @xmath51 which are fulfilled for @xmath8 lagrange finite elements on quasi - uniform meshes .
let us for this purpose introduce the ritz - projection @xmath53 . for @xmath54
the ritz - projection @xmath55 is the unique solution to the problem @xmath56 in the following , we make an assumption on the approximation quality of @xmath57 , that is that there exists a generic @xmath13-independent constant @xmath58 such that @xmath59 the second assumption is the availability of an inverse estimate , i.e. we assume that there exists a generic @xmath13-independent constant @xmath60 such that @xmath61 in addition , we assume the existence of @xmath13-independent @xmath62 with @xmath63 the above assumptions are standard in the context of finite elements if quasi - uniformity is available . for instance , for simplicial @xmath8 lagrange finite elements on a quasi - uniform mesh , the estimates and are fulfilled .
the last property can be verified by splitting @xmath64 for some @xmath16-stable clment - type quasi - interpolation operator .
the estimate then follows from inverse inequalities and @xmath65-estimates for @xmath57 . with these definitions ,
we introduce the fully discrete crank - nicolson method as follows .
[ crank - nic - gpe ] we consider the space and time discretizations as detailed above .
let @xmath66 be the initial value from problem and let @xmath67 .
then for @xmath68 , the fully discrete crank - nicolson approximation @xmath69 is given by @xmath70 for all @xmath71 and where @xmath72 .
the scheme is mass conserving and energy conserving , i.e. we have @xmath73 for all @xmath74 .
the mass conservation is verified by testing with @xmath75 in and taking the real part .
the energy conservation is verified by testing in with @xmath76 and taking the imaginary part .
the conservation properties do not immediately guarantee robustness with respect to numerical perturbations ( for instance arising from round - off errors ) , however , it still can be proved that even the perturbed approximations remain uniformly bounded .
let @xmath77 and let @xmath78 ( for @xmath79 ) be an @xmath9-perturbation of the discrete problem .
we can think of @xmath80 as representing numerical errors .
let @xmath69 for @xmath81 be any solution to the ( fully - discrete ) perturbed problem @xmath82 for all @xmath71 .
then the solutions remain uniformly bounded in @xmath9 with @xmath83 we test in the problem formulation with @xmath84 and take the real part .
this yields @xmath85 hence @xmath86 applying this iteratively gives us @xmath87
while this basic stability of the method does not require any additional smoothness assumptions , our quantified convergence and error analysis of the method relies on the regularity of @xmath88 .
we will use three types of regularity assumptions . *
assume that @xmath89,h^2(\mathcal{d}))$ ] , @xmath90 and @xmath91 .
* assume that @xmath92 and @xmath93 .
* assume that @xmath94 .
the first assumption allows the proof of convergence rates for the time - discretization and the second one is related to the optimal convergence rates for the space - discretization .
note that the high spatial regularity @xmath89 , h^2(\mathcal{d}))$ ] in ( r1 ) implies that @xmath95 for @xmath96 which is crucial for our proofs as they rely on uniform @xmath16-bounds for the discrete solutions and there is no hope for such thing if the continuous solution is unbounded in @xmath97 .
the third assumption ( r3 ) will be used to obtain optimal convergence rates for the time - discretization in the case of smooth potentials .
we can not expect it to hold in the case of rough disorder potentials @xmath98 .
still , it is possible to show that the assumptions ( r1 ) and ( r2 ) do not conflict with disorder potentials .
we discuss this aspect in more detail in appendix [ appendix - b ] .
note that even though we can not guarantee uniqueness in general , we have that every smooth solution that satisfies ( r1 ) must be unique .
any two solutions of the nls that fulfill ( r1 ) must be identical .
let @xmath99 for @xmath100 and let @xmath101 and @xmath102 denote two smooth solutions in the sense that @xmath103,h^2(\mathcal{d}))$ ] . by sobolev embedding
we can define @xmath104 \times \mathcal{d } ) } < \infty$ ] .
with @xmath105 we obtain for @xmath106 @xmath107 time integration and @xmath108 then yield @xmath109 hence , grnwall s inequality can be applied and shows @xmath110 for all @xmath111 .
the first main result of this paper states that , under the assumption of sufficient regularity , the crank - nicolson scheme admits a solution that remains uniformly bounded in @xmath16 and we obtain optimal convergence rates for the @xmath0-error , independent of the coupling between the mesh size @xmath13 and the time - step size @xmath12 . [ main - theorem-2-a ] under the regularity assumption ( r1 ) , ( r2 ) and ( r3 ) , there positive constants @xmath112 and @xmath113 such that for all partitions with parameters @xmath114 and @xmath115 there exists a unique solution @xmath69 to the fully discrete crank - nicolson scheme with @xmath116 where @xmath117 and @xmath118 .
moreover , the a priori error estimate @xmath119 holds with some constant @xmath120 that may depend on @xmath88 , @xmath121 , @xmath20 , @xmath122 and the constants appearing in - but not on the mesh parameters @xmath12 and @xmath13 .
the uniqueness of fully discrete approximations in theorem [ main - theorem-2-a ] is to be understood in the sense that any other family of approximations must necessarily diverge in @xmath16 as @xmath123 .
the second main result applies to the case of rough potentials .
[ main - theorem-2-b ] assume only ( r1 ) and ( r2 )
. then there exists @xmath113 such that for all partitions with paramters @xmath114 and @xmath124 for some @xmath125 there exists a unique solution @xmath69 to the fully - discrete crank - nicolson scheme such that @xmath116 with @xmath126 as defined in theorem [ main - theorem-2-a ] , and the a priori error estimate @xmath127 holds for some constant @xmath128 independent of @xmath13 and @xmath12 .
sections [ s : errorsemi][s : errorfull ] below are devoted to the proof of theorems [ main - theorem-2-a ] and [ main - theorem-2-b ] .
the results of theorem [ main - theorem-2-b ] are valid under the constraint @xmath124 for some @xmath125 .
this means that the mesh size needs to be small enough compared the time step size .
observe that this is a rather natural assumption if the potential @xmath20 is indeed a rough potential ( as addressed in the theorem ) .
in such a case we wish use a fine spatial mesh to resolve the variations of @xmath20 , whereas the time step size is comparably large . hence
, the constraint is not critical .
conversely , the constraints appearing in works by sanz - serna @xcite and akrivis et al .
@xcite are of a completely different nature , as they require the time step size to be small compared to the mesh size .
therefore , using a fine spatial mesh to resolve the structure of @xmath20 would impose small time steps as well .
in this section we shall consider a semi - discrete crank - nicolson approximation given as follows .
[ semi - discrete - crank - nic - gpe ] let @xmath66 be the initial value from problem and let @xmath129 .
then for @xmath68 , we define the semi - discrete crank - nicolson approximation @xmath130 as the solution to @xmath131 for all @xmath132 and where @xmath133 .
we want to prove that the above problem is well - posed and we want to estimate the @xmath9- and @xmath65-error between @xmath134 and the exact solution @xmath88 .
this requires some auxiliary results that allow us to control the error arising from the nonlinearity .
we start with introducing a truncated version of the ( possibly ) nonlinear function @xmath121 . with this truncated function
, we introduce an auxiliary problem that is central for our analysis .
[ truncation - lemma ] let @xmath135 be a constant with @xmath136 .
then there exists a smooth function @xmath137 and generic constants @xmath120 such that for all @xmath138 and all @xmath139 @xmath140 furthermore , for the antiderivative @xmath141 it holds for all @xmath142 with @xmath143 : @xmath144 before we can prove lemma [ truncation - lemma ] we need to introduce an inequality that we will frequently use in the rest of the paper .
[ lemma01ap]let @xmath145 be a three times continuously differentiable function with locally bounded derivatives .
then , for every @xmath146 with ( w.l.g . ) @xmath147 it holds @xmath148 let us define @xmath149 .
first , we observe that @xmath150 hence @xmath151 with that and using taylor expansion for suitable @xmath152 , @xmath153 with @xmath154 we observe @xmath155 since for @xmath156 @xmath157 we obtain @xmath158 in the following , we let @xmath159 denote a generic constant .
let us define @xmath160 and let @xmath161 be a curve that fulfills @xmath162 for @xmath163 and @xmath164 for @xmath165 . by polynomial interpolation we can chose @xmath166 in such a way that it is a polynomial on the interval @xmath167 $ ] and such that globally @xmath168 .
this proves .
since we have for @xmath165 that @xmath164 and @xmath169 for @xmath170 , we conclude that there exists a constant @xmath159 that only depends on the polynomial degree chosen for the interpolation such that is fulfilled .
next , we want to prove ( [ f_m_cond_3 ] ) and recall that @xmath143 .
let for simplicity @xmath171 we write the left - hand side of ( [ f_m_cond_3 ] ) as @xmath172 since @xmath173 is bounded we immediately have that @xmath174 hence , it remains to estimate the term @xmath175 which we split into three parts .
@xmath176 we start with estimating term i. we use lemma [ lemma01ap ] with the boundedness of @xmath177 and @xmath178 to obtain @xmath179 for the second term we distinguish two cases . case ii.1 : if @xmath180 we get @xmath181 and hence with the bound for @xmath182 @xmath183 case ii.2 : if @xmath184 we get with the lipschitz - continuity of @xmath182 that @xmath185 for term iii , we distinguish again two cases . case iii.1 :
if @xmath186 we see analogously to term case ii.1 that @xmath187 case iii.2 : if @xmath188 we can use lemma [ lemma01ap ] to obtain @xmath189 with the bounds for @xmath190 we get @xmath191 this shows for case iii.2 that @xmath192 combining the estimates proves ( [ f_m_cond_3 ] ) . with the function @xmath166 introduced in lemma [ truncation - lemma ] the _ truncated _ semi - discrete crank - nicolson approximation given as follows .
[ truncated - semi - discrete - crank - nic - gpe ] let @xmath193 be given as in lemma [ truncation - lemma ] and let @xmath194 .
then for @xmath68 , we define the truncated semi - discrete crank - nicolson approximation @xmath195 as the solution to @xmath196 for all @xmath132 and where @xmath197 .
since @xmath126 was chosen such that @xmath198 we have the identity @xmath199 for all @xmath200 . in order to investigate the properties of solutions to , we first need to show that there exists at least one solution . in order to show this
, we recall the following conclusion from brouwers fixed point theorem .
[ brouwer - lemma ] let @xmath201 and let @xmath202 denote the closed unit disk in @xmath203 .
then every continuous function @xmath204 with @xmath205 for all @xmath206 has a zero in @xmath207 , i.e. a point @xmath208 with @xmath209 .
second , we will also make use of following lemma that is a special case of the vitali convergence theorem ( cf .
* theorem b.101 ) ) [ vitali - convergence - theorem ] recall that @xmath210 is bounded .
a sequence @xmath211 converges strongly to @xmath212 if and only if 1 .
@xmath213 converges to @xmath214 locally in measure and 2 .
@xmath213 is @xmath215-equi - integrable , i.e. for every @xmath216 there exists a @xmath217 such that @xmath218 for all measurable subsets @xmath219 with measure @xmath220 .
this allows us to conclude that there exists at least one solution to problem .
[ existence - of - semi - disc - trunc - solutions ] for every @xmath81 there exists at least one solution @xmath195 to the truncated problem .
the proof is established in two steps .
first , we show existence in finite dimensional subspaces , then we pass to the limit to establish existence for the infinite dimensional problem . for this purpose ,
let @xmath221 denote a countable basis of @xmath52 .
we define the finite - dimensional subspaces @xmath222 _ step 1 - existence in @xmath223 .
_ let @xmath224 denote the euclidean inner product on @xmath225 and let @xmath226 denote a solution to . for @xmath68
we look for @xmath227 with @xmath228 for all @xmath229 and where @xmath230 .
we assume that @xmath231 exists and want to show existence of @xmath227 . in order to apply lemma [ brouwer - lemma ]
we define @xmath232 for @xmath233 by @xmath234 where @xmath235 is defined by @xmath236 to verify existence of @xmath237 with @xmath209 , we need to show that there exists @xmath238 such that @xmath205 for all @xmath239 with @xmath240 . we define @xmath241 and see that @xmath242 for some @xmath243-independent positive constants @xmath244 and @xmath245
. exploiting the equivalence of norms in finite dimensional hilbert spaces we conclude the existence of ( new ) @xmath243-independent positive constants such that @xmath246 .
hence , for all sufficiently large @xmath247 we have @xmath205 and therefore with lemma [ brouwer - lemma ] the existence of solutions to , provided that @xmath248 exists .
_ step 2 - existence in @xmath52 .
_ we proceed inductively to show the existence of @xmath226 ( the case @xmath249 with @xmath250 is trivially fulfilled ) .
assume hence that @xmath231 exists .
then we can apply _ step 1 _ to conclude that there exists @xmath227 which is a solution to the finite dimensional problem .
it is easy to verify that problem is energy conserving and hence @xmath251 hence , for fixed @xmath44 , the corresponding sequence of discrete solutions @xmath252 is a bounded sequence in @xmath52 with @xmath253 .
consequently there exists a subsequence of @xmath254 ( for simplicity again denoted by @xmath254 ) and a function @xmath255 such that @xmath256 here we used the rellich embedding theorem . for arbitrary @xmath132
we see that @xmath257 by the weak convergence in @xmath52 .
it remains to investigate the term @xmath258 we want to apply vitali s theorem ( lemma [ vitali - convergence - theorem ] ) to conclude that @xmath259 strongly in @xmath260 .
for that purpose , we need to verify convergence in measure and @xmath215-equi - integrability . to verify the first property
, we exploit that @xmath261 strongly in @xmath260 . using the tschebyscheff inequality
we see that @xmath262 also converges to @xmath263 in measure .
this implies in particular that from every subsequence of @xmath264 we can extract another subsequence such that @xmath262 converges to @xmath263 almost everywhere .
on the other hand , by the continuity of @xmath265 , the convergence almost everywhere is preserved when @xmath265 is applied to the sequence .
consequently , for every subsequence of @xmath266 one can extract another subsequence that converges a.e .
to @xmath267 .
this is equivalent to the property that @xmath266 converges locally in measure to @xmath268 ( since @xmath122 is bounded ) .
hence , we have the first requirement for lemma [ vitali - convergence - theorem ] .
for the second requirement we first observe that @xmath269 ( with @xmath270 and @xmath271 ) .
hence , @xmath272 is @xmath215-equi - integrable if @xmath273 is @xmath215-equi - integrable , which however follows immediately again from the strong convergence @xmath274 in @xmath260 and lemma [ vitali - convergence - theorem ] ( which works in both directions ) . in conclusion
, vitali s convergence theorem applies and yields @xmath259 strongly in @xmath260 for @xmath275 . in conclusion
we have that there exists a subsequence of @xmath276 ( still denoted by @xmath277 ) and @xmath255 such that @xmath278 with this we can pass to the limit in to obtain that @xmath255 with @xmath279 solves @xmath280 consequently , we showed iteratively the existence of a solution @xmath195 to
. goal of this section is to show that if @xmath281 , then there exists @xmath113 such that @xmath282 whenever @xmath283 .
the key to deriving such an @xmath16-bound is to first establish a uniform bound for the error between the laplacian of the exact solution @xmath88 and the laplacian of a truncated approximation @xmath284 , i.e. for @xmath285 .
we start with deriving corresponding error identities . for that purpose
we define the continuous function @xmath286 for @xmath287 by @xmath288 [ lemma - error - ids ] let @xmath195 denote a solution to and let @xmath289 . for @xmath290 , @xmath81
we define the error @xmath291 . with @xmath292
it holds @xmath293 and @xmath294 for simplicity let @xmath295 for @xmath296 .
from we have @xmath297 subtracting the term @xmath298 on both sides gives us @xmath299 testing with @xmath300 and only using the real part of the equation proves the @xmath9-norm identity .
testing with @xmath301 ( note that @xmath302 is admissible here ) and taking the imaginary part proves the error identity for @xmath303 i.e. equation .
[ lemma - bound - laplacian - hatem ] let @xmath195 denote a solution to and recall the definition of @xmath304 from lemma [ truncation - lemma ] . for @xmath305
there exists a generic constant @xmath306 ( independent of the problem and the discretization ) such that if @xmath307 , if @xmath90 , @xmath91 and @xmath308 for @xmath309 or @xmath310 , then it holds @xmath311 for some constant depending on @xmath88 , @xmath121 , @xmath20 , @xmath312 and @xmath126 .
furthermore , for @xmath309 it still holds @xmath313 let either @xmath309 or @xmath310 .
we estimate @xmath314 for the first term we obtain with the trapezoidal rule for fixed @xmath315 that @xmath316 hence for @xmath317 @xmath318 in order to estimate the second term we split the error into several contributions .
@xmath319 for @xmath320 we can apply lemma [ lemma01ap ] to obtain @xmath321 for @xmath322 we denote @xmath323 and we let @xmath324 \rightarrow [ u^{n-1},u^{n}]$ ] denote the complex valued ( linear ) curve given by @xmath325 for @xmath326 $ ] .
we have @xmath327 . with that , we get with the trapezoidal - rule and the midpoint rule that @xmath328 for term @xmath329 we can directly apply the trapezoidal rule again to obtain @xmath330 combining the estimates for @xmath320 , @xmath322 and @xmath329 and applying the young - inequality yields @xmath331 where we used @xmath332 .
it remains to estimate term @xmath333 for which we can use @xmath334 to see @xmath335 combining the estimates for i , ii and iii yields @xmath336 next , we need to show that @xmath337 is an @xmath338-term . for that reason ,
we start from the identity this time , where we observe that we have just the desired @xmath339 more in our terms . proceeding analogously as before yields @xmath340 assuming that @xmath341 is small enough so that @xmath342 we can divide by @xmath343 on both sides of . applying the arising inequality iteratively and exploiting that @xmath344 yields @xmath345 now we can plug this estimate into to get @xmath346 observe that this step exploits the quasi - uniformity of the time - discretization .
applying this inequality iteratively gives us @xmath347 this proves the estimates in the lemma .
it remains to verify @xmath348 for @xmath309 .
checking the previous estimates carefully , we see that the problematic terms @xmath349 ( which prevent the right hand side of to converge to zero ) can be replaced by @xmath350 , where @xmath351 is the error introduced by the trapezoidal rule applied to the individual integrals @xmath352 ( and summed up ) .
it can be shown that the error @xmath350 converges to zero without additional regularity assumption , cf .
* theorem 1.13 ) .
this finishes the proof of the lemma .
it holds @xmath353 hence , @xmath354 can be bounded if @xmath90 and @xmath91 .
note that the latter one implies @xmath355 .
[ lemma - l - infty - bounds - for - semidiscrete - trunc - approx ] consider the setting of lemma [ lemma - bound - laplacian - hatem ] .
there exist constants @xmath356 that only depends on @xmath122 and @xmath113 such that any solution @xmath226 to fulfills the bound @xmath357 for all partitions with @xmath283 . since @xmath122 is convex , @xmath96 and since the potential @xmath98 is non - negative , we have from elliptic regularity theory that for any @xmath358 with @xmath359 the solution @xmath360 is continuous on @xmath361 and it holds ( cf .
@xcite ) @xmath362 where @xmath363 only depends on @xmath122 , @xmath20 and @xmath364 . since @xmath365 we can apply together with lemma [ lemma - bound - laplacian - hatem ] to conclude that for every @xmath216 there exists a @xmath366 such that for all @xmath367 it holds @xmath368 this implies @xmath369 the choice @xmath370 proves the lemma . we are now prepared to proceed with the analysis of the original ( non - truncated ) semi - discrete scheme .
[ main - theorem-1 ] let @xmath90 , @xmath91 and @xmath308 for @xmath309 or @xmath310 .
there is a real number @xmath113 such that for all partitions with @xmath283 there exists a unique solution @xmath371 to the semi - discrete crank - nicolson scheme with @xmath372 where @xmath373 .
moreover , the uniform @xmath374 bound @xmath375 and the error estimate @xmath376 hold true .
any other family of semi - discrete solutions must necessarily blow up in the sense that @xmath377 as @xmath378 . from lemmas
[ existence - of - semi - disc - trunc - solutions ] and [ lemma - l - infty - bounds - for - semidiscrete - trunc - approx ] we immediately have the existence of @xmath379 and the uniform @xmath16-bound .
the uniform @xmath374-bound follows from @xmath380 , where @xmath381 for @xmath378 with lemma [ lemma - bound - laplacian - hatem ] .
the error estimates also follow directly from lemma [ lemma - bound - laplacian - hatem ] via .
it remains to show the uniqueness of @xmath382 .
let therefore @xmath383 denote two solutions to the scheme for @xmath81 with @xmath384 and the same starting value , i.e. with @xmath385 and @xmath386 . by exploiting for @xmath387 and @xmath388 and by testing with @xmath389
we obtain @xmath390 which is a contradiction for @xmath391 .
hence , we have uniqueness under the condition @xmath392 . observe that if there exists a another solution @xmath393 with @xmath394 , then the arguments remain the same and we conclude again uniqueness , which however then contradicts @xmath395 . from this
we see that the only other solutions to are the ones with a diverging @xmath16-norm and which hence can not approximate the smooth exact solution @xmath88 .
in this section we shall analyze the final fully discrete crank - nicolson discretization stated in definition [ crank - nic - gpe ] .
as for the semi - discrete method we need to take a detour over an auxiliary problem .
for that reason we consider the following scheme with truncated nonlinearity .
[ truncated - fully - discrete - crank - nic - gpe ] let @xmath396 and let @xmath12 be small enough so that the results of theorem [ main - theorem-1 ] are valid . for @xmath193 given as in lemma [ truncation - lemma ] , for @xmath397 and for @xmath68
, we define the truncated fully - discrete crank - nicolson approximation @xmath398 as the solution to @xmath399 for all @xmath400 and where @xmath401 .
again , we have existence of solutions .
[ existence - of - fully - disc - trunc - solutions ] for every @xmath81 there exists at least one solution @xmath398 to the truncated problem .
the proof is covered by the proof of lemma [ existence - of - semi - disc - trunc - solutions ] . before deriving a @xmath9-error estimate , we need one last auxiliary lemma . [ lemma - for - est - fully - disc ] for all @xmath402 with @xmath403 for @xmath404 , it holds @xmath405 we assume without loss of generality that @xmath406 . for fixed @xmath407 and @xmath106
we investigate the function @xmath408 we easily observe that @xmath409 is lipschitz continuous because we have for some @xmath410 $ ] @xmath411 now we investigate @xmath412 where we distinguish three cases .
+ case 1 : @xmath413 and @xmath414 .
we obtain with the lipschitz continuity of @xmath409 @xmath415 and @xmath416 ( and @xmath417 , otherwise everything is trivial ) . without loss of generality
let @xmath418 we obtain @xmath419 where we used that @xmath420 ; @xmath421 and that @xmath422 for @xmath423 .
+ case 3 : @xmath424 and @xmath414 we can use the results from case 1 and case 2 with the intermediate value @xmath425 to obtain @xmath426 [ lemma - l2-hatuhm - hatu ] suppose @xmath90 , @xmath91 and @xmath427 and for @xmath309 or @xmath310 .
let @xmath398 denote a solution of the fully - discrete crank - nicolson method with truncation as stated in definition [ truncated - fully - discrete - crank - nic - gpe ] and let @xmath12 be small enough for the results of theorem [ main - theorem-1 ] to hold . if @xmath371 denotes the unique solution to with the properties stated in theorem [ main - theorem-1 ] , then @xmath428 hoolds with an @xmath13-independent constant @xmath429 .
first , observe that the assumptions imply @xmath379 ( and that it is unique ) .
in the following , we denote by @xmath430 any generic constant that depends on @xmath126 , @xmath88 , @xmath121 , @xmath20 and @xmath122 .
recall the definition of the continuous function @xmath286 from and let again @xmath295 .
consider @xmath431 .
from we have @xmath432 and from @xmath433 subtracting the terms from each other and defining @xmath434 gives us @xmath435 testing with @xmath436 and taking the real part yields @xmath437 for the first term we have with theorem [ main - theorem-1 ] @xmath438 the second term can be estimates as @xmath439 where we used again theorem [ main - theorem-1 ] . for the third term we can proceed analogously since @xmath440 is bounded .
we obtain straightforwardly ( again with theorem [ main - theorem-1 ] ) that @xmath441 to bound term iv , we use lemma [ truncated - fully - discrete - crank - nic - gpe ] to estimate @xmath442 consequently , using that @xmath443 is uniformly bounded ( theorem [ main - theorem-1 ] ) we can conclude @xmath444 collecting the estimates for i , ii , iii and iv implies that @xmath445 using the inequality @xmath446 which holds for any @xmath447 with @xmath448 finishes the @xmath9-error estimate .
we can now conclude from lemma [ lemma - l2-hatuhm - hatu ] that @xmath449 remains uniformly bounded in @xmath16 which allows us to conclude @xmath450 for appropriately chosen @xmath126 . in summary we obtain theorem [ main - theorem-2-a ] . the detailed proof is given in the following .
we choose @xmath117 .
let @xmath451 and let @xmath430 denote any constant depending additionally on @xmath126 ( however , both are not allowed to depend on @xmath13 or @xmath12 ) . using the assumptions on @xmath57 , the bounds from theorem [ main - theorem-1 ] and lemma [ lemma - l2-hatuhm - hatu ] we have @xmath452 since @xmath18 and @xmath124 for @xmath309 and some @xmath125 , we conclude that there exists @xmath112 such that @xmath453 for all @xmath115 .
hence , for sufficiently small @xmath13 we have @xmath450 .
we conclude the existence of @xmath454 and the @xmath13-independent bound @xmath455 for the @xmath9-error estimate we split the error into @xmath456 the first term can be estimated with theorem [ main - theorem-1 ] for sufficiently small @xmath12 , the second term is bounded by @xmath457 ( again using theorem [ main - theorem-1 ] ) and the last term can be estimated with lemma [ lemma - l2-hatuhm - hatu ] which now holds with @xmath450 . in the setting of theorem [ main - theorem-2-a ] , this yields @xmath458 for all sufficiently small @xmath13 and @xmath12 . in the setting of theorem [ main - theorem-2-b ] ,
the order is reduced to @xmath459 .
the proof of uniqueness under some uniform bound @xmath460 independent of @xmath12 and @xmath13 is almost verbatim the same as in the semi - discrete case ( see the proof of theorem [ main - theorem-1 ] ) .
we shall conclude with some simple and illustrative numerical experiment .
the computational domain is given by @xmath461 ^ 2 $ ] and the time interval by @xmath42:=[0,1]$ ] .
we wish to approximate @xmath462 \rightarrow { \mathbb{c}}$ ] with @xmath463 and @xmath464 , \\ u & = 0 \hspace{121pt } \mbox{on } \partial \mathcal{d } \times [ 0,t].\end{aligned}\ ] ] here we have @xmath465 and and we consider a rough potential @xmath466 given by @xmath467 where `` @xmath468 '' rounds a real number @xmath469 to the largest integer smaller or equal to @xmath469 .
the potential is visualized in figure [ potential ] . as considered in the model problem . _
_ ] inspired by the discussion in appendix [ appendix - b ] we pick the initial value as the ground state of a perturbed nls . more precisely , we choose @xmath470 with @xmath471 such that @xmath472 with @xmath473 and a smooth potential perturbation @xmath474
. there exists a unique ground state @xmath475 with the above properties ( cf .
@xcite ) and it holds @xmath92 . given a finite element space @xmath476 , the discrete approximation of @xmath475 in @xmath476 is given by some @xmath477 with @xmath478 and @xmath479 i.e. , @xmath480 is an energy minimizer in @xmath476 .
such a minimizer exists and it holds @xmath481 ( independent of the smoothness of @xmath20 ) .
this means that using @xmath480 as a discrete initial value in our scheme will introduce an error that is of the same order as if using @xmath482 .
we compute the discrete minimizers @xmath480 by using the discrete normalized gradient flow method proposed in @xcite .
for @xmath483 we use a lagrange finite element space of polynomial order @xmath484 , based on a uniform ( simplicial ) triangulation of @xmath122 .
the mesh size @xmath13 is given as the diameter of the elements of the triangulation .
for @xmath485 the discrete ground state @xmath486 is depicted in figure [ reference - values ] . in the following ,
all errors are with respect to a reference solution @xmath487 computed with the crank - nicolson scheme with @xmath485 and with equidistant time steps of size @xmath488 .
the reference solution @xmath489 at @xmath490 is depicted in the right picture of figure [ reference - values ] .
obtained for @xmath491 .
right : reference solution @xmath489 at @xmath490 obtained with for @xmath491 and @xmath488.__,title="fig : " ] obtained for @xmath491 .
right : reference solution @xmath489 at @xmath490 obtained with for @xmath491 and @xmath488.__,title="fig : " ] in the following with present the discrete approximations @xmath492 obtained with in @xmath51 and with equidistant time steps . recall that the discrete initial value is given by , i.e. @xmath493 . before discussing the error evolution ,
we introduce some short - hand notation .
the rescaled mesh and time step sizes are given by @xmath494 the error compared to the reference solution a time @xmath495 is denoted by @xmath496 and the corresponding relative errors ( for real and imaginary parts ) are given by @xmath497 and analogously for the error in the gradient .
.__the table shows @xmath0- and @xmath498-errors between the reference solution @xmath489 and various crank - nicolson approximations @xmath492 obtained with for different mesh sizes and time step sizes . _ _ [ cols="^,^,^,^,^,^,^,^",options="header " , ] in order to study the accuracy of the crank - nicolson finite element method stated in , we run various computations with different constellations for the size of the mesh size @xmath13 and the time step size @xmath12 . as there is no known exact solution to our model problem
, we use the fine scale approximation @xmath489 as our reference for the computation of errors . an overview on all the computations and the arising errors
is given in table [ table - lerrors ] .
from that we can make several observations .
first , we observe a clearly convergent behavior in terms of the mesh size and time step size and we did not encounter any numerical issues ( on the solver level ) when computing the approximations @xmath492 .
we can also report that the scheme preserved the mass and the energy almost up to machine precision .
second , the computed experimental orders of convergence do not correlate with the pessimistic linear rates predicted by theorem [ main - theorem-2-b ] .
more precisely , we rather observe the quadratic rates expected under the stronger regularity assumptions of theorem [ main - theorem-2-a ] .
this is emphasized by table [ table - eocs - coupled ] , where we depict the results for the case that @xmath13 and @xmath12 are refined simultaneously . in the light of these results
, the performance of the method appears better than predicted .
this might indicate that the regularity assumption ( r3 ) is feasible also for some class of discontinuous potentials .
however , an empirical proof or disproof of this claim requires further systematic numerical studies beyond the scope of this paper that also excludes the possibility of super - convergence effects when estimating the error using some reference solution .
@xmath41 * conclusion . * in this paper we analyzed a mass- and energy conserving crank - nicolson galerkin method .
we showed that it is numerically stable under perturbations , that the scheme is well - posed in some ball ( in @xmath499)$ ] ) around zero and we derived @xmath0-error estimates under various regularity assumptions .
all our estimates are valid for general disorder potentials in @xmath500 .
however , it is not clear how or if our regularity assumptions might conflict with discontinuities in the potential . therefore we derived two graded results . in the first main result , we assume sufficient regularity of the exact solution @xmath88 and derive error estimates of optimal ( quadratic ) order in @xmath13 and @xmath12 . the novelty with respect to previous works is that our results cover a general class of nonlinearities , potential terms and we show that the method does indeed not require a time step constraint . on the contrary ,
the results in @xcite are only valid , provided that the time step size is sufficiently small with respect to the spatial mesh size . in our second main result , we relax the regularity assumptions so that they appear not to be in conflict with discontinuous potentials . under these relaxed regularity assumptions
, we can still derive @xmath0-error estimates , however , only of linear order .
furthermore , we encounter a time step constraint that was absent in the case of higher regularity .
to check the practical performance of the method , we present a numerical experiment for a model problem with discontinuous potential .
the corresponding numerical errors seem not to correlate with the pessimistic rates predicted for the low - regularity regime .
we could neither observe degenerate convergence rates nor a practical time step constraint .
instead , we observe the behavior as predicted for the high regularity regime , i.e. , convergence rates of optimal order and good approximations in all resolution regimes , independent of a coupling between mesh size and time step size .
10 g. d. akrivis , v. a. dougalis , and o. a. karakashian . on fully discrete galerkin methods of second - order temporal accuracy for the nonlinear schrdinger equation .
, 59(1):3153 , 1991 .
x. antoine , w. bao , and c. besse .
computational methods for the dynamics of the nonlinear schrdinger / gross - pitaevskii equations .
, 184(12):26212633 , 2013 .
x. antoine and r. duboscq .
robust and efficient preconditioned krylov spectral solvers for computing the ground states of fast rotating and strongly interacting bose - einstein condensates . , 258:509523 , 2014 .
w. bao and y. cai . mathematical theory and numerical methods for bose - einstein condensation . , 6(1):1135 , 2013 .
w. bao and q. du . computing the ground state solution of bose - einstein condensates by a normalized gradient flow .
, 25(5):16741697 , 2004 .
w. bao and w. tang .
ground - state solution of bose - einstein condensate by directly minimizing the energy functional .
, 187(1):230254 , 2003 .
e. cancs , r. chakir , and y. maday . numerical analysis of nonlinear eigenvalue problems . , 45(1 - 3):90117 , 2010 .
t. cazenave .
, volume 10 of _ courant lecture notes in mathematics_. new york university , courant institute of mathematical sciences , new york ; american mathematical society , providence , ri , 2003 .
d. cruz - uribe and c. j. neugebauer .
sharp error bounds for the trapezoidal rule and simpson s rule . , 3(4):article 49 , 22 , 2002 .
i. danaila and f. hecht . a finite element method with mesh adaptivity for computing vortex states in fast - rotating bose - einstein condensates .
, 229(19):69466960 , 2010 .
i. danaila and p. kazemi . a new sobolev gradient method for direct minimization of the gross - pitaevskii energy with rotation .
, 32(5):24472467 , 2010 .
l. gauckler .
convergence of a split - step hermite method for the gross - pitaevskii equation .
, 31(2):396415 , 2011 .
d. gilbarg and n. s. trudinger . .
classics in mathematics .
springer - verlag , berlin , 2001 .
reprint of the 1998 edition .
e. p. gross .
structure of a quantized vortex in boson systems .
, 20:454477 , 1961 .
p. henning and a. mlqvist .
the finite element method for the time - dependent gross - pitaevskii equation with angular momentum rotation .
arxiv e - print 1502.05025 ( submitted ) , 2015 .
p. henning , a. mlqvist , and d. peterseim .
two - level discretization techniques for ground state computations of bose - einstein condensates . , 52(4):15251550 , 2014 .
e. jarlebring , s. kvaal , and w. michiels .
an inverse iteration method for eigenvalue problems with eigenvector nonlinearities .
, 36(4):a1978a2001 , 2014 .
o. karakashian and c. makridakis .
a space - time finite element method for the nonlinear schrdinger equation : the discontinuous galerkin method .
, 67(222):479499 , 1998 .
o. karakashian and c. makridakis .
a space - time finite element method for the nonlinear schrdinger equation : the continuous galerkin method .
, 36(6):17791807 , 1999 .
g. leoni .
, volume 105 of _ graduate studies in mathematics_. american mathematical society , providence , ri , 2009 .
e. h. lieb , r. seiringer , and j. yngvason
. a rigorous derivation of the gross - pitaevskii energy functional for a two - dimensional bose gas .
, 224(1):1731 , 2001 .
dedicated to joel l. lebowitz . c. lubich . on splitting methods for schrdinger - poisson and cubic nonlinear schrdinger equations .
, 77(264):21412153 , 2008 .
b. nikolic , a. balaz , and a. pelster .
dipolar bose - einstein condensates in weak anisotropic disorder . , 88(1 ) , 2013 .
l. p. pitaevskii . .
number 13 .
soviet physics jetp - ussr , 1961 .
j. m. sanz - serna .
methods for the numerical solution of the nonlinear schroedinger equation .
, 43(167):2127 , 1984 .
m. thalhammer .
convergence analysis of high - order time - splitting pseudospectral methods for nonlinear schrdinger equations .
, 50(6):32313258 , 2012 .
m. thalhammer and j. abhau .
a numerical study of adaptive space and time discretisations for gross - pitaevskii equations .
, 231(20):66656681 , 2012 .
y. tourigny .
optimal @xmath65 estimates for two time - discrete galerkin approximations of a nonlinear schrdinger equation . , 11(4):509523 , 1991 .
j. wang .
a new error analysis of crank - nicolson galerkin fems for a generalized nonlinear schrdinger equation .
, 60(2):390407 , 2014 .
j. williams , r. walser , c. wieman , j. cooper , and m. holland . achieving steady - state bose - einstein condensation .
, 57(3):20302036 , 1998 .
i. zapata , f. sols , and a. j. leggett .
josephson effect between trapped bose - einstein condensates .
, 57(1):r28r31 , 1998 . g. e. zouraris . on the convergence of a linear two - step finite element method for the nonlinear schrdinger equation .
, 35(3):389405 , 2001 .
in this section we demonstrate that a rough potential does not exclude the regularity assumptions ( r1 ) and ( r2 ) , however their compatibility will in general rely on the choice for the initial value @xmath501 . for simplicity of the presentation
we consider the case @xmath502 ( i.e. a linear problem ) .
the nonlinear case is briefly discussed at the end of this section .
let @xmath503 denote a rough disorder potential and let @xmath504 denote a ground state or excited state to the stationary schrdinger equation @xmath505 where @xmath506 is the corresponding eigenvalue ( the chemical potential ) and @xmath501 is @xmath9-normalized , i.e. @xmath507 . from elliptic regularity theory we know that the solution to problem admits higher regularity , i.e. @xmath508 ( cf .
@xcite ) . however , since @xmath509 is rough , we can not expect any regularity beyond @xmath374 . in order to investigate the dynamics of @xmath501
, the potential trap is reconfigured . in our case
this means that we set @xmath510 , where @xmath511 is a non - negative smooth perturbation , say ( for simplicity ) @xmath512 . with this
we seek @xmath513 \rightarrow h^1_0(\mathcal{d})$ ] with @xmath514 and @xmath515 let us now assume that @xmath88 denotes a solution to that is sufficiently regular .
then , from equation we conclude that @xmath516 for @xmath517 . taking only the imaginary part of the equation
yields @xmath518 by integrating from @xmath519 to @xmath520 , we have @xmath521 this means that we have to verify that the compatibility `` @xmath522 '' is well - defined for rough potentials .
for @xmath523 we exploit the initial condition and obtain @xmath524 hence @xmath525 analogously , we obtain for @xmath526 that @xmath527 hence @xmath528 furthermore , since @xmath529 where @xmath530 ( for which we just derived corresponding bounds depending on @xmath501 ) , we can also conclude by elliptic regularity theory that @xmath531 from the equation @xmath532 we also make an important observation : we have @xmath533 , however we do _ not _ have @xmath374-regularity as this would require @xmath534 , which is clearly not available due to the roughness of @xmath509 .
therefore we can not repeat the same argument for @xmath535 .
observe that for @xmath536 we have @xmath537 which implies that @xmath538 , but it is not in @xmath260 . in order to obtain @xmath539
we would require the disorder potential to be at lest in @xmath65 ( which contradicts the notion of a disorder potential ) .
consequently , we can neither hope for @xmath539 nor for @xmath540 .
the only thing that we can hope for is to verify @xmath541 . and indeed , analogously to the proof of energy conservation we easily observe that @xmath542 with we conclude that the right - hand side is well - defined and bounded for rough potentials @xmath509 .
@xmath41 in the nonlinear case similar arguments can be used .
however , the calculations become significantly more technical since we typically do no longer have the conservation properties such as @xmath544 for @xmath545 . still for small times
it is possible to show an inequality which takes a comparable role .
for instance , in the case of the cubic nonlinearity @xmath6 ( where @xmath546 is a parameter that characterizes the type and the number of particles ) it is possible to show that there exists a minimum time @xmath21 and constant @xmath547 ( independent of the regularity of the potential ) , such that @xmath548 for @xmath549 and provided that @xmath88 is sufficiently smooth . with this
it is possible to proceed in a similar way as in the linear case and we can draw the same conclusions . | this paper analyses the numerical solution of a class of non - linear schrdinger equations by galerkin finite elements in space and a mass- and energy conserving variant of the crank - nicolson method due to sanz - serna in time .
the novel aspects of the analysis are the incorporation of weak and strong disorder potentials , the consideration of some general class of non - linearities , and the proof of convergence with rates in @xmath0 under moderate regularity assumptions that are compatible with discontinuous potentials . for sufficiently smooth potentials ,
the rates are optimal without any coupling condition between the time step size and the spatial mesh width .
crank - nicolson galerkin approximations to nonlinear schrdinger equations with disorder potentials + patrick henning , daniel peterseim + |
both standard @xcite and inflationary cosmology @xcite suffer from the initial singularity problem and hence can not yield complete descriptions of the very early universe .
if one were able to construct a non - singular bouncing cosmology , this problem would obviously disappear .
however , in order to have a chance to obtain such a non - singular cosmology , one must either go beyond einstein gravity as a theory of space - time ( see e.g. @xcite for an early construction ) , or else one must make use of matter which violates the null energy condition " ( see @xcite for a review of both types of approaches ) .
interest in non - singular bouncing cosmologies has increased with the realization that they can lead to alternatives to inflationary cosmology as a theory for the origin of structure in the universe .
a specific scenario which can arise at the level of homogeneous and isotropic cosmology is the matter bounce " paradigm which is based on the realization @xcite that vacuum fluctuations which exit the hubble radius during a matter - dominated contracting phase evolve into a scale - invariant spectrum of curvature perturbations on super - hubble scales before the bounce .
the key point is that the curvature fluctuation variable @xmath0 grows on super - hubble scales in a contracting phase , whereas it is constant on these large scales in an expanding phase . since long wavelength modes exit the hubble radius earlier than short wavelength ones
, they grow for a longer period of time .
this provides a mechanism for reddening the initial vacuum spectrum .
it turns out that a matter dominated contracting phase provides the specific boost in the power of long wavelength modes which is required in order to transform a vacuum spectrum into a scale - invariant one .
studies in the case of various non - singular bounce models @xcite have shown that on wavelengths long compared to the duration of the bounce phase , the spectrum of fluctuations is virtually unchanged during the bounce .
thus , a scale - invariant spectrum of curvature fluctuations survives on super - hubble scales at late times .
provided that the bounce can occur at energy scales much below the planck scale , non - singular cosmologies solve a key conceptual problem from which inflationary cosmology suffers , namely the
trans - planckian " problem for fluctuations @xcite : if the period of inflationary expansion of space lasts for more than @xmath1 , where @xmath2 is the hubble expansion rate during inflation ( in order to solve the key cosmological mysteries it was designed to explain , inflation has to last at least @xmath3 ) , then the physical wavelengths of even the largest - scale fluctuation modes we see today will be even smaller than the planck length at the beginning of inflation and thus in the zone of ignorance " where the physics on which inflation and the theory of cosmological perturbations are based , namely einstein gravity coupled to semiclassical field theory matter , will break down .
in contrast , in a non - singular bouncing cosmology the wavelength of modes which are currently probed by cosmological observations is never much smaller than @xmath4 ( the physical wavelength of the mode which corresponds to our current hubble radius evaluated when the temperature of the universe was @xmath5 ) and hence many orders of magnitude larger than the planck length .
thus , the fluctuations never enter the trans - planckian zone of ignorance " of sub - planck - length wavelengths .
possibly the simplest realization of the matter bounce scenario is the
quintom bounce " model @xcite and is obtained by considering the matter sector to contain two scalar fields , one of them ( the ghost field " ) having the wrong " sign of the kinetic action .
the potential of the ghost scalar field also has the opposite sign to that of regular scalar fields such that in the absence of interactions , the ghost field has a classically stable minimum .
as has been noticed in @xcite , such a quintom bounce model also arises from the scalar field sector of the lee - wick " ( lw ) lagrangian @xcite which contains higher derivatives terms .
the quintom and lee - wick bouncing cosmologies are obtained in the following way @xcite : we begin in the contracting phase with both the regular and the ghost scalar field oscillating homogeneously in space about their respective vacua .
we assume that the energy density is dominated by the regular matter field , and that hence the total energy density is positive .
once the amplitude of the regular scalar field exceeds the planck scale , the field oscillations will freeze out and a slow - climb phase will begin during which the energy density of the field only grows slowly ( this is the time reverse of the slow - roll phase in scalar field - driven inflation )
. however , the ghost field continues to oscillate and its energy density ( which is negative ) continues to grow in absolute value .
hence , the total energy density drops to zero , at which point the bounce occurs , as has been studied both analytically and numerically in the above - mentioned works . note that the energy density in this bounce model scales as matter until the regular scalar field freezes out . a major problem of bouncing cosmologies realized with matter which scales as @xmath6 as a function of the scale factor @xmath7 is the potential instability of the homogeneous and isotropic background against the effects of radiation ( which scales as @xmath8 and anisotropic stress which scales as @xmath9 .
if we simply add a non - interacting radiation component to the two scalar field system , then unless the initial energy density in radiation is tuned to be extremely small , then the radiation component will become dominant long before the bounce can arise , and will prevent the energy density in the ghost field from ever being able to become important , resulting in a big crunch singularity . similarly , unless the initial energy density in anisotropic stress is very small , it will come to dominate the energy density of the universe long before the bounce is expected . the anisotropies will destabilize the homogeneous background cosmology , and will prevent a bounce .
note that at the quantum level , there is an additional severe problem for bounce models obtained with matter fields with ghost - like kinetic terms , namely the quantum instability of the vacuum ( see e.g. @xcite ) .
in this paper we will focus on the radiation instability problem . for the purpose of this discussion we will simply assume that anisotropic stress is absent . in a recent paper ,
two of us studied the possibility that a bounce could arise if radiation is supplemented with lee - wick radiation @xcite .
however , we showed that this hope is not realized : the addition of lee - wick radiation does not prevent the big crunch singularity from occurring . in the presence of radiation ,
the only hope to obtain a bounce is to introduce a coupling between radiation and ghost scalar field matter which could effectively drain energy density from the radiation field and prevent the energy density of radiation from becoming dominant .
here we study this possibility .
however , at least for the specific lagrangian which we consider , we find that a bounce only emerges for highly fine - tuned phases of the fields and their velocities in the initial conditions .
the paper is organized as follows : in section ii , we introduce the model we study , namely the scalar field sector of lee - wick theory coupled to radiation , and write down the general equations of motion . in section iii we set up the equations of motion linearized about the bounce background , treating the entire radiation field as an inhomogeneous fluctuation . in particular , we study the different terms which contribute to the energy - momentum tensor and identify those which could assist in obtaining a non - singular bounce . in section
iv we study the solutions of the perturbed equations of motion , and in section v we analyze the evolution of the different terms in the energy - momentum tensor , identifying the conditions which would be required in order to obtain a non - singular bounce .
we have also evolved the general equations of motion for the two inhomogeneous scalar field configurations and the classical inhomogeneous radiation field in the homogeneous background cosmology .
section vi summarizes some of the numerical results .
both the analytical and numerical results confirm that we need unnatural fine - tuning of the initial conditions in order to obtain a non - singular bounce . in the final section
we offer some conclusions and discussion .
the lee - wick scalar field model coupled to electromagnetic radiation is given by the following lagrangian : [ leewick ] l&=&-_^+(^2 ) ^2-m^2 ^ 2-v ( ) + & & - f _
f^-f(,^2,f_f^ ) , where @xmath10 is the mass of the scalar field @xmath11 , and @xmath12 is its potential . here we adopt the convention that ds^2 = -dt^2 + a^2(t)(dx^2 + dy^2 + dz^2 ) , where @xmath7 is the scale factor of the universe . since it is a higher derivative lagrangian in @xmath11 , the scalar field sector contains an extra degree of freedom with the
wrong " sign kinetic term and with a mass set by the scale @xmath13 .
we choose @xmath14 , where @xmath15 is the planck mass , since we want the regular scalar field to dominate at low energies , but at the same time we do not want to worry about quantum gravity effects .
the second line of the lagrangian ( [ leewick ] ) contains the kinetic term of the radiation as well as the coupling term , where we assumed both for the sake of generality and because of foresight that the radiation field couples not only to the scalar field @xmath11 itself , but also to the higher derivative term .
the electromagnetic tensor , @xmath16 , is related to the radiation field @xmath17 through the usual definition f _ _ a_-_a _ , where @xmath18 is the covariant derivative .
it is convenient to extract the extra degree of freedom as a separate scalar field . to do this
, we use the field redefinitions & & _ 1-_2 , + _ 2 & & ^2 /m^2 .
the lagrangian ( [ leewick ] ) then takes on a simpler form : [ leewick2 ] l & = & -__1^_1+__2^_2-m^2 ^2+m^2 ^2 + & & - f _
f^-f(_1,_2,f_f^ ) , where we have chosen the potential to be zero . in this new form , the lagrangian describes two massive scalar fields with one of them ( i.e. , @xmath19 ) behaving like aghost " , and both of them coupled to the radiation field . the coupling term @xmath20 should in principle be arbitrary , however , in this paper we will take a specific form for convenience .
the form will be : [ f1 ] f(_1,_2,f_f^ ) = ( c^2+d^2)f _ f^ , where @xmath21 and @xmath22 are coupling constants which have mass dimension @xmath23 . the interaction terms are non - renormalizable . to make sure that such terms could be thought of as arising from an effective field theory which is consistent at the bounce
, we must make sure that the coefficients are chosen such that the contribution of the interaction term to the lagrangian density is smaller than that of the other terms .
this must be true even at energy densities at which the bounce occurs in the pure scalar field model .
it is easy to see that this condition will be satisfied if the coefficients @xmath21 and @xmath22 are both of the order @xmath24 .
it is the purpose of this paper to study the effects which these coupling terms have on the dynamics of the system .
we know that in the absence of coupling , i.e. when @xmath25 , a bounce will only occur if the initial radiation energy density is tuned to a very small value compared to the scalar field energy density .
this is because the positive definite energy density of radiation will scale as @xmath8 which is faster than that of the scalar fields , in particular the ghost scalar field .
generically , it will dominate the energy of the universe after some amount of contraction , it will prevent the ghost scalar field energy density from catching up and will thus prevent a bounce , leading to a big crunch singularity instead . with non - vanishing values of @xmath21 and @xmath22 , however , the scalars are in principle able to drain energy from the radiation . from the lagrangian ( [ leewick2 ] ) , one can obtain the stress - energy tensor @xmath26 by varying the action with respect to the metric @xmath27 . in the hydrodynamical limit
, we can take @xmath26 to be of the form of @xmath28 where @xmath29 and @xmath30 are energy density and pressure , respectively .
for now , we consider the general form which is : [ tmunu ] t_&=&g_l+__1__1-__2__2+(1-c_1 ^ 2-d_2 ^ 2)f_f_^ , + & = & g _ + & & + _ _
1__1-__2__2+(1-c_1 ^ 2-d_2 ^ 2)f_f_^ .
since we will be studying the contribution of plane wave perturbations of the scalar fields and we will treat radiation as a superposition of waves , we kept the space - derivative terms . by varying the lagrangian with respect to the matter fields
@xmath31 , @xmath19 and @xmath17 , we also get the equations of motion for all three fields : [ phi1eom]&&_1-(m^2-f^2)_1=0 , + [ phi2eom]&&_2-(m^2+f^2)_2=0 , + [ amueom]&&(1-c_1 ^ 2-d_2 ^ 2)(_f^+3hf^0 ) + & & -2(c_1__1+d_2__2)f^=0 which will be analyzed in detail in the rest of the paper .
since the equations of motion are nonlinear , we can not work in fourier space , and use plane wave solutions .
however , we are interested in how initially small amounts of radiation build up and possibly transfer their energy to scalar field fluctuations .
we treat radiation as a superposition of fluctuations .
therefore it makes sense to linearize our equations about the homogeneous scalar field background .
thus , we make the following ansatz for the scalar fields : [ ansatzphi ] _ 1(t , z)&=&_1^(0)(t)+_1^(1)(t , z)+^2 _
1^(2)(t ) + _ 2(t , z)&=&_2^(0)(t)+_2^(1)(t , z)+^2 _ 2^(2)(t ) , where the expansion parameter @xmath32 is taken to be much smaller than @xmath33 . should be viewed as parameterizing the initial ratio of radiation energy to background scalar field energy .
thus , the leading contribution of the radiation field is first order in @xmath32 . via the coupling terms in the lagrangian with coefficients @xmath21 and @xmath22 , the linear radiation field induces linear scalar field inhomogeneities @xmath34 and @xmath35 .
these corrections will contain a further suppression factor since @xmath21 and @xmath22 are small coefficients .
similarly , the same coupling terms in the lagrangian will lead to a perturbation @xmath36 of the rescaled radiation field @xmath37 which is of linear order in @xmath32 but suppressed by factors of @xmath21 and @xmath22 . ]
the first term on the right hand side of each line , i.e. @xmath38 correspond to the background fields , the terms @xmath39 are the fluctuations , and the second order terms @xmath40 describe the back - reaction of the fluctuations on the background and can be computed from the leading second order corrections ( averaged over space ) of the equations of motion . to simplify the analysis , we describe radiation in terms of plane waves in a fixed direction ( which we take to be the @xmath41 direction ) . without loss of generality we can restrict attention to one polarization mode which we take to be the electric field in the @xmath42 direction and the magnetic field in the @xmath43 direction . in this case , the only non - zero components of the field strength tensor are @xmath44 and @xmath45 . using the temporal gauge where @xmath46 , we find that only the first component of the gauge field is non - zero . for a single wavelength fluctuation we can make the ansatz [ ansatzgauge1 ] a_1(k , t )
= f(t)cos(kz ) ( k , t ) , or , equivalently , [ ansatzgauge2 ] a^1(k , t ) = a(t)^-2 ( k , t ) .
since in the linearized equations of motion the fourier modes are independent , we can consider @xmath34 and @xmath35 also to be plane waves propagating in @xmath41 direction , so they depend only on @xmath41 and @xmath47 . with eqs .
( [ ansatzphi]-[ansatzgauge2 ] ) in hand , we can write down the energy densities of the various fields at each order in perturbation theory .
first of all , we insert the above perturbative ansatz for the fields into the stress - energy tensor of the system . from the general expression ( [ tmunu ] ) for @xmath26
we get : [ tmunu2 ] t_&=&g _ + & & + _ _
1__1-__2__2+(1-c_1 ^ 2-d_2 ^ 2)f_f_^ .
the @xmath48 component of eq .
( [ tmunu2 ] ) denotes the energy density of the system : [ rho ] & = & ( _ 1 ^ 2+_1 ^ 2+m^2_1 ^ 2 ) -(_2 ^
2+_2 ^ 2+m^2_2 ^ 2 ) + & & + ( 1-c_1 ^ 2-d_2 ^ 2)(+f_0f_0^ ) , so at each level in perturbation theory we have : [ rho0]^(0)&=&(^2+m^2_1^(0)^2 ) -(^2+m^2_2^(0)^2 ) , + [ rho1]^(1)&=&(_1^(0)_1^(1)+m^2_1^(0)_1^(1 ) ) -(_2^(0)_2^(1 ) + & & + m^2_2^(0)_2^(1 ) ) , + [ rho2]^(2)&=&(^2+_1^(0)_1^(2 ) + ^2+m^2_1^(1)^2 + & & + m^2_1^(0)_1^(2))-(^2+_2^(0)_2^(2 ) + ^2 + & & + m^2_2^(1)^2+m^2_2^(0)_2^(2 ) ) + & & + ( 1-c_1^(0)^2-d_2^(0)^2)(+f_0f_0^ ) , + & = & ( ^2+_1^(0)_1^(2 ) + ^2+m^2_1^(1)^2 + & & + m^2_1^(0)_1^(2))-(^2+_2^(0)_2^(2 ) + ^2 + & & + m^2_2^(1)^2+m^2_2^(0)_2^(2 ) ) + & & + ( 1-c_1^(0)^2-d_2^(0)^2)(^2 + ) .
we can similarly obtain the pressure of the system from the @xmath49 components of eq .
( [ tmunu2 ] ) .
note that due to the anisotropy in @xmath26 caused by the gauge field as well as by the anisotropic fluctuations of the scalar fields , the pressures in the three directions are no longer identical .
the pressure in each direction can be written as : [ pres ] p_i&=&(_1 ^ 2-_1 ^ 2-m^2_1 ^ 2 ) -(_2 ^ 2-_2 ^ 2-m^2_2
^ 2 ) + & & -(1-c_1 ^ 2-d_2 ^ 2)(-)+ + & & - , with no summation over the index @xmath50 . from this formula
, we can see that at both zero - th and first order , the pressure is isotropic : p_i^(0)&=&(^2-m^2_1^(0)^2 ) -(^2-m^2_2^(0)^2 ) , + p_i^(1)&=&(_1^(0)_1^(1)-m^2_1^(0)_1^(1 ) ) -(_2^(0)_2^(1 ) + & & -m^2_2^(0)_2^(1 ) ) , while the second order pressure for each direction gives p_i^(2)&=&(^2+_1^(0)_1^(2 ) -^2+m^2_1^(1)^2 + & & + m^2_1^(0)_1^(2))-(^2+_2^(0)_2^(2 ) -^2 + & & + m^2_2^(1)^2+m^2_2^(0)_2^(2 ) ) + - + & & -(1-c_1^(0)^2-d_2^(0)^2)(- ) , where @xmath51 .
we can thus obtain every component of @xmath52 : p_1^(2)&=&(^2+_1^(0)_1^(2 ) -^2+m^2_1^(1)^2 + & & + m^2_1^(0)_1^(2))-(^2+_2^(0)_2^(2 ) -^2 + & & + m^2_2^(1)^2+m^2_2^(0)_2^(2 ) ) + & & + ( 1-c_1^(0)^2-d_2^(0)^2)(^2- ) , + p_2^(2)&=&(^2+_1^(0)_1^(2 ) -^2+m^2_1^(1)^2 + & & + m^2_1^(0)_1^(2))-(^2+_2^(0)_2^(2 ) -^2 + & & + m^2_2^(1)^2+m^2_2^(0)_2^(2 ) ) + & & -(1-c_1^(0)^2-d_2^(0)^2)(^2- ) , + p_3^(2)&=&(^2+_1^(0)_1^(2 ) -^2+m^2_1^(1)^2 + & & + m^2_1^(0)_1^(2))-(^2+_2^(0)_2^(2 ) -^2 + & & + m^2_2^(1)^2+m^2_2^(0)_2^(2 ) ) + & & + ( 1-c_1^(0)^2-d_2^(0)^2)(^2 + ) , and the average is : p_eff^(2)&= & + & = & ( ^2+_1^(0)_1^(2 ) -^2+m^2_1^(1)^2 + & & + m^2_1^(0)_1^(2))-(^2+_2^(0)_2^(2 ) -^2 + & & + m^2_2^(1)^2+m^2_2^(0)_2^(2 ) ) + & & + ( 1-c_1^(0)^2-d_2^(0)^2)(^2 + ) . from the above
, we can also see that in order to analyze the behavior of the energy density up to second order , we need to know the evolution of scalar fields up to second order as well as that of the gauge field up to first order , while the behavior of the gauge field to second order is not required .
it will be useful in the following to separate the contributions to the energy density and pressure in a different way , namely , + i ) the contribution from the background homogeneous part of the scalar fields , _
^(0 ) p_^h= p_^(0 ) , ii ) that of the scalar field perturbations ( in slight abuse of notation we call this the `` inhomogeneous '' term ) , _
^inh = _ ^(1)+^2 _ ^(2 ) p_^inh = p_^(1)+^2 p_^(2 ) ,
iii ) the contribution of the gauge field , _ g & = & ( ^2+^2 ) , + p_g & = & ( ^2+^2 ) , and iv ) the contribution of the coupling term , _ c & = & -(^2+^2)=-_g , + p_c&=&-(^2+^2)=-p_g , where in the last equation we define @xmath53 to be the quadratic combination of the two fields : = ( c^(0)^2 + d ^(0)^2)/2 . from the above
we can deduce the equation of state parameter for each part : w_^h&= & , + w_^inh&= & ( - ) + & & + ^2(-- ) , + w_g&=&w_c= . from the equations above , we see that for positive values of the constants @xmath21 and @xmath22 , the coupling of the scalar field with the gauge field will give rise to a contribution @xmath54 to the energy density which has the same equation of state but opposite sign to that of the gauge field .
therefore the coupling can help drain energy from the gauge field .
it is because of this mechanism that we might hope to achieve a cosmological bounce in the presence of radiation . a first indication on whether a bounce might occur can be obtained by considering the scaling of each contribution to the energy density as a function of the scale factor @xmath7 .
to find these scalings , we need the time dependence of the linear and quadratic contributions to each field . therefore , we need to solve the matter field equations of motion . in the following subsection
, we present the equations for the fields at each order , while the solutions and detailed analysis will be performed in the next sections . keeping in mind the anstze for @xmath17 , @xmath31 and @xmath19 , their equations of motion at each order
can be obtained from ( [ phi1eom ] ) , ( [ phi2eom ] ) and ( [ amueom ] ) : + a ) at zero - th order : [ eqn0 ] \ { c _
1^(0 ) + 3h _ 1^(0 ) + m^2_1^(0 ) = 0 , + + _ 2^(0 ) + 3h_2^(0 ) + m^2_2^(0 ) = 0 .
note that there is no equation at this order for @xmath17 because it is of first order in @xmath32 .
+ b ) at first order : [ eqn1 ] \ { c _
1^(1 ) + 3h _ 1^(1 ) + ( + m^2)_1^(1 ) = 0 , + + _ 2^(1 ) + 3h_2^(1 ) + ( + m^2)_2^(1 ) = 0 , + + ( 1-c_1^(0)^2 - d_2^(0)^2)(_f^+3hf^0 ) + - 2(c_1^(0)__1^(0 ) + d_2^(0)__2^(0))f^=0 , + . making use of eqs .
( [ ansatzgauge1 ] ) and ( [ ansatzgauge2 ] ) , the equation for the gauge field can also be rewritten as : [ max ] & & ( 1-c_1^(0)^2 - d_2^(0)^2)(+ h+ ) + & & - 2(c_1^(0)_1^(0 ) + d_2^(0)_2^(0))= 0 .
c ) at second order : [ eqn2 ] \ { c _
1^(2 ) + m^2_1^(2 ) -<f_f^>_1^(0 ) = 0 , + + _
2^(2 ) + m^2_2^(2 ) + <
f_f^>_2^(0 ) = 0 .
+ . here ,
pointed parentheses indicate spatial averaging ( since we are only focusing on the zero mode of the second order field fluctuations ) .
we also neglected the effect of hubble friction since it does not give an important contribution for the second order fluctuations .
in this section we will solve the equations of motion ( [ eqn0 ] ) , ( [ eqn1 ] ) and ( [ eqn2 ] ) to see if and how a bounce will happen .
it is usually useful to perform the analysis in the conformal frame where the conformal time @xmath55 is used rather than the cosmic time . additionally , to extract the dependence on the scale factor , it is convenient to use the following two variables : u_1 ( ) a()_1 ( ) , u_2 ( ) a()_2 ( ) . hereafter , we will use @xmath56 to denote the @xmath50-th order perturbation of the @xmath57-th scalar field .
moreover , for simplicity but without loosing generality , we can parameterize the scale factor @xmath7 as [ para ] a ( ) = a_0t^p = a_0||^ , with p = , where @xmath58 and @xmath59 are the initial value of the scale factor and the equation of state of the universe , respectively .
this is a self - consistent assumption when @xmath59 is nearly a constant .
the evolution of @xmath59 in our case will be shown numerically in section vi . using the parametrization ( [ para ] ) , the equations of motion at zeroth order of the two scalar fields become : [ eqn0new ] \ { c u_1^(0)+(a_0 ^ 2 m^2 ^-)u_1^(0)=0 , + + u_2^(0)+(a_0 ^ 2 m^2 ^-)u_2^(0)=0 , + . where a prime denotes the derivative with respect to conformal time @xmath62 .
their solutions are : u_1^(0 ) & ~ & h_((1-p)am|| ) , + u_2^(0 ) & ~ & h_((1-p)am|| ) , where @xmath63 represents the @xmath64-th order hankel function . far away or close to the bounce ,
i.e. for @xmath65 and @xmath66 , respectively , the approximate solutions are : + 1)oscillations for large values of the scale factor @xmath65 : [ sol0sub1 ] u_1^(0)&~&||^((1-p)a_0m||^ + & & + _ 1^(0 ) ) , + [ sol0sub2 ] u_2^(0)&~&||^((1-p)a_0m||^ + & & + _ 2^(0 ) ) , ( where @xmath67 and @xmath68 are phases set by the initial conditions ) . in terms of the non - rescaled fields @xmath69 one obtains damped ( or anti - damped ) oscillations ( depending on whether we are in an expanding or a contracting period ) [ sol0phisub1 ] _
1^(0)&~&||^((1-p)a_0m||^ + & & + _ 1^(0 ) ) , + [ sol0phisub2 ] _
2^(0)&~&||^((1-p)a_0m||^ + & & + _ 2^(0 ) ) .
2)``frozen '' evolution for small values of the scale factor @xmath66 : [ sol0sup1 ] u_1^(0)&~ & + & & + , + [ sol0sup2 ] u_2^(0)&~ & + & & + , from which it follows that the non - rescaled fields @xmath69 evolve as [ sol0phisup1 ] _ 1^(0)&~ & + & & + , + [ sol0phisup2 ] _ 2^(0)&~ & + & & + , from which we can see that the last term of @xmath70 is a constant mode while the first term is a varying one
. depending on the value of @xmath30 ( or equivalently @xmath59 ) the varying mode could be growing ( for @xmath71 or @xmath72 ) , in which case it becomes dominant , or decaying ( for @xmath73 or @xmath74 ) , in which case it becomes subdominant .
we can usually neglect the decaying part of the fields .
following the steps performed in the last subsection , we can also get the solutions for the first order components of the scalar fields . using the equations ( [ eqn1 ] ) for the first order perturbations we obtain the following equations of motion for @xmath77 and @xmath78 : [ eqn1new ] \ { c u_1^(1)+(k^2+a_0 ^ 2 m^2 ^-)u_1^(1)=0 , + + u_2^(1)+(k^2+a_0 ^ 2 m^2 ^-)u_2^(1)=0 . + . depending on the value of @xmath79
, we obtain different approximation solutions . for wavenumbers large
compared both to the hubble radius and to the mass term , we obtain oscillatory solutions with fixed amplitude .
considering now modes which are still sub - hubble ( i.e. @xmath80 ) but for which the mass term dominates over the contribution of the field tension ( i.e. the term involving @xmath79 ) , we can neglect both the @xmath81 term and the term involving @xmath82 .
the simplified equation for these modes is : \{[u1 in ] c u_1^(1 )
+ a_0 ^ 2 m^2 ^ u_1^(1 ) = 0 , + + u_2^(1 ) + a_0 ^ 2 m^2 ^ u_2^(1 ) = 0 , + . whose solutions are : u_1^(1)&~&h_((1-p)a_0m|| ) + & ~&||^((1-p)a_0m|| + & & + ^(1)_1 ) ,
+ u_2^(1)&~&h_((1-p)a_0m|| ) + & ~&||^((1-p)a_0m|| + & & + ^(1)_2 ) . for modes outside the hubble radius ( @xmath83 )
, we have : \
{ c u_1^(1 ) + ( a_0 ^ 2 m^2 ^-)u_1^(1 ) = 0 , + + u_2^(1 )
+ ( a_0 ^ 2 m^2 ^-)u_2^(1 ) = 0 , + . which have the same form as eq .
( [ eqn0new ] ) so their solution will be the same as given in eqs .
( [ sol0sup1 ] ) and ( [ sol0sup2 ] ) .
we have thus seen that the first order solutions for the scalar fields scale the same way with @xmath84 as the zeroth order solution .
this is because in the small @xmath84 region where the @xmath85 term dominates over the other ones , the equations for first order and zero - th order modes are almost the same .
thus , unless the energy density in the @xmath86 modes dominates at the initial time , it will never dominate over the background contribution from the @xmath87 terms .
thus we can conclude that the first order fluctuations of scalar fields will not prevent the bounce . in this section
, we will analyze the gauge field @xmath37 which is also considered to be of first order . the equation ( [ max ] )
can directly be transformed to conformal frame as : [ gammaeq ]
+ k^2- = 0 .
since the coefficients @xmath21 and @xmath22 are small , we can take the last term to be a source term .
in a first order born approximation , we can write the total solution as _ 0 + , where @xmath88 is the solution for the homogeneous equation obtained by setting @xmath89 , while @xmath90 is the leading correction term obtained by inserting @xmath88 into the source term ( the last term in ( [ gammaeq ] ) ) . the zero - th order ( homogeneous ) equation is easily solved and gives [ gamma ] _ 0 ~(k|| + _ ) .
for the first order equation , it is convenient to define p ( ) - , so that the equation becomes + k^2 + p()_0 = 0 , where we neglected the small term @xmath91 . inserting the solution of @xmath88 ( [ gamma ] )
, we get the following equation for @xmath90 : [ maxnew ] + k^2= -p()_0 .
we are interested in the scaling of @xmath36 as a function of time . for this purpose
, we need to work out the scaling in time of the source term in ( [ maxnew ] ) .
since the solutions for @xmath60 and @xmath61 scale differently in time in the two time intervals discussed in subsection ( [ subseca ] ) , it is necessary to analyze these two intervals separately . for times obeying @xmath92 , then by differentiating ( [ sol0phisub1 ] ) and ( [ sol0phisub2 ] ) with respect to @xmath62 we have : & ~ & ||^ + & & , + _ 2^(0)&~&||^ + & & .
note that @xmath93 is a decaying mode in the contracting phase and thus the last terms inside the square brackets in the above formulae can be neglected compared to the first ones . since p ( ) &
= & - + & & -2(c_1^(0)_1^(0)+d_2^(0)_2^(0) ) , then combining all these results we get : ~c_1||^ . for @xmath66 ,
then differentiating ( [ sol0phisup1 ] ) and ( [ sol0phisup2 ] ) with respect to @xmath62 we obtain
& ~ & , + _ 2^(0) & ~ & , when @xmath71 and ~_2^(0)0 when @xmath73
. then we can solve equation ( [ maxnew ] ) to get : & ~ & c_2||^ , p > + & ~ & c_3(k||+ _ ) .
p < in the above expressions for @xmath90 , @xmath94 , @xmath95 and @xmath96 are complicated prefactors in front of the @xmath62-dependent terms . in summary
, we see that the interactions give only a subleading correction @xmath90 to @xmath88 .
finally , let us consider the homogeneous component of the second order fluctuations of the scalars , namely , @xmath97 and @xmath98 .
if we only consider the energy density up to second order , these second order field perturbations give a contribution through their coupling to the background fields . in the following
we find the solutions of ( [ eqn2 ] ) and study the effects of the induced terms in the stress - energy tensor on a possible bounce .
given the solution for the gauge field @xmath37 obtained in the last subsection , it is easy to rewrite eqs .
( [ eqn2 ] ) as : [ eqn2new ] \
{ c u_1^(2 )
+ a^2m^2u_1^(2 ) = cu_1^(0 ) , + + u_2^(2 ) + a^2m^2u_2^(2 ) = -du_2^(0 ) , + . where we made use of the fact that f_f^ = 2(k^2 ^ 2-a^2 ^ 2)/a^4 . equation ( [ eqn2new ] ) has the same form as the zero - th order equation but with a small source term generated by the interaction with the gauge field .
this equation can be solved using the born approximation ( details are given in the appendix ) .
the general solution is the sum of the general solution of the homogeneous solution plus the solution including the source which has vanishing initial data .
the inhomogeneous term is suppressed by the coupling constants @xmath21 and @xmath22 compared to the homogeneous solution , but , as shown in the appendix , it scales as a high power of @xmath99 . via the coupling to the background scalar fields , the above second order terms enter into the expression for the energy density to second order .
the signs of the corresponding terms in the energy density are indefinite in the sense that they depend on the phases of the initial field configurations .
since it is these terms that dominate the energy density near the bounce , we find that whether a bounce occurs or not depends sensitively on the phases in the initial conditions , and that in fact in the case of many plane wave modes initially excited , a bounce requires very special phase correlations .
in the previous section we have solved all of the field equations up to second order in the amplitude of the fluctuations .
we have found the scaling in time of each field at each order .
now we are ready to look at how all of the terms in the expression for the energy density @xmath100 , @xmath101 and @xmath102 at various orders in perturbation theory ( namely , eqs . ( [ rho0])-([rho2 ] ) )
scale in time .
this analysis is straightforward but very important if we are to determine whether a bounce is possible , since in four space - time dimensional classical einstein gravity with flat spatial sections a bounce can only happen when the negative terms in the energy density catch up to the positive contributions @xcite . in the following
we give a table of how each term contained in @xmath29 scales with time as the background cosmology bounce point ( the bounce which is achieved in the absence of radiation and scalar field inhomogeneities ) is approached .
we will identify the terms which dominate in this limit .
this will give us a good indication under which conditions a bounce can occur .
the tables are structured as follows : the first line `` terms '' , indicates which term we are considering , the next set of lines `` behavior '' gives the scaling in time in the various limits and in the two relevant ranges of the parameter @xmath30 which indicates the equation of state , and the last line gives the sign with which the term contributes to the energy density . note that we focus on the growing mode solution to each field ( which is constant for small @xmath62 in the case @xmath103 ) .
we give separate tables for terms of zero - th , first and second order in @xmath32 .
\a ) for terms contained in @xmath100 : + @xmath104 & @xmath105 & @xmath106 + behavior & c @xmath107 + @xmath9 ( [ cols="^ " , ] & indefinite & indefinite + these terms to not vanish upon spatial averaging .
note that we have expressed the time dependence in terms of the dependence on the scale factor @xmath7 . at this stage
, we only need to focus on the exponent of the power - law scaling .
the more negative the power is , the more rapidly the term grows in a contracting phase ( since @xmath7 is decreasing with time ) . as mentioned earlier
, the conditions for a bounce to occur in four space - time dimensional classical einstein gravity with flat spatial sections is that the total energy density reaches zero during the contracting phase .
thus , there needs to be a negative definite term which starts out small but grows faster than the positive definite terms due to the regular scalar field and regular radiation . in the absence of radiation and scalar field inhomogeneities , it is the contribution to the energy density of the ghost field @xmath108 which plays this role . from the table
we see that there are three kinds of terms : positive definite , negative definite and indefinite ones .
the first set contains the kinetic and potential terms of the normal scalar as well as the free energy density of the gauge field , the second set is made up of the kinetic and potential terms of the ghost scalar , while the third set contains terms which arise due to the coupling terms between scalars or between scalars and gauge fields . looking first at the terms which are independent of the coupling term between the fields , we see from the first line of the `` behavior '' set of lines in the third table that , indeed , in the presence of radiation the energy density in radiation grows faster than that in the two scalar fields , thus preventing a bounce . in the presence of coupling between the fields , however , there are terms which scale with a larger negative power of @xmath7 .
the signs of some of them , however , depend on the initial phases for the linear fields @xmath37 , @xmath34 and @xmath35 .
note that the signs of the scalar coupling terms are determined by the evolution of each field and thus are hard to be identified in a general analysis .
the same is true for the gauge coupling terms ( the last two in the third table ) .
however , the coupling terms between the scalar fields and the gauge field ( the third to last in the third table ) can be made negative / positive definite easily by setting the signs of the coefficients @xmath21 and @xmath22 to be both positive / negative .
it is reasonable to assume that the contracting phase begins with the regular scalar field dominating the energy density , and that the contribution of the lee - wick scalar is much smaller . for single fourier mode initial conditions of the radiation field ,
this can be achieved with the appropriate choice of the initial phase ( see example 1 in the following section containing our numerical results ) . however , for multiple initial radiation fourier modes excited any initial phase difference between the modes will produce a contribution with the wrong sign and will thus prevent a bounce ( see example 3 in the following section ) . in the presence of an infinite set of modes , the phase correlations required to obtain a bounce thus appear to have negligible measure in initial condition space .
thus , even in the presence of coupling between scalar fields and radiation , the lee - wick bounce is unstable .
the bounce , if it exists , will happen at a time which can be chosen to be @xmath109 .
its duration ( the time interval lasting from the time the hubble radius stops decreasing in the contracting phase until when it starts expanding in the post - bounce phase ) will be denoted by @xmath110 . since the various components of the energy density scale with different powers of @xmath7 , it is clear that the duration of the bounce will be shorter or equal to the hubble radius @xmath111 ( which gives the time scale on which the ratios of energy densities in different components change ) at the beginning of the bounce phase .
for the background bounce model , we have @xmath112 .
there are two kinds of bounces according to the duration of the bounce phase i ) if the period @xmath113 , the bounce will go from the time t_b^- ~- ~- to the time t_b^+ ~ ~ with a low speed .
we call this a `` slow bounce '' . in this case
, the universe will enter the bounce period at the critical time @xmath114 , and only the @xmath115 approximate solutions of the previous tables will be applicable and not ones for the interval @xmath116 .
ii ) if the period @xmath117 , the bounce will happen in a very short time with very fast speed .
this can be called the `` fast bounce '' . in this case
, the universe evolves from the far past ( @xmath118 with @xmath119 ) to @xmath109 , passing through the point @xmath120 , then entering into the region @xmath121 before finally reaching the bounce point . in this case , both of the two approximate solutions of the field evolution will be applied .
let us now consider the necessary conditions for a bounce ( as we have indicated above and will see from the numerical analysis , these conditions are not sufficient - in addition to the conditions which follow , appropriate correlations in the initial phases are required ) .
we start in the region of time @xmath115 .
we study the conditions required to have the terms that might give a bounce grow relative to the other terms during this phase .
if the conditions are not satisfied , or the bounce does not happen even if the conditions are satisfied , then a bounce may still occure in the @xmath116 region .
the conditions for the terms in the energy which could compensate the positive radiation contribution to become dominant are then studied .
if these conditions are not satisfied , either , then a bounce is impossible .
a necessary condition for a bounce to be possible requires the growth rate of one of the indefinite sign terms in the third table above exceed all that of all of the positive definite terms . in the @xmath115 region , this requires @xmath122 , which equivalently constrains the equation of state parameter @xmath59 to be in the range @xmath123 .
if this condition is satisfied in this region , then a slow bounce may happen depending on the choice of the initial phases .
if the condition is not satisfied in the @xmath115 region , the universe may evolve into the @xmath116 region , in which the evolution of the fields are different , and new constraints on @xmath30 and @xmath59 will arise if a bounce is to be possible . following the above logic , we find that the conditions under which a bounce might happen are much looser , namely @xmath124 . to summarize this section : we have identified necessary conditions for a bounce to occur . whether one actually does occur even if the conditions are satisfied depends on the initial phases of the fields . this must be studied numerically . in the following section we will give one example of specially chosen phases for which a bounce is possible . however , when we look at a more general choice of phases , the bounce will not occur .
in order to support the analysis in the last section , we performed numerical calculations .
such numerical work is necessary because our analytical analysis is only approximate .
in particular , we worked in perturbation theory up to order second order in @xmath32 .
in addition , even in cases where our analytical analysis would indicate the possibility of a bounce , the perturbative analysis will break down near the bounce point , and there is no assurance that the trends seen in the perturbative analysis will persist .
we have numerically solved the full nonlinear equations of motion for the matter fields in the presence of a homogeneous expanding background cosmology .
the homogeneous cosmology is obtained numerically by solving the first friedmann equation h^2 = , where @xmath125 is newton s gravitational constant ( related to the planck mass used earlier ) , and @xmath29 is the total energy density , averaged over space . figures [ 1]-[6 ] are two groups of numerical results with different parameters . in both cases
we choose the initial energy density of the gauge fields to be larger than that of the lee - wick scalar , but less than that of the normal scalar .
these initial conditions correspond to the situation we are interested in , namely starting in a matter - dominated contracting phase in the presence of some radiation which is sub - dominant .
[ 1 ] , [ 2 ] and [ 3 ] show an example with parameters @xmath126 and @xmath127 .
we choose initial conditions in which a single fourier mode fluctuation is excited , and in which the phases are chosen as indicated in the figure caption .
for these initial phases , we obtain a bounce . in fig . [ 1 ] , we see that the equation of state @xmath59 begins with a value slightly larger than @xmath128 , and then evolves to some nearly fixed value . for the case of our initial condition choice
, it appears to be @xmath129 , in the region where the bounce is allowed to happen . at the bounce point , the equation of state will drop to @xmath130 , while after the bounce , the equation of state will rise again to @xmath131 .
[ 2 ] is the plot of the scale factor in this case which shows explicitly the occurrence of the bounce .
cosmic time @xmath47 ( horizontal axis ) , in the first simulation , a simulation with only a single fourier mode excited and phases chosen as indicated below .
we see that @xmath59 drops to @xmath130 , indicating that there is a bounce .
the background fields are plotted in dimensionless units by normalizing by the mass @xmath132 while the time axis is displayed in units of @xmath133 .
the mass parameters m and m were chosen to be @xmath134 and @xmath135 .
the initial conditions were chosen to be @xmath136 , @xmath137 , @xmath138 , @xmath139 , @xmath140 , @xmath141 . the coefficients @xmath21 and @xmath22 are chosen to be @xmath142 and @xmath143 . the wavenumber is @xmath144 . ] . from the plot
we see that the bounce happens at @xmath109 . ]
[ 3 ] gives a comparison of the energy densities of some components during the process .
initially , we set the energy density of the gauge field @xmath37 to be between the normal scalar and lee - wick scalar .
when the evolution of the universe enters into a region with nearly constant @xmath59 , the gauge - coupling component of energy density @xmath54 will grow very fast .
it is negative and thus enables the negative part of the energy density to catch up with the positive one , thus allowing the bounce to happen .
for the inhomogeneous fluctuation , we choose the wavenumber to be @xmath144 which corresponds to a scale which is observable by cmb and lss experiments . ,
@xmath19 and @xmath37 in the system with parameters chosen as in fig .
the curves from up to down are : @xmath145 ( black ) , @xmath146 ( red ) , @xmath54 ( dark cyan ) and @xmath147 ( blue ) , respectively .
the variables are also normalized with the mass scale @xmath148 . ]
figs . [ 4 ] , [ 5 ] and [ 6 ] give the corresponding results in the case when we choose @xmath126 while @xmath149 ( with all initial conditions identical ) .
this case seems dangerous because the contribution of the lee - wick scalar to the fluctuation terms could lead to an instability .
however , as we have mentioned before , since the effects of the lee - wick scalar are less than that of the normal scalar , it is still possible for the bounce to happen .
[ 4 ] shows the equation of state of the system .
we can see that the evolution of @xmath59 is about the same as that in fig . [ 1 ] , since the change of the sign of @xmath22 does not alter the result too much .
[ 5 ] is the behavior of scale factor in this case while fig .
[ 6 ] gives the comparison of the energy densities of all components . as a function of cosmic time @xmath47 ( horizontal axis )
the behavior that @xmath59 drops to @xmath130 indicating that a bounce takes place .
the background fields are plotted in dimensionless units by normalizing by the mass @xmath132 while the time axis is displayed in units of @xmath133 .
the mass parameters @xmath10 and @xmath13 were chosen to be @xmath134 and @xmath135 .
the initial conditions were chosen to be @xmath136 , @xmath137 , @xmath138 , @xmath139 , @xmath140 , @xmath141 .
the coefficients @xmath21 and @xmath22 are chosen to be @xmath142 and @xmath150 . the wavenumber @xmath144 . ] . from the plot
we see that the bounce happens at @xmath109 . ] , @xmath19 and @xmath37 in the system with parameters chosen as in fig .
the curves from up to down are : @xmath145 ( black ) , @xmath146 ( red ) , @xmath54 ( dark cyan ) and @xmath147 ( blue ) , respectively .
the variables are also normalized with the mass scale @xmath148 . ]
a change in the phase of the initial radiation field velocity will not change the results ( if we keep the other initial conditions fixed ) . on the other hand ,
if we flip the sign of the initial velocity of one of the two scalar fields , then the sign of the dominant contribution to the energy density as we approach the bounce will flip and this will prevent a bounce . if we use initial conditions containing two excited fourier modes , then we obtain a bounce only if the signs of the initial field velocities are both chosen as in the first run whose results are shown here . different phases for the scalar field velocities of the two modes
destroys the possibility of obtaining a bounce .
figures [ 7 ] , [ 8 ] and [ 9 ] show the results for the equation of state parameter @xmath59 , the hubble parameter @xmath2 and the contribution of the various components to the total @xmath29 in the case of a simulation in which two fourier modes are excited , with velocities of both scalar fields having opposite signs from those in the previous example .
as is obvious , a big crunch singularity occurs . as a function of cosmic time @xmath47 ( horizontal axis ) .
the background fields are plotted in dimensionless units by normalizing by the mass @xmath132 while the time axis is displayed in units of @xmath133 .
the mass parameters @xmath10 and @xmath13 were chosen to be @xmath134 and @xmath135 .
this plot is the evolution of the system with two fourier modes combined together .
the one is of which the wavenumber @xmath144 with initial conditions @xmath136 , @xmath137 , @xmath138 , @xmath139 , @xmath140 , @xmath141 , which if taken alone will give the bounce as has been shown in the previous example . the other is of which the wavenumber @xmath151 with initial conditions of the same initial values of the fields but the opposite signs of the scalar field velocity . from this plot
we can see that the combination of the two fourier mode will ( generally ) cause @xmath59 blow up , thus preventing the bounce .
this means that the bounce requires special fine - tuning of the initial phases for each fourier mode .
the coefficients @xmath21 and @xmath22 are chosen to be @xmath142 and @xmath150 . ] . from the plot
we see that there is a singularity at @xmath109 . ] , @xmath19 and @xmath37 in the system with parameters chosen as in fig .
the curves from up to down are : @xmath145 ( black ) , @xmath146 ( red ) , @xmath54 ( dark cyan ) and @xmath147 ( blue ) , respectively .
the variables are also normalized with the mass scale @xmath148 . ]
in this paper we analyzed in detail the possibility of obtaining a cosmological bounce in a model which corresponds to the scalar field sector of the lee - wick theory coupled to relativistic radiation .
it is known that the scalar field sector of the lee - wick theory in the absence of other fields can yield a cosmological bounce @xcite .
in fact , the universe will scale as non - relativistic matter with @xmath152 both before and after the bounce .
thus , this model is a possible realization of the matter bounce " scenario .
however , this background is unstable to the introduction of radiation since in the contracting phase the growth of energy density in radiation will exceed that of matter and will lead to a big crunch singularity as has been shown in previous work @xcite , the introduction of a lee - wick partner to radiation does not prevent this instability . in this paper , we introduced an interaction between the radiation field and the scalar fields .
the interaction could help drain energy from the radiation field to the lee - wick scalar , and thus could prevent the radiation from growing too fast to destroy the bounce .
we analyzed the equations describing the evolution of the three matter fields ( regular scalar field , its lee - wick partner and the radiation field ) on a cosmological background both analytically and numerically .
our analytical analysis was perturbative and made use of the second order using born approximation .
the expansion parameter is set by the initial amplitude of the gauge field .
we solved the equations of motion for each field at each order , and obtained their approximations in different cases .
we compared their contributions to the total energy density , and derived necessary conditions for a bounce to happen . to support our analysis , we also performed numerical calculations . specifically , we investigated initial conditions in which one or two fourier modes of the radiation field and the scalar field fluctuations are excited .
we found special initial conditions which indeed lead to a non - singular bounce . changing
the sign of the initial scalar field velocity will destroy the bounce solution . in the presence of two fourier modes
, we found that a bounce requires identical initial phases for the two modes . for general initial conditions ,
we conjecture that the measure of such initial conditions which lead to a bounce is very small .
we thus find that the addition of coupling terms between the scalar fields and radiation can not save the lee - wick bounce background from the instability problem with respect to the addition of radiation ( nor , for that matter , with respect to scalar field fluctuations ) .
the instability problem with respect to anisotropic stress will be even worse .
the work at mcgill is supported in part by an nserc discovery grant and by funds from the canada research chair program .
rb is recipient of a killam research fellowship .
the work at cycu is funded in parts by the national science council of r.o.c . under grant
nsc99 - 2112-m-033 - 005-my3 and no .
nsc99 - 2811-m-033 - 008 and by the national center for theoretical sciences .
the solution for the second order scalar field correction @xmath154 can be determined using the green function method .
the general solution of ( [ eqn2new ] ) is the sum of the solution @xmath155 of the homogeneous equation which solves the same initial conditions as @xmath156 and the particular solution @xmath157 wich vanishes at time @xmath158 .
the particular solution is given by [ gf ] u ( ) & = & u_a()__i^d()u_b()s( ) + & & -u_b()__i^d()u_a()s( ) , where @xmath159 and @xmath160 are two independent solutions of the homogeneous equation , @xmath161 is the wronskian : ( ) = ( u_a^ u_b - u_b^ u_a ) ^-1 , and @xmath162 is the source inhomogeneity
. we will demonstrate the analysis for the case of @xmath163 .
let us consider evolution for a short interval of time starting at some initial time @xmath158 .
then , we can neglect the expansion of the universe in the equation of motion and take @xmath164 .
we are then interested in how the result scales in @xmath158 . using this trick ,
the solutions of the homogeneous equation can be taken to be u_a ( ) & = & ( _ m ) + u_b ( ) & = & ( _ m ) and the wronskian is ( ) = - where _ m = . combining these results we obtain [ gf3 ] u_1^(2 ) & ~ & -(_m)__i^ ||^ + & & ( _ m)(2k||+2_)((1-p)am||+_1^(0 ) ) + & & + ( _ m)__i^ ||^ + & & ( _ m)(2k||+2_)((1-p)am||+_1^(0 ) ) , + u_2^(2 ) & ~ & -(_m)__i^ ||^ + & & ( _ m)(2k||+2_)((1-p)am||+_1^(0 ) ) + & & + ( _ m)__i^ ||^ + & & ( _ m)(2k||+2_)((1-p)am||+_1^(0 ) ) . note that if we only care about their scalings with respect to conformal time @xmath62 or scale factor @xmath7 , the above solutions can be reduced to : u_1,2^(2)||^a^-+ , and for the case of a matter dominated era where @xmath166 , it is straightforward to show that u_1,2^(2)||^-6a^-3 .
y. f. cai , t. t. qiu , r. brandenberger and x. m. zhang , `` a nonsingular cosmology with a scale - invariant spectrum of cosmological perturbations from lee - wick theory , '' phys .
d * 80 * ( 2009 ) 023511 [ arxiv:0810.4677 [ hep - th ] ] .
s. w. hawking , and g. f. r. ellis , _ the large scale structure of space - time _
, cambridge university press ( 1973 ) ; a. borde and a. vilenkin , phys .
lett . * 72 * , 3305 ( 1994 ) .
a. borde and a. vilenkin , `` eternal inflation and the initial singularity , '' phys .
lett . *
72 * , 3305 ( 1994 ) [ arxiv : gr - qc/9312022 ] .
v. f. mukhanov and r. h. brandenberger , `` a nonsingular universe , '' phys .
lett . * 68 * , 1969 ( 1992 ) ; + r. h. brandenberger , v. f. mukhanov and a. sornborger , `` a cosmological theory without singularities , '' phys .
d * 48 * , 1629 ( 1993 ) [ arxiv : gr - qc/9303001 ] .
m. novello and s. e. p. bergliaffa , `` bouncing cosmologies , '' phys .
rept . * 463 * , 127 ( 2008 ) [ arxiv:0802.1634 [ astro - ph ] ] .
d. wands , `` duality invariance of cosmological perturbation spectra , '' phys .
d * 60 * , 023507 ( 1999 ) [ arxiv : gr - qc/9809062 ] . f. finelli and r. brandenberger , `` on the generation of a scale - invariant spectrum of adiabatic fluctuations in cosmological models with a contracting phase , '' phys .
d * 65 * , 103522 ( 2002 ) [ arxiv : hep - th/0112249 ] .
p. peter and n. pinto - neto , `` primordial perturbations in a non singular bouncing universe model , '' phys .
d * 66 * , 063509 ( 2002 ) [ arxiv : hep - th/0203013 ] ; + f. finelli , `` study of a class of four dimensional nonsingular cosmological bounces , '' jcap * 0310 * , 011 ( 2003 ) [ arxiv : hep - th/0307068 ] ; + l. e. allen and d. wands , `` cosmological perturbations through a simple bounce , '' phys . rev .
d * 70 * , 063515 ( 2004 ) [ arxiv : astro - ph/0404441 ] ; + r. brandenberger , h. firouzjahi and o. saremi , `` cosmological perturbations on a bouncing brane , '' jcap * 0711 * , 028 ( 2007 ) [ arxiv:0707.4181 [ hep - th ] ] ; + s. alexander , t. biswas and r. h. brandenberger , `` on the transfer of adiabatic fluctuations through a nonsingular cosmological bounce , '' arxiv:0707.4679 [ hep - th ] ; + f. finelli , p. peter and n. pinto - neto , `` spectra of primordial fluctuations in two - perfect - fluid regular bounces , '' phys .
d * 77 * , 103508 ( 2008 ) [ arxiv:0709.3074 [ gr - qc ] ] ; + a. cardoso and d. wands , `` generalised perturbation equations in bouncing cosmologies , '' phys . rev .
d * 77 * , 123538 ( 2008 ) [ arxiv:0801.1667 [ hep - th ] ] ; + y. f. cai , t. qiu , r. brandenberger , y. s. piao and x. zhang , `` on perturbations of quintom bounce , '' jcap * 0803 * , 013 ( 2008 ) [ arxiv:0711.2187 [ hep - th ] ] ; + x. gao , y. wang , w. xue and r. brandenberger , `` fluctuations in a hoava - lifshitz bouncing cosmology , '' jcap * 1002 * , 020 ( 2010 ) [ arxiv:0911.3196 [ hep - th ] ] ; + t. qiu and k. c. yang , `` perturbations in matter bounce with non - minimal coupling , '' jcap * 1011 * , 012 ( 2010 ) [ arxiv:1007.2571 [ astro-ph.co ] ] ; +
c. lin , r. h. brandenberger and l. p. levasseur , `` a matter bounce by means of ghost condensation , '' arxiv:1007.2654 [ hep - th ] .
r. h. brandenberger , `` inflationary cosmology : progress and problems , '' arxiv : hep - ph/9910410 . j. martin and r. h. brandenberger , `` the transplanckian problem of inflationary cosmology , '' phys .
d * 63 * , 123501 ( 2001 ) , [ arxiv : hep - th/0005209 ] ; + r. h. brandenberger and j. martin , `` the robustness of inflation to changes in super - planck - scale physics , '' mod .
a * 16 * , 999 ( 2001 ) [ arxiv : astro - ph/0005432 ] .
y. f. cai , t. qiu , y. s. piao , m. li and x. zhang , `` bouncing universe with quintom matter , '' jhep * 0710 * , 071 ( 2007 ) [ arxiv:0704.1090 [ gr - qc ] ] .
t. d. lee and g. c. wick , `` negative metric and the unitarity of the s matrix , '' nucl .
b * 9 * , 209 ( 1969 ) ; + t. d. lee and g. c. wick , `` finite theory of quantum electrodynamics , '' phys .
d * 2 * , 1033 ( 1970 ) ; + b. grinstein , d. oconnell and m. b. wise , `` the lee - wick standard model , '' phys .
d * 77 * , 025012 ( 2008 ) [ arxiv:0704.1845 [ hep - ph ] ] .
j. khoury , b. a. ovrut , p. j. steinhardt and n. turok , `` the ekpyrotic universe : colliding branes and the origin of the hot big bang , '' phys .
d * 64 * , 123522 ( 2001 ) [ arxiv : hep - th/0103239 ] .
j. m. cline , s. jeon and g. d. moore , `` the phantom menaced : constraints on low - energy effective ghosts , '' phys .
d * 70 * , 043543 ( 2004 ) [ arxiv : hep - ph/0311312 ] .
j. karouby and r. brandenberger , `` a radiation bounce from the lee - wick construction ?
, '' phys .
d * 82 * , 063532 ( 2010 ) [ arxiv:1004.4947 [ hep - th ] ] . | it was recently realized @xcite that a model constructed from a lee - wick type scalar field theory yields , at the level of homogeneous and isotropic background cosmology , a bouncing cosmology .
however , bouncing cosmologies induced by pressure - less matter are in general unstable to the addition of relativistic matter ( i.e. radiation ) . here
we study the possibility of obtaining a bouncing cosmology if we add radiation coupled to the lee - wick scalar field .
this coupling in principle would allow the energy to flow from radiation to matter , thus providing a drain for the radiation energy .
however , we find that it takes an extremely unlikely fine tuning of the initial phases of the field configurations for a sufficient amount of radiative energy to flow into matter . for general initial conditions ,
the evolution leads to a singularity rather than a smooth bounce . |
in recent years , considerable attention has been devoted to ground - state properties of low - dimensional quantum systems .
the interest was greatly stimulated by the discovery of the high - temperature superconductivity,@xcite and by subsequent studies which revealed the importance of the cuo@xmath0 plane of high-@xmath1 materials .
however , complete understanding of the high-@xmath1 mechanism still appears to be beyond our reach .
a possible new approach to the high-@xmath1 mechanism was provided by a recent experiment@xcite on the novel material sr@xmath2cu@xmath3o@xmath4 .
this material has periodic line defects in the cuo@xmath0 plane , and thus is composed of ladders with @xmath5 legs interacting weakly with each other .
the limit @xmath6 yields the high-@xmath1 superconductors . at half - filling this material
may be described well by the antiferromagnetic heisenberg model defined on a ladder : @xmath7 ( here @xmath8 denotes the @xmath9 spin at the @xmath10th site of the @xmath11th chain .
we put @xmath12 hereafter . )
it was reported,@xcite indeed , that the double chain ( @xmath13 ) with @xmath14 develops the paring correlation upon doping .
this report was followed by the suggestion@xcite that a set of ladders interacting weakly with frustration might show the superconductivity .
thus , characterization of the ground state of the ladder models can be an essential step to understanding the high - temperature superconductivity .
it has been conjectured@xcite for @xmath15 that the ground state of the model ( [ hamil ] ) with even @xmath5 is spin liquid with the energy gap , while that with odd @xmath5 is critical , or gapless .
this conjecture is supported by a theorem,@xcite numerical calculations up to @xmath16,@xcite and a scaling theory.@xcite in order to explain this remarkable conjecture intuitively , white _ et al._@xcite proposed a resonating - valence - bond ( rvb ) picture of the ground state of the antiferromagnetic ladder models .
this was followed by a proposal@xcite of a hidden - order correlation for the rvb ground state of the double chain : @xmath17 the existence of the long - range rvb correlation supported@xcite the rvb picture for @xmath13 .
the present rvb correlation ( [ rvb_double ] ) was also shown@xcite to be very useful in discussing another issue , namely the criticality of the model ( [ hamil ] ) .
suppose that we change the value of the coupling across the ladder , @xmath18 ( the inter - chain coupling ) .
then the above conjecture arises the following question .
the model is known to be critical , or gapless@xcite for @xmath19 , whereas the model has the gap for @xmath20 according to the above conjecture . how does the energy gap emerge as @xmath18 changes ? in the previous paper@xcite we presented numerical calculations of the rvb correlations as well as another hidden - order correlation , namely the string correlation:@xcite @xmath21
we showed@xcite quite clearly for @xmath13 that the energy gap appears as @xmath22 with @xmath23 .
this conclusion is consistent with other studies.@xcite hidden correlations were first introduced to investigate the ground state of the integer-@xmath24 antiferromagnetic heisenberg chains .
the hidden correlations relevant to the model are the following string correlations,@xcite .
@xmath25 where @xmath26 denote the @xmath27 spins .
the correlations successfully detected the valence - bond - solid ( vbs ) structure@xcite of the ground state of the integer - spin chains as well as their criticality.@xcite now that we know the usefulness of the rvb correlation in the case @xmath13 , it is a challenging problem to generalize the definition ( [ rvb_double ] ) to higher @xmath5 , and to explore the ground state and the criticality of general @xmath5 ladder models . here
we consider the ground state of the ladder model with _
four _ legs .
the hamiltonian is given by ( [ hamil ] ) with @xmath16 .
we generalize the definitions ( [ rvb_double ] ) and ( [ string_double ] ) for the four - leg ladder , and show the rvb structure of the ground state in terms of the correlations .
we also confirm the prediction of the scaling theory@xcite that the ground - state phase diagram consists of two disordered phases covering @xmath28 and @xmath29 , respectively , and the critical point between them , @xmath30 .
this phase diagram is the same as in the two - leg case.@xcite the ground - state properties are known in some cases . in the limit @xmath31 , the model is decoupled to independent four - spin rungs .
the ground state is given by the product of singlets formed on the rungs.@xcite therefore , the energy gap is given by @xmath32 in this strong - coupling region . at the isotropic point @xmath33 , which is of experimental interest,@xcite the energy gap @xmath34 and the correlation length @xmath35
have been numerically estimated as @xmath36 and @xmath37.@xcite at @xmath19 the model is reduced to four @xmath38 heisenberg antiferromagnetic spin chains .
hence , it is critical as noted above . in the limit @xmath39
, the model converges to the @xmath27 heisenberg antiferromagnetic chain , which is massive according to the haldane conjecture.@xcite the haldane conjecture for @xmath27 has been confirmed numerically.@xcite the present paper is organized as follows .
we introduce hidden correlations in the next section , and present a naive discussion on the existence of the hidden correlations in terms of the rvb argument . in section 3 ,
we show numerical results of the hidden correlations with an emphasis on the case @xmath33 , where the corresponding material is available .
we also present an argument on the existence of the rvb structure by considering two double - chain ladders coupled weakly .
this argument intuitively shows the nature of the rvb ground state . in section 4
, we elaborate on the criticality of the model in terms of the hidden correlations . in the last section ,
we give a summary of the present paper .
in this section , we introduce the hidden correlations for the present four - leg - ladder model ( [ hamil ] ) .
they are defined so that they can detect expected rvb patterns .
we present schematic drawings of the expected rvb patterns in fig .
1 ( a ) for @xmath28 and ( b ) for @xmath29 , respectively .
each rvb pattern is arranged as follows .
first for @xmath28 , white _ et al._@xcite proposed that the ground state is dominated by the rvb pattern with vertical singlets and horizontal singlets .
the state is stabilized by the resonance between the configuration of two adjacent vertical singlets and the configuration of two horizontal singlets ; see fig .
this rvb picture was reported@xcite to be very useful for understanding several features of the ladder models .
second , the expected rvb pattern for @xmath29 is given as follows . in the limit @xmath39 , the system ( [ hamil ] ) with @xmath16 converges to the @xmath27 heisenberg antiferromagnetic chain .
the ground state of the chain was described@xcite well in terms of the @xmath27 valence - bond - solid state .
this vbs state is constructed in the following manner : suppose that each spin with @xmath27 consists of four spins with @xmath38 ; form a singlet using two @xmath38 spin of neighbouring sites ; arrange the singlets so that each bond may have two singlets .
all the @xmath38 spins are thus connected each other with the short - range valence bonds consequently .
we expect this state in the limit @xmath39 .
the resonance may arise among the different pairing patterns for finite and negative @xmath18 .
hence we have drawn the pattern which is shown in fig . 1 ( b ) .
we expect that this pattern dominates in the whole region @xmath29 . in order to detect the above rvb patterns , we define the following hidden correlations : @xmath40 and @xmath41 we here generalized the hidden correlations ( [ rvb_double ] ) and ( [ string_double ] ) , which have been applied to the double - chain ladder model.@xcite the angle @xmath42 appearing in ( [ rvb_double ] ) and ( [ string_double ] ) is replaced by @xmath43 in the present definitions ( [ rvb ] ) and ( [ string ] ) .
we show in the present paper that the choice @xmath44 is the most relevant to the four - chain ladder model .
we also show that the correlation ( [ rvb ] ) develops in the phase @xmath28 while the corelation ( [ string ] ) develops in the phase @xmath29 .
let us explain the reasons of these facts briefly , before going into details in 3 and 4 first , in the limit @xmath39 , the model converges to the @xmath27 heisenberg antiferromagnetic chain , as is explained in the previous section .
the correlation ( [ string ] ) then is reduced to @xmath45 where @xmath46 denote the @xmath27 spins . this correlation with @xmath44 is reported to remain finite in the limit @xmath47,@xcite as was mentioned in 1 .
the correlation ( [ rvb ] ) , on the other hand , remains finite in the limit @xmath31 ; see the next section .
second , as was done for the double - chain ladder model,@xcite we can show the existence of hidden orders schematically as follows . consider any spin configurations of the four - leg ladder , satisfying the rvb pattern of fig .
1 ( a ) . we present an example in fig .
we notice that the set of the numbers @xmath48 in fig .
2 ( a ) satisfies the condition @xmath49 for @xmath50 , @xmath51 .
this is due to the formation of two singlets over two neighboring composite spins @xmath52 .
it is easy to see that the quantity @xmath53 remains finite for any @xmath10 and @xmath54 , when it is averaged over all configurations that satisfy the condition ( [ condition]).@xcite this explains that the rvb correlation ( [ rvb ] ) remains finite if the rvb configurations as fig .
1 ( a ) are dominant in the ground state .
the above argument applies to the rvb patterns as fig . 1 ( b )
if we define the composite spin as @xmath55 ; see fig .
2 ( b ) . we can expect that local destruction of the rvb patterns does not result in vanishing of the long - range correlation.@xcite .
in this section , we investigate the ground state of the four - leg ladder by means of the hidden correlations ( [ rvb ] ) and ( [ string ] ) numerically .
the rvb picture explained in the above section is confirmed .
we show that the choice @xmath44 is the most relevant to the present system .
in this subsection , we concentrate on the four - leg ladder in the limit @xmath31 .
the limiting point is expected to be the fixed point of the phase @xmath28.@xcite .
hence , characteristics at the point may be relevant to those of the whole region @xmath28 . in this limit @xmath31 ,
the hamiltonian is reduced to a set of independent rungs .
the ground state is given by the direct product of singlets that are formed on the rungs .
we can calculate the long - range limit of the hidden correlations ( [ rvb ] ) and ( [ string ] ) exactly .
though the ground state is spin liquid.@xcite the hidden correlation ( [ rvb ] ) is long - ranged .
the correlation @xmath56 is plotted against @xmath43 in fig .
3 . the correlation has the maximum around @xmath57 , while the correlations at @xmath58 and @xmath42 are both vanishing .
these behaviors are intrinsic to the whole region of the phase @xmath28 as we see below .
it is apparent , on the other hand , that the correlation ( [ string ] ) is of short range , @xmath59 . in this subsection ,
we concentrate on the system at @xmath33 , where the interaction is spatially isotropic .
this isotropic system is of experimental interest.@xcite .
a strong - inter - chain - coupling expansion starts to fail at this point.@xcite we employed the density - matrix renormalization - group method @xcite in order to treat large systems approximately .
we conclude that ground state properties are indeed consistent with the rvb picture given in the previous section and the recent proposal that the point @xmath31 is the fixed point of the phase @xmath28.@xcite first of all , we show the precision of the present renormalization - group calculation . in fig .
4 we plotted the relative error of the ground state energy with @xmath33 and @xmath60 against the approximate level @xmath61 .
the parameter @xmath61 is the number of states kept;@xcite in this method , we treat only @xmath61 states in the course of the renormalization .
it should be noted that the precision of the correlations is worse than that of the ground - state energy .
the reason may be as follows .
because the hilbert space is restricted to the @xmath61 states , this renormalization - group method may be regarded as a kind of variational method . in many cases ,
the variational calculation yields worse estimation for the correlation functions .
the correlations @xmath62 and @xmath63 are plotted in fig .
5 ( a ) and ( b ) , respectively .
the maxima of the correlations are located around @xmath44 .
this is also reported in the previous subsection for the system with @xmath31 .
we hence observe that the most relevant angle for the correlations is given by @xmath57 as is expected from the discussions in 2 . in order to estimate the infinite distance limit of the correlations
, we plotted @xmath64 and @xmath65 against @xmath66 in fig .
6 ( a ) and
( b ) , respectively .
the correlation ( [ rvb ] ) remains finite ; @xmath67 ( the correlation would have developed fully as @xmath68 , if the ground state satisfied the condition ( [ condition ] ) completely . ) on the other hand , it is seen that the correlation ( [ string ] ) is very small .
considering the precision of the numerical calculators , we conclude that the correlation ( [ string ] ) is of short range .
finally , we show the results of the correlation ( [ rvb ] ) with the angle @xmath58 .
note that this is nothing but the nel correlation .
we present a semi - logarithmic plot of the correlation against @xmath69 ; see fig .
it is seen that it does decrease exponentially .
the correlation length is somewhat consistent with an estimate in ref .
@xcite . in this subsection
, we clarify the essential mechanism of the development of the correlation ( [ rvb ] ) with @xmath44 in the region @xmath28 . for this purpose ,
let us decouple the four - leg ladder system into two ladder systems .
the hamiltonian is given by @xmath70 the effect of the inter - ladder coupling @xmath71 is analysed with the aid of numerical simulations below .
et al._@xcite suggested that the decoupled system ( @xmath72 ) is a convenient starting point for investigating the four - leg ladder .
first , we consider the completely decoupled case @xmath72 .
the hidden correlation ( [ rvb ] ) is reduced to @xmath73 where @xmath74 and @xmath75 .
the brackets here denote the ground - state expectation value of a single two - leg ladder system . in fig .
8 , we show an example of spin configurations @xmath76 and @xmath77 for each single - ladder system .
the configurations are expressed in terms of the height of steps .
the upward ( downward ) step at the position @xmath10 of the upper drawing stands for @xmath78 ( @xmath79 ) .
the same rule applies to the lower drawing .
the example here satisfies the conditions @xmath80 for @xmath50 , @xmath51 ; therefore the sites with the magnetization @xmath81 and @xmath82 appear alternately with the sites with @xmath83 being inserted between them .
indeed , such configurations are expected to dominate the ground state of the two - leg ladder system @xcite , according to the rvb theory for the two - leg ladder.@xcite rvb pattern as in fig .
9 satisfies the condition ( [ conditions ] ) .
we calculated the two terms in eq .
( [ decoupled_rvb ] ) for the exemplified configuration in fig . 8 fixing the angle @xmath43 to @xmath44 .
the expectation values indicated in fig .
8 are the average over the depicted configuration and the reflected ( @xmath84 , @xmath85 ) one .
we can see that the correlation ( [ decoupled_rvb ] ) does not vanish only at the sites where the steps of the upper and lower ladders synchronize coincidently .
we can also show after a similar analysis that the hidden correlation ( [ rvb ] ) does not develop for @xmath58 and @xmath86 , if we restrict the configurations with the condition ( [ conditions ] ) .
this is the same as what we observe in the previous subsection .
thus , the analysis based on the decoupled system appears to be fairly relevant .
now we investigate the effect of the inter - ladder coupling in terms of the above rvb picture numerically . as a consequence , we present more detailed information on the rvb pattern . in fig .
10 , we show the hidden correlation ( [ rvb ] ) for the system with @xmath60 , @xmath33 and @xmath71 varied . we observe that the inter - ladder coupling @xmath71 stabilizes the hidden correlation .
recall that the essential mechanism of the development of the rvb correlation is the synchronization of the configurations of the upper - half ladder and the lower - half ladder ; see fig . 8 .
the rvb correlation @xmath87 for @xmath72 expresses the contribution of the coincidental synchronization .
we can observe from fig .
10 that the inter - ladder coupling enhances the synchronization .
we can explain the reason of the enhancement as follows . according to the rvb picture ,
the ground state is stabilized by the resonance between the configuration of two vertical singlets and that of two horizontal singlets.@xcite when the inter - ladder coupling @xmath71 is turned on , the resonance shown in fig .
11 becomes possible . to take advantage of the resonance , singlets along the upper - half ladder and those along the lower - half ladder may tend to appear at the same position @xmath10 .
this effect can enhance the synchronization of the configurations .
in this section , we investigate the development of the hidden correlations ( [ rvb ] ) and ( [ string ] ) and their criticality when we change the parameter @xmath18 . we show the results of the exact - diagonalization method under the periodic - boundary condition .
the result is consistent with that of the scaling theory@xcite .
we plotted the values of @xmath88 and @xmath89 for the system of the size @xmath90 and @xmath91 in fig .
12 . as is expected , we observe that the correlation @xmath92 develops in the region @xmath28 , while the correlation @xmath93 develops in the region @xmath29 .
we readily know the values of the correlations in two limiting cases . in the limit @xmath31
, we showed @xmath94 . in the limit @xmath39
, the corelation ( [ string ] ) corresponds to the string correlation of the @xmath27 heisenberg chain .
it is reported@xcite that we have @xmath95 . in order to see the development of the correlations and the criticality , we calculated the squared order parameters , @xmath96 and @xmath97 where @xmath98 and @xmath99 we plotted the results in fig . 13 ( a ) and ( b ) , scaling the data by the factor @xmath100 .
this is because of the following reason . at the critical point @xmath30
, the correlations are reduced to @xmath101 where @xmath102 denotes the string correlation for the @xmath9 heisenberg chain:@xcite @xmath103 if we know the correlation exponent of ( [ halfstring ] ) defined in @xmath104 the relation ( [ square ] ) is immediately followed by @xmath105 both for @xmath106 and @xmath107 .
generalizing hida s analysis@xcite we obtain the formula @xmath108 .
therefore , the exponent is given by @xmath109 .
the scaled order parameters are invariant at the critical point for various system sizes . in fig .
13 ( a ) and ( b ) , we observe the crossing points close to the expected critical point @xmath30.@xcite ( another crossing point which appears in fig .
13 ( b ) around @xmath110 may have arose owing to the limited system size . )
the hidden correlations appear to obey the behavior explained in 2 .
the ground state of the four - chain ladder model with both ferromagnetic and antiferromagnetic inter - chain coupling @xmath18 has been analysed by means of the hidden correlations ( [ rvb ] ) and ( [ string ] ) .
the hidden correlations can indicate the development of the corresponding rvb states .
the hidden correlation ( [ rvb ] ) develops in the phase @xmath28 , while the hidden correlation ( [ string ] ) develops in the phase @xmath29 .
though both the two phases are disordered , they are characterized in terms of the hidden long - range orders .
it indicates that the rvb structure varies drastically at the critical point @xmath30 .
these results are consistent with a recent proposal@xcite that the point @xmath31 @xmath111 is the fixed point of the region @xmath28 @xmath112 .
a detailed analysis of rvb pattern has been reported for the phase @xmath28 .
it implies that valence bonds formed along the ladder are not negligible as are in the case of the double - chain ladder model.@xcite this tendency may be more apparent as the number of the legs is increased .
we are in position to conclude that , in the phase @xmath28 , the ground state of the ladder models with two and four legs is the rvb state proposed by white _
et al._@xcite .
the rvb pattern may be common to the arbitrary ladder models with even legs .
our computer programs are partly based on the subroutine package `` titpack ver .
2 '' coded by professor h. nishimori . 99 j. g. bednorz and k. a. mller : z. phys .
b 64 * ( 1986 ) 188 . m. takano , z. hiroi m. azuma and y. takeda : in _ mechanism of superconductivity _ jap .
j. appl.phys .
* 7 * ( 1992 ) p.
e. dagotto , j. riera and d. j. scalapino : phys .
b*45 * ( 1992 ) 5744 . t. m. rice , s. gopalan and m. sigrist : europhys . lett .
* 23 * ( 1993 ) 445 .
s. gopalan , t. m. rice and m. sigrist : phys .
b*49 * ( 1994 ) 8901 .
m. sigrist , t. m. rice and f. c. zhang : phys . rev . *
b 49 * ( 1994 ) 12058 .
i. affleck : phys .
b*37 * ( 1988 ) 5186 .
a. parola , s. sorella and q. f. zhong : phys .
* 71 * ( 1993 ) 4393 .
s. r. white , r. m. noack and d. j. scalapino : phys . rev .
* 73 * ( 1994 ) 886 .
n. hatano and y. nishiyama : to appear in j. phys .
a. y. nishiyama , n. hatano and m. suzuki : j. phys .
* 63 * ( 1995 ) no . 6 .
h. watanabe : preprint .
s. r. white : preprint .
h. a. bethe : z. phys .
* 71 * ( 1931 ) 205 .
s. takada and h. watanabe : j. phys .
* 61 * ( 1992 ) 39 .
t. barnes , e. dagotto , j. riera and e. s. swanson : phys .
b*47 * ( 1993 ) 3196 .
h. watanabe : phys .
b*50 * ( 1994 ) 13442 .
k. totsuka and m. suzuki : submitted to j. phys . : condens
. matter .
m. den nijs and k. rommelse : phys .
b*40 * ( 1989 ) 4709
. s. m. girvin and d. p. arovas : phys . scr .
* t27 * ( 1989 ) 156 .
h. tasaki : phys .
* 66 * ( 1991 ) 798 .
m. oshikawa : j. phys . : condens .
matter * 4 * ( 1992 ) 7469 .
k. totsuka and m. suzuki : j. phys . : condens .
matter * 7 * ( 1994 ) 1639 .
y. nishiyama , k. totsuka , n. hatano and m. suzuki : j. phys .
* 64 * ( 1995 ) 414 .
u. schollwoeck and th .
jolicoeur : europhys .
lett . * 30 * ( 1995 ) 493 .
i. affleck , t. kennedy , e. h. lieb and h. tasaki : phys .
rev.lett .
* 59 * ( 1987 ) 799 .
i. affleck , t. kennedy , e. h. lieb and h. tasaki : commun.math .
115 * ( 1988 ) 477 .
d. p. arovas , a. auerbach and f. d. m. haldane : phys .
lett . * 60 * ( 1988 ) 531 .
m. reigrotzki , h. tsunetsugu and t. m. rice : j. phys . : condens .
matter * 6 * ( 1994 ) 9235
. f. d. m. haldane : phys .
lett . * 93a * ( 1983 ) 464 .
f. d. m. haldane : phys .
* 50 * ( 1983 ) 1153 .
n. hatano and m. suzuki : j. phys .
* 62 * ( 1993 ) 1346 .
s. v. meshkov : phys . rev .
* b 48 * ( 1993 ) 6167 .
j. deisz , m. jarrell and d. l. cox : phys . rev . *
b 48 * ( 1993 ) 10227 .
s. r. white : phys .
lett . * 69 * ( 1992 ) 2863 .
s. r. white : phys . rev . *
b 48 * ( 993 ) 10345 .
k. hida : phys . rev .
* b 45 * ( 1992 ) 2207 .
2 examples of spin configurations . the configurations ( a ) and ( b ) are arranged so as to satisfy the rvb patterns in fig .
1 ( a ) and ( b ) , respectively .
the composite magnetization @xmath113 is shown below each configurations .
the full string order develops in both cases .
4 the relative error of the ground - state energy that is calculated by means of the density - matrix renormalization - group method for the system with the parameter @xmath33 and the system size @xmath60 under the open - boundary condition . fig . 8 an example of the spin configuration for the upper double - chain ladder and the lower ladder .
two terms that appear in eq .
( [ decoupled_rvb ] ) are calculated for the configurations , respectively .
the expectation values are the average over the depicted configuration and the reflected one ( @xmath84 , @xmath85 ) . | the ground state of the four - chain heisenberg ladder model is numerically investigated .
hidden - order correlations suitable for the system are introduced and calculated with an emphasis on the spatially isotropic point , where a corresponding material exists .
the existence of a long - range hidden correlation indicates formation of a short - range rvb state in the case of the antiferromagnetic inter - chain coupling . a transition between the phase of the ferromagnetic inter - chain coupling and that of the antiferromagnetic one is discussed .
keywords : hidden correlation , ladder model , resonating valence bond , haldane phase |
already in 1937 , at the outset of modern solid state physics , de boer and verwey @xcite drew attention to the surprising properties of materials with incompletely filled 3@xmath0-bands .
this observation prompted mott and peierls @xcite to discuss the interaction between the electrons .
ever since transition metal oxides ( tmos ) were investigated intensively @xcite .
it is now well - known that in many materials with partially filled electron shells , such as the 3@xmath0 transition metals v and ni and their oxides , or 4@xmath1 rare - earth metals such as ce , electrons occupy narrow orbitals .
the spatial confinement enhances the effect of the coulomb interaction between the electrons , making them `` strongly correlated '' .
correlation effects can lead to profound quantitative and qualitative changes of the physical properties of electronic systems as compared to non - interacting particles .
in particular , they often respond very strongly to changes in external parameters .
this is expressed by large renormalizations of the response functions of the system , e.g. , of the spin susceptibility and the charge compressibility . in particular
, the interplay between the spin , charge and orbital degrees of freedom of the correlated @xmath0 and @xmath1 electrons and with the lattice degrees of freedom leads to an amazing multitude of ordering phenomena and other fascinating properties , including high temperature superconductivity , colossal magnetoresistance and mott metal - insulator transitions @xcite .
the simplest microscopic model describing interacting electrons in a solid is the one - band , spin-1/2 hubbard model @xcite where the interaction between the electrons is assumed to be so strongly screened that it is purely local .
more generally the hubbard model applies to lattice fermions with a point interaction , such as ultra - cold fermionic atoms in optical lattices where the interaction is indeed extremely short ranged .
the hamiltonian consists of two terms , the kinetic energy @xmath2 and the interaction energy @xmath3 ( here and in the following operators are denoted by a hat ) : [ vollhardt : g11.7 ] @xmath4 \hat{h}_0 & = & \sum_{i , j } \sum_{\sigma } t_{ij } \hat{c}_{i \sigma}^{+ } \hat{c}_{j \sigma}^ { } = \sum_{\bm{k } , \sigma } \epsilon_{\bm{k } } \hat{n}_{\bm{k } \sigma}^ { } \label{vollhardt : g11.7b } \\[10pt ] \hat{h}_{\mathrm{i } } & = & u \sum_{i } \hat{n}_{i \uparrow } \hat{n}_{i \downarrow } , \label{vollhardt : g11.7c}\end{aligned}\ ] ] where @xmath5 are creation ( annihilation ) operators of fermions with spin @xmath6 at site @xmath7 ( for simplicity denoted by @xmath8 ) , and @xmath9 .
the fourier transform of the kinetic energy in , where @xmath10 is the amplitude for hopping between sites @xmath8 and @xmath11 , involves the dispersion @xmath12 and the momentum distribution operator @xmath13 .
this model provides the basis for most of the theoretical research on correlated electrons during the last decades .
the hubbard model describes an interacting many - body system which can not be solved analytically , except in dimension @xmath14 for nearest - neighbor hopping @xcite . in the absence of exact solutions
there is clearly a great need for reliable , controlled approximation schemes for this model .
however , such approximations are not easy to construct as the following observation shows . when viewed as a function of time a given site of the lattice
will sometimes be empty , singly occupied or doubly occupied . for strong repulsion @xmath15 double occupations are energetically very unfavorable and are therefore strongly suppressed , implying @xmath16 . therefore approximation schemes based on the factorization of the interaction term , e.g. , hartree - fock - type mean - field theories , are generally insufficient to explain the physics of electrons in their paramagnetic phase beyond the limit of weak interactions .
this is due to the fact that in such approximations the interaction is described only as an average , static potential , whereby correlations , i.e. , dynamical many - body effects due to the interaction of individual electrons , are excluded from the beginning .
hence correlation phenomena such as the mott - hubbard metal - insulator transition can not be described by such approximations .
this clearly shows the need for comprehensive approximation schemes , which are applicable for all values of the input parameters , e.g. , coupling parameters and temperature , diagrammatically controlled , and thermodynamically consistent @xcite . there exists a well - established branch of approximation techniques which makes use of the simplifications that occur when
some parameter is taken to be large ( in fact , infinite ) , e.g. , the length of the spins @xmath17 , the spin degeneracy @xmath18 , the spatial dimension @xmath0 , or the coordination number @xmath19 , i.e. , the number of nearest neighbors of a lattice site
. is determined by the dimension @xmath0 and the lattice structure . already in @xmath20
the coordination number can be quite large , e.g. , @xmath21 for a simple cubic lattice , @xmath22 for a bcc lattice and @xmath23 for an fcc - lattice , making its inverse , @xmath24 , rather small .
it is then natural to consider the limit @xmath25 to simplify the problem . for a hypercubic lattice , obtained by generalizing the simple cubic lattice in @xmath26 to arbitrary dimensions
, one has @xmath27 .
the limit @xmath28 is then equivalent to @xmath29 .
several standard approximation schemes which are commonly used to explain experimental results in dimension @xmath20 are exact only in @xmath30 @xcite .
] investigations in this limit , supplemented if possible by an expansion in the inverse of the large parameter , often provide valuable insights into the fundamental properties of a system even when the inverse parameter is not very small .
one of the best - known mean - field theories in many - body physics is the weiss molecular - field theory for the ising model @xcite .
it is a prototypical _ single - site mean - field theory _ which becomes exact for infinite - range interaction , as well as in the limit of the coordination number @xmath31 or the dimension @xmath32 . in the latter case @xmath24 or
@xmath33 is a small parameter which can be used to improve the mean - field theory systematically .
this mean - field theory is comprehensive in the sense discussed above .
namely , it contains no unphysical singularities , is applicable for all values of the input parameters , i.e. , the coupling parameter , magnetic field , and temperature , and is diagrammatically controlled @xcite .
itinerant quantum mechanical models such as the hubbard model and its generalizations are much more complicated than classical , ising - type models .
generally there do not even exist semiclassical approximations for such models that might serve as a starting point for further investigations . under such circumstances
the construction of a mean - field theory with the comprehensive properties of the weiss molecular field theory for the ising model will necessarily be much more complicated , too . here
the limit of high spatial dimensions @xmath0 or coordination number @xmath19 has again been extremely useful since it provides the basis for the construction of a comprehensive _ dynamical _ mean - field theory ( dmft ) for lattice fermions . in this article
we will first discuss ( sec .
[ vollhardt : lattice fermions in the limit of high dimensions ] ) the limit of high spatial dimensions @xmath0 for lattice fermions , the scaling of the hopping amplitude which is necessary to obtain a meaningful limit @xmath32 as well as the simplifications of the many - body perturbation theory occurring in this limit . in sec .
[ vollhardt : dynamical mean - field theory for correlated lattice fermions ] the self - consistency equations obtained in the limit @xmath34 are derived which provide the basis for the dmft for correlated lattice fermions .
an example for the many insights gained by the dmft is the mott - hubbard metal - insulator transition discussed in sec .
[ vollhardt : the mott - hubbard metal - insulator transition ] .
the application of the dmft to real correlated materials is described in sec .
[ vollhardt : theory of electronic correlations in materials ] .
brief introductions to the dmft for correlated systems in the presence of disorder ( sec .
[ vollhardt : electronic correlations and disorder ] ) , correlated bosons in optical lattices ( sec .
[ vollhardt : dmft for correlated bosons in optical lattices ] ) , and systems in non - equilibrium ( sec . [
vollhardt : dmft for nonequilibrium ] ) are also given . in sec .
[ vollhardt : summary and outlook ] a summary and outlook is presented .
[ vollhardt : lattice fermions in the limit of high dimensions ] we consider the kinetic energy term ( [ vollhardt : g11.7a ] ) since the interaction term is purely local and is thereby completely independent of the lattice structure and the dimension . for hopping between nearest - neighbor ( nn ) sites @xmath8 and @xmath11 with amplitude @xmath35 on a @xmath0-dimensional hypercubic lattice with lattice spacing @xmath36 ,
the dispersion @xmath12 is given by @xmath37 the density of states ( dos ) corresponding to @xmath12 is @xmath38 which is the probability density for finding , the boltzmann constant @xmath39 , and the lattice constant @xmath36 equal to unity . ]
@xmath40 for a random choice of @xmath41 .
if the @xmath42 are chosen randomly , @xmath12 in is the sum of ( independent ) random numbers @xmath43 .
the central limit theorem then implies @xcite that in the limit @xmath32 the dos is given by a gaussian @xmath44 .
\label{vollhardt : g11.10}\ ] ] unless @xmath45 is scaled properly with @xmath0 this dos will become arbitrarily broad and featureless for .
clearly only the scaling @xmath46 ( `` quantum scaling '' ) yields a non - trivial limit @xmath32 for the dos @xcite .
the interaction term in is seen to be purely local and independent of the surrounding ; hence it is independent of the spatial dimension of the system .
consequently , the on - site interaction @xmath15 need not be scaled .
so we see that the scaled hubbard hamiltonian @xmath47 has a nontrivial @xmath31 limit , where both terms , the kinetic energy and the interaction , are of the same order of magnitude and are thereby able to compete ; here @xmath48 denotes nn sites @xmath8 and @xmath11 .
it is this competition between the two terms which leads to interesting many - body physics .
mathematically this is expressed by the fact that the generic matrix elements of the commutator between the kinetic and the interaction part of the hamiltonian do not vanish in the @xmath49 limit .
the quantum scaling was determined within a @xmath50-space formulation , but it can also be derived within a position - space formulation as will be discussed next .
the most important consequence of the scaling is the fact that it leads to significant simplifications in the investigation of hubbard - type lattice models @xcite . to understand this point better
we take a look at the perturbation theory in terms of @xmath15 . at @xmath51 and @xmath52 the kinetic energy of the electrons
is given by @xmath53 here @xmath54 is the one - particle density matrix which can be interpreted as the probability amplitude for hopping from site @xmath11 to site @xmath8 .
the square of its absolute value is proportional to the probability for an electron to hop from @xmath11 to @xmath8 , i.e. , @xmath55 , since site @xmath11 has @xmath56 nn sites @xmath8 .
the sum of @xmath57 over all nn sites @xmath8 of @xmath11 must then yield a @xmath19 or @xmath0 independent constant . in the limit @xmath32
we therefore find @xmath58 since the sum over nn sites in is of @xmath56 , the nn hopping amplitude @xmath45 must obviously be scaled according to for @xmath59 to remain finite in the limit @xmath60 .
hence , as expected , a real - space formulation yields the same results for the required scaling of the hopping amplitude .
the one - particle green function @xmath61 of the non - interacting system obeys the same scaling as @xmath62 .
this follows directly from its definition @xmath63 where @xmath64 is the time ordering operator , and the time evolution of the operators is given by the heisenberg representation .
the one - particle density matrix is obtained as @xmath65 .
if @xmath66 obeys ( [ vollhardt : g11.20 ] ) the one - particle green function must follow the same scaling at all times since this property does not dependent on the time evolution and the quantum mechanical representation .
the fourier transform @xmath61 also preserves this property .
it is important to realize that , although @xmath67 vanishes for @xmath32 , the particles are _ not _ localized , but are still mobile . indeed , even in the limit @xmath32 the off - diagonal elements of @xmath68 contribute , since a particle may hop to @xmath0 nearest neighbors with reduced amplitude @xmath69 . for general
@xmath70 one finds @xcite @xmath71 is the length of @xmath72 in the so - called `` new york metric '' ( also called `` taxi cab metric '' , since particles only hop along horizontal or vertical lines , never along a diagonal ) . it is the property which is the origin of all simplifications arising in the limit @xmath73 .
in particular , it implies the collapse of all connected , irreducible perturbation theory diagrams in position space @xcite
. in general , any two vertices which are connected by more than two separate paths will collapse onto the same site .
in particular , the external vertices of any irreducible self - energy diagram are always connected by three separate paths and hence always collapse . as a consequence
the full irreducible self - energy becomes a purely local quantity @xcite , but retains its dynamics @xcite [ vollhardt : g11.23 ] @xmath74 in the paramagnetic phase we may write @xmath75 .
the fourier transform of @xmath76 is then momentum - independent @xmath77 this leads to tremendous simplifications in all many - body calculations for the hubbard model and related models .
it should be noted that a @xmath50-independence of @xmath78 was sometimes _ assumed _ as a convenient approximation ( `` local approximation '' ) @xcite .
here we identified the limit where this is indeed exact . in the case of more general interactions than the hubbard interaction , e.g. , nearest - neighbor interactions such as @xmath79 the interaction constant has to be scaled , too , in the limit @xmath32 . in the case of , which has the form of a classical interaction , the `` classical '' scaling @xmath80 is required .
of course , the propagator still has the dependence .
due to all contributions , except for the hartree term , are found to vanish in @xmath81 @xcite .
hence nonlocal interactions only contribute via their hartree contribution , which is purely static .
this gives the hubbard interaction a unique role : of all interactions for fermionic lattice models only the purely local hubbard interaction remains dynamical in the limit @xmath73 @xcite . due to the @xmath50-independence of the irreducible self - energy , ,
the one - particle propagator of an interacting lattice fermion system is given by @xmath82 most importantly , the @xmath50 dependence of @xmath83 comes entirely from the energy dispersion @xmath12 of the _ non_-interacting particles .
this means that for a homogeneous system with the propagator @xmath84 its local part , @xmath85 , has the form @xcite [ vollhardt : g11.34 ] @xmath86 & \equiv & g_{\sigma}(\omega ) .
\label{vollhardt : g11.34b}\end{aligned}\ ] ] in the following we will limit our discussion to the paramagnetic phase and omit the spin index . the spectral function of the interacting system ( often referred to as the `` density of states '' ( dos ) as in the non - interacting case ) is given by @xmath87 for @xmath88 one has @xmath89 . in the limit @xmath90
two quantities then play the most important role : the local propagator @xmath91 and the self - energy @xmath92 .
we now discuss some more consequences of the @xmath50-independence of the self - energy as derived by mller - hartmann @xcite .
let us consider the hubbard model , or any one of its generalizations , in the paramagnetic phase , i.e. , without a broken symmetry .
at @xmath93 the one - particle green function then reads @xmath94 in general , even when @xmath95 is @xmath50-dependent , the fermi surface is defined by the @xmath96 limit of the denominator of as [ vollhardt : g11.40 ] @xmath97 according to luttinger and ward @xcite the volume within the fermi surface is not changed by interactions , provided the effect of the latter can be treated in infinite - order perturbation theory ( i.e. , no broken symmetry ) .
this is expressed by @xmath98 , \label{vollhardt : g11.40b}\ ] ] where @xmath99 is the particle density and @xmath100 is the step function .
in general , the @xmath50-dependence of @xmath101 in implies that , in spite of , the shape of the fermi surface of the interacting system will be quite different from that of the non - interacting system ( except for the fully rotation invariant case @xmath102 . for lattice fermion models in @xmath103 , with @xmath104 , implies that the fermi surface itself ( and hence the volume enclosed ) is not changed by interactions .
limit the notion of a fermi surface of a lattice system is complicated by the fact that the dispersion @xmath12 is not a simple smooth function . ]
the fermi energy is simply shifted uniformly from its non - interacting value @xmath105 , i.e. , @xmath106 , to keep @xmath107 in constant . from
we thus conclude that the @xmath96 value of the local propagator , @xmath108 , and hence of the spectral function , @xmath109 , is not changed by interactions .
renormalizations of @xmath110 can only come from a @xmath50-dependence of @xmath78 , i. e. , if @xmath111 .
for @xmath112 the self - energy has the property @xmath113 which implies quasiparticle ( fermi liquid ) behavior .
the effective mass @xmath114 is seen to be enhanced . in particular
, the momentum distribution @xmath115 has a discontinuity at the fermi surface , given by @xmath116 , where @xmath117 .
the limit of high spatial dimensions @xmath0 or coordination number @xmath19 provides the basis for the construction of a comprehensive mean - field theory for lattice fermions which is diagrammatically controlled and whose free energy has no unphysical singularities .
it starts from the scaled hamiltonian ( [ vollhardt : g11.18 ] ) and makes use of the simplifications in the many - body perturbation theory discussed in sec .
[ vollhardt : sec : scaling ] .
there we found that the local propagator @xmath91 , i.e. , the probability amplitude for an electron to return to a lattice site , and the local , but fully dynamical self - energy @xmath92 are the most important quantities in such a theory . since the self - energy is a dynamical variable ( in contrast to the hartree - fock theory where it is merely an average , static potential ) the resulting mean - field theory is also dynamical and can thus describe genuine correlation effects such as the mott - hubbard metal - insulator transition .
the self - consistency equations of this _ dynamical mean - field theory _ ( dmft ) for correlated lattice fermions can be derived in different ways .
all derivations make use of the fact that in the limit of high spatial dimensions hubbard - type models , i.e. , lattice models with a local interaction , reduce to a `` dynamical single - site problem '' , where the @xmath0-dimensional lattice model is effectively described by the dynamics of the correlated fermions on a single site embedded in a `` bath '' provided by the other particles
. the coordination number @xmath19 can be quite large , as in the case of a face - centered cubic lattice with @xmath118 . in the limit @xmath28 ,
i.e. , @xmath29 , the many - body lattice problem reduces to that of a single lattice site embedded in a dynamical mean field .
as shown in the inset electrons can hop onto and off that site and interact as in the finite - dimensional hubbard model .
therefore the dmft describes the dynamics of the interacting electrons correctly.,scaledwidth=80.0% ] in particular , the derivation by jani @xcite is a generalization of the coherent potential approximation ( cpa ) for disordered systems @xcite . ] to the hubbard model . in the following
we will present today s standard derivation by georges and kotliar @xcite which is based on the mapping of the lattice problem onto a self - consistent single - impurity anderson model ; this approach was also employed by jarrell @xcite . although the dmft equations derived within the cpa approach and the self - consistent single - impurity approach , respectively , are identical it is the latter formulation which was immediately adopted by the community since it makes contact with the theory of quantum impurities and kondo problems ; for a review see @xcite .
this is a well - understood branch of many - body physics @xcite for whose solution efficient numerical codes had been developed already in the 1980 s , in particular by use of the quantum monte - carlo ( qmc ) method @xcite . following the presentation of georges , kotliar , krauth , and
rozenberg @xcite the dmft equations will now be derived using the so - called cavity method used here ( see the definition in ) is opposite to that in ref .
this derivation starts by removing one lattice site together with its bonds from the rest of the lattice .
the remaining lattice , which now contains a cavity , is replaced by a particle bath which plays the role of the dynamical mean field ( see fig . [ vollhardt : dmft ] ) .
now comes a physically motivated step : the bath is coupled , via a hybridization , to the cavity .
the resulting problem then amounts to the solution of an effective single - impurity anderson model where the degrees of freedom of the bath , represented by an appropriate hybridization function , have to be determined self - consistently . to be specific
, we consider the partition function in the grand canonical ensemble @xmath119.\ ] ] the action @xmath17 for the hubbard model is given by @xmath120,\end{gathered}\ ] ] where we use grassmann variables @xmath121 , @xmath122 .
we split the action @xmath17 into three parts @xmath123 where @xmath124 is the part containing only variables on site @xmath125 @xmath126,\ ] ] @xmath127 contains the hoppings between site @xmath125 and other sites of the lattice @xmath128 @xmath129,\ ] ] and the rest , denoted by @xmath130 , is the part of the action where the site @xmath125 and its bonds are removed , i.e. , for @xmath131 one has @xmath132.\end{gathered}\ ] ] we now rewrite the partition function @xmath133 as @xmath134 \\ \times\int \prod_{i\neq0\ , \sigma } d c^{*}_{i\sigma } d c_{i\sigma } \exp[-s^{(0 ) } ] \exp[-\delta s]\end{gathered}\ ] ] and use the ensemble average @xmath135\ ] ] taken with respect to @xmath130 ( the action where the site @xmath136 is excluded ) , with @xmath137 being the corresponding partition function .
then the partition function reads @xmath138
\langle \exp[-\delta s]\rangle_{(0)}.\end{aligned}\ ] ] in the next step we expand the second exponent with respect to the action @xmath127 . a non - trivial limit @xmath139 is obtained by scaling the hopping amplitudes @xmath10 as described in sec . [
vollhardt : sec : scaling ] .
consequently , in the @xmath140 limit only the contribution @xmath141 , where @xmath142 or disconnected contributions made of products of @xmath141 s remain . applying the linked - cluster theorem and collecting only connected contributions in the exponential function one obtains the local action @xmath143,\end{gathered}\ ] ] where the rescaled hoppings are denoted with a star . introducing the hybridization function @xmath144 and employing the free ( `` weiss '' ) mean - field propagator @xmath145
one can express the dmft local action in the following form ( here the site index @xmath136 is omitted for simplicity ) @xmath146 where @xmath147 finally , we need the relation between the green function @xmath148 where the site @xmath136 is removed and the full lattice green function , i.e. , @xmath149 which holds for a general lattice . in order to obtain the full solution of the lattice problem
it is convenient to express the relation between the local green function @xmath150 and the `` weiss '' mean field , @xmath151 , or @xmath152 can be viewed as a `` dynamical mean field '' acting on particles on a site , since they all appear in the bilinear term of the local action ( [ vollhardt : local_action ] ) . ]
@xmath153 in the form of a dyson equation @xmath154^{-1 } & = & [ \mathcal{g}_{\sigma } ( i\omega _ { n})]^{-1}-\sigma _ { \sigma } ( i\omega _ { n } ) \label{vollhardt : bath_function } \\ & = & i\omega _ { n}+\mu -\delta _ { \sigma } ( i\omega _ { n})-\sigma _ { \sigma } ( i\omega _ { n } ) .
\label{vollhardt : dyson_eq}\end{aligned}\]]then the lattice green function ( in @xmath155-space ) @xmath156 is given by @xmath157after performing the so - called lattice hilbert transform we recover the local green function @xmath158 the ionic lattice on which the electrons move , and its structure , are seen to enter only via the dos of the non - interacting electrons .
after analytic continuation to real frequencies the local ( @xmath50 averaged ) propagator reads @xmath159 this completes the derivation of the self - consistent dmft equations .
namely , the functional integral determining the local propagator @xmath160\exp [ -s_{\mathrm{loc } } ] , \label{vollhardt : impurity_problem}\ ] ] where the partition function @xmath161 and the local action @xmath162 are given by @xmath163\ ] ] and , respectively , together with the expression for the lattice green function provide a closed set of equations for the local propagator @xmath164 and the self - energy @xmath165 .
these equations can be solved iteratively : starting with an initial value for the self - energy one obtains the local propagator from and thereby the bath green function @xmath166 from .
this determines the local action which is used to solve the single - impurity problem , leading to a new value for the local propagator and , by employing the old self - energy , a new bath green function , etc .
the single - impurity problem is still a complicated many - body interacting problem which can not , in general , be solved exactly . due to the purely on - site interaction in the local action the dynamics of the hubbard model , ,
remains complicated even in the limit @xmath32 .
exact evaluations are only feasible when there is no coupling between the frequencies as , for example , in the falicov - kimball model @xcite .
this model was solved analytically by brandt and mielsch @xcite soon after the introduction of the @xmath32 limit @xcite . in general ,
the local action is the most complicated part of the dmft equations . to solve the self - consistency equations different techniques ( `` impurity solvers '' )
have been developed which are either fully numerical and `` numerically exact '' , or semi - analytic and approximate .
the numerical solvers can be divided into renormalization group techniques such as the numerical renormalization group ( nrg ) @xcite and the density - matrix renormalization group ( dmrg ) @xcite , exact diagonalization ( ed ) @xcite , and methods based on the stochastic sampling of quantum and thermal averages , i.e. , quantum monte - carlo ( qmc ) techniques such as the hirsch - fye qmc algorithm @xcite and continuous - time ( ct ) qmc @xcite .
semi - analytic approximations such as the iterated perturbation theory ( ipt ) @xcite , the non - crossing approximation ( nca ) @xcite , the fluctuation exchange approximation ( flex ) @xcite , the local moment approach ( lma ) @xcite , and the parquet approximation @xcite can also provide valuable insight .
it quickly turned out that the dmft is a powerful tool for the investigation of electronic systems with strong correlations .
it provides a non - perturbative and thermodynamically consistent approximation scheme for finite - dimensional systems which is particularly valuable for the study of intermediate - coupling problems where perturbative techniques fail @xcite .
the correlation induced transition between a paramagnetic metal and a paramagnetic insulator , referred to as `` mott - hubbard metal - insulator transition ( mit ) '' , is one of the most intriguing phenomena in condensed matter physics @xcite .
this transition is a consequence of the competition between the kinetic energy of the electrons and their local interaction @xmath15 .
namely , the kinetic energy prefers the electrons to move ( a wave effect ) which leads to doubly occupied sites and thereby to interactions between the electrons ( a particle effect ) . for large values of @xmath15
the doubly occupied sites become energetically very costly .
the system may reduce its total energy by localizing the electrons .
hence the mott transition is a localization - delocalization transition , demonstrating the particle - wave duality of electrons @xcite .
mott - hubbard mits are found , for example , in transition metal oxides with partially filled bands .
for such systems band theory typically predicts metallic behavior .
the most famous example is v@xmath167o@xmath168 doped with cr @xcite .
in particular , below @xmath169k the metal - insulator transition in paramagnetic ( v@xmath170cr@xmath171)@xmath167o@xmath168 is of first order @xcite , with discontinuities in the lattice parameters and in the conductivity .
however , the two phases remain isostructural .
making use of the half - filled , single - band hubbard model the mott - hubbard mit was studied intensively in the past @xcite .
important early results were obtained by hubbard @xcite within a green function decoupling scheme , and by brinkman and rice @xcite within the gutzwiller variational method @xcite , both at @xmath172 .
hubbard s approach yields a continuous splitting of the band into a lower and upper hubbard band , but can not describe quasiparticle features .
by contrast , the gutzwiller - brinkman - rice approach gives a good description of the low - energy , quasiparticle behavior , but can not reproduce the upper and lower hubbard bands . in the latter approach
the mit is signalled by the disappearance of the quasiparticle peak . to solve this problem
the dmft has been extremely valuable since it provided detailed insights into the nature of the mott - hubbard mit for all values of the interaction @xmath15 and temperature @xmath64 @xcite .
the mott - hubbard mit is monitored by the spectral function @xmath173 of the correlated electrons @xcite .
while at small @xmath15 the system can be described by coherent quasiparticles whose dos still resembles that of the free electrons , the spectrum in the mott insulator state consists of two separate incoherent `` hubbard bands '' whose centers are separated approximately by the energy @xmath15 .
the latter originate from atomic - like excitations at the energies @xmath174 broadened by the hopping of electrons away from the atom . at intermediate values of @xmath15
the spectrum then has a characteristic three - peak structure as in the single - impurity anderson model , which includes both the atomic features ( i.e. , hubbard bands ) and the narrow quasiparticle peak at low excitation energies , near @xmath175 .
this corresponds to a strongly correlated metal .
the general structure of the spectrum ( lower hubbard band , quasiparticle peak , upper hubbard band ) is rather insensitive to the specific form of the dos of the non - interacting electrons .
the width of the quasiparticle peak vanishes for @xmath176 . on decreasing @xmath15 ,
the transition from the insulator to the metal occurs at a lower critical value @xmath177 , where the gap vanishes .
it is important to note that the three - peak spectrum originates from a lattice model with only _
one _ type of electrons .
this is in contrast to the single - impurity anderson model whose spectrum shows very similar features , but is due to _ two _ types of electrons the localized orbital at the impurity site and the free conduction band .
therefore the screening of the magnetic moment which gives rise to the kondo effect in impurity systems has a somewhat different origin in interacting lattice systems .
namely , as explained by the dmft the electrons provide both the local moments and the electrons which screen these moments @xcite .
interestingly , for any typical spectral function @xmath178 with three peaks , kramers - kronig relations and the dmft self - consistency equations imply that the self - energy @xmath95 abruptly changes slope _ inside _ the central peak at some frequency @xmath179 @xcite , once at positive and once at negative frequency .
while this behavior is not visible in @xmath178 itself , it leads to `` kinks '' in the effective dispersion relation @xmath180 of one - particle excitations , which is defined as the frequency for which the momentum - resolved spectral function @xmath181 @xmath182 @xmath183 @xmath182 @xmath184 $ ] is maximal . for frequencies below @xmath179
the dispersion is given by fermi - liquid ( fl ) theory , @xmath185 , where @xmath186 @xmath182 @xmath187 is the fl renormalization parameter .
the fl regime terminates at the kink energy scale @xmath179 .
this energy can not be obtained within fl theory itself .
namely , it is determined by @xmath186 and the non - interacting dos , e.g. , @xmath188 , where @xmath189 is an energy scale of the non - interacting system such as half the bandwidth @xcite . above
@xmath179 the dispersion is given by a different renormalization with a small offset , @xmath180 @xmath182 @xmath190 , where @xmath191 is the weight of the central peak of @xmath178 .
this theory explains kinks in the slope of the dispersion as a direct consequence of the electronic interaction @xcite .
the same mechanism may also lead to kinks in the low - temperature electronic specific heat @xcite .
these kinks have also been linked to maxima in the spin susceptibility @xcite . of course
, _ additional _ kinks in the electronic dispersion may also arise from the coupling of electrons to bosonic degrees of freedom , such as phonons @xcite or spin fluctuations @xcite .
interestingly , recent experiments @xcite have found evidence for kinks in ni(110 ) , which may be due to the electronic mechanism discussed here . the evolution of the spectral function of the half - filled frustrated hubbard model at a finite temperature is shown in fig . [ vollhardt : fig:4.1 ] . at @xmath192 ( @xmath193 : bandwidth ) in the crossover region .
the crossover from the metal to the insulator occurs via a gradual suppression of the quasiparticle peak at @xmath194 .
the inset shows the @xmath15 dependence of @xmath195 , in particular the rapid decrease for @xmath196 ; from ref .
@xcite.,scaledwidth=70.0% ] this temperature is above the temperature of the critical point so that there is no real transition but only a crossover from a metallic - like to an insulating - like solution
. the height of the quasiparticle peak at the fermi energy is no longer fixed at its zero temperature value .
this is due to the temperature dependent imaginary part of the self - energy .
the spectral weight of the quasiparticle peak is seen to be gradually redistributed and shifted to the upper ( lower ) edge of the lower ( upper ) hubbard band .
the inset of fig .
[ vollhardt : fig:4.1 ] shows the @xmath15-dependence of the spectral function at zero frequency @xmath197 . for higher values of @xmath15
the spectral density at the fermi level is still finite and vanishes only in the limit @xmath198 . for the insulating phase dmft
predicts the filling of the mott - hubbard gap with increasing temperature .
this is due to the fact that the insulator and the metal are not distinct phases in the crossover regime , implying that the insulator has a finite spectral weight at the fermi level .
this behavior has been detected experimentally by photoemission experiments @xcite .
altogether , the thermodynamic transition line @xmath199 corresponding to the mott - hubbard mit is found to be of first order at finite temperatures , and is associated with a hysteresis region in the interaction range @xmath200 , where @xmath201 and @xmath202 are the interaction values at which the insulating and metallic solution , respectively , vanish @xcite . as shown in fig .
[ vollhardt : mit_bluemer ] the hysteresis region terminates at a critical end point . at higher temperatures the transition changes into a smooth crossover from a bad metal to a bad insulator . at half filling and for bipartite lattices in dimensions @xmath203 ( in @xmath204 only at @xmath172 ) the paramagnetic phase is , in fact , unstable against antiferromagnetic long - range order .
the metal - insulator transition is then completely hidden by the antiferromagnetic insulating phase @xcite . in fig .
[ vollhardt : mit_bluemer ] it is seen that the slope of the phase transition line @xmath205 is negative down to @xmath172 , which implies that for constant interaction @xmath15 the metallic phase can be reached from the insulator by decreasing the temperature @xmath64 , i.e. , by cooling .
this anomalous behavior ( which corresponds to the pomeranchuk effect @xcite in @xmath206he , if we associate solid @xmath206he with the insulator and liquid @xmath206he with the metal ) can be understood from the clausius - clapeyron equation @xmath207 . here
@xmath127 is the difference between the entropy in the metal and in the insulator , and @xmath208 is the difference between the number of doubly occupied sites in the two phases . within the dmft
there is no exchange coupling @xmath209 between the spins of the electrons in the insulator , since the scaling implies @xmath210 for @xmath211 .
hence the insulating state is macroscopically degenerate , with entropy @xmath212 per electron down to @xmath172 .
this is larger than the entropy @xmath213 per electron in the landau fermi liquid describing the metal , i.e. , @xmath214 . at the same time
the number of doubly occupied sites is lower in the insulator than in the metal , i.e. , @xmath215 .
the clausius - clapeyron equation then implies that the phase - transition line @xmath64 vs. @xmath15 has indeed a negative slope down to @xmath172 .
however , this is an artifact of the dmft .
namely , there will usually exist an exchange coupling between the electrons which leads to a vanishing entropy of the insulator for @xmath216 .
since the entropy of the insulator vanishes faster than linearly with the temperature , the difference @xmath217 eventually becomes positive , whereby the slope also becomes positive ; this is indeed observed in cluster dmft calculations @xcite .
however , since @xmath218 for @xmath216 the phase boundary must eventually terminate at @xmath172 with infinite slope .
although the hubbard model is able to explain basic features of the phase diagram of correlated electrons it can not describe the physics of real materials in any detail . clearly , realistic theories must take into account the explicit electronic and lattice structure of the systems . until recently the electronic properties of solids were investigated by two essentially separate communities , one using model hamiltonians in conjunction with many - body techniques , the other employing density functional theory ( dft ) @xcite .
dft and its local density approximation ( lda ) have the advantage of being _ ab initio _ approaches which do not require empirical parameters as input .
indeed , they are highly successful techniques for the calculation of the electronic structure of real materials @xcite . however , in practice dft / lda is seriously restricted in its ability to describe strongly correlated materials where the on - site coulomb interaction is comparable with the band width . here , the model hamiltonian approach is more powerful since there exist systematic theoretical techniques to investigate the many - electron problem with increasing accuracy . nevertheless , the uncertainty in the choice of the model parameters and the technical complexity of the correlation problem itself prevent the model hamiltonian approach from being a suitable tool for studying real materials .
the two approaches are therefore complementary . in view of the individual power of dft / lda and the model hamiltonian approach , respectively , a combination of these techniques for _ ab initio _ investigations of correlated materials including , for example , @xmath1-electron systems and mott insulators
, would be highly desirable .
one of the first successful attempts in this direction was the lda+u method @xcite , which combines lda with a static , i.e. , hartree - fock - like , mean - field approximation for a multi - band anderson lattice model with interacting and non - interacting orbitals .
this method proved to be a very useful tool in the study of long - range ordered , insulating states of transition metals and rare - earth compounds .
however , the paramagnetic metallic phase of correlated electron systems such as high - temperature superconductors and heavy - fermion systems clearly requires a treatment that goes beyond a static mean - field approximation and includes dynamical effects , i.e. , the frequency dependence of the self - energy . here the recently developed lda+dmft method , a new computational scheme which merges electronic band structure calculations and the dmft , has proved to be a breakthrough @xcite . starting from conventional band structure calculations in the lda
the correlations are taken into account by the hubbard interaction and a hund s rule coupling term .
the resulting dmft equations are solved numerically , e.g. , with a quantum monte - carlo ( qmc ) algorithm . by construction
, lda+dmft includes the correct quasiparticle physics and the corresponding energetics .
it also reproduces the lda results in the limit of weak coulomb interaction @xmath15 .
more importantly , lda+dmft correctly describes the correlation induced dynamics near a mott - hubbard mit and beyond .
thus , lda+dmft is able to account for the physics at all values of the coulomb interaction and doping level . in the lda+dmft approach the lda band structure
is expressed by a one - particle hamiltonian @xmath219 , and is then supplemented by the local coulomb repulsion @xmath15 and hund s rule exchange @xmath209 .
this leads to a material specific generalization of the one - band model hamiltonian @xmath220 here @xmath221 and @xmath222 enumerate those orbitals for which the interaction between the electrons is explicitly included , e.g. , the three @xmath223 orbitals of the @xmath224 electrons of transition metal ions or the @xmath225 orbitals in the case of rare earth elements .
the interaction parameters are related by @xmath226 which holds exactly for degenerate orbitals and is a good approximation for @xmath223 electrons .
the actual values for @xmath15 and @xmath227 can be calculated by constrained lda @xcite .
in the one - particle part of the hamiltonian @xmath228 the energy term containing @xmath229 is a shift of the one - particle potential of the interacting orbitals .
it cancels the coulomb contribution to the lda results , and can also be calculated by constrained lda @xcite . within the lda+dmft scheme the self - consistency condition connecting the self - energy @xmath230 and
the green function @xmath231 at frequency @xmath232 reads : @xmath233^{-1}\right)_{q m , q^{\prime } m^{\prime } } . &
\label{vollhardt : dyson}\end{aligned}\ ] ] here , @xmath234 is the unit matrix , @xmath235 the chemical potential , @xmath236 is the orbital matrix of the lda hamiltonian derived , for example , in a linearized muffin - tin orbital ( lmto ) basis , @xmath237 denotes the self - energy matrix which is nonzero only between the interacting orbitals , and @xmath238^{-1}$ ] implies the inversion of the matrix with elements @xmath99 ( = @xmath239 ) , @xmath240(=@xmath241 ) , where @xmath242 and @xmath221 are the indices of the atom in the primitive cell and of the orbital , respectively may include additional non - interacting orbitals . ] .
the integration extends over the brillouin zone with volume @xmath243 . for cubic transition metal oxides eq .
can be simplified to @xmath244 provided the degenerate @xmath223 orbitals crossing the fermi level are well separated from the other orbitals @xcite . for non - cubic systems
the degeneracy is lifted . in this case
is an approximation where different @xmath245 , @xmath246 and @xmath247 have to be used for the three non - degenerate @xmath223 orbitals .
the hamiltonian is diagonalized within the dmft where , for example , quantum monte - carlo ( qmc ) techniques @xcite are used to solve the self - consistency equations . from the imaginary time
qmc green function we calculate the physical ( real frequency ) spectral function with the maximum entropy method @xcite . during the last few years
the lda+dmft and other dmft based computational schemes have provided great progress in the understanding of the electronic , magnetic and structural properties of many correlated electron materials .
these materials range from 3@xmath0 transition metals and their oxides , and @xmath1 electron systems , all the way to heusler alloys , ferromagnetic half - metals , fullerenes , and zeolites @xcite .
nevertheless , this framework still needs to be considerably improved before it becomes a truly comprehensive _ ab initio _
approach for complex correlated matter with predictive power .
in particular , the interface between the band - structure and the many - body components of the approach needs to be optimized .
this includes , for example , a solution of the double counting correction problem and a fully self - consistent treatment of the spin , orbital , and charge densities .
another important goal are realistic computations of free energies and forces , and the development of efficient methods to treat non - local correlations with quantum cluster methods @xcite .
transition metal oxides are an ideal laboratory for the study of electronic correlations in solids . among these materials ,
cubic perovskites have the simplest crystal structure and thus may be viewed as a starting point for understanding the electronic properties of more complex systems .
typically , the @xmath224 states in those materials form comparatively narrow bands with width @xmath248 ev , which leads to strong coulomb correlations between the electrons . particularly simple
are transition metal oxides with a 3@xmath249 configuration since , among others , they do not show a complicated multiplet structure .
photoemission spectra provide a direct experimental tool to study the electronic structure and spectral properties of electronically correlated materials .
intensive experimental investigations of spectral and transport properties of strongly correlated 3@xmath249 transition metal oxides started with investigations by fujimori _
et al . _
these authors observed a pronounced lower hubbard band in the photoemission spectra which can not be explained by conventional band structure theory . in photoemission spectroscopy ( pes ) a photon of a given energy is used to emit an electron whose properties ( energy , angular distribution ) are measured by a detector .
angular resolved pes is referred to as arpes .
these techniques measure the _ occupied _ electronic states , i.e. , those states which are described by the full spectral function multiplied by the fermi function @xmath250 .
by contrast , inverse photoemission spectroscopy ( ipes ) measures the _ unoccupied _ electronic states , i.e. , the states described by the full spectral function of a material multiplied by @xmath251 .
ipes is harder to perform and not as accurate as pes .
but in many situations information about the unoccupied states is also available by x - ray absorption spectroscopy ( xas ) .
spectroscopic techniques provide very valuable information about correlated electronic systems since they can directly measure the spectral function of a material , a quantity which can also be directly calculated as discussed in sec .
[ vollhardt : the mott - hubbard metal - insulator transition ] .
in particular , photoemission techniques allow one to detect the correlation induced shift of spectral weight .
in the following we will illustrate the computation of the @xmath50-integrated electronic spectra of correlated materials within the lda+dmft scheme by investigating the two simple transition metal oxides srvo@xmath168 and cavo@xmath168 .
srvo@xmath168 and cavo@xmath168 are simple transition metal compounds with a 3@xmath249 configuration . the main effect of the substitution of sr ions by the isovalent , but smaller , ca ions is to decrease the v - o - v angle from @xmath255 in srvo@xmath168 to @xmath256 in the orthorhombically distorted structure of cavo@xmath168 .
remarkably this rather strong bond bending results only in a 4% decrease of the one - particle bandwidth @xmath193 and thus in a correspondingly small increase of the ratio @xmath257 as one moves from srvo@xmath168 to cavo@xmath168 @xcite .
lda+dmft(qmc ) spectral functions of srvo@xmath168 and cavo@xmath168 were calculated by sekiyama _
_ @xcite by starting from the respective lda dos of the two materials ; they are shown in fig .
[ vollhardt : fig_ldadmft ] .
these spectra show genuine correlation effects , i.e. , the formation of lower hubbard bands at about 1.5 ev and upper hubbard bands at about 2.5 ev , with well - pronounced quasiparticle peaks at the fermi energy .
therefore both srvo@xmath168 and cavo@xmath168 are strongly correlated metals .
the small difference of the lda bandwidth of srvo@xmath168 and cavo@xmath168 is only reflected in some additional transfer of spectral weight from the quasiparticle peak to the hubbard bands , and minor differences in the positions of the hubbard bands .
the dos of the two systems shown in fig .
[ vollhardt : fig_ldadmft ] are quite similar .
in fact , srvo@xmath168 is slightly less correlated than cavo@xmath168 , in accord with their different lda bandwidths .
the inset of fig .
[ vollhardt : fig_ldadmft ] shows that the effect of temperature on the spectrum is weak for @xmath258 k. detailed spectra of srvo@xmath168 and cavo@xmath168 were also computed by pavarini _ _ et al.__@xcite .
since the three @xmath223 orbitals of this simple 3@xmath249 material are almost degenerate the spectral function has the same three - peak structure as that of the one - band hubbard model shown in fig .
[ vollhardt : fig:4.1 ] .
the temperature induced decrease of the quasiparticle peak height is also clearly seen .
we note that the actual form of the spectrum no longer resembles the lda dos used as input , i.e. , it essentially depends only on the first three energy moments of the lda dos ( electron density , average energy , band width ) .
( solid line ) and cavo@xmath168 ( dashed line ) calculated at t=300 k ; inset : effect of temperature in the case of cavo@xmath168 ; after ref .
@xcite.,scaledwidth=70.0% ] in the left panel of fig .
[ vollhardt : fig_xps ] the lda+dmft(qmc ) spectral functions at 300k are compared with experimental high - resolution bulk pes . for this purpose
the full theoretical spectra were multiplied with the fermi function at the experimental temperature ( 20@xmath259k ) and were gauss broadened with the experimental resolution of @xmath260ev @xcite .
the quasiparticle peaks in theory and experiment are seen to be in very good agreement . in particular , their height and width are almost identical for both srvo@xmath168 and cavo@xmath168 .
the difference in the positions of the lower hubbard bands may be partly due to ( i ) the subtraction of the estimated oxygen contribution ( which may also remove some @xmath224 spectral weight below @xmath261 ev ) , and ( ii ) uncertainties in the _ ab initio _ calculation of the local coulomb interaction strength . in the right panel of fig .
[ vollhardt : fig_xps ] comparison is made with xas data of inoue _ et al . _
@xcite . to this end
the full lda+dmft spectrum was multiplied with the inverse fermi function at 80k and was then gauss broadened with the experimental resolution of @xmath262ev @xcite .
the overall agreement of the weights and positions of the quasiparticle and upper @xmath223 hubbard band is good , including the tendencies when going from srvo@xmath168 to cavo@xmath168 ( in fact , ca@xmath263sr@xmath264vo@xmath168 in the experiment ) . for cavo@xmath168
the weight of the quasiparticle peak is somewhat lower than in the experiment .
in contrast to one - band hubbard model calculations , the material specific results reproduce the strong asymmetry around the fermi energy w.r.t .
weights and bandwidths .
( solid line ) and cavo@xmath168 ( dashed line ) with experiment . left : bulk - sensitive high - resolution pes ( srvo@xmath168 : circles ; cavo@xmath168 : rectangles ) . right :
1s xas for srvo@xmath168 ( diamonds ) and ca@xmath265sr@xmath264vo@xmath168 ( triangles ) @xcite .
horizontal line : experimental subtraction of the background intensity ; after ref .
@xcite.,scaledwidth=90.0% ] the experimentally determined spectra of srvo@xmath168 and cavo@xmath168 and the good agreement with parameter - free lda+dmft calculations confirm the existence of a pronounced three - peak structure in a correlated bulk material .
although the dmft had predicted such a behavior for the hubbard model @xcite it was not clear whether the dmft would really be able to describe real materials .
now it is clear that the three - peak structure not only occurs in single - impurity anderson models or the dmft for the hubbard model , but is also a characteristic feature of correlated _ bulk _ matter in @xmath26 .
the properties of real materials are influenced not only by the interaction between the electrons in the periodic crystal lattice , but also by the presence of randomness , e.g. , impurities and lattice defects @xcite . in particular , coulomb correlations and disorder are separately driving forces behind metal - insulator transitions ( mits ) connected with the localization and delocalization of particles .
the mott - hubbard mit is caused by the electronic repulsion @xcite and is characterized by the opening of a gap in the density of states at the fermi level . by contrast , the anderson localization transition is due to coherent backscattering of non - interacting particles from randomly distributed impurities @xcite . at the anderson transition the character of the spectrum at the fermi level changes from a continuous to a dense point spectrum .
both mits can be characterized by a single quantity , the local density of states ( ldos ) .
although the ldos is not an order parameter associated with a symmetry - breaking phase transition , it can distinguish between a metal and an insulator .
the dmft can easily be extended to study correlated lattice electrons with local disorder @xcite .
for this purpose a single - particle term with random local energies @xmath266 is added to the hubbard model , leading to the anderson - hubbard model @xmath267 the ionic energies @xmath266 describe the local , quenched disorder acting on the motion of the electrons .
they are drawn from a probability distribution function @xmath268 , which can be a continuous or a multi - modal function .
the dmft provides a valuable , non - perturbative theoretical framework also for the investigation of correlated electrons in the presence of disorder .
if in the dmft the effect of local disorder is taken into account through the arithmetic mean of the ldos one obtains , in the absence of interactions ( @xmath88 ) , the coherent potential approximation ( cpa ) @xcite , which does not describe the physics of anderson localization . to overcome this deficiency dobrosavljevi and
collaborators formulated a variant of the dmft where the _ geometrically _ averaged ldos is computed from the solutions of the self - consistent stochastic dmft equations @xcite which is then incorporated into the self - consistency cycle @xcite .
thereby a mean - field theory of anderson localization can be derived which reproduces many of the expected features of the disorder - driven mit for non - interacting electrons @xcite .
this scheme uses only one - particle quantities and can therefore easily be included in the dmft to treat disordered electrons in the presence of phonons @xcite or coulomb correlations @xcite .
in particular , the dmft with geometrical averaging allows one to compute the phase diagram for the anderson - hubbard model @xcite . for a continuous disorder distribution function and a half filled band
one finds that the metallic phase is _ enhanced _ at small and intermediate values of the interaction and disorder , but that the metallicity is eventually destroyed when the disorder is strong enough @xcite .
surprisingly , the mott and anderson insulators are continuously connected .
new interesting phenomena are also expected in correlated electron systems when the disorder distribution function has a binary - alloy form .
namely , it was predicted that a disorder induced splitting of the band ( `` alloy - band splitting '' ) can enhance the critical temperature for the onset of itinerant ferromagnetism @xcite .
another direct consequence of the alloy - band splitting is the fact that , if the alloy subband is effectively half - filled , a mott - hubbard metal - insulator transition can occur at non - integer filling @xcite .
the spectral - weight transfer in correlated systems with binary - alloy disorder was also investigated away from the metal - insulator transition regime @xcite . for the periodic anderson model with the binary - alloy disorder analogous behavior
was predicted @xcite .
the observation of bose - einstein condensation ( bec ) in ultra - cold atomic gases has greatly stimulated research into the properties of this fascinating quantum state of matter @xcite .
in particular , experiments with alkali atoms confined in optical lattices @xcite have renewed the theoretical interest @xcite in the physics of strongly correlated bosons on lattices , which promises significant new insights and even applications in fields such as quantum computing @xcite .
a lattice model of interacting bosons with purely local interaction , the bosonic hubbard model , has the form @xmath269 where @xmath270 and @xmath271 are bosonic operators . in the ground state
two characteristically different phases are expected to occur : a bosonic incompressible mott phase with a correlation gap and a compressible superfluid phase characterized by a non - vanishing expectation value @xmath272 which serves as an order parameter @xcite . the construction of a dmft for the bosonic hubbard model which as in the case of the fermionic dmft
becomes exact in the limit @xmath0 or @xmath140 and is valid at all temperatures is made complicated by the fact that the system can bose condense below a bose - einstein condensation temperature @xmath273 .
this has immediate consequences for the expectation value of the kinetic energy in .
namely , while in the normal state only the product @xmath274 has a finite expectation value , in the bose condensed phase even @xmath275 ( the order parameter ) is non - zero . for the expectation value of the kinetic energy in to remain finite in the limit @xmath140 the hopping amplitude then has to be scaled differently in the two phases . in the normal phase the hopping amplitude needs to be scaled as in the fermionic case ( `` quantum scaling '' ) , i.e. , as @xmath276 , while in the condensed phase a classical scaling @xmath277 is required .
such a scaling of the hopping amplitudes can not be performed on the level of the hamiltonian , but is possible in the effective action @xcite .
the bosonic dmft ( `` b - dmft '' ) derived thereby treats normal and condensed bosons on equal footing and thus includes the effects caused by their dynamic coupling .
the self - consistency equations of the b - dmft are those of a single bosonic impurity coupled to two baths , one corresponding to bosons in the normal state and one to bosons in the condensate .
the b - dmft derived in the limit @xmath0 or @xmath140 is again a comprehensive mean - field theory , i.e. , it is valid for all input parameters and all temperatures .
it not only reproduces all exactly solvable limits , such as the non - interacting ( @xmath88 ) and the atomic ( @xmath278 ) limit , but also well - known approximation schemes for interacting bosons .
for example , by neglecting all terms containing the hybridization function in the local action one obtains the mean - field theory of fisher _ et al .
_ @xcite ; for a detailed discussion see @xcite .
the solution of the self - consistent bosonic impurity problem defined by the b - dmft equations requires new theoretical / computational methods .
so far exact diagonalization @xcite and continuous - time qmc @xcite were employed .
thereby the bosonic hubbard model was solved on the bethe lattice for finite @xcite and infinite @xcite coordination number @xmath19 as well as on a simple - cubic lattice @xcite .
the phase diagram of correlated bosons on a simple - cubic lattice computed by the bosonic dmft was found to agree with that obtained by numerically exact qmc to within 2% @xcite .
the b - dmft is expected to be a valuable approximation scheme for the investigation of lattice bosons in situations where exact numerical computations are difficult to perform or inefficient , as in the case of bosons with disorder or many internal degrees of freedom , and for bose - fermi mixtures @xcite .
recently the study of strongly correlated many - body systems out of equilibrium has received much attention @xcite .
this is motivated in particular by experimental progress in the investigation of ultra - cold atomic gases @xcite and time - resolved pump - probe spectroscopy on strongly correlated materials @xcite .
in general the real - time dynamics of correlated systems can be described by the extension of dmft to nonequilibrium , provided that they are dominated by local temporal fluctuations and spatial correlations are not crucial . in nonequilibrium dmft
an effective impurity problem is formulated using the keldysh formalism @xcite , and as in equilibrium dmft this mapping onto a single site becomes exact in the limit of infinite dimensions . for the hubbard model in nonequilibrium
the single - site dmft action reads @xmath279 where the keldysh contour @xmath280 runs from @xmath281 to @xmath282 on the real time axis , back to @xmath281 , and finally to @xmath283 along the imaginary time axis @xcite .
the first term contains the local part of the hamiltonian , e.g. , @xmath284 for a time - dependent interaction .
the second term involves the hybridization function @xmath285 which couples the impurity to a time - dependent bath which must be determined self - consistently .
local contour - ordered correlation functions such as the green function @xmath286 are obtained from the action ( [ vollhardt : noneqaction ] ) as expectation values @xmath287
@xmath182 @xmath288/{\mathcal{z}}$ ] at the appropriate times , where @xmath289 is the contour - ordering operator . for the hubbard model
this evaluation is the most demanding part of nonequilibrium dmft and can so far be done with real - time quantum monte carlo methods @xcite for not too long times , and for sufficiently large @xmath15 using a self - consistent perturbation expansion around the atomic limit @xcite . for the falicov - kimball model ,
on the other hand , closed equations of motion govern the impurity green function @xcite , which can be solved on the real time axis .
the hybridization function @xmath285 in eq . must be determined self - consistently by computing the local self - energy @xmath290 from the dyson equation of the impurity model , calculating the momentum - dependent green function of the lattice model from the lattice dyson equation , integrating over momentum to obtain the local lattice green function , and finally equating it with the impurity green function .
while this procedure is necessary for a gaussian or other general dos , for a semielliptical dos the self - energy can be eliminated and @xmath291 expressed directly in terms of @xmath231 @xcite .
nonequilibrium dmft was used to obtain the response of time - resolved photoemission @xcite and optical spectroscopy @xcite in correlated systems in terms of green functions of the electronic system .
the falicov - kimball and hubbard models were studied in the presence of dc and ac electric fields @xcite , as well as for abrupt @xcite or slow changes @xcite of the interaction parameter as a function of time . as an example , we consider a sudden change ( `` quench '' ) in the interation parameter of the hubbard model from @xmath15 @xmath182 0 ( i.e. , with the non - interacting ground state as initial state ) to finite values of @xmath15 for times @xmath45 @xmath292 @xmath125 . for this case
the dmft equations were solved numerically for the paramagnetic phase and a semielliptic dos @xcite . in the following discussion
the bandwidth is equal to @xmath293 , and time is thus measured in units of @xmath294 , e.g. , on the order of femto - seconds for a bandwidth on the order of ev .
an interesting question is whether such an isolated system can thermalize due to the many - body interaction alone , i.e. , whether at least some of its properties are the same as for an equilibrium system with the same energy @xcite . before the quench
the momentum distribution is a step function , and the fermi surface discontinuity @xmath295 remains nonzero for a finite time after the quench . for quenches to @xmath15 @xmath296 @xmath297
the momentum discontinuity first reaches a so - called prethermalization plateau for @xmath45 @xmath296 5 due to the vicinity of the integrable point at @xmath88 .
this plateau in @xmath295 is given to good accuracy by @xmath298 @xcite , where @xmath19 is the fermi - liquid quasiparticle weight _ in equilibrium _ at zero temperature and for interaction @xmath15 .
this value and also the transient behavior at short times is precisely predicted by second - order unitary perturbation theory in @xmath15 @xcite . on the other hand
, the double occupation essentially relaxes to its thermal value on this timescale , showing that the potential energy ( and therefore also the kinetic energy ) relax quicker than the occupation of individual states . for large @xmath15
the behavior is different , showing strong so - called collapse - and - revival oscillations with approximate frequency @xmath299 .
they stem from the vicinity of the atomic limit ( i.e. , zero hopping amplitude ) , for which the propagator @xmath300 is exactly periodic with period @xmath299 @xcite . for finite
hopping ( small compared to @xmath15 ) these oscillations are damped and decay on timescales of order @xmath301 .
the oscillations of the double occupation are not centered at its thermal value , but rather at a different value that can be derived from strong - coupling perturbation theory @xcite .
the situation is thus similar to that at small coupling in the sense that the relaxation to the thermal state is delayed because the system is stuck in a metastable state close to an integrable point .
however , both the weak - coupling prethermalization plateau in @xmath302 as well as the strong - coupling oscillations vanish in a narrow region of interaction parameters @xmath15 near @xmath303 @xcite . for quenches of @xmath15 to approximately this value the system thermalizes rapidly : both the momentum distribution and thus also the double occupation ( due to energy conservation after quench ) relax to their thermal values . in fact the retarded nonequilibrium green function relaxes to the corresponding equilibrium function @xcite , so that all observables that can be obtained from it tend to the thermal value predicted by equilibrium statistical mechanics .
in other words , for quenches to interaction values in the vicinity of @xmath15 @xmath304 @xmath303 the isolated system indeed thermalizes rapidly due to the many - body interactions .
many interesting questions remain open in this context , e.g. , how thermalization depends on the parameters of the system .
we refer to the reviews @xcite for further discussion .
due to the intensive international research over the last two decades the dmft has quickly developed into a powerful method for the investigation of electronic systems with strong correlations .
it provides a comprehensive , non - perturbative and thermodynamically consistent approximation scheme for the investigation of finite - dimensional systems ( in particular for dimension @xmath26 ) , and is particularly useful for the study of problems where perturbative approaches are inapplicable . for this reason
the dmft has now become the standard mean - field theory for fermionic correlation problems , including cold atoms in optical lattices @xcite .
the study of models in nonequilibrium using an appropriate generalization of dmft has become yet another fascinating new research area @xcite . until a few years ago research into correlated electron systems concentrated on homogeneous bulk systems .
dmft investigations of systems with internal or external inhomogeneities such as thin films and multi - layered nanostructures are still very new @xcite .
they are particularly important in view of the novel types of functionalities of such systems , which may have important applications in electronic devices . here
the dmft and its non - local extensions @xcite will certainly become very useful . in particular , the development of the _ ab initio _ band - structure calculation technique referred to as lda+dmft has proved to be a breakthrough in the investigation of electronically correlated materials .
it provided already important insights into the spectral and magnetic properties of correlated electron materials @xcite .
clearly , this approach has a great potential for further developments .
indeed , it is not hard to foresee that the lda+dmft framework will eventually develop into a comprehensive _ ab initio _ approach which is able to describe , and even predict , the properties of complex correlated materials .
we thank jim allen , vladimir anisimov , nils blmer , ralf bulla , liviu chioncel , theo costi , vlad dobrosavljevi , peter van dongen , martin eckstein , volker eyert , florian gebhard , karsten held , walter hofstetter , vaclav jani , anna kauch , stefan kehrein , georg keller , gabi kotliar , jan kune , ivan leonov , walter metzner , michael moeckel , igor nekrasov , thomas pruschke , xinguo ren , shigemasa suga , gtz uhrig , martin ulmke , ruud vlaming , philipp werner , and unjong yu for valuable collaborations .
support by the deutsche forschungsgemeinschaft through trr 80 and for 1346 is gratefully acknowledged .
kb was also supported by the grant n n202 103138 of the polish ministry of science and education .
d. vollhardt , in _ lectures on the physics of strongly correlated systems xiv _ , _ aip conference proceedings _ , vol .
1297 , ed . by a.
avella , f. mancini ( american institute of physics , melville , 2010 ) , p. 339
m. fabrizio , in _ lectures on the physics of highly correlated electron systems xi _ , _ aip conference proceedings _ , vol .
918 , ed . by a.
avella , f. mancini ( american institute of physics , melville , 2007 ) , p. 3
k. held , i.a .
nekrasov , g. keller , v. eyert , n. blmer , a.k .
mcmahan , r.t .
scalettar , t. pruschke , v.i .
anisimov , d. vollhardt , psi - k newsletter * 56 * , 65 ( 2003 ) . reprinted in phys .
status solidi b * 243 * , 2599 ( 2006 ) a. georges , in _ lectures on the physics of highly correlated electron systems viii _ ,
_ aip conference proceedings _ , vol .
715 , ed . by a.
avella , f. mancini ( american institute of physics , melville , 2004 ) , p. 3
a. lanzara , p.v .
bogdanov , x.j .
zhou , s.a .
kellar , d.l .
feng , e.d .
lu , t. yoshida , h. eisaki , a. fujimori , k. kishio , j.i .
shimoyama , t. noda , s. uchida , z. hussain , z.x .
shen , nature * 412 * , 510 ( 2001 ) k. held , i.a .
nekrasov , g. keller , v. eyert , n. blmer , a.k .
mcmahan , r. scalettar , t. pruschke , v. anisimov , d. vollhardt , in _ quantum simulations of complex many - body systems : from theory to algorithms _
, _ nic series volume _ ,
10 , ed . by j. grotendorst , d. marks , a. muramatsu ( nic directors , forschungszentrum jlich , berlin , 2002 ) , p. 175
a. sekiyama , h. fujiwara , s. imada , s. suga , h. eisaki , s.i .
uchida , k. takegahara , h. harima , y. saitoh , i.a .
nekrasov , g. keller , d.e .
kondakov , a.v .
kozhevnikov , t. pruschke , k. held , d. vollhardt , v.i .
anisimov , phys .
* 93 * , 156402 ( 2004 ) r. van leeuwen , n.e .
dahlen , g. stefanucci , c.o .
almbladh , u. von barth , in _ time - dependent density functional theory _ , _ lecture notes in physics _
706 , ed . by m.a.l .
marques , c.a .
ullrich , f. nogueira , a. rubio , k. burke , e.k.u .
gross ( springer , berlin , 2006 ) m. takizawa , h. wadati , k. tanaka , m. hashimoto , t. yoshida , a. fujimori , a. chikamtsu , h. kumigashira , m. oshima , k. shibuya , t. mihara , t. ohnishi , m. lippmaa , m. kawasaki , h. koinuma , s. okamoto , a.j .
millis , phys .
lett . * 97 * , 057601 ( 2006 ) | the dynamical mean - field theory ( dmft ) is a widely applicable approximation scheme for the investigation of correlated quantum many - particle systems on a lattice , e.g. , electrons in solids and cold atoms in optical lattices . in particular ,
the combination of the dmft with conventional methods for the calculation of electronic band structures has led to a powerful numerical approach which allows one to explore the properties of correlated materials . in this introductory article
we discuss the foundations of the dmft , derive the underlying self - consistency equations , and present several applications which have provided important insights into the properties of correlated matter .
[ avella : vollhardt ] |
current and future radio continuum surveys typically probe redshifts out to @xmath1 and often cover a significant fraction of the sky . the large volumes accessible in these surveys provide a probe of the large - scale structure and thus can be utilised to test cosmological models .
one of the most common approaches to investigate the large - scale distribution of cosmological objects is the two - point angular correlation function ( acf ) which quantifies the projected clustering of galaxies on the plane of the sky . to gain information on the three dimensional distribution of galaxies and their evolution with time
, the redshift distribution of the sample needs to be known .
however , in general , redshifts can not be obtained from radio continuum surveys since the spectra do not show emission or absorption line features .
one way to gain redshift information of these radio sources is to match them to their optical counterparts for which the redshifts are known .
first attempts to detect clustering in radio surveys were carried out in the 1970s , but it was only in 1996 @xcite that the first high - significance detection of the clustering was made using the faint images of the radio sky at twenty centimetres ( first ) survey @xcite .
they found that on angular scales that probe large - scale structure , the acf of galaxies detected down to 1 mjy at 1.4 ghz is well - represented by a power - law , with a slope somewhat steeper than that found for typical optical surveys . a number of other studies , e.g @xcite and @xcite also measured clustering of radio sources using the acf in the first survey , in the nrao vla sky survey ( nvss , @xcite ) and in the westerbork northern sky survey ( wenss , @xcite ) .
whilst there was some disagreement about the slope of the correlation function on larger angular scales , later work by @xcite highlighted problems with their earlier results ( associated with over - cleaning of potential sidelobe sources ) and obtained results from all the surveys consistent with @xcite .
in essence , all these studies are confined to the investigation of the projected clustering signal , since many of the sources are too faint in the optical / ir to obtain accurate redshifts
. however , some information on real - space clustering can be inferred , but this relies on estimates of the average redshift distributions of the sources . during the 1990s
, @xcite developed models to infer the redshift distribution of faint radio sources extrapolating from data at much higher flux densities .
since then , a number of observations have improved our knowledge in this area .
@xcite estimated redshifts of a complete sample of 72 radio galaxies down to 1 mjy in about one square degree ( 65% with spectroscopic redshifts ) . in the combined eis - nvss survey of radio sources ( censors , @xcite ) ,
redshifts were estimated for 150 sources , in a 6 square degree region , with flux densities above 7.2 mjy in nvss ( 63% of them secure spectroscopic redshifts ) .
@xcite studied the optical matches of first sources in the 2df survey @xcite and @xcite studied nvss matches with @xmath2 mag in the 6df survey @xcite . these studies all confirmed the picture that mjy - radio surveys contain a heterogenous population of galaxies that is dominated by agn at higher flux densities and includes significant fractions of fainter star - forming galaxies at lower redshifts .
they also appeared to rule out a large ` spike ' of very low - z objects predicted by some of the dunlop and peacock models .
understanding the nature of the sources in the radio surveys contributes to our knowledge of the @xmath3 of the sources i.e. the clustering strength of the sources relative to clustering strength of the underlying dark matter . knowing the bias is essential for using clustering as a cosmological probe as it enters into measurements of autocorrelations , the integrated - sachs wolf ( isw ) effect and the lensing effect .
however , little is known about the bias of radio sources .
@xcite presented estimates of the bias based on the first sources . since then , different and sometimes contradictory prescriptions for the bias of radio sources have been used ( e.g. , * ? ? ?
* ; * ? ? ?
@xcite utilised a semi - empirical approach with a bias prescription based on the work of @xcite to predict the clustering of radio sources in future radio surveys .
the bias value in these models is artificially kept from rising to `` non - physical '' levels which underscores the lack of understanding of the bias of radio sources .
future radio surveys carried out by the square kilometre array ( ska ) will potentially reach 1 njy , providing catalogs of sources over 3@xmath4 of the sky .
ska pathfinders such as the low frequency array ( lofar ) , the australian square kilometre array pathfinder ( askap ) , the south african karoo array telescope ( meerkat ) , the westerbork synthesis radio telescope ( wsrt ) using the apertif instrument and the extended very large array ( evla ) will soon provide surveys with unprecedented depth and/or sensitivity . the resulting radio auto - correlations and cross - correlations with other datasets such as the cmb can provide valuable tests of cosmology .
they can shed light on the question of non - gaussian initial conditions in the universe @xcite and on issues concerning dark energy via the isw effect ( e.g. * ? ? ?
* ; * ? ? ?
. they may also provide strong tests of modified gravity ( e.g. * ? ? ?
* ) and be used as direct probe of dark matter through gravitational lensing effects ( e.g. * ? ?
* ; * ? ? ?
* ; * ? ? ?
it is essential for these studies to have a good understanding of the underlying bias of radio galaxies . in recent studies , ( e.g. * ? ? ?
* ) , predictions for future constraints on cosmology have been made by marginalizing over a single bias parameter but this does not capture the uncertainties in the evolution of bias which could be very important for the interpretation of measurements .
therefore , in this article we attempt to make a direct measurement of the bias of first radio sources at intermediate redshifts .
we match first sources to galaxies in the sloan digital sky survey data release 7 ( sdss - dr7 , e.g. @xcite ) and determine the redshift distribution of the matched sources .
we then create a catalog of unmatched sources to probe the higher - z population .
the format of the paper is as follows . in [ sec : data ] we discuss the data and our methodology ; in [ sec : results ] we discuss the results and present an estimate of the bias of radio sources at high redshift . finally , in
[ sec : con ] present our conclusions .
our approach to isolating a high - z sample of first sources and estimating its redshift distribution can be summarised in the following steps : 1 .
match the first sources to galaxies from the sdss survey and establish the redshift distribution of the matches from an sdss photometric redshift catalogue 2 .
use the s@xmath0 simulations @xcite to estimate an average redshift distribution for all first sources .
3 . estimate the redshift distribution of unmatched sources by removing the matched distribution from the distribution of all sources .
it is then inferred that the unmatched sources are mostly at higher redshifts .
4 . the angular clustering of the high - z sample
can then be measured and compared with what is expected for dark matter sampling the same redshift range , to obtain an estimate of the bias . in this section
we describe the sample selection of the radio sources .
table [ table : collapsing ] summarises our selection criteria quoted below .
the first survey mapped a region of the sky covering 10,000 @xmath5 in the northern galactic cap at 1.4 ghz down to 1.4 mjy .
the final catalogue contained a total of 816,331 sources with a completeness of 95% down the lower flux level used of 2 mjy .
creating our sample of first sources to be matched to sdss required various steps to minimise potential sources of contamination . in the first step , we removed objects with a high probability of being a sidelobe .
the first survey has assigned to each source a probability of being a sidelobe ranging from 0 ( indicating an object is not a sidelobe ) to 1.0 ( indicating an object is a sidelobe ) . to reduce this source of contamination
, we explored various sidelobe probability values on our initial clustering analysis .
this is discussed in more detail in
[ sec : results ] .
however , we note that for our final selection we found a sidelobe probability value of 0.7 led to results the had minimal effects from the presents for sidelobes or the over - cleaning of them . for a sidelobe probability of 0.7
we were left with 795,453 sources .
the next step required the collapsing of multiple components ( e.g. double lobes ) to a single source .
following @xcite we chose a collapsing radius of 72 " .
this is the linking length of the friends - of - friends algorithm the we use to generate the groups of sources .
we found that the average collapsed group had 2 to 3 components and a few groups that had up to 20 components . to compute the flux for each collapsed source
, the integrated flux of each component was added together .
the flux - weighted average positions were then calculated and used to match with the sdss .
this collapsing radius reduced the sample to 647,985 sources .
[ table : collapsing ] llcll + radio sample & numbers + + total first & 816,331 + * no . of sources after side - lobe removal * & * 795,453 * + collapsing sources in groups @xmath6 72 " & 253,971 + collapsed sources & 106,503 + single sources & 541,482 + * no .
of sources after collapsing * & * 647,985 * + + * no . of sources in selected area
* + * ( 130 @xmath7 ra @xmath7 240 , 5 @xmath7 dec @xmath7 55 ) * & * 307,859 * + + * no . of sources
@xmath8 2 mjy * & * 219,060 * + furthermore we only take into account sources within a region that avoided both gaps in data and the edges of the sdss and first surveys .
this region is defined by , 130 @xmath7 ra @xmath7 240 , 5 @xmath7 dec @xmath7 55 , covering a total area of 4613.43 @xmath5 .
our final catalogue of first sources to be matched with sdss contained a total of 307,859 objects . finally , in an attempt to minimize effects due to fluctuations in sensitivity noted in @xcite we applied a 2 mjy flux cut which is more than @xmath9 times the rms fluctuations in the considered region .
this leaves us with 219,060 sources . in an attempt to isolate the agn in the sample and exclude most of the low - z star - forming galaxies , we consider a sample containing only sources with flux densities greater than 7 mjy @xcite this also allows us to compare the redshift distribution to the censors survey
this leaves us with 93,202 sources in the 7 mjy subsample .
redshift distributions for the same cuts are shown as dashed lines .
the distributions correspond a sky coverage of 4613.43 @xmath5 and the s@xmath0 sample has been scaled accordingly to reflect this . , scaledwidth=49.0% ] to match our first sample to their optical counterpart we used data from the sdss - dr7 ( see e.g. * ? ? ?
* for a description of the seventh data release ) . in broad terms ,
the sdss has mapped a quarter of the entire sky with unprecedented accuracy using multi - band photometry ( @xmath10 and @xmath11 ) from the 2.5-meter telescope on apache point to a limiting magnitude of @xmath12 .
the second phase of the project is now complete and is ideally suited to our studies as it is fully contained within the first survey area .
the number density of sdss - dr7 photometric sources is orders of magnitudes greater than the density of 2 mjy first sources .
the average size of sdss galaxies is between 2 `` and 5s . to avoid erroneous matches we have chosen a relatively conservative matching radius of 2 '' to match our first sample to the sdss - dr7 photometric catalogue . to ensure accurate matches we only consider objects classified by the sdss pipeline as a galaxy , requiring that they are successfully deblended to obtain precise positions , and have reliable photometric measurements in all 5 sdss filters .
redshifts for the matched sdss galaxies are taken from @xcite .
specifically , we use the photometric redshift estimated from a neural network method inferred from the 4 galaxy colours and 3 concentration indices .
this estimate is recommended for faint ( @xmath13 ) galaxies , which dominate the matched galaxy sample .
finally , we apply a minimum redshift cut of @xmath14 to remove contamination from misidentified stars .
it should also be noted that we are likely to miss some of the optical identifications of fairly nearby multi - component radio sources as the collapsed source position may not give the position of the optical counterpart accurately enough . these sources are included in the redshift distribution of the simulations ( but not in the matched redshift distribution ) and thus will be included correctly in the unmatched redshift distribution .
our method for probing the average bias of the unmatched sample is thus still valid , but this effect could make the interpretation of the average bias more complicated .
thus , for our central analysis we use four samples : 45,883 first matched galaxies , 173,177 first unmatched galaxies with fluxes greater than 2 mjy , and similarly 15,842 matched ( 77,360 unmatched ) galaxies with fluxes greater than 7 mjy .
we probe the evolution of the bias in the matched sample by considering three redshift bins corresponding to @xmath15 , @xmath16 and @xmath17 , which were chosen such that each bin contains approximately the same number of galaxies ( see table [ table : matching ] for a summary ) .
llcllcllc + matched / unmatched samples & 7 mjy cut & 2 mjy cut & + + total number of sources & 93,202 & 219,060 + sdss matched & 15,842 & 45,883 + sdss unmatched & 77,360 & 173,177 + redshift cuts of matched : & + @xmath18 & 4334 & 14,488 + @xmath16 & 5491 & 15,533 + @xmath17 & 6017 & 15,862 + + + we now compare our matched redshift distributions to that of the publicly available semi - empirical simulation of extragalactic radio continuum sources ( s@xmath0 ) by @xcite which is part of the ska simulated skies ( s@xmath0 ) project .
the s@xmath0 covers a sky area of @xmath19 deg@xmath20 , out to a cosmological redshift of @xmath21 .
the simulated sources were drawn from observed ( or extrapolated ) luminosity functions and grafted onto an underlying dark matter density field with biases which reflect their measured large - scale clustering .
for each source , which include frii galaxies , fri galaxies , radio - quiet quasars , starburst galaxies and star forming galaxies , the database gives the radio fluxes at observer frequencies of 151 mhz , 610 mhz , 1.4 ghz , 4.86 ghz and 18 ghz , down to flux density limits of 10 njy .
a prescription for clustering that captures the clustering pattern on large scales ( larger than those where non - linear evolution of density fluctuations becomes important ) was used .
the simulations can be used to predict the redshift distribution of sources as a function of the flux cutoff of surveys .
figure [ fig : dndz ] shows the redshift distributions for our matched samples ( solid lines ) at the 7 mjy ( red ) and 2 mjy cuts ( blue ) , compared to the s@xmath0 simulation ( dotted line ) for the same flux cuts .
in general we find agreement between the observed matched and simulated redshift distributions up to @xmath22 .
however , we do note that the prominent low redshift spike observed in the s@xmath0 data at @xmath23 does not appear in our matched sample .
there are three different estimators that are used in the determination of two - point correlation function as originally developed by @xcite , @xcite and @xcite . for this work ,
we apply the @xcite estimator , as it reduces errors caused by edges of catalogues and sub - samples during error calculation .
this estimator can be written in the form : @xmath24 where @xmath25 counts the number of pairs in the observed data as a function of angular scale . similarly
, @xmath26 counts the number pairs for the random catalogue and @xmath27 is the number of cross pairs between data and random catalogue .
the integral constraint is negligable .
for our analysis we populated our random catalogue with 50 times the number of sources contained in the data for the matched and unmatched samples , and 100 times the data from the three redshift bins ( cf . , table [ table : matching ] ) .
the errors on @xmath28 were calculated using jack - knife re - sampling @xcite . in this approach
the data was split into @xmath29 bins in ra and the correlation function is recalculated repeatedly each time leaving out a different bin .
a set of @xmath30 values @xmath31 for the correlation function are obtained and the jack - knife error of the mean , @xmath32 , is calculated by @xmath33 each of the 24 bins can be considered to be fairly independent due to the physical separation at the redshift probed . in order to avoid problems associated with
the over - cleaning of sidelobes , which effects the correlation function at @xmath34 , and any potential problems associated with collapsing multi - component sources , we only examine clustering at angles @xmath35 .
we are also concerned that measurements at angles larger than @xmath36 may be unreliable ( see section 3 ) . to determine the bias of the radio population we compare their acf with the corresponding dark matter correlation function .
if @xmath37 is the normalised redshift distribution of a population of radio galaxies , the dark matter acf can then be predicted from the non linear dark matter power spectrum ( @xmath38 ) via limber s equation . for spatially flat cosmologies
one derives the following expression . @xmath39\ ] ] where @xmath40 , @xmath41 is the zeroth - order bessel function of the first kind and @xmath42 is the radial comoving distance .
here we adopt the fitting function for the non - linear cdm power spectrum by @xcite using cosmological parameters given in @xcite .
the linear bias , @xmath43 , can be written @xmath44 where @xmath45 is the power spectrum of luminous tracers of the dark matter . here
, we measure a bias parameter , @xmath46 , in the angular clustering signal which samples @xmath47 for radio sources in first : @xmath48 the derivative @xmath49 at a given redshift @xmath11 reveals the contribution of that redshift slice to the overall acf at the angle @xmath50 .
the upper panels of figure [ fig : dm_pred ] show @xmath49 as a function of redshift for the matched and unmatched samples ( left and right panel , respectively ) at three different angles ( @xmath51 , @xmath52 and @xmath53 ) .
based on that one can determine the average redshift , @xmath54 , which is probed at an angle @xmath50 for a given @xmath37 by , @xmath55 the lower panel of figure [ fig : dm_pred ] shows @xmath54 based on the redshift distributions of the sdss matched and unmatched samples and the overall set of s@xmath0 sources . for small angels , @xmath56 the ( un)matched sample probes redshifts of @xmath57 . for angles above
@xmath58 the average redshift probed is below @xmath59 irrespective of which sample is considered .
figure [ fig:2pcf ] shows the acf for the 2mjy ( left ) and 7mjy ( right ) matched ( red circles ) and unmatched ( blue squares ) samples . in each panel
the dark matter ( dm ) predictions are shown as dashed and solid lines , respectively . the bias ( equ .
[ equ : bias ] ) is computed from the ratio between the data and predicted dark matter correlation functions and is shown as a function of angle in the lower panel for the 2mjy sample . for the 2mjy cut
, we see that the matched sample is more clustered ( in angular projection ) than the unmatched sample .
this is expected since the matched sample occupies lower redshift ranges ( cf . ,
[ fig : dndz ] ) , thus a given angle corresponds to smaller physical scales where there is more clustering .
the acf for the full 1 mjy sample found by @xcite lies between our matched and unmatched curves .
the amount of clustering measured in both the matched and unmatched samples at angles greater than @xmath58 is difficult to explain when one considers the results in figure [ fig : dm_pred ] . on these scales
, one would expect to probe @xmath60 where the sample contains many fainter star - forming galaxies with a bias similar to normal galaxies i.e. @xmath61 .
instead , we see a bias @xmath62 for the unmatched and values @xmath63 for the matched sample . to explore the possibility that large angle fluctuations are due to systematic variations in source density associated with different observing epochs , we plot fractional number density variation as a function of declination in the left panel of figure [ fig : decstrips ] and note some fairly large changes in the fractional number density . to investigate the impact of this on the correlation function measurements , we divide the first sources into declination strips roughly associated with different observing epochs and
calculate the acf in each strip .
the results shown in the right panel of figure [ fig : decstrips ] indicate that , beyond @xmath58 , the results in the different declination strips start to differ .
this suggests that systematics might have a significant effect on larger scales .
however , we note that on smaller scales the measurements are consistent with each other , indicating that these scales are free of systematics related to this effect .
we discuss other possible explanations for the excess large scale - power seen in the full sample in section 3.2 .
the right panel of figure [ fig:2pcf ] shows the acf for the 7mjy sample , which should be completely dominated by agn @xcite .
the clustering of the matched sample is consistent with that of the 2mjy matched sample . in the unmatched sample , the bias is higher at large angles .
the low measurements at smaller angles may indicate that the sidelobe over - cleaning problem is more pronounced for brighter sources . to help interpret the matched acf
, we split the 2 mjy matched sample into three redshift slices , keeping the number of sources in each slice approximately constant . in the left panel of figure [ fig:2pcf_zslice ]
, we plot the acf for each of the redshift slices and in the right - hand panel we plot the bias calculated as a function of angle for each slice .
one sees that the bias for the lowest redshift slice is fairly close to @xmath64 , as one would expect for a population dominated by fairly ordinary star - forming galaxies .
sources in the highest redshift bin are much more biased , as one would expect for a population dominated by agn that trace large halo masses in the universe . the important point
to note is that according to figure [ fig : dm_pred ] the average redshift probed for the matched sample at larger angles is about @xmath65 , but we see a large bias for the matched sample , left panel in figure [ fig:2pcf ] , at these angles and this can be attributed to the more highly biased population at @xmath66 . given that our main aim in this work is to constrain the bias toward high redshifts , we choose an angle of @xmath67 to determine the clustering behaviour of the high redshift radio sources . according to figure [ fig : dm_pred ] this choice allows us to probe bias at @xmath68 .
we find that the unmatched sources are more biased than the matched sample ( at @xmath69 ) , with a value of @xmath70 , compared to @xmath71 for the matched sample at a mean redshift of @xmath72 .
.bias results measured at an angle of 0.66@xmath73 for the matched , unmatched and the three redshift bins . [ cols="<,<,^ , < , < " , ] in this paper , our results are based on measurements at angles smaller than @xmath58 but it is interesting to consider explanations for the excess power at larger angles in the @xmath74 sample . 1 . following the discussion for the matched sample , we could reason that , a highly biased population at high redshift could contribute significantly to the measurement at @xmath75 , even though figure [ fig : dm_pred ] indicates that the average redshift probed on large angles is small .
bias of @xmath76 , however , is not seen even for fairly massive clusters and additional contributors should be considered 2 .
systematics other than those discussed in section 3.1 could also contribute . the beam shown in @xcite for the nvss survey does not go to zero at large angles , suggesting that bright sources could produce artefacts at large angles due to imperfect cleaning in vla data .
however , similar ` excess power ' is observed in the sumss radio survey which was carried out using a very different kind of telescope @xcite .
nevertheless , there is a possibility that radio surveys contain spurious sources which are correlated on large angles and this is a possible explanation for the excess power observed in clustering studies .
3 . there is a low - redshift spike in the source counts , not included in the redshift distribution used for the dark matter predictions .
this would push up the clustering amplitude on all angular scales , but particularly on the larger scales .
however , the similar behaviour of the 2 and 7mjy indicate that the excess power is not due to faint , low - z star - forming galaxies .
also , the results of @xcite and @xcite appear to rule out this explanation .
the s@xmath0 redshift distribution which we use here is designed to fit these observations . a hypothetical low - redshift population which would have been missed in these studies
would need to have @xmath77 and @xmath78 , making such a low - z obscured population an unlikely explanation for much of the excess power in the unmatched sample .
our matching technique is likely to result in some low - z multi - component radio sources being missed in our matched sample and one would expect these sources to be more biased than ordinary galaxies .
this could boost the amplitude of clustering on large scales .
finally , there is the possibility that non - gaussian initial conditions could generate more clustering on large scales than in the standard model as suggested by @xcite .
further work is clearly needed to understand the excess power in the clustering signal on large angles .
we carried out a number of tests to check the robustness of our results . in the first test , we changed the matching radius to 1 " to decrease the number of false identifications .
this did not impact that acf or the average redshift distribution of the unmatched sample , indicating that the bias measurement at @xmath72 is not sensitive to the choice of matching radius . in the second test , we used the matched sample of @xcite rather than our own matching .
this sample was carefully constructed using both nvss and first and used visual identification rather than an automated `` collapse and match '' approach .
results were consistent with our matched sample , given that their sample probes a somewhat different redshift range to ours . in the third test , we considered the impact of our choice of sidelobe probability cut .
we calculated the acf for several different sidelobe probability samples and found that all samples behaved similarly at large angles .
finally , we investigated the sensitivity of our results to the photometric redshift estimates , by using different sdss photometric redshift catalogues .
we found that the results were robust to the choice of catalogue . .
we have introduced a method for measuring the bias at high redshift for a sample of radio continuum sources lacking redshift information . by matching radio sources from the first survey data to their optical counterparts in the sdss survey
, we extracted a subsample of unmatched objects .
we then used the s@xmath0 simulation to infer an average redshift distribution for all first sources and estimate the redshift distribution of unmatched sources by subtracting the matched distribution from the distribution of all sources .
we have found that the surprisingly large clustering signal at large angular scales present in the full first sample is also detected in the unmatched sample considered here , and to some extent , in the matched samples at high redshift .
we note that this could be due to systematic fluctuations in sensitivity in different observing epochs but also discuss a number of other possible explanations . using
clustering measurements at smaller angles , we estimate the bias of the unmatched first sources with flux densities over 2 mjy , at @xmath68 , to be @xmath79
. the analysis of cross - correlations with other data will be helpful in interpreting these measurements better .
these results can help constrain models of radio source evolution and are important for using radio surveys to constrain cosmological models .
the authors acknowledge the national research foundation and the ska ( south africa ) funding bodies , as well as the south african astronomical observatory ( saao ) and the institute of cosmology and gravitation ( icg ) where some of this work was carried out .
the authors thank the african institute for mathematical sciences ( aims ) for hosting the initial meeting where this project was concevied .
rj and ms would like to thank prina patel , cris sabiu and luis teodoro for valuable discussions . , m. , the 2df galaxy redshift survey team , ellis , r. , bland - hawthorn , j. , cannon , r. , cole , s. , collins , c. , couch , w. , dalton , g. , driver , s. , and _
et al . _ , , in r. m. .
w. j. couch ( ed . ) , _ looking deep in the southern sky _ , e. , dunkley , j. , nolta , m. r. , bennett , c. l. , gold , b. , hinshaw , g. , jarosik , n. , larson , d. , limon , m. , page , l. , spergel , d. n. , halpern , m. , hill , r. s. , kogut , a. , meyer , s. s. , tucker , g. s. , weiland , j. l. , wollack , e. , and wright , e. l. : 2009 , , 330 | the relationship between the clustering of dark matter and that of luminous matter is often described using the bias parameter . here , we provide a new method to probe the bias of intermediate to high - redshift radio continuum sources for which no redshift information is available .
we matched radio sources from the faint images of the radio sky at twenty centimetres ( first ) survey data to their optical counterparts in the sloan digital sky survey ( sdss ) to obtain photometric redshifts for the matched radio sources .
we then use the publicly available semi - empirical simulation of extragalactic radio continuum sources ( s@xmath0 ) to infer the redshift distribution for all first sources and estimate the redshift distribution of unmatched sources by subtracting the matched distribution from the distribution of all sources .
we infer that the majority of unmatched sources are at higher redshifts than the optically matched sources and demonstrate how the angular scales of the angular two - point correlation function can be used to probe different redshift ranges .
we compare the angular clustering of radio sources with that expected for dark matter and estimate the bias of different samples .
[ firstpage ] cosmology : methods : data analysis methods : statistical astronomical bases : miscellaneous galaxies : redshift surveys galaxies : large - scale structure of universe . |
this work is dealing with regularity , which is a property with deep implications in organisms . from the biological point of view regularity
has been related with radial symmetry , and irregularity with bilateral symmetry @xcite .
the heuristic value of radial and bilateral symmetry in biology account for taxonomic issues , however , symmetry as well as disruption symmetry have been an empirical and intuitive approach accounting for structural properties in organisms @xcite . from a mathematical point of view
, the property of regularity of a geometric form has not been formalized . based in previous results by @xcite
, we hypothesize that _ eutacticity _ provides a measure of regularity based in the following argument .
a set of @xmath1 vectors in @xmath2 , with a common origin , is called a star and a star is said to be eutactic if it can be viewed as the projection of @xmath1 orthogonal vectors in @xmath3 .
it turns out that stars associated with regular polygons , polyhedra or , in general , polytopes , are eutactic @xcite and thus regularity and eutacticity are closely linked .
a disadvantage of using eutacticity as a measure of regularity is that a star vector must be associated with the geometrical form under study .
as we shall see , this is not a problem with echinoids .
in fact , @xcite found that the flower - like patterns formed by the five ambulacral petals in 104 specimens of plane irregular echinoids ( from clypeasteroidea ) are eutactic .
here we present a deeper study that overcome the restriction to plane irregular echinoids , using the five ocular plates ( op ) to define the star vector .
additionally , we use a new criterion of eutacticity that provides a measure of the degree of eutacticity of a star which is not strictly eutactic . with these tools
we study the variability of eutacticity during geological time and to analyze pentamery variability during the evolution of sea urchins .
sea urchins are pentameric organisms with an apical structure , called the apical disc @xcite .
this structure includes five ocular plates ( op ) that can fold the vector star associate with each sea urchin species ( see fig .
[ fig : fig1 ] and section [ sec : discoap ] for a detailed description ) . in this work ,
we show that op can be useful even in ovoid echinoids , such as spatangoids , since the op are almost tangential to the aboral surface ( opposite to oral surface ) . using the op to define the star of vectors ,
we analyze the regularity and changes in a macroevolutive and taxonomic level in a collection of 157 extinct and extant sea urchins .
we conclude that evolution has preserved a high degree of regularity and , consequently , that the apical disk is a homogeneous and geometrically stable structure through the geological time .
low values of regularity were recorded in some specific families and its biological consequences are discussed .
this paper is organized as follows . in section [ sec : eutactic ] a mathematical introduction to the concept of eutactic star is presented .
section [ sec : discoap ] describes the structure of the apical disc and its biological importance , making it the obvious choice to define a vector star which characterizes each specimen .
experimental methods and results are devoted to section [ sec : resultados ] and , finally , discussion and conclusions are presented in section [ sec : discusion ] .
our main hypothesis is that the concept of regularity of a biological form may play an important role in the study of phenotipic variation in evolution . for this goal
, one must first be able to establish a formal criterion defining regularity of a geometrical form , including a measure of how regular a form is .
mathematically , this property has not been defined and here , as a first step along this direction , we adopt the concept of eutacticity that , as we shall show , is closely related to regularity
. we shall deal with a set of @xmath1 vectors @xmath4 in @xmath2 , with a common origin , called _
star_. in this case @xmath5 so the set of vector con not be linearly independent .
the star is called _ eutactic _ if its vectors are orthogonal projections of @xmath1 orthogonal vectors in @xmath6 , that is , there exist @xmath1 orthogonal vectors @xmath7 , in @xmath6 , and an orthogonal projector @xmath8 such that @xmath9 the notion of eutacticity ( from the greek _ _ eu__=good and _ _ taxy__=arrangement ) was firstly introduced by the swiss mathematician l. schlfli ( about 1858 ) in the context of regular polytopes .
later , @xcite noticed that the vectors of an eutactic star are projections from an orthogonal basis in higher dimensional spaces and proved that the star associated to a regular polytope is eutactic .
thus , eutacticity is associated with regularity and the remarkable properties of eutactic stars have been useful in different realms such as quantum mechanics , sphere packings , quasicrystals , graph and frame theory and crystal faceting ( see @xcite and references therein ) . a well known necessary and sufficient condition for a star to be eutactic is due to hadwiger himself , who proved that a star @xmath10 in @xmath2 is eutactic if and only if there is a real number @xmath11 such that @xmath12 is fulfilled for all @xmath13 . in the special case
where @xmath14 , the star is said to be _
normalized eutactic_. a more practical form of the eutacticity criterion is obtained if the so called structure matrix @xmath15 is introduced .
let @xmath15 be the matrix whose columns are the components of the vectors @xmath4 , with respect to a given fixed orthonormal basis of @xmath2 . in this case
, the matrix form of hadwiger s theorem sates that the star represented by @xmath15 is eutactic if and only if @xmath16 for some scalar @xmath11 ( here @xmath17 is the @xmath18 unit matrix ) . in this work we are dealing with stars measured in digital images of sea urchins and thus a reliable numerical criterion of eutacticity , suitable to work with experimental measurements , is need .
notice that a criterion such as ( [ eqn : aat ] ) is not useful since experimental errors may produce a matrix which is not exactly the identity matrix @xmath17 .
thus , it is desirable to obtain a numerical criterion capable of measuring the degree of eutacticity of a star which is not strictly eutactic .
this criterion has already been proposed @xcite and asserts that a star in @xmath19 , represented by the structure matrix @xmath15 , is eutactic if and only if @xmath20 where @xmath21 .
notice that the closer @xmath22 is to one , the more eutactic the star is . in the particular case of two - dimensional stars ( @xmath23 )
, it can be proved that @xmath24 @xcite .
in @xcite , vector stars were associated to the petaloid ambulacra of plane irregular echinoids .
it was reported that , for 104 specimens of the natural history museum of london , the pentagonal stars thus defined fulfill very accurately an eutacticity criterion . the calculations carried out in that work present two main restrictions : a ) stars are associated to plane or almost plane sea urchin specimen and b ) a eutacticity criterion was used that depends on the coordinate system and does not allow a measurement of the degree of eutacticity of stars which are not strictly eutactic . here
we overcome these restrictions by using the eutacticity criterion of ( [ eqn : cosfi ] ) and using the five ocular plates ( op ) to associate a star of vectors to each sea urchin .
as we shall see in what follows , besides the biological importance of the op , its use to define a star of vectors allows to study non planar echinoids .
the apical disc in sea urchins , encircled in fig . [ fig : fig1 ] , represents a crown of biological structures in the apex of the test @xcite .
it is positioned in the aboral surface of the test and is conformed by five genital plates , five ocular plates and the madreporite . the op
are located at the point of origin of the ambulacral zones .
the biological relevance of ambulacra is given by the following reasons .
firstly , each ambulacra consist of two , or even more , columns of plates extending from the margin of an op to the edge of the mouth . in most echinoids
each mature plate is perforated by two pores forming a pore pair ; each pore pair gives passage to one tube foot , which is connected internally with the water vascular system .
secondly , five ambulacra are a conspicuous sign of body plan pentamery in all extant and extinct sea urchins and , finally , ambulacral rays are seen as homologous structures in echinoderms .
thus we conclude , that op has more biological implications as the origin of ambulacra than the end of petaloid ambulacra .
in addition , the op are almost tangential to the aboral surface and thus it is useful even for ovoid echinoids .
we then define the star vector associated to a particular echinoid as the set of five vectors pointing to the op with origin at the centroid .
the star thus defined , allows us to test eutacticity in a wide range of echinoids and study changes in a macroevolutive and taxonomic level .
.list of the 47 families studied in this work .
the number of specimens considered in this work is indicated between parenthesis . [ cols="<,<,<,<",options="header " , ]
we have analyzed 157 extant and extinct specimens of sea urchins from the collection of the instituto de ciencias del mar ( universidad nacional autnma de mxico ) and from images of the natural history museum of london web site ( http://www.nhm.ac.uk ) .
as shown in table [ table : table1 ] , the analyzed sea urchins belong to 47 families in a taxonomic group of 95 families , according to the classification by @xcite .
eleven of these specimens are radial and thirty six bilateral .
to each sea urchin we associate a vector star , with vectors pointing to the op ( shown in fig .
[ fig : fig1 ] ) and origin at the centroid .
measurements were carried out on digital images of aboral surfaces , analyzed using the morphometric software packages _
makefan6 _ @xcite and _ tpsdig2 _ @xcite . with the former ,
the op is digitized and the vector star coordinates are obtained by using the second program .
once the coordinates of the star vector are available , eq . ( [ eqn : cosfi ] ) is used to calculate the value of eutacticity of the star , _
i.e. _ , @xmath25 . per oder :
a ) clypeastroida , b ) cassiduoida , c ) holectypoida , d ) pygasteroida , e ) spatangoida , f ) holasteroida , g ) disasteroida .
( b ) standar deviation of @xmath22 per family ( the bilateral families with the largest variations are only displayed . ) : a ) pygasteridae , b ) holectypidae , c ) cassiduloidae , d ) clypeasteridae , e ) arachnoidae , f ) fibulariidae , g ) neolaganidae , h ) rotulidae , i ) echinarachniidae , j ) scutellidae , k ) collyritidae , l ) toxasteridae , m ) micrasteridae , n ) brissidae , o ) hemiasteridae , p ) schizasteridae.,width=377 ] since one of our goals is to analize regularity of sea urchins through geological time , we must define a taxonomic group to define a phylogenetic reference in time .
taxonomic keys use the apical disk as a reference to describe the family level , thus family taxonomic level constitutes the best choice since it should have a low variability in the apical disk .
in fact , as shown in fig . [
fig : fig2 ] , family level shows lower variability in eutactic values as compared to order .
hence , this taxonomic level was used to represent regularity in geological time . by organizing the values of eutacticity per family , we are able to carry out a formal statistical analysis . from the properties of eutacticity , we can deduce that radial families have a high degree of eutacticity ( the stars form regular pentagons ) and , consequently , no variability ( up to experimental errors ) .
contrarily , high variability is expected in bilateral families . before proceeding with a statistical analysis of the eutacticity values per family
, we have to take into account the possibility of a stochastic nature of eutacticity .
in order to reject this possibility , an experimental set of two hundred randomly generated bilateral stars was considered .
a formal statistical analysis must then include three groups : radial , bilateral and random stars .
the values of @xmath26 of the random sample yield a mean of 0.891026 with population standard error of 0.00604 .
our experimental sample of radial stars yield a mean of 0.995187423 and standard deviation of 0.00526708 .
finally , the experimental sample of bilateral stars gives a mean of 0.96499158 and standard deviation of 0.062079301 .
the shapiro - wilk test applied to our samples yields @xmath27 and @xmath28 , for the joined radial and bilateral experimental sample , and @xmath29 and @xmath30 , for the random sample . from this result we conclude that neither the experimental or random distribution are normal and thus a non parametric statistical analysis is needed .
this non parametric test produces @xmath31 and @xmath32 , consequently , the probability of finding random significant differences between radial , bilateral and random stars is lower that 0.0001 .
the possibility of a stochastic origin of regularity is thus rejected .
now , concerning the analysis of the experimental sample per family , in fig .
[ fig : fig3 ] , a scatter plot of the eutacticity values of the 47 families is shown . from the figure
, it is observed that the lowest degree of regularity are recorded in holasteridae , corystidae , collyritidae , pourtoalesiidae , toxasteridae and nucleolitidae .
this observation will be revisited in the next section .
a wilcoxon / kruskal - wallis analysis of the 47 families was carried out , yielding @xmath33 and @xmath34 .
consequently , differences between families are also accepted . here
the variability of eutacticity per family through geological time is studied .
the experimental sample includes extant and extinct specimens .
once again we have to take into account that radial echinoids are associated with nearly eutactic stars ; the most primitive groups , like paleozoic groups , are almost always totally radial .
contrarily , the eutacticity values of species from post - paleozoic groups are less uniform and thus more than two specimens per family are required .
[ fig : fig4 ] shows the mean values of @xmath22 at four geological time intervals , namely paleozoic , triasic - jurasic , cretaceous and cenozoic .
these scales were chosen because , according the paleontological records @xcite , at the beginning of each of these intervals there was a rise in the speciation rate ; there was an increase in the numbers of families . as shown in fig .
[ fig : fig4 ] , post - paleozoic sea urchins show the highest degree of variability . a statistical analysis , however , gives the values @xmath35 and @xmath36 , implying that there are no statistical differences in regularity through geological time .
) of the experimental sample in four intervals of the geological time.,width=377 ] in fig .
[ fig : fig5 ] a plot of the eutacticity values , per family , through the geological time is shown .
the lowest values of eutacticity are recorder firstly in early mesozoic collyritidae and low values continue with late mesozoic and cenozoic holasteridae , portuolesidae and corystidae . as a matter of fact
, all these families belong to the holasteroida order which turns out to be the responsible of the prominent peak ( f ) in the standard deviation plot in fig .
[ fig : fig2](a ) . in order to have a better understanding of the singularity of holasteroida order , in fig .
[ fig : fig6 ] an evolutive cladogram showing phylogenetic relationships between orders is shown . in this cladogram
a representative star , and the mean value of eutacticity per order , is included .
it is clearly shown that the lowest values of regularity comes from disasteroida and are recorded in spatangoida and holasteroida .
in fact , from the measured values , we can say that holasteroida is an anti - eutacticgroup , _
i.e. _ , mathematically irregular .
most living representatives of holasteroida are deep - water inhabitants with exceedingly thin and fragile tests . besides that we consider that regularity and irregularity constitute two important parameters to approach ecological and evolutive topics , the observed departure from regularity could have been a way to increase the amount of complexity in sea urchin morphology . ) per family through the geological time.,width=377 ]
traditionally , radial symmetry has been associated with regularity while bilateral symmetry with irregularity . in this work
we propose the eutacticity as a measure of the regularity of a biological form which is independent of the of the radial or bilateral condition .
with this hypothesis , we have shown that regularity has dominance over irregularity in sea urchins evolution ; despite that variability increases over time , statistically sea urchins show a high degree of regularity .
this regularity is nearly perfect in the most primitive groups , belonging to the paleozoic era , which were almost totally radial .
a slight decreasing of regularity is observed in post - paleozoic sea urchins , with the notably exception of the holasteroida order which seems to constitute a critical evolutive event in sea urchins evolution .
* acknowledgments*. this work was inspired by manuel torres , whose death will not dismiss the memory of his achievements and creativity ; he will be greatly missed by us .
useful suggestions from m.e .
alvarez - buylla , g. cocho , a. laguarda and f. sols are gratefully acknowledged .
this work was financially supported by the mexican conacyt ( grant no .
50368 ) and the spanish mcyt ( grant no .
fis2004 - 03237 ) .
n. holland ( 1988 ) the meaning of developmental asymmetry for echinoderm evolution : a new interpretation , in c.r.c paul and a.b .
smith a.b .
( eds . ) , echinoderm phylogeny and evolutionary biology .
oxford university press , oxford .
pp.13 - 25 .
p. lebrun ( 2000 ) une histoire naturelle des chinides .
2me partie : anatomie , ontogense et dimorphisme , locomotion , palocologie , origine et volution des chinides , minraux & fossiles , hors - srie 10 .
melville and j.w .
durham ( 1966 ) skeletal morphology , in r.c .
moore ( ed . ) treatise on invertebrate paleontology , part u , echinodermata 3 , the geological society of america and university of kansas press , lawrence , kansas . | an eutactic star , in a @xmath0-dimensional space , is a set of @xmath1 vectors which can be viewed as the projection of @xmath1 orthogonal vectors in a @xmath1-dimensional space . by adequately associating a star of vectors to a particular sea urchin
we propose that a measure of the eutacticity of the star constitutes a measure of the regularity of the sea urchin .
then we study changes of regularity ( eutacticity ) in a macroevolutive and taxonomic level of sea urchins belonging to the echinoidea class .
an analysis considering changes through geological time suggests a high degree of regularity in the shape of these organisms through their evolution .
rare deviations from regularity measured in holasteroida order are discussed .
eutactic stars , bilateral symmetry , regularity , sea urchins |
mutually unbiased basis sets are known to provide an optimal basis for quantum tomography @xcite , to play key roles in quantum cryptography @xcite , and to be instrumental in solving the mean king problem in prime power dimensions @xcite .
the generalized pauli operators associated with mub s include the stabilizers of quantum error correcting codes @xcite , and serve as entanglement witnesses @xcite for the mub states . of interest for the foundations of quantum physics , the mub concept sharpens the concept of complementarity @xcite , and raises the question of existence in composite dimensions .
an excellent comprehensive review of mubs has recently appeared @xcite .
we deal here with hilbert spaces of prime power dimensions ( @xmath9 ) , where @xmath10 mubs are known to exist @xcite .
this is both the largest possible number , and also the number required for a complete operator basis ( in representing the density matrix , for example ) .
so , while each mub is a complete orthonormal basis in the hilbert space , the set of @xmath10 mubs is a complete ( nonorthogonal ) basis in the space of all operators , which has dimension @xmath11 . regarding terminology , to avoid reference to a `` complete set of complete sets , '' and prompted by the fact that different mubs ( or the observable sets associated with them ) are maximally complementary @xcite
, i will use the term `` full complement , '' or sometimes just `` complement , '' to denote the set of all @xmath10 mubs .
partial mub sets have been discussed in connection with composite dimensions and referred to as `` constellations '' @xcite .
the natural systems to which mubs apply consist of @xmath1 @xmath4-state objects ( _ qupits _ ) . in such systems , while mub complements exhibit only a single entanglement type for @xmath12 ( and all @xmath4 ) , the number of distinct types proliferates with increasing @xmath1 .
the variety is illustrated in a number of recent discussions , mostly on multiple qubit systems but also multiple qutrit systems @xcite .
in particular , a systematic study by romero and collaborators @xcite illustrates a broad range of entanglement patterns that occur naturally in a construction scheme for full mub complements .
such complements are catalogued for up to 4 qubits .
wieniak and collaborators @xcite have developed a construction scheme aimed at experimental implementation and discussed the total entanglement content of full mub complements of bipartite systems . with the general mub problem in mind ,
our purpose here is to develop a general framework , independent of construction schemes , for exploring mub entanglement patterns for all @xmath4 and @xmath1 .
the project begins by proving three general theorems ( the `` rules '' ) that underlie and lead quickly to an array of more specific results .
many of the latter apply to all @xmath4 , but are @xmath1-specific , as each step in @xmath1 introduces further richness .
all results refer to one of two levels - that of individual mubs and that of full complements . at the individual level
, mub types are characterized by first specifying the _ separation pattern _ - how many , and how big , are the irreducible subsets of qupits defined by the factorization of the wavefunction ? - and next , by describing the _ entanglement pattern _ - what is the nature of the entanglement within each irreducible subset ? at the level of the full complement , we ask about the possible _ mub distributions _ - what combinations of mub types can coexist within full complements . at the first level
, we will show that all conceivable separation patterns are possible , and we will show with examples how to describe the entanglement within the nonseparable factors . at the level of the full complement , we will show how to deduce constraints on the possible mub distributions . for @xmath12 and 3 ,
surprisingly , the general global constraints mentioned in the abstract suffice to determine all mub distributions for all @xmath4 .
the @xmath6 case is considerably more complex and requires the derivation of more detailed constraint equations .
let us begin with a review of basic concepts and notation in section ii . in section iii
we prove the three general theorems .
these rules are applied in section iv to obtain the entanglement patterns of individual mubs , and to deduce constraints on their possible distributions within full complements , taking the @xmath12 - 4 cases in turn . in section
v we summarize results and comment on unresolved questions .
in hilbert spaces of dimension @xmath13 , two orthonormal bases ( @xmath14 and @xmath15 ) are mutually unbiased if any state @xmath16 in basis @xmath14 has uniform probably of being found in any state @xmath17 in basis @xmath15 ; that is , if ||^2 = 1/d .
[ mub ] thus , measurements in the two bases provide no redundant information .
since measurements in any basis provide @xmath18 independent probabilities , and since @xmath19 real parameters are needed to determine an unknown quantum state ( its density matrix @xmath20 ) , it follows that @xmath10 mub s are required . in this way the mub projectors form a complete nonorthogonal basis in operator space
this required number of mubs is ( only ) known to exist in power - of - prime dimensions .
there is an intimate connection between mubs and generalized pauli operators ( hereafter called simply `` pauli operators '' ) which underlies several construction schemes ( provides a comprehensive listing @xcite ) .
these operators are conventionally written in the form of a tensor product , _
n , m = x^nz^m x_1^n_1x_2^n_2 ... x_n^n_nz_1^m_1 ... z_n^m_n , [ pauli1 ] whose factors , acting on individual qupits , are powers of the generalized ( @xmath21 ) pauli matrices , z = _
k=0^p-1 ^k 1truecm 1truecm x = _ k=0^p-1 , [ pauli2 ] where @xmath22 , and @xmath23 is the raising operator of @xmath24 . the powers @xmath25 and @xmath26 are @xmath4-nary numbers , _ eg _ , @xmath27 , whose digits take the values 0,1, ... ,@xmath28 . thus , there are @xmath29 operators @xmath30 ( including the identity @xmath31 ) , which make up a complete and orthonormal basis in operator space ( with the trace operation as inner product ) .
the desired connection with mub s is described in : the @xmath30 partition into @xmath2 internally - commuting subsets , each consisting of @xmath28 traceless operators ( excluding @xmath32 ) .
the corresponding eigenbases then form a complete complement of mub s .
the above are standard definitions and conventions
. it will be useful to adopt a couple of more special conventions for use throughout this paper .
first , the operator set @xmath30 does not form a group , because multiplication generates irreducible phase factors . however , for odd @xmath4 the set @xmath33 does form a group , of order @xmath34 , and for @xmath7 the analogous set @xmath35 forms a group of order @xmath36 .
these are called discrete heisenberg - weyl , or generalized pauli groups @xcite .
we shall not make direct use of them , but we shall take advantage of the freedom to redefine the phases of the @xmath30 in the original set : we choose phases so that the compatible subsets form groups , and we call these _ compatibility groups_. they are all isomorphic to those consisting of @xmath37 and @xmath38 , each of which is generated by the @xmath1 independent elements , @xmath39 and @xmath40 , respectively .
thus , to construct another compatibility group , we may choose a generator set @xmath41 that consists of any @xmath1 elements in the original compatible subset that do not form a subgroup , and write the resulting group elements as g^n g_1^n_1g_2^n_2 ... g_n^n_n .
[ ggroup ] thus , all of the compatibility groups are representations of the same group - the abelian group of order @xmath42 generated by @xmath1 elements .
a simple example of a compatibility group so generated is y^n = ( x_1z_1)^n_1(x_2z_2)^n_2 ... (x_nz_n)^n_n .
[ ygroup ] note that phase factors are introduced with respect to the original pauli operators because , , @xmath43 .
the generator set @xmath44 , by itself , completely determines the states of the basis ( @xmath44 ) in the hilbert space , through the eigenvalue equations @xmath45 , where @xmath46 is a @xmath4-nary representation of the state index @xmath47 .
the eigenvalues of a general group element are then given by g^n = ^n k , [ eigenvalues ] where @xmath48 , and the spectral representation of @xmath49 is therefore just the fourier transform @xcite g^n = _ k ^n k _ k ^n k ( g , k ) , [ specrep ] where @xmath50 is the projector onto state @xmath47 in basis @xmath44 .
this mub projector is then given by the inverse transform , ( g , k ) = p^-n _ n ^-n k g^n .
[ inverse ] the existence of these simple transform relationships between every compatibility group and its corresponding mub projector set is a consequence of defining the former to be a group .
the only remaining arbitrary phases are those of the generators .
in this section we establish the general rules that will form the basis for the rest of the work . for ease of reference and completeness i will state and prove these results as three separate numbered theorems . for transparency , here , in plain english , is what they will say about mub states : ( i ) a given qupit is perfectly pure or totally entangled , ( ii ) the distribution of one - qupit operator factors in the compatibility group correlates with this purity , ... , and ( iii ) in any full mub complement , every qupit appears pure @xmath2 times , and totally entangled @xmath3 times .
these theorems and the results that follow from them rely on the assumption that mub states are eigenstates of pauli operators . while this is restrictive for individual mub pairs , it is not restrictive for known mub complements or known construction schemes @xcite , allowing for unitary equivalence .
an example may help to illustrate .
consider the standard basis in 4d , and another related to it by the unitary transformation @xmath51 ( where @xmath52 0,1,2,3 ) , which is not an eigenbasis of the pauli operators of .
the two bases are mu , but a full complement can not be completed containing both of them
. however , full complements can be found containing either basis without the other : starting with the well - known full complement containing the standard basis , one could apply @xmath53 to each of its bases to obtain another full complement .
the latter are not eigenbases of the original pauli operators , but clearly they _ are _ eigenbases of transformed pauli operators , which may be thought of as corresponding to redefined parts ( and redefined quantization axes ) .
the results of this paper then apply with reference to these redefined parts . regarding the existence of a mub complement outside of this
equivalence - i believe that this question also remains unresolved @xcite .
we will return to these points in the conclusions . as a brief preliminary , one - qupit states within the @xmath1-qupit system are defined by the reduced density matrices , _
i = , [ rhoi ] where tr@xmath54 denotes the partial trace over states of all but the @xmath55-th qupit .
perfect purity means that @xmath56 is a projector , while total impurity means that @xmath57 .
one can define the purity of the state @xmath58 as p_i = ( ptr _ i^2 - 1)/(p-1 ) , [ purity ] which takes its extremal values , 1 and 0 , in the respective cases .
* theorem i : * if the system is in a pure eigenstate of pauli operators ( a generator set @xmath44 ) , then any individual qupit must exist in a state of either perfect purity , or total impurity , the same for all eigenstates of @xmath44 . *
proof : * the generators produce a compatibility group , and the @xmath1-qupit density matrix representing a pure eigenstate , @xmath59 , may be expanded as in .
considering now the pauli matrix factors that act on just the @xmath55th qupit , the generator set @xmath44 must fall into one of two categories : only one pauli matrix , say @xmath60 ( and possibly powers of it ) , appears in the generator set , or more than one appear ( including , say , @xmath61 and @xmath62 ) , that are not powers of one another .
consider the latter case , which is simpler : let @xmath63 and @xmath64 be generators that contain the factors @xmath61 and @xmath62 .
no operator of the form @xmath65 ( where @xmath66 is any one - qupit pauli matrix ) commutes with both @xmath63 and @xmath64 , and all such operators are thereby excluded from the compatibility group . as a result , the only operator with a nonvanishing partial trace tr@xmath54 is the global identity @xmath32 .
since @xmath32 enters the summation ( [ inverse ] ) with the coefficient @xmath67 , and tr@xmath54 produces a factor of @xmath68 , the reduced density matrix for the @xmath55th qupit is _
i = p^-1 _ i , [ t1 ] indicating that the @xmath55th qupit is totally impure .
now turn to the other case : if only @xmath60 ( and possibly powers ) appear in the generator set , then only @xmath60 and its powers can appear in the compatibility group ( again refering only to those factors that act on the @xmath55th qupit .
since the `` one - body '' operators @xmath69 commute with all of these , they must belong to the compatibility group .
these one - body operators are the only ones that survive the partial trace . since each of them enters the summation ( ) with coefficient @xmath70 , and
since @xmath71 produces a factor of @xmath68 in each term , we find in this case that _
i = p^-1 _ n ^- n_ik_i z_i^n_i = . [ t2 ] this shows that @xmath58 is a projector onto the eigenstate of @xmath60 whose eigenvalue is @xmath72 , that is , _
i^2 = _ i 1.2truecm 1.2truecm z_i _ i = ^k_i _ i. [ t3 ] this proof is independent of the choice of the eigenstate @xmath73 ) in the basis @xmath44 , and so clearly the @xmath55th qubit is perfectly pure for all eigenstates in this basis .
here is a related more detailed theorem on the distribution of one - qupit matrices associated with a single qupit .
* theorem ii : * in any compatibility group of @xmath1-qupit pauli operators , the distribution of one - qupit factors acting on the @xmath55th qupit must be one of two types : ( i ) only a single pauli matrix and its powers occur , and each power occurs an equal number ( @xmath74 ) of times , or ( ii ) every pauli matrix occurs , and each occurs an equal number ( @xmath75 ) of times . *
proof : * consider any set @xmath44 of @xmath1 generators of the compatibility group .
this set must be one of the two types considered in the foregoing proof : suppose first that only one pauli matrix ( say @xmath60 ) , and possibly powers of @xmath60 appear .
let @xmath63 be a generator containing @xmath60 as a factor , and let @xmath64 , @xmath76 , ...
, @xmath77 be the rest .
@xmath63 by itself generates a cyclic subgroup containing all powers of @xmath60 .
then , @xmath63 and @xmath64 by themselves generate a subgroup of order @xmath78 in which , by virtue of the rearrangement theorem , every power of @xmath60 appears @xmath4 times ( no matter which power of @xmath60 is present in @xmath64 ) .
one may repeat this argument , multiplying the order of the subgroup by @xmath4 at each stage , until the full compatibility group is generated , with each power of @xmath60 being produced @xmath74 times .
in the other case , let @xmath63 and @xmath64 be generators containing the @xmath61 and @xmath62 factors , respectively .
these two generators , by themselves , generate a subgroup of order @xmath78 in which every pauli matrix factor @xmath66 appears once and only once .
( to see this , note that @xmath61 and @xmath62 , by themselves , generate the one - qupit pauli group @xcite , but since @xmath63 and @xmath64 commute , the multiplicity of phase factors is absent . )
now , by including a third generator , @xmath76 , one generates a subgroup of order @xmath79 in which , by the rearrangement theorem , each pauli matrix factor appears @xmath4 times . repeating the process through @xmath77 , one generates the full compatibility group with each pauli matrix factor appearing @xmath75 times .
the second result is particularly striking in light of the fact that the nature of the entanglement of the @xmath55th qupit may vary widely , in the sense that its entanglement may be shared with any number of other qupits in the system
. nevertheless , only two kinds of pauli matrix distributions , with the correponding purities , are possible .
we use both of the foregoing theorems to deduce the total entanglement content - as measured by the one - qupit purities - of a full complement of mub s . this total content
is constrained by the requirement that the two types of one - qupit pauli matrix distributions be consisent with the set of all pauli operators , which must appear in the full complement .
* theorem iii : * within any full complement of @xmath0 mub s , every qupit is perfectly pure in @xmath2 basis sets , and totally entangled in the remaining @xmath3 . *
proof : * consider the @xmath55th qupit . recall that the total number of pauli operators ( excluding @xmath32 ) is @xmath80 , and that these exactly accommodate the @xmath0 compatibility groups containing @xmath81 traceless operators each .
each pauli matrix factor @xmath66 appears in @xmath82 pauli operators , except for @xmath83 which appears in @xmath84 because we are not counting @xmath32 in the individual groups .
this number must equal the sum of @xmath83 factors appearing in all of the compatibility groups . according to the previous theorem , there are @xmath85 such factors in compatibility groups in which the @xmath55th qupit is pure , and @xmath86 such factors in all other compatibility groups .
if @xmath87 is the number of compatibility groups ( or basis sets ) in which it is pure , then , in order to account for all @xmath83 factors , we must have p^2n-2 -1 = _ s^ ( p^n-1 -1 ) + ( p^n+1-_s^)(p^n-2 -1 ) . [ accounting ] solving this equation , we find the number of basis sets in which the @xmath55th qupit is pure , _
s^ = p+1 , [ nusubs ] and consequently , the number of basis sets in which it is totally entangled , _
e^ = p^n - p. [ nusube ] the following corollary arises when all qupits take their pure states simultaneously : * corollary * : the maximum number of product mubs is @xmath2 , and in any mub complement where this number is realized , all of the remaining mubs ( @xmath3 ) must be totally entangled ( in the sense that every qupit is totally entangled ) @xcite .
this is the standard distribution .
note that the probability of finding the @xmath55th qupit pure in a mub state picked at random from any full complement is equal to the averaged purity ( ) , p_i _
comp^ = _ s^ _ s^+_e^ = p+1 p^n+1 , [ piave ] which vanishes exponentially with @xmath1 .
we discuss the @xmath88 cases in turn . the first two are simpler , and we find that theorems i and iii are sufficient to determine all possible mub distributions , although ii provides useful insights . with @xmath6 , we require theorem ii in deriving more detailed constraints that apply to individual qupits . * bipartite systems * clearly , if one qupit is pure , then so must be the other . in light of theorem
i , then , both purities must be unity , or both zero . because these purities coincide
, the corollary of theorem iii applies : there are @xmath2 product bases and @xmath89 totally entangled bases - the standard distribution is inevitable .
we shall refer to all of the entangled bases as generalized bell bases , because they share the common property that their compatibility groups consist solely of two - body operators , , those containing no @xmath90 factors @xcite . to see the consequences of this , write one of the two generators as @xmath91 .
the most general eigenstates of @xmath63 may then be written as @xmath4-term expansions in the product basis of @xmath92 and @xmath93 , = 1 _ k c_k _ u _ v , [ bform ] where the eigenvalues of @xmath94 are @xmath95 and the coefficients @xmath96 are determined by the other generator , call it @xmath97 .
commutativity demands that both @xmath98 and @xmath99 , so @xmath64 induces cyclic permutations ( of order @xmath4 ) in the product states @xmath100 .
therefore the @xmath96 are unimodular , and the @xmath4 eigenvalues ( @xmath101 ) of @xmath64 are nondegenerate , like those of @xmath63 .
this confirms explicitly what we know from theorem i - namely , that measurements of one - qupit properties ( , @xmath92 or @xmath93 ) must produce random distributions over all possible outcomes .
the generalized bell states defined above are contained within a broader class definitions given elsewhere @xcite . the more restrictive definition given here - defining classes of states by the pauli operators of which they form eigenbases - applies nonetheless to all mubs that are compatible with known full complements , and
we shall employ such definitions throughout this work as we proceed to larger n. we note for future reference that the precise form of the product state expansion ( [ bform ] ) depends on the choice of basis .
a bad choice would require a @xmath78-term expansion , but even a good choice could look slightly different .
for example , if eigenstates of @xmath102 were expanded in the same product basis used in , one would find sums of @xmath103 . as a final note on bell states
, our working definition may be given in words alone : a generalized bell state is any totally entangled two - qupit eigenstate of pauli operators ( since total entanglement requires that the two pauli operators be of the form @xmath94 and @xmath104 ) .
* tripartite systems * the standard mub complement has @xmath2 product bases and @xmath105 totally entangled bases .
we shall refer to all of _ these _ totally entangled bases as generalized ghz , or @xmath44-bases , because they have common properties describable as follows : let us first illustrate with a specific example that generalizes a standard choice of generators for qubits @xcite , g ( g_1 , g_2 , g_3 ) = ( xxy , xyx , yxx ) , [ ghz1 ] to arbitrary @xmath4 . to identify an optimal product basis for an expansion , replace the latter two generators by @xmath106 and @xmath107 . recalling the usual definition @xmath108 on the @xmath55th qupit ( modulo possible phase factors ) , the result is g = ( xxy , izz^-1 , ziz^-1 ) .
[ ghz2 ] clearly the most general joint eigenstates of @xmath109 and @xmath110 are @xmath4-term expansions in the standard basis , = 1 _ k c_k , [ ghz3 ] where @xmath101 and @xmath95 are the eigenvalues of @xmath109 and @xmath110 , respectively , and the @xmath96 are determined by @xmath63 .
the @xmath96 are again unimodular because @xmath63 generates a cyclic group of order @xmath4 , of which the @xmath4 product states form a basis .
this again illustrates the randomness of one - qupit properties in totally entangled states . to demonstrate the commonality of all totally entangled three - qupit bases
, we note that at least one generator must be a three - body operator ( having no @xmath90 factors ) , which we write in complete generality as @xmath111 .
now , according to theorem ii , the inverse of each factor occurs @xmath4 times in the compatibility group , once with the inverse of @xmath63 itself , and @xmath28 times in other three - body operators in which it is the only inverse ( footnote @xcite ) .
choosing two from the latter category , one containing @xmath112 and the other containing @xmath113 , and multiplying @xmath63 by each in turn , we obtain the generator set g = ( uvw , ibc , aic ) , [ ghz4 ] where compatibility requires that @xmath114 is common to @xmath64 and @xmath76 as indicated . clearly , @xmath115 , @xmath116 , and @xmath114 define the product basis for the @xmath4-term expansions , = 1 _ k c_k _ a _ b _ c , [ ghz5 ] and each of @xmath115 , @xmath116 , and @xmath114 must differ from corresponding factors that appear in three - body operators of the compatibility group .
in other words , every three - body operator in the compatibility group induces cyclic permutations of the states composing the product basis .
the similarity of generator sets shows that all totally entangled three - qupit bases have @xmath4-term expansions in some special product basis , and that all of their compatibility groups ( of the same @xmath1 and @xmath4 ) have the same numbers of three - body and two - body operators .
again , a purely verbal definition is possible : a generalized ghz state is any totally entangled 3-qupit eigenstate of pauli operators .
a general statement for @xmath117 is possible but less categorical .
the new aspect of mubs that enters with @xmath118 is the appearance of a third ( nonstandard ) mub type , and with it , the possibility of composing a full complement with varying combinations .
the third type is biseparable , and thereby nonsymmetric with respect to qupits - one qupit separates , leaving the other two in a bell state .
we shall refer to these as `` separable - bell '' bases , with the shorthand notation @xmath119 ( or @xmath120 if we wish to identify the pure qupit ) .
such mub bases are known for @xmath7 and 3 ( ) , and to describe them for arbitrary @xmath4 , we consider a generator set s_1b = ( iia , uvi , sti ) , [ sb1 ] where @xmath94 and @xmath104 are commuting two - body operators acting on qupits 2 and 3 .
the @xmath79 joint eigenstates of this set may be written as = , [ sb2 ] which describes qupit 1 in the @xmath47th eigenstate of @xmath115 , and qupits 2 and 3 in the bell state denoted by the eigenvalues @xmath121 and @xmath4 of @xmath94 and @xmath104 , respectively .
similarly , the compatibility group of @xmath122 is a tensor product of that associated with qupit 1 ( @xmath123 ) and that of the bell basis of qupits 2 and 3 .
the tensor product is a common characteristic of all separable mubs , and the eigenstates of a particular mub all have the same character - the separation pattern involves the same entangled subsets of qupits , and the nature of the entanglement within each subset is the same .
the three mub types discussed above , including the three variations of the sb bases , exhaust all of the possibilities for three qupits .
the remaining question now is , what combinations the three types of bases may appear in the full complement ?
one can answer this question simply by conserving the number of pure qupits while conserving the number of basis sets .
we then find that we can remove a single product basis ( @xmath124 ) while adding three @xmath119 and removing two @xmath44 bases : + 2 g 3sb [ stoich1 ] table i shows the possibilities for three particles with any @xmath4 .
the cases of @xmath125 and 3 dramatize the role of totally entangled states with increasing dimension of the hilbert space .
in fact , case ( a ) , dimension @xmath126 , is the only multiparticle mub dimension in which a complement can be found with no totally entangled bases . and
more typically , a majority of mubs are totally entangled : in case ( b ) at least 4/7 of all bases are @xmath44 bases , and even for _ two _ qutrits , 6 of the 10 bases are bell bases . for @xmath118 and general @xmath4 ,
the minimum number of @xmath44 bases is given by @xmath127 , an ever - increasing fraction of the total number of bases as @xmath4 increases .
@xmath128 @xmath129
it is noteworthy that @xmath119 bases can be introduced only in steps of three , reflecting the condition that the three variations @xmath130 must balance in the full complement , since the other mub types are symmetric with respect to permutations of qupits .
this condition follows from the conservation of pure states for each qupit separately .
* quadrapartite systems * the @xmath6 case is more complex in a number of respects . most importantly ,
new mub types enter with increasing @xmath4 .
but even with @xmath7 , the number of distinct mub types exceeds the number of separation patterns .
figure 1 shows the five separation patterns that characterize all @xmath4 , and lists seven mub types , six of which account for all @xmath7 options , and a seventh which represents , but is not exhaustive for @xmath131 .
let us first discuss the mub types for general @xmath4 , and later specialize to particular cases for constraints and stoichiometries .
the separable mubs compatibility groups are tensor products of those of their constituent mubs , and their generator sets may be constructed accordingly . with @xmath132 ,
a single generator is associated with the separating particle ( for example @xmath133 ) , while three generators ( each of the form @xmath134 or their alternatives ) are associated with the three particles forming @xmath135 states .
there are four variations on this pattern corresponding to the choices of the separating particle . in the @xmath136 case
, one could pick two generators of the form @xmath137 , and two of the form @xmath138 .
there are three variations on this separation pattern , corresponding on the three ways of picking the two entangled pairs , as compared with six variations on the @xmath139 pattern from the six ways of picking a single entangled pair .
let us discuss the nonseparable bases in somewhat more detail beginning with four - particle ghz bases ( @xmath140 ) .
these are straightforward generalizations of the three - particle bases @xmath135 , and a standard generator set @xcite consists of the four operators g^(4 ) = ( xxxy , xxyx , xyxx , yxxx ) .
[ geng ] from an alternative generator set , ( @xmath141 ) , it is apparent that eigenstates may again be written as superpositions of @xmath4 product states in the standard basis .
a more general characterization of ghz states is provided in the appendix .
cluster bases ( @xmath142 ) were introduced in connection with measurement - based , one - way quantum computation @xcite , and in fact both cluster and ghz states are special cases of a broad class of @xmath1-qubit states , called graph states , which form the basis of this @xcite .
has shown that graph states may be classified in terms of curves in phase space , which provides a further connection with the mub problem .
cluster bases are defined here , for all @xmath4 , by generator sets of which a standard example , introduced for the qubit case @xcite , is c^(4 ) = ( xzxi , zxix , xixz , ixzx ) .
[ genc ] cluster states have stronger entanglement links between smaller groupings of particles , making their entanglement more robust against decoherence @xcite than ghz entanglement , which is shared equally among all particles .
this is reflected in the fact that @xmath143 has only two 2-body operators in its compatibility group , as compared with three in the @xmath144 case .
for this reason , its generator set can only be simplified to ( @xmath145 ) , and as a result , the eigenstate expansions can be reduced to no less than @xmath78 terms in the standard basis .
a general characterization of @xmath142 accompanies that of @xmath144 in the appendix , which then goes on to show that these , together with the four separable bases , exhaust all mub possibilities for four qubits . as a final example
, i have found that a new type of basis , one that has no counterpart for qubits , is necessary for the existence of full mub complements when @xmath146 , for reasons that will become apparent .
a generator set giving rise to such a basis is p^(4 ) = ( zxyw , xzwy , wyxz , ywzx ) , [ genp ] where standard definitions @xmath147 and @xmath148 are followed .
the essential point is that the generators are tensor products of four noncommuting one - body matrices , which rules out qubits , but makes possible the elimination of 2-body operators from the compatibility groups for @xmath149 .
less essential is that the four generators are related by pairwise permutations of operators ( hence the notation @xmath150 ) .
the eigenstates have bell correlations between all pairs of particles , not just the chosen pairs as in @xmath136 states .
so , unlike cluster or @xmath136 states , the entanglement is shared equally among all four particles , but unlike ghz states , the entanglement is robust .
one can perform measurements on any two particles , in any two different bases , and produce a bell state of the other two . to show that the @xmath150 basis does not exhaust the possibilities for @xmath149 , we mention another generator set involving cyclic permutations , ( @xmath151 ) , where the @xmath152 factors are inserted for compatibility
. the corresponding basis could play a role similar to that of @xmath150 in filling mub complements for @xmath153 , although it turns out to be relatively inconsequential when @xmath154 . in any case , since it would needlessly complicate the discussion without changing our conclusions , we exclude this example from the analysis .
it is interesting to note in passing , that despite the differences in appearance among the generator sets of the four ( five ) totally entangled bases , the total numbers of @xmath90 factors appearing in their compatibility groups must be the same , namely 4(@xmath155 ) , in accordance with theorem ii .
this has consequences for stoichiometry , in particular for the standard distributions , and justifies classifying @xmath136 bases as totally entangled .
let us now turn to questions of stoichiometry . while in previous cases we were able to deduce the allowed entanglement patterns from global constraints alone ( those involving total numbers of pure and entangled qupits in mub complements ) , with @xmath156 this is no longer the case .
the existence of multiple totally entangled basis types requires that we consider more microscopic constraints associated with the distributions of @xmath90 factors , as was done for qubits in . to this end
, we define a quantity that is capable of distinguishing among all mub types under consideration .
the `` @xmath25-body profile '' of a particular mub is the distribution of @xmath25-body operators ( @xmath157 , 2 , ... , @xmath1 ) in its compatibility group , where ( as implied earlier ) @xmath25-body operators are those with @xmath158 identity factors , @xmath90 .
this distribution is normalized to the total number of operators in the compatibility group , @xmath81 .
examples of @xmath25-body profiles are given in table ii , where we include the @xmath12 and 3 cases both for comparison with @xmath6 , and also to show how global information is recovered .
the number of operators in each category , summed over all mubs , must equal the numbers listed at the bottom of each column .
the latter represent the @xmath25-body profile of the set of all pauli operators , and are thus independent of the particular mub choices .
they are determined by generating all of the pauli operators as expansions in the tensor products , ( i_1,z_1,x_1, ... ,x_1z_1^(p-1 ) ) ...
( i_n , z_n , x_n, ... ,x_nz_n^(p-1 ) ) .
[ pauliik ] thus , the total number of @xmath25-body operators is @xmath159 , as shown .
the condition that the mub sums equal these bottom lines , column by column , provides ( @xmath160 ) independent constraint equations .
( in the exceptional case of @xmath12 , both equations are independent . )
@xmath161 @xmath162 it is immediately apparent from all of the first columns that the maximum number of @xmath124 bases is always given by @xmath2 , the number that defines the standard mub complement .
part ( a ) confirms that this is the only choice for @xmath12 , and its second column then determines the number of bell bases ( @xmath89 ) , in accordance with the required total number of mubs .
part ( b ) reproduces all @xmath118 results , as were summarized on table i. the three columns provide three equations , but only two are linearly independent : the first column determines all possible combinations of @xmath124 and @xmath119 bases , and the second column then determines the number of @xmath135 bases , which is again consistent with the required total number of mubs , @xmath163 .
the third column provides no further constraint . proceeding to the case of @xmath6 , the calculation of the @xmath25-body profiles for the separable bases is straightforward , since their compatibility groups are tensor products of those whose profiles
have already been calculated .
the new nonseparable bases require more thought .
we found that the more symmetrical generator sets listed in eqs .
[ geng]-[genp ] were helpful in working out the profiles for general @xmath4 .
one can see by inspection of table ii(c ) that there is a qualitative difference between @xmath6 and the other cases . consider just the first 6 mub types , which represent all possibilities for @xmath7 .
looking at the 3-body factors in column ( iii ) , we can see that as @xmath4 increases , the number of @xmath144 and/or @xmath143 mubs would have to increase as @xmath164 in order to satisfy just eq .
but then they could not satisfy eq .
( ii ) , for they would produce too many two - body operators .
clearly , one eventually needs a basis which , like @xmath150 , has no two - body operators .
this need makes itself felt already with @xmath165 , and becomes urgent with @xmath166 . with these differences in mind ,
let us consider the @xmath7 , 3 and 5 cases sequentially , to show how the general picture evolves with increasing @xmath4 . *
four qubits * @xmath167 the @xmath25-body profiles for @xmath7 are shown on table iii . to explore stoichiometries ,
consider the three equations ( i , ii , and iii ) represented by the first three columns , respectively .
equation ( i ) , by itself , determines all possible combinations of the first three mub types , 4n()+2n(s^2b)+n(sg^(3 ) ) = 12 .
[ n4p21 ] next , notice that we can isolate the @xmath136 and @xmath140 mubs because of their simple profiles . indeed , by simply adding ( i ) and ( iii ) we obtain the sum of all _ other _ mubs , n ( ) + n(s^2b ) + n(sg^(3 ) ) + n(c^(4 ) ) = 15 .
[ n4p22 ] since there are 17 mubs in total we know immediately that n(bb ) + n(g^(4 ) ) = 2 , [ n4p23 ] a result which also follows from 2(ii ) + ( iii ) - ( i ) , which reproduces the total number .
there are 16 ways to satisfy , with @xmath168 ) determined in each case by . for each of these combinations
, there are 3 ways to satisfy , for a total of 48 possible mub distributions .
to illustrate the range , a standard and a nonstandard distribution are shown in the two leftmost columns of table iv .
these examples are chosen to show the maximum and minimum numbers of the new ( nonseparable ) @xmath143 mubs , which make up a majority of mubs in 30 of the 48 possible distributions .
the dominance of @xmath143 mubs is related to the large number of 3-body operators in their profile .
similar complements were found in through an explicit construction , except that the @xmath140 mubs were not produced , so that 16 combinations were obtained with 2 @xmath136 mubs present in all of them .
@xmath169 * four qutrits * the @xmath25-body profiles for the @xmath165 case are shown in table v. again , the first column restricts the combinations of the first three mub types , 4n ( ) + 2n(s^2b ) + n(sg^(3 ) ) = 16 .
[ n4p31 ] the first and second columns together [ ( ii)@xmath1703(i ) ] restrict other combinations , 4n(bb ) + 3n(g^(4 ) ) + n(c^(4 ) ) = 72 , [ n4p32 ] and the inclusion of the third column [ ( iii)@xmath1712(ii)@xmath1706(i ) ] yields the total mub count , n ( ) + ... + n(p^(4 ) ) = 82 .
[ n4p33 ] @xmath172 there are 25 combinations of the first three mub types that satisfy , as compared with 16 such combinations in the qubit case .
but , in the absence of @xmath150 mubs , one can not solve both for all of these combinations , and we find a total of only 11 mub distributions . to trace the reasons , we subtract from [ n4p33 ] and solve for @xmath150 : n(p^(4 ) ) = 10 + 3n(bb ) + 2n(g^(4 ) ) -
[ n ( ) + n(ssb ) + n(sg^(3 ) ) ] .
[ n4p34 ] without @xmath150 mubs the left side vanishes , and there can be solutions only if the quantity in square brackets is 10 or larger .
this condition fails for the standard distribution , for which this number is @xmath173 . in this case , the minimum number of @xmath150 mubs is 6 , as shown on table iv .
the other entry maximizes the quantity in square brackets at @xmath174 . in both entries
we then minimize the number of @xmath150 mubs by maximizing the number of @xmath143 mubs .
one can increase the number of @xmath150 mubs over these minima by adding @xmath136 and/or @xmath144 and subtracting @xmath143 mubs .
although its numbers can be small , the @xmath150 mubs play a critical role in maintaining the balance of 3-body operators ( table v ) at no cost in two - body operators . with @xmath150 mubs included , the multiplicity of each of the 25 solutions of is large ( we estimate more than 200 ) , for a total of probably more than 5000 solutions .
we can not argue that all of these solutions represent realizable mub distributions , because we can not rule out the possibility of more subtle constraints .
such concerns are beyond the scope of the present paper . *
four ququints * again consulting table v for the @xmath166 case , it is striking to see how three simple equations can again emerge from appropriate combinations .
the first column gives us directly 4n ( ) + 2n(s^2b ) + n(sg^(3 ) ) = 24 , [ n4p51 ] the combination [ ( iii)@xmath1718(ii)@xmath17064(i ) ] relates the other four quantities , 8n(bb ) + 5n(g^(4 ) ) + 3n(c^(4 ) ) + 2n(4 ) = 1600 , [ n4p52 ] and still another combination [ ( iii)@xmath1712(ii)@xmath17022(i ) ] yields the total mub count , n ( ) + ... + n(p^(4 ) ) = 626 .
[ n4p53 ] there are 49 combinations of the first three mub types that satisfy , but in the absence of the @xmath150 mubs , _ none _ of these admits solutions of .
to see how this situation arises , solve the latter two equations for @xmath175 while eliminating @xmath176 : n(p^(4 ) ) = 278 + 5n(bb ) + 2n(g^(4 ) ) - 3[n()+n(sg^(3))+n(ssb ) ] .
[ n4p54 ] the quantity in square brackets has minimum and maximum values of 6 ( the standard distribution ) and 24 , as shown on table v , corresponding to lower bounds on @xmath175 of 260 and 206 , respectively .
the latter is the absolute minimum number of @xmath150 mubs in any full complement .
again , one can add @xmath150 mubs by removing @xmath143 and adding @xmath136 and/or @xmath144 mubs , so that @xmath150 can be the majority mub type in some complements . while @xmath150 is critical for both @xmath165 and 5 cases
, it plays a considerably more dominant role here .
the underlying reason is that the ratio of the numbers of 3-body to 2-body operators increases considerably in going from @xmath165 to 5 , as shown in table v. we estimate the total number of solutions of eqs .
[ n4p51]-[n4p53 ] to be in excess of @xmath177 , but again , we can not argue that all such solutions represent realizable mub distributions , or provide a revised estimate , without a further study of possible constraints .
the examples of this section have shown us that with every step in @xmath1 , and with some steps in @xmath4 , full complements require not only those mub types generated from smaller systems , but also new , nonseparable mub types that exhibit new entanglement characteristics inaccessible to smaller systems . in the step from @xmath12 to 3 , @xmath178 mubs are required for the standard distribution , although a nonstandard distribution ( @xmath119 only ) is possible with @xmath7 . with the step to @xmath165 ,
no mub complement exists without @xmath178 . in the step to @xmath6 ,
the @xmath143 mubs are indespensible to all mub distributions with @xmath7 . with the step to @xmath165 ,
the new @xmath150 mubs become possible , and they in turn make possible the standard distribution . at @xmath166 , the @xmath150 mubs become indispensable to all distributions . projecting to larger systems ,
the distinguishing feature of the @xmath150 generator set is that a different ( noncommuting ) pauli matrix factor is associated with each qupit .
the number of such factors in general is @xmath2 , and when this is equal to the number of qupits , a new type of entanglement becomes possible .
thus we predict that when @xmath1 is equal to any prime plus 1 , then that prime ( @xmath179 ) is a critical value for the emergence of new entangled states as @xmath4 is increased at fixed n. these states should play critical roles in filling mub complements for @xmath4 equal to or slightly greater than @xmath180 .
we have exploited the connections between mubs and pauli operators to develop a general framework for investigating both the entanglement properties of individual mubs , and the combinations of such mubs that can be found in full complements .
we began by proving general theorems regarding mubs as eigenbases of pauli operators : we showed that the purities of individual qupits in such eigenbases must be either 0 or 1 , that the purity alone dictates the distribution of pauli matrix factors ( including @xmath90 ) in the compatibility groups of these mubs , and that every qupit must adopt these special purities the same number of times within any mub complement : ( @xmath2 ) times pure , and ( @xmath3 ) times totally entangled .
an immediate corollary is that one may have at most @xmath2 product mubs in a full complement , and when one does , all remaining mubs must be totally entangled .
this defines the standard distribution .
armed with these theorems and the general properties of pauli operators , one quickly obtains more specific results : when @xmath12 , only product and generalized bell bases are possible , for any @xmath4 , and the standard mub distribution is inevitable . with @xmath118 ,
the unique totally entangled bases are generalized ghz bases , but a third mub type becomes possible , namely separable - bell bases .
this makes possible @xmath5 distinct mub distributions . with @xmath6 and @xmath7
there are six mub types , including two nonseparable bases and a third ( @xmath136 ) that is separable but totally entangled .
there are 48 possible mub distributions , with cluster bases making up the majority of mubs in most of these . with @xmath6 and larger @xmath4 ,
further mub types exist , and at least one such mub type ( @xmath150 ) is essential to forming a standard mub complement with @xmath165 , and to forming _ any _ mub complement with @xmath166 .
several results have emerged in the course of working the above examples , and it seems useful to synthesize these in one place : ( 1 ) a mub can exist in any separation pattern .
( 2 ) all states in a particular mub have common separation and entanglement patterns - the generator set contains all information about the nature of the entanglement , while the eigenvalues specify the states .
( 3 ) compatibility groups of separable bases are tensor products of those of the nonseparable constituent bases .
it follows that ( 4 ) within nonseparable groupings of qupits , those with two qupits must be in generalized bell states , those with three qupits - generalized ghz states .
those with 4 qupits have the same broader array of options available to 4-qupit systems .
perhaps the most important lesson to be drawn from the present examples is that , although it is easy to construct mubs from those found at lower @xmath1 , either as tensor products , or as larger-@xmath1 counterparts such as @xmath181 , the more interesting challenge is to find the new nonseparable mub types , with no counterparts at smaller @xmath1 ( or sometimes @xmath4 ) , that make full complements possible .
it may be a general feature that such mubs tend to dominate mub distributions near the @xmath1 and @xmath4 values where they first emerge , only to be superceeded by other mub types as the system size increases . in this sense , every @xmath1 is critical , but not every @xmath4 .
we predict that when @xmath1 is a prime plus 1 ( , @xmath182 , 8 , ... ) , there will be a critical value , @xmath183 , for the introduction of new entangled states that will play critical roles in mub distributions . closing thoughts on the existence question : the intimate connection between mubs and entanglement for @xmath184 highlights the way in which all known mub complements take advantage of the symmetry associated with equivalent parts , unique to dimensions @xmath42 @xcite .
theorem i , which makes no reference to dimension , can not hold in ( at least some ) composite dimensions : in the simplest counterexample , 6 dimensions , the qubit can be totally entangled , but the qutrit can not be . yet , the import of theorem i is that for all known mub complements , there exists a factorization into parts ( represented by some set of generalized pauli operators ) , in terms of which theorem i ( and the others ) hold . thus , in addition to the existence question in composite dimensions , there is also an existence question in @xmath42 dimensions - do mub complements exist that violate theorem i ? perhaps entanglement considerations such as the present ones will help in answering these persistent questions .
i would like to thank bill wootters and winton brown for many stimulating discussions , and i would like to thank the james franck institute for its hospitality during the time when this work was completed . to generalization with maximum transparency , we follow the alternative forms written in the text and define @xmath144 bases by the generator set 4 = ( abii , aici , aiid , stuv ) , [ ageng ] where every four - body factor must differ from its two - body counterpart ( @xmath185 , etc . ) .
the two - body operators provide the special product basis for the @xmath4-term expansion .
the generalization to @xmath1-qupit ghz states is apparent .
the @xmath143 bases are best defined in a similar way , although a bit less transparently because there are only two independent two - body operators , 4 = ( aici , ibid , sbui , itcv ) .
[ agenc ] again , the two - body operators provide a product basis for the expansion , which in this case requires @xmath78 terms . individual factors in the three - body operators must differ from corresponding factors in the two - body operators , except where their equality is explicit . it should be noted that there are two variations on the @xmath143 generator set , corresponding to the other ways of pairing the two - body factors .
one such variation is ( @xmath186 ) .
the three alternatives are mathematically equivalent , although one can make a physical distinction based on entanglement links between pairs .
the stronger entanglement links in a system represented by are between neighbors in the sequence ( 1 - 2 - 3 - 4 - 1 ) . in the variation given above
, the sequence is ( 1 - 3 - 2 - 4 - 1 ) , and in the other possible variation it is ( 1 - 2 - 4 - 3 - 1 ) . the various possibilities are not unphysical , as one can imagine unlike particles with tetrahedral coordination . we now argue that the two bases defined above are the only nonseparable options for qubits .
we first argue that both are the unique nonseparable representatives of their respective @xmath25-body profiles as shown on table iii .
we then show that other profiles can not exist for qubits .
it is straightforward to verify that the @xmath144 generators produce six two - body operators involving the same factors , @xmath115 - @xmath187 .
these exhaust all 12 of the available @xmath90 factors ( theorem ii demands three @xmath90 factors per qubit ) , so that all remaining operators are 4-body operators , as shown in the table . the only other basis that can share this profile is @xmath136 , in which the two - body operators have factors that do not commute individually .
but , by virtue of this fact , the @xmath136 basis has four independent two - body operators , so that these can compose the generator set .
the separability of the basis is then obvious .
turning to the @xmath143 case , it is straightforward to show that the generators of produce no further two - body operators beyond the two shown , so that remaining @xmath90 factors must appear with the 8 three - body operators .
one might wonder whether a different basis could be found with the same profile by using four - body generators in place of the three - body generators .
the answer is no - it is easy show that this would generate only @xmath144 or @xmath136 bases , depending upon whether one of the four - body generators shares factors with one of the two - body generators .
the remaining point is to rule out other four - qubit profiles .
it suffices to consider just the two - body operators , whose maximum number is six .
we will show that the numbers 4 and 0 are impossible for qubits .
the former case is very simple - it is impossible to find four commuting two - body operators that do not generate two more ( and these will immediately identify themselves as belonging to either a @xmath144 or @xmath136 compatibility group ) . as to the latter case ,
assume that there are no two - body operators .
then all 12 of the @xmath90 factors must appear in one - body operators , making three appearances on each qubit ( theorem ii ) .
consider any two of these operators that have their @xmath90 factors on the same qubit .
commutativity demands that they have exactly one other factor in common , so that their product is a two - body operator .
this forms a contradiction and shows that there is no profile without two - body operators . to briefly summarize the results of this appendix , the 6 mub types listed on table iii exhaust the possibilities for four qubits .
the 5 corresponding @xmath1-body profiles are also exhaustive ; in particular , a @xmath150-like profile does not exist for qubits .
99 i. ivanovi , j. phys .
a * 14 * , 3241 ( 1981 ) .
w.k . wootters and b.d .
fields , ann .
phys . * 191 * , 363 ( 1989 ) ; and in _
bell s theorem , quantum theory and conceptions of the universe _ , edited by m. kafatos ( kluwer academic publishers , dordrecht , 1989 ) , pp .
bennett and g. brassard , in _ proc .
conf . on computers , systems , and signal processing _ , bangalore , india ( 1984 ) , p. 175 .
d. bruss and c. macchiavello , phys .
letters * 88 * , 127901 ( 2002 ) .
ekert , phys .
letters , * 67 * , 661 ( 1991 ) .
s. grblacher et .
al . , new j. phys . * 8 * , 75 ( 2006 ) .
aravind , z. naturforsch . *
58a * , 85 ( 2003 ) .
p. shor , phys .
a * 52 * , r2493 ( 1995 ) .
d. gottesmanm , phys .
a * 54 * , 1862 ( 1996 ) .
r. calderbank , e.m .
rains , p.w .
shor , and n.j.a.sloane , phys , rev .
letters , * 78 * , 405 ( 1997 ) .
p. hyllus , o. ghne , d. bruss , and m. lewenstein , phys . rev . * 72 * , 012321 ( 2005 ) .
brukner and a. zeilinger , phys .
a * 63 * , 022113 ( 2001 ) .
j. lawrence , .
brukner , and a. zeilinger , phys .
a * 65 * , 032320 ( 2002 ) .
t. durt , b .-
englert , i. bengtsson , and k. zyczkowski , int
* 8 * , 535 ( 2010 ) .
s. brierley and s. weigert , phys .
a * 78 * , 042312 ( 2008 ) .
j. lawrence , phys .
a * 70 * , 012302 ( 2004 ) .
romero , g. bjrk , a.b .
klimov , and l.l .
snchez - soto , phys .
a * 72 * , 062310 ( 2005 ) .
m. wieniak , t. paterek , and a. zeilinger , e - print arxiv : quant - ph/1102.2080v2 .
briegel and r. raussendorf , phys .
letters , * 86 * , 910 ( 2001 ) .
a.b . klimov , c. munoz , and l.l .
snchez - soto , eprint quant - ph/1007.1751 ( 2010 ) .
see p. 90 of .
s. bandyopadhyay , p.o .
boykin , v. roychowdhury , and f. vatan , algorithmica * 34 * , 512 ( 2002 ) ; or e - print quant - ph/0103162 ( 2001 ) .
nielsen and i.l .
chuang , _ quantum computation and quantum information _
( cambridge univ .
press , cambridge , england , 2000 ) , p. 454 .
one may think of this as an @xmath1-dimensional discrete fourier transform with @xmath4 points in each dimension , or as a fourier transform over a galois field , with @xmath25 and @xmath47 as galois field variables .
boykin , m. sitharam , p.h .
tiep , and p. wojcan , quant .
* 7 * , 371 ( 2007 ) .
the one - qupit pauli groups may be written @xmath188 for odd @xmath4 , and @xmath189 for @xmath7 , where @xmath66 represents all of the @xmath78 pauli matrix factors .
these are special cases of the discussion following .
this result was derived in another way in .
a one - body pauli operator transforms only a single qupit ( for example , @xmath190 in a 2-qupit system and @xmath191 in a 3-qupit system ) .
such an operator can have no totally entangled eigenstates , because in any eigenstate the transformed qupit must be pure . a.b .
klimov , d. sych , l.l .
snchez - soto , and g. leuchs , phys . rev .
a * 79 * , 052101 ( 2009 ) .
t. durt , e - print arxiv : quant - ph/0401046 ; also see discussion on pp .
43 - 46 of .
mermin , phys .
letters , * 65 * , 1838 ( 1990 ) .
briegel and r. raussendorf , phys .
letters , * 86 * , 5188 ( 2001 ) .
briegel , d.e .
browne , w. dr , r. raussendorf , and m. van den nest , nature physics * 5 * , 19 ( 2009 ) .
this symmetry is referred to in the concluding paragraph ( p. 91 ) of . | a few simply - stated rules govern the entanglement patterns that can occur in mutually unbiased basis sets ( mubs ) , and constrain the combinations of such patterns that can coexist ( , the stoichiometry ) in full complements of ( @xmath0 ) mubs .
we consider hilbert spaces of prime power dimension ( as realized by systems of @xmath1 prime - state particles , or _ qupits _ ) , where full complements are known to exist , and we assume only that mubs are eigenbases of generalized pauli operators , without using a particular construction .
the general rules include the following : 1 ) in any mub , a particular qupit appears either in a pure state , or totally entangled , and 2 ) in any full mub complement , each qupit is pure in ( @xmath2 ) bases ( not necessarily the same ones ) , and totally entangled in the remaining ( @xmath3 ) .
it follows that the maximum number of product bases is @xmath2 , and when this number is realized , all remaining ( @xmath3 ) bases in the complement are characterized by the total entanglement of every qupit .
this `` standard distribution '' is inescapable for two qupits ( of any @xmath4 ) , where only product and generalized bell bases are admissible mub types .
this and the following results generalize previous results for qubits @xcite and qutrits @xcite , drawing particularly upon . with three qupits
there are three mub types , and a number of combinations ( @xmath5 ) are possible in full complements . with @xmath6 , there are 6 mub types for @xmath7 , but new mub types become possible with larger @xmath4 , and these are essential to realizing full complements . with this example , we argue that new mub types , showing new entanglement characteristics , should enter with every step in @xmath1 , and when @xmath1 is a prime plus 1 , also at critical @xmath4 values , @xmath8 .
such mubs should play critical roles in filling complements . |
in previous two papers @xcite , @xcite we defined and analyzed random processes with continuous time parameter describing the evolution of special trees consisting of _ living and dead nodes _ connected by _ lines_. it seems to be appropriate to repeat briefly the characteristic features of the evolution process .
the initial state @xmath3 of the tree corresponds to a single living node called _ root _ which at the end of its life is capable of producing @xmath4 new living nodes , and after that it becomes immediately dead .
if @xmath5 then the new nodes are promptly connected to the dead node and each of them _ independently of the others _ can evolve further like a root .
the random evolution of trees with continuous time parameter has not been investigated intensively recently .
the main interest since the late 1990s has been focussed on the study of non - equilibrium networks @xcite occurring in common real world . the evolution mechanism of trees with living and dead nodes may be useful in some of biological systems where the branching processes are dominant . in what follows , we will use notations applied in @xcite and @xcite .
it seems to be useful to cite the basic definitions .
the probability to find the number @xmath0 of living nodes produced by one dying precursor equal to @xmath6 was denoted by @xmath7 where @xmath8 .
is the set of non - negative integers . ] for the generating function as well as the expectation value and the variance of @xmath0 we used the following notations : @xmath9 where @xmath10_{z=1 } , \;\;\;\;\;\ ; j = 1 , 2 , \ldots\ ] ] are _ factorial moments _ of @xmath0 .
it was shown in @xcite that the time dependence of the random evolution is determined almost completely by the expectation value @xmath11 . in accordance to this
the evolution was called subcritical if @xmath12 , critical if @xmath13 and supercritical if @xmath14 . in the further considerations we are going to use four distributions for the random variable @xmath0 . as shown in @xcite the equations derived for the first and the second moments of the number of nodes are _ independent of the detailed structure of the distribution of @xmath0 _ provided that the moments @xmath11 and @xmath15 are finite .
we called distributions of this type for @xmath0 _ arbitrary _ and used the symbol @xmath16 for its notation . in many cases it seems to be enough to apply the truncated distribution of @xmath0 .
if the possible values of the random variable @xmath0 are @xmath17 and @xmath18 with probabilities @xmath19 and @xmath20 , respectively , then in the previous paper @xcite the distributions of this type were denoted by @xmath21 .
many times it is expedient to assume distributions to be completely determined by one parameter . as known the geometric and poisson distributions are such distributions . in paper
@xcite we used the symbols @xmath22 and @xmath23 to identify these distributions .
the distribution function of the lifetime @xmath24 of a living node will be supposed to be exponential , i.e. @xmath25 . in order to characterize the tree evolution two non - negative integer valued random functions @xmath26 and @xmath27
are introduced : @xmath26 is the number of living nodes , while @xmath27 is that of dead nodes at @xmath28 .
the total number of nodes at @xmath29 is denoted by @xmath30 .
clearly , the nodes can be sorted into groups according to _ the number of outgoing lines_. following the notation in @xcite the number of nodes with @xmath31 outgoing lines at time instant @xmath29 is denoted by @xmath32 .
a node not having outgoing line is called _ end - node_. it is obvious that an end - node is either live or dead .
therefore , the number of end - nodes @xmath33 can be written as a sum of numbers of living and dead end - nodes , i.e. @xmath34 . since all living nodes are end - nodes @xmath35 can be replaced by @xmath26 .
the total number of dead nodes @xmath27 is given by @xmath36 .
in this paper we are dealing with properties of @xmath37 and @xmath38 when @xmath39 .
we will call the random trees arising from a single root after elapsing infinite time _
stationary_. in section @xmath18 the basic properties of probability distributions of the number of nodes , living and end - nodes are investigated when @xmath39 .
special attention is paid in section @xmath40 to the effect of distribution law of the number of outgoing lines .
three different distributions of @xmath0 are investigated . in order to simplify the notation ,
indices referring to different distributions of @xmath0 are usually omitted in formulas .
finally , the characteristic properties of stationary random trees are summarized in section @xmath41 .
let us introduce the notion of _ tree size _ which is nothing else but the total number of nodes @xmath30 at time moment @xmath28 .
we want to analyze the asymptotic behavior of the tree size , i.e. the behavior of the random function @xmath30 when @xmath42 .
we say the _ limit random variable _
@xmath43 exists in the sense that the relation : @xmath44 is true for all positive integers @xmath2 , where @xmath3 denotes the initial state of the tree .
a randomly evolving tree is called _
`` very old '' _ when @xmath45 , and a very old tree , as mentioned already , will be named _
stationary random tree_. it is elementary to prove that if the limit probability @xmath46 exists , then the generating function @xmath47 is determined by one of the fixed points of the equation @xmath48.\ ] ] it can be shown that if @xmath49 , then the fixed point to be chosen has to satisfy the limit relation @xmath50 while if @xmath51 , then it should have the property @xmath52 and independently of @xmath11 the equation @xmath53 must hold .
the relation ( [ 4 ] ) means that the probability to find stationary tree of finite size is evidently @xmath54 , if @xmath55 , but if @xmath51 , then @xmath56 i.e. the probability to find a stationary tree of infinite size is equal to @xmath57 the proof of the proposition is simple .
let us assume that @xmath58 is a _ probability generating function _
, i.e. @xmath59 and @xmath60 .
we need the following theorem : [ th1 ] if @xmath61 , then @xmath62 ; while if @xmath63 , then there is a point @xmath64 such that @xmath65 , i.e. @xmath66 and @xmath67 .
let us introduce the function @xmath68 . since @xmath69 is convex , i.e. all derivatives are positive in the interval @xmath70 $ ] it is evident that @xmath71 is a nondecreasing function of @xmath72 $ ] . if @xmath73 , then @xmath74 , i.e. @xmath75 is nondecreasing , negative valued function of @xmath76 in @xmath77 . since @xmath78 it is obvious that @xmath79 , if @xmath80 , i.e. @xmath81 , and this is the first statement of theorem [ th1 ] . if @xmath82 , then @xmath83 , and since @xmath78 the inequality @xmath79 has to be true for all @xmath84 lying near @xmath54 . on the other side @xmath85
, hence there should exist one is excluded because @xmath69 is convex.]@xmath86 in @xmath87 which satisfies the equation @xmath88 and that implies the second statement of theorem [ th1 ] .
it seems to be important to investigate _ how the living nodes behave _ in very old , i.e. stationary trees .
intuitively one can say that stationary trees arising in subcritical evolution do not contain living nodes and they have finite average size . at the same time
, it seems to be quite obvious that stationary trees originating in supercritical evolution could have living nodes with non - zero probability , i.e. they are entities of _ `` eternal life''_. in order to give more precise answer the generating function of the random variable @xmath89 should be derived .
it is easy to show that @xmath90 is one of the fixed points of the equation @xmath91 = g^{(\ell)}(z).\ ] ] by theorem @xmath54 , equation ( [ 6 ] ) has one fixed point if @xmath92 , namely @xmath93 $ ] , and from this it follows that @xmath94 i.e. the probability that a subcritical very old tree does not have living node , is exactly @xmath54 .
if @xmath51 , then besides @xmath95 equation ( [ 6 ] ) has another fixed point given by @xmath96 $ ] , hence the probability @xmath97 should be larger than zero .
in other words , a supercritical stationary tree may evolve infinitely with certain non - zero probability . of subcritical stationary trees
the measure of subset containing trees with living nodes is zero , while in that of supercritical trees the measure of subset consisting of trees with living nodes is larger than zero . ]
the expectation value and the standard deviation of the total number of nodes can be used to characterize the size of a stationary random tree . from ( [ 3 ] ) , one obtains @xmath98_{z \uparrow 1 } = \frac{1}{1-q_{1 } } , \;\;\;\;\;\ ; \mbox{if } \;\;\;\;\ ; q_{1 } < 1.\ ] ] in a supercritical evolution , i.e. when @xmath99 , the expectation value @xmath100 does not exist , but the limit relation @xmath101 can be simply proved .
we note here that in the critical case , i.e. when @xmath102 , the average tree size becomes infinite linearly , i.e. we have the relation @xmath103 .
the standard deviation of the total number of nodes in stationary random trees is given by @xmath104 in a supercritical state the standard deviation does not exist , but it can be readily shown that @xmath105 finally , we would like to deal briefly with some of the _ properties of end - nodes _ in stationary random trees . by using equation ( @xmath106 ) of @xcite
, it can be shown that the generating function @xmath107 is nothing else than one of fixed points of the following simple equation : @xmath108 = g_{0}(z ) + ( 1-z)\;f_{0}.\ ] ] substituting @xmath109 we obtain @xmath110^{n}\right\ } = 0,\ ] ] i.e. @xmath111 , hence @xmath112 . in the sequel
we will investigate the properties of stationary trees when @xmath69 is known . in this case
we can obtain exact expressions for probabilities @xmath113 from the corresponding generating function .
the generating function of @xmath0 is given by @xmath114 with the restriction for @xmath11 and @xmath15 determined by the equality @xmath115 .
( see fig . 1 in @xcite . ) by using eq .
( [ 3 ] ) and applying ( [ 8 ] ) we have @xmath116 = g.\ ] ] the fixed point of this equation is @xmath117,\ ] ] is also the generating function of @xmath118 .
it is an elementary task to show that @xmath119 [ ht ! ] since @xmath120 performing expansion of generating function ( [ 9 ] ) into power series of @xmath121 , we can determine the probabilities @xmath122 easily .
after elementary calculations we obtain @xmath123,\ ] ] where @xmath124 @xmath125 and @xmath126 [ ht ] the probabilities @xmath127 can be seen in fig .
these are the probabilities to find @xmath128 nodes in a `` very old '' tree developed according to the distribution of @xmath129 .
the dark and light bars correspond to subcritical @xmath130 and slightly supercritical @xmath131 tree evolutions , respectively . in the last case @xmath132 and it is not surprising that the probabilities @xmath133 for finite @xmath2 are larger in sub- than in supercritical trees .
[ fig2 ] shows the dependence of @xmath133 on @xmath2 in sub- and supercritical evolutions , respectively .
if the difference @xmath134 is small enough , let us say @xmath135 , then one observes a special phenomenon , namely , an _ `` oscillation '' _ in the dependence of @xmath133 versus @xmath2 , what is clearly seen in fig .
[ fig2 ] .
[ ht ! ] in order to explain the origin of `` oscillation '' we calculated the dependence of probabilities @xmath136 and @xmath137 on @xmath11 at @xmath138 . in fig .
[ fig3 ] , we can see that @xmath139 in the interval @xmath140 . for the sake of simplicity
the upper limit @xmath141 will be called `` critical @xmath11 '' .
the reason of oscillation is trivial . clearly , if @xmath142 , then there are no random trees having nodes of even number , i.e. @xmath143 , therefore , it should be an interval @xmath144 in which @xmath145 we calculated the critical values of @xmath11 corresponding to @xmath146 at @xmath147 .
the results are shown in table 1 .
as expected @xmath148 is slightly decreasing with @xmath6 when @xmath15 is fixed .
table @xmath54 : critical values of @xmath148 at @xmath149 . [ cols="^,^,^,^",options="header " , ] [ tab4 ] [ ht ! ]
it follows from eq .
( [ 22 ] immediately that the probability @xmath133 vs. @xmath11 has a maximum at @xmath150 the appearance of maxima is well seen in fig .
[ fig13 ] in the cases of @xmath151 .
finally , we would like to discuss the _ properties of end - nodes _ in stationary random trees evolved according to poisson distribution of @xmath0 .
in order to obtain the probabilities @xmath152 we have to solve the equation @xmath153 = g_{0}(z ) + ( 1-z)\exp(-q_{1}),\ ] ] where @xmath154 by using a special expansion procedure we can get the probabilities @xmath155 @xmath156 step - by - step . introducing the notations @xmath157 @xmath158 @xmath159 @xmath160 @xmath161 the first six probabilities are given by the following eqs .
: @xmath162^{3 } , \;\;\;\ ; p_{3}^{(0 ) } = \frac{1}{6 } q_{1}^{3 } \ell_{3}(q_{1 } ) [ u(q_{1})]^{5},\ ] ] @xmath163^{7 } , \;\;\;\ ; p_{5}^{(0 ) } = \frac{1}{120 } q_{1}^{5 } \ell_{5}(q_{1 } ) [ u(q_{1})]^{9},\ ] ] @xmath164^{11}.\ ] ] [ ht ! ] [ ht ! ] by using these formulas we are able to show the dependence of @xmath165 on @xmath2 and @xmath11 , respectively .
we see in fig .
[ fig14 ] that the evolution process prefers the one end - node structures , i.e. stationary random trees in which the probability of finding @xmath166 end - nodes is very small .
the curves @xmath167 versus @xmath11 plotted in fig .
[ fig15 ] are clearly demonstrating that the sites of maxima of probabilities are shifted to higher values of @xmath11 as the number of nodes @xmath2 is increasing .
we have investigated the main properties of evolution of special trees when the continuous time parameter tends to infinity . a tree of infinite age is called _ stationary _ or simply `` very old '' .
we have stated that if the limit relations @xmath168 are true , then the random function @xmath30 giving the number of nodes in a tree at @xmath29 _ converges in distribution _ to a _ random variable @xmath118 _ which counts the number of nodes in a stationary tree . for the generating function
@xmath169 a simple equation has been derived , namely @xmath170 , \;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\ ; \mbox{{\bf ( a)}}\ ] ] where @xmath69 is nothing else than the generating function of the number of new nodes @xmath0 produced by one dying node .
it has been proved that if @xmath0 has finite first and second factorial moments , i.e. if @xmath171 and @xmath172 are finite , then @xmath173 , if @xmath49 , and @xmath174 , if @xmath51 .
it means , if @xmath49 , then the probability @xmath175 of finding infinite number of nodes in a stationary random tree is zero , while if @xmath51 , then that is larger than zero , namely , @xmath176 . here , one has to note that the expectation value @xmath177 exists only if the tree evolution is subcritical , i.e. if @xmath178 . in the case of @xmath51 ,
i.e. in supercritical evolution it has been shown that @xmath179\ } = q_{1}/(q_{1}-1),\ ] ] while if @xmath102 , then @xmath180 .
it has been shown also that the generating function @xmath181 of the number of end - nodes @xmath182 in a stationary random tree has to satisfy the equation @xmath183 - ( 1-z)f_{0 } , \;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\ ; \mbox{{\bf ( b)}}\ ] ] which can be used to study end - node properties .
when @xmath184 is known then by using appropriate method for comparison of coefficients of @xmath185 in both sides of equations * ( a ) * and * ( b ) * we could calculate step - by - step the probabilities @xmath186 to find @xmath187 nodes and end - nodes , respectively , in a stationary random tree . for the calculations , we have chosen three different distributions of @xmath0 .
the generating functions of these distributions are given by the following formulas : @xmath188 , & \mbox{if $ \nu \in { \bf g}$ , } \\
\mbox { } & \mbox { } \\ e^{q_{1}(z-1 ) } , & \mbox{if $ \nu \in { \bf p}$. } \end{array } \right.\ ] ] analyzing the results of numerical calculations the first impression is that the qualitative properties of stationary random trees depend hardly on the character of distribution of @xmath0 .
we have seen that in all cases the probability to find @xmath189 node ( or end - node ) in a stationary tree is significantly larger then to find @xmath190 nodes .
one can conclude that in the evolution process the formation of a rod - like stationary random tree is much more probable than that with many branches .
we have found special behavior in the dependence of @xmath133 on @xmath2 only in the case of @xmath21 when @xmath191 is smaller than a critical value .
the appearance of the oscillation of @xmath133 versus @xmath2 is consequence of the following trivial statement : when @xmath192 then @xmath193 .
it has been demonstrated that the probabilities @xmath133 and @xmath165 versus @xmath11 show a maximum the location of which is increasing with @xmath2 but remains always smaller than @xmath54 .
this property is best seen in the case of poisson distribution of @xmath0 .
99 l. pl , randomly evolving trees i , arxiv : cond - mat/0205650 l. pl , randomly evolving trees ii , arxiv : cond - mat/0211092 s.n .
dorogovtsev and j.f.f .
mendes , _ evolutions of networks _ ( oxford university press , 2003 ) | the properties of randomly evolving special trees having defined and analyzed already in two earlier papers ( arxiv : cond - mat/0205650 and arxiv : cond - mat/0211092 ) have been investigated in the case when the continuous time parameter converges to infinity .
equations for generating functions of the number of nodes and end - nodes in a stationary ( i.e. infinitely old ) tree have been derived . in order to solve exactly these equations
we have chosen three different distributions for the number of new nodes @xmath0 produced by one dying node . by using appropriate method
we have calculated step - by - step the probabilities of finding @xmath1 nodes as well as end - nodes in a stationary random tree .
analyzing the results of numerical calculations we have observed that the qualitative properties of stationary random trees depend hardly on the character of distribution of @xmath0 . the conclusion to be correct that in the evolution process the formation of a rod - like stationary tree is much more probable than the formation of a tree with many branches .
we have established that the probability of finding @xmath2 nodes in a stationary tree depends sensitively on the average value of @xmath0 and has a maximum the location of which is increasing with @xmath2 but remains always smaller than unity .
this is also true for the end - nodes . *
pacs : 02.50.-r , 02.50.ey , 05.40.-a * |
abundance measurements of c , n , o , ne , s , and ar in galactic and extragalactic h ii regions , planetary nebulae , and supernova remnants represent a major source of information about elemental levels in the interstellar media of spiral disks .
measured from the ground in most cases , the strengths of the numerous emission features produced by these objects can be converted in a straightforward way to ionic and elemental abundances . when the abundances for nebular objects within a single galaxy are compiled , several correlations are shown to exist either between interstellar metallicity and galactocentric distance , i.e. an abundance gradient , or between pairs of abundance ratios .
since the former is directly linked to matter distribution and star formation rates , and the latter depends on the imf and stellar yields , complete abundance profile information for a galactic disk provides important constraints on galactic chemical evolution models and hence our global understanding of how elemental composition changes within a galaxy as a function of time and location .
the purpose of this review is to provide a summary of extant data pertaining to nebular abundance profiles in disk galaxies along with brief interpretations .
readers are referred to other papers in this volume for more detailed theoretical explorations of abundance gradients .
because oxygen is readily accessible spectroscopically , its abundance provides a convenient tracer of metallicity distribution in a galactic disk .
i begin by discussing the milky way oxygen profile and then follow up with a summary of general results for a large number of other spirals .
data for oxygen in the milky way disk were taken from the following papers : shaver et al .
( 1983 ) , the first major survey of abundances in galactic h ii regions ; vlchez & esteban ( 1996 ) , a focused study of h ii regions at large galactocentric distances ; and maciel & kppen ( 1994 ) , where a large number of galactic type 2 planetary nebulae were used to map abundances in the disk .
abundances were generally derived in these papers by combining measured line strengths directly with atomic data and ionization correction factors as described in osterbrock ( 1989 ) .
figure 1 shows 12+log(o / h ) versus galactocentric distance in kpc for the milky way , where symbol shape indicates the data source .
also included are the recent b star results from smartt & rolleston ( 1997 ) .
numbers in square brackets in the legend give 12+log(o / h ) at the sun s distance ( 8.5 kpc ) and the slope in dex / kpc , respectively , for least squares fits .
parameter uncertainties are given in parentheses .
the sun s position ( grevesse & noels 1993 ) is indicated with an ` x ' .
error bars here and in subsequent figures show typical uncertainties . besides the presence of a negative metallicity gradient , fig .
1 implies the following .
( 1 ) the galactic oxygen gradient flattens beyond 10kpc , according to the vlchez h ii region data .
( 2 ) the b star oxygen profile is consistent with h ii region results inside of 10kpc and shows no flattening beyond 10kpc .
( 3 ) the oxygen abundances in planetary nebulae are systematically less than in h ii regions at corresponding distances by roughly 0.14 dex , qualitatively consistent with the idea that planetary nebulae represent an older , less metal - rich population than h ii regions . turning now to the consideration of spiral galaxies generally
, large surveys of o / h in extragalactic h ii regions include those of mccall ( 1982 ; 40 galaxies ) , vila - costas & edmunds ( 1992 ; 32 galaxies ) , and zaritsky , kennicutt , & huchra ( 1994 ; 39 galaxies ) .
detailed results for o / h in individual spirals can be found in these papers .
to show general findings here i have extracted characteristic abundances and gradient slopes from zaritsky et al . and present them in figure 2 as functions of galaxy morphological type ( t type ) , absolute blue magnitude m@xmath0 , and maximum circular velocity v@xmath1 in km / s .
all three of these independent parameters are indicators of galaxy mass , where smaller t type indices , more luminous integrated blue magnitudes , and larger rotational velocities generally correspond with more massive spirals .
gradient slopes are shown in dex / r@xmath2 .
normal and barred spirals are shown separately using filled and open symbols , respectively .
parameters from the shaver data for the milky way are indicated in fig . 2 with dashed lines , where i have adopted 11.5kpc for the isophotal radius ( de vaucouleurs & pence 1978 ) .
i note that while isophotal radius is employed as the normalization standard here , use of effective radius ( the half light radius ) or kiloparsecs produces similar results .
the choice of normalization standard is discussed by garnett in this volume .
two points are implied by fig . 2 :
( 1 ) characteristic abundances increase with galaxy mass , while gradient slopes are uncorrelated with this parameter ; and ( 2 ) characteristic abundances in normal and barred spirals are indistinguishable , but barred spirals appear to have flatter gradients . both of these results have been noted previously .
garnett & shields ( 1987 ) plotted characteristic o / h values against galaxy mass for numerous spirals and found a direct correlation between these two parameters , while pagel et al .
( 1979 ) first suggested that barred spirals may have flatter gradients .
while metallicity patterns are apparently related closely to global galaxy properties , heavy element ratios such as n / o are expected to reflect characteristics of the imf , stellar yield patterns , and star formation history .
papers on heavy element abundance ratios for the milky way included in this review are : shaver et al .
( 1983 ; n / o , s / o , ne / o , and ar / o ) , vlchez & esteban ( 1996 ; n / o , s / o ) , simpson et al .
( 1995 ; n / o , s / o , ne / o ) , maciel & kppen ( 1994 ; ne / o , s / o , ar / o ) , and fesen , blair , & kirshner ( 1985 ; n / o , s / o ) . ratios for extragalactic h ii regions are taken from papers by thurston , edmunds , & henry ( 1996 ; n / o ) , kobulnicky & skillman ( 1996 ; n / o ) , garnett ( 1989 ; s / o ) , and garnett et al . ( 1995 , 1997 ; c / o ) .
note that simpson et al .
used lines in the ir to infer their abundances .
all papers considered h ii regions with the exception of those by fesen et al . and
maciel & kppen which studied supernova remnants and planetary nebulae , respectively .
figure 3 shows log(n / o ) versus 12+log(o / h ) for both the milky way and extragalactic objects .
milky way data are shown with symbols whose shapes are interpreted in the legend of fig .
notice the segregation in the milky way data of the simpson and fesen points from the shaver and maciel points , perhaps due to the differences in the way these ratios were measured .
t type values are used as symbols for the thurston et al .
spirals , while ` d ' designates the kobulnicky & skillman dwarf galaxies .
parameters for least squares fits are indicated in the figure for the two extragalactic data sets only , where a = y - intercept , b = slope , and c = correlation coefficient .
because of the scatter in the milky way data , the following analysis applies only to the extragalactic n / o studies . the flatter n / o behavior seen at 12+log(o / h)@xmath38.0 can be ascribed to the dominance of primary nitrogen production where cno cycling is fed with carbon and oxygen derived directly from triple alpha processing within the same star .
thus , n / o is relatively independent of the star s metallicity . the steeper slope in n / o at higher metallcities is probably related to metallicity - sensitive secondary nitrogen production in which significant amounts of carbon and oxygen already present in the star enters into the cno cycle . [ see vila - costas & edmunds ( 1993 ) and thurston et al .
( 1996 ) for further discussion . ] finally , note the tendency for early type spirals ( low integer values ) to have systematically higher log(n / o ) for the same 12+log(o / h ) .
c / h , s / h , ne / h , and ar / h are plotted logarithmically against 12+log(o / h ) in fig . 4 , where the legend connects symbol shape with data source , and the parameters ( and uncertainties ) defined as in fig . 3 for the least squares
linear fits are shown in each panel .
note that open and filled symbols designate milky way and extragalactic h ii regions , respectively . the top left panel of figure 4 shows c / h values from garnett et al .
( 1995 ; 1997 ) for extragalactic h ii regions , along with results for galactic objects m 8 ( peimbert et al .
1993 ) and the orion nebula ( walter et al . 1992 ) . the two results for orion stem from the assumption of line - of - sight temperature fluctuations in the case of the higher value .
the two points at low o / h are for i zw 18 . for purposes of the fit , the orion collisional line abundance and the two points for i zw 18 were omitted .
a clear correlation exists between c and o , and with a slope significantly greater than unity , a metallicity - sensitive production mechanism for carbon is suggested .
another strong correlation , assisted considerably by the extragalactic data , exists between s / h and o / h ( top right panel ) .
it is interesting to note the greater scatter for planetary nebula data than for h ii region data . in all but the simpson et al .
work , s abundances were determined by combining s@xmath4 abundances with an ionization correction factor .
however , the excitation level of planetary nebulae is generally higher than in h ii regions , so planetaries should have proportionately less s@xmath4 , and this method will be less accurate for them . the strong correlation shown in this plot , along with a slope close to unity , implies that s / o is generally constant .
the average log(s / o ) value is -1.76 , close to the solar value of -1.66 ( grevesse & noels 1993 ) .
this result agrees with garnett ( 1989 ) , although daz ( 1989 ) found a negative correlation between log(s / o ) and 12+log(o / h ) .
in addition to ionization correction factor uncertainties , sulfur abundance studies also suffer from the effects of severe telluric absorption on the ir [ s iii ] 9069 and 9532 lines .
ne / h and ar / h appear to be correlated with o / h ( bottom two panels ) , although the slopes are less than unity .
nevertheless , average ratios of @xmath5 and @xmath6 , compare favorably with solar values of -0.80 and -2.27 ( grevesse & noels 1993 ) , respectively .
tight correlations such as those between s , ne , ar , and o suggest that these elements are produced by primary nucleosynthesis in massive stars with yields that are invariant with metallicity . this issue was explored in the case of ne and o in particular by henry ( 1989 ) , who found the interstellar ne / o ratio to be constant in several different galaxies .
_ the effects of bars on abundance gradients _ has recently been studied extensively in particular by martin , roy , walsh , belley , and julien . the bottom
three panels of fig .
2 show the tendency for barred spirals to possess shallower gradients .
the paper by martin & roy ( 1994 ) further relates gradient slope to bar strength , a quantity which measures bar ellipticity .
they find direct relations between the slope of the oxygen abundance gradient of a barred spiral and the galaxy s bar strength ( ellipticity ) and length .
this empirical result is consistent with radial flow models of chemical evolution in which the presence of a bar enhances large - scale mixing over the galaxy s disk , damping radial abundance variations . _ a negative vertical gradient in o / h in the milky way _
is suggested by planetary nebula studies .
abundance data compiled by kaler ( 1980 ) for pne ranging in height above the disk from less than 0.4 kpc to greater than 1 kpc show a decrease in o / h with increasing height above the plane .
a comparison of more recent studies of pne close to the plane ( perinotto 1991 ) , greater than 300pc above the plane ( cuisinier et al . 1996 ) , and in the halo ( howard & henry 1997 ) shows averages of 12+log(o / h ) for these three samples of 8.68 , 8.52 , and 8.02 respectively , qualitatively consistent with kaler .
thorough tests for _ azimuthal gradients in spiral disks _ have yet to be carried out .
one example of apparent o / h asymmetry is discussed by kennicutt & garnett ( 1996 ) in their study of m101 .
they find that h ii regions located along a spiral arm southeast of the major axis have a lower oxygen abundance by 0.2 - 0.4 dex compared with h ii regions on the opposite side .
_ global metallicities in low surface brightness galaxies _ are generally found to be subsolar by roughly a factor of three , according to mcgaugh ( 1994 ) , indicating that these galaxies evolve very slowly and form few stars during a hubble time .
apparently , they also lack detectable gradients .
this , despite the fact that these objects are similar in mass and size to prominent spirals defining the hubble sequence .
mcgaugh suggests that a galaxy s environment and surface mass density are more relevant to galaxy evolution than gross size . _
effects of cluster environment _ on the chemical evolution of galaxies has been investigated by skillman et al .
( 1996 ) , who studied oxygen profiles in several virgo spirals representing a range in h i deficiency ( taken as a gauge of cluster environmental interactions ) .
their results imply that global metal abundances in disks tend to be higher in stripped galaxies , presumably because reduced infall of metal - poor h i gas means less dilution of disk material .
henry et al .
( 1996 and references therein ) investigated metallicity and heavy element abundance ratios ( n / o , s / o ) in three cluster spiral disks with normal h i and found no clear signatures of environmental effects .
thus , cluster environment alone is apparently not a sufficient condition for altered chemical evolution . _
the mathematical form of abundance profiles in spiral disks , _ has been investigated recently by henry & howard ( 1995 ) , who fit line strength behavior over the disks of m33 , m81 , and m101 using photoionization models .
their best fits for o / h versus galactocentric distance were produced using exponential profiles , although power law forms could not be ruled out . however , linear profiles poorly reproduced the observations .
henry and howard also concluded that there is currently no strong observational case for gradient flattening in the outer parts of some disks , although such flattening has been proposed by several authors ( see moll et al .
since the conference review on this same subject by daz ( 1989 ) both the average number of observed and analyzed h ii regions per galaxy and the number of galaxies sampled have steadily risen . also , a larger number of elements is now being studied .
the following points seem solid : ( 1 ) global metallicity in spirals is influenced by galaxy mass .
evidence in vila - costas & edmunds ( 1992 ) also suggests that interstellar metallicity scales directly with total surface density , perhaps through a star formation rate which is a function of local density , and thus an abundance gradient merely traces matter distribution .
( 2 ) n / o behavior with metallicity clearly suggests a primary then secondary origin of n with chemical evolution .
( 3 ) finally , constant ratios of s / o , ne / o , and ar / o are ubiquitous and appear to suggest primary nucleosynthesis in massive stars with metallicity - insensitive yields .
while significant progress has been made over the past decade , more questions remain to be addressed .
is the abundance scatter at a specific galactocentric distance real or observational ? why do collisional and recombination lines give abundances which often differ by as much as a factor of two ?
do gradients really flatten in outer regions of disks ? finally , there is the problem of measuring good abundances in low excitation h ii regions , where auroral line electron temperature determinations are difficult / impossible to obtain , preventing adequate probes of early spirals as well as the inner disks of late ones . clearly , there is still plenty of work to do .
cuisinier , f. , acker , a. , & kppen , j. 1996 , , 307 , 215 de vaucouleurs , g. , & pense , w.d .
1978 , , 83 , 1163 daz , a.i .
1989 , in evolutionary phenomena in galaxies , j.e .
beckman & b.e.j .
pagel , eds .
( cambridge university press ) , 377 fesen , r.a . ,
blair , w.p . , & kirshner , r.p .
1985 , , 292 , 29 garnett , d.r . , & shields , g.a .
1987 , , 317 , 82 garnett , d.r .
1989 , , 345 , 282 garnett , d.r . ,
1995 , , 443 , 64 garnett , d.r . ,
1997 , , in press grevesse , n. , & noels , a. 1993 , in origin and evolution of the elements , eds .
n. prantzos , e. vangioni - flam , & m. casse ( cambridge university press ) , 15 henry , r.b.c .
1989 , , 241 , 453 henry , r.b.c . , &
howard , j.w .
1995 , , 438 , 170 henry , r.b.c .
1996 , , 283 , 635 howard , j.w . , & henry , r.b.c .
1997 , , 284 , 465 kaler , j.b . 1980 , , 239 , 78 kennicutt , r.c .
, & garnett , d.r . 1996 , , 456 , 504 kobulnicky , h.a . , & skillman , e.d . 1996 , , 471 , 211 maciel , w.j . , & kppen , j. 1994 , , 282 , 436 martin , p. , & roy , j - r . 1994 , , 424 , 599 mccall , m.l . 1982 , phd thesis , university of texas , austin mcgaugh , s.s . 1994 , , 426 , 135 .
moll , m. , ferrini , f. , & diaz , a.i .
1996 , , 466 , 668 osterbrock , d.e .
1989 , astrophysics of gaseous nebulae and active galactic nuclei , ( university science books ) pagel , b.e.j . ,
1979 , , 189 , 95 peimbert , m. , torres - peimbert , s. , & dufour , r.j .
1993 , , 418 , 760 perinotto , m. 1991 , , 76 , 687 shaver , p.a . , et al . 1983 , , 204 , 53 simpson , j.p . ,
1995 , , 444 , 721 skillman , e.d . ,
1996 , , 462 , 147 smartt , s.j . , & rolleston , r.j .
1997 , , thurston , t.r . ,
edmunds , m.g . , & henry , r.b.c .
1996 , , 283 , 990 vila - costas , m.b . , & edmunds , m.g .
1992 , , 259 , 121 vila - costas , m.b .
, & edmunds , m.g .
1993 , , 265 , 199 vlchez , j.m .
, & esteban , c. 1996 , , 280 , 720 walter , d.k . , dufour , r.j .
, & hester , j.j .
1992 , , 397 , 196 zaritsky , d. , kennicutt , r.c . , &
huchra , j.p .
1994 , , 420 , 87 | galactic and extragalactic abundance studies of h ii regions , planetary nebulae , and supernova remnants are compiled and their implications briefly reviewed .
galaxy global metallicity appears to be related directly to total mass and morphology , while gradient slopes are uncorrelated with these parameters , although barred spirals tend to have flatter profiles than non - barred spirals .
n / o behavior is consistent with primary , then secondary , production of n as metallicity increases , and empirical evidence for a metallicity - sensitive c yield is seen .
metallicity - invariant primary production mechanisms are suggested for s , ne , ar , and o. |
the weak value amplification of the observables is finding increasing number of applications in the study of a variety of physical systems @xcite .
although originally formulated for quantum systems , many past and current applications include applications to classical light beams .
for example the first observation of the weak value amplification was in the context of a gaussian beam propagating through a birefringent medium @xcite .
other important applications of weak value amplification include observation of spin hall effect of light @xcite , goos - hnchen shifts and various generalizations @xcite , angular shifts of light beams @xcite , enhancement of interferometric results @xcite .
weak value amplification has been used to measure the state of polarization of light beam on poincare sphere using optical vortex beams @xcite .
it is intriguing that a concept formulated for quantum systems has so many profound applications in the context of classical light beams .
aiello showed in a formal way how weak value amplification works for beams of light @xcite .
lundeen and coworkers used weak value amplification to get the wavefront of a single photon @xcite .
steinberg @xcite proposed the applications in the measurement of interaction between two fermions .
weak value amplification has been proposed to measure the presence of an additional charge in ahranov bohm interferometer @xcite .
experimentally determined weak values for a single photon s polarization via a weak value amplification @xcite .
starling et .
al . used the weak value amplification to enhance frequency shift resolution in a sagnac interferometer @xcite .
while most examine the amplification of the small shifts , several have examined the question of improvement in sensitivity @xcite of the measurement .
the weak value amplification can worsen the metrological performance @xcite for example the technical noise or decoherence can not be completely overcome @xcite although a significant improvement can be obtained @xcite .
an optimal strategy would be to retain full data @xcite and do a weak measurement .
however advantages of this technique are not exclusive to this technique @xcite . in this paper we show the great advantage offered by weak value amplification for studying quantum mechanical cat states for atoms .
the cat states are the linear superposition of two coherent states on the bloch sphere @xmath1 and @xmath2 represent the probability amplitudes for the atomic system to be in the atomic coherent states @xmath3 and @xmath4 , respectively .
the quantum interferences in cat state are most prominent if the two coherent states are close on the bloch sphere @xcite .
the study of quantum interferences is greatly aided by the weak value amplification otherwise these are difficult to observe .
the weak value amplification gives us the capability to resolve two close lying coherent states .
we look at the interaction of a single photon with an ensemble of atoms prepared in a coherent state @xcite .
the interaction produces an entangled state of the photon polarization variables with the coherent states of the atomic ensemble .
we use preselection and postselection of the polarization states of the photon .
the postselected polarization is nearly orthogonal to the input polarization .
this enables us to magnify the weak values associated with the measurements of the phase @xmath5 .
although in our work we produce cat states by heralding i.e. by detection of a photon , there are many methods to produce cat states @xcite .
the most prominent method is to use atomic systems dispersively interacting with a cavity field @xcite .
the importance of cat states in quantum optics hardly needs to be emphasized as these have important nonclassical properties @xcite and are important in precision measurements @xcite .
the organization of this paper is as follows : in section 2 , we introduce the model of the interacting atom - field system . in section 3
, we make a weak value amplification on the atom - field system so that the post - selected atomic cat state is generated . in section 4 ,
we present the variation of the wigner distribution of the post - selected atomic cat state when the overlap of the initial and final states of the field changes . in section 5 ,
we show that the small phase shift in the post - selected atomic cat state can be amplified by choosing nearly orthogonal pre - selection and post - selection of the single photon field . in this section
we also discuss how the state tomography of the post - selected cat state can be done . in section 6 ,
we discuss the weak value amplification for our atomic cat states using the quantum fisher information and show that the fisher information in the meter and the classical fisher information yields the quantum fisher information of the full meter system state .
this is in agreement with recent conclusions in several papers @xcite .
we conclude our paper in the final section .
we consider an ensemble of @xmath0 identical two - level atoms interacting with two orthogonally polarized modes of a single photon field with frequency @xmath6 denoted by creation ( annihilation ) operators @xmath7 , @xmath8 ( @xmath9 , @xmath10 ) as shown in fig .
[ fig1](a ) .
the two - level atoms have degenerate ground states @xmath11 and excited states @xmath12 , separated by an energy of @xmath13 .
according to the angular - momentum selection rules , the transitions @xmath14 and @xmath15 are forbidden , only the transitions @xmath16 and @xmath17 are allowed .
moreover , the levels @xmath18 and @xmath19 are coupled by the field mode @xmath9 , and the levels @xmath20 and @xmath21 are coupled by the field mode @xmath10 .
their coupling strengthes are @xmath22 and @xmath23 , respectively .
the hamiltonian of the combined system of the atoms and the field @xcite takes the form @xmath24 in eq .
( [ 2 ] ) , the first two terms are the atomic excitation hamiltonian , the third term is the free field hamiltonian , the last two terms are the atom - field interaction hamiltonian . in the dispersive limit in which the detuning between the atomic transitions and the field modes is much larger than the atom - field coupling strengthes i.e. , @xmath25 , the hamiltonian equation ( [ 2 ] ) can be reduced to @xmath26 note that there are no couplings between the ground states @xmath27 ( @xmath28 ) and the excited ones . for simplicity , we assume that the coupling strengths have identical amplitudes @xmath29 . if one starts from the initial ground state , the relevant interaction hamiltonian between the atoms and the field modes is written as @xmath30 where @xmath31 is the field operator , @xmath32 is the collective atomic operator , and @xmath33 is the coupling constant between @xmath34 and @xmath35 .
another important example of optical transitions leading to the interaction hamiltonian ( [ 4 ] ) would be three level atoms @xcite with ground levels @xmath36 coupled to an excited state @xmath37 by a far off resonant field , as shown in fig .
[ fig1](b ) . under the assumption that the frequency separation between @xmath36 is such that the two photon raman coupling between @xmath38 and @xmath39 is negligible ,
the effective interaction between the atoms and the field is described by eq .
( [ 4 ] ) .
in what follows we assume that the initial state of the atomic sample is an atomic coherent state @xmath40 with a definite angular momentum value @xmath41 , which can be created by rotating the ground state @xmath42 on a bloch sphere by an angle @xmath43 around an axis defined as @xmath44 @xcite @xmath45 where @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , and @xmath51 indicates a state for which all atoms are in the ground state .
the atomic coherent state also can be expressed as @xcite @xmath52 where @xmath53 is the simultaneous eigenstate of @xmath54 and @xmath35 , @xmath55 where @xmath56 , @xmath57 , and @xmath0 is a total number of atoms in the sample .
the atomic coherent states are in general not orthogonal except for antipodal points .
the modulus squared of the inner product of two atomic coherent states is @xmath58 is the angle between the directions @xmath59 and @xmath60 on the bloch sphere , and @xmath61 .
for the special cases ( @xmath62 , @xmath63 ; or @xmath64 , @xmath65 ) , @xmath66 , @xmath67 , @xmath68 , the two atomic coherent states are orthogonal .
the atomic coherent states can be produced by classical driving fields of constant amplitude @xcite .
we assume that the initial state of the single photon field is given by @xmath69 and @xmath70 are the probability amplitudes to find the light in the states @xmath71 and @xmath72 , respectively , and @xmath73 .
note that the @xmath74 corresponds to an elliptically polarized light . when @xmath75 or @xmath76 , the @xmath74 is circularly polarized .
when @xmath77 , the light is @xmath78 polarized .
note that single photons can be heralded @xcite or a single nitrogen - vacancy color center in a diamond nanocrystal @xcite .
thus the initial state of atom - field system is @xmath79 , which is a product state of the two subsystems of the field and the atomic ensemble .
we next study the evolution of the system under the interaction ( [ 4 ] ) .
this is given by @xmath80 where @xmath81 is the atom - field interaction time .
expression ( [ 12 ] ) indicates that the atoms are entangled with the single photon field due to the atom - field coupling .
the two coherent states in ( [ 12 ] ) differ in phase by @xmath82 , @xmath83 .
the interaction of the single photon with the medium is expected to be weak and hence @xmath84 is small .
thus the question of distinguishing between two such coherent states arises .
it is here that the weak value amplification offers great advantage .
we make a preselection and postselection of the states of the polarization of the single photon . for simplicity
we assume that initially the field is linearly polarized with @xmath85 i.e. , @xmath86 .
we take the post - selected polarization state as @xmath87 in which @xmath88 is a small angle , and is controlled by the polarizer .
hence the overlap between the pre - selected and post - selected states of the field is @xmath89 which depends on the parameter @xmath88 .
when @xmath90 , @xmath91 , the pre - selected and post - selected states of the field are orthogonal . for small @xmath88 , the pre and post - selected states are nearly orthogonal .
after the weak interaction between the atoms and the field , we project the state of the system @xmath92 onto a final state of the single photon field @xmath93 , we obtain the normalized atomic schrdinger cat state @xmath94,\end{aligned}\ ] ] which is a superposition of two distinct atomic coherent states @xmath95 and @xmath96 rotating in opposite directions in phase space .
thus the weak interaction of the atoms with the single photon field , followed by the state - selective measurement on the single photon field , has transformed the initial atomic coherent state @xmath40 into a post - selected atomic cat state with two components @xmath95 and @xmath96 with small opposite phase shifts @xmath84 . here
@xmath97 is the normalization factor given by @xmath98.\nonumber\\\end{aligned}\ ] ] in what follows we study the post - selected state for @xmath99 and @xmath100 .
this lies in the equatorial plane and has maximum coherence . in spin language
this means that the spin is @xmath78 polarized at @xmath101 .
in this section , we use the wigner function to quantify the nonclassicality of the post - selected atomic cat state .
the wigner function of the atomic cat state is defined as @xcite @xmath102 where @xmath103 is the state - multipole operators @xcite , @xmath104 is the wigner @xmath105 symbol , @xmath106 is an integer taking values 0 , 1 , 2 , @xmath107 , @xmath108 , @xmath109 , and @xmath110 is the spherical harmonic .
it can be interpreted as a quasiprobability distribution in phase space .
it satisfies the normalization condition @xmath111 after some calculations , we obtain the following result for the wigner function @xmath112(-1)^{j - m'}(2k+1)^{\frac{1}{2}}\nonumber\\ & & \times\left(\begin{array}{ccc}j & k & j\\ -m ' & q & m'-q\end{array}\right)^{*}\left(\begin{array}{c}2j\\j+m'-q\\\end{array}\right)^{\frac{1}{2}}y_{kq}(\alpha,\beta)\bigg\},\end{aligned}\ ] ] in which the first term and the second term correspond to the contributions of the individual coherent states @xmath113 and @xmath114 in the atomic cat state , respectively , the last term represents the interference between the two atomic coherent states .
+ we now demonstrate the importance of the weak value amplification and postselections in discriminating two coherent states lying nearby . for illustration purpose
we choose the small phase shift @xmath115 and the total number of the atoms @xmath116 ( @xmath117 ) .
the wigner function for the atomic cat state on the bloch sphere for several different post - selected values of @xmath88 is plotted in fig .
we notice that the post - selected state for @xmath118 can be hardly distinguished from the initial state .
the post selection is useful here .
we show the wigner functions of several post - selected states by choosing @xmath88 values such that the pre - selected and post - selected polarization states are nearly orthogonal .
for decreasing values of @xmath88 one sees more and more negative regions in the wigner function signifying that the post - selected state becomes more and more nonclassical .
the wigner function can be determined by tomographic reconstruction @xcite .
in this section , we show that how the small phase shift @xmath84 in the post - selected atomic cat state induced by the weak interaction of the atomic sample with the single photon field can be measured by studying the phase distribution of the atomic cat state .
the phase distribution @xmath119 for the atomic cat state @xcite , i.e. , the probability distribution @xmath119 of finding the system in the state @xmath120 , is given by @xmath121 the figure [ fig3 ] shows that the phase distribution @xmath119 as a function of @xmath5 for different values of @xmath88 for @xmath122 and @xmath117 . as the value of @xmath88
is decreased from @xmath123 to @xmath124 , the phase distribution @xmath119 gradually changes from a single peak at @xmath125 to two symmetric peaks with a dip at @xmath125 , which is due to the destructive interference between two weakly separated coherent states in the atomic cat state .
the scaled left peak shift @xmath126 from @xmath125 as a function of @xmath88 is shown in fig .
it is noted that the scaled peak shift @xmath126 increases with decreasing @xmath88 . when @xmath127 , @xmath128 , the amplification factor for the weak value @xmath84 is about 15 .
if there is 100 atoms in the atomic sample ( @xmath129 ) , the phase distribution @xmath119 as a function of @xmath5 for different values of @xmath88 is shown in fig .
the scaled left peak shift @xmath126 from @xmath125 as a function of @xmath88 is shown in fig .
[ fig4 ] . as the value of @xmath88
is decreased , the scaled left peak shift @xmath126 increases . when @xmath127 , @xmath130 , the amplification factor for the weak value @xmath84 is about 5.8 .
note that the amplification factor for the weak value @xmath84 decreases with increasing the total number @xmath0 of the atoms .
this is because the modulus squared of the overlap of the two atomic coherent states @xmath113 and @xmath114 decreases when the total number @xmath0 of the atoms becomes larger , which is shown below . using eq .
( [ 10 ] ) , one gets latexmath:[\[\label{22 } when @xmath122 , one has @xmath132 and @xmath133 .
we now find the observable whose weak value is being displayed in figs .
[ fig3][fig5 ] . to see this we write the numerator in (
[ 21 ] ) as @xmath134 where @xmath135 to simplify ( [ 24 ] ) for small enough @xmath84 , we do a taylor series expansion @xmath136[f(\varphi)+\omega f'(\varphi)a],\label{26}\\ a&=&\frac{-\sin(\gamma-\frac{\pi}{4})+\cos(\gamma-\frac{\pi}{4})}{\sin(\gamma-\frac{\pi}{4})+\cos(\gamma-\frac{\pi}{4})}=\cot(\gamma).\label{27}\end{aligned}\ ] ] thus we can rewrite ( [ 26 ] ) as @xmath137.\label{28}\end{aligned}\ ] ] the expansion ( [ 28 ] ) is valid unless @xmath88 is very small .
thus the shift of the distribution is @xmath138 where @xmath139 is the weak value of the observable defined by the stokes operator for the photon field @xmath140 where @xmath74 and @xmath93 are defined respectively by eqs .
( [ 11 ] ) and ( [ 13 ] ) . here
@xmath74 is the initial state of the photon and @xmath93 is the post - selected state of the photon . before concluding this section we present a method of studying the post - selected cat state ( [ 15 ] ) .
the best way to study this cat state is via the distributions of the spin variables .
note that the basic interaction ( [ 4 ] ) produced the heralded cat state ( [ 15 ] ) by using a single photon .
now imagine that after heralding a second classical field with linear polarization is sent through the ensemble , then the polarization of the classical field will be rotated .
this rotation of polarization will depend on the quantum variable @xmath35 .
thus the studies in the fluctuations of the polarization of the classical field applied after initial heralding will measure the quantum characteristics of the cat state .
such techniques were initially pioneered by julsgaard et .
al . @xcite and
more recently applied to heralded cat states by mcconnell et .
further information on the cat state ( [ 15 ] ) can be obtained by using a rf field so as to work in a different basis @xcite .
this way one can do a complete tomography of the state .
the precision in estimating the parameter @xmath84 is limited by the quantum cramer - rao bound , which gives the minimum achievable variance @xmath141 , depending on the quantum fisher information @xcite . in this section ,
we examine the question of the estimation of the parameter @xmath84 via the quantum fisher information .
the utility of the weak value amplification in comparison of the standard metrological protocols has been questioned @xcite .
however , alves et .
@xcite have shown that for most preselected states , full information on the coupling constant ( analog of @xmath84 ) can be obtained from the meter data while for a small fraction of the pre - selected states , the full information must be obtained from the post - selection statistics . in a comprehensive paper jordan et .
al . @xcite have shown theoretically that in presence of many different sources of noise , the weak value amplification outperforms the standard metrological protocol .
these theoretical results have been given experimental support @xcite .
the quantum fisher information @xmath142 in the state ( [ 12 ] ) defined by @xmath143\end{aligned}\ ] ] is given by @xmath144 the quantum fisher information in the post - selected state ( [ 15 ] ) is given by @xmath145\nonumber\\ & = & \frac{0.5n}{p}\big[1+\cos(2\gamma)\cos^{n-2}\omega(1-n\sin^{2}\omega)\nonumber\\ & & -\frac{0.5n}{p}\cos^{2}(2\gamma)\cos^{2n-2}\omega\sin^{2}\omega\big],\end{aligned}\ ] ] and @xmath146,\end{aligned}\ ] ] where @xmath147 is the probability of success .
the classical fisher information in the post - selected process is defined by @xmath148 we find @xmath149 in figs .
[ 6]-[8 ] , we show the quantum fisher information @xmath150 of the atomic cat state , the probability @xmath147 of the successful post selection , and the classical fisher information @xmath151 in the post - selected process as a function of @xmath88 for @xmath122 and @xmath129 . first of all , we note that although @xmath152 ( eq . (
[ 32 ] ) ) , as expected from the analysis of refs .
@xcite . on the other hand
, the missing information can be obtained from the study of both @xmath150 and @xmath151 as evidenced from figs .
[ 6]-[8 ] .
this has been emphasized in refs .
@xcite , see especially fig . 2 in ref .
we have demonstrated how the nonclassical interference between two slightly separated atomic coherent states in a post - selected atomic cat state can be observed via making a weak value amplification on a system containing @xmath0 two - level atoms weakly coupled to a single photon field .
we find that the negative region of the wigner distribution of the atomic cat state is increased when the pre - selected and post - selected states of the single photon field are closer to orthogonal .
we show that the weak value amplification can lead to peak shift in the phase distribution of the atomic cat state that can be 15 times [ for 10 atoms ] the phase shift in the atomic cat state when the initial and final states of the signal photon field are nearly orthogonal .
we have presented an experimental scheme to do tomography of the post - selected cat state .
we have also discussed different aspects of the quantum and the classical fisher information for the states of the previous sections .
this work was supported by the singapore national research foundation under nrf grant no .
nrf - nrff 2011 - 07 and the oklahoma state university .
gsa thanks the hospitality of the director , tata institute of fundamental research where this work was completed .
he also thanks a. jordan for discussions on weak values .
aharonov y , albert d z and vaidman l 1988 _ phys .
lett . _ * 60 * 1351 duck i m , stevenson p m and sudarshan e c g 1989 _ phys .
d _ * 40 * 2112 leggett a j 1989 _ phys .
lett . _ * 62 * 2325 dressel j , malik m , miatto f m , jordan a n and boyd r w 2014 _ rev .
phys . _ * 86 * 307 kofman a g , ashhab s and nori f 2012 _ phys .
rep . _ * 520 * 43 ritchie n w m , story j g and hulet r g 1991 _ phys .
lett . _ * 66 * 1107 hosten o and kwiat p 2008 _ science _ * 319 * 787 jayaswal g , mistura g and merano m 2013 _ opt . lett . _
* 38 * 1232 jayaswal g , mistura g and merano m 2014 _ opt .
lett . _ * 39 * 2266 goswami s , pal m , nandi a , panigrahi p k and ghosh n 2014 _ opt .
* 39 * 6229 magana - loaiza o s , mirhosseini m , rodenburg b and boyd r w 2014 _ phys .
. lett . _ * 112 * 200401 salazar - serrano l j , valencia a and torres j p 2014 _ opt .
* 39 * 4478 brunner n and simon c 2010 _ phys .
lett . _ * 105 * 010405 strbi g and bruder c 2013 _ phys .
_ * 110 * 083605 kobayashi h , nonaka k and shikano y 2014 _ phys .
a _ * 89 * 053816 jordan a n , tollaksen j , troupe j e , dressel j and aharonov y 2015 _ quantum stud . : math . found . _ * 2 * 5 zhang l , datta a and walmsley i a 2015 _ phys .
lett . _ * 114 * 210801 combes j , ferrie c , jiang z and caves c m 2014 _ phys .
a _ * 89 * 052117 tanaka s and yamamoto n 2013 _ phys .
a _ * 88 * 042116 jordan a n , martnez - rincn j and howell j c 2014 _ phys .
4 * 011031 viza g i , martnez - rincn j , alves g b , jordan a n and howell j c 2014 arxiv : 1410.8461 bi alves g , escher b m , de matos filho r l , zagury n and davidovich l 2015 _ phys . rev . a _ * 91 * 062107 calsamiglia j , gendra b , munoz - tapia r and bagan e 2014 arxiv : 1407.6910 knee g c , briggs g a d , benjamin s c and gauger e m 2013 _ phys . rev . a _ * 87 * 012115 knee g c and gauger e m 2014 _ phys .
x _ * 4 * 011032 gerry c c and grobe r 1997 _ phys . rev .
a _ * 56 * 2390 agarwal g s , puri r r and singh r p 1997 _ phys .
a _ * 56 * 2249 mcconnell r , zhang h , uk s , hu j , schleier - smith m h and vuleti v 2013 _ phys
a _ * 88 * 063802 agarwal g s , lougovski p and walther h 2005 _ j. mod
. opt . _ * 52 * 1397 rauschenbeutel a , bertet p , osnaghi s , nogues g , brune m , raimond j m and haroche s 2001 _ phys .
a _ * 64 * 050301(r ) brune m , hagley e , dreyer j , matre x , maali a , wunderlich c , raimond j m and haroche s 1996 _ phys . rev . lett . _ * 77 * 4887 gerry c c and grobe r 1998 _ phys .
a _ * 57 * 2247 davidovich l , brune m , raimond j m and haroche s 1996 _ phys .
a _ * 53 * 1295 lau h w , dutton z , wang t and simon c 2014 _ phys . rev . lett . _ * 113 * 090401 opatrn t and mlmer k 2012 _ phys .
a _ * 86 * 023845 dooley s and spiller t p 2014 _ phys .
a _ * 90 * 012320 bhaktavatsala rao d d , bar - gill n and kurizki g 2011 _ phys . rev .
_ * 106 * 010404 leibfried d , knill e , seidelin s , britton j , blakestad r b , chiaverini j , hume d b , itano w m , jost j d , langer c , ozeri r , reichle r and wineland d j 2005 _ nature _ * 438 * 639 arecchi f t , courtens e , gjtlmore r and thomas h 1972 _ phys .
a _ * 6 * 2211 recamier j , castaos o , juregui r and frank a 2000 _ phys
a _ * 61 * 063808 recamier j , castaos o , juregui r and frank a 2000 _ int . j. quantum chem . _ * 80 * 1129 brida g , genovese m , gramegna m and predazzi e 2004 _ phys . lett .
a _ * 328 * 313 babichev s a , brezger b and lvovsky a i 2004 _ phys .
lett . _ * 92 * 047903 allaume r , treussart f , messin g , dumeige y , roch j f , beveratos a , brouri - tualle r , poizat j p and grangier p 2004 _ new j. phys . _
* 6 * 92 beveratos a , khn s , brouri r , gacoin t , poizat j p and grangier p 2002 _ eur .
j. d _ * 18 * 191 agarwal g s 2012 _ quantum optics _ ( cambridge university press ) sec .
11.7 agarwal g s 1981 _ phys .
* 24 * 2889 agarwal g s 1998 _ phys .
a _ * 57 * 671 auccaise r , araujo - ferreira a g , sarthour r s , oliveira i s , bonagamba t j and roditi i 2015 _ phys .
. lett . _ * 114 * 043604 signoles a , facon a , grosso d , dotsenko i , haroche s , raimond j m , brune m and gleyzes s 2014 nat . phys .
* 10 * 715 julsgaard b , kozhekin a and polzik e s 2001 _ nature _
* 413 * 400 mcconnell r , zhang h , hu j , uk s and vuleti v 2015 _ nature _ * 519 * 439 hofmann h f , goggin m e , almeida m p and barbieri m 2012 _ phys . rev .
a _ * 86 * 040102 ( r ) | we show the utility of the weak value amplification to observe the quantum interference between two close lying atomic coherent states in a post - selected atomic cat state , produced in a system of @xmath0 identical two - level atoms weakly interacting with a single photon field . through the observation of the negative parts of the wigner distribution of the post - selected atomic cat state , we find that the post - selected atomic cat state becomes more nonclassical when the post - selected polarization state of the single photon field tends toward becoming orthogonal to its pre - selected state .
we show that the small phase shift in the post - selected atomic cat state can be amplified via measuring the peak shift of its phase distribution when the post - selected state of the single photon field is nearly orthogonal to its pre - selected state .
we find that the amplification factor of 15 [ 5 ] can be obtained for a sample of 10 [ 100 ] atoms .
this effectively provides us with a method to discriminate two close lying states on the bloch sphere .
we discuss possible experimental implementation of the scheme , and conclude with a discussion of the fisher information .
+ keywords : weak value amplification , the atomic cat state , the wigner distribution , the phase distribution , interference , the fisher information |
recently , verlinde @xcite presented a remarkable new idea that gravity can be explained as an entropic force caused by the information changes when a material body moves away from the holographic screen .
this idea implies that gravity is not fundamental . with the holographic principle and the equipartition law of energy
, verlinde showed that newton s law of gravitation can arise naturally and unavoidably in a theory in which space is emergent through a holographic scenario , and a relativistic generalization leads to the einstein equations .
in fact , the similar idea can be traced back to sakharov s work @xcite . on the other side , using the equipartition law of energy for the horizon degrees of freedom together with the thermodynamic relation @xmath3 , padmanabhan also obtained the newton s law of gravity @xcite .
subsequently , with the idea of entropic force , some applications have been carried out .
the friemdann equations and the modified friedmann equations for friedmann - robertson - walker universe in einstein gravity @xcite , @xmath4 gravity @xcite , deformed horava - lifshitz gravity @xcite , and braneworld scenario @xcite were derived with the help of holographic principle and the equipartition rule of energy .
the newtonian gravity in loop quantum gravity was derived in ref .
@xcite . in ref .
@xcite the coulomb force was regarded as an entropic force . in ref .
@xcite , it was shown that the holographic dark energy can be derived from the entropic force formula .
it was pointed out in @xcite that verlinde s entropic force is actually the consequence of a specific microscopic model of spacetime .
the similar ideas were also applied to the construction of holographic actions from black hole entropy @xcite . while ref .
@xcite showed that gravity has a quantum informational origin . on the other hand ,
a modified entropic force in the debye s model was gave in @xcite . in ref .
@xcite , some of the problems of verlinde s proposal on the thermodynamical origin of the principle of relativity was presented , and the thermodynamic origin of the principle of relativity was explained by hidden symmetries of thermodynamics .
other applications can be seen in @xcite . according to verlinde s idea ,
the holographic screens locate at equipotential surfaces , where the potential is defined by a timelike killing vector .
then the local temperature on a screen can be defined by the acceleration of a particle that is located very close to the screen .
the energy on the screen is calculated by the holographic principle and the equipartition rule of energy @xmath5 with @xmath6 the bit density of information on the screen . in this paper
, we apply these formulas to investigate the temperature and energy on holographic screens for 4-dimensional static spherically symmetric and the kerr black hole , and look for experiment methods for testing the entropic force .
the paper is organized as follows . in sec .
[ sec2 ] , we review verlinde s idea about the temperature and the energy from an entropic force in general relativity .
then , with the idea of entropic force together with the equipartition law of energy , we calculate the temperature and the energy for 4-dimensional static spherically symmetric black holes and the kerr black hole in sec .
[ sec3 ] and sec .
[ sec4 ] , respectively .
finally , we give a brief discussion and present that the entropic force could be tested by experiments .
we first review the idea of verlinde about the temperature and the energy from an entropic force in general relativity .
consider a static background with a global time like killing vector @xmath7 .
one can relate the choice of this killing vector field with the temperature and the energy . in order to define the temperature ,
we first need to introduce the potential @xmath8 via the timelike killing vector @xmath7 : @xmath9 where @xmath10 satisfies the killing equation @xmath11 the redshift factor is denoted by @xmath12 and it relates the local time coordinate to that at a reference point with @xmath13 , which we will take to be at infinity .
the potential is used to define a foliation of space , and the holographic screens are put at surfaces of constant redshift .
so the entire screen has the same time coordinate .
the four velocity @xmath14 and the acceleration @xmath15 of a particle that is located very close to the screen can be expressed in terms of the killing vector @xmath7 as @xmath16 note that because the acceleration is the gradient of the potential , it is perpendicular to screen @xmath17 .
so we can turn it into a scalar quantity by contracting it with a unit outward pointing vector @xmath18 normal to the screen @xmath17 and to @xmath7 . with the normal vector @xmath19 , the local temperature @xmath20 on the screen
is defined by @xmath21 where a redshift factor @xmath12 is inserted because the temperature @xmath20 is measured with respect to the reference point at infinity .
we will call the temperature defined in ( [ t ] ) as unruh - verlinde temperature .
assuming that the change of entropy at the screen is @xmath22 for a displacement by one compton wavelength normal to the screen , one has @xmath23 now the entropic force , which is required to keep a particle with mass @xmath24 at fixed position near the screen , is turned out to be @xmath25 where @xmath26 is the relativistic analogue of newton s acceleration , and the additional factor @xmath12 is due to the redshift , which is a consequence of the entropy gradient .
we now consider a holographic screen on a closed surface of constant redshift @xmath8 .
the number of bit @xmath27 of the screen is assumed to proportional to the area of the screen and is given by @xcite @xmath28 then , by assuming that each bit on the holographic screen contributes an energy @xmath29 to the system , and by using the equipartition law of energy , we get @xmath30 with @xmath6 the bit density on the screen . inserting the expressions of @xmath20 and @xmath27 into ( [ energy_a ] ) results in @xmath31 in the following , we will calculate the unruh - verlinde temperature and the energy on holographic screens for 4-dimensional black holes with static spherically symmetric metric and stationary axisymmetric metric , respectively .
we first consider the case of 4-dimensional general static spherically symmetric black holes . in the general case
, the metric can be taken as the form @xmath32 where @xmath33 is the metric on a unit 2-sphere .
we assume @xmath34 for ensuring that the metric is asymptotically flat .
the event horizon radius @xmath35 is usually determined by the largest solution of @xmath36 . by using the killing equation @xmath37 and the static spherically symmetric properties of the metric ( [ general1 ] ) @xmath38 as well as the condition @xmath39 at infinity ,
the timelike killing vector of the general black hole is solved as @xmath40 which is zero at the event horizon . according to ( [ phi ] ) , ( [ amu ] ) and
( [ t ] ) , the potential , the acceleration and the unruh - verlinde temperature on the holographic screen put at the spherical surface with radius @xmath41 are calculated as in the first version of our manuscript . ]
@xmath42 the energy on the screen is @xmath43 next , we focus on the schwarzschild black hole and the rn black hole . for the schwarzschild black hole ,
the metric has the form of ( [ general1 ] ) with @xmath44 the event horizon of the balk hole is located at @xmath45 , and the hawking temperature @xmath46 on the event horizon is given by @xmath47 the acceleration and the unruh - verlinde temperature on the screen are calculated as @xmath48 note that the unruh - verlinde temperature on the event horizon @xmath49 is just the hawking temperature @xmath46 : @xmath50 then , with the expressions ( [ e_spherically ] ) and ( [ f(r)_schwarz ] ) , we obtain the energy on the screen : @xmath51 it is interesting to note that this energy is independent of the radius of the screen .
in fact this is a consequence of the gauss law . as an important example of spherically symmetric black holes , the rn black hole has the metric form of ( [ general1 ] ) with @xmath52 there are two horizons , the event horizon with radius @xmath53 and the cauchy horizon with radius @xmath54 , where @xmath55 the extremal rn black hole corresponds to the case @xmath56 or , equivalently , @xmath57 .
the hawking temperatures @xmath58 on the horizons are @xmath59 the potential , the acceleration and the unruh - verlinde temperature on the screen read @xmath60 on the horizons @xmath61 , we have @xmath62 which shows that the unruh - verlinde temperatures on both the horizons equal to the hawking temperatures @xmath58 . the solution of @xmath63 is @xmath64 .
so for @xmath65 , i.e. , @xmath66 , the maximum of @xmath20 in the range @xmath67 is at @xmath68 . for @xmath69 , the maximum of @xmath20 in the range @xmath67 is at @xmath70 and is given by @xmath71 the energy on the screen is @xmath72 which is the well - known komar energy of the rn black hole .
now this energy is dependent on the radius of the screen , which is also a consequence of the gauss laws of gravity and electrostatics .
for the screen at infinity , the energy is reduced to that of the schwarzschild black hole : @xmath73 .
it is interesting to note that the energies on the horizons @xmath74 are the same : @xmath75 for the extremal rn black hole with @xmath76 , the corresponding unruh - verlinde temperature and energy are @xmath77 they vanish at the horizon @xmath78 .
this result is in agreement with the statements that the extremal black hole has a unique internal state @xcite and its temperature is zero due to the vanishing of surface gravity on horizon .
in this section , we extend this work to the kerr black hole .
the kerr solution describes both the stationary axisymmetric asymptotically flat gravitational field outside a massive rotating body and a rotating black hole with mass and angular momentum .
the kerr black hole can also be viewed as the final state of a collapsing star , uniquely determined by its mass and rate of rotation . moreover ,
its thermodynamical behavior is very different from the schwarzschild black hole and the rn black hole , because of its much more complicated causal structure .
hence its study is of great interest in understanding physical properties of astrophysical objects , as well as in checking any conjecture about thermodynamical properties of black holes . in terms of boyer
lindquist coordinates , the euclidean kerr metric reads @xmath79 where @xmath80 is the kerr horizon function @xmath81 and @xmath82 here @xmath83 is the angular momentum for unit mass as measured from the infinity ; it vanishes in the schwarzschild limit .
the nonextremal kerr black hole has the event horizon @xmath84 and the cauchy horizon @xmath54 at @xmath85 the extreme case corresponds to @xmath86 or @xmath87 .
the hawking temperatures on the horizons are @xmath88 the solution for the timelike killing vector @xmath10 is read as @xcite @xmath89 where @xmath90 .
the corresponding potential is @xmath91 which shows that equipotential surfaces are dependent with two parameters @xmath41 and @xmath92 , hence the holographic screens are not spherically symmetric but axisymmetric .
however , when @xmath93 or @xmath94 , the holographic screens are approximate spherically symmetric , which can be seen from the contour plots of the potential @xmath8 in the @xmath95-@xmath96 plane showed in figs . [
fig : contourphia ] and [ fig : contourphib ] .
it can be seen that , when @xmath97 , the effect of the angular momentum are remarkable near the horizons . when far away from the event horizon , the equipotential line can be approximated as a spherical surface .
-4 mm -4 mm the non - zero components of the acceleration are @xmath98 } , \nonumber \\
a^\theta & = & \frac{mra^2(a^2+r^2)\sin(2\theta ) } { \varrho^4[2mr(a^2+r^2)+\delta\varrho^2]}. \label{a_kerr}\end{aligned}\ ] ] the unruh - verlinde temperature is given by @xmath99 ^ 2 } { \varrho ^4 \left[2 m r \left(a^2+r^2\right ) + \delta
\varrho ^2\right]^3 } \right . \nonumber \\ & & \quad\quad~ + \left . \frac{\delta \left[a^2 m r ( a^2+r^2 ) \texttt{sin}(2\theta ) \right]^2 } { \varrho ^4 \left[2 m r \left(a^2+r^2\right ) + \delta
\varrho ^2\right]^3 } \right\ } ^{\frac{1}{2 } } .
\label{t_kerr}\end{aligned}\ ] ] on the horizons @xmath61 ( @xmath100 ) , the unruh - verlinde temperatures become @xmath101 so , the unruh - verlinde temperatures on both horizons are equal to the hawking temperatures . for vanishing @xmath83 ,
the above result is reduced to @xmath102 , which is just the hawking temperature ( [ t_schwarzschild ] ) on the horizon of the schwarzschild black hole .
contour plots of the temperature @xmath20 in the @xmath95-@xmath96 plane are shown in figs .
[ fig : contourta ] and [ fig : contourtb ] .
-4 mm -4 mm now , by using the equipartition law of energy and the holographic principle @xmath103 , we get the energies on the horizons @xmath104 which is the reduced mass @xmath1 of the kerr black hole .
when far away from the event horizon , or equivalently @xmath93 , the screen is approximated as a sphere and the energy can be expressed as @xmath105 for @xmath106 , the result is @xmath73 , which is independent of the angular of the kerr black hole . for @xmath107 , the expression ( [ e_kerr_approx ] )
reduce to @xmath73 , which is just the energy expression ( [ e_schwarzschild ] ) for the schwarzschild black hole .
in this paper , with the holographic principle and the equipartition law of energy , we investigate the unruh - verlinde temperature and energy on holographic screens for several 4-dimensional black holes with static spherically symmetric metric and stationary axisymmetric metric . on the event horizon of a static spherically symmetric black hole
, the hawking temperature @xmath46 is related to the surface gravity @xmath108 by the relation @xmath109 , while the unruh - verlinde temperatures @xmath20 is related to the acceleration @xmath110 by @xmath111 , which shows that the unruh - verlinde temperatures @xmath20 for a static spherically symmetric black hole is identical to the hawking temperature .
hence , outside the horizons , the unruh - verlinde temperature @xmath20 can be considered as a generalized hawking temperature on the holographic screen . for the kerr black case , we also consider @xmath20 as the generalized hawking temperature . on the screen located at the infinity , the surface gravity or the acceleration of a particle ( which is free there ) vanishes , hence the temperature is zero . the entropic force ( [ fmu ] ) for a static spherically symmetric black hole can be calculated as @xmath112 .
the magnitude of the force is @xmath113 . for the schwarzschild black hole ,
the entropic force @xmath114 is just the newton force @xmath115 here we recover the newton s gravitational constant @xmath116 . while , for the rn case , the entropic force is turned out to be @xmath117 which shows that there is a corrected term caused by the charge of the black hole .
this force is related to the pure gravitational effect .
the gravity or the geometry near the event horizon of the black hole is affected by the energy of the electric field .
when far away form the horizon , where the electric field becomes weak , the entropic force will recover to the newton one .
note that , since @xmath118 , we have @xmath119 and @xmath120 . similarly , the angular momentum of the kerr black hole will also affect the structure of the spacetime near the event horizon .
the entropic force of the kerr black hole is read as @xmath121 and @xmath122 at @xmath68 and @xmath93 , respectively .
it also reduces at large @xmath41 to the newton force because the effect of the field " caused by the angular momentum of the black hole becomes weaker with the increase of @xmath41 .
the energy got from the holographic principle is indeed the komar energy on the screen . with the same explanation above
, we also can understand that the energy @xmath123 on the screen will trend to the adm mass @xmath0 when @xmath106 for anyone of the black holes considered in the paper .
this is because the field " produced by the charge or the angular momentum is strong on the screen near the event horizon and weak when far away from the black hole . on the screen at infinity , the effect of them vanishes and
the energy is naturally equal to the black hole mass @xmath0 according to the gravitational gauss law .
we see clearly from ( [ e_rn_h ] ) and ( [ e_kerr_h ] ) that when applying the entropic force idea to the black hole horizon , it is the reduced mass @xmath1 that takes the place of the black hole mass @xmath0 . for the schwarzschild black hole
, there is no other parameter beside @xmath0 , so the reduced mass @xmath1 is just the black hole mass @xmath0 . for the rn black hole , the reduced mass is read from ( [ e_rn ] ) as @xmath124 , which is @xmath125 on the event horizon . for the kerr case ,
the reduced mass is @xmath126 on the event horizon and is @xmath127 at @xmath93 .
therefore , with the black hole mass @xmath0 replaced by the reduced mass @xmath1 , we can rewrite the entropic force of a black hole as @xmath128 which has the same form as the newton s law of gravity .
this formula shows that the force between two neutral particles is exactly the newton s force of gravity , while the force between a neutral particle and a charged particle would departure from the newton s law of gravity .
hence , the idea of entropic force could be tested by experiments according to eq .
( [ f_entropic ] ) .
explicitly , the force between a neutral particle and a charged particle is @xmath129 according entropic force idea . therefore , by comparing the high accurate experiment results of the force between a neutral particle and a charged particle with the theoretical predictions of the entropic force , one could check whether the entropic idea is right or wrong .
this work was supported by the program for new century excellent talents in university , the national natural science foundation of china ( no . 10705013 ) , the key project of chinese ministry of education ( no . 109153 ) , and the fundamental research funds for the central universities ( no . lzujbky-2009 - 54 , no . lzujbky-2009 - 122 and no .
lzujbky-2009 - 163 ) . | we investigate the temperature and energy on holographic screens for 4-dimensional black holes with the entropic force idea proposed by verlinde .
we find that the `` unruh - verlinde temperature '' is equal to the hawking temperature on the horizon and can be considered as a generalized hawking temperature on the holographic screen outside the horizons .
the energy on the holographic screen is not the black hole mass @xmath0 but the reduced mass @xmath1 , which is related to the black hole parameters . with the replacement of the black hole mass @xmath0 by the reduced mass @xmath1
, the entropic force can be written as @xmath2 , which could be tested by experiments . |
it is a truth universally acknowledged , that random objects are asymmetric .
it was shown by wright @xcite that for @xmath11 a random @xmath8 graph has , whp , a trivial automorphism group .
he actually worked with the @xmath12 model , but the reduction to @xmath8 is well - known and follows easily from the chernoff bound .
also , a graph and its complement clearly have the same automorphism group , so we can restrict ourselves to the range @xmath13 .
wright s bound is tight , since a graph @xmath5 of slightly smaller density is likely to have isolated vertices , which can be swapped by a @xmath5-automorphism . this paper concerns the range of smaller @xmath7 by showing that for @xmath14 whp all of @xmath5 s automorphisms are essentially trivial . here is our main result : theoremmaintheorem[thm : maintheorem ] let @xmath15 be a @xmath8 graph with @xmath16 . then whp its @xmath6-core has a trivial automorphism group .
this shows that for this range of @xmath7 , whp @xmath3 is generated by : * automorphisms of rooted trees that are attached to the @xmath6-core .
* automrophisms of the tree components and swaps of such components .
the most interesting range of this statement is @xmath17 . for larger @xmath7
the @xmath6-core is the whole graph , in which range ours is just a new proof for the rigidity of sufficiently dense random graphs .
* general strategy of the proof : * we denote the vertex set of @xmath5 s @xmath6-core by @xmath18 .
it is easy to see that @xmath3 fixes @xmath18 setwise and our proof shows first that @xmath3 actually fixes @xmath18 pointwise . in order to prove the theorem in full we show that this rigidity does not result from boundary effects of vertices near @xmath19 .
the neighbor set of @xmath20 and its degree are denoted by @xmath21 and @xmath22 .
if @xmath23 are the neighbors of @xmath24 , we denote by @xmath25 the _ multiset _ @xmath26 .
clearly @xmath27 is preserved by automorphisms .
we fix some @xmath28 and consider two directed rooted cycles @xmath29 and @xmath30 in @xmath5 .
we show that whp every two such cycles have many _ incompatible _ pairs @xmath31 for which @xmath32 .
this already implies that @xmath18 is fixed pointwise . in the full proof of the theorem we find , for every two such cycles , an incompatible pair @xmath31 , where both @xmath33 and @xmath34 are at distance @xmath35 from @xmath19 . such a pair is not only incompatible in @xmath5 , but also in @xmath18 , proving the theorem .
it turns out that theorem [ thm : maintheorem ] yields some interesting insights on the well - known graph reconstruction conjecture which we now recall .
let @xmath5 be an @xmath36-vertex graph .
when we delete a vertex of @xmath5 we obtain an @xmath37-vertex graph . by doing this separately for each vertex in @xmath5 we obtain the @xmath36 graphs that make up @xmath5 s _ deck_. the _ graph reconstruction conjecture _ ( @xcite , @xcite ) posits that every two graphs of @xmath38 or more vertices that have identical decks must be isomorphic . a graph @xmath5 is said to be _
reconstructible _ if every graph with the same deck is isomorphic to @xmath5 .
bollobs proved @xcite that whp @xmath8 graphs are reconstructible for all @xmath39 .
we show that this is in fact true for every @xmath40 .
one reason why this extension of range is of interest has to do with the _ edge reconstruction conjecture _ @xcite which states that every graph can be reconstructed from its deck of edge - deleted subgraphs .
this leads to the notion of _ edge - reconstructible _ graphs .
we recall two facts from this theory : ( i ) every reconstructible graph with no isolated vertices is edge - reconstructible ( e.g. , @xcite ) ( ii ) every @xmath36-vertex graph with at least @xmath41 edges is edge reconstructible .
our result applies to the range @xmath42 where the edge reconstruction problem is still open .
we turn to discuss the _
canonical labeling problem _ @xcite .
let @xmath43 be a class of graphs .
a canonical labeling of @xmath44 assigns distinct labels to the vertices of @xmath5 , where the labeling is uniquely determined by @xmath5 s isomorphism class . in the probabilistic version of this problem , @xmath43 is a probability space of graphs and we seek to efficiently find a canonical labeling for almost all graphs in @xmath43 .
such a canonical labeling algorithm clearly solves in particular the _ random graph isomorphism problem _ for @xmath43 .
specifically we ask for which values of @xmath7 there is a polynomial time canonical labeling in @xmath8 . by considering the complementary graph it suffices to consider the range @xmath45 . such an algorithm is known @xcite for @xmath46 $ ] .
our proof of theorem [ thm : maintheorem ] yields a polynomial time algorithm for @xmath47 , whence a polynomial time solution exists for @xmath48 $ ] .
[ [ graph - theory ] ] graph theory : + + + + + + + + + + + + + graphs are denoted @xmath15 and usually @xmath49 .
the neighbor set of @xmath50 is denoted by @xmath51 . for @xmath52 ,
we denote @xmath53 and @xmath54 .
the set of cross edges between two subsets @xmath55 is denoted @xmath56 , and @xmath57 ( to wit : even if @xmath58 , we consider every relevant edge exactly once ) . for a singleton @xmath59
, we use the shorthand @xmath60 . also , @xmath61 . for @xmath52
we denote @xmath62 , the set of those vertices not in @xmath63 that have exactly one neighbor in @xmath63 .
we denote by @xmath64 the subgraph of @xmath5 induced by @xmath52 .
let @xmath65 denote the _ multiset of integers _ @xmath66 .
we denote the vertex set of @xmath5 s @xmath6-core by @xmath18 .
[ [ asymptotics ] ] asymptotics : + + + + + + + + + + + + a property of @xmath8 graphs is said to hold _ whp _ ( _ with high probability _ ) if its probability tends to @xmath67 as @xmath68 .
[ [ probability ] ] probability : + + + + + + + + + + + + for a discrete random variable @xmath69 , let @xmath70 if @xmath69 is multinomial with parameters @xmath71 , we denote @xmath72 by @xmath73 .
the following lemmas provide a description of @xmath73 .
[ lem : multinomialmaximizer]let @xmath69 be a multinomial random variable with parameters @xmath71 and suppose that @xmath74
. then , @xmath75 for every @xmath76 , or , in other words @xmath77 . without loss of generality ,
assume by contradiction that @xmath78 .
then , since @xmath79 , there exists some index @xmath80 , say @xmath81 such that @xmath82 .
@xmath83 contrary to the assumed maximality of @xmath84 .
[ lem : multinomialbound]for an integer @xmath85 , a constant @xmath86 and a probability vector @xmath87 , such that @xmath88 for each @xmath89 , it holds that @xmath90 we first show how to reduce the proof to the case where @xmath91 for each @xmath89 .
assume that the lemma holds in this case .
for a real vector @xmath92 and two coordinate indices @xmath93 , let @xmath94 be the vector obtained by eliminating the coordinates @xmath95 and introducing a new coordinate of @xmath96 .
let @xmath69 and @xmath97 be multinomial random variables with parameters @xmath98 , @xmath99 , respectively .
note that for every @xmath100 there holds @xmath101 thus , @xmath102 .
we generate a sequence of probability vectors that start from @xmath103 and proceed as follows . at each step
we replace , as described , the two smallest coordinates in the present probability vector by one coordinate that is their sum .
we continue with this process until the first time at which this vector @xmath104 has at most one coordinate that is smaller than @xmath105 .
if the smallest coordinate in @xmath104 is @xmath106 , then , since each of the above steps can only increase @xmath107 , the reduction is complete . otherwise , @xmath104 has exactly one coordinate , say @xmath108 that is @xmath109 .
but then in @xmath110 all coordinates vary between @xmath105 and @xmath111 .
the reduction is again complete .
we now turn to proving the lemma for the case where @xmath112 for @xmath113 .
clearly @xmath114 .
set @xmath115 and suppose that @xmath116 by lemma [ lem : multinomialmaximizer ] , @xmath117 for all @xmath89 .
now @xmath118 by stirling s bound , @xmath119 .
thus , @xmath120 which can be stated as @xmath121 but @xmath122 and @xmath123 therefore , @xmath124 [ lem : multinomialwithbinomialprobsbound ] let @xmath125 be an integer , @xmath126 and let @xmath127 , where @xmath128 . then , for every @xmath129 there holds @xmath130 it is well known that @xmath131 therefore , by lemma [ lem : multinomialbound ] , @xmath130
we recall that @xmath18 stands for @xmath5 s @xmath6-core .
we also denote @xmath132 . [
lem : humongous2core ] let @xmath5 be a @xmath133 graph where @xmath134 .
for every @xmath135 there holds @xmath136 let @xmath137 be the set of those vertices in @xmath5 with degree at most @xmath38 . we claim that @xmath138 .
clearly @xmath139 , since @xmath140 is acyclic .
also , @xmath141 since a vertex in @xmath140 can have at most one neighbor in @xmath142 .
hence , @xmath143 as claimed .
thus , it is enough to bound the probability that @xmath144 .
we fix a set @xmath145 of @xmath146 vertices and note that a vertex @xmath147 has @xmath148 only if @xmath149 , which holds with probability @xmath150 .
thus , the probability that all vertices in @xmath145 have degree @xmath151 is at most @xmath152 .
therefore , the probability that such a set @xmath145 exists is at most @xmath153 finishing the proof .
let @xmath15 be an @xmath36-vertex graph , and @xmath154 an integer .
an order @xmath125 _ configuration _ of @xmath5 is a pair of functions @xmath155\to v$ ] .
if @xmath156 we say that @xmath89 is a _ confluence _ of @xmath157 .
* a confluence - free configuration @xmath157 is said to have _
type i _ when @xmath158 and @xmath159 and @xmath160 are simple cycles ( in this order ) .
* we say that @xmath157 is a _ type ii _
configuration when @xmath161 and @xmath162 are each a simple path or a simple cycle .
also , @xmath158 , and @xmath163 are the only confluences .
[ lem : configurationbound ] let @xmath15 be a random @xmath8 graph and let @xmath28 .
pick the functions @xmath164\to v$ ] uniformly at random .
consider the events * @xmath165 that @xmath157 is a type i configuration . *
@xmath166 that @xmath157 is a type ii configuration .
then : 1 .
@xmath167 2 .
@xmath168 we only prove the first claim .
the same argument applies as well to the second case .
+ denote @xmath169 and @xmath170 . for @xmath171
, we estimate the probability of the events @xmath172 that @xmath173 for every @xmath174 and @xmath175 is a simple cycle in @xmath5 .
clearly , @xmath176 .
we complete the proof by showing that @xmath177 . indeed , suppose that @xmath178 for some @xmath179 . in this case
it is possible that @xmath180 are neighbors since @xmath181 coincides with either @xmath182 or with @xmath183 , but that happens with probability @xmath184 . otherwise , they are neighbors with probability @xmath7 . [
lem : compatiblebound]let @xmath185 for some @xmath186 .
pick a random @xmath8 graph @xmath15 and two random maps @xmath164\to v$ ] where @xmath187 .
let @xmath80 denote the number of indices @xmath188 such that @xmath189 .
then : @xmath190 @xmath191 we only prove the type i case .
the same argument applies to type ii configurations as well .
the argument below and all relevant calculations take place in the space _ conditioned on _ @xmath165 .
let @xmath192 and @xmath193 .
for each index @xmath89 let @xmath194 be the set of those neighbors of @xmath195 that have no other neighbor in @xmath196 .
we expose the subgraph induced on @xmath196 , thus revealing the sets @xmath194 .
the following proposition comes in handy : [ prop : compatibleboundprop ] with probability @xmath197 there holds : * @xmath198 * there are at least @xmath199 indices @xmath200 for which @xmath201 .
we proceed under the conditioning that the conclusion of this proposition holds .
we next reveal the edges connecting @xmath202 and @xmath203 .
this determines @xmath204 for all @xmath179 . on the other hand , @xmath205 is completely determined by the neighbor sets of vertices from @xmath194 in @xmath203 .
consequently the family of multisets @xmath206 is independent .
we are concerned with the event that @xmath189 . at this stage
this may already be impossible , and if possible , this uniquely determines the multiset of degrees @xmath207 over @xmath208 .
the elements of this multiset are drawn from a binomial distribution , so by lemma [ lem : multinomialwithbinomialprobsbound ] , if @xmath201 , then @xmath209 note that for @xmath210 to hold , the equality @xmath189 must hold for at least @xmath211 of the indices @xmath89 for which @xmath212 .
hence , @xmath213 as stated
. _ proof of proposition [ prop : compatibleboundprop ] : _ the first claim follows from chernoff s bound , as we observe that @xmath214 so that @xmath215 for the second claim @xmath216 and so @xmath217 let @xmath145 denote the event that @xmath218 and @xmath219 . by chernoff s bound , @xmath220 now , @xmath221 let @xmath222 denote the event that @xmath145 holds , and , in addition , @xmath223 .
note that @xmath145 implies @xmath224 hence , @xmath225 and so @xmath226 let @xmath227 .
note that @xmath222 implies that at least @xmath228 vertices in @xmath229 have no neighbor in @xmath230 , and thus , @xmath231 .
clearly , @xmath232 .
let @xmath233 denote the event that @xmath234 .
for @xmath235 , chernoff s bound implies @xmath236 note that given @xmath237 , the event @xmath233 is negatively correlated with every event of the form @xmath238 where @xmath239\setminus{i}$ ] .
thus , for every @xmath240 $ ] , @xmath241 in particular , the event @xmath242 that at most @xmath243 of the @xmath233 hold satisfies @xmath244 which implies @xmath245 let @xmath246 be the event that at most @xmath243 of the @xmath194 s satisfy @xmath247 .
a similar argument shows that given @xmath248 and we conclude that @xmath249 at this stage we have already established the following whp : for @xmath250 every automorphism of a @xmath8 graph pointwise fixes its @xmath6-core .
however , we seek to prove the stronger statement that the @xmath6-core has no nontrivial symmetries . as before consider two random maps @xmath164\to v$ ] where @xmath251 .
let @xmath192 and define the events @xmath252 as above .
clearly @xmath253 , since @xmath254 is a union of cycles , and now we need to control the effect of non-@xmath6-core vertices on @xmath3
. this effect is mediated by the set @xmath255 of @xmath254 s _ peripheral _ vertices , namely those within distance @xmath6 of @xmath256 . as we show , the above - mentioned effect is not large , since @xmath257 tends to be small .
we prove [ lem : peripherialbound ] @xmath258 @xmath259 we only prove case i. the same argument applies as well to case ii .
all our arguments below are made _ conditioned on _ @xmath165 .
let @xmath128 and @xmath193 .
clearly , @xmath260 .
reveal the subgraph @xmath261 of @xmath5 , induced by @xmath262 .
denote @xmath263 .
let @xmath264 .
by lemma [ lem : humongous2core ] , @xmath265 we henceforth condition on this event . note that @xmath266 , and thus , it is enough to bound the number of vertices in @xmath254 at distance @xmath267 from @xmath268 .
we denote @xmath269 by @xmath270 .
we claim that @xmath271 , since every vertex in @xmath270 has a neighbor in @xmath268 , whereas every vertex in @xmath268 has at most one neighbor in @xmath270 .
( note that @xmath272 and a vertex with more than one neighbor in @xmath270 is in @xmath273 as well ) . to understand the set @xmath274 of peripheral vertices
, we define three sets @xmath275 with @xmath276 and show that whp all @xmath277 are small .
let @xmath278 be the set of those vertices in @xmath254 with a neighbor in @xmath268 .
let @xmath279 be the set of those vertices in @xmath254 with a neighbor in @xmath278 .
finally , @xmath280 is the set of those vertices in @xmath254 with a neighbor in @xmath270 .
now reveal the set of cross edges @xmath281 . for @xmath282 , the probability that @xmath24 has a neighbor in @xmath268 is at most @xmath283 .
thus , @xmath284 and similarly , @xmath285 . in what follows we condition on the event that @xmath286 .
we finish by bounding @xmath287 .
reveal the edge set @xmath288 . by assumption
, @xmath254 is the image of a type i configuration , namely two simple cycles , possibly with some overlaps .
this implies the existence of certain edges in @xmath288 , at most @xmath289 in number .
in addition , the random variable @xmath290 is a sum of at most @xmath291 independent bernoulli-@xmath7 random variables . by assumption @xmath292 , so that @xmath293 only if at least @xmath294 of these bernoulli trials succeed . therefore @xmath295 clearly , @xmath296 , and so , @xmath297 with probability at least @xmath298 .
a configuration @xmath157 of @xmath5 is said to be _ compatible _ if there exists an automorphism @xmath299 of @xmath18 such that @xmath300 for each @xmath89 .
[ lem : nobadconfigs]let @xmath5 be a random @xmath8 graph with @xmath301 .
then whp @xmath5 contains no compatible configuration of type i or ii .
we prove the claim for type i configurations .
the proof for type ii follows the same argument . in the coming paragraph
we denote @xmath25 by @xmath302 .
we also consider the 2-neighborhood of @xmath303 in the subgraph induced by @xmath18 and denote @xmath25 in that graph by @xmath304 .
clearly , a configuration @xmath164\to v$ ] can be compatible only if @xmath305 for each @xmath306 .
let @xmath307 and pick two functions @xmath164\to v$ ] uniformly at random .
by lemma [ lem : configurationbound ] , the probability that @xmath157 is a configuration is at most @xmath308 .
conditioned on this event , let @xmath309\mid y(\phi(i))\neq y(\psi(i))\}.\ ] ] by lemma [ lem : compatiblebound ] , @xmath310 .
let @xmath311\mid y(\phi(i))=z(\phi(i))\text{~and~ } y(\psi(i))=z(\psi(i))\}.\ ] ] note that @xmath312 when both @xmath195 and @xmath313 are non - peripherial .
hence , by lemma [ lem : peripherialbound ] , @xmath314 .
but @xmath315 , so they must intersect , say @xmath316 .
then @xmath317 , which makes @xmath157 incompatible .
clearly this holds with probability @xmath298 . if @xmath318 is the number of compatible type i configurations we can now estimate its expectation : @xmath319 and so @xmath320 which completes the proof .
we can now finish up the proof of our main theorem .
let @xmath261 denote the @xmath6-core of @xmath5 .
it is known ( @xcite ) that whp @xmath321 , which we henceforth assume .
suppose that @xmath322 for some @xmath323 , and a vertex @xmath303 . by lemma [ lem : nobadconfigs ] ,
it is enough to show that this assumption implies that @xmath5 has a compatible configuration .
it is easy to see that if @xmath299 fixes all vertices of @xmath261 contained in cycles , then @xmath299 is trivial , so let @xmath324 be a cycle that contains @xmath24 .
the bound on @xmath5 s diameter implies that such a @xmath324 exists of length at most @xmath325 .
the argument splits now according to whether @xmath299 fixes some vertex in @xmath324 .
if it does not , then @xmath326 and @xmath327 that map @xmath328 $ ] to @xmath324 and to @xmath329 respectively form a compatible type i configuration , and we are done . otherwise , consider an arc @xmath330 ( possibly @xmath331 ) of @xmath324 so that : @xmath332 , and the only @xmath299-fixed points in @xmath333 are @xmath334 .
we obtain a compatible type ii configuration by letting @xmath326 map @xmath328 $ ] to @xmath333 and @xmath327 map @xmath328 $ ] to @xmath335 .
the purpose of this section is to prove : [ thm : reconstruct ] for every @xmath40 whp a @xmath8 graph is reconstructible .
we may clearly restrict ourselves to the range @xmath336 , since a graph is reconstructible iff its complement is reconstructible .
we may further restrict our attention to the range @xmath337 since the theorem is known for the two complementary ranges . for @xmath338
this was done by bollobs @xcite .
also , disconnected graphs are reconstructible @xcite , which takes care of the range @xmath339 .
one further simplification is that for @xmath7 in the above range , @xmath5 almost surely has no @xmath340 subgraph .
so we can and will be assuming this below .
our line of argument resembles the first part of the proof of theorem [ thm : maintheorem ] .
however , we need to adapt lemma [ lem : compatiblebound ] , a key step in that proof .
this lemma gives an upper bound on @xmath341 , while here this equality gets replaced by an _
approximate _ equality as we now define . for two multisets of integers
we say that @xmath342 if they can be made equal by applying some of the following operations to each of them .
( here @xmath69 refers to either @xmath145 or @xmath222 ) . *
decrease some elements of @xmath69 by @xmath67 or @xmath6 .
the total subtracted sum must be @xmath343 . * delete one or two elements of @xmath69 .
a configuration @xmath157 is _ acceptable _ if there exist vertex sets @xmath55 of size @xmath344 such that @xmath345 , @xmath346 , and @xmath64 and @xmath347 are isomorphic through a graph isomorphism @xmath299 that maps @xmath195 to @xmath313 for every @xmath89 .
[ lem : noacceptableconfigs ] whp , @xmath5 contains no acceptable configurations of type i or ii .
we first claim that @xmath348 , for every @xmath349 for @xmath63 , @xmath268 and @xmath299 as above .
this is so , since the property of @xmath299 implies @xmath350 .
these are subgraphs of @xmath344 vertices and the effect of the two missing vertices is limited due to @xmath351-freeness . since @xmath5 is @xmath340 free , @xmath352 for every @xmath353 .
hence , by removing @xmath24 from @xmath5 the possible changes in @xmath65 are : ( i ) decreasing one or two elements of @xmath65 by @xmath67 : each vertex in @xmath354 ( of which there are at most two ) may lose one neighbor , ( ii ) removal of a single element from @xmath65 ( the element corresponding to @xmath24 itself , if @xmath355 ) . to prove the lemma , we first strengthen lemma [ lem : compatiblebound ] , and replace the condition @xmath189 by @xmath356 .
the proof is essentially the same , with one change : clearly the multiset @xmath357 is uniquely determined by the condition @xmath189 .
now we operate under the weaker condition @xmath356 . rather than the above multiset , we consider a multiset where at most two of the entries are @xmath358 which stand for the possibly deleted vertices .
this multiset can take on only @xmath359 possible values .
lemma [ lem : multinomialwithbinomialprobsbound ] and a union bound argument yield : @xmath360 by lemma [ lem : configurationbound ] and the stronger version of lemma [ lem : compatiblebound ] proved here , the expected number of acceptable type i or type ii configurations in @xmath5 is at most @xmath361 we say that a vertex pair @xmath362 is _ interior _ if @xmath363 .
[ lem : nopartialsisomorphisms ] whp , for every interior vertex pair @xmath364 it holds that ( i ) every automorphism of @xmath365 fixes @xmath366 and ( ii ) for every interior vertex pair @xmath367 , the graphs @xmath368 and @xmath365 are non - isomorphic .
we may assume that @xmath369 , as this holds whp @xcite . also , + ( likewise for @xmath370 ) since the removal of a vertex at most doubles the diameter .
by lemma [ lem : noacceptableconfigs ] , we may also assume that @xmath5 has no acceptable type i or ii configurations .
we prove both parts of the lemma together by considering as well the case @xmath371 .
assume that there exists an isomorphism @xmath299 between @xmath368 and @xmath365 that does not pointwise fix the @xmath6-core . to prove the lemma , it is enough to show that there exists an acceptable type i or ii configuration in @xmath5 .
we consider two cases , first where @xmath299 moves some vertex in @xmath372 that resides in a cycle .
since @xmath373 , we may assume that this cycle has length @xmath374 .
the existence of an acceptable type i or ii configuration follows from the same argument as that in theorem [ thm : maintheorem ] . in the second case ,
@xmath299 fixes pointwise every cycle of @xmath372 .
thus , it must map some path between two vertices in cycles , fixed by @xmath299 , to a different path between these two vertices . due to
the bound on the diameter of @xmath368 , the length of these paths must be @xmath375 , which yields a type ii acceptable configuration . we may and will assume that @xmath5 satisfies the conclusion of lemma [ lem : nopartialsisomorphisms ] . for @xmath349 ,
we denote @xmath376 .
we claim that the cardinality @xmath377 is reconstructible . indeed ,
it is known @xcite that the degree sequence of @xmath5 is reconstructible , and thus , the property @xmath378 is recognizable .
now , assume that @xmath379 .
it is clearly possible to determine @xmath380 from @xmath381 .
also @xmath382 when @xmath383 . since @xmath384 iff @xmath385
, we can determine whether @xmath384 by observing @xmath381 .
we also note that the degree sequence of @xmath5 s @xmath6-core is reconstructible . indeed , if @xmath386 this follows from the reconstructibility of @xmath5 s degree sequence
otherwise , the @xmath6-core itself is reconstructible , as above .
let @xmath387 .
note that every vertex pair in @xmath145 is interior .
it is not hard to determine whether @xmath388 given @xmath381 , based on the reconstructibility of the @xmath6-core s degree sequence .
we claim that @xmath145 contains almost all vertices . by lemma [ lem : humongous2core ]
, there holds whp @xmath389 . for @xmath20 @xmath390 and by the union bound @xmath391 so ,
let @xmath392 and @xmath393 be such that @xmath394 and @xmath395 are interior pairs , and there exists an isomorphism @xmath299 between @xmath396 and @xmath397 . by lemma [ lem : nopartialsisomorphisms ]
, this holds only when @xmath398 , @xmath399 and @xmath299 fixes the @xmath6-core pointwise . using this property
, we can identify the vertices @xmath24 and @xmath400 respectively in the graphs @xmath381 and @xmath401 and identify each vertex in the @xmath6-core of one graph with its counterpart in the other .
this allows us to reconstruct @xmath5 up to the question of whether @xmath355 . since @xmath402 is reconstructible , this last question
can be answered as well .
in this section we describe a polynomial time random graph canonical labeling algorithm for graphs in @xmath8 where @xmath47 .
let @xmath403 be the collection of all _ rooted , oriented _ cycles of length @xmath307 in an @xmath36-vertex graph @xmath15 .
we use @xmath404 to denote the lexicographic ordering between multisets of integers , where the elements in @xmath145 and in @xmath222 appear in increasing order .
we equip @xmath405 with the semi - order @xmath406 where short cycles precede longer ones . for two cycles @xmath407 we say that @xmath408 if for some @xmath89 there holds @xmath409 and @xmath410 for every @xmath411 . we claim that for the relevant range of @xmath7 , a @xmath8 graph satisfies the following conditions whp : 1 .
each connected component of @xmath412 is a tree of size @xmath413 .
2 . @xmath321 .
@xmath414 is totally ordered by @xmath406 .
property ( 1 ) is easy to derive by a first - moment argument . for property
( 2 ) , see @xcite .
a proof of property ( 3 ) follows from property ( 2 ) by a simple variation of the proof of theorem [ thm : maintheorem ] .
we now explain how to canonically label a graph @xmath15 with these three properties . to a vertex @xmath24 that is contained in a cycle we assign the @xmath406-smallest label @xmath415 over all cycles for which @xmath416 .
this label can be found in polynomial time . if @xmath125 is the length of the shortest cycle through @xmath416 , then it is easy to show that there are at most @xmath417 such cycles .
we scan all of them and pick the @xmath406-smallest one .
note next , that a vertex @xmath303 that is not contained in a cycle must reside on the unique path between two vertices @xmath418 , each contained in a cycle .
therefore @xmath24 is uniquely defined by its distances from @xmath400 and from @xmath419 .
this , and the labels of @xmath400 and @xmath419 , give us a unique label for @xmath24 .
finally we find labels for vertices in @xmath420 . by property ( 1 )
, such a vertex belongs either to ( i ) a tree of size @xmath413 rooted at some vertex of @xmath18 or ( ii ) an acyclic connected component of size @xmath413 .
let @xmath24 be a vertex of type ( i ) , belonging to a tree @xmath254 rooted at @xmath384 .
there are only @xmath421 rooted trees of size @xmath413 @xcite , so we can list them and give a unique polynomial - length label to each vertex of each such class .
we label @xmath24 by a pair @xmath422 , where @xmath423 is the label of the vertex corresponding to @xmath24 in @xmath254 s isomorphism class in the list , and @xmath424 is @xmath400 s label .
that is , @xmath24 is labeled as `` the vertex of type @xmath423 in the tree rooted at @xmath400 '' .
type ( ii ) vertices are likewise handled , using a list of all isomorphism classes of non - rooted trees . to deal with vertices on acyclic connected components , collect all connected components of the same isomorphism class and give each of them a unique number .
the label of @xmath24 consists of the type of tree that contains it , that tree s ordinal number in its isomorphism class , and @xmath24 s location in that tree .
for smaller values of @xmath7 the structure of @xmath3 may become somewhat more complicated . for @xmath425 ,
a @xmath8 graph has , with probability bounded away from zero and one , some small symmetric components , e.g. , an isolated triangle .
moreover , with probability @xmath426 even the @xmath6-core of the graph s giant component , has a nontrivial symmetry .
this may result e.g. , from a triangle that `` hangs off '' the @xmath6-core .
however , as shown in @xcite , whp this @xmath6-core has a unique biconnected component of @xmath427 vertices .
we suspect that this giant biconnected component is rigid whp . for @xmath428
it was shown by bollobs @xcite that not only is @xmath5 reconstructible whp , such graphs have reconstruction number three .
we do not know whether this holds as well for smaller and substantially smaller values of @xmath7 . | a graph with a trivial automorphism group is said to be _
rigid_. wright proved @xcite that for @xmath0 a random graph @xmath1 is rigid whp .
it is not hard to see that this lower bound is sharp and for @xmath2 with positive probability @xmath3 is nontrivial .
we show that in the sparser case @xmath4 , it holds whp that @xmath5 s @xmath6-core is rigid .
we conclude that for all @xmath7 , a graph in @xmath8 is reconstrutible whp .
in addition this yields for @xmath9 a canonical labeling algorithm that almost surely runs in polynomial time with @xmath10 error rate .
this extends the range for which such an algorithm is currently known @xcite . |
recent measurements from some type ia supernovae ( sne ) at intermediate and high redshifts ( perlmutter et al . 1999 , riess et al . 1998 ) indicate that the bulk of energy in the universe is repulsive and appears like a quintessence " component , that is , an unknown form of dark energy ( in addition to the ordinary cdm matter ) probably of primordial origin ( see turner 2000 for a review ) .
together with the observations of cmb anisotropies ( de bernardis 2000 ) , such results seem to provide an important piece of information connecting an early inflationary stage with the astronomical observations .
this state of affairs has stimulated the interest for more general models containing an extra component describing this dark energy , and simultaneously accounting for the present accelerated stage of the universe .
however , the absence of a convincing evidence on the nature of the dark component gave origin to an intense debate and mainly to theoretical speculations .
a possible list of old and new candidates for quintessence " now include : \(i ) a decaying vacuum energy density , or a time varying @xmath0-term ( zer and taha 1987 , freese at al .
1987 , chen and wu 1990 , carvalho et al .
1992 , waga 1993 ; for reviews see overduin and cooperstoock 1998 ; sahni and starobinski 2000 ) \(ii ) the so - called x - matter " , an extra component simply characterized by an equation of state @xmath1 , where @xmath2 ( turner and white 1997 , chiba et al .
1997 , efstathiou 1999 , lima and alcaniz 2000 , turner 2000 , alcaniz and lima 2001 ) , which describes , as a particular case , cosmologies with a constant @xmath0-term ( @xmath0cdm models ) .
generically , the @xmath3 parameter may be a function of the redshift ( cooray and huterer 2000 ) \(iii ) a rolling scalar field ( ratra and peebles 1988 , caldwell et al .
1998 , wang et al .
2000 ) .
here we are interested in the first class of models .
the basic reason is the widespread belief that the early universe evolved through a cascade of phase transitions , thereby yielding a vacuum energy density which at present is at least @xmath4 orders of magnitude smaller than in the planck time ( weinberg 1989 ) .
such a discrepancy between theoretical expectation ( from the modern microscopic theory of particles and gravity ) and empirical observations constitutes a fundamental problem in the interface uniting astrophysics , particle physics and cosmology , which is often called `` the cosmological constant problem '' ( weinberg 1989 ; jack ng 1992 ; dolgov 1997 ) .
this puzzle inspired some authors ( lima and maia 1994 , lima and trodden 1996 ) to propose a class of phenomenological deflationary cosmologies driven by a decaying vacuum energy density where the present value , @xmath5 , is a remnant of the primordial inflationary stage ( from now on the subscript
o " denotes the present day quantities ) .
the basic scenario has an interesting cosmological history that evolves in three stages .
initially , an unstable de sitter configuration , with no matter and radiation is supported by the largest values of the vacuum energy density .
this nonsingular de sitter state evolves to a quasi - frw vacuum - radiation - dominated phase , and , subsequently , the universe changes continuously from vacuum - radiation to the present vacuum - dust dominated phase
. the first stage harmonizes the scenario with the cosmological constant problem , while the transition to the second stage solves the horizon and other well - know problems in the same manner as in inflation .
finally , the universe enters in the present vacuum - dust phase with a negative deceleration parameter as required by the sne type ia observations . in this article , we focus our attention on this class of deflationary decaying vacuum models .
the effective time dependent cosmological term is regarded as a second fluid component with energy density , @xmath6 , which transfers energy continuously to the material component .
the main goal is to investigate the basic kinematic tests in the present vacuum - dust dominated phase , or equivalently , how the classical cosmological tests may constrain the physical parameters of such models .
the paper is organized as follows : in section 2 , we set up the basic equations for deflationary cosmologies driven by a decaying @xmath7-term . in section 3 , the expressions for classical cosmological tests are derived and compared with the conventional expressions without the @xmath0-term .
section 4 gives the conclusion of the main results , and , in the appendix a , the exact expression yielding the dimensionless radial coordinate as a function of the redshift is deduced .
we shall consider a class of spacetimes described by the general frw line element ( @xmath8 ) @xmath9 where @xmath10 is the scale factor , @xmath11 , @xmath12 is the curvature parameter of the spatial sections , and @xmath13 , @xmath14 and @xmath15 are dimensionless comoving coordinates . in that background ,
the einstein field equations ( efe ) with a nonvacuum component plus a cosmological @xmath7-term are : @xmath16 @xmath17 where an overdot means time derivative , @xmath18 and @xmath19 are the energy density and pressure , respectively . as usual , we consider that the nonvacuum component obeys the @xmath20-law " equation of state @xmath21 where @xmath22 $ ] specifies if the fluid component is radiation ( @xmath23 ) or dust ( @xmath24 ) .
phenomenologically , we also assume that the effective @xmath7-term is a variable dynamic degree of freedom so that in an expanding universe it relaxes to its present value according with the following ansatz ( lima and trodden 1996 ) @xmath25 where @xmath26 is the vacuum density , @xmath27 is the total energy density , @xmath28 is the hubble parameter , @xmath29 is the arbitrary time scale characterizing the deflationary period , and @xmath30 $ ] is a dimensioneless parameter of order unity .
it is worth noticing that for @xmath31 , the above definition reduces to @xmath32 so that we have inflation with no matter - radiation component ( @xmath33 ) . in particular ,
if @xmath29 is the order of the planck time , that is , @xmath34s , one may show that such models are in accordance with the cosmological constant problem in the sense that @xmath35 . at late times ( @xmath36 )
, we see from ( 5 ) that @xmath37 , as required by the recent supernovae observations . to be more precise , if the deflationary process begins at planck time , the ratio @xmath38 while the remaining terms are of order unity .
indeed , even if deflation begins much later , say at @xmath39s , or even at @xmath40s ( the respective scales of grand and electroweak unification in the standard model ) , one obtains @xmath41 and @xmath42 , respectively .
this means that to a high degree of accuracy , the scale @xmath43 is unimportant during the vacuum - dust dominated phase . therefore , since in this work we mainly interested in the classical cosmological tests , henceforth we consider only this limiting behavior with the material component described by a pressureless fluid ( @xmath24 ) .
let us now consider the evolution of the scale factor . combining equations ( [ rho])-([ansatz ] ) and taking the limit @xmath44 , it is readily seen that the scale factor during the vacuum - dust phase obeys the slightly modified frw equation @xmath45 the first integral of which is @xmath46 where the constant @xmath47 in order that @xmath18 be positive definite in this phase ( see eq.(2 ) ) . by expressing the constant @xmath48 in terms of the present day parameters ,
it is straightforward to show that the above equation can be rewritten as @xmath49\quad .\ ] ] where @xmath50 is the present value of the matter density parameter . for @xmath51
the above equation reproduces the standard cold dark matter frw result ( kolb and turner 1990 ) . whereas for @xmath52 , it describes the influence of the decaying vacuum in the present phase . following standard lines
we also define the deceleration parameter @xmath53 . using equations ( [ ddr ] ) and ( [ dr ] )
one may show that @xmath54 then , for any value of @xmath55 , we see that the deceleration parameter @xmath56 with decaying vacuum energy is always smaller than the corresponding one of the frw model .
the critical case ( @xmath57 ) , describes a coasting cosmology " . however , instead of being supported by
k - matter " ( kolb 1989 ) , this kind of model is obtained in the present context for a vacuum - dust filled universe , and the corresponding solutions hold regardless of the value of @xmath58 .
it is also interesting that even negative values of @xmath56 are allowed since the constraint @xmath59 can always be satisfied provided @xmath60 .
these results are in line with recent measurements of the deceleration parameter @xmath56 using type ia supernovae ( perlmutter et al .
1999 ; riess et al .
1998 ) . such observations indicate that the universe may be accelerating today , which corresponds dynamically to a negative pressure term in the efe .
for a fixed @xmath58 , this means that the universe with decaying vacuum is older than the corresponding frw model with the usual deceleration parameter @xmath61 .
this behavior also reconcile other recent results ( freedman 1998 ) , pointing to a hubble parameter @xmath62 larger than @xmath63 km s@xmath64 mpc@xmath64 .
now , combining equations ( [ rho ] ) , ( [ ansatz ] ) and ( [ dr ] ) for ( @xmath65 ) one finds that vacuum and the matter energy density can be expressed as @xmath66 @xmath67 where @xmath68 . in virtue of the decaying vacuum energy , we also see that the explicit dependence of the energy density on the scale factor @xmath10 is slightly modified in comparison with the standard case . as should be expected from eqs .
( [ ddr ] ) and ( [ dr ] ) , some dynamic expressions may be obtained from the standard frw ones simply by replacing the index " @xmath69 by an effective parameter @xmath70 ( see the above scale laws ) .
the kinematical relation distances must be confronted with the observations in order to put limits on the free parameters of the deflationary class of models presented in the last section . as remarked before ,
@xmath43 can not be constrained by the classical tests since it is unimportant in the present phase .
it is adjusted in order to put the models in accordance to the cosmological constant problem ( lima and trodden 1996 ) .
it thus follows that we have only the pair of free parameters ( @xmath71 ) satisfying @xmath72 , where @xmath73 is the present value of the total density parameter . 0.1 in \a ) _ lookback time - redshift diagram _ the lookback time , @xmath74 , is the difference between the age of the universe at the present time ( @xmath75 ) and the age of the universe when a particular light ray at redshift @xmath76 was emitted . by integrating ( [ dr1 ] ) such a quantity
is easily derived @xmath77 which generalizes the standard frw result ( kolb and turner 1990 ) .
the age of the universe is obtained by taking the limit @xmath78 in the above equation .
we find @xmath79 for @xmath51 the above expressions reduce to the ones of the standard frw models ( kolb and turner 1990 ) .
generically , we see that decaying vacuum increases the dimensionless parameter @xmath80 while preserving the overall expanding frw behavior .
the lookback time curves as a function of the redshift for some selected values of @xmath81 and @xmath82 are displayed in fig .
1 . for completeness , in fig .
2 we show the age of the universe ( in units of @xmath83 ) as a function of @xmath56 for some selected values of @xmath82 . 0.1 in \b ) _ luminosity distance - redshift _
the luminosity distance of a light source is defined as the ratio of the detected energy flux @xmath84 , and the apparent luminosity , i.e. , @xmath85 . in the homogeneous and isotropic frw metric ( 1 ) it takes the form below ( sandage 1988 ) @xmath86 where @xmath87 is the radial coordinate distance of the object at light emission . inserting @xmath87
derived in the appendix , it follows that @xmath88}{h_o \left(\frac{\omega_o}{1-\beta } - 1 \right)^{\frac{1}{2}}}\quad .\ ] ] where @xmath89 , @xmath90 and @xmath91 . 0.1 in
as one may check , expressing @xmath58 in terms of @xmath56 from ( 9 ) , and taking the limit @xmath92 , the above expression reduces to @xmath93\quad , \ ] ] which is the usual frw result ( weinberg 1972 ) . expanding eq.(15 ) for small @xmath76 gives @xmath94 z^{2 } + ... \quad , \ ] ] which depends explicitly on the @xmath82 parameter .
however , replacing @xmath58 from ( 9 ) we recover the usual frw expansion for small redshifts , which depends only on the effective deceleration parameter @xmath56 ( weinberg 1972 ) .
this is not a surprising result since expanding @xmath95 in terms of @xmath58 , the `` 00 '' component of einstein s equations has implicitly been considered , while the expansion in terms of @xmath56 comes only from the form of the frw line element .
the luminosity distance as a function of the redshift for closed and open models with decaying vacuum energy is shown in figures 3 and 4 , respectively .
as expected for all kinematic tests , different cosmological models have similar behavior at @xmath96 , and the greatest discrimination among them comes from observations at large redshifts . a more quantitive analysis based on the sne observations will be presented elsewhere . 0.1 in \c )
_ angular size - redshift _ another important kinematic test is the angular size - redshift relation @xmath97 . as widely known , the data concerning the angular - size are until nowadays somewhat controversial ( see buchalter et al .
1998 and references therein ) . here
we are interested in angular diameters of light sources described as rigid rods and not as isophotal diameters .
these quantities are naturally different , because in an expanding world the surface brightness varies with the distance ( sandage 1988 ) .
the angular size of a light source of proper size @xmath98 ( assumed free of evolutionary effects ) located at @xmath99 and observed at @xmath100 is @xmath101 inserting the expression of @xmath87 given in the appendix a it follows that 0.1 in for small @xmath76 one finds @xmath102\quad .\ ] ] hence , decaying vacuum cosmologies as modeled here also requires an angular size decreasing as the inverse of the redshift for small @xmath76 .
however , for a given value of @xmath58 , the second order term is a function only of the @xmath82 parameter . in terms of @xmath56 , inserting ( 9 ) into ( 20 ) it is readily obtained @xmath103\quad , \ ] ] which is formally the same frw result for small redshifts ( sandage 1988 ) . at this limit
only the effective deceleration parameter may be constrained from the data , or equivalently , at small redshifts one can not extract the values of @xmath58 and @xmath82 separately . the angular size - redshift diagram for closed and open models and selected values of the @xmath82 parameter
is displayed in figures 5 and 6 , respectively . 0.1 in \d ) _ number counts - redshift _
the final kinematic test considered here is the galaxy number count per redshift interval .
the number of galaxies in a comoving volume is equal to the number density of galaxies ( per comoving volume ) @xmath104 , times the comoving volume element @xmath105 @xmath106 0.1 in by using @xmath87 as derived in appendix , it follows that the general expression for number - counts can be written as @xmath107 } \over ( 1 + z)z^{2}f(\omega_o , \beta , z)}\quad , \ ] ] where @xmath108^{1/2}$ ] . for small redshifts
@xmath109z + ... \quad .\ ] ] in figures 7 and 8 we have displayed the number counts - redshift relations for flat , closed and open universes for some selected values of @xmath81 and @xmath82 .
it is worth mentioning the tendency of decaying vacuum models to have larger volumes per redshift interval than the standard frw models with the same @xmath81 .
this feature is similar to the one found in `` quintessence '' and @xmath0cdm cosmologies and could be advantageous if the observational data indicate an excess count of high - redshift objects .
the limits on the @xmath82 parameter obtained from all kinematic tests are shown in table 1 .
0.1 in at this point , it is also interesting to compare our results with previous studies of kinematic tests for @xmath7 cosmologies .
observable expressions for a large class of flat decaying vacuum models were discussed by waga ( 1993 ) . in his paper , the time varying cosmological term was defined by @xmath110 , where @xmath111 , @xmath82 and @xmath20 are arbitrary constants .
although rather different from the deflationary ansatz assumed here , we see that in the limit @xmath112 , our expression ( 5 ) reduces to @xmath113 , or equivalently , @xmath114 .
therefore , by considering only the flat case ( @xmath11 ) and taking @xmath115 , the late stages of the deflationary cosmologies has exactly the same behavior of the models examined by waga .
note also that our @xmath82 parameter must be identified with @xmath116 of that one appearing in his paper .
more recently , some aspects involving the angular size , luminosity distance , and sne type ia observations in @xmath7 models were also investigated by vishwakarma ( 2000 , 2001 ) . by generalizing earlier papers ( chen and wu 1990 , carvalho et al .
1992 ) , he considered the interesting ansatz @xmath117 , where @xmath118 is a dimensionless constant and @xmath119 is the density parameter of the fluid component .
although proposed as a new cosmology , if we rewrite this @xmath7 in terms of the total energy density , this decaying law corresponds precisely to the late stages of the deflationary universes for @xmath120 .
a more quantitative analysis accounting for angular sizes , the existence of old high redshifts galaxies , and ages constraints from globular clusters in this framework will be discussed in a forthcoming communication .
we also remark that models driven by a gravitational adiabatic matter creation process ( calvo et al .
1992 , lima and alcaniz 1999 , alcaniz and lima 1999 ) present some kinematic expressions , like the dimensionless radial coordinate , which are similar to the ones discussed here for @xmath7 cosmologies ( see appendix ) . in these models ,
the present day matter creation rate , @xmath121 nucleons @xmath122 , is also nearly the same rate transferred from the decaying vacuum to the fluid component .
however , these two spacetimes are completely different from a physical viewpoint because the negative creation pressure of the former scenario can not be interpreted as a genuine decaying vacuum component . [ tabelabeta ] rlrll + test & & open & closed + + + luminosity distance - redshift & & @xmath123 & @xmath124 + + angular size - redshift & & @xmath125 & @xmath126 + + number counts - redshift & & @xmath127 & @xmath128 + +
the recent observational evidences for an accelerated state of the present universe , obtained from distant sne ia ( perlmutter et al .
1999 , riess et al . 1998 ) gave a strong support to the search of alternative cosmologies .
as demonstrated here , a variable @xmath0-term or a decaying vacuum energy density is also an ingredient accounting for this unexpected observational result . in the present paper
we have analyzed all kinematic expressions for flat , closed , and open cosmologies when the decaying vacuum is a fraction of the total energy density . at the later stages of the evolution ,
the rather slight changes introduced by @xmath7 , which is quantified by the @xmath82 parameter , provides a reasonable fit of several cosmological data .
kinematic tests like luminosity distance , angular diameter and number - counts versus redshift relations constrain perceptively the decaying vacuum ( see table 1 ) . for models characterized by the pair ( @xmath129 ) , the age of the universe is always greater than the corresponding frw model ( @xmath51 ) , and even values bigger than @xmath130
are allowed for all values of the curvature parameter .
some observable quantities in the standard frw model are easily determined expressing the radial dimensionless coordinate @xmath13 of a source light as a function of the redshift ( mattig 1958 ) . in this appendix
, we derive a similar equation to the decaying vacuum energy scenario discussed in this paper . now consider a typical galaxy located at @xmath131 emitting radiation to an observer at @xmath132 .
if the waves leave the source at time @xmath133 and reach the observer at time @xmath134 , the null geodesic equation defining the light track yields @xmath135 since @xmath136 , changing variable to @xmath137 and using ( 8) , the above result reads this integral depends on the values of the @xmath58 and @xmath82 parameters . for @xmath138
one finds the same results of the coasting cosmology ( kolb 1989 ) . for @xmath82
different of @xmath139 , we introduce a new auxiliary variable @xmath140x^{(1 - 3\beta)}$ ] , in terms of which the above equation becomes @xmath141 where @xmath142 , @xmath143^{\frac{1}{2}}$ ] , and @xmath144 .
the right hand side of the above integral is the same appearing in ( a.1 ) for @xmath145 .
hence , replacing in ( a.3 ) the value of @xmath146 given by ( [ rho ] ) and ( 7 ) , it is readily seen that @xmath147}{r_oh_o(\frac{\omega_o}{1-\beta } - 1)^{\frac{1}{2}}}\quad .\ ] ] in particular , the limit for a flat universe ( @xmath148 ) yields @xmath149 which could have been obtained directly from ( a.2 ) . in terms of the deceleration parameter ( a.4 )
may be rewritten as @xmath150}{r_oh_o(\frac{2q_o}{1 - 3\beta } - 1)^{\frac{1}{2}}}\quad , \ ] ] which in the limit @xmath151 reduces to the usual frw result ( weinberg 1972 ) @xmath152 equation ( a.4 ) , or equivalently ( a.6 ) , plays a key role in the derivation of some astrophysical quantities discussed in this paper . * acknowledgements : * the authors are grateful to j. s. alcaniz for helpful discussions and a critical reading of the manuscript .
this work was partially supported by the project pronex / finep ( no .
41.96.0908.00 ) , and cnpq ( 62.0053/01 - 1-padct iii / milenio ) . 30 alcaniz j.s . ,
lima j.a.s .
2001 , apj lett .
550 , l133 alcaniz j.s .
, lima j.a.s .
1999 , a&a 349 , 729 buchalter et al .
1998 , apj 494 , 503 caldwell r.r . ,
dave r. , steinhardt p.j . 1998 , phys . rev . lett .
80 , 1582 calvo m.o .
, lima j.a.s . , waga i. 1992 , phys .
lett . a 162 , 233 carvalho j.c . ,
lima j.a.s .
, waga i. 1992 , phys .
d 46 , 2404 chen w. , wu y.s . 1990 ,
phys . rev .
d 41 , 695 chiba t. , sugiyama n. and nakamura t. 1997 , mnras 289 , l5 cooray a.r .
, huterer d. 2000 , apj lett .
513 , l95 de bernardis p.
2000 , nat 404 , 955 dolgov a.d . , a lecture presented at the 4th colloque cosmologie , paris , 1997 ( astro - ph/9708045 ) efstathiou g. 1999 , mnras 310 , 842 freedman w.l . in proceedings of the 18th texas symposium on relativistic astrophysics , edited by a. olinto , j. frieman , d. schramm .
singapura : world scientific ( 1998 ) freese k. , adams f.c . , frieman j.a .
, mottola e. 1987 , nucl .
b 287 , 797 gurvits l.i . , kellermann k.i . , frey s. 1999 , a&a 342 , 378 jack ng y. , 1992 , int
d 1 , 145 kolb e.w .
1989 , apj 344 , 543 kolb e.w .
, turner m.s .
1990 , the early universe .
addison - wesley , redwood city lima j.a.s . ,
alcaniz j.s .
1999 , a&a 348 , 1 lima j.a.s .
, alcaniz j.s .
2000 , a&a 357 , 393 lima j.a.s , maia j.m.f .
1994 , phys . rev .
d 49 , 5597 lima j.a.s .
, trodden m. 1996 , phys .
d 53 , 4280 loh e. , spillar e. 1986 , apj 307 , l1 mattig w. 1958 , astron . nachr . 284 , 109 overduin f.m . ,
cooperstock f.i .
1998 , phys .
d 58 , 043506 zer m. , taha m.o .
1987 , nucl .
b 287 , 776 perlmutter s. , aldering g. , goldhaber g. et al . 1999 , apj 517 , 565 perlmutter s. et al .
1995 , apj 440 , l41 ratra b. and peebles p.j.e .
1988 , phys .
d 37 , 3406 riess a. g. et al .
1998 , aj 116 , 1009 sandage a. 1988 , ara&a 26 , 561 sahni v. , starobinsky a.a .
2000 , int .
d 9 , 373 turner m.s .
, white m. 1997 , phys .
d 56 , r4439 turner m.s .
2000 , phys . rep .
333 , 619 vishwakarma r.g . 2000 , class
. q. grav .
17 , 3833 vishwakarma r.g . 2001 , class .
18 , 1159 waga i. 1993 , apj 414 , 436 wang l. and kamionkowski m. 2000 , phys . rev .
d 61 , 063504 weinberg s. 1972 , cosmology and gravitation , john wiley sons , n.y .
weinberg s. 1989 , rev . | we discuss the classical cosmological tests for a large class of frw type models driven by a decaying vacuum energy density .
analytic expressions for the lookback time , age of the universe , luminosity distance , angular diameter , and galaxy number counts versus redshift are derived and their meaning discussed in detail . it is found that the standard frw results are significantly altered , showing that such tests may constrain the physical parameters of these models which are also in agreement to the accelerated expansion suggested by the latest sne type ia observations . |
the importance of beam polarization for the ilc experiments can be illustrated by fact that the effective luminosity is increasing by approximately @xmath0 in the case of both beam polarized @xcite .
furthermore a suitable combination of polarized electron and positron beams suppresses significantly unwanted background processes and enhances signal rates .
there are two important aspects which should be taken into account for polarized beams .
the first one is a delivery of polarized beams from the source to the interaction point .
the spin transport for the different areas of the ilc were already studied @xcite and the installation of spin rotators before and after damping ring was recommended .
the examples of possible layouts of spin - rotators for the ilc can be found in @xcite .
the second problem arising from the presence of polarized beams is the requirement of fast helicity reversal .
the helicity pattern of the electron beam can be adjusted by changing the helicity of the laser . for the positron beam
this is a non - trivial task , as the polarization of the positron beam depends on helicity of the undulator .
the possibility of spin manipulation was considered at @xcite where two post - damping ring spin rotators were included . on the other side ,
the spin manipulation of post damping ring beams is limited by the emittance preservation constraints .
in addition the spin rotator used in the tdr design can not provide a fast helicity reversal in the time scale desirable for the ilc , i.e. from train to train .
the idea of using a pre - damping ring spin rotator section for some beam helicity manipulations has been already suggested in @xcite , but no detailed lattice was produced .
meanwhile , the layout of the central region of the ilc provides enough space before damping ring for a combined spin rotation with a possibility of quick switch between two helicities .
a possible layout of a pre - damping ring spin rotator / spin flipper section is presented below .
the positron linac to damping ring ( pltr ) is a section of the ilc transport positron beam to the damping ring ( dr ) . the schematic layout of the pltr is given in fig . [ f1 ] .
it serves for the extraction of the positrons from the positron linac booster , energy compression and spin rotation .
in general , the desirable spin rotation can be produced by spin precession around the field direction . in the dipole field
the rate of spin precession is directly proportional to the orbit deflection angle @xmath1 while in the solenoidal field spin precession rate is directly proportional to the field @xmath2 and the length @xmath3 of solenoid and inversely proportional to the the magnetic rigidity @xmath4 . at 5 gev the orbital deflection angle of @xmath5 rotates spins by @xmath6 . in section e the spin rotation from the longitudinal to the transversal direction is done by the means of horizontally bending dipoles with the total orbital rotation angle of @xmath7 which corresponds to @xmath8 of spin rotation .
the total length of section d is 123.595 m. the suggested combined spin flipper / spin rotator design is only 80 m long .
a new modified section d can fulfill two tasks simultaneously , namely spin rotation and train - by train helicity reversal .
the energy compression in section d matches the beam energy spread to the dr acceptance .
then the transversal beam polarization can be rotated to vertical in the solenoid with a field integral of 26.18 [ t m ] .
two different superconducting solenoids design were considered : 8.32 m long solenoid with an integrated field of 26.18 @xmath9 $ ] ( solenoid 1 ) and a shorter 5 m long superconducting solenoid with integrated field of 26.2 @xmath9 $ ] ( solenoid 2 ) . the pre - damping ring position of the spin- rotator makes the emittance preservation constrains less challenging .
the suggested combined spin flipper / spin rotator consists of two parallel beam lines for spin rotation equipped with two solenoids of opposite polarities , i.e. setting the spin parallel ( one beam line ) or antiparallel ( second beam line ) to the field in the damping ring , fig .
this spin- flipper design is based on the concept of branch splitter / merger used for the post - damping ring positron lines @xcite .
the first lattice cell is an irregular fodo cell which include fast kickers and separate the branches horizontally .
the total length of the splitter section is approximately 26 m in order to fit the available space , 2 m of two horizontal branches separation is taken .
the shortening of the splitter section is achieved by using stronger bending magnets .
each branch consists of a first order achromat fodo dogleg , a solenoid section and another dogleg to recombine the line back to the design orbit .
the achromat design assures that no dispersion suppressors would be required . the simple solenoid rotator design is considered , similar to the one used in @xcite .
the advantage of this design is the possibility of quick and random switching between two helicities for the positrons . in order to save some transversal space
an asymmetry can be introduced in the relative position of solenoids at two branches .
the section consists of the first irregular fodo - like cell with a pulsed kicker and a combined function defocusing / bending magnet , followed by 4 regular fodo cells with @xmath10 phase advance , forming together an achromat dogleg , a solenoid matching section and a solenoid with an integrated field of 26.18 or 26.2 @xmath9 $ ] . in the solenoid beta functions @xmath11 and they are reaching the minimum in the middle of the solenoid . the rest of the section is a mirror image of the first part with respect to the middle of solenoid .
the second branch of the lattice can be obtained by switching the sign of the kick in the pulsed kicker and the bending angles in the following dogleg .
the section was optimized by mad8 package @xcite to meet the constraints on the length of total d section .
then this spin - rotator part of section d was inserted to the pltr lattice developed by w.liu @xcite thus including two extra matching sections . in fig .
[ f3]-fig .
[ f4 ] the results of the optics matching using mad8 package is given for plrt section .
the lattice matching for the section d of the plrt was done for both versions of superconducting solenoids .
the optics for solenoid 1 setting is cross - checked with elegant code@xcite code .
spin tracking with bmad code @xcite is done by v.kovalenko @xcite .
two solenoid sections in the opposite branches were placed with @xmath12 m shift producing a horizontal offset of 0.54 m for each branch with respect to design orbit .
one or two extra fodo cells were added in front of the solenoid section for one branch and the same number of extra fodo cells were added after the solenoid section for another branch .
these changes lead to an increase of the total length of the whole spin rotator section .
rematching was done in order to fit the total length of section d ( 123.595 m ) and the total pltr length . in fig .
[ f5 ] the design of the new spin rotation section for two version of super - conducting solenoids is given .
the suggested spin rotator design confirms that fast helicity switching for the positron beam is possible .
the train - to - train polarity selection for electron and positron beams at the ip can be achieved . in particular : * the suggested optic design for the fast helicity reversal spin rotator section satisfies the pltr section requirements . * both versions of superconducting solenoid for spin rotator
are used and two versions of optic files are available for the symmetric lattice * an asymmetric design for the solenoid position in two parallel lines of spin rotator is produced for two version of superconducting solenoid parameters . * the optic design is cross - checked with different accelerator design codes * depolarization effects in a new lattice are estimated by bmad and no significant depolarization connected with beam optics is discovered .
the authors are grateful to the all members of the spin - management group for the fruitful discussions .
we would like to thank dr .
w. liu for close collaboration and for providing the matching parameters of the pltr lattice .
malysheva also thanks mr .
j. jones and dr .
p. williams from astec for help and advice concerning elegant running .
g.moortgat-pick et al . ,
phys.report.460:131-243,(2008 ) www.ippp.dur.ac.uk / gudrid / power/. ilc technical design report //www.linearcollider.org / ilc / publications / technical - design - report l.i .
malysheva and d.p .
depolarisation in the damping rings of the ilc.proceedings of international linear collider workshop lcws2007 , hamburg , germany , june 2007 .
bailey et al , spin tracking at the international linear collider .
proc . of pac
2007 , albuquerque new mexico 25 - 29 jun 2007 , pp .
2901 - 2903 .
p.emma , a spin rotator system for the nlc , slac nlc - note-07,1994 .
p.schmidt , a spin rotator for the ilc , eurotev - report-2005 - 024 - 01 f.zhou et al.,start - to - end beam optics development and multiparticle tracking for tha ilc undulator based positron spource , slac - pub-12239 , jan.2007 . k.moffeit et al . , spin rotation schemes at the ilc for two interaction regions and positron polarisation with both helicities , slac - tn-05 - 045,2005 .
n. solyak , rtml general layout and parameters .
; ilc source / rtml / bds+mdi baseline technical review , desy , october 2011 ` http://mad.web.cern.ch/mad/ ` w.liu , private conversation
. m. borland .
elegant : a flexible sdds - comliant code for accelerator simulation , proceedings of the @xmath13 international computational accelerator physics conference , icap2000 , september 2000 www.lepp.cornell.edu/ dcs / bmad v.kovalenko et al , spin tracking simulation of a future international linear collider proc . of the 3nd international particle accelerator conference ,
( ipac12 ) , new orleans louisiana , usa , 20 - 25 may 2012 , pp.1807 | the use of polarized beams enhance the possibility of the precision measurements at the international linear collider ( ilc ) @xcite . in order to preserve the degree of polarization during beam transport spin rotators
are included in the current tdr ilc lattice @xcite . in this report
some advantages of using a combined spin rotator / spin flipper section are discussed .
a few possible lattice designs of spin flipper developed at desy in 2012 are presented . |
abundance determinations of stars in galactic globular clusters can provide valuable information about important astrophysical processes such as stellar evolution , stellar structure , galactic chemical evolution and the formation of the milky way .
surface stellar abundances of c , n , o , and often na , mg , and al are found to be variable among red giants within a globular cluster .
the physical process responsible of these star - to - star element variations is still uncertain ( see * ? ? ?
* paper i ) . of particular importance to the present study
are the results of @xcite , who found that [ fe / h ] among m92 subgiants is a factor of two smaller than [ fe / h ] from red giants in the same cluster @xcite .
if this result is not due to systematic differences arising from the analysis procedures of the different groups that handle the red giant branch ( rgb ) and the subgiant samples , then the results of @xcite would suggest some modification of photospheric fe abundances which would be quite difficult to explain . clearly the present day clusters stars are incapable of modifying their own fe abundances through nucleosynthesis . yet evolution - driven changes in [ fe / h ] could be possible if the outer envelopes of the subgiants were somehow infused with fe - poor external material .
thus as evolution progesses up the rgb the deeping convective envelope would dilute the photospheric [ fe / h]with more fe - rich `` unmodified '' material from the interior causing an increase in [ fe / h ] with luminosity .
we note the unlikelyhood of this scenario as fe , unlike the lighter elements , is not involved in normal mass transfer processes that might ocurr in binary systems or in normal stellar winds , and is only produced in supernovae .
moreover , it is difficult to understand how cluster stars could be contaminated by material with _
less _ fe , presumably at a time after the present subgiants had formed . in order to study the origin of the star - to - star abundance variations and to address the issues raised by @xcite
, we have started a program to determine chemical abundances of the nearer galactic globular cluster stars . in this paper , we present our results for the iron abundance of m71 , the nearest globular cluster reachable from the northern hemisphere .
similar programs are underway at eso by @xcite and @xcite taking advantage of the fact that the nearest globular cluster accessible from a southern site ( ngc 6397 ) is 1.8 times close than m71 and has a lower reddening .
our m71 sample includes stars over a large range in luminosity : 19 giant stars , 3 horizontal branch stars , and 3 stars near the main sequence turnoff , in order to study in a consistent manner red giants , horizontal branch stars , and stars at the main sequence turnoff .
details on the star sample , observations , data reduction and determination of stellar parameters are described in paper i. previous high dispersion abundance analysis for m71 involve studies of red giants only , resulting in [ fe / h ] of @xmath10.70 @xcite , @xmath10.80 @xcite , between @xmath10.6 and @xmath11.0 @xcite , and @xmath10.79 @xcite .
we begin our analysis with fe , as many fe lines are identified in our hires spectra over a wide range of excitation potentials and line strengths , as well as over two ionization states ( fe i and fe ii ) .
the iron abundance analysis is done using a current version of the lte spectral synthesis program moog @xcite .
a line list specifying the wavelengths , excitation potentials , gf values , damping constants , and equivalent widths for the observed fe i and fe ii lines is required .
the provenance of the gf values and the measurement of equivalent widths are discussed below .
the damping constants for all fe i and fe ii lines were set to twice that of the unsld approximation for van der waals broadening following @xcite .
the use of the blackwell approximation for the damping constants gives the same result within the errors , when comparing [ fe / h ] obtained with the `` good line set '' of fe i ( see definition of line sets on sec .
in addition , a model atmosphere for the effective temperature and surface gravity appropriate for each star and a value for the microturbulent velocity are also required .
we use the grid of model atmospheres from @xcite with a metallicity of [ fe / h ] = @xmath10.5 dex , based on earlier high dispersion abundance analysis of m71 red giants @xcite .
the final result for [ fe / h ] is not sensitive to small changes in the metallicity of the model atmosphere . in particular ,
the error introduced by using a model with [ fe / h ] = @xmath10.5 instead of @xmath10.7 dex is very small ( see tables 2 & 3 ) .
the effective temperatures and surface gravities are derived from the photometry of the stars as described in paper i. the error in the photometric @xmath0 is 75 k for giants and 150 k for the dwarfs and the error in the photometric log(@xmath2 ) is 0.2 dex ( paper i ) .
the microturbulent velocity is derived spectroscopically ( see below ) .
the stellar parameters are listed in table 1 .
transition probabilities for the fe i lines were obtained from several laboratory experiments , including studies of fe i absorption lines produced by iron vapor in a carbon tube furnace @xcite ( oxford group ) , measurement of radiative lifetimes of fe i transitions by laser induced fluorescence @xcite , fe i emission line spectroscopy from a low current arc @xcite , and emission lines of fe i from a shock tube @xcite .
we also considered solar gf
values from @xcite when needed .
we compare the gf
values obtained by the different experiments in an attempt to place them onto a common scale with respect to the results from @xcite , who provided the longest list of gf values .
we considered for the comparison only the set of lines present in our data , which have a wavelength coverage roughly from 5380 to 7900 .
we found that the values of @xcite and of the oxford group were on the same scale ; the mean difference in log(@xmath3 ) between the two experiments is @xmath4 for 21 lines in common .
similar results are found when considering 21 lines in common between @xcite and @xcite ( see also * ? ? ?
* ) , where the mean difference in log(@xmath3 ) is @xmath5 .
considering 34 lines in common between @xcite and @xcite , the mean difference in log(@xmath3 ) is @xmath6 . when comparing the 14 lines in common between @xcite and @xcite we found a mean difference in log(@xmath3 ) of @xmath7 .
we also compared the results from @xcite with solar gf
values obtained by @xcite .
we found that the mean difference in log(@xmath3 ) is @xmath8 when comparing 68 lines in common between these latest works .
similar offsets are found by @xcite .
values for our fe i lines were taken when possible from laboratory data in the following order of priority : from @xcite , from the oxford group , from @xcite , and from @xcite corrected by @xmath9 dex . in the cases where no laboratory data were available
, we used solar gf
values from @xcite corrected by + 0.05 dex .
transition probabilities for the fe ii lines were taken from the solar analysis of @xcite , @xcite , and from the semiempirical calculations of @xcite .
while restricting ourselves to lines present in our spectra , we compared the gf
values from @xcite and @xcite to the study of @xcite .
there are 9 lines in common between @xcite and @xcite .
the mean difference between the two in log(@xmath3 ) is @xmath10 .
a similar offset was found by @xcite .
we found no significant difference between the results of @xcite and @xcite , since the 9 lines in common result in a mean difference of @xmath11 .
values for our fe ii lines were taken in the following order of priority : from @xcite , from @xcite corrected by 0.14 dex , and from @xcite .
our sample contains many stars observed in mutiple orders , with many detectable absorption features in these high s / n spectra .
for example , in the coolest m71 star in our sample , m71 - 1 - 45 , 1407 absorption lines have been identified . a fortran code to automatically search for absorption features and measure their equivalent width ( @xmath12 ) , ewdet , was developed for this project .
the code is available upon request to svr .
ewdet determines the continuum location of the hires spectra by fitting a curve to the spectra performing several iterations of point rejection above and below sigma levels given by the user
. then , ewdet identifies lines above the noise level defined by a factor of two of the continuum rms dispersion .
each of the identified lines is fit by a gaussian profile and then the @xmath12 are computed by the integration of the fitted gaussian .
the error in @xmath12 is computed by adding quadratically the error at each point of integration , @xmath13 , times the step of the integration .
the error at each point of integration is given by : @xmath14\ ] ] where @xmath15 is the gaussian profile , @xmath16 is the peak of the gaussian , @xmath17 is the error in the peak of the gaussian , @xmath18 is the central wavelength of the gaussian , @xmath19 is the error in the central wavelength , @xmath20 is the dispersion of the gaussian , @xmath21 is the error in the dispersion of the gaussian , and @xmath22 is the error in the continuum .
the errors of the gaussian parameters are from the covariance matrix of the gaussian fit .
the expression for the error in @xmath12is deduced by propagating the errors of the gaussian parameters and assuming that the continuum level is equal to one ( see appendix ) .
the fit by a gaussian profile is reasonable even for the strongest lines we use , as shown in figure 1 , where the observed line is plotted with a solid curve and its corresponding gaussian profile fit is shown with a dashed curve . the line list identified and measured by ewdet is then correlated to the line list with the atomic parameters to specifically identify the fe i and fe ii lines . the detailed lists of @xmath12 and gf values will be given in the next paper .
the spectral resolution , @xmath23 , of an echelle is fixed , unlike a low incidence , angle of low order grating spectrograph where @xmath24 is constant , independent of @xmath25 .
hence a line of constant central depth , @xmath26 , will have an equivalent width proportional to @xmath27 .
we construct the relationship between @xmath27 and @xmath12 for the fe i lines to look for possible blends and for saturation effects .
the @xmath27 versus @xmath12 relation for three stars are plotted in figure 2 ; m71-i is one of the most luminous and coolest stars of our sample , m71-g53476_4543 is a medium luminosity and temperature star , and m71-g53392_4624 is one of the faintest and hottest stars of the sample .
we fit a second order polynomial to the @xmath28@xmath12 relationship for each star , performing several iterations of point rejection above and below the two sigma level .
the second order term is needed to account for line saturation at large @xmath12 and the rejection of points is needed to eliminate blended lines . the second order fit is plotted as a solid line in figure 2 , and the points considered for such a fit are shown in black .
for fe i , we use two sets of lines .
the first set , subsequently called `` the weak line set '' , contains those fe i lines which are within two sigma levels of the @xmath28@xmath12 fit , have @xmath12 @xmath29 60 m , and have errors less than a third of the @xmath12 .
this set of lines produces a sample of the best weak fe i lines with the most accurate @xmath12 and the abundances derived using them will have a minimal dependence on the choice of microturbulent velocity .
the second set , subsequently called `` the good line set '' , consists of all the fe i lines with errors less than a third of the @xmath12 and with @xmath12 computed from the fit for @xmath28@xmath12 determined for each star . in future papers ,
the @xmath12 of lines of other elements will be determined in the same manner as the fe i lines of `` the good line set '' . this way a consistent comparison can be done among the resulting abundances without a restriction on the strength of the lines used .
the `` weak line set '' is different for each star .
actually the weak lines for the stars near the main sequence are no longer weak for the cooler stars in our sample .
we compare the results from 20 lines common to 15 stars over almost the whole range in @xmath0 with the results from the `` weak line set '' ; there is no difference within the errors , nor a trend in @xmath0 .
for fe ii , the @xmath12 of the lines are also determined using the fit to the @xmath28@xmath12 relation of the fe i lines of `` the good line set '' .
the fe ii lines follow the relationship determined from fe i lines well , as shown in figure 3 . additional fe ii lines ,
not picked up automatically , were measured by hand for the stars near the turn off and for m71-g53425_4612 and m71-g53457_4709 .
the set of @xmath12 for the hand selected fe ii lines is computed from the @xmath28@xmath12 relation of the fe i lines , after determining their observed depth from the spectra . the number of fe i lines , for both sets of lines , and the number of fe ii lines utilized in this analysis are listed in table 1 .
the effective temperature ( @xmath0 ) of a star can be determined spectroscopically by requiring the abundance to be independent of the lower excitation potential .
this technique can be applied to 20 of our stars where we have fe i lines with enough range in lower excitation potential to do so . for the spectroscopic @xmath0 determination
we are using `` the weak line set '' of fe i lines , because its resulting abundance and spectroscopic @xmath0 will be only weakly dependent on the choice of microturbulent velocity .
we find that the spectroscopic @xmath0 is in good agreement with the photometric @xmath0 derived in paper i , as shown in figure 4 .
the solid line in figure 4 shows the ideal case when the spectroscopic and the photometric @xmath0 are equal .
the scatter around the solid line is about 150 k , which is comparable to the error of the photometric @xmath0 of 75 k for giants and of 150 k for dwarfs ( paper i ) , also shown in figure 4 .
the scatter around the solid line is symmetric , not above or below , indicating the lack of systematic effects with the photometric temperatures .
the microturbulent velocity ( @xmath30 ) of a star can be determined spectroscopically by requiring the abundance to be independent of the strength of the lines measured as the equivalent width .
we apply this technique for both sets of fe i lines .
the resulting @xmath30 and the [ fe / h ] computed with it for `` the weak line set '' of fe i lines are listed in table 1 .
only 18 of our stars have enough weak fe i lines to derive @xmath30 spectroscopically .
we also compute @xmath30 using all the good fe i lines for 20 of our stars .
the difference of the resulting [ fe / h ] computed with the derived @xmath30 for the two set of lines is plotted with respect to @xmath0 in figure 5 .
the mean difference is @xmath31 0.011 , hence ,
as expected , the [ fe / h ] results from both sets of lines show a very good agreement .
the relationship between @xmath30 determined for the set of all good fe i lines and the photometric @xmath0 is shown in figure 6 .
the solid line corresponds to a linear least squares fit of the data , excluding the red horizontal branch ( rhb ) stars , marked with circles .
the best fit line is given by : @xmath32 the scatter around the solid line is about 0.2 km s@xmath33 , which is a reasonable estimation of the error in @xmath30 . for the rest of the analysis , we will use the set of all the good fe i and fe ii lines , with @xmath30 computed from the @xmath30-@xmath0 fit . for the rhb stars we use a value of 1.61 km s@xmath33 , which corresponds to the mean value determined for the three rhb stars .
the microturbulent velocity used for our stars is listed in table 1 .
given the stellar parameters from table 1 , we determined the iron abundance using the equivalent widths of all the good fe i and fe ii lines identified in the hires spectra .
we employ the grid of stellar atmospheres from @xcite to compute the iron abundance using the four stellar atmosphere models with the closest @xmath0 and log(@xmath2 ) to each star s parameters .
the [ fe / h ] listed in table 1 is an interpolation of the results from the closest stellar model atmospheres to the appropriate @xmath0 and log(@xmath2 ) for each star .
the results of [ fe / h ] from fe i lines are listed in column 9 of table 1 and plotted against the photometric @xmath0 in figure 7a .
@xmath0 is used for the x - axis as a convenient parameter for characterizing the position of the stars in the color - magnitude diagram as it also ranks the stars in luminosity ( except for the rhb stars ) .
the errors listed in table 1 correspond to the larger of the statistical uncertainty , given by the standard deviation of the iron abundance from different lines divided by the square root of the number of lines , or a minimum value of 0.03 dex .
these errors are lower limits to the actual uncertainties in the abundances , since they do not include uncertainties due to the stellar parameters nor any systematic effects that might be present .
we estimate the sensitivity of [ fe / h ] derived from fe i lines with respect to the stellar parameters in three cases 4250/1.0/1.0 , 5000/2.5/1.0 and 5500/4.0/1.0 , where the three numbers correspond to @xmath0/log(@xmath2)/@xmath30 .
the results are listed in table 2 , where the range adopted for each parameter is representative of its uncertainty .
our determination of [ fe / h ] from fe i lines is most sensitive to errors in @xmath0 , which is less than @xmath34 0.1 dex for @xmath35@xmath0 of @xmath36 100 k , and have a minimal sensitivity on the choice of metallicity of the model atmosphere grid for plausible changes in [ fe / h ] ( @xmath360.2 dex ) .
the solid line , shown in figure 7a , is a linear fit weighted by the errors of [ fe / h ] versus @xmath0 .
the slope of the fit is @xmath37 dex / k , which is consistent with [ fe / h ] being constant , independent of @xmath0 ( ie , of luminosity or equivalently position in the color - magnitude diagram ) .
we divide our sample in four groups of stars : giant stars at or above the rhb , stars on the rhb , giant stars below the rhb , and main sequence stars near the turnoff .
the mean [ fe / h ] for each group is listed on table 4 .
we found no significant difference in the mean [ fe / h]obtained from fe i lines among the defined groups of stars .
the mean [ fe / h ] weighted by the errors of all 25 stars is @xmath38 , in very good agreement with earlier determinations @xcite .
the determinations of [ fe / h ] from fe ii lines are listed in column 11 of table 1 and plotted against the photometric @xmath0 in figure 7b .
the errors listed in table 1 corresponds to the statistical uncertainty or a value of 0.05 dex , whichever is larger .
we estimate the sensitivity of [ fe / h ] derived from fe ii ines with respect to the stellar parameters in the same manner as the sensitivity of [ fe / h ] from fe i lines .
the results are listed in table 3 , where the range adopted for each parameter is representative of its uncertainty .
we see a stronger sensitivity on the stellar parameters from the fe ii lines than from the fe i lines .
the [ fe / h ] determination from fe ii lines is most sensitive to the systematic error ( note that the internal uncertainty in log(@xmath2 ) is @xmath39 0.1 dex ) in log(@xmath2 ) , as well as to @xmath0 among the coolest m71 giants .
the sensitivity on the choice of metallicity of the model atmosphere grid is small for reasonable changes in metallicity .
the solid line , shown in figure 7b , is a linear fit weighted by the errors of [ fe / h ] versus @xmath0 .
the slope of the fit is @xmath40 dex / k , which is consistent with [ fe / h ] being constant , independent of @xmath0 .
we found that there is no significant difference in the mean [ fe / h ] obtained from fe ii among stars from different luminosity groups , listed in table 4 .
the mean [ fe / h ] weighted by the errors is @xmath41 , in very good agreement with our result from fe i lines and earlier determinations @xcite .
the iron abundance could be affected by departures from lte .
the main nlte effect in late - type stars is caused by overionization of electron donor metals by ultraviolet radiation @xcite .
recently , @xcite and @xcite studied nlte effects in fe abundances in metal - poor late - type stars .
@xcite found that nlte corrections for fe lines are very small in dwarfs of any @xmath0 , and only small corrections ( @xmath29 0.1 dex ) are expected for stars on the red giant branch .
@xcite found that nlte corrections become less important as [ fe / h ] increases , being less than 0.1 dex for stars with [ fe / h]@xmath420.75 dex , and that ionized lines are not significantly affected by nlte .
one way to explore possible nlte effects present in our data is by comparing the results from fe i and fe ii lines .
the difference between [ fe / h ] from fe ii and fe i lines is plotted in figure 8 against @xmath0 .
the solid line is a linear fit weighted by the errors .
the slope of the fit is @xmath43 dex / k , which is nearly flat .
the mean difference is @xmath44 .
we conclude that nlte effects are negligible in our iron abundance determination , as expected from results of earlier studies @xcite .
our [ fe / h ] abundance results involve for the first time a wide luminosity sample of stars , which includes at the same time stars from luminous giants to stars near the turnoff .
we find that the [ fe / h ] abundance , from both fe i and fe ii lines , is independent of @xmath0 , and equivalently luminosity .
our result is in agreement with the work of @xcite .
they present abundances from high dispersion spectra from the vlt of stars in ngc 6397 and ngc 6752 .
they found that the [ fe / h ] obtained for stars at the base of the subgiant branch agrees within a few percent with the [ fe / h ] obtained for stars at the main sequence turnoff , and further compare this value with analysis of the rgb stars in this cluster by other groups .
note that the luminosity range of the sample presented in our work is several orders of magnitude wider than the luminosity range of @xcite s sample .
our results , and those of @xcite , appear to be in disagreement with inhomogeneities in [ fe / h ] found earlier by @xcite .
they obtained [ fe / h][email protected] dex for a sample of subgiant stars in m92 , which is a factor of two smaller than [ fe / h ] measurements using red giants in the same cluster @xcite .
@xcite compare their result for the m92 subgiants with analysis of rbg stars by other groups , who may have determined the stellar parameters and performed the abundance determinations in a different way .
this possible difference in the analysis of the giant and subgiant sample may account for the difference in [ fe / h ] found by @xcite or perhaps the determination of the stellar parameters by @xcite is flawed .
our result , on the other hand , is robust , because we have determined both the stellar parameters and the fe abundance in a homogeneous and consistent manner for all our stars .
@xcite also found that ngc 6397 is homogeneous in both o and fe , while an o - na anticorrelation is present among unevolved stars in ngc 6752 , which is very difficult to explain by the deep mixing scenario .
lines from many additional species , including o , na , mg , ti , sc among others , are observed in our hires spectra .
we plan to present additional information in the matter of light , iron - peak , and heavy elements in the near future .
we present results of a high dispersion analysis of fe i and fe ii lines to obtain [ fe / h ] for 25 members of the galactic globular cluster m71 .
our sample of stars includes 19 giant stars , 3 horizontal branch stars , and 3 stars near the turnoff .
our conclusions are summarized as follows : * the [ fe / h ] obtained from fe i lines agrees very well with the [ fe / h ] obtained from fe ii lines . *
the mean [ fe / h ] obtained from fe i and fe ii lines of all 25 stars is in good agreement with earlier determinations . *
the [ fe / h ] obtained from both fe i and fe ii lines is independent of @xmath0 , and equivalently luminosity . *
no difference is found among the mean [ fe / h ] from giant stars located at or above the rhb , rhb stars , giant stars located below the rhb and stars near the main sequence turnoff . in the near future
, we will present the result from the analysis now underway of additional elements .
the entire keck / hires and lris user communities owes a huge debt to jerry nelson , gerry smith , steve vogt , bev oke , and many other people who have worked to make the keck telescope and hires and lris a reality and to operate and maintain the keck observatory .
we are grateful to the w. m. keck foundation for the vision to fund the construction of the w. m. keck observatory .
we thank r. gratton for providing a detailed description of his automatic equivalent width measuring program and r. pogge for providing the gaussian profile fitting routine .
partial support to mmb was provided by a theodore dunham , jr .
grant for research in astronomy and the national science foundation under grant ast-9624680 to mmb and grant ast-9819614 to jgc .
the equivalent width of a line is defined as : @xmath45 where , @xmath46 is the flux of the spectrum , @xmath47 is the continuum level , and @xmath24 is the step that can be made arbitrarily small to better approximate the integral . in this case
, @xmath46 is the gaussian profile of the line , given by : @xmath48\ ] ] where @xmath16 is the peak of the gaussian , @xmath18 is the central wavelength , and @xmath20 is the dispersion of the gaussian . if the equivalent width is computed through out the summation then the error in the equivalent width , @xmath49 will be the quadratic summation of the errors in the individual points times the step : @xmath50 where : @xmath51^{2 } + \left [ \frac{\partial w}{\partial cont } \sigma_{cont } \right]^{2}\ ] ] @xmath52^{2 } + \left [ \frac{g(\lambda_{i})}{cont(\lambda_{i})^{2 } } \sigma_{cont } \right]^{2}\ ] ] if we assume that the continuum level is equal to one then : @xmath53 since the gaussian depends on three parameters : @xmath54^{2 } + \left [ \frac{\partial g}{\partial \lambda_{cen } } \sigma_{\lambda_{cen } } \right]^{2 } + \left [ \frac{\partial g}{\partial \sigma } \sigma_{\sigma } \right]^{2}\ ] ] @xmath55^{2 } + \left [ \frac{(\lambda_{i } - \lambda_{cen})}{\sigma^{2 } } g(\lambda_{i } ) \
\sigma_{\lambda_{cen } } \right ] ^{2 } + \left [ \frac{(\lambda_{i } - \lambda_{cen})^{2}}{\sigma^{3 } } g(\lambda_{i } ) \
\sigma_{\sigma } \right]^{2}\ ] ] @xmath56\ ] ] finally : @xmath57\ ] ] gratton , r. g. , bonifacio , p. , bragaglia , a. , carretta , e. , castellani , v. , centurion , m. , chieffi , a. , claudi , r. , clementini , g. , dantona , f. , desidera , s. , francois , p. , grundahl , f. , lucatello , s. , molaro , p. , pasquini , l. , sneden , c. , spite , f. , & straniero , o. , 2001 , , submitted lcccccccccl [ tab1 ] 145 & 3950 & 0.90 & 1.48 & 59 & 0.74 @xmath36 0.03 & 1.46 & 187 & 0.78 @xmath36 0.03 & 6 & 0.61 @xmath36 0.11 + i & 4150 & 1.00 & 1.00 & 67 & 0.69 @xmath36 0.03 & 1.37 & 186 & 0.76 @xmath36 0.03 & 6 & 0.88 @xmath36 0.07 + 166 & 4250 & 1.35 & 1.80 & 68 & 0.83 @xmath36 0.03 & 1.32 & 179 & 0.71 @xmath36 0.03 & 6 & 0.86 @xmath36 0.09 + 164 & 4200 & 1.35 & 1.57 & 61 & 0.76 @xmath36 0.03 & 1.34 & 187 & 0.74 @xmath36 0.03 & 5 & 0.73 @xmath36 0.09 + 156 & 4525 & 1.60 & 0.81 & 25 & 0.63 @xmath36 0.04 & 1.19 & 127 & 0.56 @xmath36 0.03 & 2 & 0.83 @xmath36 0.14 + 195 & 4550 & 1.65 & 1.00 & 79 & 0.68 @xmath36 0.03 & 1.18 & 184 & 0.67 @xmath36 0.03 & 8 & 0.87 @xmath36 0.05 + 181 & 4550 & 1.75 & 1.50 & 77 & 0.71 @xmath36 0.03 & 1.18 & 180 & 0.64 @xmath36 0.03 & 6 & 1.05 @xmath36 0.05 + 11 & 4700 & 2.05 & 0.89 & 55 & 0.62 @xmath36 0.03 & 1.11 & 134 & 0.67 @xmath36 0.03 & 5 & 0.86 @xmath36 0.05 + 180 & 5300 & 2.45 & ... & 28 & 0.68 @xmath36 0.04 & 1.61 & 71 & 0.69 @xmath36 0.03 & 5 & 0.92 @xmath36 0.05 + 187 & 5300 & 2.45 & 1.62 & 68 & 0.58 @xmath36 0.03 & 1.61 & 128 & 0.60 @xmath36 0.03 & 9 & 0.83 @xmath36 0.05 + 194 & 5300 & 2.45 & ... & 42 & 0.78 @xmath36 0.03 & 1.61 & 94 & 0.77 @xmath36 0.03 & 6 & 0.81 @xmath36 0.05 + 160 & 4900 & 2.30 & 0.70 & 49 & 0.77 @xmath36 0.03 & 1.02 & 119 & 0.80 @xmath36 0.03 & 6 & 0.70 @xmath36 0.05 + 159 & 4600 & 2.30 & 1.50 & 62 & 0.79 @xmath36 0.03 & 1.16 & 141 & 0.78 @xmath36 0.03 & 5 & 0.70 @xmath36 0.05 + g53476_4543 & 4900 & 2.65 & 0.83 & 94 & 0.64 @xmath36 0.03 & 1.02 & 174 & 0.68 @xmath36 0.03 & 7 & 0.84 @xmath36 0.05 + 2160 & 5100 & 2.70 & 1.10 & 68 & 0.59 @xmath36 0.03 & 0.92 & 145 & 0.54 @xmath36 0.03 & 5 & 0.97 @xmath36 0.08 + g53447_4707 & 5175 & 2.75 & 1.35 & 90 & 0.62 @xmath36 0.03 & 0.89 & 155 & 0.57 @xmath36 0.03 & 7 &
0.86 @xmath36 0.05 + g53445_4647 & 5050 & 2.85 & 0.54 & 50 & 0.54 @xmath36 0.03 & 0.95 & 112 & 0.65 @xmath36 0.03 & 6 & 0.85 @xmath36 0.05 + g53447_4703 & 5000 & 3.00 & 0.90 & 62 & 0.72 @xmath36 0.03 & 0.97 & 125 & 0.77 @xmath36 0.03 & 4 & 0.80 @xmath36 0.05 + g53425_4612 & 5150 & 3.15 & 1.40 & 36 & 0.77 @xmath36 0.03 & 0.90 & 80 & 0.73 @xmath36 0.03 & 2 & 0.91 @xmath36 0.08 + g53477_4539 & 5150 & 3.15 & ... & 56 & 0.66 @xmath36 0.03 & 0.90 & 119 & 0.70 @xmath36 0.03 & 5 & 0.90 @xmath36 0.05 + g53457_4709 & 5200 & 3.35 & 1.24 & 58 & 0.78 @xmath36 0.03 & 0.88 & 93 & 0.78 @xmath36 0.03 & 5 & 0.76 @xmath36 0.11 + g53391_4628 &
5100 & 3.35 & ... & 55 & 0.74 @xmath36 0.03 & 0.92 & 106 & 0.84 @xmath36 0.03 & 5 & 0.81 @xmath36 0.07 + g53417_4431 & 5800 & 4.05 & ... & 19 & 0.66 @xmath36 0.03 & 0.60 & 38 & 0.68 @xmath36 0.04 & 3 & 0.61 @xmath36 0.12 + g53392_4624 & 5800 & 4.05 & ... & 23 & 0.81 @xmath36 0.04 & 0.60 & 36 & 0.81 @xmath36 0.03 & 3 & 0.66 @xmath36 0.08 + g53414_4435 & 5900 & 4.15 & ... & 5 & 0.82 @xmath36 0.12 & 0.55 & 13 & 0.83 @xmath36 0.04 & 2 & 0.58 @xmath36 0.17 + ccccc [ tab4 ] rgb at or above rhb & 13.46 & 10 & @xmath600.07 & @xmath610.12 + rhb & 14.50 & 3 & @xmath620.07 & @xmath630.05 + rgh below rhb & 15.92 & 9 & @xmath640.09 & @xmath650.06 + ms to & 17.76 & 3 & @xmath660.06 & @xmath670.13 + | we present [ fe / h ] abundance results that involve a sample of stars with a wide range in luminosity from luminous giants to stars near the turnoff in a globular cluster .
our sample of 25 stars in m71 includes 10 giant stars more luminous than the rhb , 3 horizontal branch stars , 9 giant stars less luminous than the rhb , and 3 stars near the turnoff .
we analyzed both fe i and fe ii lines in high dispersion spectra observed with hires at the w. m. keck observatory .
we find that the [ fe / h ] abundances from both fe i and fe ii lines agree with each other and with earlier determinations .
also the [ fe / h ] obtained from fe i and fe ii lines is constant within the rather small uncertainties for this group of stars over the full range in @xmath0 and luminosity , suggesting that nlte effects are negligible in our iron abundance determination . in this globular cluster ,
there is no difference among the mean [ fe / h ] of giant stars located at or above the rhb , rhb stars , giant stars located below the rhb and stars near the turnoff . |
test data generation ( tdg ) aims at automatically generating test - cases for interesting test _
coverage criteria_. the coverage criteria measure how well the program is exercised by a test suite .
examples of coverage criteria are : _ statement coverage _ which requires that each line of the code is executed ; _ path coverage _ which requires that every possible trace through a given part of the code is executed ; etc .
there are a wide variety of approaches to tdg(see @xcite for a survey ) .
our work focuses on _ glass - box _
testing , where test - cases are obtained from the concrete program in contrast to _ black - box _
testing , where they are deduced from a specification of the program . also , our focus is on _ static _ testing , where we assume no knowledge about the input data , in contrast to _ dynamic _ approaches @xcite which execute the program to be tested for concrete input values .
the standard approach to generating test - cases statically is to perform a _
symbolic _ execution of the program @xcite , where the contents of variables are expressions rather than concrete values .
the symbolic execution produces a system of _ constraints _ consisting of the conditions to execute the different paths .
this happens , for instance , in branching instructions , like if - then - else , where we might want to generate test - cases for the two alternative branches and hence accumulate the conditions for each path as constraints . the symbolic execution approach is usually combined with the use of _ constraint solvers _ in order to : handle the constraints systems by solving the feasibility of paths and , afterwards , to instantiate the input variables .
tdg for declarative languages has received comparatively less attention than for imperative languages . in general ,
declarative languages pose different problems to testing related to their own execution models , like laziness in functional programming ( fp ) and failing derivations in constraint logic programming ( clp ) .
the majority of existing tools for fp are based on black - box testing ( see e.g. @xcite ) .
an exception is @xcite where a glass - box testing approach is proposed to generate test - cases for curry . in the case of clp ,
test - cases are obtained for prolog in @xcite ; and very recently for mercury in @xcite .
basically the test - cases are obtained by first computing constraints on the input arguments that correspond to execution paths of logic programs and then solving these constraints to obtain test inputs for such paths . in recent work @xcite ,
we have proposed to employ existing _ partial evaluation _ ( pe ) techniques developed for clpin order to automatically generate _ test - case generators _ for glass - box testing of bytecode .
pe @xcite is an automatic program transformation technique which has been traditionally used to specialise programs w.r.t . a known part of its input data and , as futamura predicted , can also be used to compile programs in a ( source ) language to another ( object ) language ( see @xcite ) .
the approach to tdgby pe of @xcite consists of two independent clp pe phases .
( 1 ) first , the bytecode is transformed into an equivalent ( decompiled ) clp program by specialising a bytecode interpreter by means of existing pe techniques .
( 2 ) a second pe is performed in order to supervise the generation of test - cases by execution of the clp decompiled program .
interestingly , it is possible to employ control strategies previously defined in the context of clp pe in order to capture _ coverage criteria _ for glass - box testing of bytecode .
a unique feature of this approach is that , this second pe phase allows generating not only test - cases but also test - case _ generators_. another important advantage is that , in contrast to previous work to tdgof bytecode , it does not require devising a dedicated symbolic virtual machine . in this work
, we study the application of the above approach to tdgby means of pe to the prolog language .
compared to tdgof an imperative language @xcite , dealing with prolog brings in as the main difficulty to generate test - cases associated to failing computations .
this happens because an intrinsic feature of pe is that it only produces results associated to the _ non - failing _ derivations .
while this is what we need for tdgof an imperative language ( like bytecode above ) , we now want to capture non - failing derivations in prolog and still rely on a standard partial evaluator .
our proposal is to transform the original prolog program into an equivalent prolog program with explicit failure by partially evaluating a prolog interpreter which captures failing derivations w.r.t .
the input program .
this transformation is done in the phase ( 1 ) above . as another difference , in the case of bytecode , the underlying constraint domain only manipulates integers .
however , the above phase ( 2 ) should properly handle the data manipulated by the program in the case of prolog .
compared to existing approaches to tdgof prolog @xcite , our approach basically is of interest for bringing the advantages which are inherent in tdgby pe to the field of prolog : * it is _ more powerful _ in that we can produce test - case generators which are clp programs whose execution in clp returns further test - cases on demand without the need to start the tdg process from scratch ; * it is more _ flexible _ , as different coverage criteria can be easily incorporated to our framework just by adding the appropriate local control to the partial evaluator .
* it is _ simpler _ to implement compared to the development of a dedicated test - case generator , as long as a clp partial evaluator is available .
the rest of the paper is organized as follows . in the next section ,
we give some basics on pe of logic programs and describe in detail the approach to tdgby pe proposed in @xcite .
[ sec : control_flow ] discusses some fundamental issues like the prolog control - flow and the notion of computation path .
then , sect .
[ sec : explicit_failure ] describes the program transformation to make failure explicit , sect .
[ sec : gener - test - cases ] outlines existing methods to properly handle symbolic data during the tdg phase , and finally sect .
[ sec : future ] concludes and discusses some ideas for future work .
in this section we recall the basics of partial evaluation of logic programming and summarize the general approach of relying on partial evaluation of clp for tdg of an imperative language , as proposed in @xcite .
we assume familiarity with basic notions of logic programming and partial evaluation ( see e.g. @xcite ) .
partial evaluation is a semantics - based program transformation technique which specialises a program w.r.t .
given input data , hence , it is often called _
program specialisation_. essentially , partial evaluators are non - standard interpreters which evaluate goals as long as termination is guaranteed and specialisation is considered profitable . in logic programming ,
the underlying technique is to construct ( possibly ) _ incomplete _ sld trees for the set of atoms to be specialised . in an incomplete tree , it is possible to choose _ not _ to further unfold a goal .
therefore , the tree may contain three kinds of leaves : failure nodes , success nodes ( which contain the empty goal ) , and non - empty goals which are not further unfolded .
the latter are required in order to guarantee termination of the partial evaluation process , since the sld being built may be infinite . even if the sld trees for fully instantiated initial atoms ( as regards the _ input _ arguments ) are finite , the sld trees produced for partially instantiated initial atoms may be infinite .
this is because the sld for partially instantiated atoms can have ( infinitely many ) more branches than the actual sld tree at run - time .
the role of the _ local control _ is to determine how to construct the ( incomplete ) sld trees . in particular , the _ unfolding rule
_ decides , for each resolvent , whether to stop unfolding or to continue unfolding it and , if so , which atom to select from the resolvent . on the other hand ,
partial evaluators need to compute sld - trees for a number of atoms in order to ensure that all atoms which appear in non - failing leaves of incomplete sld trees are `` covered '' by the root of some tree ( this is known as the closedness condition of partial evaluation @xcite ) .
the role of the _ global control _ is to ensure that we do not try to compute sld trees for an infinite number of atoms .
the usual way of achieving this is by applying an _ abstraction operator _ which performs `` generalizations '' on the atoms for which sld trees are to be built .
the global control returns a set of atoms @xmath0 .
finally , the partial evaluation can then be systematically extracted from the set @xmath0 ( see @xcite for details ) .
traditionally , there have been two different approaches regarding the way in which control decisions are taken , _ on - line _ and _ off - line _ approaches . in _ online _ pe , all control decisions are dynamically taken during the specialisation phase . in _ offline _ pe , a set of previously computed annotations ( often manually provided ) gives information to the control operators to decide , 1 ) when to stop unfolding ( _ memoise _ ) in the local control , and 2 ) how to perform generalizations in the global control .
the development of pe techniques has allowed the so - called `` interpretative approach '' to compilation which consists in specialising an interpreter w.r.t . a fixed object code .
interpretive compilation was proposed in futamura s seminal work @xcite , whereby compilation of a program @xmath1 written in a ( _ source _ ) programming language @xmath2 into another ( _ object _ ) programming language @xmath3 is achieved by partially evaluating an interpreter for @xmath2 written in @xmath3 w.r.t .
the advantages of interpretive ( de-)compilation w.r.t.dedicated ( de-)compilers are well - known and discussed in the pe literature ( see , e.g. , @xcite ) . very briefly , they include : _ flexibility _ , it is easier to modify the interpreter in order to tune the decompilation ( e.g. , observe new properties of interest ) ; _ easier to trust _ , it is more difficult to prove that ad - hoc decompilers preserve the program semantics ; _ easier to maintain _ , new changes in the language semantics can be easily reflected in the interpreter . in recent work
, we have proposed an approach to test data generation ( tdg ) by pe of clp @xcite and used it for tdgof bytecode .
the approach is generic in that the same techniques can be applied to tdg other both low and high - level imperative languages . in figure
[ fig : overview ] we overview the main two phases of this technique . in * phase
i * , the input program written in some ( imperative ) language @xmath4 is compiled into an equivalent clp program @xmath5 .
this compilation can be achieved by means of an ad - hoc decompiler ( e.g. , an ad - hoc decompiler of bytecode to prolog @xcite ) or , more interestingly , can be achieved automatically by relying on the first futamura projection by means of pe for logic programs as explained above ( e.g. , @xcite ) .
now , the aim of * phase ii * is to generate test - cases which traverse as many different execution paths of @xmath6 as possible , according to a given coverage criteria . from this perspective
, different test data will correspond to different execution paths . with this aim , rather than executing the program starting from different input values ,
the standard approach consists in performing _ symbolic execution _ such that a single symbolic run captures the behavior of ( infinitely ) many input values .
the central idea in symbolic execution is to use constraint variables instead of actual input values and to capture the effects of computation using constraints .
hence , the compilation from @xmath4 to clp allows us to use the standard clp execution mechanism to carry out this phase .
in particular , by running the @xmath5 program without input values , each successful execution corresponds to a different computation path in @xmath6 . rather than relying on the standard execution mechanism , we have proposed in @xcite to use pe of clp to carry out * phase ii*. essentially , we can rely on a clp partial evaluator which is able to solve the constraint system , in much the same way as a symbolic abstract machine would do .
note that performing symbolic execution for tdgconsists in building a finite ( possibly unfinished ) evaluation tree by using a non - standard execution strategy which ensures both a certain coverage criterion and termination .
this is exactly the problem that _ unfolding rules _ , used in partial evaluators of ( c)lp , solve .
in essence , partial evaluators are non - standard interpreters which receive a set of partially instantiated atoms and evaluate them as determined by the so - called unfolding rule .
thus , the role of the unfolding rule is to supervise the process of building finite ( possibly unfinished ) sld trees for the atoms .
this view of tdgas a pe problem has important advantages .
first , we can directly apply existing , powerful , unfolding rules developed in the context of pe .
second , it is possible to explore additional abilities of partial evaluators in the context of tdg . in particular , the generation of a residual program from the evaluation tree returns a program which can be used as a _ test - case generator _
, i.e. , a clp program whose execution in clp returns further test - cases on demand without the need to start the tdgprocess from scratch .
in the rest of the paper , we study the application of this general approach to tdgof prolog programs .
as we have already mentioned , test data generation is about producing test - cases which traverse as many different execution paths as possible . from this perspective
, different test data should correspond to different execution paths .
thus , a main concern is to specify the computation paths for which we will produce test - cases .
this requires first to determine the control flow of the considered language . in this section
, we aim at defining the control flow of prolog programs that we will use for tdg .
test data will be generated for the computation paths in the control flow .
as usual a prolog program consists of a set of predicates , where each predicate is defined as a sequence of clauses of the form @xmath7 : - @xmath8 with @xmath9 .
a predicate is univocally determined by its _ predicate signature _
@xmath10 , being @xmath11 the name of the predicate and @xmath12 its arity . throughout the rest of the paper we will consider prolog programs with the following features : * rules are normalized , i.e.
, arguments in the head of the rule are distinct variables .
the corresponding bindings will appear explicitly in the body as unifications .
* atoms appearing in the bodies of rules can be : unifications ( considered as builtins ) , calls to defined predicates , term checking builtins ( ` = = /2 ` , ` \==/2 ` , etc ) , and arithmetic builtins ( ` is/2 ` , ` < /2 ` , ` = < /2 ` , etc )
. other typical prolog builtins like ` fail/0 ` , ` ! /0 ` , ` if/3 ` , etc , have been deliberately left out to simplify the presentation . *
all predicates must be moded and well - typed .
we will assume the existence of a `` ` : - pred ` '' declaration associated with each predicate specifying the type expected for each argument ( see as example the declarations in fig .
[ fig : cfgs ] ) . note that this assumption is sensible in the context of tdg(as the aim is the automatic generation of test _ input _ ) .
also , it should not be a limitation as analyses that can automatically infer this information exist .
the control flow in prolog programs is significantly more complex than in traditional imperative languages .
the declarative semantics of prolog implies some additional features like : 1 ) several forms of backtracking , induced by the failure of a sub - goal , or by non - deterministic predicates ; or 2 ) forced control flow change by the predicate `` cut '' . traditionally , control - flow graphs ( cfgs for short ) are used to statically represent the control - flow of programs . typically , in a cfg
, nodes are blocks containing a set of sequential instructions , and edges represent the flows that the program can follow w.r.t .
the semantics of the corresponding programming language . in the literature ,
cfgs for prolog ( and mercury ) have been used for the aim of tdgin @xcite ( @xcite for mercury ) . in particular , cfgs
determine the computation paths for which test - cases will be produced .
our framework relies on the cfgs of @xcite which are known as _ p - flowgraph _ s .
as will be explained later , there are some differences between these cfgs and the ones in @xcite which lead to different test - cases . [ cols="^,^ " , ]
figure [ fig : cfgs ] depicts the prolog code together with the corresponding cfgs for predicates ` foo/2 ` and ` sorted/1 ` .
predicate ` foo/2 ` , given a number in its first argument , returns , in the second one , the value ` pos ` if the number is positive and ` zero ` if it is zero .
if the number is negative , it just fails .
predicate ` sorted/1 ` , given a list of numbers , checks whether the list is strictly sorted , in that case it succeeds , otherwise it fails .
the cfgs contain the following nodes : * a non - terminal node associated to each atom in the body of each clause , * a set of terminal nodes `` t@xmath13 '' representing the success of the @xmath14-th clause , and * the terminal node `` f '' to represent failure .
as regards edges , in principle all non - terminal nodes have two output flows , corresponding to the cases where the builtin or predicate call succeeds or fails respectively .
they are labeled as `` yes ''
or `` no '' for builtins ( including unifications ) , and as `` * * rs * * '' ( _ return - after - success _ ) or `` * * rf * * '' ( _ return - after - failure _ ) for predicate calls . there is an exception in the case of unifications where one of the arguments is a variable , in which case the unification can not fail .
this can be known statically by using the mode information .
see for example nodes `` ` z = pos ` '' and `` ` z = zero ` '' in the ` foo/2 ` cfg . both `` yes '' and `` * * rs * * '' edges point to the node representing the next atom in the clause or to the corresponding `` t@xmath13 '' node if the atom is the last one . finally , each `` t@xmath13 '' node has an output edge labeled as `` redo '' to represent the case in which the predicate is asked for more solutions .
all `` no '' , `` * * rf * * '' and `` redo '' edges point either to the node corresponding to the first previous non - deterministic call in the same clause , or the first node of the following clause , or the `` f '' node if no node meets the above conditions . see as an example the `` * * rs * * '' and `` * * rf * * '' edges from the non - terminal node for sorted([y|r ] ) . in order to define the computation paths determined by the cfgs
, every edge in every cfg is labeled with a unique natural number . an special edge labeled with `` 0 '' and @xmath10 represents the entry of predicate @xmath10 .
given the cfg for predicate @xmath1 , a _ computation sub - path _ is a sequence of numeric labels ( natural numbers ) @xmath15 s.t .
: * @xmath16 corresponds to either an entry , an `` * * rs * * '' , an `` * * rf * * '' or a `` redo '' edge , * @xmath17 leads to a terminal node or to a predicate call , and * for all consecutive labels @xmath18 , there exists a node corresponding to a builtin in the cfg of @xmath1 , for which @xmath19 is an input flow and @xmath20 is an output flow .
given the cfgs corresponding to the set of predicates defining a program , a _ computation path _
( cp for short ) for predicate @xmath11 is a concatenation @xmath21 ( @xmath22 ) of computation sub - paths such that : * first label in @xmath23 is either @xmath24 , in which case we say it is a _ full _ cp , or corresponds to a `` redo '' edge , in which case we say it is a _ partial _ cp ( pcp for short ) .
* last label in @xmath25 leads to a terminal node in the cfg of @xmath11 .
if it is a @xmath0 node the cp is said to be _ successful _ otherwise it is called _
failing_. * for all @xmath26 whose last label leads to a node corresponding to a predicate call , @xmath27 , @xmath28 is a cp for the called predicate , and : * * if @xmath29 is successful then the first label in @xmath30 corresponds to an `` * * rs * * '' edge , * * otherwise ( @xmath29 is failing ) , it corresponds to an * rf * edge .
* for all @xmath26 whose first label corresponds to a `` redo '' edge flowing from a `` t@xmath31 '' node in the cfg of predicate @xmath32 , @xmath33 , @xmath34 , whose first label corresponds either to an entry edge or to a `` redo '' edge flowing from `` t@xmath35 '' , @xmath36 , of the cfg of @xmath32 .
if a cp contains at least one label corresponding to a `` redo '' flow , then the cp is said to be an _ after - retry _ cp .
the rest of the cps are _ first - try _ cps .
for example in ` foo/2 ` , @xmath37=@xmath38 and @xmath39=@xmath40 are first - try successful cps ; @xmath41=@xmath42 is a first - try failing branch ; @xmath43=@xmath44 is an after - retry successful cp ( although this one is unfeasible as @xmath45 and @xmath46 are disjoint conditions ) , and @xmath47=@xmath48 is an after - retry failing branch . in ` sorted/1 ` , @xmath49=@xmath50 is a first - try successful cp and @xmath51=@xmath52 is a first - try failing cp .
it is interesting to observe the correspondence between the cps and the test data that make the program traverse them . in ` foo/2 ` , @xmath37 is followed by goal ` foo(1,z ) ` , @xmath39 by goal `
foo(0,z ) ` , @xmath41 by ` foo(-1,z ) ` , @xmath43 is an unfeasible path , and @xmath47 is followed by ` foo(0,z ) ` when we ask for more solutions . as regards ` sorted/1 ` , @xmath49 is followed by the goal ` sorted([0,1 ] ) ` and @xmath51 by ` sorted([0,1,0 ] ) ` . as we will see in sect .
[ sec : gener - test - cases ] , these will become part of the test - cases that we automatically infer .
a key feature of our cfgs is that they make explicit the fact that after failing with a clause the computation has to re - try with the following clause , unless a non - deterministic call is left behind .
e.g. , in ` foo/2 ` the cfg makes explicit that the only way to get a first - try failing branch is through the cp @xmath53 , hence traversing , and failing in , both conditions @xmath45 and @xmath46 .
therefore , a test data to obtain such a behavior will be a negative number for argument @xmath54 .
other approaches , like the one in @xcite , do not handle flows after failure in the same way .
in fact , in @xcite , edge `` 3 '' in ` foo/2 ` goes directly to node `` f '' .
it is not clear if these approaches are able to obtain such a test data . as another difference with previous approaches to tdgof prolog
, we want to highlight that we use cfgs just to reason about the program transformation that will be presented in the following section and , in particular , to clarify which features we want to capture .
however , in previous approaches , test - cases are deduced directly from the cfgs .
as we outlined in sect .
[ sec : intro ] , an intrinsic feature of the second phase of our approach is that it can only produce results associated to non - failing derivations .
this is the main reason why the general approach to tdg by pe sketched in sect .
[ sec : basics - tdg - partial ] is directly applicable only to tdg of imperative languages . to enable its application to prolog , we propose a program transformation which makes failure explicit in the prolog program .
the specialisation of meta - programs has been proved to have a large number of interesting applications@xcite .
futamura projection s to derive compiled code , compilers and compiler generators fall into this category .
the specialization of meta - interpreters for non - standard computation rules has also been studied .
furthermore , language extensions and enhancements can be easily expressed as meta - interpreters which perform additional operations to the standard computation . in short
, program specialisation offers a general compilation technique for the wide variety of procedural interpretations of logic programs . among them , we propose to carry out our transformation which makes failure in logic programs explicit by partially evaluating a prolog meta - interpreter which captures failing derivations w.r.t . the original program .
first , in sect . [ sec : interpreter ]
we describe such a meta - interpreter emphasizing the prolog control features which we want to capture . then , sect .
[ sec : control_pe ] describes the control strategies which have to be used in pe in order to produce an effective transformation . given a prolog program and given a goal , our aim is to define an interpreter in which the computation of the program and goal produces the same results as the ones obtained by using the standard prolog computation but with the difference that failure is never reported . instead
, an additional argument @xmath55 will be bound to the value `` yes '' , if the computation corresponds to a successful derivation , and to `` no '' if it corresponds to a failing derivation .
predicate ` solve/4 ` is the main predicate of our meta - interpreter whose first and second arguments are the predicate signature and arguments of the goal to be executed ; and its third argument is the answer ; by now we ignore the last argument .
for instance , the call ` solve(foo/2,[0,z],answer , _ ) ` succeeds with @xmath56 and @xmath57 , and ` solve(foo/2,[-1,z],answer , _ ) ` also succeeds , but with @xmath58 .
the interpreter has to handle the following issues : 1 .
the prolog _ backtracking _ mechanism has to be explicitly implemented .
to this aim , a stack of _ choice points _ is carried along during the computation so that : * if the derivation fails : ( 1 ) when the stack is empty , it ends up with success and returns the value `` no '' , ( 2 ) otherwise , the computation is resumed from the last choice point , if any ; * if it successfully ends : ( 1 ) when the stack is empty , the computation finishes with answer `` yes '' , ( 2 ) otherwise , the computation is resumed from the last choice point .
2 . when backtracking occurs , all variable bindings , between the current point and the choice point to resume from , have to be undone .
3 . the interpreter has to be implemented in a _ big - step _ fashion .
this is a requirement for obtaining an effective decompilation .
more details are given in sect .
[ sec : control_pe ] . ....
solve(p / ar , args , answer , tncps ) : - pred(p / ar , _ ) , build_s0(p / ar , args , s0,outvs ) , exec(args , s0,sf ) , sf = st(_,_,_,outvs',answer , tncps/ _ ) , outvs ' = outvs .
exec(_,s , sf ) : - s = st(_,[],[],outvs , yes , ncps ) , sf = st(_,_,_,outvs , yes , ncps ) .
exec(_,s , sf ) : - s = st(_,[],[_|_],outvs , yes , ncps ) , sf = st(_,_,_,outvs , yes , ncps ) .
exec(_,s , sf ) : - s = st(_,_,[],outvs , no , tncps/0 ) , sf = st(_,_,_,outvs , no , tncps/0 ) .
exec(args , s , sf ) : - s = st(_,[],[cp|cps],_,yes , tncps/0 ) , build_retry_state(args , cp , cps , tncps , s ' ) , exec(args ,
exec(args , s , sf ) : - s = st(_,_,[cp|cps],_,no , tncps/0 ) , build_retry_state(args , cp , cps , tncps , s ' ) , exec(args , s',sf ) . .... .... exec(args , s , sf ) : - s = st(pp,[a|as],cps , outvs , yes , tncps / encps ) , pp = pp(p / ar , clid , pt ) , internal(a ) , functor(a , a_f , a_ar ) , a = .. [ a_f|a_args ] , next(pt , pt ' ) , solve(a_f / a_ar , a_args , ans , encps ' ) , tncps ' is tncps + encps ' , encps '' is encps + encps ' , pp ' = pp(p / ar , clid , pt ' ) , s ' = st(pp',as , cps , outvs , ans , tncps'/encps '' ) , exec(args , s',sf ) .
exec(args , s , sf ) : - s = st(pp,[a|as],cps , outvs , yes , ncps ) , pp = pp(p / ar , clid , pt ) , builtin(a ) , next(pt , pt ' ) , run_builtin(pp , a , ans ) , pp ' = pp(p / ar , clid , pt ' ) , s ' = st(pp',as , cps , outvs , ans , ncps ) , exec(args , s',sf ) .
.... figure [ fig : interpreter ] shows an implementation of a meta - interpreter which handles the above issues .
the fourth argument of the main predicate ` solve/4 ` , named ` tncps ` , contains upon success the total number of choice points not yet considered , whose role will be explained later .
the interpreter assumes that the program is represented as a set of ` pred/2 ` and ` clause/3 ` facts .
there is a ` pred/2 ` fact per predicate providing its predicate signature , number of clauses and mode information ; and a ` clause/3 ` fact per clause providing the actual code and clause identifier .
predicate ` solve/4 ` basically builds an initial state on ` s0 ` , by calling ` build_s0/4 ` , and then delegates on ` exec/3 ` to obtain the final state ` sf ` of the computation . the output information , `
outvs ` , is taken from ` sf ` .
the state carried along is of the form ` st(pp , g , cps , outvs , ans , ncps ) ` , where ` pp ` is the current program point , ` g ` the current goal , ` cps ` is the stack of choice points ( list of program points ) , ` outvs ` the list of variables in ` g ` corresponding to the output parameters of the original goal , ` ans ` the current answer ( `` yes ''
or `` no '' ) and ` ncps ` the number of choice points left behind .
a program point is of the form ` pp(p / ar , clid , pt ) ` , where ` p / ar ` , ` clid ` and ` pt ` are the predicate signature , the clause identifier and the program point of the clause at hand .
predicate ` exec/3 ` implements the main loop of the interpreter .
given the current state in its second argument it produces the final state of the computation in the third one .
it is defined by the seven clauses which are applied in they following situations : @xmath59 : : _ the current goal is empty , the answer `` yes '' and there are no pending choice points . _ then , the computation finishes with answer `` yes '' . the current answer is actually used as a flag to indicate whether the previous step in the computation succeeded or failed ( see the last two ` exec/3 ` clauses ) .
@xmath60 : : _ as @xmath59 but having at least one choice point .
_ this clause represents the solution in which the computation ends .
the 4@xmath61 clause takes the other alternatives .
@xmath62 : : _ the previous step failed and there are no pending choice points_. then , the computation ends with answer `` no '' .
@xmath63 : : _ the current goal is empty , the answer `` yes '' and there is at least one pending choice point .
_ this is the same situation as in the @xmath64 clause , however in this case the alternative of resuming from the last choice point is taken .
the corresponding state ` s ` is built by means of ` build_retry_state/5 ` and the computation is resumed from ` s ` by recursively calling ` exec/3 ` .
@xmath65 : : _ the previous step failed and there is at least one pending choice point . _ then , the computation is resumed from the last choice point in the same way as in the previous clause .
@xmath66 : : _ the first atom to be solved is user - defined . _ a call to ` solve/4 ` handles the atom , and the computation proceeds with the next program point of the same clause which was the current one before calling ` solve/4 ` .
this way of solving a predicate call makes the interpreter _ big - step _ ( issue ( 3 ) above ) .
@xmath67 : : _ the first atom to be solved is a builtin .
_ then , ` run_builtin/3 ` produces the corresponding answer , and the computation proceeds with the following program point .
an interesting observation ( also applicable for the previous clause ) is that the answer obtained from ` run_builtin/3 ` ( or ` solve/4 ` ) is now set up as the answer of the next state .
this will make the computation go through the 3@xmath68 or 5@xmath61 clauses in the following step , if the obtained answer was `` no '' .
the correspondence between these clauses and the flows in the cfgs is as follows : clauses @xmath69 , @xmath64 and @xmath70 represent the output edges from every `` t '' node .
clause @xmath71 represents the `` no '' edges to `` f '' nodes and @xmath72 the `` no '' edges to non - terminal nodes
. finally clauses @xmath73 and @xmath74 represents the execution of builtins and predicate calls in non - terminal nodes and their corresponding `` yes '' edges .
let us now explain how the interpreter handles the above three issues . to handle ( 1 ) , a stack of choice points is carried along within the state , initialised to contain all initial program points of each clause defining the predicate to be solved , except for the first one .
e.g. , the initial stack of choice points for ` sorted/1 ` is ` [ pp(sorted/1,2,1),pp(sorted/1,3,1 ) ] ` .
how this stack is used to perform the backtracking is already explained in the description of the 4@xmath61 and 5@xmath61 ` exec/3 ` clauses above . as regards issue ( 2 ) , a quite simple way to implement this in prolog is to produce the necessary fresh variables every time the computation is resumed .
this is done inside ` build_retry_state/5 ` .
the corresponding unification to link the fresh variables with the original goal variables is made at the end ( see last line of ` solve/4 ` ) .
this is the reason why 1 ) the list of the actual variables used in the current goal needs to be carried along within the state ; and 2 ) the original arguments are carried along as the first argument of ` exec/3 ` , as the original ground arguments provided , have to be used when resuming from a choice point .
finally , it is worth mentioning that ` solve/4 ` does not return the actual stack of choice points but only the number of them .
this means that during a computation the interpreter only considers choice points of the predicate being solved .
the question is then , how can the interpreter backtrack to the last choice point , including those induced by other computations of ` solve/4 ` ?
e.g. , how can the interpreter follow edge `` 13 '' in the cfg of ` sorted/1 ` ? the interpreter performs the backtracking in the following way : 1 ) the total number of choice points left behind , ` tncps ` , is carried along within the state and finally returned in the last argument of ` solve/4 ` .
2 ) the number of choice points corresponding to invoked predicates , ` encps ` , is also carried along .
it is updated right after the call to ` solve/4 ` in the 6@xmath61 clause of ` exec/3 ` .
both numbers are stored in the last argument of the state as ` tncps / encps ` .
3 ) execution is resumed from choice points of the current predicate only if @xmath75 , as it can be seen in the 4@xmath61 and 5@xmath61 clauses .
otherwise , the computation just fails and prolog s backtracking mechanism is used to ask the last invoked predicate for more solutions .
this indeed means that the non - determinism of the program is still implicit .
the specialisation of interpreters has been studied in many different contexts , see e.g. @xcite .
very recently , @xcite proposed control strategies to successfully specialise low - level code interpreters w.r.t . non trivial programs . here
we demonstrate how such guidelines can be , and should be , used in the specialisation of non - trivial prolog meta - interpreters .
they include : 1 .
_ big - step _ interpreter .
this solves the problem of handling recursion ( see @xcite ) and enables a compositional specialisation w.r.t .
the program procedures ( or predicates ) . note that an effective treatment of recursion is specially important in prolog programs where recursion is heavily used .
optimality _ issues .
optimality must ensure that : a ) the code to be transformed is traversed exactly once , and b ) residual code is emitted once in the transformed program . to achieve optimality , during unfolding ,
all atoms corresponding with _ divergence _ or _ convergence points _ in the cfg of the program to be transformed , has to be _ memoised _ ( see sect . [
sec : pe - basics ] ) . a divergence ( convergence )
point is a program point from ( to ) which two or more flows originate ( converge ) .
we already explained that the interpreter in fig .
[ fig : interpreter ] is big - step . as regards optimality , by looking at the cfgs of fig .
[ fig : cfgs ] , we can observe : 1 ) all program points are divergence points except those corresponding with unifications in which one argument is a variable , and 2 ) the first program point of every clause , except for the one of the first clause , is a convergence point .
we assume that and denote , respectively , the set of convergence points and divergence points of a predicate .
we follow the syntax of @xcite for pe annotations .
an annotation is of the form `` @xmath76 \rightarrow ann ~pred$ ] '' where @xmath77 is an optional precondition defined as a logic formula , @xmath78 is the kind of annotation ( only * memo * in this case ) , and @xmath79 is a predicate descriptor , i.e. , a predicate function and distinct free variables .
then , to achieve an effective transformation , we specialise the interpreter in fig . [
fig : interpreter ] w.r.t .
the program to be transformed by using the following annotation for each predicate ` p / ar ` in the program : @xmath80 additionally ` solve/4 ` and ` run_builtin/3 ` are also annotated to be memoised always to avoid code duplications .
this already describes how the specialisation has to be steered in the local control .
as regards the global control , the only predicate which can introduce non - termination is ` exec/3 ` .
its first and third arguments contain a fixed structure with variables .
the second one might be problematic as it ranges over the set of all computable states at specialisation time .
note that the number of computable states remains finite thanks to the big - step nature of the interpreter .
still , it can happen that the same program point is reached with different values for the ` ncps ` sub - term of the state .
therefore , if one wants to achieve the optimality criterion above , such argument has to be always generalised in global control .
figure [ fig : transformed ] depicts the transformed code we obtain for predicate ` foo/2 ` .
it can be observed that there is a clear correspondence between the transformed code and the cfg in fig .
[ fig : cfgs ] .
thus , predicate ` solve/4 ` represents the node `` ` x>0 ` '' , ` exec_1/5 ` implements its continuation , whose three clauses correspond to the three sub - paths @xmath81 , @xmath82 and @xmath83 respectively .
predicate ` exec_2/4 ` represents the node `` ` x=0 ` '' and ` exec_3/5 ` implements its continuation , whose two clauses correspond to the sub - paths @xmath84 and @xmath85 .
note that edge `` 8 '' is not considered in the meta - interpreter ( nor in the transformed program ) as it is meaningless for tdg .
it is worth mentioning that the transformed program captures the way in which variable bindings are undone .
for instance in ` solve(foo/2,[c , d],\ldots ) ` , if we keep track of variables ` c ` and ` d ` , it can be seen that ` d ` , which corresponds to variable ` z ` in the original code , is only used for the final unification ` f=[d ] ` , while new fresh variables are used for the unifications with ` pos ` and ` zero ` .
however , variable ` c ` , which corresponds to variable ` x ` in the original code , is actually used for the checks in ` run_builtin_1/2 ` and ` run_builtin_2/2 ` .
this turns out to be fundamental when trying to obtain test data associated to the _ first - try failing _
. it must be the same variable the one which , at the same time , is not `` @xmath86 '' and not `` = 0 '' .
otherwise we can not obtain a negative number as test data for such cp . finally , observe that the original prolog arithmetic builtins have been ( automatically ) transformed into their ` clpfd ` counterparts .
c|c ....
solve(foo/2,[c , d],a , b ) : - run_builtin_1(e , c ) , exec_1(c , e , f , a , b ) , f = [ d ] .
exec_1(a , no , f , g , h ) : - exec_2(a , f , g , h ) .
exec_1(_,yes,[pos],yes,1 ) .
exec_1(a , yes , f , g , h ) : - exec_2(a , f , g , h ) .
exec_2(a , g , h , i ) : - run_builtin_2(k , a ) , exec_3(k , g , h , i ) . .... & .... exec_3(no,[_],no,0 ) .
exec_3(yes,[zero],yes,0 ) .
run_builtin_1(yes , a ) : - a#>0 .
run_builtin_1(no , a ) : - \+ a#>0 . run_builtin_2(yes , a ) : - a#=0 .
run_builtin_2(no , a ) : - \+ a#=0 . ....
once the original prolog program has been transformed into an equivalent prolog program with explicit failure , we can use the approach of @xcite to carry out * phase ii * ( see fig . [
fig : overview ] ) and generate test data both for successful and failing derivations . as we have explained in sect .
[ sec : general - scheme ] , the idea is to perform a second pe over the clp transformed program where the unfolding rule plays the role of the coverage criterion . in
@xcite an unfolding rule implementing the _ block - count(k ) _ coverage criterion was proposed .
a set of computation paths satisfies the _ block - count@xmath87 criterion _ if it includes all terminating computation paths which can be built in which the number of times each block is visited does not exceed the given @xmath88 .
the blocks the criterion refers to are the blocks or nodes in the cfgs of the original prolog program .
as the only form of loops in prolog are recursive calls , the `` @xmath88 '' in the _ block - count@xmath87 _ actually corresponds to the number of recursive calls which are allowed .
unfortunately , the presence of prolog s negation in our transformed programs complicates this phase .
the negation will appear in the transformed program for `` no '' branches originating from nodes corresponding to a ( possibly ) failing builtin .
see for example predicates ` run_builtin_1/3 ` and ` run_builtin_2/3 ` in the transformed code of ` foo/2 ` in fig .
[ fig : transformed ] .
while prolog s negation works well for ground arguments , it gives no information for free variables , as it is required in the evaluation performed during this tdg phase .
in particular , in the ` foo/2 ` example , given the computation which traverses the calls `` ` \+ a#>0 ` '' and `` ` \+ a#=0 ` '' ( corresponding to the path @xmath89 in the cfg ) , we need to infer that `` ` a<0 ` '' . in other words , we need somehow to turn the _ negative _ information into _ positive _ information .
this transformation is straightforward for arithmetic builtins : we just have to replace `` ` \+ e_1#=e_2 ` '' by `` ` e_1#\=e_2 ` '' and `` ` \+ e_1#>e_2 ` '' by `` ` e_1#=<e_2 ` '' , etc .
this transformation allows us to obtain the following set of test - cases for ` foo/2 ` : @xmath90,[pos],yes / first - try}\rangle , & \langle\mbox{\tt [ 1],[\_],no / after - retry}\rangle , \\
\langle\mbox{\tt [ 0],[zero],yes / first - try}\rangle , & \langle\mbox{\tt [ -100],[\_],no / first - retry}\rangle \end{array}\right\}\ ] ] they correspond respectively ( reading by rows ) to the cps @xmath91 , @xmath92 , @xmath40 and @xmath42 .
each test - case is represented as a 3-tuple @xmath93 being @xmath94 the list of input arguments , @xmath95 the list of output arguments and @xmath96 the answer .
the answer takes the form @xmath97 with @xmath98 and @xmath99 , we decided not include in the interpreter the support to calculate the ` first - try / after - retry ` value .
] , so that we obtain sufficient information about the kind of cp to which the test - case corresponds ( see sect .
[ sec : control_flow ] ) .
as there are no recursive calls in ` foo/2 ` such test - cases are obtained using the _ block - count@xmath87 _ criterion for any @xmath88 ( greater than @xmath24 ) . the domain used for the integer number is @xmath100 .
however , it can be the case that negation involves unifications with symbolic data .
for example , the transformed code for ` sorted/1 ` includes the negations `` ` \+ l= [ ] ` '' and `` ` \+ l= [ _ _ ] ` '' .
as before , we might write transformations for the negated unifications involving lists , so that at the end it is inferred that `` ` l= [ _ , _ _ ] ` '' . however this would be too an ad - hoc solution as many distinct term structures , different from lists , can appear on negated unifications .
a solution for this problem has been recently proposed for mercury in the same context @xcite .
it roughly consists in the following : 1 ) it is assumed that each predicate argument is well - typed .
2 ) a domain is initialised for each variable , containing the set of possible functors the variable can take .
3 ) when a negated unification involving an output variable is found ( in their terminology a negated _ decomposition _ ) , the corresponding functor is removed from the variable domain .
it is crucial at this point the assumption that complex unifications are broken down into simple ones .
4 ) finally , a search algorithm is described to generate particular values from the type definition and final domain for the variable .
the technique is implemented using chr and can be directly used in principle for our purposes as well . on the other hand ,
advanced declarative languages like toy @xcite make possible the co - existence of different constraint domains .
in particular , the co - existence of boolean and numeric constraint domains enables the possibility of using _ disequalities _ involving both symbolic data and numbers .
this allows for example expressing the negated unifications `` ` \+ l= [ ] ` '' and `` ` \+ l= [ _ _ ] ` '' as disequality constraints `` ` l/= [ ] ` '' and `` ` l/= [ _ _ ] ` '' . additionally , by relying on the boolean constraint solver , the negated arithmetic builtins `` ` \+ a#>0 ` '' and `` ` \+
a#=0 ` '' can be encoded as `` ` ( a#>0 ) = = false ` '' and `` ` ( a#=0 ) = = false ` '' .
this is in principle a more general solution that we want to explore , although a thorough experimental evaluation needs to be carried out to demonstrate its applicability to our particular context .
now , by using any of the techniques outlined above , we obtain the following set of test - cases for ` sorted/1 ` , using _ block - count@xmath101 _ as the coverage criterion : @xmath102,[],yes / first - try}\rangle , & \langle\mbox{\tt [ [ 0]],[],yes / first - try}\rangle,\\ \langle\mbox{\tt [ [ 0,1]],[],yes / first - try}\rangle , & \langle\mbox{\tt [ [ 0,1,2]],[],yes / first - try}\rangle,\\ \langle\mbox{\tt [ [ 0,1,2,0\textbar\_]],[],no / first - try}\rangle , & \langle\mbox{\tt [ [ 0,1,0\textbar\_]],[],no / first - try}\rangle,\\ \langle\mbox{\tt [ [ 0,0\textbar\_]],[],no / first - try}\rangle \end{array}\right\}\ ] ] they correspond respectively ( reading by rows ) to the cps `` @xmath103 '' , `` @xmath104 '' , `` @xmath105 '' , `` @xmath106 '' , `` @xmath107 '' , `` @xmath108 '' , `` @xmath109 '' .
they are indeed all the paths that can be followed with no more than @xmath110 recursive calls .
this time the domain has been set up to @xmath111 .
very recently , we proposed in @xcite a generic approach to tdg by pe which in principle can be used for any imperative language . however , applying this approach to tdg of a declarative language like prolog introduces some difficulties like the handling of failing derivations and of symbolic data . in this work
, we have sketched solutions to overcome such difficulties . in particular , we have proposed a program transformation , based on pe , to make failure explicit in the prolog programs . to handle prolog s negation in the transformed programs ,
we have outlined existing solutions that make it possible to turn the negative information into positive information .
though our preliminary experiments already suggest that the approach can be very useful to generate test - cases for prolog , we plan to carry out a thorough practical assessment .
this requires to cover additional prolog features like the module system , builtins like ` cut/0 ` , ` fail/0 ` , ` if/3 ` , etc . and also to compare the results with other tdg systems .
we also want to study the integration of other kinds of coverage criteria like _ data - flow _ based criteria .
finally , we would like to explore the use of static analyses in the context of tdg .
for instance , the information inferred by a _ failure analysis _ can be very useful to prune some of the branches that our transformed programs have to consider .
this work was funded in part by the information society technologies program of the european commission , future and emerging technologies under the ist-15905 _ mobius _ project , by the spanish ministry of education under the tin-2005 - 09207 _ merit _ project , and by the madrid regional government under the s-0505/tic/0407 _ promesas _ project .
e. albert , m. gmez - zamalloa , and g. puebla .
est data generation of bytecode by clp partial evaluation . in _
18th international symposium on logic - based program synthesis and transformation ( lopstr08 ) _ , lncs .
springer - verlag , july 2008 . to appear .
f. degrave , t. schrijvers , and w. vanhoof .
automatic generation of test inputs for mercury . in _
18th international symposium on logic - based program synthesis and transformation ( lopstr08 ) _ , lncs .
springer - verlag , 2008 . to appear .
m. gmez - zamalloa , e. albert , and g. puebla .
odular decompilation of low - level code by partial evaluation . in _
8th international working conference on source code analysis and manipulation ( scam08)_. ieee computer society , september 2008 . to appear .
kim s. henriksen and john p. gallagher . abstract interpretation of pic programs through logic programming . in _
scam 06 : proceedings of the sixth ieee international workshop on source code analysis and manipulation _ , pages 184196 .
ieee computer society , 2006 .
g. luo , g. bochmann , b. sarikaya , and m. boyer .
control - flow based testing of prolog programs . in _ in proc . of the 3rd international symposium on software reliability engineering _ ,
pages 104113 , 1992 .
m. mndez - lojo , j. navas , and m. hermenegildo .
lexible ( c)lp - based approach to the analysis of object - oriented programs . in _
17th international symposium on logic - based program synthesis and transformation ( lopstr07 ) _ , august 2007 . | in recent work , we have proposed an approach to test data generation ( tdg ) of imperative bytecode by _ partial evaluation _ ( pe ) of clp which consists in two phases : ( 1 ) the bytecode program is first transformed into an equivalent clp program by means of interpretive compilation by pe , ( 2 ) a second pe is performed in order to supervise the generation of test - cases by execution of the clp decompiled program .
the main advantages of tdgby pe include flexibility to handle new coverage criteria , the possibility to obtain test - case generators and its simplicity to be implemented .
the approach in principle can be directly applied for tdgof any imperative language . however , when one tries to apply it to a declarative language like prolog , we have found as a main difficulty the generation of test - cases which cover the more complex control flow of prolog .
essentially , the problem is that an intrinsic feature of pe is that it only computes non - failing derivations while in tdgfor prolog it is essential to generate test - cases associated to failing computations .
basically , we propose to transform the original prolog program into an equivalent prolog program with _
explicit failure _ by partially evaluating a prolog interpreter which captures failing derivations w.r.t .
the input program .
another issue that we discuss in the paper is that , while in the case of bytecode the underlying constraint domain only manipulates integers , in prolog it should properly handle the symbolic data manipulated by the program .
the resulting scheme is of interest for bringing the advantages which are inherent in tdgby pe to the field of logic programming . |
hoava - lifshitz gravity @xcite is attracting much attention as one of candidates for the theory of quantum gravity because of its power - counting renormalizability , which is realized by the lifshitz scaling @xmath2 with the dynamical critical exponent @xmath3 in the ultraviolet ( uv ) .
there are many attempts to investigate properties and implications of this theory @xcite .
it is natural to suppose that not only gravitational fields but also other fields exhibit the same lifshitz scaling in the uv .
even if they classically have different scalings , quantum corrections should render them to have the same scaling .
a lifshitz scalar with @xmath4 can obtain scale - invariant , super - horizon field fluctuations even without inflation @xcite , thus can source the primordial curvature perturbations through mechanisms such as the curvaton scenario @xcite or the modulated decay @xcite .
it is noteworthy that this value of @xmath5 is the minimal value for which gravity is power - counting renormalizable . in order to discern this production mechanism of the primordial perturbation from others , we need to investigate distinct features in observables such as the cosmic microwave background . in this respect ,
non - gaussianity has been considered as one of the promising approaches to distinguish production mechanisms .
for this reason , there are on - going efforts to detect or constrain non - gaussian nature of the primordial perturbation @xcite . towards identification of the production mechanism by future observations
, theoretical analyses of non - gaussianity in various cosmological scenarios have been performed @xcite . in this paper , we focus on primordial non - gaussianity from a lifshitz scalar and calculate its bispectrum and trispectrum . with the dynamical critical exponent @xmath4 ,
the scaling dimension of the lifshitz scalar is zero and , thus , nonlinear terms in the action are unsuppressed unless forbidden by symmetry or driven to small values by renormalization .
it is those nonlinear terms that we expect to produce non - gaussianity .
even when the lifshitz scalar s field fluctuations are linearly transformed to the curvature perturbations ( which can be realized by the curvaton mechanism or / and modulated decay ) , it turns out that the produced bispectrum can be large enough to be observed in future observations .
we find three independent cubic terms dominant in the uv , each of which gives different shape dependence of the bispectrum .
roughly speaking , they correspond to local , equilateral and orthogonal shapes , respectively . the rest of this paper is organized as follows . in section [ sec : review ] we review generation of scale - invariant cosmological perturbations from a lifshitz scalar . in section [ sec :
estimation ] we estimate the size of non - gaussianity and see that the nonlinear parameter @xmath6 can be as large as @xmath7 . in section [ sec : shape ] we concretely show the momentum dependence of the bispectrum and trispectrum .
section [ sec : summary ] is devoted to a summary of this paper and discussions . in appendix
[ appendix ] we derive the set of independent cubic and quartic terms dominant in the uv .
in this section , we review the mechanism for generation of scale - invariant cosmological perturbations from a lifshitz scalar @xcite .
the action for a lifshitz scalar @xmath8 in minkowski background is @xmath9 , \label{lifshitzscalarminkowski}\end{aligned}\ ] ] where @xmath10 @xmath11 , @xmath12 and @xmath13 are mass scales and @xmath14 are dimensionless constants . here
, it is supposed that the time kinetic term is already canonically normalized , and thus nonlinear terms in the action indicated by @xmath15 do not include time derivatives . on the other hand
, those nonlinear terms can include spatial derivatives .
also , the sign of the first term in the right hand side of ( [ actionphi ] ) is set by requiring stability in the uv . in the uv ,
the first term in @xmath16 is dominant and the field @xmath8 described by the action ( [ actionphi ] ) exhibits the lifshitz scaling ( [ scaling ] ) with @xmath17 we find that for @xmath4 , the scaling dimension of @xmath8 is zero and thus the amplitude of quantum fluctuations of @xmath8 is expected to be independent of the energy scale of the system of interest .
this indicates that the power spectrum of quantum fluctuations of @xmath8 in an expanding universe should be scale - invariant .
intriguingly , the minimal value of @xmath5 for which hoava - lifshitz gravity is power - counting renormalizable is also @xmath18 .
hereafter , we consider the @xmath4 case .
now let us consider the lifshitz scalar @xmath8 , specialized to the case with @xmath4 , in a flat frw background @xmath19 to investigate generation of cosmological perturbations .
we just need to replace the volume element @xmath20 by @xmath21 and the spatial laplacian @xmath22 by @xmath23 in the action ( [ lifshitzscalarminkowski ] ) with @xmath4 .
we expand the scalar field @xmath8 around a homogeneous v.e.v . @xmath24 as @xmath25 . throughout this paper
we consider the uv regime in which the hubble expansion rate @xmath26 is much higher than mass scales in the scalar field action . in this regime ,
the hubble friction is so strong that the time dependence of the background @xmath24 is unimportant .
for this reason , hereafter , we treat @xmath24 as a constant .
the action for the perturbation @xmath27 is then written as @xmath28,\end{aligned}\ ] ] where @xmath29 @xmath30 and @xmath31 are mass scales and @xmath32 and @xmath33 are dimensionless constants . in the uv ,
the quadratic action for @xmath27 is simply @xmath34 as discussed after ( [ eqn : lifshitz - scaling ] ) , the scaling dimension of @xmath8 and thus @xmath27 is zero , @xmath35 and its power - spectrum should be scale - invariant .
since @xmath27 is scale - invariant and there is only one scale @xmath30 in the uv quadratic action ( [ s2 ] ) , we expect that the power - spectrum should be roughly @xmath36 now let us calculate the power spectrum concretely . by solving the heisenberg equation obtained from the quadratic action ( [ s2 ] )
, operator @xmath27 can be expanded as @xmath37 and @xmath38 where @xmath39 the mode function @xmath40 is chosen so that its asymptotic behavior in the minkowski limit ( @xmath41const . )
is the same as the positive - frequency mode function in minkowski background .
the vacuum state @xmath42 is defined as usual by @xmath43 the mode function @xmath40 approaches a constant value in the @xmath44 limit if the integral @xmath45 converges , where @xmath46 is the time corresponding to @xmath44 .
the power - law expansion @xmath47 with @xmath48 satisfies this condition . under this condition ,
when the physical wavelength @xmath49 becomes as long as the size of the sound horizon @xmath50 , the mode function @xmath40 stops oscillating and freezes out .
( see in fig.([fig : wave.pdf ] ) . )
note that the physical wavelength at sound horizon exit is super - horizon size in the uv , i.e. when @xmath51 . with @xmath52 .
the physical wavelength exits the sound horizon @xmath50 in the uv and re - enters the sound horizon @xmath53 in the ir .
we suppose that the scale - invariant perturbations of the lifshitz scalar are converted into those of radiation by curvaton mechanism or / and modulated decay when physical wavelengths of interest are outside the sound horizon .
thus , strictly speaking , the sound horizon in the uv is for the lifshitz scalar and that in the ir is for radiation.,height=377 ] the commutation relations of operators are defined in the usual manner as @xmath54=i\delta^3(\boldsymbol{x}-\boldsymbol{y } ) , \qquad [ \phi(\boldsymbol{x } ) , \phi(\boldsymbol{y})]= [ \pi(\boldsymbol{x } ) , \pi(\boldsymbol{y } ) ] = 0,\end{aligned}\ ] ] where @xmath55 and hence the operators @xmath56 and @xmath57 satisfy @xmath58= ( 2\pi)^3\delta^3(\boldsymbol{k}-\boldsymbol{k } ' ) , \qquad [ a_{\boldsymbol{k}},a_{\boldsymbol{k}'}]= [ a_{\boldsymbol{k}}^\dagger , a_{\boldsymbol{k}'}^\dagger]=0.\end{aligned}\ ] ] power spectrum @xmath59 is defined as @xmath60 so that @xmath61 substituting eqs.([phi ] ) and ( [ modefunction ] ) into eq.([power ] ) , we obtain @xmath62 in this paper , we suppose that the primordial fluctuations of the lifshitz scalar field @xmath27 are almost linearly transformed to the curvature perturbations and that large non - gaussianity is not created in the transformation itself . in scenarios like the curvaton mechanism @xcite and the modulated decay @xcite , this assumption
can be realized naturally , while violation of the assumption is also possible in these scenarios . with the assumption
, the curvature perturbation @xmath63 is related to the field fluctuation @xmath27 as @xmath64 where @xmath65 is an energy scale .
therefore , the power spectrum of the curvature perturbation takes over the scale - invariance as @xmath66 the cobe normalization @xcite sets @xmath67 , and thus @xmath68
let us move on to discussions about non - gaussianity .
in the previous section we have seen that a free lifshitz scalar with the dynamical critical exponent @xmath4 can produce scale - invariant perturbations .
the essential reason for this is that the scaling dimension of the scalar is zero at high energy .
this implies that at high energy , some nonlinear operators could be as important as leading operators in the quadratic action and that large non - gaussianity could be generated by those nonlinear operators . on the other hand , because of power - counting renormalizability , those nonlinear operators do not completely dominate over the quadratic terms and thus do not spoil the analysis of scaling dimensions and the scale - invariance of the power - spectrum unless the theory gets really strongly coupled . in this section
, we shall present order estimates for the bispectrum of curvature perturbations and the corresponding nonlinear parameter @xmath6 , deferring detailed calculations until the next section . as presented in ( [ phi - to - zeta ] ) , throughout this paper we assume that perturbations of the lifshitz scalar are almost linearly transformed to curvature perturbations . in our calculation of the bispectrum and trispectrum of curvature perturbations , we thus ignore nonlinear terms in the right hand side of ( [ phi - to - zeta ] ) and take into account the linear term only .
in particular , @xmath69 this treatment is justified by the fact that perturbations of the lifshitz scalar have large non - gaussianity . actually , the bispectrum and trispectrum would obtain corrections from nonlinear terms in ( [ phi - to - zeta ] ) , but those corrections would be smaller than large contributions from the lifshitz scalar s intrinsic non - gaussianity . in terms of the nonlinear parameter @xmath6 ,
corrections due to nonlinear terms in ( [ phi - to - zeta ] ) are typically @xmath70 but we shall see below that the scalar field s intrinsic non - gaussianity can lead to @xmath71 .
we shall adopt the so called in - in formalism @xcite to calculate the bispectrum and trispectrum of the lifshitz scalar . the leading contribution to the bispectrum
is given by the following formula ( see the next section for details ) @xmath72 \right\rangle , \label{in - in - bispectrum}\ ] ] where @xmath73 represents cubic terms in the interaction hamiltonian .
dominant terms in @xmath74 are marginal ones , i.e. those terms whose scaling dimensions are zero
. actually , as shown in appendix [ appendix ] , there are three ( and only three ) independent marginal cubic operators in the action in the uv : @xmath75 where @xmath76 are dimensionless parameters .
( the fist term can be forbidden by the shift symmetry if one likes . ) evidently , validity of perturbative expansion ( in the in - in formalism ) requires @xmath76 be smaller than unity .
the corresponding cubic operators in the interaction hamiltonian are @xmath77 each of these dominant cubic terms includes six spatial derivatives and gives zero scaling dimension to @xmath74 . combining this with the fact that the scaling dimension of @xmath27 is zero ,
we conclude that the bispectrum of @xmath27 given by ( [ in - in - bispectrum ] ) should be scale - independent , and thus @xmath78 where @xmath79 stands for the most dominant one among @xmath76 ( @xmath80 ) . roughly speaking , the non - linear parameter @xmath6 is defined so that @xmath81 thus , combining this with ( [ phiphi - to - zetazeta ] ) , ( [ phiphi - roughly ] ) and ( [ phiphiphi - roughly ] ) , we obtain @xmath82 here , we have used the cobe normalization ( [ cobenormalization ] ) . in the next section
, we obtain a more precise expression for @xmath6 in terms of @xmath76 .
as already stated , validity of perturbative expansion requires that the dimensionless parameters @xmath76 be smaller than unity .
we find from the order estimate ( [ eq : estimation ] ) that @xmath6 can be large , e.g. as large as @xmath7 , even if @xmath76 are reasonably small .
we have seen that the lifshitz scalar s non - gaussian intrinsic fluctuations can leave observable non - gaussianities in the sky . in this section ,
we compute the bispectrum and trispectrum of lifshitz scalar fluctuations and discuss their shapes . since a lifshitz scalar with @xmath4 can have all possible self - coupling terms containing six spatial derivatives ( cf .
( [ s3 ] ) , ( [ s4 ] ) ) , the resulting correlation functions can take various configurations in momentum space .
we will see especially that the generated bispectrum includes local , equilateral , and orthogonal shapes .
we also discuss non - gaussianity of the primordial curvature perturbations sourced by the lifshitz scalar s field fluctuations . by comparing with the latest observational constraints , we obtain bounds on the lifshitz scalar s self - coupling strengths .
we make use of the prescription of the in - in formalism ( see e.g. @xcite ) for calculating expectation values of a product @xmath83 of field operators at time @xmath84 , @xmath85 q^i(t ) \left [ t \exp \left(- i \int^t_{t_0 } h_i ( t')dt ' \right ) \right ] \right\rangle , \label{inin}\ ] ] where @xmath86 denotes time - ordering , @xmath87 is anti - time - ordering , @xmath88 is the product @xmath83 in the interaction picture , and @xmath89 is the interaction part of the hamiltonian in the interaction picture .
we take the time @xmath90 to be at very early times when the fluctuation wavelengths are well inside the sound horizon . in the present case , introducing the time @xmath91 , then the integral can be taken in terms of @xmath92 from @xmath93 to some time after the sound horizon exit . )
is shifted to @xmath94 on the right side of @xmath95 , and to @xmath96 on the left side , so that the oscillating exponents become exponentially decreasing .
this prescription corresponds to picking up the vacuum state . ] in order to obtain the bispectrum of the lifshitz scalar fluctuations , we expand ( [ inin ] ) in terms of @xmath97 and compute terms which are first order in the 3-point interaction hamiltonian ( [ h3 ] ) .
we then find @xmath98 \rangle \\ & = -i ( 2 \pi)^3 \delta^{(3 ) } ( \boldsymbol{k_1 } + \boldsymbol{k_2 } + \boldsymbol{k_3 } ) f(\boldsymbol{k_1 } , \boldsymbol{k_2 } ) \\ & \qquad \qquad \times u_{\boldsymbol{k_1}}(t )
u_{\boldsymbol{k_2}}(t ) u_{\boldsymbol{k_3}}(t ) \int^t_{t_0 } \frac{dt'}{m^5 a(t')^3 } u_{-\boldsymbol{k_1}}^*(t ' ) u_{-\boldsymbol{k_2}}^*(t ' ) u_{-\boldsymbol{k_3}}^*(t ' ) + \mathrm{c.c . } \\ & = - \frac{(2\pi)^3 m^3}{2 ^ 2 } \frac { \delta^{(3 ) } ( \boldsymbol{k_1 } + \boldsymbol{k_2 } + \boldsymbol{k_3 } ) f(\boldsymbol{k_1 } , \boldsymbol{k_2 } ) } { ( k_1 k_2 k_3)^3 ( k_1 ^ 3 + k_2 ^ 3 + k_3 ^ 3 ) } , \label{bispectrum}\end{aligned}\ ] ] where we have defined @xmath99 with @xmath100 . in order to study its configuration in momentum space , let us express the bispectrum ( [ bispectrum ] ) as @xmath101 and plot @xmath102 as a function of @xmath103 and @xmath104 .
contributions from the @xmath105 , @xmath106 , and @xmath107 terms in ( [ fij ] ) are plotted respectively in figures [ fig : a1 ] , [ fig : a2 ] , and [ fig : a3 ] .
we have assumed @xmath108 to avoid showing the same configuration twice , and @xmath109 further follows from the triangular inequality .
contribution to @xmath110 , where @xmath111 . ]
contribution to @xmath110 , where @xmath111 . ]
contribution to @xmath110 , where @xmath112 . ] here , one should note that only the contribution from the @xmath105 term blows up in the `` squeezed '' triangle limit , i.e. @xmath113 , giving a shape similar to that of the local form @xcite @xmath114 this can be understood as follows : among the three interaction terms in ( [ s3 ] ) , only the @xmath105 term breaks the shift symmetry @xmath115 such a shift symmetry makes the theory indifferent to infinitely large scale fluctuations , and suppresses interactions with long wavelength modes .
this is why @xmath110 from the @xmath106 and @xmath107 contributions disappear in the squeezed limit .
moreover , the shapes from the @xmath106 and @xmath107 terms respectively look similar to that of the orthogonal @xcite and equilateral @xcite forms which are given by @xmath116 @xmath117 the permutations act only on the last terms in the parentheses . to give a bit more quantitative discussion on shapes , let us express ( [ bispectrum ] ) in terms of the above forms . following @xcite
, we introduce a scalar product between two distributions @xmath118 and @xmath119 , @xmath120 where summation is taken over all @xmath121 s which form a triangle , and @xmath122 denotes the power spectrum
. then one can `` expand '' @xmath123 of ( [ f ] ) in terms of ( [ local ] ) , ( [ orthog ] ) , ( [ equil ] ) and obtain a template function : @xmath124 where @xmath125 @xmath106 and @xmath107 are absent in ( [ tildefnllocal ] ) since the denominator @xmath126 blows up . which roughly corresponds to the current observable limit in the cmb .
] here , since ( [ local ] ) , ( [ orthog ] ) , and ( [ equil ] ) do _ not _ form a complete basis set , one should consider ( [ template ] ) as an indicator roughly telling which combination of the 3-point interaction terms in ( [ s3 ] ) gives which bispectrum shape . using the above template function , one can estimate the non - gaussianity parameter @xmath6 of the primordial curvature perturbations . as discussed around ( [ phiphi - to - zetazeta ] ) , we assume that the lifshitz scalar fluctuations are linearly converted to the curvature perturbations , i.e. @xmath127 with some mass scale @xmath65 . for the bispectrum of bardeen s curvature perturbations @xmath128 ( which is related to the primordial curvature perturbations by @xmath129 in the matter - dominated era ) @xmath130
we define its non - gaussianity parameter @xmath6 as @xmath131 here , @xmath132 is the amplitude of the power spectrum @xmath133 which in the present case is @xmath134 the value on the far right hand side is set by the cobe normalization ( [ cobenormalization ] ) . then , by substituting the template function ( [ template ] ) into the lifshitz scalar fluctuations ( [ f ] ) , we obtain @xmath135 where @xmath136 local , orthog . ,
we can now use this relation to set constraints on the self - coupling strengths , by computing the necessary conditions for ( [ tildefnllocal ] ) , ( [ tildefnlorthog ] ) , and ( [ tildefnlequil ] ) to satisfy the wmap 7yr @xcite constraints on the non - gaussianity parameters : @xmath137 , @xmath138 , and @xmath139 ( each at 95% cl ) , respectively .
thus we arrive at @xmath140 we expect that more rigorous treatments ( such as using observational data to constrain combinations of different shapes , instead of individual ones ) would not change the results significantly . ) are conservative bounds , since they are necessary conditions for satisfying observational constraints .
one way to obtain more strict bounds is by imposing observational constraints under an _ ad hoc _ assumption that two out of the three @xmath76 s are zero .
however , such procedure does not change the results significantly , except for the strong upper limit on @xmath141 tightening the upper bound on @xmath106 and lower bound on @xmath107 roughly by an order of magnitude . ] the bounds in ( [ alphabound ] ) show that observational constraints require @xmath142 .
note that such smallness of the coupling strengths is required anyway for validity of the perturbative expansions in terms of @xmath97 , which we have carried out in computing the bispectrum .
in other words , as @xmath106 or @xmath107 become of order unity and our procedure for computing the bispectrum breaks down , the curvature perturbations saturate the current observational limit for the orthogonal / equilateral form bispectra .
the most stringent bounds are obtained for @xmath105 , i.e. @xmath143 where the required level of tuning depends on its sign .
this is mainly due to the rather tight constraints on the local form bispectrum it produces . in order to suppress @xmath105 ,
some sort of symmetry may be required in the theory .
for example , as we have stated below ( [ local ] ) , a shift symmetry for @xmath27 forbids such self - coupling terms producing local - type bispectra .
here we have focused on non - gaussianity sourced by the intrinsic fluctuations of the lifshitz scalar , but we should remark that further non - gaussianities in the resulting curvature perturbations can be generated through non - linear conversion processes . the trispectrum can be obtained in a similar manner to the previous subsection .
we compute contributions from the scalar - exchange diagram ( figure [ fig : se ] ) , and from the contact - interaction diagram ( figure [ fig : ci ] ) .
the 4-point interaction terms of the lifshitz scalar action are given in appendix [ appendix ] as @xmath144 where @xmath145 are dimensionless parameters .
therefore the 4-point interaction hamiltonian becomes @xmath146 even if the 3-point interactions are suppressed , i.e. @xmath147 , the 4-point interaction terms can produce a large trispectrum through the contact - interaction diagram .
the contribution from the scalar - exchange diagram are obtained as follows ( note that we only compute contributions from connected diagrams . ) , @xmath148 \right ] \rangle \right ) \right|_{\mathrm{connected } } \\ & = \frac{(2 \pi ) ^3 m^4}{2 ^ 3 } \frac{\delta^{(3 ) } ( \boldsymbol{k_1 } + \boldsymbol{k_2 } + \boldsymbol{k_3 } + \boldsymbol{k_4 } ) ( k_1 ^ 3 + k_2 ^ 3 + k_3 ^ 3 + k_4 ^ 3 + k_{12}^3 ) f(\boldsymbol{k_1 } , \boldsymbol{k_2 } ) f(\boldsymbol{k_3 } , \boldsymbol{k_4 } ) } { ( k_1 k_2 k_3 k_4 k_{12})^3 ( k_1 ^ 3 + k_2 ^ 3 + k_{12}^3 ) ( k_3 ^ 3 + k_4 ^ 3 + k_{12}^3 ) ( k_1 ^ 3 + k_2
^ 3 + k_3 ^ 3 + k_4 ^ 3 ) } \\ & \qquad + \biggl ( \mathrm{2\ , \ , terms\ , \
, with\ , \ , } ( 1,2)(3,4 ) \longrightarrow ( 1,3)(2,4),\ , ( 1,4)(2,3 ) \biggr ) , \label{tri - se}\end{aligned}\ ] ] where @xmath149 is defined in ( [ fij ] ) .
one can also compute the contribution from the contact - interaction diagram , @xmath150 \rangle \right ) \right|_{\mathrm{connected } } \\ & = - \frac{(2\pi)^3 m^4}{2 ^ 3 } \frac { \delta^{(3 ) } ( \boldsymbol{k_1 } + \boldsymbol{k_2 } + \boldsymbol{k_3 } + \boldsymbol{k_4 } ) r(\boldsymbol{k_1 } , \boldsymbol{k_2 } , \boldsymbol{k_3 } , \boldsymbol{k_4 } ) } { ( k_1 k_2 k_3 k_4)^3 ( k_1 ^ 3 + k_2 ^ 3 + k_3 ^ 3 + k_4 ^ 3 ) } , \label{tri - ci}\end{aligned}\ ] ] where we have defined @xmath151 expressing the trispectra as @xmath152 let us plot @xmath153 in the `` equilateral '' limit : @xmath154 .
in this limit , the shape of the tetrahedron formed by @xmath121 s depends on two independent variables , e.g. , @xmath155 and @xmath156 .
contributions to @xmath157 ( [ tri - se ] ) from the @xmath158 , @xmath159 , @xmath160 , @xmath161 , @xmath162 , and @xmath163 terms are plotted respectively in figures [ fig : a1a1 ] , [ fig : a2a2 ] , [ fig : a3a3 ] , [ fig : a1a2 ] , [ fig : a1a3 ] , and [ fig : a2a3 ] .
contributions to @xmath164 ( [ tri - ci ] ) from the @xmath165 , @xmath166 , @xmath167 terms are plotted in figures [ fig : b1 ] , [ fig : b5 ] , and [ fig : b6 ] .
we have omitted contributions from the @xmath168 , @xmath169 , @xmath170 terms since in the equilateral limit they become flat and are equivalent to that from @xmath165 up to overall constant factors . the condition @xmath171 is required for @xmath121 s to close , and we further assume @xmath172 in order to avoid showing the same configuration twice .
contribution to @xmath173 , where @xmath174 . ]
contribution to @xmath173 , where @xmath174 . ]
contribution to @xmath175 , where @xmath176 . ]
contribution to @xmath175 , where @xmath176 . ]
contribution to @xmath175 , where @xmath177 . ]
contribution to @xmath175 , where @xmath177 . ]
contribution to @xmath178 , where @xmath179 . ]
contribution to @xmath178 , where @xmath179 . ]
contribution to @xmath178 , where @xmath180 . ]
in this paper , we have studied non - gaussianity in the intrinsic fluctuations of a lifshitz scalar which follows an anisotropic scaling with @xmath181 .
our work is based on @xcite , which pointed out that its special dispersion relation in the uv can lead to generation of super - horizon field perturbations .
since the scaling dimension of a lifshitz scalar with @xmath4 is zero , the resulting field perturbations become scale - invariant whether or not the scalar s self - couplings are small .
this leads to our main point that curvature perturbations generated from such field fluctuations necessarily leave large non - gaussianity in the sky , unless the field s self - couplings are forbidden by some symmetry , or the field exhibits some sort of asymptotic freedom .
this is to be contrasted with perturbations generated through cosmic inflation , where largely non - gaussian intrinsic fluctuations are in most cases incompatible with scale - invariance .
the lifshitz scalar s self - coupling terms containing spatial derivatives produce non - gaussianities with various configurations in momentum space .
in particular , the bispectrum of the field fluctuations includes shapes which are similar to that of the local , equilateral , and orthogonal forms .
( however , we emphasize that the local , equilateral , and orthogonal shapes do not form a complete basis set for the bispectrum obtained in this paper .
we also note that the results of the effective field theory approach in @xcite do not apply to our case , where lorentz symmetry is explicitly broken , and non - gaussianity is sourced by marginal terms in the action . ) upon computing the correlation functions , we have carried out expansions in terms of the interaction hamiltonian . within the domain of applicability of such perturbative expansion ,
i.e. the self - couplings less than unity , we have seen that the lifshitz scalar s field fluctuations can lead to significant non - gaussianity in the primordial curvature perturbations . in particular ,
when curvature perturbations are sourced linearly from the field fluctuations as in ( [ linconv ] ) , their bispectrum saturates the current observational limit for the orthogonal and equilateral forms , as the self - couplings @xmath106 and @xmath107 in ( [ s3 ] ) approach unity .
since naively there is no reason for such self - couplings to be suppressed , we can expect large non - gaussianity to be produced from lifshitz scalar fluctuations , which may be detected by upcoming cmb observations . on the other hand , for the local - type bispectrum , observational constraints require @xmath105 to be as small as @xmath182 ( the level of tuning depends on @xmath105 s sign ) .
however , as we have remarked , such self - couplings sourcing local - type non - gaussianity can be forbidden by a shift symmetry .
the field fluctuations generated in the mechanism of @xcite obtain a scale - invariant spectrum .
however , when one takes into account the renormalization - group flow of the parameters of the theory ( e.g. @xmath30 in ( [ s2 ] ) ) , the spectrum may become tilted .
a time - dependent background value @xmath183 may also give rise to similar effects .
how strong the tilt becomes , as well as the scale - dependence of the non - gaussianity , remains to be understood . while in this paper we have studied fluctuations of scalar fields , the scalar graviton which can show up in hoava - lifshitz gravity may also obtain fluctuations in a similar manner .
it would be interesting to investigate the possibility that such scalar graviton generates the primordial curvature perturbations .
@xcite works in this direction .
see @xcite for some issues related to the scalar graviton , including non - perturbative continuity of the limit in which general relativity is supposed to be recovered .
see also @xcite for a recent attempt to eliminate the scalar graviton from the theory . ) furthermore , when considering cosmic inflation in hoava - lifshitz gravity , due to the field fluctuations freezing - out at the time of sound horizon @xmath184 exit , the well - known relations in slow - roll inflation between various cosmological observables and the slow - roll parameters are expected to be modified .
aspects of cosmic inflation in hoava - lifshitz gravity are also worthy of study in details .
we would like to thank eiichiro komatsu and takahiro nishimichi for helpful conversations .
is also grateful to masahiro kawasaki for his continuous support .
the work of t.k . was supported by grant - in - aid for jsps fellows
21@xmath1858966 . the work of s.m .
was supported in part by grant - in - aid for young scientists ( b ) no .
17740134 , grant - in - aid for creative scientific research no .
19gs0219 , grant - in - aid for scientific research on innovative areas no . 21111006 , grant - in - aid for scientific research ( c ) no .
21540278 , and the mitsubishi foundation .
this work was supported by world premier international research center initiative ( wpi initiative ) , mext , japan .
in this appendix we construct nonlinear , marginal terms in the action in the uv , specializing to the @xmath4 case .
as mentioned after ( [ actionphi ] ) , those nonlinear terms do not include time derivatives but can include spatial derivatives .
we demand that each term in the action can include only up to six spatial derivatives .
this treatment is justified since with @xmath4 , terms with more than six spatial derivatives would be non - renormalizable and thus would not be generated by quantum corrections . in this case the most important terms in the uv are marginal ones , i.e. those with six spatial derivatives . in the following we write down
all independent combinations of cubic and quartic terms which are marginal in the uv . in general we can write down fourteen cubic terms with six spatial derivatives as follows ,
@xmath186 some linear combinations of these terms turn out to be total derivatives .
thus , not all of these fourteen terms lead to independent terms in the action , provided that surface terms are dropped . apparently , there are thirteen total derivatives with six spatial derivatives as follows , @xmath187=a_1 + 2a_2 , & \partial_i \left[\left(\delta^2 \phi\right ) \left ( \partial_i\phi\right ) \phi \right]=a_2+a_3+a_7 , \\ \partial_i \left[\left(\delta\partial_i \partial_j \phi\right ) \left(\partial_j \phi \right ) \phi \right]=a_2+a_4+a_8 , & \partial_i \left [ \left(\delta\partial_i \phi\right ) \left(\delta \phi\right ) \phi \right]=a_3+a_5+a_9 ,
\\ \partial_i \left[\left(\delta\partial_j \phi\right ) \left(\partial_i\partial_j \phi\right ) \phi \right]=a_4+a_5+a_{10 } , & \partial_i \left[\left(\partial_i\partial_j\partial_k \phi\right ) \left(\partial_j \partial_k\phi\right ) \phi \right]=a_4+a_6+a_{11 } , \\ \partial_i
\left[\left(\delta \partial_i \phi\right ) \left ( \partial_j\phi\right)\left(\partial_j \phi\right ) \right]=a_7 + 2a_{10 } , & \partial_i \left[\left(\delta\partial_j \phi\right ) \left ( \partial_i\phi \right)\left(\partial_j\phi\right ) \right]=a_8+a_{9}+a_{10 } , \\
\partial_i \left[\left(\partial_i\partial_j\partial_k \phi\right ) \left ( \partial_j \phi\right)\left(\partial_k \phi\right ) \right]=a_8 + 2a_{11 } , & \partial_i \left [ \left(\delta\phi\right)\left(\delta \phi\right ) \left(\partial_i \phi\right ) \right]=2a_9+a_{12 } , \\
\partial_i \left[\left(\delta \phi\right ) \left(\partial_i \partial_j \phi\right)\left(\partial_j \phi\right ) \right ] = a_9+a_{10}+a_{13},\qquad & \partial_i \left [ \left(\partial_i\partial_j\phi\right ) \left ( \partial_j \partial_k \phi\right ) \left(\partial_k
\phi\right ) \right ] = a_{10}+a_{11}+a_{14 } , \\
\partial_i \left[\left(\partial_j\partial_k \phi\right ) \left(\partial_j\partial_k \phi\right)\left(\partial_i \phi\right ) \right ] = 2a_{11}+a_{13}. \end{array } \label{totalderivative}\end{aligned}\ ] ] actually , two of the combinations in ( [ totalderivative ] ) are not independent and ( [ totalderivative ] ) include only eleven independent combinations . as a result
, we can represent arbitrary marginal cubic terms in the action as linear combinations of three among fourteen in ( [ a ] ) . concretely , in this paper we shall use the following three terms . @xmath188
an advantage of this representation is that contributions from these three terms roughly correspond to local , equilateral and orthogonal shape of bispectrum , respectively .
if we impose the shift symmetry for @xmath27 then the number of independent interaction terms is reduced . with the shift symmetry
, we have to consider only the last eight terms in ( [ a ] ) and only the last seven combinations in ( [ totalderivative ] ) .
one of the last seven combinations in ( [ totalderivative ] ) is not independent . as a result
, the marginal cubic terms in the action with the shift symmetry can be written as ( [ appendixs3 ] ) with @xmath189 .
similarly , we can write down twenty quartic terms with six spatial derivatives as follows , @xmath190 apparently , there are sixteen total derivatives as follows , @xmath191=b_{1}+3b_{2 } , & \partial_i \left [ \left(\delta^2 \phi\right)\left(\partial_i\phi\right ) \phi^2
\right]=b_{2}+b_{3}+2b_{5 } , \\ \partial_i
\left [ \left(\delta\partial_i\partial_j \phi\right)\left(\partial_j\phi\right ) \phi^2 \right]=b_{2}+b_{4}+2b_{6 } , & \partial_i \left [ \left(\delta\partial_i \phi\right)\left(\delta\phi\right ) \phi^2 \right]=b_{3}+b_{7}+2b_{9 } , \\ \partial_i \left [ \left(\delta\partial_j \phi\right)\left(\partial_i\partial_j\phi\right ) \phi^2 \right]=b_{4}+b_{7}+2b_{10 } , & \partial_i \left [ \left(\partial_i\partial_j\partial_k \phi\right ) \left(\partial_j\partial_k\phi\right)\phi^2 \right]=b_{4}+b_{8}+2b_{11 } , \\
\partial_i \left [ \left(\delta\partial_i \phi\right)\left(\partial_j\phi\right ) \left(\partial_j \phi\right ) \phi \right]=b_{5}+2b_{10}+b_{15 } , & \partial_i \left [ \left(\delta\partial_j \phi\right)\left(\partial_i\phi\right ) \left(\partial_j \phi\right ) \phi \right]=b_{6}+b_{9}+b_{10}+b_{15 } , \\ \partial_i \left [ \left(\partial_i\partial_j\partial_k \phi\right ) \left(\partial_j\phi\right)\left(\partial_k \phi\right ) \phi \right]=b_{6}+2b_{11}+b_{16 } , & \partial_i \left [ \left(\delta \phi\right)\left(\delta\phi\right)\left(\partial_i \phi\right ) \phi \right]=2b_{9}+b_{12}+b_{17 } ,
\\ \partial_i \left [ \left(\delta \phi\right)\left(\partial_i \partial_j\phi\right ) \left ( \partial_j\phi\right ) \phi
\right]=b_{9}+b_{10}+b_{13}+b_{18 } , & \partial_i \left [ \left(\partial_i\partial_j \phi\right ) \left(\partial_j \partial_k\phi\right)\left(\partial_k
\phi\right ) \phi \right]=b_{10}+b_{11}+b_{14}+b_{20 } , \\ \partial_i
\left [ \left(\partial_j\partial_k \phi\right ) \left(\partial_j\partial_k\phi\right)\left(\partial_i \phi\right ) \phi \right]=2b_{11}+b_{13}+b_{19 } , & \partial_i \left [ \left(\delta \phi\right)\left(\partial_i\phi\right ) \left(\partial_j
\phi\right)\left(\partial_j \phi\right ) \right]=b_{15}+b_{17}+2b_{18 } , \\ \partial_i
\left [ \left(\partial_j\partial_k \phi\right)\left(\partial_i\phi\right ) \left(\partial_j \phi\right)\left(\partial_k \phi\right ) \right]=b_{16}+b_{18}+2b_{20 } , & \partial_i \left [ \left(\partial_i\partial_j \phi\right)\left(\partial_j\phi\right ) \left(\partial_k
\phi\right)\left(\partial_k \phi\right ) \right]=b_{15}+b_{19}+2b_{20}. \\
\end{array } \label{const.b}\end{aligned}\ ] ] two of them are not independent . thus , marginal quartic terms in the action can be written as a linear combination of six independent terms . in this paper , we use the following terms , @xmath192 with the shift symmetry , we have to consider only the last six terms in ( [ b ] ) and only the last three combinations in ( [ const.b ] ) .
therefore , the number of the independent terms is three . in the notation ( [ appendixs4 ] ) , the shift symmetry imposes the constraints as @xmath193 and then the action can be transformed into the following form , @xmath194 this manifestly has the shift symmetry .
p. horava , phys .
d * 79 * , 084008 ( 2009 ) [ arxiv:0901.3775 [ hep - th ] ] .
p. horava , phys .
lett . * 102 * , 161301 ( 2009 ) [ arxiv:0902.3657 [ hep - th ] ] .
s. mukohyama , jcap * 0906 * , 001 ( 2009 ) [ arxiv:0904.2190 [ hep - th ] ] .
g. calcagni , jhep * 0909 * , 112 ( 2009 ) [ arxiv:0904.0829 [ hep - th ] ] .
r. brandenberger , phys .
d * 80 * , 043516 ( 2009 ) [ arxiv:0904.2835 [ hep - th ] ] .
e. kiritsis and g. kofinas , nucl .
b * 821 * , 467 ( 2009 ) [ arxiv:0904.1334 [ hep - th ] ] .
s. maeda , s. mukohyama and t. shiromizu , phys .
d * 80 * , 123538 ( 2009 ) [ arxiv:0909.2149 [ astro-ph.co ] ] .
s. mukohyama , k. nakayama , f. takahashi and s. yokoyama , phys .
b * 679 * , 6 ( 2009 ) [ arxiv:0905.0055 [ hep - th ] ] .
t. takahashi and j. soda , phys .
* 102 * , 231301 ( 2009 ) [ arxiv:0904.0554 [ hep - th ] ] .
s. mukohyama , phys .
d * 80 * , 064005 ( 2009 ) [ arxiv:0905.3563 [ hep - th ] ] .
s. mukohyama , jcap * 0909 * , 005 ( 2009 ) [ arxiv:0906.5069 [ hep - th ] ] .
m. henneaux , a. kleinschmidt and g. l. gomez , phys .
d * 81 * , 064002 ( 2010 ) [ arxiv:0912.0399 [ hep - th ] ] . c. charmousis , g. niz , a. padilla and p. m. saffin , jhep * 0908 * , 070 ( 2009 ) [ arxiv:0905.2579 [ hep - th ] ] .
a. wang and r. maartens , phys . rev .
d * 81 * , 024009 ( 2010 ) [ arxiv:0907.1748 [ hep - th ] ] .
d. blas , o. pujolas and s. sibiryakov , phys .
. lett . * 104 * , 181302 ( 2010 ) [ arxiv:0909.3525 [ hep - th ] ] .
k. koyama and f. arroja , jhep * 1003 * , 061 ( 2010 ) [ arxiv:0910.1998 [ hep - th ] ] .
k. izumi and s. mukohyama , phys .
d * 81 * , 044008 ( 2010 ) [ arxiv:0911.1814 [ hep - th ] ] . c. doran , phys .
d * 61 * , 067503 ( 2000 ) [ arxiv : gr - qc/9910099 ] .
k. i. maeda , y. misonoh and t. kobayashi , arxiv:1006.2739 [ hep - th ] .
d. h. lyth and d. wands , phys .
b * 524 * , 5 ( 2002 ) [ arxiv : hep - ph/0110002 ] .
k. enqvist and m. s. sloth , nucl .
b * 626 * , 395 ( 2002 ) [ arxiv : hep - ph/0109214 ] .
t. moroi and t. takahashi , phys .
b * 522 * , 215 ( 2001 ) [ erratum - ibid .
b * 539 * , 303 ( 2002 ) ] [ arxiv : hep - ph/0110096 ] .
g. dvali , a. gruzinov and m. zaldarriaga , phys .
d * 69 * , 023505 ( 2004 ) [ arxiv : astro - ph/0303591 ] .
l. kofman , arxiv : astro - ph/0303614 .
a. d. linde and v. f. mukhanov , phys .
d * 56 * , 535 ( 1997 ) [ arxiv : astro - ph/9610219 ] .
n. bartolo , s. matarrese and a. riotto , phys .
d * 65 * , 103505 ( 2002 ) [ arxiv : hep - ph/0112261 ] . f. bernardeau and j. p.
uzan , phys .
d * 66 * , 103506 ( 2002 ) [ arxiv : hep - ph/0207295 ] .
f. bernardeau and j. p. uzan , phys .
d * 67 * , 121301 ( 2003 ) [ arxiv : astro - ph/0209330 ] .
g. dvali , a. gruzinov and m. zaldarriaga , phys .
d * 69 * , 023505 ( 2004 ) [ arxiv : astro - ph/0303591 ] .
n. bartolo , e. komatsu , s. matarrese and a. riotto , phys .
rept . * 402 * , 103 ( 2004 ) [ arxiv : astro - ph/0406398 ] .
d. seery and j. e. lidsey , jcap * 0506 * , 003 ( 2005 ) [ arxiv : astro - ph/0503692 ] .
d. seery and j. e. lidsey , jcap * 0509 * , 011 ( 2005 ) [ arxiv : astro - ph/0506056 ] .
m. sasaki , j. valiviita and d. wands , phys .
d * 74 * , 103003 ( 2006 ) [ arxiv : astro - ph/0607627 ] .
x. chen , m. x. huang , s. kachru and g. shiu , jcap * 0701 * , 002 ( 2007 ) [ arxiv : hep - th/0605045 ] .
t. suyama and m. yamaguchi , phys .
d * 77 * , 023505 ( 2008 ) [ arxiv:0709.2545 [ astro - ph ] ] .
f. arroja and k. koyama , phys .
d * 77 * , 083517 ( 2008 ) [ arxiv:0802.1167 [ hep - th ] ] .
f. arroja , s. mizuno and k. koyama , jcap * 0808 * , 015 ( 2008 ) [ arxiv:0806.0619 [ astro - ph ] ] .
m. sasaki , prog .
phys . * 120 * , 159 ( 2008 ) [ arxiv:0805.0974 [ astro - ph ] ] .
a. naruko and m. sasaki , prog .
phys . * 121 * , 193 ( 2009 ) [ arxiv:0807.0180 [ astro - ph ] ] .
t. suyama and f. takahashi , jcap * 0809 * , 007 ( 2008 ) [ arxiv:0804.0425 [ astro - ph ] ] .
x. gao , jcap * 0806 * , 029 ( 2008 ) [ arxiv:0804.1055 [ astro - ph ] ] .
h. r. s. cogollo , y. rodriguez and c. a. valenzuela - toledo , jcap * 0808 * , 029 ( 2008 ) [ arxiv:0806.1546 [ astro - ph ] ] . k. ichikawa , t. suyama , t. takahashi and m. yamaguchi , phys . rev .
d * 78 * , 023513 ( 2008 ) [ arxiv:0802.4138 [ astro - ph ] ] .
c. t. byrnes , jcap * 0901 * , 011 ( 2009 ) [ arxiv:0810.3913 [ astro - ph ] ] .
d. langlois , f. vernizzi and d. wands , jcap * 0812 * , 004 ( 2008 ) [ arxiv:0809.4646 [ astro - ph ] ] . c. hikage , k. koyama , t. matsubara , t. takahashi and m. yamaguchi , mon . not .
soc . * 398 * , 2188 ( 2009 ) [ arxiv:0812.3500 [ astro - ph ] ] .
m. kawasaki , k. nakayama , t. sekiguchi , t. suyama and f. takahashi , jcap * 0811 * , 019 ( 2008 ) [ arxiv:0808.0009 [ astro - ph ] ] .
q. g. huang , jcap * 0811 * , 005 ( 2008 ) [ arxiv:0808.1793 [ hep - th ] ] .
q. g. huang , jcap * 0905 * , 005 ( 2009 ) [ arxiv:0903.1542 [ hep - th ] ] .
q. g. huang , jcap * 0906 * , 035 ( 2009 ) [ arxiv:0904.2649 [ hep - th ] ] .
c. t. byrnes and g. tasinato , jcap * 0908 * , 016 ( 2009 ) [ arxiv:0906.0767 [ astro-ph.co ] ] . c. cheung , a. l. fitzpatrick , j. kaplan and l. senatore , jcap * 0802 * , 021 ( 2008 ) [ arxiv:0709.0295 [ hep - th ] ] .
x. chen , m. x. huang , s. kachru and g. shiu , jcap * 0701 * , 002 ( 2007 ) [ arxiv : hep - th/0605045 ] .
m. alishahiha , e. silverstein and d. tong , phys .
d * 70 * , 123505 ( 2004 ) [ arxiv : hep - th/0404084 ] .
d. langlois , s. renaux - petel , d. a. steer and t. tanaka , phys . rev .
* 101 * , 061301 ( 2008 ) [ arxiv:0804.3139 [ hep - th ] ] .
d. langlois , s. renaux - petel , d. a. steer and t. tanaka , phys .
d * 78 * , 063523 ( 2008 ) [ arxiv:0806.0336 [ hep - th ] ] .
s. li , y. f. cai and y. s. piao , phys .
b * 671 * , 423 ( 2009 ) [ arxiv:0806.2363 [ hep - ph ] ] .
x. gao and b. hu , jcap * 0908 * , 012 ( 2009 ) [ arxiv:0903.1920 [ astro-ph.co ] ] .
y. f. cai and h. y. xia , phys .
b * 677 * , 226 ( 2009 ) [ arxiv:0904.0062 [ hep - th ] ] .
d. langlois , s. renaux - petel and d. a. steer , jcap * 0904 * , 021 ( 2009 ) [ arxiv:0902.2941 [ hep - th ] ] .
j. khoury and f. piazza , jcap * 0907 * , 026 ( 2009 ) [ arxiv:0811.3633 [ hep - th ] ] .
s. mizuno , f. arroja , k. koyama and t. tanaka , phys .
d * 80 * , 023530 ( 2009 ) [ arxiv:0905.4557 [ hep - th ] ] .
m. x. huang , g. shiu and b. underwood , phys .
d * 77 * , 023511 ( 2008 ) [ arxiv:0709.3299 [ hep - th ] ] .
d. seery and j. e. lidsey , jcap * 0701 * , 008 ( 2007 ) [ arxiv : astro - ph/0611034 ] .
d. seery , m. s. sloth and f. vernizzi , jcap * 0903 * , 018 ( 2009 ) [ arxiv:0811.3934 [ astro - ph ] ] . y. rodriguez and c. a. valenzuela - toledo , phys .
d * 81 * , 023531 ( 2010 ) [ arxiv:0811.4092 [ astro - ph ] ] .
x. chen , b. hu , m. x. huang , g. shiu and y. wang , jcap * 0908 * , 008 ( 2009 ) [ arxiv:0905.3494 [ astro-ph.co ] ] .
x. gao , m. li and c. lin , jcap * 0911 * , 007 ( 2009 ) [ arxiv:0906.1345 [ astro-ph.co ] ] .
x. chen , b. hu , m. x. huang , g. shiu and y. wang , jcap * 0908 * , 008 ( 2009 ) [ arxiv:0905.3494 [ astro-ph.co ] ] .
x. chen , m. x. huang and g. shiu , phys .
d * 74 * , 121301 ( 2006 ) [ arxiv : hep - th/0610235 ] .
f. arroja , s. mizuno , k. koyama and t. tanaka , phys .
d * 80 * , 043527 ( 2009 ) [ arxiv:0905.3641 [ hep - th ] ] .
s. mizuno , f. arroja and k. koyama , phys .
d * 80 * , 083517 ( 2009 ) [ arxiv:0907.2439 [ hep - th ] ] .
s. renaux - petel , jcap
* 0910 * , 012 ( 2009 ) [ arxiv:0907.2476 [ hep - th ] ] . d. m. regan , e. p. s. shellard and j. r. fergusson , phys .
d * 82 * , 023520 ( 2010 ) [ arxiv:1004.2915 [ astro-ph.co ] ] .
n. bartolo , m. fasiello , s. matarrese and a. riotto , arxiv:1006.5411 [ astro-ph.co ] . s. mizuno and k. koyama , arxiv:1007.1462 [ hep - th ] . s. weinberg , phys .
d * 72 * , 043514 ( 2005 ) [ arxiv : hep - th/0506236 ] .
e. komatsu and d. n. spergel , phys .
d * 63 * , 063002 ( 2001 ) [ arxiv : astro - ph/0005036 ] .
l. senatore , k. m. smith and m. zaldarriaga , jcap * 1001 * , 028 ( 2010 ) [ arxiv:0905.3746 [ astro-ph.co ] ] .
p. creminelli , a. nicolis , l. senatore , m. tegmark and m. zaldarriaga , jcap * 0605 * , 004 ( 2006 ) [ arxiv : astro - ph/0509029 ] .
d. babich , p. creminelli and m. zaldarriaga , jcap * 0408 * , 009 ( 2004 ) [ arxiv : astro - ph/0405356 ] . c. cheung , p. creminelli , a. l. fitzpatrick , j. kaplan and l. senatore , jhep * 0803 * , 014 ( 2008 ) [ arxiv:0709.0293 [ hep - th ] ] .
b. chen , s. pi and j. z. tang , jcap * 0908 * , 007 ( 2009 ) [ arxiv:0905.2300 [ hep - th ] ] .
s. mukohyama , arxiv:1007.5199 [ hep - th ] .
horava and c. m. melby - thompson , arxiv:1007.2410 [ hep - th ] .
x. gao , arxiv:0904.4187 [ hep - th ] . | a lifshitz scalar with the dynamical critical exponent @xmath0 obtains scale - invariant , super - horizon field fluctuations without the need of an inflationary era .
since this mechanism is due to the special scaling of the lifshitz scalar and persists in the presence of unsuppressed self - couplings , the resulting fluctuation spectrum can deviate from a gaussian distribution .
we study the non - gaussian nature of the lifshitz scalar s intrinsic field fluctuations , and show that primordial curvature perturbations sourced from such field fluctuations can have large non - gaussianity of order @xmath1 , which will be detected by upcoming cmb observations .
we compute the bispectrum and trispectrum of the fluctuations , and discuss their configurations in momentum space . in particular , the bispectrum is found to take various shapes , including the local , equilateral , and orthogonal shapes .
intriguingly , all integrals in the in - in formalism can be performed analytically .
= 1 |
the theoretical description of adsorption is a long - standing complex problem in surface science that presently does not have a general solution @xcite . in 1918
langmuir derived the monolayer adsorption isotherm kinetically for gas monoatomic molecules adsorbed on the homogeneous surface of adsorbents without attractions among the adsorbed molecules @xcite .
later , some theories have been proposed to describe multilayer adsorption in equilibrium @xcite . among them , the one of brunauer - emmett - teller ( bet ) @xcite and the one of frenkel - halsey - hill @xcite are the simplest which provide the basis to construct more elaborate approaches .
those more elaborate analytic approaches take into account lateral interactions between the admolecules , differences between the energy of the first and upper layers , surface energetic heterogeneity and so forth .
these leading models have played an important role in the characterization of solid surfaces by means of gas adsorption .
a more accurate description of multilayer adsorption should account for the fact that , in practice , most adsorbates are polyatomic .
even the simplest nonspherical molecules such as @xmath0 and @xmath1 may adsorb on more than one site depending on the surface structure @xcite .
this effect , called multisite - occupancy adsorption , introduces a new complexity to the adsorption theory .
the difficulty in the analysis of the multisite statistics is mainly associated with three factors which differentiate the @xmath2-mers statistics from the usual single - particle statistics .
namely , @xmath3 no statistical equivalence exists between particles and vacancies ; @xmath4 the occupation of a given lattice site ensures that at least one of its nearest - neighbor sites is also occupied ; and @xmath5 an isolated vacancy can not serve to determine whether that site can ever become occupied .
however , several attempts were done in the past in order to solve the @xmath2-mers problem . among them , onsager @xcite , zimm @xcite and isihara @xcite made important contributions to the understanding of the statistics of rigid rods in dilute solution .
the fh theory , due independently to flory @xcite and to huggins @xcite , has overcome the restriction to dilute solution by means of a lattice calculation .
the approach is a direct generalization of the theory of binary liquids in two dimensions or polymer molecules diluted in a monomeric solvent .
it is worth mentioning that , in the framework of the lattice - gas approach , the adsorption of @xmath2-mers on homogeneous surfaces is an isomorphous problem to the binary solutions of polymer - monomeric solvent .
more recently , two new theories to describe adsorption with multisite occupancy have been introduced . in the first , ramirez - pastor et al .
@xcite presented a model to study the adsorption of linear adsorbates on homogeneous surfaces .
the model is based on exact forms of the thermodynamic functions of linear adsorbates in one dimension and its generalization to higher dimensions . in the second , which is called the fractional statistical theory of the adsorption of polyatomics ( fsta ) , the configuration of the molecule in the adsorbed state is incorporated as a model parameter @xcite . the treatments in refs .
@xcite are limited in their application because they are valid only for monolayer adsorption .
there are few studies that take into account multisite occupancy at the multilayer regime @xcite .
aranovich and donohue @xcite presented a multilayer adsorption isotherm that should be capable to include multisite occupancy .
later , a closed exact solution for the multilayer adsorption isotherm of dimers was reported in ref .
it demonstrates that the determinations of surface areas and adsorption energies from polyatomic adsorbate adsorption ( also called @xmath2-mers ) may be misestimated , if this polyatomic character is not properly incorporated in the thermodynamic functions from which experiments are interpreted .
this treatment has recently been extended @xcite to include the two - dimensional nature of the substrate .
there are another two important physical facts which have not been sufficiently studied : @xmath6 the effect of the lateral interactions between the ad - molecules and @xmath7 the effect of surface heterogeneity in presence of multisite and multilayer adsorption . in the first case
, a recent paper @xcite extends the bet equation to include nearest - neighbor lateral interactions between the molecules adsorbed in the first layer .
following the configuration - counting procedure of the bragg - williams approach and the quasi - chemical approximation , ref .
@xcite gives us a simple statistical mechanical framework for studying multilayer adsorption of interacting polyatomics . in the second case , multilayer adsorption of polyatomics on heterogeneous surfaces is a topic of interest in surface science since it comprises most features present in experimental situations .
surfaces generally present inhomogeneities due to irregular arrangement of surface and bulk atoms , the presence of various chemical species , etc . in this context , a recent job shows how the monolayer volume predicted by bet equation differs from its real value when considering both the adsorbate size and the surface topography @xcite . on the other hand ,
combined effects coming from lateral interactions , multisite occupancy and surface heterogeneity have been analyzed in the interesting paper by nikitas @xcite . in ref .
@xcite , the author concludes that : @xmath3 one can obtain an underestimation of the true monolayer capacity of the order of @xmath8 when the adsorbate occupies more than one lattice site , and this underestimation will become worse if the effect of the multisite occupancy is coupled with heterogeneity effects and @xmath4 the attractive interactions in the gas adsorption lead always to a weak overestimation of the monolayer volume .
the previously discussed issues can be outlined in fig .
1 , where dimensions of complexity are sketched . in each axis
we see a particular characteristic of the adsorption process that can ( or can not ) be present in a particular model .
usually , the more features a model has , the more is its generality ; and more complex the process it describes . in the multilayer regime , we can see models that include adsorption with lateral interactions , but they do not allow to multisite occupancy , or incorporate surface energetic heterogeneity , but they overlook lateral interactions in the adsorbed layer , etc . with fig .
1 we try to show that , at present , some models venture in some of these dimensions of complexity but , the vast majority , lack the complete set of complexities .
it is also clear that the farther away we are from the origin the more general ( and usually the more complex ) the model is .
even more , the upper plane ( including multisite occupancy ) , is a set of models developed in the last decade that represent a leap forward in the lattice - gas theory . in this work
we will attempt to take one step further away from the origin , into a more general and encompassing scheme , presenting a model that considers the totality of these dimensions of complexity ( multilayer adsorption , lateral interactions , multisite occupancy , energetic heterogeneity and surface topography ) .
for this purpose , a theoretical formalism is presented based upon the analytical expression of the adsorption isotherm with lateral interaction weighted by the characteristic length of the surface heterogeneous .
in addition , monte carlo ( mc ) simulations are performed in order to test the validity of the theoretical model .
the new theoretical scheme allows us @xmath9 to obtain an accurate approximation for multilayer adsorption on two - dimensional substrates accounting multisite occupancy , lateral interactions , energetic heterogeneity and surface topography , and @xmath10 to provide a simple model from which experiments may be reinterpreted .
the present work is organized as follows . in section
ii , we present the theoretical formalism .
section iii is devoted to describing the simulation scheme .
the analysis of the results and discussion are given in section iv .
finally , the conclusions are drawn in section v.
in this section , we will present the model and make a revision of some existing theories describing the adsorption process in the framework of the lattice - gas approximation .
we will focus on those contributions that give contextualized understanding of the work that we will present .
the substrate is modeled by a regular lattice of @xmath11 sites with periodic boundary conditions , where the adsorption energy of the first layer @xmath12 depends on each site @xmath13 of the surface and the adsorbate is represented by @xmath2-mers ( linear particles that have @xmath2 identical units ) .
a @xmath2-mer adsorbed occupies @xmath2 sites of the lattice and can arrange in many configurations .
this property is called multisite - occupancy adsorption . on the other hand , for higher layers ,
the adsorption of a @xmath2-mer is exactly onto an already absorbed one , with an adsorption energy of @xmath14 .
thus , the monolayer structure reproduces in the remaining layers .
the mechanism used to describe the adsorption in the multilayer regime mimics the phenomenon called pseudomorphism .
this was observed , by using low - energy electron diffraction technique @xcite , in the case of adsorption of straight chain saturated hydrocarbon molecules on metallic surfaces .
however , recent synchrotron x - ray scattering measurements show a different and more complex growth process in these systems @xcite .
nevertheless , for simplicity , here we study a variant of the model that maintains the same mechanism to describe the adsorption in the higher layers . because we analyze isotherms in the low coverage regime
, we expect that this choice will not affect our results .
finally , attractive and repulsive lateral interactions are considered in the first layer and horizontal interactions are ignored in higher layers . in this sense , although the contribution from the secondary adsorption can already be significant , the density of the molecules in the second and higher adlayers is expected to be much lower than that in the first adsorbed layer [ bet equation can be applied at coverage not greatly exceeding ( statistically ) monolayer coverage ] .
therefore , it seems to be satisfactorily enough to take into account only the interactions between the primarily adsorbed molecules @xcite . under these considerations
we can write the hamiltonian as : @xmath15 where the first term of the right - hand side ( rhs ) represents the adsorption energy of the @xmath16 @xmath2-mers adsorbed in the first layer ( adjacent to de adsorbent ) and the second term is the energy of the @xmath17 @xmath2-mers adsorbed on top of the first layer ( second layer , third layer , and so on ) .
the third and fourth terms correspond to the lateral interaction energy , where @xmath18 is the interaction energy between two nearest - neighbor ( nn ) units belonging to different @xmath2-mers adsorbed in the first layer ( we use @xmath19 for repulsive and @xmath20 for attractive interactions ) ; @xmath21 is the occupation variable which can take the values 0 if the site @xmath13 is empty or 1 if the site @xmath13 is occupied and @xmath22 represents pairs of nn sites . since the summation in the third term overestimates the total energy by including @xmath23 bonds belonging to the @xmath16 adsorbed @xmath2-mers , the fourth term subtracts this exceeding energy .
our study will be restricted to the class of lattice - gas models in which the substrate is a regular array of individual adsorption sites where molecules can be deposited ( also in a discrete manner ) . in the following sections we will review some analytical deductions of those models . a more in depth treatment
is carried out in the original publications . one way to begin
our analysis is to review the behavior of a system of identical particles , which will be adsorbed in a regular lattice of @xmath11 identical sites of adsorption @xcite .
the supposition is that , like in the bet model , adsorption is done on the surface of the solid or onto an already adsorbed particle . in this case , in stead of adsorbing spherically symmetric ad - atoms , we will use entities ( that we will call @xmath2-mers ) that occupy @xmath2 consecutive lattice sites .
therefore , we will only have two possible unit - adsorption processes : @xmath9 a @xmath2-mer occupies @xmath2 consecutive empty surface sites ; and @xmath10 a @xmath2-mer adsorbs on top of an already adsorbed @xmath2-mer .
this adsorptive process will form columns of @xmath2-mers on the solid .
we must notice that a unit - adsorption where a @xmath2-mer adsorbs on top of two already adsorbed @xmath2-mers is prohibited . in the case of the unit - desorption processes , we will only be able to desorb a @xmath2-mer that is on the top of one @xmath2-mer column or a @xmath2-mer that is adsorbed on the solid surface , but has no other @xmath2-mers adsorbed on top of it . under these conditions ,
the grand partition function of the system is : @xmath24 where @xmath25 is the maximum number of columns that can be formed , @xmath26 it is the total number of distinguishable configurations of @xmath27 columns in @xmath11 sites and @xmath28 is the grand partition function of a unique column of @xmath2-mers that has at least one @xmath2-mer in the first layer . on the other hand
, we have that the grand partition function for the monolayer ( @xmath29 ) is : @xmath30 being @xmath27 , in this case , the number of @xmath2-mers adsorbed on the surface of the solid ( first layer ) , and @xmath31 the fugacity of the monolayer .
@xmath32 is still the number of possible configurations with @xmath27 @xmath2-mers in @xmath11 sites .
this quantity must be equal in eqs .
( [ eq : newtheoricalmulticapa ] ) and ( [ eq : newtheoricalmonocapa ] ) .
if now we compare eqs .
( [ eq : newtheoricalmulticapa ] ) and ( [ eq : newtheoricalmonocapa ] ) we can observe that they have a similar form . this allows us to write : @xmath33 if , in addition , inspired by this similarity , we propose the following ansatz : @xmath34 then @xmath35 where @xmath36 $ ] is the ratio between the partition function of a particle in the first layer and a particle in any other layer [ @xmath37 , being @xmath38 the boltzmann constant ] . and , then , we can write the monolayer coverage as : @xmath39 now , the total coverage ( @xmath40 ) can be written in terms of the coverage of the monolayer ( @xmath41 ) : @xmath42 the theoretical procedure in eqs .
( [ eq:4nt])-([eq:7nta ] ) provides the isotherm in the multilayer regime from the isotherm in the monolayer regime .
in fact : \(1 ) by using @xmath43 as a parameter ( @xmath44 ) , the relative pressure is obtained by using eq .
( [ eq:4nt ] ) .
this calculation requires the knowledge of an analytical expression for the monolayer adsorption isotherm .
\(2 ) the values of @xmath43 and @xmath45 are introduced in eq .
( [ eq:7nta ] ) and the total coverage is obtained .
the items ( 1 ) and ( 2 ) are summarized in the following scheme : @xmath46 @xmath47 following the previous scheme , it is possible to obtain the exact multilayer isotherm for @xmath2-mers in 1d homogeneous surfaces .
in fact , in ref .
@xcite the expression for the monolayer coverage is proved to be : @xmath48^{k-1}}{kc(1-\theta _ { 1})^k+\left[1-\frac{(k-1)}{k } \theta _ { 1}\right]^{k-1}}. \label{eq:21nt}\ ] ] equations ( [ eq:7nta ] ) and ( [ eq:21nt ] ) represent the exact solution of the 1d model and , as it is expected , retrieve the bet equation for the case @xmath49 . also , the previous scheme can be used to obtain an accurate approximation for multilayer adsorption on 2d substrates accounting multisite occupancy . in this case
, the semi - empirical monolayer adsorption isotherm @xcite can be used @xmath50^{\left(k-1 \right ) \theta_1 } \left [ 1- \frac{2 \left ( k-1 \right)}{\zeta k } \theta_1 \right]^{\left(k-1 \right ) \left ( 1- \theta_1 \right)}}{z k c \left ( 1-\theta_1\right)^k + \theta_1 \left [ 1- \frac{\left ( k-1 \right)}{k } \theta_1 \right]^{\left(k-1 \right ) \theta_1 } \left [ 1- \frac{2 \left ( k-1 \right)}{\zeta k } \theta_1 \right]^{\left(k-1 \right ) \left ( 1- \theta_1 \right ) } } , \label{pre2d}\ ] ] where @xmath51 is the connectivity of the lattice and @xmath52 represents the number of available configurations ( per lattice site ) for a linear @xmath2-mer at zero coverage .
thus , @xmath53 for @xmath49 and @xmath54 for @xmath55 .
note that , for @xmath56 , eq . ( [ pre2d ] )
is identical to the eq .
( [ eq:21nt ] ) . therefore , eqs .
( [ eq:7nta ] ) and ( [ pre2d ] ) represent the general solution of the problem of multilayer adsorption in homogeneous surfaces with multisite occupancy . in the following section
our aim will be to generalize this expression to include the lateral interactions following the methodology used in ref .
clearly , the complexity of the isotherm is greatly increased , but the expression is still a manageable one and in the case of zero interactions the previous case is retrieved . due to the complexity introduced in the analytical expressions because of
the lateral interactions is that approximations are used to deal with this feature of the adsorption process .
the most commonly used approximations are the _ mean - field approximation _ ( mfa ) @xcite and the _ quasi - chemical approximation _ ( qca ) @xcite .
as shown in previous work@xcite , the configuration - counting procedure of the qca allows us to obtain an approximation that is significantly better than the mfa for polyatomics .
based on this finding , we restrict the rest of our discussion to the estimates obtained under qca . in order to apply the theoretical scheme described in previous section , we start with the monolayer adsorption isotherm of interacting @xmath2-mers adsorbed on a lattice of connectivity @xmath51 obtained from the formalism of qca @xcite , @xmath57 \left [ \frac{(1-\theta_1)^{k(\zeta-1)}\left [ k - ( k-1 ) \theta_1 \right]^{k-1 } \left[\frac{z \theta_1}{2k}- \alpha \right]^{z/2}}{\left[\frac{\zeta k}{2}-(k-1)\theta_1 ) \right]^{k-1 } \left[\frac{\zeta}{2}(1-\theta_1)- \alpha \right]^{k \zeta /2 } \left(\frac{z \theta_1}{\zeta k } \right)^{z}}\right ] , \label{mu}\ ] ] where @xmath58 \label{lam},\ ] ] @xmath59 } , \label{alfa}\ ] ] @xmath60 ^ 2 - \frac{z \zeta}{k } a \theta_1 ( 1-\theta_1 ) \right\}^{1/2 } , \label{b}\ ] ] and @xmath61 replacing eq .
( [ mu ] ) into eq .
( [ eq:4nt ] ) , we obtain @xmath62^{k-1 } \left[\frac{\zeta}{2}(1-\theta_1)- \alpha \right]^{k \zeta /2 } \left(\frac{z \theta_1}{\zeta k } \right)^{z } } { \theta_1 \exp { \left ( \beta w z /2 \right)}(1-\theta_1)^{k(\zeta-1)}\left [ k - ( k-1 ) \theta_1 \right]^{k-1 } \left[\frac{z \theta_1}{2k}- \alpha \right]^{z/2 } } \label{pqca}.\ ] ] eqs .
( [ eq:7nta ] ) and ( [ pqca ] ) represent the solution describing the multilayer adsorption of interacting @xmath2-mers on homogeneous surfaces in the framework of the qca .
this method is presented in more depth in @xcite .
now we can see that we have moved upward in the plane of complexity and further more included interactions .
the next section revises the role of surface heterogeneity and lateral interactions . in the two previous sections
, we obtained the multilayer isotherm from the monolayer isotherm .
it is possible demonstrate that this formalism still holds for interacting @xmath2-mers and a given surface heterogeneity .
however , this strategy leads to a complex solution that is not useful for practical purposes . to build a simpler function ( easier to analyze ) , we have chosen to approximate the multilayer heterogeneous isotherm by a weighted sum of multilayer homogeneous isotherms .
the heterogeneous surface is modeled by two kinds of adsorption sites in the first layer ( bivariate surface ) : strong sites with adsorption energy @xmath63 and weak sites with adsorption energy @xmath64 .
as seen later , these sites can be spatially distributed in different ways ( different topographies ) .
then , the total adsorption energy for an isolated @xmath2-mer on the first layer with @xmath65 monomers located over strong sites and @xmath66 monomers located over weak sites is @xmath67 under these considerations , and using the formalism of the integral equation of the adsorption isotherm @xcite , the mean coverage @xmath40 can be written as @xmath68 where @xmath69 is the fraction of @xmath2-uples of @xmath65 strong sites and @xmath66 weak sites ( @xmath70 ) and @xmath71 represents the local multilayer adsorption isotherm corresponding to an adsorptive energy @xmath72 .
this local isotherm can be well approximated by using the multilayer adsorption isotherm associated to an homogeneous surface characterized by an effective value of @xmath73 given by @xmath74.\ ] ] this value of @xmath75 can also be expressed as function of @xmath76 and @xmath77 , the values of @xmath73 for homogeneous surfaces whose adsorption energies are @xmath63 and @xmath64 , respectively .
thus , if the @xmath13-th term in eq .
( [ isoh ] ) corresponds to a @xmath2-mer with @xmath65 units located over strong sites and @xmath66 units located over weak sites , then @xmath78 as an example , let us consider the multilayer adsorption of non - interacting dimers on a bivariate linear lattice , where strong and weak sites are spatially distributed in alternating homotattic patches of size @xmath79 ( @xmath80 ) . in this case , eq .
( [ isoh ] ) has three different terms , being each one of them a dimer isotherm and @xmath81 , it is possible to solve eqs .
( [ eq:7nta ] ) and ( [ eq:21nt ] ) to obtain single expressions of the multilayer isotherms corresponding to monomers and dimers . ] with a particular value of @xmath73 , @xmath82^{1/2 } \right\ } + \nonumber\\ & + & \left ( \frac{1}{l } \right ) \frac{1}{\left ( 1- p / p_0 \right ) } \left\ { 1 - \left [ \frac{1-p / p_0}{1 + \left ( 4 \sqrt{c_1 c_2 } -1 \right)p / p_0 } \right]^{1/2 } \right\ } + \nonumber\\ & + & \left ( \frac{l-1}{2l } \right ) \frac{1}{\left ( 1- p / p_0 \right ) } \left\ { 1 - \left [ \frac{1-p / p_0}{1 + \left ( 4c_2 - 1
\right)p / p_0 } \right]^{1/2 } \right\}. \label{isoh1dk2}\end{aligned}\ ] ] the first [ third ] term in the rhs of eq .
( [ isoh1dk2 ] ) represents the adsorption within a strong [ weak ] patch , on a pair of sites ( @xmath83 ) [ ( @xmath84 ) ] , with @xmath76 [ @xmath77 ] .
the fraction of ( @xmath83 ) [ ( @xmath84 ) ] pairs on the lattice is @xmath86 @xcite . the remaining term of eq .
( [ isoh1dk2 ] ) corresponds to a dimer isotherm on a ( @xmath87 ) [ or ( @xmath88 ) ] pair ( @xmath89 ) , being @xmath90 the fraction of this kind of pairs on the lattice . as it can be observed , eq .
( [ isoh1dk2 ] ) depends on @xmath79 and the dimer isotherm _ sees _ the topography . in general , the number of terms in eq .
( [ isoh ] ) increases as the adsorbate size @xmath2 is increased and this equation leads to a complex solution .
this scheme can be notoriously simplified following the results in ref .
@xcite . in this paper
, the authors showed that : * if @xmath91 ( with @xmath92 ) , the multilayer adsorption isotherm can be represented by a single homogeneous isotherm @xmath93 * for a topography where @xmath94 , the isotherm is @xmath95 * the details of the topography are relevant only when @xmath96 . in this case , it is possible to consider a simpler expression of the multilayer isotherm given by @xmath97 eq .
( [ isoh1daprox ] ) @xmath3 captures the extreme behaviors eqs .
( [ l1 ] ) and ( [ lpt ] ) , and @xmath4 approximates very well the complete eq .
( [ isoh ] ) in 1d and 2d . in the 2d case , the local isotherm in eq .
( [ isoh1daprox ] ) is obtained from eqs .
( [ eq:7nta ] ) and ( [ pre2d ] ) , with @xmath98 , @xmath99 and @xmath100 for honeycomb , square and triangular lattices , respectively . finally , we propose to extend eq .
( [ isoh1daprox ] ) as to describe multilayer adsorption of interacting polyatomic molecules on heterogeneous surfaces . for this purpose ,
we propose to obtain the local isotherms from eqs .
( [ eq:7nta ] ) and ( [ pqca ] ) .
the advantages of using this simple description as a tool for interpreting multilayer adsorption data and characterization of the adsorption potential will be shown in section 4 by analyzing simulation results .
the adsorption process is simulated through a grand canonical ensemble monte carlo ( gcemc ) method . for a given value of the temperature @xmath101 and chemical potential @xmath102 , an initial configuration with @xmath103 @xmath2-mers adsorbed at random positions ( on @xmath104 sites )
is generated .
then , an adsorption - desorption process is started , where each elementary step is attempted with a probability given by the metropolis @xcite rule : @xmath105}\right\},\ ] ] where @xmath106 and @xmath107 represent the difference between the hamiltonians and the variation in the number of particles , respectively , when the system changes from an initial state to a final state . in the process
there are four elementary ways to perform a change of the system state , namely , adsorbing one molecule onto the surface , desorbing one molecule from the surface , adsorbing one molecule in the bulk liquid phase and desorbing one molecule from the bulk liquid phase . in all cases , @xmath108 .
the algorithm to carry out one mc step ( mcs ) , is the following : * set the value of the chemical potential @xmath102 and the temperature @xmath101 . *
set an initial state by adsorbing @xmath103 molecules in the system .
each @xmath2-mer can adsorb in two different ways : @xmath109 on a linear array of ( @xmath2 ) empty sites on the surface or @xmath110 exactly onto an already adsorbed @xmath2-mer . *
introduce an array , denoted as @xmath111 , storing the coordinates of @xmath112 entities , being @xmath112 , @xmath113 where @xmath114 is the sum of two terms : @xmath109 the number of @xmath2-uples of empty sites on the surface and @xmath110 the number of columns of adsorbed @xmath2-mers - uple for the adsorption of one @xmath2-mer . ] .
* choose randomly one of the @xmath112 entities , and generate a random number @xmath115 $ ] * * if the selected entity is a @xmath2-uple of empty sites on the surface then adsorb a @xmath2-mer if @xmath116 , being @xmath117 the transition probability of adsorbing one molecule onto the surface .
* * if the selected entity is a @xmath2-uple of empty sites on the top of a column of height @xmath13 , then adsorb a new @xmath2-mer in the @xmath118 layer if @xmath119 , being @xmath120 the transition probability of adsorbing one molecule in the bulk liquid phase . * * if the selected entity is a @xmath2-mer on the surface then desorb the @xmath2-mer if @xmath121 , being @xmath122 the transition probability of desorbing one molecule from the surface . *
* if the selected entity is a @xmath2-mer on the top of a column then desorb the @xmath2-mer if @xmath123 , being @xmath124 the transition probability of desorbing one molecule from the bulk liquid phase . *
if an adsorption ( desorption ) is accepted in @xmath125 , then , the array @xmath111 is updated . *
repeat from step @xmath125 @xmath11 times .
in the present case , the equilibrium state could be well reproduced after discarding the first @xmath126 .
then , averages were taken over @xmath127 successive configurations .
the total coverage was obtained as simple averages , @xmath128 where @xmath129 is the mean number of adsorbed particles , and @xmath130 means the time average over the mc simulation runs .
in the present section , we will analyze the main characteristics of the multilayer adsorption isotherms given by eq .
( [ isoh1daprox ] ) [ with local isotherms obtained from eqs .
( [ eq:7nta ] ) and ( [ pqca ] ) ] , in comparison with simulation results for a lattice - gas of interacting @xmath2-mers on heterogeneous one - dimensional and square lattices .
heterogeneity is introduced by considering bivariate surfaces , i.e. , surfaces composed by two kinds of sites in the first layer , strong and weak sites , with adsorptive energies @xmath63 and @xmath64 , respectively .
recent developments in the theory of adsorption on heterogeneous surfaces , like the supersite approach @xcite , and experimental advances in the tayloring of nanostructured adsorbates @xcite , encourage this kind of study .
a special class of bivariate surfaces , with a chessboard structure , has been observed recently to occur in a natural system @xcite , although it was already intensively used in modeling adsorption and surface diffusion phenomena @xcite .
bivariate surfaces may also mimic , to a rough approximation , more general heterogeneous adsorbates . just to give a few examples
, we may mention the surfaces with energetic topography arising from a continuous distribution of adsorptive energy with spatial correlations , like those described by the dual site - bond model @xcite , or that arising from a solid where a small amount of randomly distributed impurity ( strongly adsorptive ) atoms are added @xcite . in both cases the energetic topography could be roughly represented by a random spatial distribution of irregular patches ( with a characteristic size ) of weak and strong sites . in the particular case studied in this article , the surface
is modeled in two different ways : @xmath9 as a chain of alternating patches of size @xmath79 ( see fig .
2a ) ; and @xmath10 as a collection of finite homotattic patches in a chessboard - like array , where each patch is assumed to be a domain of equal size , @xmath131 sites ( see fig .
2b ) . in this model , the energy correlation length is simply given by the patch size .
the computational simulations have been developed for one - dimensional chains of @xmath132 sites , and square @xmath133 lattices with @xmath134 , and periodic boundary conditions . with this lattice size
we verified that finite - size effects are negligible .
note , however , that the linear dimension @xmath135 has to be properly chosen in such a way that it is a multiple of @xmath79 .
we start analyzing what happens when the topography is changed .
3 shows the behavior of the multilayer adsorption isotherms for @xmath81 , @xmath136 and different topographies in 1d as indicated .
the energy difference between different patches has been chosen to be high ( @xmath137 and @xmath138 ) in order to emphasize the effects of the surface heterogeneity .
we have identified the different topographies as @xmath139 for patches of size @xmath79 , and @xmath140 for the case of a surface with two big patches ( @xmath141 ) .
symbols represent simulation results and lines correspond to theoretical data [ eq .
( [ isoh1daprox ] ) ] .
it can be seen that all curves are contained between the two limit ones : the one corresponding to @xmath142 and the one corresponding to @xmath140 . for @xmath143 ,
the adsorption energy of a dimer in the first layer is @xmath144 for all configuration . in this condition
, the system corresponds to a 1d lattice - gas of interacting dimers on a homogeneous surface and , consequently , eq .
( [ isoh1daprox ] ) is exact . in general , for @xmath145
, this equation is approximate .
for @xmath146 , the analytic isotherms agree very well with the simulation data .
however , for @xmath45 ranging between 0 and 0.15 , some differences between theoretical and numerical data are observed .
this happens because eq .
( [ isoh1daprox ] ) has been built assuming that the three different pairs of sites are filled simultaneously and independently .
however , for @xmath147 , the real process occurs in 3 stages : @xmath3 the pairs of sites ( @xmath83 ) are covered ; @xmath4 the pairs ( @xmath84 ) begin to be filled and @xmath5 the multilayer is formed .
note that in the first stage all the pair of sites ( @xmath87 ) and ( @xmath88 ) are removed .
for this regime , a better approximation can be obtained by a semisum of two isotherms with @xmath76 y @xmath77 . when @xmath148 , the agreement between the analytic isotherms and the simulation data is very good . in this case , for @xmath147 the first stage does not eliminate all the pairs of sites ( @xmath87 ) and ( @xmath88 ) , because each dimer occupies only two sites in the strong patches .
for this reason , the range of validity of eq .
( [ isoh1daprox ] ) is wider than in the previous case .
now , if @xmath149 or @xmath150 , the behaviors are similar to those observed for @xmath146 or @xmath148 , respectively . in general , for even @xmath79 ,
the first stage eliminates almost completely the pairs of sites ( @xmath87 ) and ( @xmath88 ) , while this does not happen for odd @xmath79 . finally , when @xmath151 , the fraction of pair ( @xmath87 ) and ( @xmath88 ) goes to zero and eq .
( [ isoh1daprox ] ) is exact .
this limit corresponds to the called large patches topography ( @xmath140 surface in our model ) , where the surface is assumed to be a collection of homogeneous patches , large enough to neglect border effects between neighbor patches with different adsorption energies .
we now analyze the effect of the lateral interactions on the behavior of the system . for this purpose , fig .
4 shows the adsorption isotherms for @xmath81 , @xmath137 , @xmath138 and two different values of the lateral interactions @xmath136 ( attractive case ) and @xmath152 ( repulsive case ) .
in addition , for each value of @xmath153 , the limit topographies ( @xmath142 and @xmath140 ) have been considered ( as seen in fig .
3 , all curves corresponding to all topographies are contained between them ) .
for repulsive couplings , the interactions do not favor the adsorption on the first layer and the isotherms shift to higher values of pressure .
on the other hand , attractive lateral interactions facilitate the formation of the monolayer .
consequently , the isotherms shift to lower values of @xmath154 and their slope increases as the ratio @xmath155 increases .
in both cases ( @xmath136 and @xmath152 ) , the agreement between theoretical and simulation data is excellent . from the curves in fig . 4 ( and from data not shown here for the sake of clarity )
it is observed that : there exists a wide range of @xmath153 s ( @xmath156 ) , where the theory provides an excellent fitting of the simulation data .
in addition , most of the experiments in surface science are carried out in this range of interaction energy .
then , the present theory not only represents a qualitative advance in the description of the multilayer adsorption of interacting @xmath2-mers on heterogeneous surfaces , but also gives a framework and compact equations to consistently interpret thermodynamic multilayer adsorption experiments of polyatomics species such as alkanes , alkenes , and other hydrocarbons on regular surfaces .
the effect of energetic heterogeneity ( ratio between @xmath76 and @xmath77 ) is analyzed in fig .
5 , where the degree of heterogeneity is varied by changing the value of @xmath77 between 1 and 100 with @xmath76 fixed ( @xmath137 ) . as in fig . 4 ,
@xmath81 and @xmath136 .
lines represent theoretical results and symbols correspond to simulation data ( @xmath138 : circles ; @xmath157 : triangles ; and @xmath158 : squares ) . for each set of values of the parameters ,
the limit cases corresponding to @xmath142 ( open symbols ) and @xmath140 ( solid symbols ) topographies are studied . as it can be observed from the simple inspection of the figure , the effect of topography is important in a range of @xmath159 between @xmath160 and @xmath161 and is practically negligible for @xmath162 . to complete
the discussion started in fig .
3 , we now evaluate the effect of the @xmath2-mer size on the adsorption isotherms .
this study is shown in fig .
6 , where the @xmath142 and @xmath140 multilayer adsorption isotherms are plotted for @xmath136 , @xmath137 , @xmath138 and two different values of @xmath2 ( @xmath81 and @xmath163 )
. one important conclusion can be drawn from the figure .
namely , the effects of topography and energetic heterogeneity tends to disappear as the size @xmath2 is increased .
the study in figs .
3 - 6 was repeated for surfaces in 2d .
the behavior of the curves ( not shown here for brevity ) is very similar to the one observed in 1d . summarizing
, we have shown that just by using an expression of three terms , eq .
( [ isoh1daprox ] ) , we can approach very well the multilayer isotherm in 1d and 2d for the adsorption of interacting polyatomics on heterogeneous surfaces . in the next
, we will use this approximation and monte carlo simulations to study how lateral interactions , multisite occupancy and surface heterogeneity affect the determination of monolayer volume predicted by the bet equation . in a typical experiment of adsorption ,
the adsorbed volume of the gas , @xmath164 , is measured at different pressures and at a given fixed temperature , the total coverage is @xmath165 .
analyzing an isotherm with the bet equation @xmath166 it is possible to estimate the monolayer volume if we rewrite the previous equation as : @xmath167 this equation is a linear function of @xmath45 .
if we denote with @xmath168 and @xmath169 , the @xmath170-intercept and the slope of this straight line , respectively , we obtain @xmath171 and @xmath172 the asterisk has been added in order to indicate that the quantities given by eqs .
( [ vas ] ) and ( [ cas ] ) correspond to the prediction of the bet theory .
then , by means of a plot ( the so - called bet plot ) of the experimental data of @xmath173 vs @xmath45 , we can obtain an estimate of the monolayer volume and the parameter @xmath73 .
nevertheless , in the experiments it is commonly found that there are deviations from linearity in the bet plot . following the scheme described in previous paragraphs
, we carry out numerical experiments to determine , in different adsorption situations , how much the value of the monolayer volume predicted by the bet equation differs from its real value , @xmath174 . with this purpose ,
analytic and simulation isotherms were analyzed as experimental data . in this way
, we have determined how adsorbate size , surface heterogeneity and lateral interactions , affect the standard determination of the monolayer volume . in fig .
7 we plot the calculated value of @xmath175 as a function of @xmath2 for fixed values of @xmath76 and @xmath77 ( namely @xmath137 and @xmath138 ) and different values of the attractive lateral interaction .
the repulsive interactions , not shown in the figure , show a marked deviation from the predicted bet equation , therefore wiping out compensation effects due to the surface heterogeneity .
we can see three sets of plots in fig . 7 .
each set ( 1 ) is the result of the effect of different lateral interactions strengths : @xmath176 , circles ; @xmath177 , squares and @xmath178 , triangles ; and ( 2 ) depicts the limiting cases for the surface topography : open and solid symbols correspond to @xmath142 and @xmath140 surfaces , respectively .
all other topographies must lay in between those two plots .
if we analyze the behavior of the open symbols , we are analyzing , essentially , the homogenous case @xcite where an already known feature for the homogeneous scenario can be distinguished : the compensation effect of @xmath2 is lower as @xmath2 increases , as @xcite stated for the non - interacting case . on the other hand , for @xmath140 surfaces , the compensation effect increases with @xmath2 as seen in @xcite . in all cases , for stronger ( attractive ) interaction strength the plots move upwards showing greater compensation effects .
as shown in previous work @xcite for non - interacting @xmath2-mers on homogeneous surfaces , the monolayer volume from the bet model diminishes with increasing values of the @xmath2-mer size .
the data presented in fig . 7 demonstrate that attractive lateral interactions and surface heterogeneity play a key role in the compensation of @xmath2-mer size effects .
this finding is very important because most of the experiments in surface science are carried out in these conditions .
finally , the study in fig .
7 is repeated for 2d surfaces ( see fig .
in this case , both the compensation effect for attractive interactions and the underestimation of the monolayer volume for repulsive interactions ( not shown here ) are more important than in the 1d case .
the explanation is simple : in 1d systems , particles will interact only at their ends , regardless of @xmath2 ; on the other hand , in 2d systems , each particle interacts at their ends , but also interacts along its @xmath2 monomers .
this makes the interaction energy grow linearly with @xmath2 ( if all first neighbors are occupied ) , and indicates that lateral interactions will play a more important role in two - dimensional adsorption systems than in one - dimensional ones .
in this work , we have studied the multilayer adsorption of interacting polyatomic molecules onto heterogeneous surface .
the polyatomic character of the adsorbate was modeled by a lattice gas of @xmath2-mers . with respect to lateral interactions ,
the ad - ad couplings in the monolayer were explicitly considered in the solutions .
the range of validity of analytical isotherms was analyzed by comparing theoretical and mc simulation results .
we also analyzed the 1d and 2d bet plots obtained using the theoretical and simulation isotherms .
for non - interacting @xmath2-mers , we found that the use of bet equation leads to an underestimate of the true monolayer volume : this volume diminishes as @xmath2 is increased .
the situation is different for the case of interacting molecules over heterogeneous surface .
thus , attractive lateral interactions favor the formation of the monolayer and , consequently , compensate the effect of the multisite occupancy . in this case
, the monolayer volume predicted by bet equation agrees very well with the corresponding true value . in the case of repulsive couplings ,
the lateral interactions impede the formation of the monolayer and the bet predictions are bad ( even worse than those obtained in the non - interacting case ) .
both the compensation effect for attractive interactions and the underestimation of the monolayer volume for repulsive interactions are more important for 2d systems .
this work was supported in part by conicet ( argentina ) under project number pip 112 - 200801 - 01332 ; universidad nacional de san luis ( argentina ) under project 322000 ; universidad tecnolgica nacional , facultad regional san rafael ( argentina ) under projects pqprsr 858 and pqcosr 526 and the national agency of scientific and technological promotion ( argentina ) under project 33328 pict 2005 .
fig . 1 : schematic representation of the evolution of complexity in the theoretical adsorption models . fig .
2 : schematic representation of heterogeneous bivariate square surfaces with chessboard topography .
the black ( white ) symbols correspond to strong ( weak ) adsorption sites .
( a ) one - dimensional lattice and ( b ) square lattice .
the patch size in this figure is @xmath179 .
3 : adsorption isotherms for dimers on one - dimensional lattices with @xmath136 and different topographies 1d as indicated . the energy difference between different patches has been chosen to be high ( @xmath137 and @xmath138 ) in order to emphasize the effects of the surface heterogeneity .
solid lines and symbols represent theoretical and simulation results , respectively . for each set of values of the parameters ,
the limit cases corresponding to @xmath142 ( open symbols ) and @xmath140 ( solid symbols ) topographies are shown .
4 : adsorption isotherms for dimers on one - dimensional lattices with @xmath137 , @xmath138 and two different values of the lateral interactions @xmath136 ( attractive case ) and @xmath152 ( repulsive case ) .
solid lines and symbols represent theoretical and simulation results , respectively .
for each set of values of the parameters , the limit cases corresponding to @xmath142 ( open symbols ) and @xmath140 ( solid symbols ) topographies are shown .
fig . 5 : adsorption isotherms for dimers on one - dimensional lattices with @xmath136 .
the degree of the surface heterogeneity is varied by changing the value of @xmath77 between 1 and 100 with @xmath76 fixed ( @xmath137 ) .
lines represent theoretical results and symbols correspond to simulation data ( @xmath138 : circles ; @xmath157 : triangles ; and @xmath158 : squares ) . for each set of values of the parameters ,
the limit cases corresponding to @xmath142 ( open symbols ) and @xmath140 ( solid symbols ) topographies are shown .
fig . 6 : adsorption isotherms for dimers on one - dimensional lattices with @xmath136 , @xmath137 , @xmath138 and two different values of @xmath2 ( @xmath81 and @xmath163 ) .
solid lines and symbols represent theoretical and simulation results , respectively . for each set of values of the parameters ,
the limit cases corresponding to @xmath142 ( open symbols ) and @xmath140 ( solid symbols ) topographies are shown .
fig . 7 : results of the bet plots for the adsorption in 1d heterogeneous surfaces with @xmath137 , @xmath138 .
dependence on @xmath2 of the fraction @xmath180 for three different values of @xmath153 : @xmath176 , circles ; @xmath181 , squares and @xmath178 , triangles . for each set of values of the parameters , the limit cases corresponding to @xmath142 ( open symbols ) and @xmath140 ( solid symbols ) topographies are shown .
fig . 8 : as fig . 7 for 2d surfaces : @xmath176 , circles ; @xmath182 , squares and @xmath177 , triangles .
symbols connected by ( solid ) dotted lines correspond to results obtained from ( theoretical ) mc simulation isotherms . | in the present work we introduce a generalized lattice - gas model to study the multilayer adsorption of interacting polyatomics on heterogeneous surfaces . using an approximation in the spirit of the well - known brunauer
emmet
teller ( bet ) model , a new theoretical isotherm is obtained in one- and two - dimensional lattices and compared with monte carlo simulation .
in addition , we use the bet approach to analyze these isotherms and to estimate the monolayer volume . in all cases
, we found that the use of the bet equation leads to an underestimate of the true monolayer capacity .
however , significant compensation effects were observed for heterogeneous surfaces and attractive lateral interactions . ,
, equilibrium thermodynamics and statistical mechanics , surface thermodynamics , adsorption isotherms , monte carlo simulations |
the study of extended radio galaxies in clusters of galaxies is important to understand the evolution of radio sources . in particular , the low - brightness lobes are at a late stage of evolution and are typically confined by the intergalactic medium , therefore synchrotron losses dominate in these regions .
studies of extended radio sources often suffer from limitations of the high - resolution interferometric observations , which can properly map the low - brightness regions only at low frequencies ( 610 327 mhz or lower ) . on the other hand , information on the relativistic particle ageing and on the polarization behaviour of the extended low - brightness regions
is only obtained with multiple frequencies .
the effelsberg 100-m radio telescope is at present the best available instrument to map the outermost regions of extended radio galaxies with good angular resolution , overcoming the problem of the lack of short spacings ( see e.g. andernach et al .
1992 and mack et al .
thanks to its high sensitivity , it is possible to map the extended structures at 10.6 ghz . at this high frequency ,
faraday effects are generally negligible , thus the intrinsic degree of polarization and the orientation of magnetic field can be derived .
moreover , energy losses in relativistic electrons are severely affecting the emission spectrum , allowing a determination of the particle ageing at different distances from the core . in this paper
we present new multi - frequency radio data of the extended tailed radio galaxies ic310 , ngc1265 , 3c129 , and 3c465 , revealing the spectral index trends and the polarization behaviour up to the outermost regions . the spectral steepening
can thence be used to determine the particle ages , following e.g. the procedure of carilli et al .
( 1991 ) . tailed radio sources are particularly suited for this purpose .
in fact , according to currently accepted models , the morphology of these sources is affected by the drag action exerted by the intergalactic medium on the moving galaxy , and the low - brightness tails consist of aged relativistic electrons , which have reached pressure equilibrium with the ambient gas and are loosing energy essentially through synchrotron radiation and the inverse compton effect . since the radio emitting electrons
are expected to be older at greater distance from the nucleus , the variation of the physical conditions with distance is related to the time evolution of the radio plasma .
lcclrccccc source & ra(b1950 ) & dec(b1950 ) & z & freq .
& # cov . & map size & & @xmath0 & @xmath1 + & [ @xmath2 @xmath3 @xmath4 & [ @xmath5 @xmath6 & & [ mhz ] & & & & [ mjy / beam ] & [ mjy / beam ] + ic310 & 03 13 25.2 & 41 08 30 & 0.0183 & 327 & - & - & 69 & 3.2 & - + & & & & 610 & - & - & 69 & 0.4 & - + & & & & 10550 & 12 & @xmath7 & 69 & 0.5 & 0.3 + ngc1265 & 03 14 56.8 & 41 40 32 & 0.0183 & 327 & - & - & 147 & 5.1 & - + & & & & 610 & - & - & 147 & 0.3 & - + & & & & 4750 & 13 & @xmath8 & 147 & 5.0 & 2.5 + & & & & 10550 & 22 & @xmath9 & 69 & 0.6 & 0.5 + 3c129 & 04 45 21.0 & 44 56 48 & 0.021 & 2695 & 30 & @xmath10 & 258 & 4.4 & 3.3 + & & & & 4750 & 30 & @xmath11 & 147 & 3.4 & 1.5 + & & & & 10550 & 21 & @xmath12 & 69 & 1.0 & 0.4 + 3c465 & 23 35 59.0 & 26 45 16 & 0.0322 & 610 & - & - & 147 & 5.0 & - + & & & & 4750 & 11 & @xmath13 & 147 & 2.3 & 1.3 + & & & & 10550 & 20 & @xmath9 & 69 & 0.9 & 0.5 + in sect . 2 we briefly describe the observations and data reduction techniques .
section 3 presents the results , which are discussed in sect . 4 , with emphasis on the particle lifetime and velocity , and the magnetic field morphologies .
in sect . 5
we summarize our results and present our conclusions .
a hubble constant h@xmath14 = 100 km s@xmath15 mpc@xmath15 and a q@xmath16 are assumed .
the observations at 10.6 , 4.8 , and 2.7 ghz have been carried out using the 100-m radio telescope at effelsberg . in table 1
we have listed the radio galaxies , together with their position and redshift , the observing frequency , the angular resolution in the final maps , the r.m.s .
noise level in the i and u , q maps
. fields were mapped in azimuth and elevation ( see emerson et al .
1979 for details ) for each source ensuring sufficiently large areas to also account for the beam throws envolved .
the number of coverages and field sizes for each source are given in table .
the 2.7-ghz map of 3c129 was obtained by scanning in right ascension and declination 30 times , as this receiver has a single feed so that mapping in az - el is not mandatory . at the two highest frequencies ,
difference maps were obtained to remove atmospheric fluctuations in the signals and the conventional restoration technique of emerson et al .
( 1979 ) was applied .
the resulting stokes i , u , and q maps were subsequently transformed into the equatorial coordinate system , and individual coverages finally averaged ( with weights proportional to the inverse squares of the rms noise values ) .
maps at 10.6 ghz were finally cleaned , with the algorithm described by klein & mack ( 1995 ) .
calibration of the flux density scale and the polarization parameters was achieved by frequently cross - scanning and mapping the point sources 3c 48 , 3c 138 , 3c 286 , and 3c 295 , with the flux density scale adopted from baars et al .
( 1977 ) . for the radio galaxies ic310 and ngc1265 , maps at 610 and 327 mhz were extracted from the image of the perseus cluster ( sijbring 1993 ) , obtained with the westerbork synthesis radio telescope ( wsrt ) .
the shortest baseline at both frequencies was 36 m , therefore only structures larger than 48@xmath17 at 610 mhz and @xmath18 at 327 mhz are missed .
this ensures that these measurements are not affected by the lack of short baselines , and a comparison between effelsberg and wsrt data is possible .
similarly , the wsrt map of the radio galaxy 3c465 , obtained by jgers ( 1987 ) with the wsrt at 610 mhz , was used in the present work .
the wsrt maps , available only in total intensity , were convolved to the resolution of the effelsberg maps .
the images were analyzed with the aips package . to compare observations of the same source at different frequencies we convolved all the maps to the lowest resolution with a gaussian .
thereafter , the maps were re - gridded by means of the task ` hgeom ' in aips to have identical pixel size .
ic310 is located in the perseus cluster .
previous observations of this radio galaxy are listed by mack et al .
( 1993 ) , who first mapped the large - scale structure of this source at a high frequency .
the 10.6-ghz map presented here in fig .
1 has a higher sensitivity , owing to additional scans . for the spectral comparison
, we used the wsrt maps at 327 and 610 mhz by sijbring ( 1993 ) , convolved to the same resolution of 69@xmath19 .
the trend of the spectral index ( @xmath20 and @xmath21 ; i@xmath22 ) derived along the maximum brightness ridge is given in fig .
the low - frequency spectral index @xmath20 is 0.4 , almost constant up to 2@xmath17 from the core , suggesting that the particles radiating in this region and at low frequencies have not suffered synchrotron losses . a clear steepening is visible at high frequencies and with increasing distance from the core .
the tail is strongly polarized , up to a distance of 3@xmath17 from the core , with the polarization percentage increasing along the tail ( see fig .
the orientation of the polarization vectors is perpendicular to the tail ( see fig . 1 ) .
ngc1265 is located in a peripheral region of the perseus cluster , at about 27@xmath17 north - west of the active galaxy ngc1275 ( 3c84 ) .
previous studies of this source are listed in mack et al .
( 1993 ) , who obtained the first image at 10.6 ghz . the 10.6-ghz map presented here ( fig .
2 ) has a much higher sensitivity than that of mack et al .
( 1993 ) , owing to a longer observing time .
we also obtained observations at 4750 mhz and extracted wsrt maps at 327 and 610 mhz from the perseus cluster image obtained by sijbring ( 1993 ) .
we note that the very - low - brightness feature first reported by gisler & miley ( 1979 ) and observed with higher sensitivity and dynamic range by burns et al .
( 1992 ) and by sijbring ( 1993 ) is just indicated in our 10.6-ghz map if smoothed to a 15 beam , especially in the ( more sensitive ) map of linear polarization . at the other frequencies
the feature , if present , is concealed by the sidelobe structure which we could not yet clean at this wavelength .
all maps were smoothed to the common resolution of 147@xmath19 for a multifrequency comparison .
the trend of the spectral index along the ridge of maximum brightness for the 3 frequency pairs is presented in fig .
5 . between 0.3 and 4.8
ghz it is in the range 0.5 0.8 , while beyond 4.8 ghz a strong steepening is evident , which increases with distance from the core , reaching a value @xmath23 2 at the tail s end .
strong linear polarization is found in the external tail region ( see fig .
the polarization percentage at 10.6 ghz increases along the tail up to @xmath2440% . at 4.8 ghz ,
the degree of polarization is comparable to that at 10.6 ghz up to 5@xmath17 from the core , and is lower beyond that distance .
the electric vector at both frequencies is perpendicular to the tail orientation .
this radio source is identified with an e galaxy in the galactic plane ( _ l_@xmath25=160.4@xmath5 , _
b_@xmath25=0.1@xmath5 ) .
it was studied by many authors ( e.g. miley 1973 , rudnick & burns 1981 , van breugel 1982 , van breugel & jgers 1982 ) with interferometric observations .
the effelsberg map at 10.6 ghz is given in fig .
maps were also obtained at 2.7 ghz and 4.8 ghz .
the tail of this source , which at lower frequencies is about 30@xmath17 long ( van breugel 1982 ) , is detected out to about 20@xmath17 at 2.7 ghz and to about 12@xmath17 at 10.6 ghz .
the spectral index along the source was obtained with an angular resolution of about 4@xmath17 ( fig .
the spectrum strongly steepens from the core to the outer regions , reaching values up to @xmath26@xmath243 .
the polarization percentage ( fig . 6 ) in the outermost tail region reaches values of @xmath2460 - 70% at high frequencies and is slightly lower at 2.7 ghz .
the polarization vector at 10.6 ghz is perpendicular to the tail , and shows significant rotation at lower frequencies . this source is identified with a cd galaxy at the centre of the rich cluster a2634 .
it is the prototype of wide - angle tailed radio sources ( see e.g. eilek et al .
1984 and references therein ) , and a model of its dynamical evolution has been proposed by leahy ( 1984 ) .
high - resolution vla maps show an unresolved core , and two asymmetric jets which terminate in a bright spot and after a sharp bending , in extended lobes . in our maps at 4.8 ghz ( not shown here ) and 10.6 ghz ( fig . 4 ) , both the n - s and and e - w tails are well resolved .
we derived the point - to - point spectrum in the tails , with an angular resolution of 147@xmath27 , using also the 610 mhz wsrt map of jgers ( 1987 ) .
the spectral index between 0.6 and 4.8 ghz is lower than that between 4.8 and 10.6 ghz , except for the outermost points of the n - s lobe , where however the two values are consistent within the errors .
a remarkable high - frequency steepening is present in the e - w lobe , at increasing distance from the core ( fig .
the polarization percentage at both 10.6 and 4.8 ghz is asymmetric and is higher in the n - s lobe ( fig .
the spectral behaviour in all the sources presented here is similar .
the spectrum between two given frequencies shows a progressive steepening with increasing distance from the core . moreover , the steepening is stronger in the high - frequency range than at lower frequencies . since the radio emitting electrons
are expected to be older at larger distance from the core , the variation of the spectrum with distance is related to its evolution with time under the effect of radiation losses .
crrrrrrr & @xmath28(kp ) & @xmath28(jp ) & h@xmath29 & age(kp)&age(jp)&vel(kp)&vel(jp ) + @xmath30$]&[ghz ] & [ ghz ] & [ @xmath31 g ] & [ @xmath3210@xmath33yr ] & [ @xmath3210@xmath33yr ] & [ 100 km s@xmath15 ] & [ 100 km s@xmath15 ] + 0.5 & @xmath3530.0 & @xmath3530.0 & 4.3 & @xmath362.0 & @xmath362.0 & @xmath353.8 & @xmath353.8 + 1.0 & @xmath3530.0 & @xmath3530.0 & 3.6 & @xmath362.2 & @xmath362.2 & @xmath356.9 & @xmath356.9 + 1.5 & 30.0 & 23.9 & 3.3 & 2.4 & 2.6 & 9.6 & 8.6 + 2.0 & 12.1 & 10.3 & 3.3 & 3.7 & 4.0 & 5.6 & 5.5 + 2.5 & 5.5 & 5.6 & 3.1 & 5.7 & 5.6 & 3.9 & 4.8 + 3.0 & 3.6 & 4.5 & 3.1 & 6.9 & 6.2 & 5.9 & 12.0 + 3.5 & 2.2 & 1.8 & 2.9 & 9.2 & 10.3 & 3.4 & 1.9 + 4.0 & 1.7 & 1.4 & 2.8 & 10.5 & 11.7 & 5.8 & 5.4 + 4.5 & 1.2 & 1.0 & 2.7 & 12.7 & 13.9 & 3.4 & 3.4 + 5.0 & 0.7 & 0.6 & 2.3 & 17.5 & 18.2 & 1.6 & 1.8 + 5.5 & 0.4 & 0.5 & 2.2 & 21.8 & 21.1 & 1.8 & 2.6 + 6.0 & 0.3 & 0.4 & 1.9 & 25.7 & 22.7 & 1.9 & 4.7 + crrrrrrr & @xmath28(kp ) & @xmath28(jp ) & h@xmath29 & age(kp ) & age(jp ) & vel(kp)&vel(jp ) + @xmath30$]&[ghz]&[ghz ] & [ @xmath31 g ] & [ @xmath3210@xmath33yr ] & [ @xmath3210@xmath33yr ] & [ 100 km s@xmath15 ] & [ 100 km s@xmath15 ] + 11 & @xmath3530.0 & @xmath3530.0 & 3.5 & @xmath362.3 & @xmath362.3 & @xmath356.6 & @xmath356.6 + 12 & @xmath3530.0 & 26.0 & 3.5 & @xmath362.3 & 2.5 & @xmath3513.2 & 12.3 + 13 & 17.6 & 14.5 & 3.2 & 3.1 & 3.4 & 14.6 & 15.4 + 14 & 12.9 & 11.0 & 3.1 & 3.7 & 4.0 & 26.5 & 26.9 + 15 & 10.9 & 9.5 & 3.1 & 4.0 & 4.3 & 45.7 & 50.3 + 16 & 9.1 & 8.2 & 3.0 & 4.4 & 4.7 & 35.1 & 38.7 + 17 & 7.8 & 7.2 & 2.9 & 4.9 & 5.1 & 35.9 & 39.7 + 18 & 6.5 & 6.2 & 2.7 & 5.5 & 5.6 & 25.6 & 29.0 + 19 & 5.5 & 5.3 & 2.6 & 6.0 & 6.1 & 28.5 & 29.6 + 10 & 4.8 & 4.6 & 2.3 & 6.6 & 6.7 & 26.5 & 25.1 + crrrrrrr & @xmath28(kp ) & @xmath28(jp ) & h@xmath29 & age(kp ) & age(jp ) & vel(kp)&vel(jp ) + @xmath30$]&[ghz]&[ghz]&[@xmath31g]&[@xmath3210@xmath33yr]&[@xmath3210@xmath33yr ] & [ 100 km s@xmath15 ] & [ 100 km s@xmath15 ] + 11 & @xmath3530.0 & @xmath3530.0 & 4.1 & @xmath362.1 & @xmath362.1 & @xmath358.4 & @xmath358.4 + 12 & @xmath3530.0 & @xmath3530.0 & 3.9 & @xmath362.1 & @xmath362.1 & @xmath3516.2 & @xmath3516.2 + 13 & @xmath3530.0 & @xmath3530.0 & 4.2 & @xmath362.0 & @xmath362.0 & @xmath3525.6 & @xmath3525.6 + 14 & 26.0 & 21.0 & 3.6 & 2.4 & 2.7 & 28.8 & 25.9 + 15 & 13.0 & 11.5 & 3.1 & 3.7 & 3.9 & 13.7 & 14.1 + 16 & 7.6 & 7.7 & 3.0 & 4.8 & 4.8 & 14.7 & 19.1 + 17 & 5.3 & 6.3 & 2.8 & 5.9 & 5.4 & 15.4 & 26.1 + 18 & 4.4 & 5.7 & 2.7 & 6.6 & 5.8 & 26.9 & 52.2 + 19 & 3.4 & 5.0 & 2.6 & 7.5 & 6.2 & 18.5 & 39.1 + 10 & 2.9 & 4.5 & 2.4 & 8.4 & 6.7 & 20.5 & 35.1 + 11 & 2.5 & 3.5 & 2.3 & 9.1 & 7.6 & 23.3 & 18.7 + 12 & n.d . & 2.8 & 2.1 & - & 8.6 & - & 17.8 + 13 & n.d . & 2.3 & 1.9 & - & 9.5 & - & 18.1 + 14 & n.d . & 1.9 & 1.6 & - & 10.3 & - & 22.7 + crrrrrrr & @xmath28(kp ) & @xmath28(jp ) & h@xmath29 & age(kp ) & age(jp ) & vel(kp)&vel(jp ) + @xmath30 $ ] & [ ghz ] & [ ghz ] & [ @xmath31 g ] & [ @xmath3210@xmath33yr ] & [ @xmath3210@xmath33yr ] & [ 100 km s@xmath15 ] & [ 100 km s@xmath15 ] + n - s 1 & @xmath3530.0 & @xmath3530.0 & 4.5 & @xmath361.9 & @xmath361.9 & @xmath3513.7 & @xmath3513.7 + 2 & @xmath3530.0 & @xmath3530.0 & 4.7 & @xmath361.8 & @xmath361.8 & @xmath3528.4 & @xmath3528.4 + 3 & @xmath3530.0 & @xmath3530.0 & 5.2 & @xmath361.7 & @xmath361.7 & @xmath3546.4 & @xmath3546.4 + 4 & 26.0 & 20.9 & 6.3 & 1.5 & 1.7 & 69.3 & 62.2 + 5 & n.d . & n.d . & 5.2 & - & - & - & - + 6 & n.d . & n.d . &
4.9 & - & - & - & - + 7 & n.d . & n.d . &
3.8 & - & - & - & - + 8 & n.d . & n.d . & 2.7 & - & - & - & - + e - w 1 & @xmath3530.0 & @xmath3530.0 & 5.0 & @xmath361.7 & @xmath361.7 & @xmath3514.9 & @xmath3514.9 + 2 & @xmath3530.0 & @xmath3530.0 & 6.5 & @xmath361.3 & @xmath361.3 & @xmath3538.5 & @xmath3538.5 + 3 & 27.9 & 22.6 & 7.7 & 1.2 & 1.3 & 67.3 & 60.5 + 4 & 13.0 & 11.3 & 6.2 & 2.1 & 2.3 & 26.1 & 25.3 + 5 & 5.5 & 6.5 & 4.7 & 4.2 & 3.9 & 12.3 & 16.1 + 6 & 3.5 & 4.3 & 3.9 & 6.1 & 5.5 & 14.1 & 16.3 + 7 & 3.5 & 3.8 & 3.4 & 6.6 & 6.3 & 53.8 & 31.9 + 8 & 4.6 & 4.4 & 3.2 & 5.9 & 6.0 & - & - + the synchrotron loss mechanism for an ensemble of electrons with an initial power - law energy distribution n(e)de @xmath37 e@xmath38de produces a curvature of the originally straight power - law emission spectrum i(@xmath39 ) @xmath40 , with @xmath26 = ( @xmath41 1)/2 .
the curvature manifests itself at the critical frequency @xmath28 , related to the electron age .
the shape of the expected synchrotron spectrum can be computed analytically as a function of the critical frequency , the electron energy distribution index @xmath41 and the evolution of the electron pitch - angle distribution with time ( pacholczyk 1970 ) .
the model of kardashev - pacholczyk ( kp ) is obtained in the case that electrons maintain the same pitch angle throughout their radiative lifetime ( kardashev 1962 ) .
the model of jaffe - perola ( jp ) assumes that there is a redistribution of electron pitch angles on short time scales in comparison with their radiative lifetimes , due to their scattering on magnetic field irregularities ( jaffe & perola 1973 ) .
we have fitted the spectra of our sources obtained at different distances from the core ( see previous section ) using both the kp and jp models , since we do not have observational evidence in favour of one of them .
we also used different values of the exponent @xmath41 of the energy distribution function of the electrons , when no information about this parameter could be obtained from the low - frequency spectrum .
the fitting procedure was developed by murgia & fanti ( 1996 ) . in fig .
7 we present an example of spectral fits . in tables 2 to 5 we give the break frequencies obtained from the fitting procedure .
the values of the critical frequencies of the kp model have been divided by 4/9 for convenience , for the computation of ages ( see next subsection ) .
typical errors of the break frequencies are of the order of 400 mhz .
we note that in the above analysis , the energy losses due to expansion are not considered
. their effect would be to increase the effect of synchrotron losses , without modifying the shape of the spectrum . in the case of constant expansion rates , the values of the break frequencies would be shifted by the same amount and the differential behaviour of @xmath42 along the sources would be unchanged .
however , since low brightness lobes in tailed radio galaxies are likely to be confined by the outer medium ( see e.g. feretti et al .
1992 ) , expansion losses should be negligible .
the results of our spectral fits do not allow us to argue in favour of either the kp or jp model .
they are discussed for each source below .
* ic310 * - the low - frequency spectral index of this source up to 2@xmath17 from the core is @[email protected] .
since we do not expect any effect of self - absorption or ageing in this region we can assume that this spectrum reflects the original energy distribution of the radiating electrons .
this implies an index @xmath41 = 1.8 .
the data are well fit with both the kp and the jp models .
the break frequencies are given in table 2 .
* ngc1265 * - we fitted the spectra of this source at different distances from the core using flux densities at 4 frequencies ( see fig . 5 ) .
the low - frequency spectrum within a few arcmin from the nucleus has a spectral index in the range of @xmath43 , therefore only a range for the parameter @xmath41 can be inferred .
we have used @xmath41 = 1.8 , 2.0 , and 2.2 .
the value of 1.8 gives a bad fit with both models .
fits with @xmath41 = 2.0 and @xmath41 = 2.2 yield good results to both the kp and jp model , but @xmath44 is lower for @xmath41 = 2.2 ( table 3 ) .
the jp model gives a slightly better fit up to 5@xmath17 from the core , while the kp model seems more suitable to describe the steepening in the most external region .
* 3c129 * - no information is available for this source about the low - frequency spectrum , therefore we fitted our data assuming 4 values of @xmath41 : 1.8 , 2.0 , 2.2 , 2.4 . the best agreement with the data and the jp and kp models
have been obtained using @xmath41 = 2.4 , which corresponds to an intrinsic spectral index of @xmath45 .
the values of the derived break frequencies are given in table 4 .
the kp model gives a slightly better fit for the points within 7@xmath17 from the core but it is unable to fit the very steep spectrum of the more external points where the jp model fit is significantly better . * 3c465 * - in the innermost source region , the spectral index @xmath46 is 0.6 , therefore the electron energy index @xmath41 is likely to be @xmath47 2.2 .
we fitted the spectra using @xmath41 = 2.0 and 2.2 , and found that the last value gives a better fit ( see table 5 ) . in the innermost points ,
both the kp and jp models provide a good fit to the spectral data , while at larger distances from the core , the kp model is increasingly better . in the outermost points of the n - s tail ,
no satisfactory fit to the spectra was found , unless using a different electron energy index .
this implies that a simple model of synchrotron emission can not account for the spectral flattening at high frequency .
synchrotron radiation losses lead to a steepening of the emission spectrum at high frequencies with time . from the value of the critical frequency @xmath28 and of the magnetic field , it is possible to get information on the lifetime of radiating electrons suffering synchrotron and inverse compton losses against the 3k - background radiation . as the magnetic field strength , we use the equipartition value estimated at different locations along each source .
the calculation of the equipartition magnetic field involves a number of assumptions ( pacholczyk 1970 ) .
we assume the same volume for relativistic particles and magnetic field ( filling factor @xmath48 = 1 ) , the same energy in electrons and protons ( k = 1 ) , and upper and lower cutoff frequencies of 10 mhz and 100 ghz , respectively .
the derived values of h@xmath29 for each source are given in the corresponding table .
the previous equation is valid for the jp model , where a continuous isotropization of the pitch angle is allowed with time . in the kp model , where particles will maintain their initial pitch - angle over time ,
the expression for the critical frequency differs from that valid in the jp model by @xmath36sin@xmath51 , i.e. 4/9 for an isotropic pitch - angle distribution .
therefore , the right - hand side of eq .
( 1 ) must be divided by 4/9 when using break frequencies obtained in the kp model .
to use eq .
( 1 ) also for the kp model , the kp break frequencies in tables 2 to 5 have been divided by 4/9 .
the derived ages at different distances from the core for each source are given in tables 2 to 5 for the two models and are plotted in fig .
typical errors in lifetimes are of the order of 5% . from the age , we obtained the projected drift velocity of emitting particles at different distances from the core .
these are also reported in tables 2 to 5 .
* ic310 * seems to be old , since the electron lifetimes in the outer regions appear to be 22.5 @xmath52 yrs .
this is a bit surprising as the source is smaller than the others studied here .
this could suggest the existence of strong projection effects , but this is in contrast with the low velocity ( 180 km s@xmath15 ) of ic310 with respect to the perseus cluster .
we note that the radio structure shows no evidence of any transverse expansion .
this suggests that the source is well confined by the external medium and that electrons have not suffered significant expansion losses , and can thus survive for such a long time .
the projected velocity of radiating electrons decreases along the tail from @xmath24900 km / s to @xmath24200 km / s .
projection effects would increase these values .
miley ( 1973 ) pointed out that the radio plasma of ic310 , being @xmath2410 times less luminous than ngc1265 , may have been ejected with less kinetic energy .
this may comply with the relatively low drift velocity inferred for its tail .
* ngc1265 * reaches an age of 6.7 @xmath53 yr .
the average electron velocity is @xmath243000 km / s , and increases with distance , up to 7@xmath17 from the core .
the radial velocity of this galaxy with respect to the perseus cluster is 2055 km s@xmath15 .
we note that odea & owen ( 1987 ) argued , based on the magnetic field in the jets , that ngc1265 should be within 45@xmath5 of the plane of the sky .
if this is the case the total galaxy velocity would be consistent with the average value estimated from spectral aging . * 3c129 * is @xmath54 yrs old in the more external regions .
we remember that for the outer regions of this source only the jp model is able to fit the spectral index data .
the particle velocity is around 2000 km / s , showing variations along the tail .
it seems that there is an increase of velocity at about 6@xmath177@xmath17 from the core , where a large bend in the radio structure occurs . in * 3c465
* the particle lifetimes in the e - w lobe reach 66.5 @xmath53 yrs and the average velocity of emitting electrons is @xmath243500 km / s
. the velocity decreases at 4@xmath17 6@xmath17 from the core , then it increases in the outermost source regions . in the n - s tail ,
the points where the break frequencies have been estimated are too few to derive the trend of electron age and velocity .
the velocity implied at 4@xmath17 from the nucleus is about 6500 km s@xmath15 .
the ages obtained for the present galaxies are of the same order as in other objects for which similar studies were made , such as 3c31 , 3c449 ( andernach et al .
1992 ) , and ngc4869 ( feretti et al . 1990 ) .
the relativistic electron velocities are higher than expected , as we would expect that the particles diffuse at the typical sound speed ( some hundred km s@xmath15 ) or that their motion reflects the proper motion of the parent galaxy within the cluster , as in ngc1265 .
we note that there could be a selection effect , in that tailed radio sources with very long tails may be those associated with cluster galaxies with high proper motion .
however , in all cases we detect significant local variations , with values up to 50006000 km s@xmath15 , at large distance from the core .
the inferred trends could be explained by : a ) projection effects , b ) the existence of reacceleration processes , which allow the electrons to extend their lifetimes , c ) the presence of bulk streaming motions in the radio - emitting plasma . the existence of bulk streaming motions in radio galaxies has been suggested by liu et al .
( 1989 ) , on the basis of similar electron ageing arguments .
we have also considered the possibility that the magnetic field , which enters in the computation of the age , is not the equipartition value .
we have noted , however , that the equipartition magnetic fields in these sources are always close to the values which give the maximum lifetimes ( van der laan & perola 1969 ) , therefore the use of significantly different values of the magnetic field would make the velocities still higher .
we have determined the linear polarization in all sources at 10.6 ghz , and at 4.8 ghz , and 2.7 ghz ( when available ) .
the degree of polarization generally increases along the tail , reaching values larger than 50% at 10.6 ghz . towards the head ,
the degrees of polarization is lower , owing to beam depolarization within the regions where the twin jets ( and hence the magnetic fields tied to them ) get strongly bent backwards .
for the same reason , the measured fractional polarization can also be significantly lower at the longer wavelengths .
synchrotron radiation theory predicts that the intrinsic polarization percentage is given by @xmath55 where @xmath26 is the observed emission spectral index . at the tail s
end the spectral index typically increases to @xmath26@xmath242 ; then the intrinsic fractional polarization expected there is @xmath2480 % . the observed polarization percentage is related to the intrinsic theoretical one by the expression @xmath56 where h@xmath57 and h@xmath58 are the ordered and random component of the magnetic field , respectively .
the high linear polarization percentage seen at 10.6 ghz implies a high , though not complete , degree of ordering of the magnetic field .
the increase of the polarization percentage with distance from the core is much stronger than that expected from the spectral index variation ( eq . 2 ) .
therefore , the degree of ordering of the magnetic field is higher as the radio emitting plasma is getting older . in 3c129 , the degree of polarization at the tail s end is close to the theoretical value .
we derive that 91% of the magnetic field in this region is ordered on a scale of @xmath2420 kpc . at @xmath245@xmath17 from the core ,
the ordered component of the magnetic field is @xmath2465% . in the other sources , the uniform component of the magnetic field is @xmath247075% at the tails ends on a 2030 kpc scale .
if the increase in the ordering of the magnetic field is due to reconnection of the magnetic field ( e.g. soker & sarazin 1990 ) , then this would be accompanied by energy release which could also be responsible for reaccelerating the electrons . in all sources , the magnetic field
is aligned with the tails .
this alignment has been interpreted as due to the anisotropic effect of ram pressure ( pacholczyk & scott 1976 ) and is consistent with the increase of magnetic field ordering along the tails .
we have presented a multi - frequency study of 4 tailed radio galaxies .
the results are summarized in the following .
point - to - point spectra using 3 or 4 frequencies , in the range from 327 mhz to 10.6 ghz , have been obtained along the tails , up to the outermost regions .
we find that the spectrum in each radio galaxy progressively steepens with distance from the core , and at each location it is steeper at higher frequencies . except for the n - s tail of 3c465 , the spectra
are well fitted by models involving synchrotron energy losses , assuming both that the electron pitch angle is constant over time ( kp model ) , and that there is continuous isotropization of the pitch - angle distribution ( jp model ) .
the spectral fits do not allow us to argue in favour of either the kp or jp model .
the critical frequency obtained by spectral fits decreases with increasing distance from the core . assuming that the magnetic field is at the equipartition value , we show that the lifetimes of radiating electrons after synchrotron and inverse compton losses are generally between 10@xmath33 and 10@xmath59 yr , with the exception of ic310 where the electrons in the outer tail are older than 10@xmath59 yr .
electron drift velocities are generally larger than 2000 km / s , and show significant variations along the tails .
this implies that reacceleration processes or bulk motions should be present .
the polarization data at 10.6 ghz indicate that in all the sources the magnetic fields are oriented along the tails .
the polarization percentage increases along the tails , reaching values of @xmath6050% . in 3c129
the polarization percentage in the outermost regions is @xmath2475% , close to the theoretical synchrotron value .
the degree of ordering of the magnetic field increases along the tails , viz .
up to @xmath2475% in ic310 , ngc1265 , and
3c465 . in 3c129 the ordered component of the magnetic field exceeds 90% at the tail s end .
we thank matteo murgia for his help in the application of the code for the spectral fits .
we are grateful to the referee , chris odea , for helpful suggestions .
part of this work was supported by the deutsche forschungsgemeinschaft , grant kl533/4 - 2 and by european commission , tmr programme , research network contract erbfmrxct97 - 0034 `` ceres '' .
alexander p. , leahy j.p .
, 1987 , mnras 225 , 1 andernach h. , feretti l. , giovannini g. et al .
1992 , a&as 93 , 331 baars j.w.m .
, genzel r. , pauliny - toth i.i.k . ,
witzel a. , 1977 , a&a 61 , 99 burns j.o . ,
sulkanen m.e . ,
gisler g.r . ,
perley r.a . , 1992 , apj 388 , l49 carilli c.l . , perley r.a .
, dreher j.w . , leahy j.p . , 1991 , apj 383 , 554 eilek j.a . , burns j.o .
, odea c.p .
, owen f.n . , 1984 ,
apj 278 , 37 eilek , j.a . ,
arendt , p.n .
, 1996 , apj 457 , 105 emerson d.t .
, klein u. , haslam c.g.t .
, 1979 , a&a 76 , 92 feretti l. , dallacasa d. , giovannini g. , venturi t. , 1990 , a&a 232 , 337 feretti , l. , perola , g.c . , fanti , r. , 1992 , a&a 265 , 9 gisler g.r .
, miley g.k . , 1979 , a&a 76 , 109 jaffe w.j . ,
perola g.c . , 1973 ,
a&a 26 , 423 jgers , w.j . , 1987 , a&as 71 , 603 leahy , j.p .
, 1984 , mnras 208 , 323 kardashev n.s . , 1962 ,
sov . astron .
aj 6 , 317 klein , u. , mack , k .- h . , 1995 , in _ multifeed systems for radio telescopes _ , d.t .
emerson & j.m .
payne eds . , asp conference series , vol .
75 , p. 318
liu r. , riley j. , warner p. , pooley g. , alexander p. , 1989
, mnras 240 , 501 mack k .- h .
, feretti l. , giovannini g. , klein u. , 1993 , a&a 280 , 63 miley g.k . , 1973 ,
a&a 26 , 413 murgia m. , fanti r. , 1996 , internal technical report ira 228/96 odea , c.p . , owen , f.n . , 1987 , apj 316 , 95 pacholczyk a.g . , 1970 , radio astrophysics , freeman , san francisco pacholczyk a.g .
, scott j.c . ,
1976 , apj 203 , 313 rudnick l. , burns j.o .
, 1981 , apj 246 , l69 sijbring l.g . , 1993 , ph.d .
thesis , univ .
groningen soker , n. , sarazin , c.l . , 1990 , apj 348 , 73 van breugel w. , 1982 , a&a 110 , 225 van breugel w. , jgers w. , 1982 , a&as 49 , 529 van der laan , h. , perola , g.c . , 1969 , a&a 3 , 468 wiita , p.j . , gopal - krishna , 1990 , apj 353 , 476 | high - frequency observations of the tailed radio galaxies ic310 , ngc1265 , 3c129 , and 3c465 have been performed with the effelsberg 100-m telescope . for the radio galaxies
ic310 ,
ngc1265 and 3c465 , radio data obtained at low frequencies with the westerbork synthesis radio telescope are also available .
these new radio data allow us to map the extended structure of the radio galaxies and obtain spectral and polarization information in the outermost regions .
the multi - frequency spectra were used to study the synchrotron ageing of relativistic electrons with increasing distance from the active nucleus .
we found that the spectrum in each radio galaxy progressively steepens with distance , and at each location it is steeper at high frequencies .
the spectra are fitted by models involving synchrotron energy losses and the critical frequency is obtained at increasing distance from the core . assuming that the magnetic field is the equipartition value , we obtain the radiating electron lifetimes and consequently their drift velocities .
our results imply the existence of reacceleration processes or bulk motions along the tails . the polarization data at 10.6 ghz give information on the intrinsic degree of polarized flux and the orientation of the magnetic field .
we find that the polarization percentage increases along the tails , reflecting an increase of the degree of ordering of the magnetic field .
the magnetic field in the tails is longitudinal .
27a 40l 0 1 |
predicting the macroscopic properties of composite or porous materials with random microstructures is an important problem in a range of fields @xcite .
there now exist large - scale computational methods for calculating the properties of composites given a digital representation of their microstructure ; eg .
permeability @xcite , conductivity @xcite and elastic moduli @xcite .
a critical problem is obtaining an accurate three - dimensional ( 3d ) description of this microstructure @xcite .
for particular materials it may be possible to simulate microstructure formation from first principles .
generally this relies on detailed knowledge of the physics and chemistry of the system , with accurate modeling of each material requiring a significant amount of research .
three - dimensional models have also been directly reconstructed from samples by combining digitized serial sections obtained by scanning electron microscopy @xcite , or using the relatively new technique of x - ray microtomography @xcite . in the absence of sophisticated experimental facilities , or a sufficiently detailed description of the microstructure formation ( for computer simulation ) ,
a third alternative is to employ a statistical model of the microstructure .
this procedure has been termed `` statistical reconstruction '' since the statistical properties of the model are matched to those of a two - dimensional ( 2d ) image @xcite .
statistical reconstruction is a promising method of producing 3d models , but there remain outstanding theoretical questions regarding its application .
first , what is the most appropriate statistical information ( in a 2d image ) for reconstructing a 3d image , and second , is this information sufficient to produce a useful model ? in this paper we address these questions , and test the method against experimental data . modeling a composite and numerically estimating its macroscopic properties is a complex procedure .
this could be avoided if accurate analytical structure - property relations could be theoretically or empirically obtained .
many studies have focussed on this problem @xcite . in general , the results are reasonable for a particular class of composites or porous media . the self - consistent ( or effective medium )
method of hill @xcite and budiansky @xcite and its generalization by christensen and lo @xcite is one of the most common for particulate media @xcite .
no analogous results are available for non - particulate composites .
a promising alternative to direct property prediction has been the development of analytical rigorous bounds ( reviewed by willis @xcite , hashin @xcite and torquato @xcite ) .
there is a whole hierarchy of these bounds , each set tighter than the next , but depending on higher and higher order correlation functions of the microstructure .
the original hashin and shtrikman @xcite bounds that have been widely used by experimentalists implicitly depend on the two - point correlation function of the microstructure , although the only quantities appearing in the formulas are the individual properties of each phase and their volume fractions . to go beyond these bounds to higher - order
, more restrictive ( i.e. , narrower ) bounds , it is necessary that detailed information be known about the composite in the form of three - point or higher statistical correlation functions @xcite , which do appear explicitly in the relevant formulas .
evaluation of even the three point function is a formidable task , so use of these bounds has in the past been restricted to composites with spherical inclusions .
it is now possible to evaluate the bounds for non - particulate composites @xcite , and it is interesting to compare the results with experimental and numerical data . if the properties of each phase are not too dissimilar the bounds are quite restrictive and can be used for predictive purposes @xcite .
sometimes experimental properties closely follow one or the other of the bounds , so that the upper or lower bound often provides a reasonable prediction of the actual property even when the phases have very different properties @xcite .
it is useful to test this observation . in this study
we test a generalized version @xcite of quiblier s @xcite statistical reconstruction procedure on a well - characterized silver - tungsten composite .
computational estimates of the young s moduli are compared to experimental measurements .
the composite is bi - continuous ( both phases are macroscopically connected ) and therefore has a non - particulate character .
as such the microstructure is broadly representative of that observed in open - cell foams ( such as aerogels ) , polymer blends , porous rocks , and cement - based materials . by comparing our computations of the moduli to the results of the self - consistent method we can test its utility for non - particulate media .
an advantage of the reconstruction procedure we use is that it provides the statistical correlation functions necessary for evaluating the three - point bounds .
comparison of the young s modulus to the bounds therefore allows us to determine the bounds range of application for predictive purposes .
the two basic models we employ to describe two - phase composite microstructure are the overlapping sphere model and the level - cut gaussian random field ( grf ) model . in this section
we review the statistical properties of these models which are useful for reconstructing composites .
the simplest , and most common , quantities used to characterize random microstructure are @xmath3 , the volume fraction of phase 1 , @xmath4 , the surface area to total volume ratio and @xmath5 , the two - point correlation function ( or @xmath6/[p - p^2]$ ] the auto - correlation function ) .
@xmath5 represents the probability that two points a distance @xmath7 apart lie in phase 1 .
here we only consider isotropic materials where @xmath8 does not depend on direction . also note that @xmath9 and @xmath10 .
realizations of the overlapping sphere model @xcite are generated by randomly placing spheres ( of radii @xmath11 ) into a matrix .
the correlation function of the phase exterior to the spheres ( fraction @xmath3 ) is @xmath12 for @xmath13 and @xmath14 for @xmath15 where @xmath16 and the surface to volume ratio is @xmath17 . with modification
it is also possible to incorporate poly - dispersed and/or hollow spheres .
the overlapping sphere model is the most well characterized of a wider class called boolean models , which have been recently reviewed by stoyan _
et al . _
@xcite .
the internal interfaces of a different class of composites can be modeled by the iso - surfaces ( or level - cuts ) of a stationary correlated gaussian random field ( grf ) @xmath18 ( so called because the value of the field at randomly chosen points in space is gaussian distributed ) . moreover ,
if @xmath19 is fixed , the distribution over an ensemble will also be gaussian .
correlations in the field are governed by the field - field correlation function @xmath20 which can be specified subject to certain constraints [ @xmath21 , @xmath22 .
invariably @xmath23 is taken as unity .
a useful general form for @xmath24 is @xcite @xmath25 the resulting field is characterized by a correlation length @xmath26 , domain scale @xmath27 and a cut - off scale @xmath28 .
the cut - off scale is necessary to ensure @xmath29 as @xmath30 ; fractal iso - surfaces are generated if @xmath31 .
there are many algorithmic methods of generating random fields .
a straight forward method is to sum @xmath32 ( @xmath331000 ) sinusoids with random phase and wave - vector @xmath34 where @xmath35 is a uniform deviate on @xmath36 and @xmath37 is uniformly distributed on a unit sphere .
the magnitude of the wave vectors @xmath38 are distributed on @xmath39 with a probability ( spectral ) density @xmath40 .
the density is related to @xmath41 by a fourier transform [ @xmath41=@xmath42 .
note that @xmath43 specifies an additional constraint on @xmath41 .
although this formulation of a grf is intuitive , the fast fourier transform method is more efficient @xcite . following berk @xcite one
can define a composite with phase 1 occupying the region in space where @xmath44 and phase 2 occupying the remainder .
the statistics of the material are completely determined by the specification of the level - cut parameters and the function @xmath41 ( or @xmath40 ) .
the volume fraction of phase 1 is @xmath45 berk @xcite and teubner @xcite have shown that the two point correlation function is @xmath46 where @xmath47.\end{aligned}\ ] ] the auxiliary variables @xmath48 and @xmath49 are needed below .
the singularity at @xmath50 can be removed with the substitution @xmath51 .
the specific surface is @xmath52 where @xmath53 with @xmath24 given by eqn .
( [ modelg ] ) . many more models ( for which @xmath5 can be simply evaluated ) can be formed from the intersection and union sets of the overlapping sphere and level - cut grf models . here
we define a few representative models which have been shown to be applicable to composite and porous media .
a normal model ( n ) corresponds to berk s formulation .
models can also be formed from the intersection ( i ) and union ( u ) of two statistically identical level - cut grf s .
another model , i@xmath54 , formed from the intersection of @xmath55 primary models , has also been found useful .
the statistical properties ( @xmath3 , @xmath4 and @xmath8 ) of each model are given in terms of the properties of berk s model [ eqns .
( [ gamma ] ) , ( [ h2 ] ) and ( [ hdash ] ) ] in table [ tabmods ] @xcite . .
the volume fraction @xmath3 , two - point correlation function @xmath56 and surface to volume ratio @xmath4 of models
n , i , u and i@xmath54 in terms of the properties [ @xmath48 , @xmath49 and @xmath57 of berk s two - level cut gaussian random field model . the formula @xmath58 is used for calculating the level - cut parameters ( see table [ tabab ] ) .
[ tabmods ] [ cols="^,^,^,^,^",options="header " , ] for the purposes of computing the elastic properties of the model we maintain the length scale parameters ( @xmath26 , @xmath28 and @xmath27 ) and alter the level cut parameters ( @xmath59 and @xmath60 ) of the model such that @xmath61=20% ( in accord with the experimental composite ) .
the young s modulus , computed using the finite element method , is compared with the experimental data of umekawa _ et al . _ in fig .
[ cf_young ] .
for the temperature region below the melting point of silver the maximum error is 4% , a very good result . above the melting point of silver , when the silver phase is taken to have a zero bulk and shear modulus , the error is only 3% .
the agreement may actually be better than that , however .
since the elastic measurements were dynamic measurements , the liquid silver can be considered as being trapped on the time scale of the experimental measurement , before any significant flow could take place , and so could still contribute to the effective moduli via its non - zero liquid bulk modulus . just before melting
, the silver had a bulk modulus of about 35gpa .
if we take the bulk modulus to be somewhat lower , in analogy to the ice - water difference around 0@xmath1c , then a bulk modulus of 23.1gpa causes the n ( @xmath62=0 ) model to agree perfectly with experiment at temperature points above the melting point of silver . the bounds are also shown in fig .
[ cf_young ] . for model n ( @xmath62=0 )
the microstructure parameters are @xmath63 and @xmath64 .
the results bound the experimental data and provide a reasonable prediction of the young s modulus below the melting point of silver . note that even if the silver phase is given a non - zero bulk modulus past its melting point , the zero shear modulus causes the lower bounds for shear modulus and therefore young s modulus to be identically zero .
unfortunately , there was no reported poisson s ratio results for the composite , so we can not compare to the model results for this quantity .
we now compare the different analytical predictions of effective moduli with the finite element data .
we have chosen to study the two most common models and only consider the moduli appropriate for the w - ag composite studied above .
the results are shown in fig . [
cf_thy](a ) for overlapping spheres and in fig .
[ cf_thy](b ) for the single level - cut grf or excursion set of quiblier [ model n(@xmath62=0 ) ] .
the self - consistent method provides a very good estimate of @xmath65 for model n(@xmath62=0 ) , but not for overlapping spheres .
this might be expected because the n(@xmath62=0 ) model is symmetric with respect to phase - interchange ( like the scm ) while the overlapping sphere model is not . as stated above the application of the generalized
scm is difficult because it is not obvious which phase should be chosen as the ` inclusion ' phase . for the overlapping sphere model
the tungsten phase is comprised of spheres ( at 80% volume fraction ) , so is the more likely choice for the inclusion phase .
nevertheless , we report both estimates ( 80% w inclusions and 20% ag matrix or 20% ag inclusions and 80% w matrix ) for both models . for either choice
, the gscm fails to provide an accurate estimate .
indeed , above the melting point of silver the gscm vanishes for the case 20% ag matrix case since the matrix phase is now completely soft . for the overlapping sphere model the beran , molyneux ,
milton and phan thien bounds are calculated using the microstructure parameters @xmath66 , @xmath67 @xcite .
below the melting point of silver ( where the contrast between the phases is moderate ) the upper bounds provide a very good estimate of the effective moduli . a brief discussion of the effect of elastic contrast is necessary here .
we have already noted that the analytical predictions of effective moduli do not explicitly depend on microstructure , but have a built - in " microstructure .
the elastic contrast , the ratio between the phase moduli , will determine how sensitive the effective moduli actually are to microstructure .
for example , in the case of a two - phase composite having equal shear moduli but different bulk moduli , there is a simple exact formula for the effective bulk modulus which is totally insensitive to microstructure @xcite . in the case of small contrast , the effective moduli can be expressed exactly as a power series in the moduli differences @xcite . up to second order in this difference , at any volume fraction
, the coefficients of the power series are not dependent on anything but the volume fractions and the individual phase properties .
therefore at small contrast , analytical predictions of effective moduli that explicitly depend only on volume fractions and phase moduli should all work well .
( a ) ( b )
throughout this paper , we have treated the finite element computation method as being perfectly accurate , so that comparisons of elastic results to experiment were solely a test of how well the reconstructed microstructure compared to the real microstructure .
this is not exactly true , since there are numerical errors in the finite element method @xcite .
these are small , however , and are generally of about the same size or less than the differences seen between model computations and experimental data for the elastic moduli .
there are also statistical sampling errors associated with the finite size ( @xmath6840@xmath69 m ) of the models we employ to estimate the elastic properties .
since this is much greater than the correlation length of the samples ( @xmath685@xmath69 m - see fig .
[ cf_p2 ] ) we again assume these errors to be small . therefore , the good agreement between model prediction and experimental data seen in this paper is good evidence that the model considered is indeed capturing the main aspects of the experimental microstructure .
we have compared various theoretical results to finite element computations of the effective young s modulus @xmath65 for non - particulate media : a w - ag composite and two model media ( overlapping spheres and a single - cut gaussian random field ) .
the generalized self - consistent method ( derived for particulate composites ) did not provide a good estimate of @xmath65 for the bi - continuous materials considered here . the standard self - consistent method provided a good estimate for the single - cut grf and w - ag composite .
since the method predicts zero moduli for porosity above 50% but the solid phase of the single - cut grf remains connected up to porosities of around 90% @xcite such agreement can not be general .
upper bounds , calculated using three - point statistical correlation functions , provided a good prediction at low contrast ( @xmath70/@xmath71 6 ) for each composite .
when one of the phases was completely soft the bounds lost predictive value . therefore , for general composites , it is important to employ numerical computations of the effective moduli . for accurate numerical prediction of composite properties
it is important that a realistic model be used .
model - based statistical reconstruction , based on the joshi - quiblier - adler approach , appears to be a viable route for microstructural simulation .
however , it is important that the models underlying the procedure be capable of mimicking the composite microstructure .
we have shown how several different models can be employed to find a useful reconstruction .
thanks the australian - american educational foundation ( fulbright commission ) for financial support and the department of civil engineering and operations research at princeton university where this work was completed .
we also thank the partnership for high - performance concrete program of the national institute of standards and technology for partial support of this work . | we statistically reconstruct a three - dimensional model of a tungsten - silver composite from an experimental two - dimensional image .
the effective young s modulus ( @xmath0 ) of the model is computed in the temperature range 251060@xmath1c using a finite element method .
the results are in good agreement with experimental data . as a test case
, we have reconstructed the microstructure and computed the moduli of the overlapping sphere model .
the reconstructed and overlapping sphere models are examples of bi - continuous ( non - particulate ) media .
the computed moduli of the models are not generally in good agreement with the predictions of the self - consistent method .
we have also evaluated three - point variational bounds on the young s moduli of the models using the results of beran , molyneux , milton and phan thien .
the measured data were close to the upper bound if the properties of the two phases were similar ( @xmath2 ) .
= -27.0 mm = -30 mm * keywords : * structure - property relationships ; microstructures ( a ) ; inhomogeneous material ( b ) ; finite elements ( c ) ; probability and statistics ( c ) . |
the 2800 and [ oii]3727 emission line luminosity densities have decreased by a factor of @xmath4 10 from z = 1 to the present day ( lilly et al .
1996 ; hammer et al .
this has led madau et al .
( 1996 , 1998 ) to suggest that the cosmic star formation density has decreased by the same factor within that redshift interval , and that most of the stars seen now were formed during the first half of the universe s existence .
the uv emission from galaxies is produced by a complex mix of short and moderately long - lived stars , the latter ( late b and a0 stars ) contributing more at longer uv wavelengths .
even old stars in luminous early - type and quiescent galaxies can contribute to the observed uv luminosity density .
however the most important uncertainty in estimating the star formation density from the uv luminosity density is due to the extinction which can show large variations from one galaxy to another .
for example in iras star - forming galaxies , most of the energy is reemitted at far - ir ( fir ) wavelengths and these objects are either missed or their star formation rates are severely underestimated when derived by uv measurements .
this is why it is often thought that the uv luminosity density is likely to provide only a lower limit to the actual star formation density .
the situation is complicated further by the expected contamination by agn to the uv light density .
+ in an attempt to better estimate the cosmic star formation density , tresse and maddox ( 1998 ) have calculated the extinction - corrected @xmath5 luminosity density at z @xmath4 0.2 .
their result is in agreement with the uv ( 2800 ) at z = 0.35 ( lilly et al , 1996 ) if an extinction of 1 mag is assumed for the uv continuum .
a preliminary study of more distant galaxies indicates that the situation might be similar at z@xmath4 1 ( glazebrook et al , 1998 ) , but it is limited by the difficulty of measuring the near - ir redshifted @xmath5 line of faint galaxies with 4 m telescopes .
+ multi - wavelength analyses can provide , in principle , a detailed budget of the energy output in each wavelength range for the entire galaxy energy distribution .
it has been shown for local galaxies that fir luminosities are tightly correlated with radio luminosities ( helou et al .
1987 ; condon 1992 ) , and that bolometric luminosities are most closely proportional to 12 @xmath0 m luminosities ( spinoglio and malkan , 1989 ; spinoglio et al , 1995 ) .
these trends hold over a wide range of galaxy luminosities , despite the large variety of galaxy energy distributions .
only agns which are believed to be associated to supermassive black holes , appear not to follow those relations ( condon et al .
+ recent observational advances allow now to study distant galaxies from the uv to the radio , sampling a wavelength range which covers most of the domain where their energy is being emitted .
vla deep surveys are able to detect sources down to 10 @xmath0jy ( e.g. the 5ghz surveys of fomalont et al .
1992 and richards et al . 1998 ) and isocam ( csarky _ et al .
_ 1996 ) aboard the infrared space observatory ( iso , kessler _
1996 ) can reach detection limits of 100@xmath0jy at 15@xmath0 m ( elbaz et al .
1998 ) . in the range 60200@xmath0 m
the detection limits are 0.2 jy at 60@xmath0 m from the iras faint source catalog ( moshir et al .
1989 ) and 0.1 jy at 175@xmath0 m from the firback survey carried out with iso ( clements et al . 1998
; puget et al . 1998 ) .
fir detections thus appear to be not sensitive enough to reach the same depth as radio and mid - ir ( mir ) deep surveys .
for example , if we consider a strong and highly reddened starburst ( sbh in the terminology of schmitt et al .
1998 ) , a @xmath6=250 @xmath0jy source would correspond to a 0.009 jy source at 60@xmath0 m , and a z=1 redshifted sbh with @xmath7=16@xmath0jy would have 0.022 jy at 175@xmath0 m .
+ the sensibility and high spatial resolution of isocam allow the study of distant field galaxies at mir wavelengths ( 2 @xmath8 20@xmath0 m ) .
star - forming galaxies and agn are easily detectable in the wavelength range 5 - 18@xmath0 m , even at large distances ( franceschini 1991 ) .
the canada - france redshift survey ( cfrs ) field at 1415 + 52 ( lilly et al .
1995a ) is the second most observed field at all wavelengths after the hubble deep field ( hdf ) . while it does not go as deep
, it is @xmath4 18.5 larger in area and thus is more suited for source statistics when a volume - limited ( z @xmath1 1 ) sample is considered .
it has been observed to very faint magnitudes in the bvik bands ( photometric completeness down to @xmath9=23.5 mag , lilly et al .
1995b ) , possesses spectroscopic data for galaxies brighter than @xmath9 = 22.5 mag from the cfrs , and deep radio observations ( s@xmath10 16 @xmath0jy , fomalont _ et al .
the cfrs sample can be considered complete in the sense that it contains all luminous ( @xmath11 -20.5 ) galaxies in the volume out to z = 1 .
+ this paper presents a major follow - up study of the above cfrs field , by gathering and studying representative samples of galaxies selected at radio and mir wavelengths . with sensitivity limits of 250@xmath0jy at 15@xmath0 m ( isocam ) and 16@xmath0jy at 5ghz ( vla ) , these samples should include all strong and reddened starbursts up to z=1 , with star formation rates larger than 100 and 70 @xmath12 @xmath13 respectively .
these samples should not miss any luminous fir source in the cfrs field as the sources were selected using observations which cover wavelengths on either side of the 60 - 100@xmath0 m bump .
they can thus be used ultimately to estimate the star formation density which has been missed by uv flux measurements .
+ the nature of the @xmath0jy radio sources in the field and of their optical counterparts has been extensively discussed by hammer et al .
recently this field has been imaged by isocam in the lw2 ( 5 - 8.5@xmath0 m ) and lw3 ( 12 - 18@xmath0 m ) filters .
we have presented the 6.75@xmath0 m lw2 observations in a previous paper ( flores et al .
there , we have discussed the details of the data reduction , the astrometry and the confidence level for each source .
fifty four sources with s / n @xmath14 3 were detected with s@xmath15 @xmath14 150 @xmath0jy , 21 of which possess spectra from the cfrs .
of the latter , 7 were stars . among the non - stellar sources , 42% were classified as agn and 50% as @xmath16 galaxies , i.e. star - forming galaxies with a significant population of a stars .
the relatively high fraction of agn is not unexpected because strong agn are generally associated with a hot dust component and have generally bluer near - ir colors than starbursts . while the 6.75@xmath0 m data appears not to be optimal for selecting starbursts , they are useful for constraining galaxy seds , as galaxies ( and especially seyfert 2 galaxies ) show large variations in their mir color properties .
we present here the 15@xmath0 m lw3 observations .
for galaxies in the redshift range 0.25 @xmath1 z @xmath1 1 , this filter samples the 8 - 12@xmath0 rest wavelength region .
it can thus simultaneously provide samples of normal , starburst and active galaxies which are complete to a well - defined bolometric flux limit ( spinoglio et at , 1995 ) .
mir measurements are sensitive to dust thermal emission ( @xmath17 ) , its broad emission being interpreted as due to pahs or unidentified infrared bands ( uibs ) carriers ( @xmath18 ) , and to nuclear non - thermal radiation .
there have been recent suggestions that pahs arise from photo - dissociation regions around hii regions , the latter showing red near - ir continua ( laurent and mirabel , 1998 , in preparation ) .
genzel et al .
( 1998 ) have used the relative strengths of pah features and near - ir colors to discriminate agn dominated objects from starbursts .
lutz et al . ( 1998 ) found that the fraction of agn powered objects is relatively small at moderate ir luminosities ( typically @xmath19 2 @xmath20 ) , but reaches half at higher luminosities .
+ in section 2 , we discuss the data reduction , the astrometry and the construction of the catalogs . in section 3 we present the redshift distribution of the isocam lw3 objects , and their optical properties including their morphologies from hst images .
section 4 presents spectral energy distributions from uv to radio wavelengths and a classification scheme for these and for radio - selected galaxies .
derivations of the uv and ir luminosity densities are presented in section 5 .
section 6 describes the global star formation density at z @xmath1 1 and compares it to previous estimates based on the uv luminosity density .
the cfrs field at 1415 + 52 was mapped with the isocam lw channel ( pfov 6 arcsec pixel ) and the filter lw3 ( 12 - 18@xmath0 m ) .
twelve individual images were obtained using the micro - scanning aot mode ( cam01 ) and resulting in a total integration time of @xmath4 1200 sec pixel@xmath21 .
the micro - scanning mode provides the best spatial resolution by superposition of images .
the same pixel of the sky was placed in different parts of the camera in order to minimize and detect any systematic effects .
the micro - scanning aot technique also allows an accurate flat - field image to be generated and yields a pixel size of 15 in the final integrated image .
the detection and removal of transients and glitches , integration of images , absolute flux calibration , and source detection were carried out using the method described by dsert et al .
this method has been found to be particularly well adapted to our observational strategy , i.e. coadding the twelve images , without redundancy within each image .
special attention was paid to possible error propagation in the flux values .
the photometric accuracy has been discussed by desert _
( 1998 ) . from the stellar energy distributions
, we find the photometry to be accurate to a few percent for sources with s / n @xmath22 10 , and to @xmath450% for sources with s / n = 3 .
+ figure 1 displays the final 15@xmath0 m image of the cfrs 1415 + 52 field .
individual images were carefully registered with each other in order to optimize the image quality of the brightest compact objects ( see flores et al , 1998 ) .
the final image of the whole iso field has a resolution equivalent to a median fwhm@xmath411 ( calculated with daophot under iraf ) .
it can be seen that the noise structure ( figure 1 ) is relatively homogeneous ( the standard deviation is lower than one tenth of the mean ) in @xmath4 85% of the area of the image , except near the edges .
+ point sources are iteratively extracted with a gaussian psf of 9 arcsecond fwhm ; a correction factor of 1.39 ( deduced from a detailed modelling of the effective iso psf , cesarsky _ et al .
_ , 1996 ) is applied to the measured flux ( calibrated with isocam user s manual conversion table ) to account for losses in the wings of the true psf .
source detections were made on the basis of s / n and repeatability in three independent combinations of the 12 individual images ( for details about the source detection repeatability and classification , see f.x .
the repeatability test is based on the redundancy factor , which is the number of times that the sky pixel was seen by different pixels on the camera .
the software built three independent projection subrasters , and for each source candidate the flux and error are measured at the same position in each subrasters . the quality factor is based on flux measurements and varies from the best confidence index (= 4 ) to the worse confidence level (= 0 , see eqs .
7 to 11 of dsert et al.,1998 ) .
we have considered only sources with a confidence level higher than 3 , which means that the final source flux is within 3 @xmath23 ( 2 @xmath23 in the case of 4 ) of the source fluxes in subrasters , where @xmath23 is the error in the final source flux .
+ altogether , 78 sources with s / n @xmath24 3 fulfill these detection criteria ( table 1 ) .
we have considered as secure detections those sources with s / n @xmath24 4 ( 41 sources , catalogs 1 , 2 and 3 ) .
the 37 sources listed in catalogs 4 , 5 and 6 with 3 @xmath25 s / n @xmath25 4 are considered `` less '' secure . that a s / n @xmath14 4 is a good detection criterion is confirmed by studies in the lockman hole deep survey ( csarsky et al .
1998 , in prep ) which show that sources with s / n @xmath26 in individual frames are confirmed in 95% of cases in the final integrated image ( s /
n @xmath22 10 ) .
figure 2 shows the flux distribution and the comparison to results from other fields suggests a completeness down to @xmath27 350 @xmath0jy for sources with s / n @xmath14 4 , and to @xmath4 250 @xmath0jy for sources with s / n @xmath14 3 , at least for the central 85% of the 10x10 field .
these correspond to number densities of 1590 sources per square degree , for @xmath28 350 @xmath0jy , slightly larger than the number density of 1260 source per square degree found by elbaz et al .
( 1998 ) in their isocam survey of the hubble deep field .
+ surveys of low s / n ratio sources can be affected by several biases , the main ones being the possible unreliability of sources near the survey limit and the eddington bias . near the flux density limit , completeness falls off and possible false sources may be introduced .
the fraction of false sources can be controlled through a random match control test based on the identification rate of isocam sources with optical sources ( see next section ) and should be small .
we have checked whether our higher number count relative to the hdf data could be due to the eddington bias .
our survey is intermediate in depth between the lockman hole deep survey ( from 450 to 1400 microjy ) and the hubble deep field survey ( from 125 to 350 microjy ) .
elbaz et al .
( 1998 ) quote a slope of 1.3 and 2.1 for the hdf and lockman hole respectively , and we should expect an intermediate slope near @xmath4 1.7 , lower than the value of 2.2 derived from our s@xmath24250 and s@xmath24350 @xmath0jy counts .
we have attempted to estimate the fraction of sources which may have entered our sample erroneously because statistical uncertainties bring them over the detection threshold , while they actually possess a lower flux density than s=250 microjy .
we have adopted a slope of 1.7 and perform monte carlo simulations of 30 000 sets , assuming a gaussian distribution for the noise of faint sources .
the simulations show that @xmath4 15% of the sources with s / n @xmath24 3 can be sources with fluxes lower than 250 microjy , but that fraction drops to only @xmath4 2% for sources with s
/ n @xmath24 4 .
this is probably close to the truth since after accounting for the eddington bias ( i.e. removing 15% of the 78 sources with s / n @xmath22 3 as well as 2% of the 41 s / n@xmath22 4 sources ) , the slope becomes 1.7 , in good agreement with other surveys .
+ biases in source counts thus affect mainly the faintest sources in our sample ( those with 3 @xmath1 s / n @xmath25 4 ) .
they are the least powerful sources , and as such can not introduce significant uncertainties on global luminosity density estimates .
the positional accuracy of iso sources is affected by both the pixel size and the distortion ( see flores et al , 1998 ) .
the superposition of the isocam field onto the cfrs 1415 + 52 optical field was done by matching the positions of the 6 brightest sources ( which were also detected at 6.75@xmath0 m ) . by comparing the astrometry of the optical and isocam lw3 sources ,
we find a median difference of @xmath43.7 arcsec .
this is reasonable given that the iso pixel size is @xmath46.0 arcsec .
+ the astrometric accuracy of the cfrs 1415 + 52 optical field is 0.15 , based on the comparison between the optical and radio frames ( hammer _ et al .
_ , 1995 ) .
we first compare the 15@xmath0 m image with the vla @xmath0jy radio map ( fomalont et al , 1991 ) , and calculate the probability of a pure coincidence , assuming poisson statistics : + @xmath29 where d is the angular distance between the isocam lw3 source and the radio source in degrees and
n is the integrated density of radiosources with flux @xmath7 ( @xmath30 ) = 83520 @xmath31 ) .
ten isocam lw3 sources are thus identified ( see table 1 ) .
they all possess optical counterparts ( hammer _ et al .
+ astrometry of isocam lw3 sources not detected in radio has been derived by comparison to the optical sources in the cfrs 1415 + 52 field , and @xmath32 is the probability of a pure coincidence , where @xmath33 are integrated counts derived from the cfrs . in table 1 , six catalogs have been defined with different s / n values ( @xmath34 ) and p values ( @xmath35 , and sources without optical counterparts within 12 arcsec ) .
objects which are also radio sources are indicated .
these all have very low probabilities of accidental coincidence .
+ we also compare the 15@xmath0 m map with that described in flores et al .
( 1998 ) at 6.75@xmath0 m .
we found 17 of the 78 15@xmath0 m sources to be also detected at 6.75@xmath0jy ( with flux densities above 150@xmath0jy ) .
the fraction of 15@xmath0 m sources also detected at 6.75@xmath0 m increases with s / n ( see table 2 ) .
+ the final noise structure in isocam images is not strictly gaussian , because of possible residuals glitches , and because the 15@xmath0 m positions could not have been fully corrected for image distortions . to calibrate the probabilities given in table 1 in an empirical way
, we have applied a random match control test to the isocam image by rotating it successively by 45 , 90 , 180 and 270 degrees relative to the optical image .
only 3@xmath361 of the 78 isocam sources were found to be randomly associated with an optical counterpart with @xmath3722.5 and p @xmath25 0.02 .
this should be compared to the 37 @xmath3722.5 and [email protected] counterparts found in table 1 .
these experiments suggest that non - gaussian effects can not affect our probability calculations by more than a factor 2 .
+ table 1 lists the sources in six catalogs , the confidence level decreasing in each successive catalog .
the isocam lw3 sources and their optical counterparts are given in columns ( 1 ) & ( 2 ) .
column ( 3 ) provides the optical source redshift when available , `` star '' indicates that the iso source is stellar , while `` '' indicates that no redshift is available . in columns ( 4 ) , ( 5 ) and
( 6 ) are the i , v and k isophotal magnitudes in the ab system , and `` ''
indicates that no photometry is available .
column ( 7 ) gives the angular distance in arcsec between the isocam lw3 source and optical ( or radio ) identification , while column ( 8) gives the associated probability of coincidence .
columns ( 9 ) and ( 10 ) give the flux density at 1218 @xmath0 m and its error in @xmath0jy .
the positions and flux densities of the 9 non - identified iso sources are also listed in table 1 ( catalogues 3 and 6 ) . among the sources with @xmath38 4 , only 8 ( 19% ) have counterparts fainter than @xmath9 = 22.5 , or no optical counterparts .
these sources could be at redshifts higher than 1 .
the fraction of very faint optical counterparts increases for the fainter isocam lw3 sources ( 41% for sources with 3 @xmath25 s
/ n @xmath25 4 ) , which is to be expected . on the other hand , in the most uncertain catalog ( catalog 5 )
, one can expect @xmath4 4.6 sources out of 19 to be pure coincidental projection . in seven cases ,
more than one optical counterpart appear to be related to an isocam lw3 source . in the following analyses , we have used the `` best '' optical identification ( i.e. , those with lowest coincidental probability ) , but in table 1 , we have listed all possible optical identifications with a probability within a factor two of the smallest probability .
the @xmath39 color distribution of the 15@xmath0 m optical counterparts ( figure 3 ) is significantly redder than that of the cfrs survey ( top panel ) , with a median 0.5 magnitude redder than that of cfrs galaxies ( @xmath40 ) .
twenty six of the sources detected at 15@xmath0 m and with @xmath41 have spectroscopy available in the cfrs database . among these are 4 stars which are also detected at 6.75@xmath0 m .
figure 4 shows the redshift histogram of iso sources superposed on the cfrs redshift distribution .
the median redshift value of the isocam lw3 galaxies ( @xmath42 @xmath4 0.76 ) is higher than that of the cfrs ( @xmath42 @xmath40.58 ) , but coincides with that of @xmath43 16@xmath0jy radio sources ( hammer et al .
since 70% of the 15@xmath0 m optical sources have @xmath4422.5 , we estimate from the redshift distribution that more than 63% of the sources with @xmath28 250@xmath0 jy are at z @xmath1 1 . the spectrophotometric classification of the @xmath45 m optical counterparts is based on optical and emission - line properties using diagnostic diagrams and spectral templates ( see hammer et al .
1995 , 1997 ) .
the following spectrophotometric types are found : fifteen emission - line galaxies , three active galaxies ( including the qso cfrs 14.0198 at z=1.6 , the most distant cfrs object ) , one quiescent galaxy , one spiral galaxy , and one hii galaxy ( cfrs 14.1103 ) .
figure 5 shows the spectra of all extragalactic objects , classified by their spectrophotometric type .
+ most ( 71% ) of the 15@xmath0 m optical counterparts are classified as @xmath16 galaxies , which are galaxies with moderate @xmath46 $ ] emission ( @xmath47 } \sim$ ] 20a ) , and characterized by large d(3550 - 3850 ) indices ( see table 3 ) , as defined in hammer _
large d(3550 - 3850 ) values can be caused by either a large a star population ( d(3550 - 3850 ) is correlated with the @xmath48 equivalent width , d(3550 - 3850)=0.2 corresponding to @xmath49=5 - 7 a for a non - extincted galaxy ) or by very large extinctions . as we shall see , it is the a star hypothesis which is the most plausible .
this suggests that in most extragalactic @xmath45 m sources , there were star formation occurring about 0.5 gyr prior to the observed event .
the importance of the s+a population is also supported by the fact that at high redshift , galaxies with large d(3550 - 3850 ) are also detected at 6.75@xmath0 m and at 1.44ghz . about
30% of the cfrs field at 1415 + 52 has been observed by the hst .
public domain hst images of objects in the cfrs and autifib / low dispersion survey spectrograph survey ( the cfrs / ldss deep survey , brinchmann et al . ,
1997 ) and the groth survey ( groth _ et al . _ , 1994 ) are available . among the 55 @xmath50 iso sources with optical counterparts , 16 ( 30% ) with have been observed with the hst f814w filter .
hst images of optical counterparts of @xmath0jy radio sources can be find in hammer _
( 1996 ) .
figure 6 displays a mosaic of hst images of optical counterparts with @xmath51 22.5 of 15@xmath0 m sources . for each source
the cfrs name is indicated as well as its morphological classification by either brinchmann _ et al . _
( 1997 ) or us .
two sources are unresolved by the hst ( one qso and one hii region , 14.1103 ) .
six galaxies ( 37% ) are found in strongly interacting systems , four ( 25% ) are e / s0 galaxies and four ( 25% ) are disk - dominated galaxies .
most of the 14 resolved sources show irregularities which might be interpreted as pre- or post - merging events .
several groups have been gathering multi - wavelength observations of local galaxies ( spinoglio et al . 1995 ; schmitt et al . 1998 ) . the main limitations of the data sets are the assumed aperture corrections and possible agn variability .
seds of galaxies in the same class often show large differences , especially in the infrared .
this is especially true for seyfert 2 galaxies which show a wide range of colors in the near and far ir ( spinoglio et al .
this large dispersion motivated us to use mean seds of local galaxies to compare with distant sources , so as to avoid biases for or against a given class of galaxies .
for example , some seyfert 2 galaxies may be misclassified as starbursts , but the effect on global quantities would be compensated by true starburst galaxies which are misclassified as seyfert 2 galaxies .
however average properties can be affected by a single object with extreme properties .
for example , in the schmitt et al .
( 1998 ) sample , the average radio luminosities for elliptical and spiral galaxies appears to be overestimated because of the large radio powers of ngc 1316 and of ngc 598 , respectively .
indeed the mean values are significantly ( @xmath4 8 times ) larger than those derived from the much larger and more complete sample of local ellipticals of wrobel and heeschen ( 1991 ) . using the median rather than the mean would provide a value in much closer agreement with wrobel and heeschen , and also provide a sed for spirals which follows the radio - far infrared correlation ( condon 1992 ) . in the following
we will use only median points for determining the sed of local templates .
+ another concern is whether using local templates to analyze the properties of distant sources is appropriate .
we have compared the two samples of local galaxies which have been classified and studied at several wavelengths .
the schmitt et al .
sample has been mainly selected from the iue archives and the spinoglio et al .
sample from the 12@xmath0 m faint iras catalogue .
our 15 micron sample has been selected at the wavelengths 15/(1+z ) microns and 0.835/(1+z ) microns , so that both local samples are a priori appropriate for use as templates for our sources .
the two local samples show somewhat different properties in their seds , especially concerning the average ir properties of seyfert 2 galaxies .
the schmitt et al .
galaxies are bright infrared galaxies and include most of the starburst and agn templates used by genzel et al .
( 1998 ) to compare with ultraluminous iras galaxies .
conversely , the spinoglio et al .
sample includes galaxies varying over a 4 to 5 magnitude range of luminosities in the infrared , and their average seds are mostly dominated by fainter galaxies .
figure 7 shows the distribution of luminosities for the vla / isocam / cfrs objects .
rest frame luminosities at 7@xmath0 have been interpolated from broad band observations at 0.835 , 2.2 , 6.75 and 15@xmath0 , and are available for most objects , except those detected only at radio wavelengths . at z @xmath24
0.5,these galaxies have luminosities comparable to those of the schmitt et al .
galaxies , and are 1.5 to 2.5 magnitudes brighter than the mean magnitude of the spinoglio et al . galaxies .
it is reasonable to think that the discrepancies at ir wavelengths between the spinoglio et al . and schmitt et al .
galaxies are mainly due to the large ir luminosity differences .
+ the large dispersions displayed by individual schmitt et al . galaxies in each class ( especially for seyfert 2 galaxies ) , ensure that this local sample is the most appropriate one for comparison with the intrinsically bright distant galaxies . according to schmitt et al.(1997 ) , the uv to optical range constitutes the best discriminator between seyfert 2 and sbh or sbl ( starburst with low extinction ) galaxies .
seyfert 2 galaxies are generally early - type spirals , with uv colors typically bluer than those of spirals ( see e.g. kennicutt 1992 ) .
figure 8 shows the color distribution of the radio-15@xmath0 m samples . in the ( b - k , 2800 -b )
diagram , their colors are reasonably well fitted by bruzual and charlot ( 1995 ) models , except for three objects which are either a qso or a seyfert 1 galaxy .
they display a range of colors which are well fitted by starburst and spiral templates .
the same conclusion can be drawn from the ( b-7@xmath0 m , 2800 -b ) diagram , with a few objects having b-7@xmath0 colors consistent with those of seyfert 2 templates .
the latter are likely to be associated with hot dust , and have redder optical - mir colors than starbursts . in summary ,
most of the galaxies detected at 15@xmath0 m have colors from the uv to the mir consistent with those of star - forming objects : starburst and spiral galaxies .
+ seds for the 27 extragalactic 15@xmath0m - radio sources with z @xmath1 1 have been constructed , using the observed fluxes at visible ( b@xmath52 , v@xmath52 and i@xmath52 magnitudes ) , near - ir ( k@xmath52 when available ) , mir ( 6.7@xmath53 m , flores et al , 1998 ; 15@xmath0 m , table 1 ) and radio wavelengths ( 1.4 ghz and 5 ghz from fomalont _ et al .
_ , 1985 ) , and shifted to rest wavelengths .
each sed has then been compared to the average sed of well - known local objects ( a sample of 59 galaxies , schmitt _
et al . _ , 1997 ) .
the latter include e ( ellipticals ) , sp ( spirals ) , sbh ( highly reddened starbursts ) , sbl ( low extinction starbursts ) , seyfert 2 and liners . schmitt _
( 1997 ) also provided standard deviations of the mean , which allows us to optimize the comparison . +
in order to classify the seds , we have performed monte carlo simulations , assuming independent variations of our flux measurements within gaussian error bars . in each of the 5000 monte carlo sets ,
every object sed has been fitted to template seds spanning all the classes defined by schmitt et al , providing a @xmath54 value weighted by the standard deviations of the mean given by those authors .
because galaxies often show properties intermediate between one class and another ( see figure 8) , we also defined a hybrid class which consists of a linear superposition of starburst and spiral sed templates , with the contribution of the spiral sed varying from 10% to 90% in 10% steps .
this hybrid class describes the current mix of old and young stellar populations in a galaxy .
we have not defined other hybrid classes such as the seyfert 2 + starburst class , as schmitt et al .
have found that these two populations have rather similar radio to infrared properties .
27 objects were successfully classified with this technique ( the averaged - weighted @xmath55 is less than or equal to one ) except for seven ( table 3 ) . among the latter , four are unambiguously powerful agn as evidenced from their optical spectra ( see figure 5 and hammer et al , 1995 ) , including a qso ( 14.1303 ) and seyfert 1 galaxies ( 14.0573 , 14.1302 and 14.1567 ) , and one is a hii region ( 14.1103 ) .
the 1 sigma dispersion around the @xmath54 value for each fit gives an idea of the reliability of our classification scheme ( figure 9 ) . among the 27 objects selected either at radio or 15 microns
, we find seven ambiguous cases , generally in the classes between seyfert 2 and s + sbh ( table 3 ) .
table 4 summarizes all the informations derived from their optical spectra and radio observations ( spectral index and imagery ) . for 3 objects , emission lines or radio properties remove the ambiguity of the classification : 14.0779 shows a low ionization spectra and both 14.1028 and 14.1041 have negative radio spectral indexes and radio emission extending much beyond their optical sizes ) .
+ the validity of our classification scheme is illustrated in figure 8 , which emphasizes the relevance of the ( b-7@xmath0 m , 2800a - b ) color - color diagram as a diagnostic diagram for distinguishing starbursts from agn . in that diagram , starbursts lie in a well defined color - color region , while agn - powered sources are around the starburst region . the object with the smallest ( b-7@xmath0
m ) color is 14.9025 which is classified as a liner .
+ five galaxies , all within the 15@xmath0 m sample , are classified as pure sbhs .
they are generally radio sources , and were classified by hammer et al .
( 1995 ) as being star - forming s+a objects .
figure 10 displays their seds on which is superposed the sbh sed from schmitt et al . the interpolated luminosity at the 60@xmath0 m bump and the 60@xmath0 m luminosity derived from the radio - fir correlation ( @xmath56 , franceschini _
1994 ) agree to within 20% on average .
nine objects , seven of which are in the 15@xmath0 m sample , are classified as the superposition of a starburst with a spiral sed ( figure 11 ) .
for all of them but one ( 14.0846 ) , the starburst component dominates , providing more than 70% of the object bolometric luminosity ( see table 3 ) .
only half of these objects are detected at radio wavelengths ( figure 7 ) , so they have likely lower star formation rates than pure sbhs .
+ it has been widely argued that powerful starbursts such as those discovered by iras could contain an agn which can contribute to their infrared luminosities ( sanders et al , 1988 ) .
in addition to the fact that their energy distributions and colors are typical of those of starbursts , it is unlikely that the starbursts described here are significantly contaminated by an agn , because : + -their infrared luminosities ( @xmath57 2 @xmath58 @xmath59 , see table 4 ) are lower than those of the ultra - luminous iras galaxies . local galaxies with those luminosities are mainly star - forming galaxies ( lutz et al , 1998 ) .
+ -their radio spectral indices range from 0.4 to 1 , between the observed frequencies of 1.4 and 5 ghz .
this implies a thermal to non - thermal energy ratio in the rest frame of 0 to 0.2 , in good agreement with the starburst population studied by condon ( 1992 ) .
+ -their radio angular sizes ( @xmath25 2 arcsec ) are always in agreement with their optical sizes , as expected if star formation was distributed over the galaxy .
+ -five of them have z @xmath1 0.7 and their spectra from cfrs have [ oiii]5007 and @xmath60 emission lines ; they all show low ionization spectra , with prominent [ oii]3727 line , and for all but one , no [ oiii]5007 line .
figure 12 shows the sum of the five spectra , which reveals a red spectrum with @xmath463727/h\beta$ ] @xmath4 1 and @xmath615007]/h\beta$ ] @xmath1 0.05 .
these galaxies undoubtedly have emission - line spectra typical of hii regions , not of agn .
+ six objects , three of which are in the 15@xmath0 m sample , are classified as seyfert 2 galaxies and four , two of which are in the 15@xmath0 m sample , are powerful agn ( seyfert 1 and qso ) .
their seds are shown in figure 13 as well as that of the radio liner 14.9025 , on which are superposed schmitt et al .
templates or radio - quiet qso seds from sanders et al ( 1989 ) . +
the two remaining objects are 14.1103 , an hii galaxy , and 14.0820 , an elliptical or a spiral galaxy with moderate star formation activity ( @xmath62=16 ) .
the object 14.1103 ( z=0.21 ) is detected at 15@xmath0 m but not at radio wavelengths , and its sed is consistent with that of a local hii region .
this is in agreement with the absence of an old stellar component as noted by tresse _
( 1993 ) from the very large @xmath63 equivalent width ( @xmath22 2500 ) .
a hst image ( figure 6 ) shows an unresolved object , giving a diameter less than 250 pc .
we are probably witnessing here a compact system undergoing one of its very first bursts of star formation : @xmath46_{3727}$ ] is not detected while @xmath61_{5007}$ ] is very prominent .
this galaxy shows spectral properties similar to those of a primordial galaxy according to tresse _
those authors derived a low heavy - element abundance ( @xmath64 0.05 the solar value ) and a large effective temperature ( t@xmath2250 000k based on the @xmath61_{4363}$]/@xmath61_{5007}$ ] ratio ) .
@xmath65 values have been interpolated from @xmath66 and @xmath67 ( or @xmath68 ) colors , using a grid of models with exponentially decreasing sfr ( @xmath69=1 gyr , bruzual and charlot , 1995 ) and a salpeter imf ( dn(m)/dm=-2.35 over m=0.1 - 100 @xmath12 ) .
these calculations are found to be rather independent of the exact value of @xmath69 , except for very small values ( @xmath70 0.1 gyr ) .
we have calculated the total l@xmath71 ( see table 5 ) in the cfrs 1415 + 52 field .
it is based on 143 galaxies which have been spectroscopically identified and are representative of the 558 @xmath72 galaxies in the field ( see lilly et al , 1995b ) . converted to a luminosity density , this value is in excellent agreement ( to within 6% ) with that found for the whole cfrs field ( lilly et al , 1996 ) .
this is consistent with the poisson error of 8% for the 143 representative galaxies .
infrared ( 8 - 1000@xmath0 m ) luminosities are interpolated from our mir and radio flux measurements , by using for each object the template from schmitt et al .
( 1997 ) with the smallest @xmath54 .
the template gives 5 to 7 points within the wavelength range of interest , and errors can be estimated for each object using our monte carlo simulations .
our quoted errors take into account flux measurements errors ( 13% on average ) as well as uncertainties in our classification scheme ( 5% on average ) .
the interpolated infrared luminosity for each object along with its associated error is given in table 4 .
galaxies detected at 15@xmath0 m and/or at radio wavelengths have infrared luminosities in the range of 5 @xmath73 to 2 @xmath58 @xmath59 ( table 4 ) , comparable to those of typical local starbursts and seyfert 2 galaxies ( genzel et al . 1998 ) .
these should be compared to their 2800 luminosities which range from 4 @xmath74 to 2 @xmath73 @xmath59 and are not very different from those of galaxies not detected at 15@xmath0 m and at radio wavelengths .
thus , taking into account only the uv luminosity of the 15@xmath0 m and radio galaxies which are luminous infrared sources would lead to severe underestimates of their actual star formation rates .
nearly half of the global infrared luminosity of 15@xmath0 m and radio galaxies is coming from 7 sources ( 4 sbhs and 3 seyfert 2 ) .
these are detected at both wavelengths , except for the seyfert 2 galaxy 14.0937 .
estimates of infrared luminosities of these bright galaxies should thus not be affected strongly by source count biases or modelling uncertainties .
+ accounting for the global luminosity is slightly complicated by the current observational status of the data .
hammer et al .
( 1995 ) and recent follow - up work have nearly completed the redshift identifications of the radio @xmath0jy sources . among the 45 @xmath44 22.5 15@xmath0 m sources which are not detected at radio wavelengths ,
only 16 possess a redshift from the cfrs .
the latter have not been selected _ a priori _ for their infrared properties , and can be considered as representative of the whole sample . in the calculation of the global luminosity ,
we have assumed two different scaling factors : 1 for luminous galaxies detected at radio wavelengths , and 45/16 = 2.8 for the less luminous galaxies undetected in radio , with an additional statistical error of 25% .
table 5 presents the different values of the global luminosities at 2800 and in the infrared , after excluding powerful agn ( qso and seyfert 1 ) . at ir wavelengths , seyfert 2
contribute to more than a third of the global luminosity .
we consider here the subsample of the 49 galaxies which possess a redshift , are not detected at 15 @xmath0 m or radio wavelengths , and have been observed at k. this subsample can be taken as representative of the 489 @xmath51 22.5 galaxies not detected at 15@xmath0 m or at radio wavelengths in the cfrs 1515 + 52 field .
for these galaxies , we use bvik photometry and their 15 @xmath0 m and radio upper limits to investigate their seds .
we found that 19 of them have seds consistent with those of local template seds for sbh galaxies and 16 with local template seds for sbl galaxies .
the relatively large fraction of sbls ( starbursts with low extinction ) is compatible with the fact that they are not detected in the mir range .
other objects have their seds consistent with those of local template seds for e / s galaxies ( 9 objects ) or agn galaxies ( 5 objects ) .
the fraction of agn is in reasonable agreement with hammer et al .
( 1997 ) who find that @xmath4 8% of all galaxies in the field at z @xmath4 0.5 , are seyfert 2 galaxies as indicated by their emission line ratios .
figure 14 displays four examples of fits for these objects . all objects but
four have been reasonably classified by this method .
the 4 remaining objects have been finally classified as starbursts .
+ the luminosity density of star - forming galaxies may be simply estimated by summing up the contribution of all starburst galaxies ( sbh , sbl and s+sbh ) identified in the field .
uncertainties which include errors in source counts , flux errors and uncertainties in the identification , would come from our monte - carlo simulations .
+ however several complications could affect the validity of such a calculation . first , the origin of the uv and infrared light from seyfert 2 galaxies is unclear , and can be attributed either to the agn or to star formation or to both . indeed seyfert 2 galaxies show a wide range in their infrared properties , and few color diagrams are able to distinguish them from starbursts .
contamination of the luminosity density by agn is potentially the most serious problem for any estimate of the star formation density .
second , cirrus can contribute to the infrared luminosity , and this is likely the case for the objects classified as s+sbh .
it is often assumed that the 60@xmath0 m luminosity of apparently quiescent early - type galaxies is related to cirrus excited by the emission of the underlying stellar population ( sauvage & thuan , 1992 ) .
this is however disputed by devereux and hameed ( 1997 ) who present some counter - examples , and argued that young stars can be an important source of the fir light of early - type galaxies .
one may wonder however about the exact nature of early - type galaxies with a significant population of young stars embedded in dust .
might they not be equivalent to our s+sbh galaxies ?
in that case , the cirrus contribution to the infrared luminosity of these galaxies can not be significant as the starburst component contribute to most of their bolometric luminosities .
+ recall that former estimates of the sfr co - moving density from the uv light are uncertain , because : 1 .
they include contributions at 2800 from agn and old stellar populations .
2 . they do not account for the uv light reprocessed by dust into fir radiation .
we wish here to take into account the deficiencies mentioned above and calculate _ upper and lower limits to the luminosity density related to star formation_. the upper limit is obtained with the following assumptions : + first , the 15@xmath0 m and radio fluxes of all objects not detected at those wavelengths are set equal to their detection limit .
this provides an upper limit to their infrared flux .
+ second , the 2800 and 60@xmath0 m luminosities of agn - classified objects are assumed to be powered by star formation .
+ third , the infrared luminosities of s+sbh galaxies are assumed to be only coming from star formation .
+ the lower limit is obtained with the following assumptions : + first , all objects not detected at 15@xmath0 m and at radio wavelengths have zero extinction , and hence the star formation rate estimated at infrared wavelengths equals that estimated at uv wavelengths .
+ second , the 2800 and 60@xmath0 m luminosities of agn and of quiescent ( e ) galaxies are assumed to be related respectively to the active nucleus and to the old stellar component .
+ third , the infrared luminosities of s+sbh galaxies are assumed to be related to both cirrus and star formation , in the proportion of the ratio of the bolometric luminosities of the two components .
+ table 5 displays the global luminosities in the cfrs 1415 + 52 field , as derived from uv ( 2800 ) and infrared ( 8 - 1000@xmath0 m ) fluxes , for the sample of objects detected either at 15@xmath0 m or at radio wavelengths , and for that containing objects not detected at both wavelengths .
the agn contributions are given in the upper limit case .
they can attain non negligible values ( @xmath4 20% ) , including at uv wavelengths .
this represents a major uncertainty for the sfr densities derived from uv fluxes .
the errors take into account uncertainties in source counts , flux measurements errors and uncertainties in the classification scheme .
the star formation history has been estimated by various authors ( see madau _ et al .
_ 1996 ; lilly _ et al .
_ 1996 ; hammer _ et al .
_ 1997 ) on the basis of the cfrs and/or the hdf surveys , using star formation rates derived from 2800 or @xmath46_{3727}$ ] fluxes . since the uv light is dominated by emission from more or less massive stars , the total sfr as derived only from the flux at 2800 is somewhat uncertain , as the extrapolation to the low mass end ( @xmath75 ) of the imf is not constrained
. moreover hammer _ et al . _
( 1997 ) have suggested that dust and metallicity can severely affect these estimates and mask the true evolution of the cosmic star formation rate .
+ we adopt in the calculation of the star formation rate ( sfr ) from the uv light the calibration of madau et al ( 1998 ) and kennicutt ( 1998 ) , assuming a salpeter ( 1955 ) imf with mass limits 0.1 and 100 @xmath12 : @xmath76 the sfr can also be inferred from infrared luminosities , assuming that they result mostly from dust heating by young stars , in the optically thick limit .
the sfr calibration depends mainly on the burst duration and on the imf slope at both high and intermediate mass ranges .
following kennicutt ( 1998 ) we adopt the same imf as that assumed to calculate the sfr derived from uv light and : + @xmath77 the above equation is based on the models of leitherer & heckman ( 1995 ) for a continuous burst with age 10 - 100 myr . condon ( 1992 ) calibrated the sfr from the non - thermal radio luminosity ( @xmath78 ) , assuming the galactic relation between @xmath78 and the radio supernova rate .
he then derived the @xmath79 calibration from the fir - radio correlation , showing that @xmath4 2/3 of the uv - optical emission is re - emitted between 40 and 120 microns ( see also helou 1988 ) .
after rescaling his value by assuming a salpeter imf ( 0.1 - 100 @xmath12 ) , condon ( 1992 ) obtains : @xmath80 the difference between the two authors is at the 28% level , in agreement with the kennicutt uncertainty estimate of @xmath36 30% on his calibration , after comparison with other models ( hunter et al , 1986 , lehnert & heckman , 1996 , meurer et al , 1997 ) .
+ the key parameter in these estimations is the star formation rate of massive stars ( typically @xmath24 5 @xmath12 ) , which are mainly responsible for the uv continuum , the non - thermal radio continuum and fir luminosities re - radiated by dust .
the @xmath81 ratio calculated in the following , is essentially independent of the imf slope , and in particular of any extrapolation towards low stellar masses .
we aim to calculate here the cosmic sfr from the infrared emission , and compare it with that derived from the uv emission .
we first consider the `` directly observed '' sfr , based on observations of the @xmath51 22.5 galaxies , as presented by lilly _
( 1996 ) . in section 5.1
, we have calculated the total @xmath65 contributed by all galaxies with @xmath51 22.5 and @xmath82 . in section 5.2 , we have computed the total @xmath83 from those galaxies with either @xmath28 250 @xmath0jy or @xmath84 16 @xmath0jy and from cfrs galaxies not detected at 15@xmath0 m and at radio wavelengths .
+ we have calculated the @xmath85 ratio accounting for all galaxies not classified as powerful agn .
it ranges from 2.5 @xmath360.95 in the lower limit case , to 5.4 @xmath36 1.9 when galaxies undetected at 15@xmath0 m and in radio have their fluxes at these wavelengths equal to the detection limit .
these values can however be affected by uncertainties on the calibration of the sfr in the uv and fir wavelength ranges .
nevertheless the values we find for @xmath86 and @xmath79 show some consistency , since for all galaxies , @xmath79 is larger than @xmath86 .
these ratios give extinctions ranging from @xmath2=0.49 to 0.87 , if we assume a standard galactic extinction law .
if the true sfr density is given by the fir estimation , then from 35% to 85% of the global sfr co - moving density for z @xmath64 1 are not taken into account when only the uv flux density is considered .
+ + figure 15 shows the co - moving sfr density evolution with lookback time ( for @xmath87=50 km @xmath88 and @xmath89=0.5 ) .
the vla - iso - cfrs points ( filled dots ) represent the average values of the upper and lower limits defined above .
since there is no evidence for a change in the extinction for z @xmath64 0.5 as compared to for 0.5 @xmath25 z @xmath64 1 , we have adopted the same redshift bins as those used in previously deriving the sfr from the uv fluxes .
vla - iso - cfrs data are [email protected] larger than former uv estimated values , even though latter were accounting for all sources , including strong agn .
the error bars are accounting for incompleteness errors ( lilly et al , 1996 ) as well as uncertainties in the determination of the sfr density at ir wavelengths .
+ error bars in figure 15 are very large mostly because we have assumed that the blue galaxies undetected at 15@xmath0 m and at radio wavelengths have no extinction ( lower limit case ) , or have flux densities equal to the detection limits ( upper limit case ) , both of which are clearly unrealistic assumptions .
it is interesting to check whether our average value makes sense .
starbursts detected by both isocam and the vla are rare .
they have large extinctions ranging from @xmath2= 1.5 to 2 , and show red continua ( figure 12 ) .
tresse & maddox ( 1998 ) predict an extinction of 1 mag at 2800 ( or @xmath2= 0.48 ) for the blue field galaxy population at z @xmath1 0.3 .
if the same value holds on average for blue galaxies up to z = 1 , we can predict in principle the infrared luminosity of the field galaxy population not detected at 15@xmath0 m and at radio wavelengths . for them ,
table 5 gives an average value @xmath65= 196 @xmath73 @xmath59 in the field . according to equation 3 ,
this corresponds to a global @xmath86 of 980 @xmath90 . adopting the tresse & maddox extinction estimate
, the global infrared luminosity would be equal to 1445 @xmath73 @xmath59 ( equation 4 ) .
this is in very good agreement with the average value in the fourth column of table 5 , and adds more confidence in the vla - iso - cfrs average values shown in figure 15 .
+ these average values imply that the stellar mass formed since z = 1 would be 3.3 @xmath91 @xmath92 , slightly larger than the estimation of the present day stellar mass of 3 @xmath91 @xmath93 ( cowie et al . , 1996 ; glazebrook et al . , 1995 ) .
of course , the uncertainties on the sfr estimations are large , and our lower estimate is still consistent with the present day stellar mass .
however we believe that this will constitute a serious problem , when uncertainties will be lowered by larger and deeper surveys at infrared and radio wavelengths . the problem gets worse if one accounts for the additional star formation at z@xmath22 1 .
there may be two ways out .
first , half of the sfr density comes from @xmath4 10% of the field galaxies , those detected at 15@xmath0 m and at radio wavelengths . these objects with large sfr ( especially those with @xmath22 100 @xmath12@xmath13 ) may have imfs which significantly deviate from the salpeter law .
alternatively , the local stellar mass density may have been systematically underestimated .
the number densities of faint mir and radio sources found in the cfrs 1415 + 52 field and in the hdf appear to be a reasonable agreement .
however our results are not consistent with those derived for the hdf by rowan robinson _ et al . _
the hdf is 18.5 smaller in area than the cfrs 1415 + 52 field , and from our results , we would expect 0.3 starburst with sfr larger than 100 @xmath94 , as compared to the four starbursts with sfrs from 500 to 1010 @xmath94 claimed by rowan - robinson _
no simple explanation can account for this discrepancy of more than one order of magnitude .
the calibration of the @xmath95 by rowan - robinson et al(1997 ) differs slightly from ours ( their proportionality factor in equation 4 is 2.6 @xmath96 instead of our 1.7 @xmath96 ) .
we simply stress that : a ) : : our source detection and identification procedures are robust ( see also flores et al , 1998 ) , while source identifications can be difficult in the hdf , since several galaxies in the hdf can lie within a single isocam pixel .
b ) : : rowan robinson _
( 1997 ) have used a three - component model ( a stellar component giving rise to the uv to mir emission plus cirrus and starburst components giving rise to the mir to radio emission ) to fit galaxy seds .
this type of model has several degrees of freedom , including the relative energy ratio between the different components .
the number of free parameters is larger than the number of available data points ( 1 - 3 ) in the mir to radio wavelength range .
c ) : : among the twelve hdf galaxies whose seds were fitted by rowan robinson _ et al . _ ( 1997 ) , five have limits on their radio luminosities which are significantly lower than the starburst model predictions ( see their figure 1 ) . instead of force - fitting the data points by multiple - parameter starburst models ,
we propose here a more conservative empirical approach of fitting seds of galaxies with @xmath82 with sed templates of well - studied local galaxies ( see schmitt et al .
the fits are generally quite good , especially for starburst ( sb ) seds which are consistent with the standard radio - fir correlation .
a pure starburst has its energy distribution dominated mostly by star formation .
its sed includes massive star light with or without extinction at uv wavelengths , reemission of the absorbed uv light at mir and fir wavelengths , and starburst thermal emission and synchrotron radiation from supernova remnants at radio wavelengths .
locally , schmitt _
( 1997 ) show that starbursts , when compared to normal spirals or ellipticals , have a smaller spread in their seds from fir to radio wavelengths .
there is no evidence that these properties change significantly in the redshift range 0 @xmath97 .
strong starbursts ( with sfrs larger than 70 @xmath94 ) should be detected by deep radio surveys to @xmath0jy levels , such as those by fomalont _
( 1992 ) in the cfrs 1415 + 52 field , or by fomalont
_ et al . _
( 1997 ) in the hdf .
the fact that these are not detected in the hdf at radio wavelengths casts some doubts on the large sfrs derived by rowan robinson et al .
( 1997 ) for objects without radio counterparts . as for the 3 radio sources with @xmath43 16@xmath0jy and z @xmath1 1 which are found in the hdf ( fomalont et al , 1997 ) , one ( isohdf 12 36 46 + 62 14 06 , z=0.96 ) is found by rowan - robinson _ et al . _
( 1997 ) to have the highest sfr ( 1010 @xmath94 ) in their sample .
it is at z=0.96 , has an elliptical morphology , is undetected at 15 @xmath0 m ( as most of the elliptical galaxies of the hammer _
( 1995 ) radio sample ) , and shows highly ionized emission typical of a seyfert galaxy ( mgii@xmath982799 , [ nev]@xmath983426 and [ neiii]@xmath983868 ) .
it is likely a seyfert 1 galaxy since the mgii line is broad , and its radio emission shows some variability ( richards et al .
it is doubtful that its emission is powered by star formation .
+ to derive fir luminosities ( and hence sfrs ) requires complete information from uv to radio wavelengths .
it seems reasonable to use templates based on well - known local objects , especially for the most powerful star - forming galaxies .
observations of distant field galaxies have been obtained with isocam down to unprecedented flux levels at 6.75@xmath0 m ( s@xmath24150@xmath0jy ) and at 15@xmath0 m ( s@xmath24250@xmath0jy ) .
we have attempted to make our samples at both those wavelengths as complete as possible , by a careful reduction of the data . that the samples are complete is suggested by the good positional correlation between source identifications at 6.75@xmath0 m , 15@xmath0 m and radio wavelengths . + + source densities are comparable at 6.75@xmath0 m ( 1944 s@xmath22150@xmath0jy sources per square degree , flores et al .
1998 ) , 15@xmath0 m ( 2808 s@xmath22250@xmath0jy sources per square degree , this paper ) and 5 ghz ( 1440 s@xmath2216@xmath0jy sources per square degree , fomalont et al , 1991 ) .
star - forming objects contribute respectively , 50% , 73% and 26% of the extragalactic counts at 6.75@xmath0 m , 15@xmath0 m and 5 ghz .
this suggests that the 60@xmath0 m luminosity density is strongly dominated by star - forming galaxies .
the fraction of z @xmath22 1 objects is found to be @xmath25 32% , @xmath25 43% and @xmath25 40% of the extragalactic counts at 6.75@xmath0 m , 15@xmath0 m and 5 ghz , respectively .
the 15@xmath0 m survey is found to be rather efficient in selecting high - redshift objects , since sources with @xmath4422.5 and @xmath99 250 @xmath0jy have a median redshift of 0.76 as compared to 0.59 for the whole cfrs .
+ the 15@xmath0 m survey , combined with radio and optical data , allows to identify the most powerful star - forming objects ( with @xmath100 larger than 100 @xmath94 ) in the field , at least up to z = 1 .
four such objects ( 0.7% ) are found among the 578 @xmath101 22.5 galaxies , and they contribute to 18% of the sfr density . from their uv or @xmath46_{3727}$ ] emission line properties ,
these objects can not be distinguished from galaxies with more modest rates of star formation .
if we correct for extinction , assuming that their optical spectra from 2500 to 4000 is reddened by the standard galactic extinction curve ( with @xmath2 derived from the @xmath79 to @xmath102 ratio ) , they would appear as young starbursts with a moderate population of a stars ( @xmath103= 3.5 in the combined spectrum ) .
the 15@xmath0 m sample contains highly reddened young starbursts as well as a larger number of galaxies ( s+sb ) with lower sfrs , which all contain a significant population of a stars ( s+a galaxies ) .
this is consistent with the scenario of strong starburst episodes followed by the last phases of the burst , where the ir emission is still high due to dust heating by intermediate - mass stars ( m= 1 - 3 @xmath12 , lisenfeld et al , 1997 ) .
+ combination of 15@xmath0 m and radio samples probably gives a good representation of the galaxy population in a very deep 60 @xmath0 m survey .
interpolated infrared ( 8 - 1000 @xmath0 m ) luminosities based on fits of galaxy seds by local templates from radio to uv wavelengths , imply that 75% ( -40%,@xmath10410% ) of the star formation rate density for @xmath82 is hidden by dust . no evidence has been found for an evolution of that fraction in the above redshift range .
the global opacity of the universe up to z = 1 ranges from @xmath2= 0.5 to 0.85 .
a subsample of sixteen 15@xmath0 m galaxies observed by the hst indicates that more than a third of the star formation hidden by dust is associated with interacting galaxies or mergers .
+ + estimates of the sfr from uv fluxes carry some uncertainties because of two main reasons , one related to the uv light reprocessed by dust into fir radiation , and the other , to the probably important contribution of agn light at uv wavelengths . in spite of the small statistics of the sample considered here , our work allows a first glimpse of the true sfr density for @xmath82 . from our careful data analysis , and the use of multi - wavelength data from radio to uv , we believe that the true sfr density lies within the region delimited by the still large error bars in figure 15 .
our average value , 75% of the star formation rate density hidden by dust , is consistent with : + - a one mag absorption at 2800 for the blue galaxy population ( see tresse & maddox , 1998 ) , and , + - less than 4% of the galaxies are highly reddened , detected at 15@xmath0 m and at radio wavelengths and contribute to 50% of the global ir luminosity density .
+ although our values are lower by a factor 2.8 than those of rowan robinson et al .
( 1997 ) , they might be too high when the corresponding stellar mass formed since z = 1 is compared to the present - day stellar mass density
. this could raise an important question about the universality of the imf , especially in the high sfr galaxies detected by isocam and the vla , and described in this paper .
studies of other cfrs fields with the same multi - wavelength technique are needed to improve the source statistics , determine more accurately the sfr density , and study in more detail its redshift evolution .
we thank marc sauvage and david elbaz for useful discussions .
we are also grateful to david schade who made available the hst images of the cfrs 1415 + 52 field in stamp format .
comments and criticisms from an anonymous referee has led us to greatly improve the manuscript .
brinchmann , j. , abraham , r. , schade , d. , et al . , 1997 ,
apj , 499 , 112 .
bruzual , g. and charlot , s. 1995 .
cesarsky , c.j . ,
abergel , a. , agnese , p. et al , 1996 , a & al , 315 , 32 .
clements , d. , puget j. , lagache g. , et al , 1998 , astro - ph/9809054 .
condon , j.and roderick j. , 1988 , aj , 96 , 30 .
condon , j. , 1992 , ara&a * 30 * 575 .
connolly a. , szalay a. , dickinson m. , subbarao m. , brunne r. , 1997 , apjl , 486 , 11 .
cowie , l. , hu , e. songaila , a. , and cohen , j. , 1996 , apj , 112 , 839 .
dsert , f. x. , puget , j . , clements , d. l. , et al .
1998 , a&a , to appear .
devereux n. and hameed s. , 1997 , 113 , 599 .
devereux n. and young j. , 1991 , apj , 371 , 515 .
edelson r. and malkan m. , 1986 , apj , 308 , 59 .
elbaz , d. , aussel , h. , baker , a. , cesarsky , c. et al , astro - ph/9807209 flores , h. , hammer , f. , desert , f.x .
, et al.,1998 , a&a , to appear .
fomalont , e.b . ,
windhorst , r.a . ,
kristian , j.a . and kellerman , k.i . , 1991 ,
aj , 102 , 1258 .
fomalont , e.b . ,
kellermann , k. , richards , e.,et al . , 1997 ,
apjl , 475 , 5 .
franceschini a. , mazzei p. , zotti g. and danesse l. , 1994 , apj , 427 , 140 .
franceschini a. , toffolatti l. , et al , 1991 , a&ass , 89 , 285 .
hammer f. , crampton , d. , lilly , s. , le fvre , o. , kenet , t. , 1995 , mnras , 276 , 1085 .
hammer f. , schade , d. , crampton , d. , lilly , s. , le fvre , o. , 1996 , in `` science with the hst ii '' , proceedings of a workshop held in paris , dec .
4 - 8 , 1995 , ed .
et al _ , p. 101 .
hammer f. , flores h. , lilly s. , crampton d. , le fvre o. , rola c. , mallen - ornelas g. , schade d. and tresse l. , 1997 , apj , 480 , 59 .
helou g. , khan i. , malek l. and boehmer l. 1988 , apjs 68 , 151 .
hunter j. sandford m. , whitaker r. and klein r. 1986 , apj 305 , 309 .
genzel r. , lutz d. , sturn e. , el al , 1997 , astro - ph/9711255 .
groth e. , kristian j. , lynds r. , _ et al .
_ , 1994 , baas , 185 , 5309 .
gallagher j. , bushouse , h. , hunter , 1989 , aj , 97 , 700 .
gallego j. , zamorano , j. , aragon - salamanca , a. , rego , m. , 1995 , apj , 455 , 1 .
glazebrook k. , peacok j. , miller l. , collins c. , 1995 , mnras , 275 , 169 glazebrook k. , blake c. , economou f. , et al , 1998 , astro - ph/9808276 .
kennicutt , r. 1992 , , 388 , 310 .
kennicutt , r. jr .
, 1998 , to appear in , astro - ph/9712213 .
kessler m.f . ,
steinz , j.a . , anderegg , m.e .
et al . , 1996 ,
a&al , 315 , 27 .
lehnert m. and heckman t. , 1996 , apj 472 , 546 .
leitherer m. fergunson h. and heckman t. , 1995 , aas 186 , 3004 .
lilly s. , le fvre o. , crampton d. , hammer f. , and tresse l. , 1995 , apj * 455 * , 50 .
lilly , s.j .
, hammer , f. , le fvre , o. , crampton , d. , 1995 , , 455 , 75 .
lilly s. , le fvre o. , hammer f. , crampton , d. , 1996 , apjl * 460 * , l1 .
lisenfeld u.,volk h. and xu c. , 1996 ,
astro - ph/9605118 .
lutz d. , genzel r. , rigopoulou d. , et al , 1998 , to appear in , astro - ph/9808330 .
madau p. , ferguson h. and dickinson m. , _ et al .
_ , 1996 , mnras , 283 , 1388 .
madau p. , pozzetti l. and dickinson m. , 1998 , apj , 498 , 106 .
meurer g. , heckman t. , lehnert m. et al , 1997 , aj 114 , 54 .
moshir m. , `` iras faint source survey , explanatory supplement '' version 1 , 1989 .
osterbrock , d. , 1989 , astrophysics of gaseous nebulae and active galactic nuclei " , university science books .
puget j. and lger , 1989 araa , 27 , 169 .
puget j. , lger a. and boulanger , f. , 1985 , a&al , 142 , 9 .
puget j. , et al , 1998 , in preparation richards e. , kellermann k. , fomalont e. et al , 1998 , aj 116 , 1039 .
rowan - robinson , m. , mann , r. , oliver , s. , et al . , 1987 , mnras .
salpeter e. , 1955 , apj 121 , 161 .
sauvage m. and thuan t.x .
, 1992 apjl , 396 , 69 .
scoville and young j. , 1983 , apj * 265 * 148 .
sanders d. , soifer b. , elias j , et al , 1988 , 328 , 35 .
sanders d. , phinney e. , neugebauer g. , 1989 , _ et al .
_ , apj , 347,29 .
schade d. , crampton d. , hammer f. , le fvre , o. , lilly s.j .
, 1996 , mnras , 278 , 95 .
schmitt h. , kinney a. , calzetti d. and storchi - bergmann t. , 1997 , aj , 114 , 592 .
spinoglio l. and malkan m. 1989 , esa irsa 343 .
spinoglio l , malkan m. , rush b. et al , 1995 , apj 453 , 616 .
tresse l. , hammer f. , le fevre o. and proust d. 1993 , a&a 277 , 53 .
tresse l. and maddox , 1998 , 495 , 691 .
wrobel j. and heeschen d. , 1988 , apj , 335 , 677 .
rllllcccrlrr 0 & 14:17:41.8 & 52:28:23.3 & 14.1157 & 1.150 & 20.54 & 22.45 & & 0.65 & 0.000000@xmath105 & 1653 . & 57 . 5 & 14:17:41.9 & 52:30:23.2 & 14.1139 & 0.660 & 20.20 & 21.49 & 18.92 & 2.94 & 0.000383@xmath105 & 562 . & 48 . 9 & 14:17:40.4
& 52:28:21.1 & 14.1192 & & 23.49 & 24.40 & & 0.92 & 0.012702 & 399 . & 58 .
13 & 14:17:52.0 & 52:25:32.8 & 14.0855 & & 20.92 & 22.60 & & 0.72 & 0.000998 & 487 . & 69 .
32 & 14:17:56.6 & 52:31:58.6 & 14.0711 & & 21.44 & 22.29 & 20.26 & 1.50 & 0.006552 & 274 .
42 & 14:18:20.9 & 52:25:53.0 & 14.0098 & star & 14.66 & 16.44 & & 1.53 & 0.000029 & 362 . & 69 .
43 & 14:17:42.6 & 52:28:46.3 & 14.1129 & & 21.05 & 22.35 & 20.30 & 0.57 & 0.000694 & 209 . & 52 .
44 & 14:17:34.9 & 52:27:51.0 & 14.1329 & 0.375 & 19.52 & 20.60 & & 1.89 & 0.000294@xmath105 & 347 . & 52 .
51 & 14:17:24.3 & 52:30:24.0 & 14.1567 & 0.479 & 19.79 & 20.04 & 18.62 & 2.99 & 0.006937 & 459 . & 57 .
84 & 14:17:45.8 & 52:30:31.2 & 14.1028 & 0.988 & 21.57 & 23.84 & 19.69 & 1.52 & 0.007463@xmath105 & 295 . & 50 .
145 & 14:18:07.1 & 52:28:37.3 & 14.0446 & & 20.00 & 21.33 & & 0.31 & 0.000089 & 314 . & 54 .
206 & 14:18:13.4 & 52:31:47.3 & 14.0272 & 0.668 & 20.51 & 21.70 & 19.12 & 1.51 & 0.003157 & 297 . & 55 .
233 & 14:17:58.5 & 52:27:14.8 & 14.0663 & 0.743 & 20.88 & 22.34 & & 2.51 & 0.002081 & 288 .
282 & 14:18:18.9 & 52:29:05.4 & 14.0138 & star & 15.77 & 16.80 & 15.43 & 5.66 & 0.000985 & 214 . & 53 .
294 & 14:17:47.0 & 52:29:11.9 & 14.0998 & 0.430 & 20.58 & 21.84 & 19.06 & 2.44 & 0.008694@xmath105 & 245 . & 53 .
303 & 14:17:46.4 & 52:33:50.8 & 14.1006 & & 20.66 & 21.74 & & 2.12 & 0.007003 & 270 . & 53 .
308 & 14:17:26.7 & 52:32:20.9 & 14.1511 & & 20.71 & 21.61 & & 0.78 & 0.000990 & 284 . & 61 .
369 & 14:18:4.0 & 52:27:47.6 & 14.9154 & 0.812 & 21.57 & 23.06 & & 3.25 & 0.000489@xmath105 & 304 . & 51 . 8 & 14:17:23.8 & 52:27:49.3 & 14.1598 &
& 19.84 & 21.00 & & 8.87 & 0.061772 & 367 . & 73 .
24 & 14:17:23.7 & 52:34:33.7 & 14.1582 & & 22.78 & 23.44 & & 3.92 & 0.123123 & 341 . & 77 .
138 & 14:17:26.3 & 52:32:51.5 & 14.1527 & & 23.00 & 24.24 & & 5.54 & 0.268762 & 252 . & 62 .
139 & 14:18:18.3 & 52:29:16.3 & 14.0150 & & 22.23 & 24.19 & 19.98 & 1.64 & 0.014691 & 284 . & 57 .
160 & 14:17:37.4 & 52:31:41.9 & 14.1278 & & 22.08 & 28.36 & 20.31 & 11.04 & 0.448282 & 401 . & 55 .
190 & 14:17:54.1 & 52:33:56.2 & 14.0779 & 0.578 & 22.01 & 23.10 & & 2.94 & 0.039091 & 255 . & 53 .
195 & 14:17:24.6 & 52:30:40.1 & 14.1569 & & 20.61 & 21.65 & 19.51 & 7.05 & 0.071948 & 364 . & 61 .
258 & 14:17:41.1 & 52:30:22.3 & 14.1166 & 1.015 & 22.46 & 23.88 & 20.46 & 5.24 & 0.166123 & 188 . & 47 .
258 & & & 14.1178 & 9.999 & 22.47 & 24.66 & 20.24 & 6.38 & 0.237751 & 188 . & 47 .
278 & 14:17:42.9 & 52:28:00.8 & 14.1103 & 0.209 & 22.33 & 22.64 & & 7.84 & 0.306807 & 235 . & 55 . 278 & & & 14.1145 & & 22.64 & 25.72 & & 7.72 & 0.365875 & 235 . & 55 .
278 & & & 14.1091 & & 22.02 & 22.79 & & 10.70 & 0.412821 & 235 . & 55 .
278 & & & 14.1125 & & 22.62 & 25.75 & & 8.45 & 0.415531 & 235 . & 55 .
296 & 14:17:56.0 & 52:32:55.9 & 14.0741 & & 23.00 & 23.16 & & 6.09 & 0.314943 & 207 . & 49 .
296 & & & 14.0743 & & 21.65 & 22.74 & & 10.58 & 0.320898 & 207 . & 49 .
304 & 14:17:27.3 & 52:32:09.6 & 14.1489 & & 20.91 & 20.54 & & 9.14 & 0.147525 & 284 . & 61 .
307 & 14:17:47.9 & 52:34:06.6 & 14.0975 & & 21.21 & 23.19 & & 7.45 & 0.126166 & 252 . & 55 .
325 & 14:18:12.5 & 52:32:48.4 & 14.0291 & & 22.19 & 24.51 & & 3.88 & 0.077093 & 225 . & 51 .
361 & 14:17:49.9 & 52:33:43.2 & 14.0909 & 0.978 & 22.34 & 24.30 & & 2.12 & 0.026649 & 248 . & 56 .
372 & 14:18:02.7 & 52:27:59.6 & 14.0557 & & 21.63 & 22.51 & & 3.57 & 0.042434 & 225 . & 52 .
421 & 14:17:52.5 & 52:35:13.4 & 14.9907 & & 22.95 & 19.41 & & 7.78 & 0.447372 & 259 . & 55 .
434 & 14:17:44.1 & 52:25:50.5 & 14.1070 & & 20.86 & 21.33 & & 5.82 & 0.060281 & 277 . & 65 .
440 & 14:17:39.3 & 52:28:46.1 & 14.1232 & & 22.06 & 23.19 & 20.05 & 3.91 & 0.070782 & 245 . & 55 .
440 & & & 14.1212 & & 20.20 & 20.65 & 20.30 & 8.87 & 0.081572 & 245 . & 55 .
457 & 14:17:52.1 & 52:30:49.8 & 14.0846 & 0.989 & 21.81 & 23.15 & 20.34 & 2.97 & 0.034073 & 250 . & 50 .
021 & 14:17:23.2 & 52:32:58.3 & & & & & & & 1.0000000 & 270 . & 62 .
202 & 14:17:57.3 & 52:32:29.8 & & & & & & & 1.0000000 & 209 . & 49 .
288 & 14:17:53.6 & 52:33:44.6 & & & & & & & 1.0000000 & 231 . & 55 .
439 & 14:17:54.7 & 52:32:54.9 & & & & & & & 1.0000000 & 248 . & 53 .
171 & 14:18:16.1 & 52:29:39.2 & 14.0198 & 1.603 & 20.04 & 20.21 & 19.86 & 1.52 & 0.002196 & 211 . & 56 .
183 & 14:17:30.9 & 52:33:44.0 & 14.1400 & star & 15.61 & 16.13 & & 5.59 & 0.000845 & 208 . &
228 & 14:17:23.3 & 52:30:33.4 & 14.1609 & star & 18.93 & 20.83 & 18.30 & 2.54 & 0.002518 & 193 . & 61 .
243 & 14:17:35.9 & 52:32:46.6 & 14.1302 & 0.548 & 20.85 & 21.72 & & 3.19 & 0.018359 & 188 . & 55 .
310 & 14:18:15.0 & 52:31:22.2 & 14.0227 & 0.772 & 20.84 & 22.08 & 19.47 & 3.42 & 0.020906 & 196 . & 53 .
326 & 14:17:53.5 & 52:25:51.5 & 14.0818 & 0.899 & 21.02 & 22.14 & & 2.74 & 0.010601 & 165 . & 54 .
351@xmath106 & 14:17:40.7 & 52:33:57.9 & 14.9504 & & 23.90 & 99.99 & & 8.44 & 0.005986@xmath105 & 228 . & 59 .
354 & 14:17:47.8 & 52:32:52.8 & 14.0968 & & 23.39 & 24.88 & & 1.07 & 0.015834 & 172 . & 54
. 396 & 14:18:18.2 & 52:33:04.9 & 14.0151 & & 17.92 & 18.19 & & 1.93 & 0.000647 & 185 . & 53 .
408 & 14:17:40.5 & 52:27:14.7 & 14.1190 & 0.754 & 20.99 & 22.49 & & 0.95 & 0.001837@xmath105 & 185 . & 56 .
427 & 14:17:59.7 & 52:26:01.4 & 14.0645 & & 22.44 & 24.40 & & 9.72 & 0.000824@xmath105 & 247 . & 67 .
449 & 14:17:42.7 & 52:32:21.6 & 14.1117 & & 20.79 & 21.15 & & 3.77 & 0.024364 & 213 . & 54 .
475 & 14:17:53.9 & 52:31:37.0 & 14.0820 & 0.976 & 21.69 & 24.27 & 19.37 & 11.82 & 0.012867@xmath105 & 203 . & 53 .
137 & 14:17:23.9 & 52:25:54.1 & 14.1597 & & 23.05 & 25.95 & & 4.89 & 0.224187 & 284 . & 93 .
174 & 14:17:45.3 & 52:29:47.6 & 14.1042@xmath107 & 0.722 & 21.49 & 23.38 & 19.79 & 3.67 & 0.040133 & 197 . & 50 .
183 & 14:17:30.8 & 52:33:54.2 & 14.1403 & & 21.48 & 24.21 & & 4.96 & 0.071532 & 208 . & 60 .
228 & 14:17:23.7 & 52:30:29.7 & 14.1591 & & 22.76 & 23.65 & 21.60 & 5.51 & 0.225442 & 193 . & 61 .
243 & 14:17:36.0 & 52:32:55.4 & 14.1300 & & 23.25 & 24.15 & & 5.65 & 0.328192 & 188 . & 55 .
276 & 14:18:12.7 & 52:31:41.8 & 14.0287 & & 22.29 & 24.19 & 19.98 & 7.56 & 0.281069 & 175 . & 51 .
277 & 14:17:24.7 & 52:26:56.2 & 14.1554 & & 21.90 & 23.07 & & 5.51 & 0.120360 & 240 . & 76 .
341 & 14:17:23.9 & 52:32:28.0 & 14.1576 & & 22.33 & 23.83 & & 9.93 & 0.444487 & 214 . & 64 .
341 & & & 14.1596 & & 24.21 & 26.44 & & 4.52 & 0.422721 & 214 . & 64 .
367 & 14:17:25.7 & 52:31:32.2 & 14.1533 & & 22.42 & 23.19 & 21.81 & 7.46 & 0.299945 & 177 . & 54 .
377 & 14:17:40.8 & 52:34:13.0 & 14.1170 & & 21.83 & 23.92 & & 5.14 & 0.100134 & 233 . & 60 .
384 & 14:17:42.6 & 52:29:54.3 & 14.1120 & & 22.60 & 23.57 & 20.78 & 6.25 & 0.251090 & 187 . & 51 .
396 & 14:18:17.7 & 52:33:02.6 & 14.0157 & & 22.35 & 23.01 & & 7.03 & 0.258736 & 185 . & 53 .
399 & 14:18:05.8 & 52:32:10.2 & 14.0477 & & 23.01 & 24.73 & & 4.36 & 0.177522 & 179 . & 53 .
408 & 14:17:40.3 & 52:27:20.6 & 14.1195 & & 22.67 & 24.42 & & 5.85 & 0.235036 & 185 . & 56 .
424 & 14:18:12.3 & 52:29:45.4 & 14.0302 & & 20.87 & 22.20 & 19.22 & 5.07 & 0.046449 & 191 . & 50 .
426 & 14:18:19.9 & 52:30:47.2 & 14.0112 & & 23.53 & 24.86 & 20.61 & 6.04 & 0.433874 & 160 . & 53 .
442 & 14:17:38.5 & 52:34:43.2 & 14.1252 & & 21.91 & 21.79 & & 10.02 & 0.347868 & 221 . & 69 .
449 & 14:17:43.3 & 52:32:19.5 & 14.1100 & & 24.46 & 24.37 & & 2.75 & 0.220019 & 213 . & 54 .
474 & 14:17:31.7 & 52:29:11.0 & 14.1393 & & 22.23 & 23.67 & 21.28 & 2.42 & 0.031712 & 160 . & 52 .
231 & 14:17:33.4 & 52:34:21.1 & & & & & & & 1.0000000 & 209 . & 59
. 268 & 14:17:31.4 & 52:25:37.5 & & & & & & & 1.0000000 & 224 . & 70 .
348 & 14:18:06.3 & 52:25:32.0 & & & & & & & 1.0000000 & 193 . & 67 .
359 & 14:17:23.0 & 52:28:59.2 & & & & & & & 1.0000000 & 234 . & 73 .
483 & 14:17:29.9 & 52:29:21.3 & & & & & & & 1.0000000 & 167 . & 53 . | the cfrs 1452 + 52 field has been deeply imaged with the infrared space observatory ( iso ) using isocam through the lw3 filter ( 12 - 18@xmath0 m ) .
careful data analysis and comparison to deep optical and radio data have allowed us to generate a catalog of 78 15 @xmath0 sources with both radio and optical identifications .
they are redder and lie at higher redshift than i - band selected galaxies , with most of them being star - forming galaxies .
+ we have considered the galaxies detected at radio and 15@xmath0 m wavelengths which potentially include all strong and heavily extincted starbursts , up to z=1 .
spectral energy distributions ( sed ) for each of the sources have been derived using deep radio , mid - ir , near - ir , optical and uv photometry .
the sources were then spectrally classified by comparing to seds of well known nearby galaxies . by deriving their fir luminosities by interpolation , we can estimate their star formation rate ( sfr ) in a way which does not depend sensitively on the extinction .
75% ( -40% , + 10% ) of the star formation at z@xmath1 1 is related to ir emission and the global extinction is in the range @xmath2=0.5 0.85 . while heavily extincted starbursts , with sfr in excess of 100 m@xmath3 constitute less than a percent of all galaxies , they contribute about 18% of the sfr density out to z=1 .
their morphologies range from s0 to sab , and more than a third are interacting systems .
+ the sfr derived by fir fluxes is likely to be @xmath4 2.9 times higher than those previously estimated from uv fluxes .
the derived stellar mass formed since the redshift of 1 could be too high when compared to the present day stellar mass density .
this might be due to an imf in distant star - forming galaxies different from the solar neighborhood one , or to an underestimate of the local stellar mass density . |
a great deal of effort has gone into observing and analyzing disintegration of sunspots ( and starspots ) .
the sunspot decay is usually characterized by the rate of decrease of the sunspot area @xmath3 , and numerous observations appear to be consistent with a parabolic decay law , with @xmath4 a decreasing quadratic function of time @xmath5 .
early theories invoked turbulent diffusion of the magnetic field within the spot to model the observed rate of decay , yet such models predicted a linear decay law , corresponding to a constant area decay rate @xmath6 @xcite . in order to explain a parabolic decay
, @xcite developed a model of sunspot disintegration by turbulent `` erosion '' of penumbral boundaries , which occurs when bits of magnetic field are sliced away from the edge of a sunspot and swept to the supergranular cell boundaries by supergranular flows @xcite .
a key feature of the erosion model is that the turbulent diffusivity , associated with the flows , is suppressed within the spot .
the assumption is justified by the theoretical prediction that the diffusivity @xmath0 should rapidly decrease if the magnetic field @xmath2 exceeds an energy equipartition value @xcite , which is why the diffusivity in the turbulent erosion model may be assumed to be a decreasing function of the magnetic field strength .
@xcite presented observational evidence in support of the parabolic decay law and its theoretical explanation by turbulent erosion .
@xcite recently revisited the theory of sunspot decay by turbulent erosion , considered as a moving boundary problem . while some of the earlier results were confirmed for moderate sunspot magnetic field strengths , the new analytical and numerical solutions yielded a significantly improved theoretical description of sunspot disintegration .
in particular , the dependence of the spot area was shown to be a nonlinear function of time , which in a certain parameter regime can be approximated by a parabola .
more accurate expressions for the spot lifetime in terms of an initial magnetic field were derived analytically and verified numerically .
following @xcite , @xcite assumed in their study that the turbulent diffusivity @xmath7 within a decaying sunspot is much less than that outside the spot .
a more realistic model should incorporate a more realistic dependence of the turbulent diffusivity on the field strength within the spot .
our aim in this paper is further to develop the theory of turbulent erosion by exploring the effect of a non - vanishing diffusivity within a sunspot on the rate of its disintegration . theoretical mechanisms and observable features of the diffusive transport of the photospheric magnetic field have been a subject of intense research activity .
our detailed analysis of a simple nonlinear model , reinforced by numerical solutions , can complement more detailed magnetohydrodynamic simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ) and guide empirical models @xcite in studies of sunspot and starspot evolution .
the determination of lifetimes of spots is a topic of general and current astrophysical interest .
quantitative models for starspot evolution may help estimate the magnetic diffusion timescale and thus yield important constraints in stellar dynamo models , which provides further motivation for our exploration of the turbulent erosion model .
following @xcite and @xcite , we model a decaying sunspot as a cylindrically symmetric flux tube .
the evolution of the magnetic field @xmath8 is governed by the following nonlinear diffusion equation : @xmath9 where @xmath5 is time , and @xmath10 is the distance from the @xmath11-axis . the turbulent diffusivity @xmath0 is suppressed in a magnetic field exceeding an energy equipartition value @xmath12 @xcite , where @xmath13 g for typical parameters of the solar photosphere @xcite .
it follows that @xmath0 is a decreasing function of @xmath2 , although the exact functional dependence remains uncertain . below
we investigate several choices for @xmath7 , which yield analytical solutions .
as previously @xcite , we choose dimensionless units so that @xmath14 , where @xmath15 and @xmath16 is the initial radius of the fluxtube . consequently the time is measured in units of @xmath17 .
the initial value problem is specified by the dimensionless magnetic field profile @xmath18 at @xmath19 , where @xmath20 is the maximum field within the spot .
another important parameter is the total magnetic flux of the spot @xmath21 an initially present sunspot corresponds to @xmath22 ( or @xmath23 in dimensional units ) . in the following
we define the dimensionless fluxtube radius @xmath24 by the condition that @xmath25 at its edge .
our choice of the length scale @xmath16 to be the initial radius of the spot means that @xmath26 and it follows that @xmath27 finally , we define the spot decay time @xmath28 by the condition @xmath29 which is equivalent to @xmath30
similarity solutions to partial differential equations have a large number of applications ( e.g. * ? ? ?
in particular self - similar solutions to nonlinear diffusion equations have been considered for several forms of the diffusion coefficient ( e.g. * ? ? ?
these similarity reductions not only lead to exact solutions of specific initial - value problems but also serve as intermediate asymptotics that approximate solutions of a much larger class of problems . here
we consider a similarity reduction of the nonlinear two - dimensional diffusion equation ( [ eq - diffusion ] ) , assuming that the dimensionless diffusion coefficient has a power - law form : @xmath31 where we take @xmath32 in order to model the supression of magnetic diffusivity in the strong magnetic field of a sunspot . the self - similar solution to equation
( [ eq - diffusion ] ) , which satisfies the flux conservation condition @xmath33 , is known to take the form @xmath34 with @xmath35 @xcite . on substituting this form into equation ( [ eq - diffusion ] ) , solving for @xmath36 , and using equation ( [ eq - phi0-def ] ) to specify an integration constant , we get the following expression for an evolving field profile : @xmath37^{-1/\nu } .
\label{eq - essim - solution}\ ] ] it follows that the maximum field ( at @xmath38 ) is given by @xmath39 and so @xmath40 now equation ( [ eq - b-1 - 0 ] ) yields the magnetic flux @xmath41 to find the decay time @xmath28 for a given initial magnetic field @xmath42 , we express the parameter @xmath43 in terms of @xmath42 and substitute the resulting expression into equation ( [ eq - b-0-t - selfsim ] ) . on solving equation ( [ eq - b-0-t ] ) for @xmath28 and using equation ( [ eq - phi0-selfsim ] ) to eliminate @xmath44 , we obtain the sunspot decay time in terms of the parameters @xmath42 and @xmath45 : @xmath46 in the limit @xmath47 , equation ( [ eq - t - selfsim ] ) reduces to the expression for the case of linear diffusion : @xmath48 note that the self - similar solution has a curious feature : equation ( [ eq - t - selfsim ] ) predicts that @xmath49 as @xmath50 ( for any value of @xmath45 ) , whereas it is physically obvious that @xmath51 when @xmath52 .
this singular limit behavior is related to the fact that the magnetic flux @xmath53 as @xmath50 . in practice
this does not cause any problems since we always assume the initial field @xmath22 in order to model a sunspot .
the self - similar solution above is applicable only for @xmath54 since the flux integral in equation ( [ eq - phi0-def ] ) diverges for larger values of @xmath45 , which physically corresponds to an instanteneous flux transfer to infinity @xcite .
more generally , @xmath55 is required to avoid the divergence of the flux integral for diffusion in @xmath56 dimensions .
mathematical issues of existence and uniqueness of solutions were analyzed by @xcite .
while solutions for @xmath57 formally violate the total flux conservation , they are mathematically correct and may provide a useful local description of nonlinear diffusion . as an illustration ,
consider the case @xmath58 .
it is straightforward to derive a separable solution to equation ( [ eq - diffusion ] ) . by assuming @xmath59 , we reduce the problem to a second - order ordinary differential equation for the spatial part @xmath60 .
the solution is as follows : @xmath61 where @xmath62 is an integration constant , and the other integration constant is determined by the requirement that @xmath63 be finite at @xmath38 . in a different context , equation ( [ eq - fnu1 ] ) in a particular case @xmath64 was given by @xcite . on using equations ( [ eq - b-0 - 0 ] ) and ( [ eq - phi0-def ] ) to express @xmath28 and @xmath62 in terms of @xmath42 and @xmath44
, we obtain @xmath65 here @xmath66 is the initial magnetic flux of the spot ( at @xmath19 ) , which decreases with time according to @xmath67 the localized magnetic field profile is seen to shrink with time until the spot vanishes at @xmath68 the termination of the process in a finite time was referred to by @xcite as `` superfast '' diffusion .
this unusual feature of the solution is related to the singular behavior of @xmath69 as @xmath70 : in sharp contrast to the solution of a linear problem , the continuity flux @xmath71 is independent of time and approaches a constant value , @xmath72 , as @xmath73 , which results in `` flux suction at infinity '' @xcite .
physically , because @xmath70 as @xmath74 , the diffusivity @xmath75 .
consequently the diffusion time scale @xmath76 , and so a diffusive description breaks down as @xmath74 .
we now consider a different model for the diffusivity suppression ( quenching ) within a spot , which complements the analysis by @xcite .
we assume that the evolution of the magnetic field is described by the nonlinear two - dimensional diffusion equation ( [ eq - diffusion ] ) with a step dependence in the dimensionless diffusion coefficient : @xmath77 and @xmath78 @xcite considered the case @xmath79 , which corresponds to a very strong suppression of turbulent diffusivity within the spot .
a more realistic model would correspond to a less severe turbulent suppression of the diffusivity within the spot .
therefore we generalize our nonlinear diffusion model to incorporate the effect of a non - vanishing diffusivity within a sunspot on the rate of its disintegration . to obtain an analytical solution for the case of a moderate diffusivity suppression within the spot , we use the heat - balance integral method @xcite , which proved to yield accurate approximations in problems of nonlinear diffusion @xcite .
the basic idea is to require that an approximate solution satisfy an integral of a nonlinear equation rather than the equation itself .
we apply the method to describe the turbulent erosion of a sunspot , modeled as a cylindrically symmetric fluxtube .
we assume that @xmath80 and seek an approximate solution of the form @xmath81 .
\label{eq - b - form}\ ] ] the fluxtube radius @xmath24 is defined by equation ( [ eq - edge ] ) .
consequently we have @xmath82 integration of equation ( [ eq - diffusion ] ) over @xmath10 from @xmath83 to @xmath84 and substitution of the self - similar form ( [ eq - b - form ] ) into the resulting equation yields @xmath85 = 4 \epsilon g r_e^2 , \label{eq - re - ode}\ ] ] where equation ( [ eq - edge ] ) was used to simplify the right - hand side , and equation ( [ eq - fg - rel ] ) was used to eliminate @xmath86 . in the linear case @xmath87 ,
the solution of the initial value problem with a gaussian profile at @xmath19 is given by @xmath88 here the parameters of an evolving gaussian profile are chosen to satify the initial conditions , given by equations ( [ eq - b-0 - 0 ] ) and ( [ eq - b-1 - 0 ] ) .
the decay time , defined by equations ( [ eq - t - def ] ) and ( [ eq - b-0-t ] ) , is easily shown to be @xmath89 where , as previously , we assume @xmath22 in order to exclude solutions with an infinite magnetic flux . note for clarity that , if @xmath90 , we have @xmath91 at @xmath19 in the linear solution , and so diffusion causes the fluxtube radius to increase until it reaches a maximum at @xmath92 and then to decrease . motivated by the form of the linear solution
, we substitute @xmath93 into the approximate equation ( [ eq - b - form ] ) describing a weakly nonlinear case @xmath94 .
thus equation ( [ eq - re - ode ] ) becomes an ordinary differential equation for the fluxtube radius @xmath24 , which should be solved subject to the initial condition given by equation ( [ eq - r=1 ] ) .
equation ( [ eq - t - def ] ) then yields the spot decay time @xmath28 . keeping in mind that @xmath95 is assumed to be small
, we solve equation ( [ eq - re - ode ] ) by iteration .
on substituting a simple linear function @xmath96 into the right - hand side of equation ( [ eq - re - ode ] ) and integrating from @xmath83 to @xmath5 , we get an approximate analytical expression for @xmath97 : @xmath98 \right\ } . \label{eq - re - approx}\ ] ] on setting @xmath99 , we obtain an algebraic equation for the decay time @xmath28 : @xmath100 = 1 .
\label{eq - t - algebr}\ ] ] alternatively , integration of equation ( [ eq - re - ode ] ) over @xmath5 from @xmath83 to @xmath28 yields an expression for @xmath28 , which makes clear that @xmath101 leads to a slower decay : @xmath102 now substitution of equations ( [ eq - g - fun ] ) and ( [ eq - re2-lin ] ) into the right - hand side of equation ( [ eq - t - integral ] ) yields equation ( [ eq - t - algebr ] ) . to solve equation ( [ eq - t - algebr ] ) in the case of a small @xmath95 , we replace @xmath28 by @xmath103 from equation ( [ eq - t - lin ] ) in all terms containing @xmath95 .
the result is as follows : @xmath104 .
\label{eq - t - approx}\ ] ] for a fixed @xmath105 , @xmath106 is an increasing function of @xmath42 .
we note that the assumed form for the magnetic diffusivity @xmath0 , given by equations ( [ eq - dc - step1 ] ) and ( [ eq - dc - step2 ] ) , is finite in the limiting case @xmath107 , whilst the real magnetic diffusivity is expected to vanish in this limit .
however , since @xmath108 remains finite in our solution for any @xmath109 , the behavior of @xmath0 as @xmath107 is irrelevant .
our assumption of a suppressed but nonzero diffusivity inside the spot is physically meaningful as long as the magnetic field @xmath2 remains finite , as it does in our solution .
we have previously considered a solution for @xmath110 valid when the diffusivity vanishes within the spot @xcite , which is accurate for a very strong magnetic field @xmath42 in the spot .
the new solution presented here quantifies the effects of a nonvanishing @xmath0 .
to quantify the accuracy of the analytical results , equation ( [ eq - diffusion ] ) is solved numerically .
we use an explicit scheme which maps the region @xmath111 $ ] in radius @xmath10 to the region @xmath112 $ ] in a transformed independent variable @xmath113 , as explained in the appendix .
the approach allows an exact boundary condition to be imposed at @xmath114 , which provides more accurate solutions than an approximate boundary condition at a finite radius .
we present solutions with a grid spacing @xmath115 in @xmath113 and with a time step one quarter of the stability limit identified in the appendix . to test the numerical method , and to illustrate the properties of the analytic solutions presented in section 3 , we consider solution with the power - law form for the diffusion coefficient , equation ( [ eq - plaw - dcoeff ] ) , which admits self - similar solutions .
first we consider a flux - conserving case ( @xmath116 ) , with @xmath117 .
the solid curves in the upper panel of figure 1 show the analytic magnetic field profile @xmath110 given by equation ( [ eq - essim - solution ] ) at the three times during the spot evolution @xmath19 , @xmath118 , and @xmath119 , where @xmath28 is the decay time defined by equation ( [ eq - t - selfsim ] ) .
the numerical solutions are also shown in this panel by dashed curves , but they coincide with the solid curves and are not visible . the lower panel shows the absolute error in the numerical solution at the times @xmath118 ( circles ) and @xmath119 ( plus signs ) .
the maximum error is @xmath120 .
the numerical solution conserves flux throughout the time evolution ( @xmath121 time steps ) to within @xmath122 .
second we consider the non - flux conserving case @xmath123 , which exhibits superfast diffusion , i.e. vanishing of the spot at the time @xmath28 defined by equation ( [ eq - t - sfast ] ) .
the upper panel of figure 2 shows the analytic and numerical solutions at the three times @xmath19 , @xmath118 , and @xmath119 , for the case @xmath124 , with a log - linear display used to illustrate the behaviour for large @xmath10 .
the analytic solutions are the solid curves , and the numerical solutions at times @xmath118 and @xmath119 are shown by the circles and by the plus signs , respectively .
the analytic solution at time @xmath119 is identically zero .
the numerical solution is accurate initially , but becomes inaccurate as @xmath125 .
the reason for the error is that the boundary condition at infinity , equation ( [ eq - bc2 ] ) , does not reproduce the behaviour of the continuity flux at large @xmath10 , as shown in the lower panel .
in particular the numerical solution does not maintain a large positive value of the continuity flux as @xmath126 , which produces the `` flux suction at infinity '' .
this example demonstrates the need to accurately represent the boundary conditions at large @xmath10 in the solutions .
the numerical solution for the case with a diffusion coefficient defined by equations ( [ eq - dc - step1 ] ) and ( [ eq - dc - step2 ] ) provides a test of the heat - balance integral method results presented in section 4 , and in particular of the expression ( [ eq - t - approx ] ) for the lifetime . in this case
we do not have an exact solution to ensure accuracy , but the numerical solutions presented here conserve flux during the time evolution to within @xmath128 , in the worst case .
first we consider the field profile @xmath110 for the heat - balance solution , specified by equations ( [ eq - b - form ] ) , ( [ eq - fg - rel ] ) , ( [ eq - g - fun ] ) , and ( [ eq - re - approx ] ) .
figure 3 compares the heat - balance method profiles ( solid curves ) with the numerical solutions ( dashed curves ) at the three times @xmath19 , @xmath118 , and @xmath119 .
the solutions assume an initial magnetic field strength @xmath117 .
the left panel shows the case @xmath129 , and the right panel shows the case @xmath130 . for the smaller value of @xmath95
the heat - balance integral method provides a good approximation to the field profile throughout the evolution .
the method is somewhat less accurate for the larger value of @xmath95 .
figure 4 repeats the display in figure 1 , but shows the case @xmath131 .
the accuracy of the field profile obtained with the heat - balance integral method is not strongly dependent on the choice of @xmath42 .
second , we consider the accuracy of the heat - balance integral estimate for the spot lifetime , equation ( [ eq - t - approx ] ) .
the estimate has the form @xmath132 where @xmath133 is the lifetime for the linear case ( @xmath134 ) , given by equation ( [ eq - t - lin ] ) .
figure 5 compares equation ( [ eq - t - approx - taylor ] ) ( solid curves ) with the numerical solution ( circles ) as a function of @xmath42 , for the range @xmath135 .
the upper panel shows @xmath28 for the three cases @xmath87 , @xmath136 , and @xmath130 ( bottom to top ) and the lower panel shows @xmath137 for the nonlinear cases @xmath129 and @xmath130 .
figure 5 demonstrates that the heat - balance method provides a good approximation to the lifetime of the spot for the choice @xmath129 , over the range of initial field strengths considered .
the approximation is worse for the larger value of @xmath95 , as expected .
it is interesting to consider replacing equation ( [ eq - t - approx - taylor ] ) by the @xmath138 pad approximant : @xmath139 the dotted curves in figure 5 show the lifetimes given by equation ( [ eq - t - approx - pade ] ) , and the results show that the pad expression provides a better approximation for the lifetime .
this might be expected based on the final steps in the derivation of equation ( [ eq - t - approx ] ) .
the physical mechanism of sunspot erosion was proposed by @xcite . @xcite introduced a nonlinear diffusion equation of a magnetic flux tube as a model of turbulent erosion .
@xcite identified the maximum magnetic field @xmath42 in a sunspot as the key parameter that determines the lifetime @xmath28 of the spot and derived a sunspot decay law due to turbulent erosion .
as @xcite demonstrated , however , the accuracy of the predictions was limited : for instance , @xcite have shown that the sunspot lifetime in the model is about a half of that originally predicted .
we have presented in this paper a further development of the quantitative theory of sunspot disintegration by turbulent erosion .
our analysis makes it clear that a variety of scalings @xmath140 are theoretically possible , depending on initial conditions and the dependence of the turbulent diffusivity on the magnetic field strength within the spot .
the results may have implications for the use of the model for estimating the magnetic diffusion timescale in starspots , which is an important parameter in stellar dynamo models .
experimental studies typically attempt to infer the anomalous , turbulent - driven magnetic diffusivity by equating the lifetime of a spot to a theoretical diffusion time scale @xcite .
we expect that the more detailed analytical models we have developed may help to improve the accuracy of the procedure .
more generally , we expect that the new models may allow insight into differences and similarities between sunspots and starspots .
the original formulation of the turbulent erosion model predicted a parabolic dependence of the sunspot area on time @xcite . yet
the deviations from the parabolic decay are large for any one spot @xcite , motivating the use of continuous linear piecewise functions in modeling of the starspot growth and decay @xcite .
such an approach is consistent with the more general decay laws predicted in our analysis , ultimately controlled by the dependence of the magnetic diffusivity on the magnetic field strength within a spot .
as pointed out by the referee , however , apparent deviations from the parabolic decay law for individual spots can be caused by the difficulty of identifying the individual spots within a decaying sunspot group or by the effect of a varying external plage field outside the spots @xcite .
the turbulent erosion model can be further improved .
recall that the solution for @xmath79 in @xcite describes turbulent erosion of a sunspot as a moving boundary problem in which the rate of sunspot decay is controlled by the inward speed of a current sheet around the spot .
in other words , the decay rate is determined by the local diffusion rate of magnetic field within the sheet , modeled as a tangential discontinuity at @xmath141 . by contrast , the new solution given by equations ( [ eq - re - approx ] ) and ( [ eq - t - approx ] ) describes the sunspot decay determined by global magnetic field diffusion , and the field discontinuity is ignored in the smooth profile of the evolving magnetic field , which should be a reasonable assumption if @xmath94 . the diffusive evolution of exact self - similar solutions is also a global process . yet
physically the sunspot disintegration rate is likely to be influenced by both mechanisms , and so a more accurate solution for intermediate values of @xmath95 should incorporate both local and global diffusion processes by considering a more general initial field profile in the spot and more realistic diffusivity dependence on the magnetic field strength .
more detailed models of spot decay should also quantify the effects of flux cancellation , caused by photospheric magnetic reconnection ( e.g. * ? ? ?
application of the model to the data on sunspot and starspot decay may shed light on the physics of turbulent diffusion in magnetized astrophysical plasmas .
an anonymous referee s comments and suggestions are gratefully acknowledged .
@xcite numerically solved equation ( [ eq - diffusion ] ) using a crank nicolson scheme .
the method imposed a boundary condition at a finite outer boundary @xmath142 which allowed flux transport across the boundary , using one - sided spatial derivatives .
this approach was an improvement over the assumption of zero flux at an outer radius used by @xcite , but it was still a source of error for the long time integrations necessary to determine the lifetimes for the model spots .
here we use a simpler explicit scheme , but with a transformation of the infinite @xmath10-domain to a finite domain , to allow an exact outer boundary condition .
we consider solution of equation ( [ eq - diff - xr ] ) on a uniformly spaced grid in @xmath113 defined by @xmath147 , with @xmath148 , where @xmath149 is the grid spacing .
similarly we consider discrete times @xmath150 , with @xmath151 , where @xmath152 is a constant time step . a suitable forward - time , centred - space ( ftcs ) discretisation of equation ( [ eq - diff - xr ] ) is @xcite : @xmath153 where @xmath154 and @xmath155 , and where @xmath156 . the diffusion coefficients at intermediate grid points may be approximated by @xmath157 equations ( [ eq - diff - xr - ftcs ] ) and ( [ eq - dcoeff - approx ] ) give the numerical scheme @xmath158 .
\label{eq - numerical - scheme}\ ] ] equation ( [ eq - numerical - scheme ] ) is an explicit prescription for time evolution at grid locations @xmath159 . for @xmath160 ( @xmath38 )
the physical boundary condition is @xmath161 for all times .
equation ( [ eq - bc1 - 1 ] ) may be enforced using a one - sided ( forward ) approximation to the derivative with respect to @xmath113 at @xmath162 : @xmath163 giving the update for the grid point @xmath160 : @xmath164 this approach requires that the initial conditions satisfy equation ( [ eq - bc1 - 2 ] ) , i.e. @xmath165
. under the transformation ( [ eq - xtor ] ) the total magnetic flux , defined by equation ( [ eq - phi0-def ] ) , becomes : @xmath169b[r(x),t)\,dx .
\label{eq - phi0-def - x1}\ ] ] equation ( [ eq - phi0-def - x1 ] ) may be evaluated in our discrete version of the problem using the trapezoidal rule @xcite : @xmath170 where @xmath171 for @xmath172 , and @xmath173 .
equation ( [ eq - phi0-def - x2 ] ) is used to check that flux is approximately conserved by the numerical solution .
a simple estimate of the stability condition for the method may be made as follows .
we expect that an ftcs discretisation of equation ( [ eq - diffusion ] ) is stable at a given time step subject to @xcite : @xmath174 where @xmath175 is the grid spacing .
( strictly this requires a uniform grid in @xmath10 . ) from equation ( [ eq - xtor ] ) we have @xmath176 so equation ( [ eq - ftcs - stability1 ] ) becomes @xmath177 in the case of the model with a step dependence of @xmath0 on @xmath2 ( section 4 ) we can take @xmath178 numerical experimentation suggests that equation ( [ eq - ftcs - stability - hbal ] ) provides a good estimate for the actual stability constraint . | quantitative models of sunspot and starspot decay predict the timescale of magnetic diffusion and may yield important constraints in stellar dynamo models .
motivated by recent measurements of starspot lifetimes , we investigate the disintegration of a magnetic flux tube by nonlinear diffusion .
previous theoretical studies are extended by considering two physically motivated functional forms for the nonlinear diffusion coefficient @xmath0 : an inverse power - law dependence @xmath1 and a step - function dependence of @xmath0 on the magnetic field magnitude @xmath2 .
analytical self - similar solutions are presented for the power - law case , including solutions exhibiting `` superfast '' diffusion . for the step - function case ,
the heat - balance integral method yields approximate solutions , valid for moderately suppressed diffusion in the spot .
the accuracy of the resulting solutions is confirmed numerically , using a method which provides an accurate description of long - time evolution by imposing boundary conditions at infinite distance from the spot .
the new models may allow insight into differences and similarities between sunspots and starspots . |
in human societies social life consists of the flow and exchange of norms , values , ideas , goods as well as other social and cultural resources , which are channeled through a network of interconnections . in all the social relations between people
_ trust _ is a fundamental component @xcite , such that the quality of the dyadic relationships reflects the level of trust between them . from the personal perspective social networks
can be considered structured in a series of layers whose sizes are determined by person s cognitive constraints and frequency and quality of interactions @xcite , which in turn correlate closely with the level of trust that the dyad of individuals share . as one moves from the inner to the outer layers of an individual s social network , emotional closeness diminishes , as does trust . despite its key role in economics , sociology , and social psychology ,
the detailed psychological and social mechanisms that underpin trust remain open . in order to provide a systematic framework to understand the role of trust
, one needs to create metrics or quantifiable measures as well as models for describing plausible mechanisms producing complex emergent effects due to social interactions of the people in an interconnected societal structure .
one example of such social interaction phenomena , in which trust plays an important role , is trading between buyers and sellers .
such an economic process is influenced by many apparently disconnected factors , which make it challenging to devise a model that takes them into account .
therefore , models that have been proposed , necessarily select a subset of factors considered important for the phenomena to be described .
for instance , there are studies of income and wealth distribution @xcite , using gas like models @xcite , life - cycle models @xcite , game models @xcite , and so on . for a review of various agent based models we refer to @xcite .
in addition , we note that detailed studies of empirical data and analysis of the distribution functions @xcite seem to lend strong support in favour of gas - like models for describing economic trading exchanges . in order to consider the role of trust in trading relations we focus on the simplest possible situation in which trust clearly plays a definite role .
this is the case of trading goods or services for money through dyadic interactions or exchange , which takes place either as a directional flow of resources from one individual to another individual or vice versa .
when an agent is buying , trust plays a role , as people prefer to buy from a reliable and reputable selling agent , i.e. agent they trust .
it should be noted that the dyadic relationship does not have to be symmetric , i.e. a seller does not need to trust the buyer .
a key ingredient in the trading interactions is profit that an agent makes when providing goods or services , and it can realistically be assumed that a seller wants to put the highest possible price to its goods , while the buyer tends to perform operations with agents offering a low price . in this study
we propose an agent based `` gas - like '' model to take into account the above mentioned important features of trading .
the model describes dyadic transactions between agents in a random network .
the amount of goods and money are considered conserved in time , but the price of goods and trust , we measure as reputation , vary according to the specific situation in which trade is made . in section [ model ]
we describe the model and set up the dynamic equations of the system . in section [ results ]
we present the results of extensive numerical calculations and explore their dependence on the parameters of the model . here
we also compare our numerical results with available real data and discuss the predictions of the model as well as possible extensions to it .
finally , in section [ conclusion ] we conclude by making remarks concerning the role of trust in trade and social relations .
first we introduce the basic model , which describes the dynamic development of a random network of @xmath0 agents such that the state of agent @xmath1 is defined by two time - dependent state variables , @xmath2 , where @xmath3 stands for the amount of money and @xmath4 for the amount of goods or services .
the pairwise connectivities between agents in the network are described by @xmath5 adjacency matrix @xmath6 .
it is necessary to distinguish the direction of the flow of goods and money in the network , since agent @xmath1 could buy from agent @xmath7 , or vice versa . at time
@xmath8 we define two symmetric matrices , @xmath9 and @xmath10 , with an average of @xmath11 random entries per row , for the flow of money or goods , respectively .
then the adjacency matrix is simply @xmath12 , and @xmath13 stands for the mean degree .
the elements of @xmath14 and @xmath15 are defined as the normalised probabilities of transactions per unit time @xmath16 and @xmath17 , respectively and they could become asymmetric .
these matrices represent the buying or selling transactions , according to the individual agent instantaneous situation .
the dynamic equations for the state variables @xmath18 ( money ) and @xmath19 ( goods ) initialised randomly @xmath20 $ ] are , [ nm ] @xmath21 \label{second } \\
\frac{\mathrm{d}y_{i}}{\mathrm{d}t } & = \sum_{j } \left [ \frac{x_i \beta_{ij}}{s_j } + y_j\alpha_{ji}- \frac{x_{j}\beta_{ji}}{s_i}-y_i\alpha_{ij}\right].\label{third } \end{aligned}\ ] ] where @xmath22 is the price of the goods as decided by seller @xmath1 , and its value depends on time . in both eqs .
( [ nm ] ) the first and second terms on the right represent the transactions in which agent @xmath1 is buying goods from agent @xmath7 .
note that there is an outflow of money ( negative @xmath23 ) and an inflow of goods ( negative @xmath24 ) .
the third and last terms represent selling goods to @xmath7 .
observe that we could simply use the first and third terms in eq .
( [ second ] ) , to represent the exchange of money , or use the second and fourth terms , if the exchanged money is represented as its transaction value @xmath25 ( @xmath26 has units of money per unit good ) .
the same reasoning is applied to eq .
( [ third ] ) .
we preferred to keep the equations in a symmetric form as regards to goods and money .
the elements of the matrices @xmath16 and @xmath17 in the equations represent the proportion of money and goods , respectively , that an agent is distributing amongst its links in a given transaction .
therefore , the simplest expressions for the corresponding transaction coefficients are linear functions of the respective variables , thus we propose , @xmath27 provided @xmath28 for @xmath16 and @xmath29 for @xmath17 , respectively . the unit time @xmath30 could conveniently be taken as one .
the reason to include the heaviside functions @xmath31 is that agent @xmath1 buys goods from agent @xmath7 only if it has enough money to pay for the price agent @xmath7 is asking for its goods ( in @xmath23 ) or if agent @xmath7 has enough goods to satisfy agent @xmath1 ( in @xmath24 ) . by the same token
, agent @xmath1 can not sell to agent @xmath7 if it has not enough goods ( as in @xmath32 ) or if the buyer has not enough money ( as in @xmath33 ) .
the diagonal elements of these matrices represent the quantity of money or goods that are not used in the transaction .
if they are set to zero , all the money that an agent has is used in all the transactions , meaning that there are no savings .
the elements of these matrices can not be negative , hence , whenever an element is about to become negative , it should be set to zero , which is equivalent to say that the link between agents is irrelevant for that particular transaction.this amounts to reshaping the trading network .
these dynamical equations constitute the basic model for trading transactions .
it should be noted that the model conserves the amount of goods @xmath34 and money @xmath35 .
this restriction could be easily relaxed , if needed , by adding up sources and sinks to the equations .
we decided to keep the conserved version for simplicity .
as emphasised above trust plays a key role in all social and trading transactions . in order to include trust into the transactions , we need first to assume that trust is something that you either gain or lose with time
therefore , it has to be related to a quantity that measures the performance of agents as traders , which we here assume to be _ reputation _ @xmath36 .
accordingly , we write @xmath37 \mathrm{d } t ' . \end{aligned } \label{r}\ ] ] observe that the numerator of eq .
( [ r ] ) at a given time increases with the number of successful transactions , since the elements of the trading matrices @xmath16 and @xmath17 are positive definite . in order to maintain the scale of variation of all the variables between zero and one , we normalise the variable @xmath38 by dividing it with the maximum value of @xmath38 , encountered in the network . note that the reputation of a given agent could decrease in time because of the time dependent normalisation .
let us now introduce a non - symmetrical matrix @xmath39 , whose elements @xmath40 regulate the amount of goods and money in each transaction between a pair of agents .
the entries of the matrix @xmath39 range from zero ( no transaction possible ) to unity ( larger amount of goods and money exchanged ) .
we assume that agent @xmath1 prefers to buy if the value of the goods in its possession is smaller than the amount of money it has , and favours selling otherwise , in order to maintain a balanced stock of money and goods , and the same for agent @xmath7 . to reflect this , one could define the matrix elements of @xmath39 as follows , @xmath41 where @xmath42 is the average price over the whole network of the goods at time @xmath43 , which could be very different from the price the individual @xmath1 is pricing its goods .
here it is assumed that agent @xmath1 is buying from agent @xmath7 , and therefore the reputation of agent @xmath7 is the factor that matters .
trust on sellers when buying is important because one buys from a reliable agents only . on the contrary , when selling one is happy to do it to whomever is able to pay , independently of the reputation of the buyer .
note that the transpose @xmath44 , should be used on the transactions in which agent @xmath1 sells goods , in which case the reputation of agent @xmath1 matters .
taking this new feature into account , the dynamical equations of our model read as follows , @xmath45 , \\ & \\
\frac{\mathrm{d}y_{i}}{\mathrm{d}t } = \sum_{j } & \left [ \frac{x_i \beta_{ij}f_{ij}}{s_j } + y_j\alpha_{ji}f_{ij } \right .
\\ & \qquad \left .
- \frac{x_{j}\beta_{ji}f_{ji}}{s_i}-y_i\alpha_{ij}f_{ji}\right ] . \end{aligned } \label{nmf}\ ] ] notice that these model equations reduce to those of the basic model ( eqs .
( [ nm ] ) ) if one eliminates the effect of trust by setting @xmath46 . in this sense , the effect of trust is to regulate the amount of goods and money exchanged in each transaction , but also in deciding if one sells or buys .
the main quantity that regulates the dynamics of goods and money in our trading model is the time dependence of the price of the goods .
therefore , one needs to suggest a mechanism by which an agent modifies the price of its goods .
we assume that in general people would like to sell at the highest possible price , but we need to consider to what extend it is limited by the agent s reputation .
if the agent is reputed to be a successful trader , it could increase its prices by a large amount , provided that the price the agent @xmath1 is using is smaller than the average price the agents connected to it are offering , otherwise lowering the value of @xmath22 is a way of attracting more transactions . in a mathematical language
, one can represent this situation by the following coupled first order equations , @xmath48 , \\ & = -\frac{1}{t_0 } r_i(t ) \left [ s_i(t ) - s^k_i(t ) \right ] , \\
\frac{\mathrm{d } s_i}{\mathrm{d}t } & = \frac{1}{t_0 } \left [ r_i \tau_i(t ) - g s_i(t ) \right ] , \end{aligned}\ ] ] where @xmath49 is a variable that enables agent @xmath1 to decide how to modify its prices according to trust and the information available to it .
the parameter @xmath50 is the time scale for price changes and @xmath51 is a constant that represents reluctance for the agent to lower the prices , which for simplicity is assumed to be very small and the same for all the agents .
we can eliminate the variable @xmath52 from this equations and obtain a second order equation for @xmath26 , which we identify as a set of damped harmonic oscillators , which admits an analytical solution , namely , @xmath53 + s_i^k(t),\ ] ] where @xmath54 and @xmath55 .
notice that this solution is approximate and valid only when the prices do not change appreciably within a time step @xmath56 , which means that @xmath50 is large and @xmath51 is small . in any case
one could find the solution of this non - linear system of coupled oscillators numerically .
the solutions of eqs .
( [ sd1 ] ) produce an apparently chaotic behaviour for @xmath47 showing qualitatively similar to the price variations of real markets ( there are even some models of the stock market that use damped oscillators to analyse the price variation @xcite ) .
the numerical calculations were made in a random network of 500 agents with the average degree of 12 , using a simple euler method with random initial conditions for @xmath18 , and @xmath57 $ ] .
the appropriate step size for convergence was found to be @xmath58 , reached within 200000 iterations . at time @xmath59
we set @xmath60 randomly using a flat distribution with the mean @xmath61 and the width @xmath62 , and set a time scale @xmath63 and @xmath64 . for this @xmath50
the system reaches dynamical equilibrium within the running time .
the diagonal elements of the trading matrices @xmath14 and @xmath15 were set to zero . in fig .
[ fig1 ] we depict the structure of the final network by showing only the active links , where the height of the 3d plot is stratified according to the wealth of the agent @xmath1 , i.e. @xmath65 . here
it is seen that there are poor agents and only a few very rich ones .
in addition we observed in numerous calculations that if the width of the distribution of prices is increased , the rich become richer and the number of poor agents increases . and
@xmath66 , red @xmath67 and @xmath68 , and blue @xmath69 and @xmath66 .
the colour of the nodes is chosen according to the wealth ( @xmath70 such that light green , red , dark green , blue , grey , black are in decreasing order of @xmath71.,width=226 ] in fig .
[ fig2 ] we show the time histories of the variables for the typical calculation of the previous figure .
note that most agents have converged to certain amounts of money and goods , but the ones who posses more wealth are prevented from taking it all , which is an effect of trust . also observe oscillations of prices and apparently chaotic behaviour of trust . with the chosen value of @xmath51 ( reluctance for the agent to lower the prices ) the average price
is maintained around a constant value , but with increasing @xmath51 the average price diminishes , because one is less reluctant to lower its prices .
if all prices are set to the same value , the dynamical behaviour of @xmath38 turns out to be less chaotic , and shows less variations . in fig .
[ fig3 ] the behaviour of the average reputation of sellers and price of goods are depicted .
here we see that the mean value of reputation of agents decreases with time , which reflects the fact that there are less successful transactions , while the reputation of some selected individuals increases with time . in order to assess the effect of including trust in the transactions
, we made a calculation in which no trust variables were included , that is @xmath72 and @xmath73 , @xmath74 , and another calculation including the dynamics of the trust variables . in fig .
[ fig4 ] we show a representation of the network as in fig . [ fig1 ] and a 2d graph of the two calculations .
here we observe that trust has two main effects : 1 ) it reinforces the trade network , the number of active links is increased noticeably , while in the calculation without trust ( indiscriminate trading ) the network is dismembered in isolated subgraphs with very few active links .
2 ) the wealth distribution between agents turns out to be more fair , in the sense that there are less poor people and quite a large proportion in the middle class .
the backbone network , in which active trading took place , is noticeably larger and with more `` black '' or active interactions ( @xmath16 and @xmath66 ) , the dramatic effect of trust is : a disconnected trading network without trust becomes a robust and connected network when trust is affecting agents in deciding transactions .
we have also observed that in the calculations with trust the state variables remain positive or zero , while in the calculations without trust a small number of agents have negative values .
this means that these agents are not only poor but in debt but the number of such agents is very small , such that on average there are 160 agents with negative values in a population of 5000 agents after 200000 time steps . as one of our main research foci
we compare the distribution of wealth calculated from our model with that of the actual statistical data . for comparison
we depict in fig .
[ fig5a ] the histograms of the actual wealth distribution in the usa ( in blue ) , together with our results from 10 realisations of the network with 500 agents , without and with trust included .
it should be noted that our model predicts a concentration of wealth in the hands of few agents , regardless of the role of trust , although the concentration of such agents is much less pronounced than what we see in reality . in the calculation including trust the distribution amongst middle and bottom percentages agree quantitatively with the data , but we are not able to find a good match with data for the upper 40@xmath75 of the wealth .
this could be due to the small size and randomness of our model network , nevertheless the trend is already noticeable . and the spread @xmath62 . ] in ref .
@xcite one finds an interesting exercise in which people are asked to construct wealth distribution that they consider ideal , and also an estimated distribution , based on their information , and these data are compared with actual data for wealth distribution in the usa . in fig .
[ fig5b ] we compare the published results with our calculations . here
we find that our calculation results seem to compare very well with the ideal distribution when trust is included and no spread of prices is allowed .
this situation probably represents a country that has strict control of unique prices for goods , set by the government or some monopoly .
the estimated distribution is well reproduced with a rather small spread of prices and having trust included . however , the actual situation is somewhat disappointing , since the best fit is with a large spread of prices . in the bottom right panel
we show the situation in which the network agents lack trust in their transactions thus rendering the outcome unrealistic .
the situation is quite different for such egalitarian societies as denmark , in which case we have found that for very small spread of prices and including trust our model agrees extremely well with the actual data ( taken from @xcite ) of the income distribution for the 1992 statistics . using data from the money variable
only , we show the comparison in fig .
[ fig6 ] , which turns out to be very favourable . .
] another maybe better way to compare our results with real statistical data is to investigate the distribution of money in different systems , since it depends on the actual mechanisms of acquiring money .
for instance , there are data on the annual income of people in europe , which ranges from 0 to millions of euros . in fig .
[ fig7](a ) we compare the actual data ( in red ) with a histogram from our numerical calculation with 1000 agents but without including trust . in fig .
[ fig7](b ) we compare the same data with the results of a calculation with 250 agents and a dispersion of prices twice as large as in ( a ) but once again without including trust .
both these calculations do not seem to fit with the real data , neither do averages over many realisations .
+ however , a rather different situation is encountered in a closed system as we have found out when we investigated data from the annual salaries that all the players in the nfl earned in two different years . in fig .
[ fig7](c ) we show in red the distribution of salaries in 1998 and compare it with a numerical result without including trust . in fig . [ fig7](d ) we show the numerical results for the same system including trust and compare them with the nlf data from 2011 . here
we can observe that both distributions fit the data fairly well , which allows us to think that trust does not play much of a role in the mechanisms of deciding salaries in a system like nfl , in which a few of the star players began to receive exaggeratedly good salaries . in order to investigate such a situation in more detail
we have adopted the approach presented by jun - ichi inoue et al .
@xcite , in which they define indices to measure social inequality in various fields , including income and trading . in fig .
[ fig8 ] we present the inequality for our model calculations .
we see that these lorentz inequality curves vary quite sensitively with the spread of prices of the goods .
we also see that the real situation presents itself as rather unequal such that for the same spread of prices , the lack of trust generates more inequity .
for the curve tagged with an arrow we found @xmath76 and @xmath77 , which is very similar to the indexes calculated for usa : @xmath78 and @xmath79 @xcite .
we have also detected that the results do not vary much when the size of the network is increased to 1000 .
as for the dynamical behaviour of trading there are data available for the wealth share of various countries @xcite . in fig .
[ fig9 ] we show the top 1 , 5 , and 10% countries income share , and compare it with the numerical calculations for a network size of 500 agents .
all the calculations were set to run up to 200000 iterations , of which only half of them were selected for the comparisons .
the parameters of the model @xmath80 and @xmath50 were varied to find the best fit .
for all the cases a value of @xmath63 in the model calculation is found to fit well with the data , except in the case of italy , where we chose @xmath81 for the best fit .
it is interesting to notice that the case of china is the only one that fits better with the 100000 initial iterations of the calculation , and the dispersion of prices is smaller ( @xmath82 ) .
all the other countries are best fitted once the variables attain a final distribution , which occurs during the last 100000 iterations .
this could reflect the fact that china is a newly emergent economic power and that their rules of trading are tighter . for japan and australia
the dispersion of prices is larger than for china ( @xmath83 ) .
this could be interpreted such that these countries have more free trading rules , in which case one could vary prices more widely without loosing competitiveness . for developed countries with long history in economic traditions
, @xmath62 is quite large , probably reflecting the influence of many strategies to allow prices to vary without the loss of competitiveness .
for instance the dispersion of gas prices nationally in the usa ranges from 1.31 to 2.37 dollars per gallon , and the prices are unevenly distributed geographically @xcite .
observe that usa and canada are practicable indistinguishable , which is to be expected as they are similar and tightly linked .
the case of italy , the only european country selected , is interesting , since it is the only one in which a good fit is obtained by increasing the time scale for price changes .
this seems to suggest that in italy and similar european countries prices tend to change more slowly than in the very dynamic american economies .
interestingly , the fit for the 1% top population is not as good as the others , meaning that our model predicts less extremely rich people than in reality there is .
the extreme social and economic inequality of the present world is most likely due to the fact that the economic alliances are not random ( as in our model ) and impose some trading preferences , other than the ones considered in our model .
also , our approximation of a conserved system result in a constraint on the amount of goods or money an individual could gather , and this constraint is not present in the actual economic picture , in which money could be printed and goods could be produced and destroyed . in the real world countries can not be considered as closed systems , although the global economy could be considered as such .
in conclusion our agent - based trading model gives rise to results that overall seem to compare very favourably with the findings from the real data , even though the model takes into account only a subset of the known factors affecting trading .
one striking result is the effect of trust in trading relations .
first of all , it was found that trust reinforces trading transactions in such a way that the network with active links remains fully connected when trust is included , and becomes a set of disconnected graphs if it is not .
secondly , trust helps to make society more even and it is seen that the distribution of wealth is fairer when trust plays a role in trading . if trust is not included then the society seems to have a number of poor people in debt , unlike in a society with trust .
one important conclusion of this work is that even in the simplified case of having a conserved system , agreement with real data on the distribution of wealth is not only possible but also quite good .
this could be interpreted indicating that including the production and deterioration of goods and money , which is essential to the idea of creation of wealth @xcite , does not seem to be a fundamental property of the economy in general .
furthermore , as far as the distribution of wealth is concerned , the fundamental issue seems to be the spread of prices , rather than the production of wealth .
it also turns out that our model predicts inequality in a closed economy without production from simple rules of buying and selling between agents , which illustrates the fact that inequality can arise naturally in a rudimentary economy .
it should be noted that contrary to the general trend in classical economic models to consider various representative classes of agents ( consumers or producers ) @xcite we have here considered a single class of heterogeneous trading agents .
the behaviour emerging from such economy is an aggregate of individual decisions .
hence it seems that an unbalanced economy emerges from particular decisions of individuals @xcite .
it is also important to recognise that the structure of real trading networks is more similar to a scale free networks than to random networks , which stays fixed during the dynamic trading process . in order to test the effect of topologically changing network structure we have introduced a rewiring scheme in which the agents whose connections are not working
are deleted from the network , and new agents are added following a scale free method . as a result
we have found that the final network structure evolved after a long run of the dynamics resembling roughly a scale free network .
it is found that the structure of the final network still depends very much on the trust variable @xmath38 , namely if @xmath84 for all agents , then agents that become `` hubs '' at a certain time are very likely to be deleted , and eventually the network is fragmented into small isolated trading groups . on the other hand , if one includes @xmath38 in the dynamics , the hubs that are formed remain trading and the network , although changing , remains cohesive .
the network that one obtains after many rewiring processes has always all the links working perfectly ( black lines ) .
however , the wealth distribution in the network is very similar to the ones reported here without rewiring , and thus we decided to leave the detailed study of the rewiring problem for the future , since the above - described rewiring scheme breaks the conservation of wealth , which is an essential assumption in our current model .
our justification to do so is that in the calculations shown here the time span for transactions is not long enough to rewire the network .
+ as the main message of this work we would like to suggest that in trade the prices seem to be the main cause of impoverishment
. wide spread of prices tends to augment the differences between poor and rich people , and the results seem much more sensitive to a change in the allowed spread of prices than to the average price .
also we conclude that trust seems to be important in regulating trading , since it is a way in which agents decide to trade preferentially amongst the agents they are linked with . in indiscriminate transactions without including trust , the network turns out to be disrupted , while with trust few links are always reinforced . in this way trust could be considered to favour the appearance of monopolies , since the agents that have a good history seem better off regardless of the high prices and quality of goods , and are thus in position of engulfing small traders . as a final remark
we could conclude that our simple model conserving the amount of goods and money is able to reproduce some salient features found in trading networks , with the additional advantage that we could fairly easily add new features to the model and analyse them in depth .
rb , erg , and tg would like to acknowledge financial support from conacyt ( mexico ) through project 179616 and kk from academy of finland through project 276439 .
we are grateful to prof .
larissa adler , whose original ideas were the primary source of inspiration for this work .
we are also grateful to prof .
juan m. hernndez for careful reading of the manuscript and valuable comments and criticisms .
chakraborti a , challet d , chatterjee a , marsili m , zhang y , chakrabarti bk , _ statistical mechanics of competitive resource allocation using agent - based models _ , physics reports , * 552 * , 1 - 25 ( 2015 ) , issn 0370 - 1573 , doi:10.1016/j.physrep.2014.09.006 inoue j , ghosh a , chatterjee a , chakrabarti bk , _ measuring social inequality with quantitative methodology : analytical estimates and empirical data analysis by gini and indices _ , physica a * 429(1 ) * 184 - 204 ( 2015 ) issn 0378 - 4371 , doi:10.1016/j.physa.2015.01.082 | we present a simple dynamical model for describing trading interactions between agents in a social network by considering only two dynamical variables , namely money and goods or services , that are assumed conserved over the whole time span of the agents trading transactions .
a key feature of the model is that agent - to - agent transactions are governed by the price in units of money per goods , which is dynamically changing , and by a trust variable , which is related to the trading history of each agent .
all agents are able to sell or buy , and the decision to do either has to do with the level of trust the buyer has in the seller , the price of the goods and the amount of money and goods at the disposal of the buyer . here
we show the results of extensive numerical calculations under various initial conditions in a random network of agents and compare the results with the available related data . in most cases the agreement between the model results and real data turns out to be fairly good , which allow us to draw some general conclusions as how different trading strategies could affect the distribution of wealth in different kinds of societies .
+ : social networks , agent - based model , wealth distribution , nonlinear dynamical systems , price effects , trust , reputation + |
observations of fluctuations in the temperature of the cosmic microwave background ( cmb ) are now providing us with a direct view of the primordial inhomogeneities in the universe .
the power spectrum of temperature fluctuations yields a wealth of information on the nature of the primordial perturbations , and the values of the cosmological parameters .
mapping the polarization of the cosmic microwave sky is an important next step , offering a great deal of complementary information , especially regarding the character of the primordial inhomogeneities @xcite .
one of the most interesting questions to resolve is whether the primordial perturbations possessed a tensor ( gravitational wave ) component , as predicted by simple inflationary models . here ,
polarization measurements offer a unique probe @xcite .
polarization of the cosmic microwave sky is produced by electron scattering , as photons decouple from the primordial plasma .
linear polarization is produced when there is a quadrupole component to the flux of photons incident on a scattering electron .
scalar ( density ) perturbations generate an ` electric ' ( gradient ) polarization pattern on the sky due to gradients in the velocity field on the surface of last scattering . for scalar perturbations
the velocity field is curl - free , and this leads directly to the production of an entirely ` electric ' pattern of linear polarization .
in contrast , tensor perturbations ( gravitational waves ) produce polarization by anisotropic redshifting of the energy of photons through decoupling . in this case
the polarization has ` magnetic ' ( i.e. curl ) and ` electric ' ( i.e. gradient ) components at a comparable level .
a magnetic signal can also be produced by weak lensing of the electric polarization generated by scalar modes .
detection and analysis of the lensing signal would be interesting in itself , but a detection of an additional tensor component would provide strong evidence for the presence of primordial gravitational waves , a generic signature of simple inflationary models . detecting or excluding a magnetic component is clearly of fundamental significance in cosmology .
but there is a significant obstacle to be faced .
the problem is that for the foreseeable future , the primordial sky polarization will only be observable over the region of the sky which is not contaminated by emission from our galaxy and other foreground sources of polarization .
thus we shall only be able to measure the polarization over a fraction of the sky .
but the electric / magnetic decomposition is inherently _ non - local _ , and _ non - unique _ in the presence of boundaries . to understand this ,
consider the analogous problem of representing a vector field @xmath0 ( in two dimensions ) as a gradient plus a curl : @xmath1 the electric and magnetic components respectively . from this equation
, one has @xmath2 , and @xmath3 . for a manifold without a boundary , like the full sky
, the laplacian may be inverted up to a constant zero mode , and the two contributions to @xmath0 are uniquely determined .
but for a finite patch , one can always think of adding charged sources for the potentials @xmath4 and @xmath5 outside of the patch on which @xmath0 is measured , which alter @xmath4 and @xmath5 without changing @xmath0 .
for example one can add to @xmath4 and @xmath5 pieces with equal but perpendicular gradients so there is no net contribution to @xmath0 .
since full sky observations are unrealistic , so is the hope of a unique decomposition of the sky polarization into electric and magnetic components .
however , this does not at all mean that the hypothesis of a magnetic signal can not be tested .
one possibility is to construct a local measure of the magnetic signal by differentiating the measured polarization ( i.e. @xmath6 vanishes if @xmath0 is pure electric in the analogue example above ) , but this is problematic for noisy , sampled data .
a more promising alternative , which avoids differentiating the data , is to construct line integrals of the polarization @xcite .
for example , in the vector analogy above , any line integral @xmath7 is guaranteed to vanish if @xmath0 is purely electric . however
, the problem with these line integrals is that there are an infinite number of them , and they are not statistically independent .
one would therefore prefer a set of ` magnetic ' variables to which the ` electric ' component does not contribute , but which are finite in number and statistically independent , for a rotationally symmetric statistical ensemble . since polarization from a primordial scale invariant spectrum of gravitational waves
is predominantly generated on scales of a degree or so ( the angle subtended by the horizon at last scattering ) , we would expect to be able to characterize the cosmic magnetic signal by a set of statistically independent variables roughly equal in number to the size of the patch in square degrees
. however the signal within a degree or so of the boundary can not be unambiguously characterized as magnetic , and hence one loses a number of variables proportional to the length of the boundary .
the amount of information about the magnetic signal therefore increases as the patch area minus the area of this boundary layer . in this paper we shall find the set of observable ` magnetic ' variables explicitly for circular sky patches : the method may be generalized to non - circular patches if necessary . as mentioned above , the electric component of the polarization ( due primarily to scalar perturbations ) is expected to be much larger than the magnetic signal . therefore to detect the latter it may be useful to construct observables which suffer no electric contamination . we show how to construct such variables , and use them to estimate what magnitude of magnetic signal the planned planck satellite might be able to detect .
we also discuss the optimal survey size for future experiments aimed at detecting tensor modes via magnetic polarization , including the effects of ` magnetic noise ' due to weak lensing of the dominant electric polarization @xcite . even for observations that do not expect to detect the magnetic signal the magnetic - only observables are likely to be very useful in checking consistency of any residual polarization with noise or indeed in identifying foreground contamination .
they may also be useful for studying the small scale weak lensing signal @xcite . to construct variables that depend only on the electric or magnetic polarization
we integrate the polarization field over the observed patch with carefully chosen spin - weight 2 window functions .
we present a harmonic - based approach for constructing these window functions which is exact in the limit of azimuthally - symmetric patches .
the method is expected still to perform well for arbitrary shaped patches of the sky , but the separation will no longer be exact in that case . constructing the window functions with our harmonic method
automatically removes redundancy due to the finite size of the patch , keeps the information loss small ( except for very small patches ) , and ensures that for idealized noise in the polarization map ( isotropic and uncorrelated ) , the noise on the electric and magnetic variables preserves these properties . in this respect the construction is analogous to the orthogonalized harmonics approach used in the analysis of temperature anisotropies @xcite .
however in the polarized case there is no simple interpretation in terms of a set of orthogonalized harmonics . in ref .
@xcite it was shown how the lossless quadratic estimator technique can be applied to polarization . there , no attempt was made to separate the electric and magnetic contribution to the estimators , so the resulting window functions for the power displayed considerable leakage between the electric and magnetic modes .
the authors of ref .
@xcite showed how the leakage could be reduced , but it is arguably still too large to allow robust estimation of the magnetic signal in the presence of an electric signal that is orders of magnitude larger .
we are able to perform a much cleaner separation at the level of the harmonic components in the map , and as we shall see the information loss in our approach is quite small for full sky surveys with a galactic cut .
the electric - magnetic decomposition of the polarization field is exactly analogous to the corresponding decomposition of projected galaxy ellipticities induced by weak lensing @xcite .
@xcite shows how to construct local real - space correlation functions for measuring the magnetic component .
these are useful for distinguishing the purely electric signal due to gravitational lensing from intrinsic correlations in galaxy alignments , and the method has the advantage of working for arbitrarily shaped regions of sky .
however the method assumed a flat sky approximation , and includes only the two - point information .
for polarization observations the sky curvature will be important and we aim to extract a set of statistically independent observables that contain as much of the magnetic information as possible .
this may also prove useful for weak lensing studies .
the paper is arranged as follows . in sec .
[ sec : ebpol ] we present the spin - weight 2 window technique for separating electric and magnetic polarization on the sphere , generalizing results in refs .
section [ sec : sep ] describes our harmonic - based technique for constructing window functions with the properties required to ensure separation of the electric and magnetic modes while keeping information loss small .
classical techniques for testing the hypothesis that there is no magnetic signal are discussed in sec .
[ sec : hyp ] , and estimates of the detection limits with the planck satellite and future experiments are also given .
lossless methods for estimation of the polarization power spectra are contrasted with methods using the separated variables in sec .
[ lossless ] . in a series of appendices we outline our conventions for spin weight functions and their spherical harmonics .
in addition we present a number of the standard integral theorems on 2-dimensional manifolds in convenient spin weight form , and present recursive methods for the fast computation of the spin weight spherical harmonics and their inner products over azimuthally symmetric patches of the sphere .
a further appendix discusses the statistics of detecting weak signals from tensor modes .
the observable polarization field is described in terms of the two stokes parameters @xmath8 and @xmath9 with respect to a particular choice of axes about each direction on the sky . in this paper
we take these axes to form a right - handed set with the incoming radiation direction ( following ref .
the real stokes parameters are conveniently combined into a single complex field that represents the observed polarization @xmath10 the values of stokes parameters depend on the choice of axes ; since @xmath8 is the difference of the intensity in two orthogonal directions it changes sign under a rotation of @xmath11 .
the @xmath8 field is related to the @xmath9 field by a rotation of @xmath12 .
more generally under a right - handed rotation of the axes by an angle @xmath13 about the incoming direction the complex polarization transforms as @xmath14 and is therefore described as having spin minus two ( see appendix [ app : eth ] for our conventions for spin weight functions ) .
the analysis of polarized data is therefore rather more complicated than for the temperature which does not depend on making a choice of axes in each direction on the sky .
as described in appendix [ app : eth ] , one can define spin raising and lowering operators that can be used to relate quantities of different spin @xcite .
the spin raising operator is denoted @xmath15 and the lowering operator @xmath16 .
since the polarization has spin - weight -2 it can be written as the action of two spin lowering operators on a spin zero complex number @xmath17 the underlying real scalar ( spin - zero ) fields @xmath18 and @xmath19 describe electric and magnetic polarization respectively @xcite .
they are clearly non - local functions of the stokes parameters .
one can define a spin zero quantity which is local in the polarization by acting with two spin raising operators .
using some results from appendix [ app : eth ] one obtains @xmath20 where @xmath21 is the covariant derivative on the sphere .
the real and imaginary parts of this equation can therefore be used to determine the electric and magnetic parts of the polarization . performing a surface integral we define @xmath22 where @xmath23 is a complex window function defined over some patch @xmath24 of the observed portion of the sky .
it follows that @xmath25 provide a measure of the electric and magnetic signals .
note that @xmath26 with an equivalent result for @xmath27 . using the integral theorem in appendix [ app : eth ] we can write @xmath28 where @xmath29 is now a spin 2 window function
, @xmath30 is a spin @xmath31 window function , and @xmath32 is the spin 1 element of length around the boundary @xmath33 of @xmath24 . clearly we do not wish to take derivatives of noisy observed data and hence it is usually useful to choose the window function to eliminate the derivative terms on the boundary . for cmb polarimetry
we are interested in the polarization defined on the spherical sky .
the surface integrals vanish if we choose @xmath23 such that @xmath34 , which will be true if @xmath23 is a linear combination of the spherical harmonics with @xmath35 or 1 , since these possess no spin 2 component .
if we then set @xmath36 on the boundary , so as to eliminate the derivatives of the polarization , we are forced to consider circular patches @xmath24 , in which case a combination of the two @xmath37 harmonics works .
this implies that the electric and magnetic signals can be probed by performing line integrals around circles , as emphasized in refs .
these line integrals can be performed around any circle that is enclosed in the observed region of the sky , and it is unclear how to obtain a complete set of statistically independent observables in order to extract all of the available information . also for current experiments , performing one - dimensional line integrals on pixelized maps is unlikely to be a good way to extract information robustly . in this paper , we suggest choosing the window functions so that the line integrals around @xmath33 that appear in the construction of @xmath38 and @xmath27 contain no contribution from the magnetic and electric polarization respectively . in the absence of special symmetries ( see below for exceptions that arise in the case of circular patches ) this requires that @xmath23 , @xmath39 , and @xmath40 all vanish on the boundary .
these conditions are equivalent to demanding that the window function and its normal derivative vanish on @xmath33 . with such a choice of window
we can measure the electric and magnetic signals using only the surface integrals @xmath41 since the window functions are scalar functions on the sphere we can expand them in spherical harmonics , @xmath42 ( the square root factor is included for later convenience . )
we need not include @xmath35 and 1 spherical harmonics since they do not contribute to the spin - weight @xmath43 window functions , and the boundary integral terms automatically separate for these multipoles . in practice , we are only interested in probing scales to some particular @xmath44 ( e.g. the magnetic signal from tensor modes has maximal power for @xmath45 and decreases rapidly with @xmath46 ) , so the sum in eq .
( [ eq : hwindow ] ) can be truncated at some finite @xmath44 .
we shall focus on the case where the observed sky patch is azimuthally symmetric in which case the construction of exact window functions becomes particularly simple .
the harmonic - based method we describe in sec .
[ sec : sep ] provides a practical solution to constructing a non - redundant set of window functions that separate the electric and magnetic modes exactly .
in addition , for the special case of isotropic , uncorrelated noise on the observed polarization , these simple properties are preserved in the variables @xmath47 and @xmath48 . for observations over non - azimuthally symmetric patches our method can , of course , be used over the largest inscribed circular patch , but in this case there is inevitable information loss since we use only a subset of the observed data .
however , we expect that the method presented in sec .
[ sec : sep ] could also be applied directly to the full observed region to construct window functions that achieve approximate separation of electric and magnetic polarization .
consider the case of an azimuthally - symmetric patch so the boundary @xmath49 consists of one or two small circles .
for each azimuthal dependence on @xmath50 we can construct combinations @xmath51 that satisfy the necessary boundary conditions .
for @xmath37 it is easy to see that @xmath47 and @xmath48 contain no contribution from @xmath19 and @xmath18 respectively for any choice of the @xmath52 [ i.e. the boundary integrals that distinguish @xmath47 ( @xmath48 ) from @xmath38 ( @xmath27 ) vanish if the polarization is pure magnetic ( electric ) ] .
it follows that for @xmath37 there are @xmath53 linearly independent window functions that satisfy the boundary conditions . for @xmath54
it will be shown in the next section that there is only one independent linear constraint per boundary circle , so there are @xmath55 possible window functions ( @xmath56 for a boundary composed of two circles ) . for @xmath57
there are two linear constraints per boundary circle which can be taken to be the vanishing of @xmath58 and its normal derivative . in this case
there are @xmath59 ( @xmath60 ) window functions for boundaries consisting of one ( two ) small circles .
since we are only considering a fraction of the sky not all of the window functions counted above may return observables @xmath47 and @xmath48 containing independent information .
this arises because for large @xmath44 , or small patches , there will generally arise non - zero window coefficients @xmath61 that produce spin 2 window functions that are poorly supported over the patch .
( see e.g. ref .
@xcite for a discussion of the equivalent problem in the case of scalar functions . ) the redundancy in the set of acceptable window functions can be removed by expanding the spin 2 window functions in a smaller set of functions which are ( almost ) complete for band - limited signals over the region @xmath24 .
the construction of such a set by singular value methods ( e.g. refs .
@xcite ) forms the starting point of the method we present in the sec .
[ sec : sep ] .
we construct window functions in harmonic space , so as a useful preliminary we consider the harmonic expansion of spin - weight 2 fields over the full sphere @xcite .
the polarization @xmath62 is spin @xmath31 and can be expanded over the whole sky in terms of the spin two harmonics ( see appendix [ app : harmonics ] for our conventions and some useful results ) @xmath63 reality of @xmath8 and @xmath9 requires @xmath64 , so that @xmath65 with an equivalent result for @xmath66 . under parity transformations @xmath67 but @xmath68 , since @xmath69 . from the orthogonality of the spherical harmonics over the full sphere
it follows that @xmath70 in a rotationally - invariant ensemble , the expectation values of the harmonic coefficients define the electric and magnetic polarization power spectra : @xmath71 if the ensemble is parity - symmetric the cross term is zero , @xmath72 .
the form of the harmonic expansion ( [ eq : hwindow ] ) of the window function ensures that the spin - weight @xmath43 windows are @xmath73 where the sum is over @xmath74 and @xmath75 .
evaluating the surface integrals in eq .
( [ eq : ewbw ] ) we find @xmath76 where the pseudo - harmonics are obtained by restricting the integrals in eqs .
( [ eq : elm ] ) and ( [ eq : blm ] ) to the region @xmath24 : @xmath77 , \label{nosepeqe } \\
\tilde{b}_{lm } & = & \frac{i}{2}\sum_{l'm'}\int_s \text{d}s\ , \left [ ( e_{l'm ' } - i b_{l'm'}){}_{-2}y_{l'm'}{}_{-2}y_{lm}^\ast - ( e_{l'm ' } + i b_{l'm ' } ) { } _ 2y_{l'm'}{}_2y_{lm}^\ast \right ] .
\label{nosepeqb}\end{aligned}\ ] ] defining hermitian coupling matrices @xmath78 where @xmath79 we can write @xmath80 in the limit @xmath81 , @xmath82 become projection operators as a consequence of the completeness of the spin - weight harmonics .
the matrix @xmath83 controls the contamination of @xmath47 and @xmath48 with magnetic and electric polarization respectively .
our aim is to construct window functions @xmath61 that remove this contamination for all @xmath84 and @xmath66 .
some elements of the matrices @xmath85 are shown in fig .
[ windows ] . for azimuthally - symmetric patches the coupling matrices
are block diagonal ( @xmath86 ) , and so window functions can be constructed for each @xmath50 separately [ see eq .
( [ eq : hwindowm ] ) ] . for @xmath37
we have @xmath87 so @xmath88 and we have clean separation for any azimuthally - symmetric window function .
the set of azimuthally symmetric window functions gives @xmath53 separated variables that contain the same information as would be obtained by computing line integrals around all those circles concentric with the boundary of the azimuthal patch . for general @xmath50
there is leakage of @xmath84 into @xmath89 ; for parity - symmetric cuts there is only leakage between modes with different parity ( i.e. for even @xmath46 the pseudo - harmonics @xmath89 depend on @xmath90 only for odd @xmath91 ) .
we showed in the previous section that , for a general window function , the contamination of e.g. @xmath47 by the magnetic polarization is due entirely to boundary terms .
this implies that @xmath83 can always be written as a line integral around the boundary of @xmath24 .
( we show in appendix [ app : ints ] that the matrices @xmath92 can be transformed into line integrals for @xmath93
. however @xmath83 can be written as a line integral for all @xmath46 and @xmath91 . ) making use of eq .
( [ eq : appcdint ] ) , it is straightforward to show that @xmath94\nonumber\\ & & \mbox{}+ \oint_{\partial s } { { } _ 1\text{d}l\:}\ , \left[\sqrt{l(l+1 ) } { } _ { -1}y_{lm}^\ast { } _ { -2}y_{l'm ' } + \sqrt{(l'-1)(l'+2 ) } y_{lm}^\ast { } _ { -1}y_{l'm ' } \right ] \bigr ) .
\label{eq : wminusint}\end{aligned}\ ] ] this can be put in manifestly hermitian form using the recursion relation derived from the action of @xmath95 on @xmath96 @xmath97 for a circular boundary at constant latitude @xmath98 ( i.e. the boundary of an azimuthal patch ) , we find @xmath99 , \label{eq : wminusdcmp}\ ] ] where the vectors @xmath100 , \\
v_l(m ) & = & \sqrt{\frac{(l-2)!}{(l+2)!}}\frac{\sqrt{(m^2 - 1)}}{\sin\theta } y_{lm}(\theta,\phi)\end{aligned}\ ] ] for @xmath74 and some arbitrary @xmath4 .
[ note that @xmath101 and @xmath102 will not generally be orthogonal so eq . ( [ eq : wminusdcmp ] ) is not the spectral decomposition of @xmath103 . ]
any window @xmath61 whose inner products with @xmath101 and @xmath102 both vanish , i.e.@xmath104 will achieve clean separation of electric and magnetic polarization . for @xmath105 such window functions and their normal
derivative necessarily vanish on the boundary . as noted earlier , for @xmath54
there is actually only one constraint to be satisfied which now follows from the fact that @xmath106 .
in this section we give a practical method for constructing a non - redundant set of window functions @xmath107 where @xmath108 labels the particular window , that achieve exact separation for azimuthal patches .
the corresponding ( cleanly separated ) electric and magnetic observables will be denoted @xmath109 and @xmath110 .
we will make use of a notation where vectors are denoted by bold roman font , e.g. @xmath111 has components @xmath110 , and @xmath112 has components @xmath66 , and matrices are denoted by bold italic font , e.g. @xmath113 have components @xmath85 .
we present the method in a form that is applicable ( though no longer exact ) to arbitrary shaped regions @xmath24 ; for azimuthal patches the method is exact . for the azimuthal case
all matrices are block diagonal and the window functions can be constructed for each @xmath50 separately . in matrix form ,
( [ eq : ewbw_harm ] ) is @xmath114 where @xmath115 is the matrix whose @xmath108th row contains the harmonic coefficients of the @xmath108th window function , and recall @xmath116 for an azimuthally - symmetric sky patch the block - diagonal matrices @xmath117 ( with components @xmath118 ) from which @xmath113 are constructed can be computed very quickly using the recursion relations given in appendix [ app : ints ] .
alternatively , @xmath119 can be computed directly from eq .
( [ eq : wminusdcmp ] ) . in the limit of full sky coverage @xmath120 and @xmath121 .
we know that for @xmath122 , the range of @xmath50th submatrix of @xmath119 is two - dimensional [ spanned by @xmath101 and @xmath102 ] , so that all but two of the eigenvalue of the submatrix are zero .
equivalently , all but two linear combinations of the @xmath89 are independent of @xmath90 .
the @xmath54 submatrices of @xmath119 have only one non - zero eigenvalue ; the associated eigenvectors are @xmath124 .
the @xmath37 submatrix is identically zero .
the essence of our method for constructing the window functions is to choose @xmath125 to project out of the range of @xmath119 .
we first diagonalize @xmath126 by performing a singular value decomposition @xcite . here
, @xmath127 is a positive ( semi-)definite diagonal matrix whose elements are the eigenvalues of @xmath128 .
the columns of the unitary matrix @xmath129 are the normalized eigenvectors of @xmath128 .
the singular value decomposition allows us to identify the linear combinations @xmath130 that are poorly determined by @xmath131 those corresponding to the small diagonal elements of @xmath127 .
the eigenvectors with very small eigenvalues correspond to polarization patterns that essentially have no support inside the observed patch of the sky , and would lead to a set of redundant window functions if not removed from the analysis .
the distribution of eigenvalues of @xmath128 is approximately bimodal as illustrated in fig .
[ evals ] , and the exact definition of ` small ' is not critical when considering the range of the matrix .
this bimodality arises because @xmath132 are approximately projection operators for large @xmath44 , and the fact that the range of @xmath119 is a rather small subspace . to remove redundant degrees of freedom from the spin 2 window functions ,
we define an operator @xmath133 which projects onto the eigenvectors of @xmath128 whose eigenvalues are close to one .
this amounts to removing the appropriate columns of @xmath129 .
since @xmath129 is orthogonal , @xmath133 is column orthogonal and hence @xmath134 ( but @xmath135 ) .
the matrix @xmath136 is the corresponding smaller square diagonal matrix , and we have @xmath137 we now multiply by @xmath138 , defined by @xmath139_{ij } \equiv \delta_{ij } [ { \tilde{{\bm{d}}}}_+]_{ii}^{-1/2}$ ] , to give @xmath140 including the factor @xmath141 in the window functions @xmath125 is equivalent to constructing the spin 2 window functions from a reduced basis that is orthonormal and ( almost ) complete over the region @xmath24 . the remaining step is to project out the unwanted boundary terms that contaminate @xmath131 with @xmath142 . for the case of an azimuthal patch ,
the ranges of the submatrices of @xmath119 are all two dimensional ( or lower ) .
it follows that the same is true of @xmath143 , so we can remove the unwanted boundary term by ensuring that @xmath125 also includes a factor that projects out of the range of @xmath143 . in practice
, we perform this projection by constructing its singular value decomposition , which for a non - square matrix takes the form @xmath144 . here , @xmath145 is a diagonal matrix with the same dimension as @xmath136
, @xmath146 is a unitary matrix of the same dimension , and @xmath147 is a column orthogonal rectangular matrix .
there are at most two non - zero singular values ( elements of the diagonal matrix @xmath145 ) per @xmath50 , and the corresponding left singular vectors ( columns of @xmath146 ) form an orthonormal basis for the range of @xmath143 .
we can project out of this range by defining @xmath148 as the matrix obtained by removing the columns of @xmath146 where the corresponding singular value is non - zero .
thus , choosing the window functions @xmath149 we guarantee separation of the electric and magnetic polarization for azimuthally - symmetric patches .
our separated polarization observables become @xmath150 for azimuthal patches the separation is exact ; the approximation sign arises only from our use of @xmath151 in simplifying the matrix that premultiplies @xmath142 and @xmath112 in eqs .
( [ eq : bvar ] ) and ( [ eq : evar ] ) . for observations over non - azimuthal patches
, one can either apply the exact separation over the largest inscribed azimuthal region or attempt to apply the method outlined above to the entire patch . in the latter case the structure of @xmath119 is less clear , but we can still expect a significant number of its eigenvalues to be very small ; the associated eigenvectors correspond to window functions that satisfy @xmath152 on the boundary .
the number of independent window functions that achieve separation can be estimated as the number of pixels of linear size @xmath153 contained in the patch ( roughly the number of modes that survive the diagonalization of @xmath128 ) minus twice the number of pixels on the boundary ( roughly the number of constraints in setting @xmath23 and @xmath39 to zero on the boundary ) .
such window functions will only give an approximate separation of electric and magnetic polarization .
in practice , the accuracy of the separation , and the number of independent window functions constructed , will depend on the choice of threshold for retaining the singular values of @xmath143 .
if we are only interested in constructing variables that depend on the magnetic polarization , the maximum multipole @xmath44 in the window functions can be chosen rather small ( of the order of a few hundred ) .
the relation @xmath154 will only hold for square @xmath119 if @xmath44 is chosen to include all the significant power in the electric polarization , so for smaller @xmath44 we can not assume that @xmath119 is square .
however , for azimuthal patches , the range of each submatrix of @xmath119 is still guaranteed to be of dimension two or less , irrespective of its shape , and so the exact separation can proceed with @xmath119 treated as square . for non - azimuthal patches it would be prudent to monitor the effect of varying the number of columns in @xmath119 on the range of this matrix .
it is straightforward to project the errors on the stokes parameters to find the noise in the separated variables @xmath155 and @xmath111 . in this section
we consider the simple case of maps with idealized noise properties .
we assume that the noise correlation between pixels is negligible , and that the noise on the stokes parameters @xmath156 and @xmath157 is un - correlated .
the neglect of noise correlations between pixels amounts to assuming white noise in the time stream of the measurement . in principle , correlations between stokes parameters can be eliminated with a careful choice of polarimeter directions in the experiment @xcite . with these assumptions , we have @xmath158 and the noise correlation is given by @xmath159 it follows that the noise @xmath160 on the pseudo - multipoles @xmath131 has correlations @xmath161 - { { \textstyle \frac{1}{2}}}\int_s \text{d}s\ , ( \sigma^2_q - \sigma^2_u){{\textstyle \frac{1}{2}}}[{}_{-2 } { \mathbf{y}}^\ast ( { } _ { 2}{\mathbf{y}}^\ast)^\dag + { } _ 2 { \mathbf{y}}^\ast ( { } _ { -2 } { \mathbf{y}}^\ast)^\dag ] , \label{eq : noise1}\ ] ] with a similar result for the noise @xmath162 on @xmath163 . here , vectors of spin - weight @xmath164 functions @xmath165 have components @xmath96 . the polarimeter arrangements that give uncorrelated errors between stokes parameters also ensure that @xmath166 so that the last integral in eq .
( [ eq : noise1 ] ) is zero .
here we concentrate on the simple case where @xmath167 is isotropic in which case @xmath168 the covariance of the noise @xmath169 on @xmath111 is therefore given by @xmath170 and hence the noise is diagonal ( and isotropic ) ; similarly @xmath171 .
what is more , the noise on @xmath155 and @xmath111 are uncorrelated for isotropic noise since @xmath172 and hence @xmath173 as @xmath174 annihilates @xmath175 . for isotropic noise
our polarization variables therefore have the same desirable diagonal properties as the scalar diagonalized harmonic coefficients used in the analysis of the cut - sky cmb temperature @xcite . in the presence of a symmetric beam , white noise in the time stream of the experiment projects to white ( though generally non - isotropic ) noise on the beam - convolved polarization field . in multipole space ,
the convolved fields have electric and magnetic multipoles that are related to the unconvolved @xmath84 and @xmath66 by spin - weight 2 beam window functions @xmath176 @xcite . for @xmath44 of a few hundred ,
appropriate for probing magnetic polarization , the beam window functions will be negligible for experiments with resolution much better than one degree . for lower resolution experiments , or higher @xmath44 , the effect of the beam window function should be included in the theoretical ( signal ) covariance of the variables @xmath155 and @xmath111 ( see below ) .
in general , non - uniform coverage of the sky will lead to variations in @xmath167 . in this case
it is still possible to define harmonic variables @xmath177 that have isotropic noise .
( here @xmath178 where @xmath179 , @xmath180 is unitary and @xmath181 is diagonal . ) for azimuthal patches with general ( everywhere finite ) noise patterns we can still construct the magnetic - only variables for each value of @xmath50 but the noise will now couple variables with different @xmath50 ( unlike the signal ) . while presenting no fundamental obstacles , this @xmath50-mode coupling does increase the computational overhead considerably .
sky patches of general shape are equivalent to azimuthal patches with regions of infinite noise , and the general case of a non - azimuthally symmetric survey region with arbitrary noise can be treated by re - defining the pseudo - harmonics @xmath163 , @xmath131 and coupling matrices @xmath128 , @xmath119 with a factor of @xmath182 inside the integral .
all the above results then follow with the new definitions , though computing the matrices and manipulating them may become computationally challenging .
the assumption of isotropic noise is therefore not fundamental to our analysis , and the following results could be generalized for more realistic situations .
the expected magnetic signal correlation matrix is given by @xmath183 where the diagonal magnetic power spectrum matrix is given by @xmath184_{(lm)(l'm')}=\delta_{m m'}\delta_{l l'}c_{l}^{bb}$ ] .
since the noise correlation is proportional to the identity matrix for isotropic noise we can perform any rotation @xmath185 , where @xmath186 is unitary , and still have a set of variables with uncorrelated errors .
the rotated variables are derived from window functions @xmath187 . for a particular theoretical model we can rotate to the frame where the signal matrix is diagonal .
the rotated @xmath188 will then be fully statistically independent . in fig .
[ realspace ] we plot the window functions for the @xmath188 which give the largest contributions to the signal for a typical flat @xmath189cdm model with a scale invariant tensor initial power spectrum and no reionization .
the window functions are plotted as line segments of length @xmath190 at angle @xmath191 to the @xmath192 direction where the real quantities @xmath193 and @xmath194 are defined in terms of the real part @xmath195 of the ( rotated ) scalar window function @xmath196 as @xmath197 this definition ensures that @xmath198 for the imaginary part of @xmath188 , @xmath193 and @xmath194 should be defined as @xmath199 . for the case of azimuthal patches , as considered in fig .
[ realspace ] where the windows are constructed for each @xmath50 , the imaginary part would produce a plot that is rigidly rotated by @xmath200 ( @xmath201 ) about the centre .
plotting the window functions in this form is useful since the length of the line segment gives the sampling weight assigned to that point , and the orientation gives the direction of the linear polarization that contributes at each point .
we could repeat the exercise for the @xmath155 in which case for the real part we would define @xmath202 . in fig .
[ contribs ] we show the signal to noise in the magnetic variables for two azimuthal patches .
as the patch size increases the signal in the modes with large @xmath50 also increases , reflecting the fact that for small patches the diagonalization of @xmath128 removes a greater relative fraction of the modes at each @xmath50 as @xmath50 increases .
for small patches of the sky most of the signal at each @xmath50 is compressed into a small number of modes , whereas for larger patches the signal is distributed more uniformly . for cosmological models with reionization
the signal for large patches is distributed less uniformly , with a small number of modes giving big contributions due to the greater large scale power .
we are now in a position to use the magnetic observable @xmath111 to constrain the magnetic signal without having to worry about contamination with the much larger electric signal .
the simplest thing to do would be to test the null hypothesis that the magnetic signal is due entirely to noise ( this hypothesis is unlikely to be ruled out pre - planck ) .
if the signal were not consistent with noise it could indicate various things : the presence of cmb magnetic polarization , the presence of polarized foregrounds , that havent been removed successfully , systematic leakage into the magnetic mode in the analysis ( e.g. due to unaccounted for pointing errors , or pixelization effects ) , or @xmath8-@xmath9 leakage in the observation ( e.g. due to unaccounted for cross - polarization in the instrument optics ) .
magnetic polarization can originate from tensor modes , but also by weak lensing of the scalar electric polarization @xcite .
the lensing signal should be dominant on small scales , and the magnetic variables could certainly be used to observe this signal . of more interest here is the larger scale contribution from tensor modes . in order to identify this component
we shall have to model the lensing contribution , which becomes increasingly important as one tries to observe smaller tensor contributions @xcite .
in the first three of the following subsections we assume that the magnetic signal is generated purely from the tensor modes , then in subsection [ lensing ] we show how our results can be adapted to account for the lensing signal .
if the noise and signal are gaussian the @xmath111 will be gaussian and the simplest thing to do is a @xmath203 test by computing @xmath204 ( for isotropic noise this is just @xmath205 ) . whilst the cmb magnetic polarization signal is from tensor modes
is expected to be gaussian , the lensing signal and any spurious or unexpected signal may not be .
one may therefore also wish to do a more sophisticated set of statistical tests at this point .
assuming that the signal is as expected any @xmath206 signal present is gaussian and would have a power spectrum as predicted for a near scale - invariant tensor initial power spectrum one can account for the expected form of the power spectrum and thereby increase the chance of a detection .
we assume that the main parameters of the universe are well determined by the time magnetic polarization comes to be observed , so the shape of the magnetic polarization power spectrum is known to reasonable approximation ( the only significant freedom arising from the shape of the primordial tensor power spectrum ) .
we compute the expected signal correlation @xmath207 for some particular tensor amplitude and say that the real signal is @xmath208 . assuming gaussian signal and noise the likelihood in this case is then given by @xmath209}{|{\bm{n}}+ r{\bm{s}}|^{1/2}}. \label{likelihood}\ ] ] the likelihood distribution can be computed numerically from the @xmath111 observed , and gives the posterior probability distribution on the value of @xmath210 after multiplying by the prior @xmath211 . the magnetic signal is expected to be weak , and the detailed statistics for analysing such a signal are given in appendix [ app : stats ] .
there we show that @xmath212 gives a measure of the number of ` sigmas ' of the detection the number of standard deviations of the maximum likelihood @xmath213 from pure noise ( @xmath214 ) assuming low signal to noise .
we use this as a test statistic in monte - carlo simulations to compute detection probabilities at a given significance .
we have checked at isolated points that using optimal statistics gains very little except for very small sky patches ( where there are only a small number of magnetic modes , each of which must have fairly high signal to noise in order to get a detection ) . using the @xmath111 variables
is clearly not optimal as we have thrown away some well determined linear combinations of @xmath215 and @xmath206 .
however in the idealized situation considered here they should provide a robust way for testing for magnetic polarization .
the number of modes thrown away is in any case quite small not more than two per @xmath50 mode for azimuthal patches .
we quantify this information loss further in sec .
[ lossless ] . of the current funded experiments ,
only planck is likely to detect magnetic polarization if the levels are as predicted by standard cosmological models . as a toy model we consider the @xmath216 and @xmath217 ghz polarized channels of the planck high frequency instrument .
we approximate the noise as isotropic and ignore the variation of beam width ( 7.1 and 5.0 arcmin full width at half maximum respectively ) between these channels .
combining maps from these two channels with inverse variance weighting , we find @xmath218 , where @xmath8 and @xmath9 are expressed as dimensionless thermodynamic equivalent temperatures in units of the cmb temperature .
we apply an azimuthally - symmetric galactic cut of @xmath219 degrees either side of the equator . the expected magnetic polarization power spectrum peaks at @xmath220 , and there is therefore no need to consider high resolutions so we can use @xmath221 without significant loss of power . in fig .
[ planck_detect ] we show the probability of obtaining a detection with planck as a function of the true underlying scale - invariant tensor power spectrum amplitude @xmath222 ( defined as in ref .
@xcite ) assuming a standard flat @xmath189cdm model .
a tensor amplitude of @xmath223 would contribute about 1/10 of the large scale temperature @xmath224 detected by cobe , and is likely to be detected by planck if our model is at all realistic .
this corresponds to being able to detect the signal from inflationary models with energy scale at horizon crossing @xmath225 .
such models include the simple @xmath226 potentials , with @xmath227 . for a given detector sensitivity the magnitude of the signal that can be detected depends on the size of the sky patch that is observed .
the signal to noise in each observable increases in proportion to the observation time per unit area .
the noise covariance is proportional to @xmath228 which varies in proportion to the observed area for a given survey duration .
for large areas the number of observables varies approximately in proportion to the area , which would make the number of ` sigmas ' of a chi - squared detection scale with the square root of the area . combining these two effects
, the expected detection is therefore proportional to one over the square root of the area , and is larger if a fixed survey time is spent observing a smaller area .
however for smaller areas the signal to noise on each observable becomes larger , and the number of variables decreases . with fewer variables the probability of obtaining no detection increases significantly .
this is just the fact that if you observe a small patch of sky you have a larger chance of being unlucky and having a patch which has a small magnetic polarization signal everywhere .
also the existence of the boundary becomes increasingly important for small patches and a larger fraction of the information is lost in order to obtain clean separation of the magnetic observables .
the question of ` optimal ' survey size is somewhat delicate , as it depends on the probability distribution for the detection significance that one thinks is optimal . in fig .
[ patchsize_probs ] we plot the probability of detecting various tensor amplitudes at 95 per cent and 99 per cent confidence for different survey sizes . in fig .
[ patchsize ] we show the minimum gravitational wave ( tensor ) amplitude that might be detected at 99 per cent confidence as a function of the radius of the survey size .
it is clear that radii in the range @xmath229@xmath230 are optimal , though one can not be more precise without defining more specifically the aims of the observation .
a radius of about @xmath231 would be a good compromise between being able to place good upper limits if there was no detection ( which favours radii closer to @xmath230 ) and having a better chance of detecting small amplitudes ( which favours smaller radii ) . the solid curves in fig .
[ patchsize ] fully take account of the need to separate the magnetic signal from the ( much larger ) electric signal . by way of comparison ,
the dashed curves show the minimum detectable amplitude obtainable if one could do perfect lossless separation , which is clearly impossible on an incomplete sky ( see sec .
[ lossless ] ) . with lossless separation ,
the best upper bounds are obtained for smaller patches since the size of the boundary is no longer important .
the dashed curves in fig .
[ patchsize ] can be compared with those given in ref .
@xcite where perfect separation was assumed , the effects of finite sky coverage were treated only approximately , and a less rigorous approach to hypothesis testing was employed . ref .
@xcite gives an improved analysis along the lines of ref .
@xcite , and also performs calculations properly taking account of the mixing of electric and magnetic polarization through a ( brute - force ) fisher analysis in pixel space .
unlike most of the foreground signals that might contaminate the observation , the magnetic signal from the lensing of scalar electric polarization has the same frequency spectrum as the primordial magnetic signal and so can not be removed easily by use of multi - frequency observations . in order to isolate the tensor contribution to the magnetic signal we can incorporate knowledge of the expected lensing power spectrum @xcite into the null - hypothesis covariance matrix @xmath232 ( we neglect the non - gaussianity of the lensed polarization ) . for the multipoles @xmath233 of interest for the tensor signal
the lensing signal is approximately white , with @xmath234 if the cobe signal is entirely generated by scalar modes . for large patches of sky , where the matrix @xmath136 is nearly proportional to the identity matrix
, the lensing signal contributes like an additional isotropic noise with @xmath235 .
we have checked this approximation by computing the following results exactly in particular cases , with agreement to within a few percent for patch sizes with @xmath236 .
the effect of the lensing is therefore simply to increase the effective noise by a constant amount . for the planck satellite the effect is small , reducing the @xmath222 that could be observed by about @xmath237 per cent .
however for the smaller surveys with better sensitivity , considered in figs .
[ patchsize_probs ] and [ patchsize ] , the effect is much more important . for a one year survey of radius @xmath192 with sensitivity @xmath238 the noise variance
is given by @xmath239 where @xmath240 is the fraction of the sky which is observed and @xmath241 .
this noise gives the tensor amplitudes @xmath242 plotted in fig .
[ patchsize ] . incorporating the lensing effect the actual tensor amplitude one could detect in an experiment with sensitivity @xmath164 and duration
@xmath243 is @xmath244 where @xmath242 is the amplitude for a one year mission with @xmath238 and lensing ignored ( i.e. the amplitude plotted in fig .
[ patchsize ] ) .
this allows our previous results to be modified for inclusion of the lensing signal .
we have plotted the modified results for various survey sensitivities in fig .
[ patchsize_lens ] .
the optimal survey size now depends on the sensitivity as sensitivity improves the lensing signal becomes more important and one needs to survey larger scales in order to accurately measure the difference in variance expected with the tensor signal . for large patch
sizes the tensor amplitude @xmath222 that can be detected in the absence of lensing is proportional to @xmath245 . allowing for lensing
, there is an optimal survey size at @xmath246 [ if there is a solution with @xmath247 , in other words when the variance of the instrument noise is equal to the lensing signal .
there is a lower limit of @xmath248 that can be measured even with perfect sensitivity , when the tensor contribution can not be distinguished from random sampling variations in the lensing signal distribution .
this corresponds to an inflation model with energy scale @xmath249 , in broad agreement with ref .
@xcite for a three sigma detection .
this situation could only be improved if one could find ways to obtain information about the particular realization of the lensed signal .
we have assumed that component separation and source subtraction can be performed exactly so that the observed signal is only lensed cmb .
polarized thermal dust emission is expected to generate a significant magnetic signal @xcite at roughly the level shown by the line in fig . [ patchsize_lens ] at @xmath250 .
separation of this signal from the cmb signal should be possible with multi - frequency observations , and it should then not have a significant effect on our results .
we now compare the above analysis with truly lossless methods .
lossless , likelihood analysis for cmb polarization in pixel space has been considered recently in ref . @xcite . in this section
we consider lossless and nearly lossless methods in harmonic space .
a simple way to incorporate most of the magnetic signal for constraining the tensor modes is to use the unprojected variables @xmath251 , where @xmath136 and @xmath133 were defined in sec .
[ sec : sep ] .
the null - hypothesis covariance matrix @xmath232 can be computed including the expected signal from the electric polarization , and the analysis can be performed as before .
this is marginally superior to using the projected variables if the tensor amplitude is quite high , as shown in fig .
[ planck_detect ] for the planck mission . for smaller tensor amplitudes the entangled linear combinations of @xmath215 and @xmath206 modes
are dominated by the electric component and performing the projection looses very little .
using the projection gives one clean separation , and there is no need to know the electric polarization power spectrum . by identifying variables that depend only on the electric and magnetic signals at the level of the map we also do not need to assume gaussianity , so we could for example perform gaussianity tests on the two physically distinct polarization types independently .
we now consider the full joint analysis of the electric and magnetic polarization , with the pseudo - multipoles @xmath163 and @xmath131 forming our fundamental data vector @xmath252 since we are no longer worrying about @xmath215-@xmath206 separation so we can equally well use the block - diagonal frame where @xmath253 performing a singular value decomposition of the block - diagonal matrix @xmath254 ( where the matrices @xmath146 and @xmath145 should not be confused with those defined in sec .
[ sec : sep ] ) , we can identify the well determined linear combinations as before @xmath255 this diagonalization is equivalent to defining harmonic coefficients @xmath256 with respect to a complete set of spin two harmonics which are orthonormal over the patch of sky , in the same way as for the spin zero cut - sky temperature analysis @xcite .
as before , this construction ensures that for isotropic noise the noise correlation is diagonal @xmath257 the signal correlation is given by @xmath258 where the @xmath259 and @xmath260 are the diagonal electric and magnetic power spectrum matrices respectively , and we have assumed that @xmath261 . if the noise and signal are gaussian we can proceed to do a likelihood analysis for the power spectra using @xmath262}{|{\bm{n}}+{\bm{s}}|^{1/2}}.\ ] ] the coupling matrices @xmath132 can be computed quickly for an azimuthally symmetric patch of sky , as described in appendix [ app : ints ] , and modes with different @xmath50 decouple .
the problem is therefore tractable .
however it is not nearly so simple to find a maximum likelihood estimate of the magnetic amplitude , and in general there will be complicated correlations between the recovered power spectra . by using the lossy projection in the previous sections we have essentially shown that this likelihood function is ` nearly ' separable . making it separable costs something in terms of lost information , but
it significantly simplifies the problem . using the projected variables also reduces the size of the matrices , so performing the matrix inversions is significantly faster .
if the @xmath206 signal is determined to be negligible one would want to apply an efficient nearly - lossless method to estimate the electric power spectrum @xmath263 ( or to do parameter estimation ) , so we now consider the case when one polarization type is absent .
if the @xmath206 signal can be neglected we have @xmath264 where we have done a singular value decomposition as before so that we can find the well determined linear combinations of the @xmath84 : @xmath265 the matrices one has to invert to do a likelihood analysis are now one half the size of those in the optimal method when both polarization types are present , and so the problem is numerically much faster . however isotropic noise no longer gives the simple diagonal noise covariance , though this can always be rectified by using @xmath266 .
in practice a nearly optimal method would probably be more appropriate using only the unprojected variables @xmath267 , where @xmath136 and @xmath133 were defined in sec .
[ sec : sep ] .
these variables have diagonal noise properties like the @xmath155 for isotropic noise , and the computational saving may be significant when analysing high resolution polarization maps .
we have checked numerically that including @xmath131 in the analysis gains very little even for low @xmath44 and small patches . for large area surveys at high resolution
the information loss will probably be negligible .
there are exactly equivalent relations for the well determined magnetic variables in the case when @xmath215 vanishes .
this case is of little practical interest , since the @xmath215 signal would have to be removed to within the magnitude of the @xmath206 signal , and this is impossible on an incomplete sky since the two are not unambiguously distinguishable without accurate boundary data .
however , supposing that @xmath215 could be removed is useful theoretically as we can then compute the best obtainable magnetic signal to compare with what we obtain using our projected variables . the information lost due to the projection depends on the cosmology .
models with reionization have more power on large scales and a greater fraction of the power is lost due to removal of the boundary terms . for our toy model of the planck satellite
we find that the amplitude that could be detected at given significance and probability is reduced by about @xmath268 per cent by the projection for a cosmology with reionization at @xmath269 , but only by @xmath270 per cent for a zero reionization model . in the reionization model one
is loosing a lot of the additional information in the low multipoles that in the absence of the projection would have high signal to noise .
the net result is that the reionization model has an only slightly higher chance of giving a tensor detection despite having more large scale power .
by using the unprojected variables and incorporating the expected electric polarization contamination as an extra noise term one can approximately halve this loss .
the lossless result is compared to the realistic projected result for general circular sky patches in fig .
[ patchsize ] for an observation with much higher sensitivity .
the cost we incur by using the non - optimal method in terms of slightly larger error bars on the @xmath206 signal , or a less powerful test of detection at a given significance , is small for large survey areas though it does increase for small sky patches .
for these sensitive observations the electric signal is much larger than the magnetic signal and essentially nothing is lost by performing the projection rather than including electric contamination as a large extra noise . to make a detection of the magnetic signal on such a small sky patch with the planned long duration balloon observations the tensor / scalar ratio would need to be significantly larger than one , which is too large to be allowed by the current temperature anisotropy observations @xcite
of course , seeing if there is only a small magnetic signal is an important consistency check for current models with low tensor / scalar ratio to pass .
one simple way to reduce the information loss in our method would be to use data objects that include not only the surface integrals @xmath271 , but also those parts of the boundary terms in eqs .
( [ eq : i2wp ] ) and ( [ eq : im2wp ] ) that do not depend on @xmath272 on the boundary .
such objects would separate electric and magnetic polarization exactly if the scalar window functions were constructed to vanish on the boundary .
the problem of producing a non - redundant set of such windows could be tackled with a simple variant of the harmonic - based method presented in sec .
[ sec : sep ] . the additional boundary contribution would cancel that part of @xmath119 that couples to the normal derivative of the window function on the boundary , leaving a single non - zero singular value ( for @xmath273 ) to project out .
the net effect would be that for azimuthal patches we would gain one extra variable per @xmath50 for @xmath105 , though the noise properties of these variables would not be as simple as if the line integrals were not included , and the problem of performing line integrals with pixelized data is non - trivial .
for reionization models ( which have significant large scale power ) the reduction in information loss may be worth the effort required to overcome these obstacles , though a full analysis with the non - separated variables would probably work better .
we have considered the problem of producing statistically independent measures of the electric and magnetic polarization from observations covering only a portion of the sky .
although the separation of the polarization field into electric and magnetic modes is not unique in the presence of boundaries , we have shown how to construct window functions that are guaranteed to probe separately the electric and magnetic polarization exactly over azimuthally - symmetric patches of the sky .
we presented a harmonic - based method for efficient construction of the windows that automatically removes redundancy due to the finite sky coverage .
in addition , our window functions return separated electric and magnetic variables that have very simple diagonal noise correlations for idealized noise on the polarization map . for azimuthal patches separating the electric and magnetic polarization
comes at the cost of losing two pieces of information per @xmath50 mode , or roughly twice the number of pixels of area @xmath274 on the boundary of the patch . for large patches
this information loss is small unless there is large scale power due to reionization , but for smaller patches it can be more severe due to the limited support of the high @xmath50 spin - weight 2 harmonics in the patch .
although we have proved that our method gives exact separation for azimuthal patches , the harmonic - based construction should produce window functions that give approximate separation for arbitrarily shaped patches with similar information loss to the azimuthal case .
we showed how the variables constructed from our window functions could be used to constrain the amplitude of the magnetic signal without contamination from the much larger electric signal .
for the first time , we made predictions for the tensor amplitude that planck should be able to detect taking proper account of excluding the galactic region .
if other non - negligible foregrounds can be removed using the other frequency channels , planck should be able to detect the magnetic signal predicted by some simple inflationary models . for less sensitive observations
, our window functions should nevertheless be useful to set upper limits on the magnetic signal , and may also aid the identification of systematic effects in the instrument or analysis pipeline .
if the magnetic signal is shown to be consistent with noise then we showed how one can use all the well determined polarization pseudo - multipoles to analyse the electric polarization power spectrum without loss of information .
the analysis using these variables is no more complicated than the analysis of temperature anisotropies using cut - sky orthogonalized scalar harmonic functions @xcite .
we have only considered isotropic noise here , however , as long as the noise is azimuthally symmetric the separation of @xmath50 modes will still work , and the problem remains computationally tractable though rather less simple . in practice , there will be several other complications in real - life cmb polarimetry observations that will impact on the map - making and subsequent analysis stages .
further careful investigation of the propagation of instrument effects such as beam asymmetries , straylight , cross - polar contamination , and pointing instabilities through the map - making stage will be required before the programme for analysing magnetic polarization outlined in this paper will be realizable .
we acknowledge use of the lapack package for performing the matrix decompositions .
we acknowledge the support of pparc via a pptc special program grant for this work .
ac acknowledges a pparc postdoctoral fellowship .
we also thank pparc and hefce for support of the cosmos facility .
in general a spin - weight @xmath164 quantity @xmath275 is defined over a two - dimensional riemannian manifold with respect to an orthonormal diad field @xmath276 . the local freedom in the choice of diad amounts to the transformations @xmath277 of the ( complex ) null vectors @xmath278 .
a quantity is defined to be of spin - weight @xmath164 if under the transformation ( [ eq : appa0a ] ) @xmath279 . to every spin - weight @xmath164 object @xmath275 we can associate a ( complex ) symmetric trace - free , rank-@xmath280 tensor @xmath281 : for @xmath282 , @xmath283 where the irreducible tensor product @xmath284 .
the inverse relation is @xmath285 for @xmath286 we define @xmath287 .
the spin raising and lowering operators @xmath15 and @xmath15 are defined by the null diad components of the covariant derivatives of @xmath288 : @xmath289 ( the minus signs are conventional )
. in cmb polarimetry we are concerned with fields defined over the sphere , in which case the transformation in eq .
( [ eq : appa0a ] ) corresponds to a _ left_-handed rotation of the diad about the outward normal @xmath290 . choosing the orthonormal diad to be aligned with
the coordinate basis vectors @xmath291 and @xmath292 of a spherical polar coordinate system , we have @xmath293 and @xmath294 .
it follows that for this choice of diad the spin raising and lowering operators reduce to @xmath295 an elegant interpretation of the spin raising and lowering operators on the sphere can be obtained by considering spin - weight @xmath164 objects @xmath296 defined on a diad at ( position - dependent ) angle @xmath297 to the coordinate directions , so that @xmath298 where @xmath299 is defined on @xmath291 and @xmath292 . in this case , the spin raising and lowering operators can be related to the angular momentum operators for a rigid body @xcite .
working in a representation where the orientation of the body is specified in terms of euler angles @xmath300 follows ref .
@xcite , i.e. , successive right - handed rotations by @xmath301 , @xmath302 , and @xmath13 about the @xmath303 , @xmath304 , and @xmath303-axes respectively . the use of @xmath297 , which is minus the third euler angle , as a configuration variable for the rigid body is necessary to relate the angular momentum operators directly to the spin raising and lowering operators with the ( consistent ) conventions we have adopted here . ] , the angular momentum operators on the _ body - fixed axes _
take the form @xmath305 where @xmath306 .
these operators satisfy the commutation relations @xmath307 = \mp k_\pm , \qquad [ k_+,k_-]=-2 k_z , \label{eq : appa0}\ ] ] so that @xmath308 are lowering / raising operators with respect to the eigenvalues of @xmath309 .
note that the signs in these commutation relations are different from those for angular momentum operators on a fixed frame since on the body - fixed axes we have @xmath310 = - i k_z$ ] @xcite .
the action of the spin raising and lowering operators can then be formulated in terms of the angular momentum operators as @xmath311 several useful results for the spherical raising and lowering operators follow from the commutation relations ( [ eq : appa0 ] ) .
[ similar relations on a general manifold can be found from eqs .
( [ eq : appa1c ] ) and ( [ eq : appa1d ] ) . ] for a spin - weight @xmath164 quantity defined on the coordinate basis , @xmath312 { } _ s\eta(\theta,\phi ) & = & ( \nabla^2 - s^2 \csc^2\theta + 2i s \cot\theta\ , \csc\theta\ , \partial_\phi ) { } _ s\eta(\theta,\phi),\end{aligned}\ ] ] where we have used @xmath313 to derive the last identity .
applying these relations repeatedly to a spin - weight 0 quantity we find @xmath314 and the useful relation @xmath315 the integral of the spin - weight 0 quantity @xmath316 over some portion @xmath24 of the two - dimensional manifold is determined by the integral around the boundary @xmath33 : @xmath317 where @xmath318 is the spin - one element of length around the boundary : @xmath319 on the sphere in the spherical polar coordinate frame @xmath320 .
equation ( [ inttheorem ] ) is the complex representation of stokes theorem and the divergence theorem .
an equivalent result holds for spin - weight one quantities by forming the complex conjugate of the above , with @xmath321 .
the spin raising and lowering operators obey leibnitz rule , so there is a ` green s identity ' @xmath322 where @xmath323 and @xmath8 have definite spin weight . for @xmath324 with spin - weight zero integrating over a surface gives the integral theorem @xmath325 a similar result is obtained using @xmath326 for @xmath324 of spin weight @xmath31 which gives @xmath327 with a similar result for a spin - weight @xmath270 quantity .
the spin - weight @xmath164 spherical harmonics @xmath328 are defined in terms of the usual spherical harmonics @xmath329 by @xmath330 where @xmath331 .
they are non - zero for @xmath332 . by making use of eqs .
( [ eq : appa1 ] ) and ( [ eq : appa2 ] ) , and the properties of the @xmath333 operators when acting on wigner @xmath334-matrices - matrices follow refs .
we adopt the condon - shortley phase for the spherical harmonics , which differs from that used by goldberg et al .
@xcite by a factor of @xmath335 . ] @xcite , it is straightforward to show that @xcite @xmath336 with the conventions adopted here , @xmath337 .
orthonormality of the spin weight harmonics over the full sphere , @xmath338 follows from the orthogonality of the @xmath334-matrices over the @xmath339 group manifold . applying @xmath340 to eq .
( [ eq : appb1 ] ) , and using eqs .
( [ eq : appa1 ] ) and ( [ eq : appa2 ] ) , one can show that the spin weight harmonics satisfy the differential equation @xmath341\ , { } _ sy_{lm}. \label{eq : appb4}\ ] ] [ an alternative proof of this result follows from eq .
( [ commident ] ) . ]
the spin weighted harmonics are separable and can be written as @xmath342 the @xmath343 satisfy the self - adjoint equation @xmath344 ' - \frac{m^2+s^2 + 2 m s x}{1-x^2 } { } _ s\lambda_{lm } = -l(l+1){}_s\lambda_{lm } , \label{diffequation}\ ] ] where a prime denotes differentiation with respect to @xmath345 . the @xmath343 can be evaluated recursively for @xmath346 starting with @xmath347 and the recursion relation ( derived from standard results for the wigner @xmath334-matrices ; see e.g. ref .
@xcite ) @xmath348 where @xmath349 the harmonics for @xmath350 can be obtained from @xmath351 .
a straightforward way to evaluate @xmath343 for @xmath352 is to compute @xmath353 for @xmath354 and then use the symmetry @xmath355 .
further useful results for the spin weight harmonics can be found in refs .
@xcite , and expressions for the spin - weight @xmath356 harmonics in terms of the associated legendre functions are given in ref .
setting @xmath357 and @xmath358 in eq . , and using the differential equation ( [ eq : appb4 ] ) to simplify the integrand on the left - hand side , we have @xmath359\int_{s } \text{d}s\ , { } _ sy_{l'm'}^\ast\ , { } _ sy_{lm } = \oint_{\partial s } { { } _ 1\text{d}l\:}{}_sy_{l'm'}^\ast \,{\,\overline{\eth}\,}{}_sy_{lm } - \oint_{\partial s } { { } _ 1\overline{\text{d } l}\:}{}_sy_{lm}{\,\eth\,}{}_sy_{l'm'}^\ast .
\label{eq : appc1}\end{aligned}\ ] ] this expresses the overlap integral for @xmath360 in terms of a line integral around the boundary of @xmath24 . for an azimuthally symmetric surface the integrals are particularly straightforward as the spin - weight harmonics with different @xmath50 are orthogonal over the patch .
the overlap integral for the same @xmath50 but different @xmath46 can be determined from eq . to be @xmath361\int_a^b \text{d}x\,{}_s\lambda_{l'm}\ , { } _ s\lambda_{lm } = ( 1-x^2 ) [ { } _ s\lambda_{l'm } ' { } _ s\lambda_{lm}-{}_s\lambda_{lm } ' { } _ s\lambda_{l'm}]_a^b.\ ] ] [ this result also follows directly from eq . .
] the derivatives can be removed while maintaining homogeneity in the spin weight by using @xmath362 to write the @xmath363 integral as @xmath364_a^b.\end{aligned}\ ] ] note that @xmath365 .
the @xmath366 integrals can be evaluated recursively using @xmath367 which follows from eq . .
for @xmath368 , the starting values @xmath369 can be obtained from @xmath370_a^b}{2m+1 } + \frac{s}{\sqrt{(2m+1)(m^2-s^2 ) } } a^{s(m-1)}_{m(m-1 ) } , \label{eq : appd2 } \\
a^{s(-m)}_{mm } & = & a^{s(-m+1)}_{(m-1)(m-1 ) } + \frac{2\pi [ x { } _ s\lambda_{m(-m)}^2]_a^b}{2m+1 } - \frac{s}{\sqrt{(2m+1)(m^2-s^2 ) } } a^{s(-m+1)}_{m(m-1 ) } \qquad(m > |s| ) , \label{eq : appd3}\end{aligned}\ ] ] which follow from the explicit form of @xmath371 given in eq . .
for @xmath372 the integrals have the analytic solutions @xmath373_a^b , \qquad a^{(-m)m}_{mm } = \left[\left(\frac{x+1}{2}\right)^{2m+1}\right]_a^b.\ ] ] for @xmath374 , the starting values @xmath375 for the recursion in eq .
can be obtained from the symmetry @xmath376 and the recursion relations and .
if one is also generating the spin - weight zero overlap integrals for analysing the temperature field an alternative approach is to use eq . to relate the integrals of the spin - weight two and zero harmonics . in general integrals for different spin weights
can be related by @xmath377 where @xmath378 , and we have used the results @xmath379 which follow from eq . .
for @xmath380 and @xmath381 one can obtain the spin @xmath43 integrals in terms of the spin zero integrals .
the spin zero integrals are computed using the above relations with @xmath380 , in agreement with the relations given in ref .
@xcite . for a small patch of sky a large number of the overlap integrals are going to be very close to zero .
this makes sense intuitively , and is easy to see more quantitatively . from the differential equation the character of the harmonics changes from oscillatory to decaying at the point where
@xmath382 for @xmath383 nearer the poles than the critical value the harmonics become very small . for a sky patch extending from the north pole to @xmath383 , for large @xmath46
we see that for @xmath384 larger than @xmath385 the harmonics will be small within the patch .
hence the overlap integrals with @xmath386 will be small , corresponding to one ( or both ) of the harmonics being localized out of the sky patch .
after obtaining a vector of observed data @xmath111 two questions one might ask are : ( 1 ) what is the probability that the signal is just noise ? ; ( 2 ) given the signal is not noise , what can we say about the amplitude @xmath210 ? the likelihood function encapsulates all the information in the data concerning the amplitude @xmath210 , and gives the posterior probability distribution of @xmath210 after multiplying by the prior .
question ( 1 ) is really asking for a comparison of two models , one in which @xmath214 ( or @xmath387 where @xmath388 is small ) , and one in which @xmath389 .
given there is a prior probability @xmath390 that the signal is pure noise ( @xmath214 ) and probability @xmath391 that @xmath392 , distributed with the normalized prior probability distribution @xmath211 , the posterior probability that @xmath214 is given by @xmath393 the posterior probability tells us the probability that the signal is pure noise once we have a particular set of data .
however , it does nt immediately tell us that we might expect to obtain from a given observation , just what we know once the observation has been performed .
the posterior distribution can depend strongly on the prior .
the posterior probability is a monotonic function of the bayes factor @xmath394 in classical hypothesis testing the likelihood ratio @xmath395 is the most powerful test statistic for distinguishing a model with @xmath396 from one with @xmath214 ( for details see ref .
the largest fraction @xmath397 of the values of the likelihood ratio under the null - hypothesis ( @xmath214 ) determine a region , which , if the observed ratio falls in it , rejects the null hypothesis at significance @xmath13 .
the distribution of the likelihood ratio is straightforward to compute using monte - carlo techniques .
in general we do not have some fixed alternative hypothesis @xmath396 ; one possibility is to use the the statistic @xmath398 , formed by marginalizing over some prior , in place of the likelihood ratio .
often the prior is fixed to @xmath399 , where @xmath213 is the maximum likelihood estimate of @xmath210 , so the test reduces to a likelihood ratio test with the alternative hypothesis @xmath400 . by the time magnetic polarization
comes to be observed we should have some prior information about the tensor amplitude from accurate measurement of the temperature power spectrum .
however at this point we do not have very useful information about the prior distribution , and to consider the possible results of future experiments it is useful to assume no information . the maximum entropy prior ( the uniform prior ) is probably not appropriate in this case do we really think that all values of @xmath210 ( below a certain bound ) are equally likely rather than , say , all values of @xmath401 being equally likely ?
the uniform prior gives radically different answers depending on the choice of variable . the jeffreys prior @xmath402^{1/2},\ ] ] is reparameterization invariant @xcite , and therefore does not suffer from this problem .
the jeffreys prior goes like @xmath403 if @xmath404 and in general is improper ( does not have a finite integral ) , though this is not a problem for evaluating test statistics since the integral of the product with the likelihood function will be finite .
it is also not a true prior in the sense that it depends on what data is going to be collected , but this is really a good thing as it concentrates the prior probability in the region where we need it in order to obtain a detection . since the prior is not localized the posterior probability of the null - hypothesis will depend on where one cuts off the prior .
however if one only ever compares the bayes factors @xmath398 the cut - off is not very important since the likelihood function will be localized . for this reason performing classical hypothesis tests using @xmath398
is rather more independent of the prior information than considering the values of the posterior probability , though clearly it is the latter which is rigorous and contains all the available information .
classical hypothesis tests are useful for assessing probability of getting a detection at a given significance , though it should be remembered that getting a detection at @xmath405 per cent confidence does _ not _ mean a the probability of @xmath406 per cent that the signal is pure noise .
however for sensible priors there will be a close correspondence ; classical and bayesian techniques agree that high values of @xmath398 correspond to the null - hypothesis being less likely .
using classical hypothesis tests to compute the probability of getting a detection at a given significance for a given true @xmath210 is significantly simpler than computing and interpreting the corresponding distribution of the posterior probability distributions . for full sky surveys
there are a large number of statistically - independent magnetic variables , which can be thought of as the eigenvectors of @xmath407 .
those variables corresponding to large eigenvalues have high expected signal to noise and are most useful for obtaining detections .
if @xmath111 has dimension @xmath408 there will be @xmath408 independent variables .
if the eigenvalues of @xmath407 are distributed fairly uniformly , and we consider expected chi - squared detections of order one sigma , all the eigenvalues will be @xmath409 and we have @xmath410 . using a second order approximation we then have @xmath411 and @xmath412 in this approximation the entire likelihood distribution is simple to compute since the matrix manipulations only need be performed once , and is of the gaussian form @xmath413 where @xmath414 and the maximum likelihood estimate for @xmath210 is @xmath415.\ ] ] the maximum likelihood estimate for @xmath210 is only weakly biased for small signal to noise .
consider fixing the signal hypothesis to @xmath416 , which gives the largest likelihood ratio possible for any prior .
the likelihood ratio is then a monotonic function of the test statistic @xmath417 the quantity @xmath418 gives the number of standard deviations the maximum likelihood is from pure noise , which is an good intuitive measure of the number of ` sigmas ' at which the magnetic signal has been detected . if any eigenvalues become large enough then the small @xmath210 approximation will fail and one needs to compute the full likelihood ratio to obtain optimal results .
the @xmath419 statistic is a function only of the vector @xmath420 and the matrix @xmath421 , so working in the frame in which the matrix is diagonal the statistic is simple to compute .
note that @xmath422 and @xmath213 can be negative and correspond to a non - detection .
the second order approximation can also be used to speed up computing the full likelihood function for integration against a prior . in the diagonal frame
it is straightforward to identify any modes with high signal to noise if there are some then the likelihood for these modes can be computed exactly .
the likelihood for the remaining low signal to noise modes can be computed quickly using the approximation in which the likelihood is gaussian . multiplying these together
allows the full likelihood function to be computed , and hence the the probability distribution of @xmath398 . whilst this is slower than using the @xmath419 statistic it is a useful check , and
may be essential when there are very high signal to noise modes . for good detections
the prior is not very important as long as it is not small over the bulk of the likelihood .
if the prior is approximately constant over most of the likelihood integral , the second order approximation is valid , and @xmath423 , then @xmath424 where @xmath425 depends weakly on @xmath213 and @xmath426 compared to the exponential . in this case @xmath398 depends almost entirely ( and monotonically ) on the value of @xmath427 , which is why @xmath419 is a good statistic to use .
we found that detection probabilities computed using full likelihood results with the jeffreys prior tend to agree very closely with those computed using @xmath419 .
the advantage of using @xmath419 rather than computing the full likelihood distribution is that it is computationally significantly simpler and faster , which is useful , though by no means essential , for performing accurate monte - carlo computations .
there is no problem computing using the full likelihood distribution exactly from single samples of actual observed data . in the limit that @xmath428
the likelihood function can be approximated by the first two terms in its taylor series in @xmath210 .
the likelihood ratio for any given ( small ) @xmath210 is then a monotonic function of @xmath429 , which is independent of @xmath210 and therefore provides the uniformly most powerful test . for gaussian signal and noise
it is proportional to @xmath430 . after dividing by the root of the variance in the null hypothesis @xmath431
we obtain the quantity @xmath432 which is the optimal quadratic ` null - buster ' statistic introduced to the cmb literature in ref . @xcite . in the limit in which all the eigenvalues of @xmath407 tend to zero
this is the optimal test statistic . in general
it is not a signal that can be detected will violate this assumption .
for a very large number of approximately equal eigenvalues the value of @xmath426 will approximate @xmath433^{-1/2}$ ] , the fisher curvature for small @xmath210 . in this limit
the null - buster remains a good statistic and @xmath434 .
however in general @xmath419 performs significantly better when the eigenvalues are distributed less evenly , or when there are not that many eigenvalues . as the signal to noise increases @xmath419 also performs much less sub - optimally than the null - buster
. for the case of magnetic polarization observations
the @xmath419 statistic outperforms the null - buster for a wide range of patch sizes in realistic reionization models .
the large scale magnetic signal coming from low redshift ( @xmath435 ) reionization gives a small number of modes with relatively high signal to noise ( see fig .
[ contribs ] ) , and the conditions under which the null - buster is a good statistic are therefore not satisfied .
the qualitative reason that the null - buster performs poorly is that the position of the maximum and the curvature of the likelihood function are correlated , so dividing the actual maximum by the expected curvature does not give you an accurate measure of the number of ` sigmas ' from zero for a particular observation .
this makes the null hypothesis distribution unnecessarily broad at large values , and therefore makes it harder to rule out the null hypothesis with good significance .
the @xmath419 statistic has a much sharper distribution than the null - buster ( which has a distribution similar to chi - squared ) in the alternative hypothesis , and the value of @xmath419 corresponds much more closely to the significance ( measured in gaussian - like ` sigmas ' ) of the result .
a few points can be made in the null - buster s defence .
firstly it is slightly easier to compute than @xmath419 .
secondly , since we motivated the @xmath419 by assuming a gaussian signal it is conceivable that the null - buster could perform better with certain non - gaussian signals .
lastly , the null - buster is quadratic which makes it easy to calculate the mean and variance analytically .
however it is clear that with gaussian signals using the null - buster is in general significantly sub - optimal . | the full sky cosmic microwave background polarization field can be decomposed into ` electric ' and ` magnetic ' components .
working in harmonic space we construct window functions that allow clean separation of the electric and magnetic modes from observations over only a portion of the sky .
we explicitly demonstrate the method for azimuthally symmetric patches , but also present it in a form in principle applicable to arbitrarily - shaped patches . from the window functions we obtain variables that allow for robust estimation of the magnetic component without risk of contamination from the probably much larger electric signal .
the variables have a very simple noise properties , and further analysis using them should be no harder than analysing the temperature field . for an azimuthally - symmetric patch , such as that obtained from survey missions when the galactic region is removed , the exactly - separated variables are fast to compute .
we estimate the magnetic signal that could be detected by the planck satellite in the absence of extra - galactic foregrounds .
we also discuss the sensitivity of future experiments to tensor modes in the presence of a magnetic signal generated by weak lensing , and give lossless methods for analysing the electric polarization field in the case that the magnetic component is negligible .
a series of appendices review the spin weight formalism and give recursion relations for fast computation of the spin - weighted spherical harmonics and their inner products over azimuthally - symmetric patches of the sphere .
a further appendix discusses the statistics of weak signal detection . |
the hubble - sandage diagram for type ia supernovae ( hereafter sneia ) , as measured by the supernova cosmology project@xcite and the high - z supernova search team@xcite , provided the first evidence that the present universe is undergoing a phase of accelerating expansion driven by an exotic component with negative pressure ( in addition to the cold dark matter ) , usually called dark energy .
the idea of a dark energy - dominated universe is a direct consequence of a convergence of independent observational results , and constitutes one of the greatest challenges for our current understanding of fundamental physics@xcite . among a number of possibilities to describe this dark energy component , the simplest and most theoretically appealing way is by means of a cosmological constant @xmath2 , which acts on the einstein field equations as an isotropic and homogeneous source with a constant equation of state , @xmath13 .
although cosmological scenarios with a @xmath2 term might explain most of the current astronomical observations , from the theoretical viewpoint they are plagued with at least a fundamental problem , namely , it is really difficult to reconcile the small value of the vacuum energy density required by observations ( @xmath14 ) with estimates from quantum field theories ranging from 50 - 120 orders of magnitude larger@xcite .
this problem sometimes called the cosmological constant problem ( pcc ) has inspired many authors to propose decaying @xmath2 models@xcite and other alternative approaches for describing dark energy@xcite .
nevertheless , the present cosmic concordance model ( ccm ) which is supported by all the existing observations is a flat @xmath2cdm cosmology with a matter fraction of @xmath15 and a vacuum energy contribution of @xmath16@xcite . on the other hand ,
the real universe is not perfectly homogeneous , with light beams experiencing mass inhomogeneities along their way thereby producing many observable phenomena .
for instance , light lines traversing in the universe are attracted and refracted by the gravitational force of the galaxies on their path , which bring us the signal of lensing , one of which is the multiple images of a single far galaxy@xcite .
nowadays , gravitationally lensed quasars and radio sources offer important probes of cosmology and the structure of galaxies . the optical depth for lensing depends on the cosmological volume element out to moderately high redshift . in this way
, lens statistics can in principle provide valuable constraints on the cosmological constant or , more generally , on the dark energy density and its equation of state @xcite . in this context , one of the most important issues in the modern cosmology is to quantify from the present observations the influence of such inhomogeneities on the evolution of the universe .
an interesting possibility to account for such effects is to introduce the smoothness parameter @xmath0 which represents the magnification effects experienced by the light beam .
when @xmath17 ( filled beam ) , the frw case is fully recovered ; @xmath18 stands for a defocusing effect ; @xmath19 represents a totally clumped universe ( empty beam ) .
the distance relation that takes the mass inhomogeneities into account is usually named dyer - roeder distance@xcite , although its theoretical necessity had been previously studied by zeldovich@xcite and kantowski@xcite . in this way
, we label it here as zeldovich - kantowski - dyer - roeder ( zkdr ) distance formula ( for an overview on cosmic distances taking into account the presence of inhomogeneities see the paper by kantowski@xcite ) .
several studies involving the zkdr distances in dark energy models have been published in the last few years .
useful analytical expressions for @xmath2cdm models have been derived by kantowski _
_ @xcite and demianski _
et al._@xcite .
working in the empty beam approximation ( @xmath20 ) , sereno _
et al._@xcite investigated some effects of the zkdr distance for a general background . by assuming that both dominant components may be clustered they also discussed the critical redhift , i.e.
, the value of @xmath21 for which @xmath22 is a maximum ( or @xmath23 minimum ) , and compared to the homogeneous background results as given by lima and alcaniz@xcite , and further discussed by lewis and ibata@xcite .
demianski and coworkers derived an approximate solution for a clumped concordance model valid on the interval @xmath24 .
additional studies on this subject is related to time delays@xcite , gravitational lensing@xcite , and even accelerated models driven by particle creation have been investigated@xcite . in a previous paper@xcite ,
we have applied the zkdr equation in the framework of phantom cosmology in order to determine cosmological constraints from a sample of milliarcsecond compact radio sources . by assuming a gaussian prior on the matter density parameter , i.e. , @xmath25 , the best fit model for a phantom cosmology with @xmath26 occurs at @xmath27 and @xmath28 when we marginalize over the characteristic size of the compact radio sources .
such results suggested that the zkdr distance can give important corrections to the so - called background tests of dark energy . 0.1 in in this article ,
the pair of cosmic parameters , @xmath29 and @xmath0 , are constrained from supernovae observations by applying the zkdr distance - redshift relation for a flat ( @xmath2cdm ) model .
as we shall see , the @xmath0 parameter is not well constrained by the 115 sne observed by astier _
however , the 182 sne type ia sample of riess _
( 2007 ) , constrains the pair of parameters to be @xmath8 and @xmath9 ( @xmath5 ) . as a general result , even considering the existence of inhomogeneities described by the @xmath0 parameter
, the einstein - de sitter model is ruled out by the two samples with a high degree of statistical confidence ( @xmath10 and @xmath11 , respectively ) .
the paper is organized as follows . in section
ii , we present the basic equations and the distance description taking into account the inhomogeneities as described by the zkdr equation . in section iii , we determine the constraints on the cosmic parameters from the two supernovae samples . finally , we summarize the main conclusions in section iv .
in a clumpy universe model , the local geometry is inhomogeneous , but its global aspect can be described by the frw type geometry ( @xmath30 ) @xmath31 where @xmath32 is the scale factor and @xmath33 denotes the metric in the 2-sphere .
as it is widely known , the idea of clumpy universe is still a ill - defined notion since we do not have a clear mathematical recipe to separate the global properties from the local inhomogeneous aspects of the universe .
after dyer & roeder@xcite , it is usual to introduce a phenomenological parameter , @xmath34 , called the smoothness " parameter .
such a parameter quantifies the portion of matter in clumps ( @xmath35 ) relative to the amount of background matter which is uniformly distributed ( @xmath36 ) . in general , due to the structure formation process , it should be dependent of the redshift , as well as , on the direction along the line of sight ( see , for instance , @xcite and refs . there in )
. however , in the majority of the works @xmath0 is assumed to be a constant parameter . from a mathematical viewpoint
the treatment is based on the optical - scalar equation for light propagation in the so - called geometric optics approximation@xcite @xmath37 where a prime denotes differentiation with respect to the affine parameter @xmath38 , @xmath39 is the cross - sectional area of the light beam , @xmath40 the ricci tensor , and @xmath41 the photon four - momentum . in this form
, it is implicit that the influence of the weyl tensor ( shear ) can be neglected .
this means that the light rays are propagating far from the mass inhomogeneities so that the large - scale homogeneity implies that their shear contribution are canceled .
the proportionality factor between the cross - sectional length @xmath42 and the angular distance @xmath43 can be defined to be constant . actually , the above optical scalar equation
is usually written in terms of the dimensionless angular diameter distance @xmath44 .
further , by recalling the existence of a simple relation between the luminosity distance , and the angular - diameter distance ( from etherington principle@xcite , @xmath45 ) , it is easy to show that the zkdr ( dimensionless ) luminosity distance for @xmath2cdm cosmology satisfies the following differential equation@xcite @xmath46 which satisfies the boundary conditions : @xmath47 where @xmath48 , @xmath49 end @xmath50 are functions of the cosmological parameters , expressed in terms of the redshift by : @xmath51 where as remarked before , the @xmath0 parameter appearing in the @xmath50 expression ( here assumed to be a constant ) quantifies the clustered fraction of the pressureless matter .
the standard frw models contain only homogeneously and isotropically distributed perfect fluid gravity sources , and the present ccm is assumed to represent both the large scale " geometry of the universe and the matter content .
however , the universe appears homogeneous only in a statistical sense , when one is describing the largest scales .
therefore , although making very useful predictions , our cosmological models are somewhat inadequate at small and moderate scales .
this means that relations like @xmath52 , the distance modulus for a standard candle , commonly assumed to be valid on average can be incorrect even for observations including sne ia . in particular if the underlying mass density approximately follows luminous matter ( i.e. , associated with bounded galaxies ) , the effects of inhomogeneities on relations like @xmath53 must be taken into account . 0.1 in in fig . 1
, we display the effects of the inhomogeneities in the reduced hubble - sandage diagram for the astier _
( 2006 ) and riees _ et al . _
( 2007 ) samples for some selected values of the smoothness parameter .
the plots correspond to several values of @xmath1 and @xmath0 as indicated in the panels .
the difference between the data and models from an empty universe case ( ocdm ) prediction is also displayed there .
for the sake of comparison , we also show the einstein - de sitter ( e - ds ) model , i.e. @xmath54 and @xmath17 , as well as the present cosmic concordance ( @xmath55 , @xmath56 , @xmath57 ) .
note that the @xmath0 parameter contributes in the right direction i.e. , the sne type ia become dimmer when it increases on the allowed range . in
what follows , a @xmath3 minimization will be applied for the two sets of sne data with the parameters @xmath58 and @xmath0 spanning the interval [ 0,1 ] in steps of 0.01 , for all numerical computations .
let us now discuss the bounds arising from sne ia observations on the pair of parameters ( @xmath59 ) defining the zkdr luminosity distance .
the current data from supernova legacy survey ( snls ) collaboration correspond to the first year results of its planned five year survey .
the total sample includes 71 high-@xmath21 sne ia in the redshift range @xmath60 plus 44 low-@xmath21 sne ia as published by astier _
et al._@xcite .
although in a better agreement with wmap 3-years results @xcite than the _ gold _ sample @xcite ( for a more detailed discussion see e.g. , jassal _
et al._@xcite ) , the most distant sne ia of these 115 events has redshift smaller than unity .
following standard lines , the maximum likelihood estimator , @xmath61 $ ] , is determined by a @xmath62 statistics @xmath63 where @xmath64 is the complete set of parameters that we want to fit , @xmath65 , @xmath66 are , respectively , the errors associated with the observational techniques in determining the distance moduli ( includes a peculiar contribution ) and the intrinsic dispersion of sne ia .
the corresponding errors are reported in astier _
_ paper @xcite .
marginalizing our likelihood function over the nuisance parameter , @xmath67 , we obtain the likelihood function for the @xmath68 plane . in order to determine the cosmological parameters ( @xmath1,@xmath0 ) ,
a @xmath3 minimization for the range of [ 0,1 ] in steps of @xmath69 has been applied .
the @xmath70 , @xmath71 and @xmath72 confidence levels are defined by the conventional two - parameters @xmath3 levels @xmath73 , @xmath74 and @xmath75 , respectively .
it is very important to note that we do not consider any prior in @xmath1 , as usually required by the sne ia test .
0.1 in the basic results are shown in figs .
2a , 2b and 2c . from fig .
2a we see that all range for @xmath0 is accepted , while a @xmath76 is obtained . in fig .
3b we see the likelihood for the smoothness parameter .
the best fit adjustment occurs for values of @xmath77 and @xmath55 with @xmath78 and @xmath79 degrees of freedom ( @xmath80 ) , thereby showing that the model provides a good fit to these data .
it is also interesting that for any @xmath0 value , we also find no evidence for a high @xmath1 parameter as required by a flat einstein - de sitter universe ( @xmath81 ) .
actually , the einstein - de sitter scenario has a very small statistical significance @xmath82 ( @xmath10 outside ) .
however , since the astier _
_ data are not restrictive for the @xmath0 parameter , let us now consider the enlarged sneia sample observed by the high - z supernovae search team @xcite . the so - called _ gold _ sample from the hzs team @xcite is a selection of 182 sne ia events distributed over the redshift interval @xmath83 , and constitutes the compilation of the best observations made so far by them and by the supernova cosmology project events observed by hubble space telescope ( hst ) . as before , constrains on the cosmological parameters ( @xmath1,@xmath0 ) , are determined from a @xmath3 minimization within the range of [ 0,1 ] spanned by such parameters . in fig .
3a one can see that @xmath84 and @xmath85 with @xmath86 of statistical confidence .
the best fit adjustment occurs for values of @xmath87 and @xmath88 with @xmath89 and @xmath90 degrees of freedom the reduced ( @xmath91 ) .
therefore , the model provides a very good fit to the riess _
et al . _ sample . in fig .
3b we see the likelihood for the smoothness parameter . as previously remarked , the riess _
data set is much more restrictive for the smoothness parameter than astier _
et al . _ sample . within @xmath5 , the allowed range for
the @xmath0 falls on the interval @xmath84 ( cf fig 2a ) . in fig 3c
we show the probability for the density matter parameter . in this case a small region is permitted @xmath85 with of the confidence level ( @xmath5 ) . in the analysis for the einstein - de sitter universe ( @xmath92 , @xmath93 ) , the @xmath94 is too bad ( @xmath11 c.l .
outside for 1 degree of freedom ) , and guarantee us to exclude this model with high confidence .
cosmology is in an exciting period . a considerable set of rather sophisticated experiments , until a few years ago regarded as futuristic ,
have now been completed with spectacular success .
the results of the first observations almost one decade ago have been confirmed what was long surmised , namely , that most of the matter is nonbaryonic and that we live in an accelerating expanding universe dominated by dark energy . in this article , we have discussed the influence of inhomogeneities on the expansion rate of the universe . in particular , if the smoothness @xmath0 parameter could be constrained through a statistical analysis involving two large sets of sne ia data . as we have seen , in the case of the astier _
_ sample , the entire interval of @xmath0 is allowed while a @xmath76 is obtained . within the existing uncertainties ,
these results are consistent with the constraints obtained from angular diameter of compact radio sources with basis on the gurvits _ et al .
_ data @xcite ) .
therefore , although in close agreement with rather different analysis , this sne data set is uncapable to constrain the smoothness parameter . actually , at this moment , the sample of riess _
provides a more stringent constrain with the allowed range for @xmath0 falling on the interval @xmath84 ( @xmath5 ) .
basically , this occurs because the riess _
_ sample extends to appreciably higher redshifts . in general ,
both analysis suggest that a large range for @xmath0 is permitted , and that the einstein - de sitter model is strongly excluded ( @xmath10 and @xmath11 , respectively ) . as we have seen , the necessity of the dark energy and a transition from deceleration to an accelerating phase is maintained even when one takes into account the clustering phenomenon .
however , at the level of the sne ia observations discussed here , these results suggest that the clumpiness of matter distribution can mimic at least a small fraction of the dark energy component .
finally , we would like to stress that measurements from sne ia combined with the zkdr inhomogeneous approach adopted here may provide an independent and more rigorous cosmological test for the cosmic concordance model in the near future . in this concern
, it should be very important to investigate whether the @xmath0 parameter can be constrained using independent observations , among them : the cosmic microwave background anisotropies , the physics of galaxies clusters , sunyaev - zeldovich effect , time delay and statistical of gravitational lensing .
some studies along these lines will be presented in a forthcoming communication .
the present results based only on the hubble - sandage diagram show that the riess _
sample more restrictive than the astier _
_ sample , thereby reinforcing the interest to observe more supernova events at higher redshifts .
we are very grateful to the organizers of the iwara ( international workshop on relativistic astrophysics , natal , brazil ) where the preliminary results of this work were presented .
the authors would also like to thank l. marassi , f. j. jesus , r. f. l. holanda , and v. busti for helpful discussions .
rcs is supported by cnpq no .
15.0293/2007 - 0 and jvc is supported by fapesp no .
2005/02809 - 5 .
jasl is partially supported by cnpq and fapesp no .
04/13668 - 0 ( brazilian research agencies ) .
99 s. perlmutter _
et al . _ ,
nature * 391 * , 51 ( 1998 ) ; s. perlmutter _
_ , astrophys
. j. * 517 * , 565 ( 1999 ) .
a. g. riess _
et al . _ ,
j. * 116 * , 1009 ( 1998 ) .
t. padmanabhan , phys . rept . * 380 * , 235 ( 2003 ) ; p. j. e. peebles and b. ratra , rev . mod . phys .
* 75 * , 559 ( 2003 ) ; j. a. s. lima , braz .
phys . * 34 * , 194 ( 2004 ) , [ astro - ph/0402109 ] ; v. sahni and a. starobinsky , ijmp * d15 * , 2105 ( 2006 ) .
s. weinberg , rev .
phys . * 61 * , 1 ( 1989 ) .
w. chen and y - s .
wu , phys .
d * 41 * , 695 ( 1990 ) ; d. pavn , phys . rev .
d * 43 * , 375 ( 1991 ) ; j. c. carvalho , j. a. s. lima and i. waga , phys . rev .
d * 46 * , 2404 ( 1992 ) ; i. waga , astrophys . j. * 414 * , 436 ( 1993 ) ; j. a. s. lima and j. m. f. maia , phys . rev d * 49 * , 5597 ( 1994 ) ; j. a. s. lima and m. trodden , phys .
d * 53 * , 4280 ( 1996),[astro - ph/9508049 ] ; j. a. s. lima , phys .
d * 54 * , 2571 ( 1996 ) , [ gr - qc/9605055 ] ; a. i. arbab and a. m. m. abdel - rahman , phys .
d * 50 * , 7725 ( 1994 ) ; l. f. bloomfield torres and i. waga , mon . not .
279 * , 712 ( 1996 ) ; j. m. overduin and f. i. cooperstock , phys . rev .
d * 58 * , 043506 ( 1998 ) ; j. m. overduin , astrophys .
j. * 517 * , l1 ( 1999 ) ; r. g. vishwakarma , class .
* 17 * , 3833 ( 2000 ) ; m. v. john and k. b. joseph , phys .
d * 61 * , 087304 ( 2000 ) ; o. bertolami and p. j. martins , phys . rev .
d * 61 * , 064007 ( 2000 ) ; j. a. s. lima and j. s. alcaniz mnras * 317 * , 893 ( 2000 ) , [ astro - ph/0005441 ] ; r. g. vishwakarma , gen .
* 33 * , 1973 ( 2001 ) ; a. s. al - rawaf , mod .
a * 14 * , 633 ( 2001 ) ; r. g. vishwakarma , class .
18 * , 1159 ( 2001 ) ; j. a. s. lima and j. m. f. maia , phys .
d * 65 * , 083513 ( 2002 ) , [ astro - ph/0112091 ] ; m. k. mak , j.a .
belinchon , and t. harko , ijmpd * 11 * , 1265 ( 2002 ) ; m. r. mbonye , ijmpa * 18 * , 811 ( 2003 ) ; j. v. cunha and r. c. santos , ijmp d * 13 * , 1321 ( 2004),[astro - ph/0402169 ] ; s. carneiro and j. a. s. lima , ijmpa * 20 * 2465 ( 2005),[gr - qc/0405141 ] ; i. l. shapiro , j. sola and h. stefancic , jcap * 0501 * , 012 ( 2005 ) ; e. elizalde , s. nojiri , s.d .
odintsov and p. wang , phys .
d * 71 * , 103504 ( 2005 ) ; j. s. alcaniz and j. a. s. lima , phys .
d * 72 * , 063516 ( 2005 ) , [ astro - ph/0507372 ] ; a. friaca , j. s. alcaniz and j. a. s. lima , mon . not .
. soc . * 362 * , 1295 ( 2005 ) , [ astro - ph/0504031 ] ; i. l. shapiro , j. sola and h. stefancic , jcap * 0501 * , 012 ( 2005 ) .
t. padmanabhan and t. r. choudhury , mon . not .
* 344 * , 823 ( 2003 ) ; p. t. silva and o. bertolami , astrophys . j. * 599 * , 829 ( 2003 ) ; j. a. s. lima , j. v. cunha and j. s. alcaniz , phys .
d * 68 * , 023510 ( 2003 ) , [ astro - ph/0303388 ] ; z .- h .
zhu and m .- k .
fujimoto , , 52 ( 2003 ) ; s. nesseris and l. perivolaropoulos , phys .
d * 70 * , 043531 ( 2004 ) ; y. wang and m. tegmark , phys .
* 92 * , 241302 ( 2004 ) ; j. a. s. lima and j. s. alcaniz , phys .
b * 600 * , 191 ( 2004 ) , [ astro - ph/0402265 ] ; t. r. choudhury and t. padmanabhan , astron .
astrophys . * 429 * 807 ( 2005 ) ; d. rapetti , s. w. allen and j. weller , mon . not .
. soc . * 360 * , 546 ( 2005 ) ; r. colistete jr and j. c. fabris , class .
* 22 * , 2813 ( 2005 ) ; b. feng , x. wang and x. zhang , phys .
b * 607 * , 35 ( 2005 ) ; r. lazkoz , s. nesseris and l. perivolaropoulos , [ astro - ph/0503230 [ ; r. lazkoz , s. nesseris and l. perivolaropoulos , [ astro - ph/0503230 ] ; a. shafieloo , u. alam , v. sahni , a. a. starobinsky , mon . not .
. soc . * 366 * , 1081 ( 2006 ) ; j. v. cunha , j. s. alcaniz and j. a. s. lima , phys .
d * 69 * , 083501 ( 2004 ) , [ astro - ph/0306319 ] ; f. c. carvalho _
et al . _ ,
* 97 * , 081301 ( 2006 ) , [ astro - ph/0608439 ] ; [ arxiv:0704.3043 ] ; l. r. abramo and n. pinto - neto , phys .
d * 73 * , 063522 ( 2006 ) ; j. v. cunha , l. marassi and r. c. santos , ijmp d * 16 * , 403 ( 2007 ) , [ astro - ph/0306319 ] ; a. shafieloo , astro - ph/0703034 ( 2007 ) .
p. astier et al . , astron . astrophys . * 447 * , 31 ( 2006 ) .
d. n. spergel et al .
. ser . * 170 * , 377 ( 2007 ) ; m. s. turner and d. huterer , arxiv:0706.2186 ( 2007 ) .
a. g. riess et al . ,
j. * 659 * , 98 ( 2007 ) .
s. w. allen et .
al . , arxiv:0706.0033v1 ( 2007 ) . p. schneider , j. ehlers and e. e. falco , _ gravitational lenses _ , springer - verlag , berlin ( 1992 ) . c. s. kochanek , b. mochejska , n. d. morgan and k. z. stanek , astrophys . j. * 637 * , l73 ( 2006 ) ; m. oguri et al
arxiv:0708.0825 ( 2007 ) . m. fukugita , t. futamase and m. kasai , mon . not
246 * , 24 ( 1990 ) ; m. fukugita and e. l. turner , mon . not .
. soc . * 253 * , 99 ( 1991 ) ; e. l. turner astrophys .
j. * 365 * , l43 ( 1990 ) ; l. m. krauss and m. white , astrophys .
j. * 394 * , 385 ( 1992 ) ; d. maoz and h. w. rix , astrophys .
j. * 416 * , 425 ( 1993 ) ; c. s. kochanek , astrophys
. j. * 466 * , 638 ( 1996 ) ; e. e. falco , c. s. kochanek and j. a. muoz , astrophys .
j. * 494 * , 47 ( 1998 ) ; d. rusin and c. s. kochanek , astrophys .
j. * 623 * , 666 ( 2005 ) ;
a. r. cooray and d. huterer , astrophys . j. * 513 * , l95 ( 1999 ) ; i. waga and a. p. m. r. miceli , phys .
d * 59 * , 103507 ( 1999 ) ; i. waga and j. frieman , phys .
d * 62 * , 043521 ( 2000 ) ; q .-
j . zhang , l .- m .
cheng , y .- l .
wu , arxiv:0708.2164 ( 2007 ) .
m. kasai , t. futamase and f. takahara , phys .
a * 147 * , 97 ( 1990 ) ; e. v. linder , astron . astrophys . * 206 * , 190 ( 1988 ) ; * ibdem * astrophys . j. * 497 * , 28 ( 1998 ) .
p. schneider and a. weiss , astrophys .
j. * 327 * , 526 ( 1988a ) ; p. schneider and a. weiss , astrophys .
j. * 330 * , 1 ( 1988b ) ; m. bartelmann and p. schneider , astron .
astrop . * 248 * , 349 ( 1991 ) ; k. watanabe , m. sasaki and k. tomita , astrophys . j. * 394 * , 38 ( 1992 ) .
c. c. dyer and r. c. roeder , astrophys .
j. * 174 * , l115 ( 1972 ) ; astrophys .
j. 180 , l31 ( 1972 ) . ya .
b. zeldovich , sov . astron .
* 8 * , 13 ( 1964 ) .
r. kantowski , astrophys .
j. * 155 * , 89 ( 1969 ) .
r. kantowski , phys .
d * 68 * , 123516 , ( 2003 ) .
r. kantowski , astrophys .
j. 507 , 483 ( 1998 ) .
r. kantowski , j. k. kao , r. c. thomas , astrophys . j. , 545 549 ( 2000 ) .
m. demianski , r. de ritis , a. a. marino , e. piedipalumbo , astron .
a * 411 * , 33 ( 2003 ) .
m. sereno , g. covone , e. piedipalumbo and r. de ritis , mon . not .
. soc . * 327 * , 517 ( 2001 ) .
j. a. s. lima and j. s. alcaniz , astron
. astrophys . * 357 * , 393 ( 2000 ) , [ astro - ph/0003189 ] ; * ibdem * gen
. relativ .
gravit . * 32 * , 1851 ( 2000 ) , [ astro - ph/9904182 ] .
lewis , g. f.and ibata , r. a. 2002 , mon . not .
r. astron .
337 , 26 f. giovi and l. amendola , mon . not
. soc . * 325 * , 1097 ( 2001 ) . c. s. kochanek , astrophys .
j. * 578 * , 25 ( 2002 ) . c. s. kochanek , and p. l. schechter , astro - ph/0306040 ( 2003 ) .
m. campos and j. a. de souza , astron . and astrophys .
422 , 401 ( 2004 ) .
r. c. santos and j. a. s. lima , astro - ph/0609129v2 ( 2007 ) .
p. r. k. sachs , proc .
london a * 264 * , 309 ( 1961 ) ; p. jordan , j. ehlers and r. k. sachs , akad .
mainz * 1 * , 1 ( 1961 ) .
i. m. h. etherington , phil . mag . * 15 * , 761 ( 1933 ) .
k. jassal , j. s. bagla and t. padmanabhan , astro - ph/0601389 ( 2006 ) .
j. s. alcaniz , j. a. s. lima , r. silva , ijmp d * 13 * , 1309 ( 2004 ) . | the presence of inhomogeneities modifies the cosmic distances through the gravitational lensing effect , and , indirectly , must affect the main cosmological tests . assuming that the dark energy is a smooth component , the simplest way to account for the influence of clustering is to suppose that the average evolution of the expanding universe is governed by the total matter - energy density whereas the focusing of light is only affected by a fraction of the total matter density quantified by the @xmath0 dyer - roeder parameter . by using two different samples of sne
type ia data , the @xmath1 and @xmath0 parameters are constrained by applying the zeldovich - kantowski - dyer - roeder ( zkdr ) luminosity distance redshift relation for a flat ( @xmath2cdm ) model .
a @xmath3-analysis using the 115 sne ia data of astier _ et al .
_ sample ( 2006 ) constrains the density parameter to be @xmath4(@xmath5 ) while the @xmath0 parameter is weakly limited ( all the values @xmath6 $ ] are allowed even at 1@xmath7 ) . however , a similar analysis based the 182 sne ia data of riess _ et al . _
( 2007 ) constrains the pair of parameters to be @xmath8 and @xmath9 ( @xmath5 ) .
basically , this occurs because the riess _
et al .
_ sample extends to appreciably higher redshifts . as a general result , even considering the existence of inhomogeneities as described by the smoothness @xmath0 parameter
, the einstein - de sitter model is ruled out by the two samples with a high degree of statistical confidence ( @xmath10 and @xmath11 , respectively ) .
the inhomogeneous hubble - sandage diagram discussed here highlight the necessity of the dark energy , and a transition deceleration / accelerating phase at @xmath12 is also required
. 0.5 cm |
the differential cross - section @xmath6 for a lepton ( @xmath7 , @xmath8 ) with four - momentum @xmath9 scattering off a nucleon with four - momentum @xmath10 can be expressed as @xmath11 \nonumber\end{aligned}\ ] ] where @xmath4 is the four - momentum transfer squared , @xmath12 is the bjorken scaling variable , @xmath13 is the inelasticity variable and @xmath14 .
the contribution from @xmath1 dominates the cross - section .
the contribution from @xmath3 is a qcd correction which is important only at large @xmath15 and that from @xmath16 is negligible for @xmath17 . to investigate sensitivity to @xmath3 at large @xmath15 or @xmath16 at large @xmath4
, the reduced cross - section @xmath18 is adopted . in the quark parton model ( or in the dis scheme of nlo qcd ) and for @xmath17 @xmath19
is the charge - weighted sum of the quark densities @xmath20 where @xmath21 $ ] is the _ singlet _ summed quark and anti - quark distributions .
similarly , the charged current cross - section @xmath22 at hera can be expressed as @xmath23 .
\nonumber\end{aligned}\ ] ] for fixed - target @xmath24 experiments , @xmath25 , and in the qpm @xmath26 where @xmath27 $ ] is the _ non - singlet _ difference of these distributions . in fig .
[ kine ] , the kinematic plane covered by the @xmath28 measurements is shown , including the new preliminary datasets from h1 and zeus which are seen to extend to : low @xmath15 ( @xmath29 ) providing overlap with the fixed - target experiments ; very low @xmath0 ( @xmath30 ) at low @xmath4 exploring the transition region from soft to hard physics ; high @xmath15 ( @xmath31 ) giving sensitivity to @xmath3 ; high @xmath32 probing sensitivity to electroweak effects in @xmath1 and @xmath16 as well as constraining the valence quarks at large @xmath4 .
the fixed - target experiments nmc , bcdms , e665 and slac experiments have provided final measurements at higher @xmath0 and lower @xmath4 . new information was presented at ichep98 from ccfr , e866 and the tevatron which also constrain the medium - high @xmath0 partons . theoretically , the directions in @xmath2 can be mapped out according to the dominant dynamical effects , as illustrated in fig .
[ evol ] . given a phenomenological input as a function of @xmath0 , the parton distributions
are evolved to different physical scales ( @xmath4 ) via the dglap evolution equations .
the alternative bfkl approach is to attempt to calculate the @xmath0 dependence directly from pqcd , where the running of the effective coupling constant is neglected to leading order .
bfkl predicts an @xmath33 dependence of @xmath1 at small @xmath0 .
the bfkl equation has recently been calculated to nlo .
@xcite these corrections are numerically very large in the experimentally accessible low ( @xmath34 ) range , resulting in @xmath35 being reduced to @xmath36 .
the understanding of the nlo corrections to the bfkl equation is therefore an active area of study .
@xcite in the dlla ( ` double leading log approximation ' ) , non - leading @xmath37 terms can also be evaluated , but a method which reliably maps the complete region of @xmath2 in terms of pqcd is still not known .
the region of high parton density may be reached at very low @xmath0 , where these approaches are not strictly valid .
the expectation from the glr equation @xcite is that the region where the partons overlap is accessible at slightly higher @xmath0 for decreasing @xmath4 .
however , @xmath4 should also be sufficiently large that higher twist and non - perturbative effects parameterised in terms of regge exchanges can be neglected . in the dglap approach
, the non - singlet contribution evolves as @xmath38 where @xmath39 and the @xmath40 s represent the nlo dglap splitting probabilities for radiating a parton with momentum fraction @xmath0 from a parton with higher momentum @xmath41 .
quantities such as @xmath16 provide a measure of @xmath42 which is insensitive to the a priori unknown gluon distribution .
similarly , the singlet quark and gluon densities are coupled via @xmath43 \otimes \left(\begin{array}{c}\sigma(x',q^2 ) \\
g(x',q^2 ) \end{array}\right)\ ] ] and quantities such as @xmath1 provide input for @xmath44 as well as coupled knowledge of @xmath42 and the gluon , @xmath45 . at the starting scale , @xmath46 , the light valence quarks ( @xmath47 ) and the sea of quark and anti - quarks ( @xmath48 ) as well as the gluon ( @xmath49 )
are attributed a given functional form .
for example in the mrst parameterisations @xmath50 where some parameters are set to 0 or fixed by sum rules and differences of @xmath51 and @xmath52 quarks may additionally be constrained .
the heavy quark ( sea ) contributions are calculated explicitly at nlo and their uncertainty is typically determined by the range allowed for the effective mass of the heavy quark .
the measured structure functions are then described by the convolutions of the light quark densities with the appropriate nlo matrix elements .
the outlined procedure defines the structure of a nucleon in terms of its constituent quarks and gluons .
however , guided by the new datasets and analyses which were presented to this conference , it should be noted that the following assumptions are made : * @xmath53 ( dglap ) terms are large compared to @xmath54 ( bfkl ) terms in the perturbative splitting functions ; at small @xmath0 are presently unable to distinguish these bfkl terms .
the status of forward jet production searches which enhance sensitivity to these effects is reviewed by j. huston .
@xcite ] * higher - twist ( ht ) contributions ( suppressed by factors of @xmath55 ) are negligibly small ; * nuclear binding effects are treated as small corrections or ignored in analyses of deuteron data . in table
[ tab : mrst ] , the experimental datasets considered in the mrst analysis are listed along with the underlying physics process and the parton behaviour which is being probed .
the experiments denoted by @xmath56 correspond to final measurements reported at ichep98 .
those denoted by @xmath57 correspond to new preliminary measurements reported at this conference .
the latest global fits of mrst @xcite , grv98 @xcite and cteq4 @xcite are used here to compare with the data .
[ cols= " < , < , < " , ]
the hera @xmath58 collider has improved its performance in successive years , providing large @xmath59 datasets from 1994 - 97 running . during this period 27.5
gev positrons collided with 820 gev protons resulting in data samples of 37pb@xmath60 and 46pb@xmath60 for h1 and zeus , respectively .
as a measure of the progress which has been made in the study of the scaling violations of @xmath1 , fig . [ f2vq ] illustrates how the latest hera data , presented for the first time at ichep98 , extend the reach in @xmath4 beyond 10,000 gev@xmath61 with @xmath0 extending up to 0.65 . at high @xmath4 values , the effects of @xmath62 exchange are significant and the value of @xmath63 is quoted where @xmath64 where @xmath63 , @xmath65 and @xmath66 are the contributions due to photon exchange , @xmath67 interference and @xmath68 exchange , respectively
. the data also extend to very low @xmath0 below @xmath69 : here it should be noted that the precision of the low-@xmath70 data is now comparable to that of the fixed - target data at higher @xmath0 .
there is a region of overlap where the hera and fixed - target experiments can be compared .
in particular , the analysis of the zeus results at low @xmath15 may resolve the ccfr - nmc discrepancy discussed earlier in the context of the 5/18ths rule " .
however , this comparison ( not shown ) is currently inconclusive , with the zeus data lying between the nmc and ccfr data .
_ transition region : _ the rise of @xmath1 with decreasing @xmath0 or , equivalently , the rise of @xmath71 with increasing @xmath72 has stimulated significant experimental and theoretical developments in the understanding of qcd at high energies .
one challenge is to explore how and where the transition occurs from soft to hard physics and interpret low-@xmath4 data .
measurements have been performed using dedicated low - angle taggers ( e.g. the zeus bpc ) and shifted vertex ( svx ) data @xcite as well as qed initial state radiation ( isr ) data @xcite in order to map out this region . in fig .
[ xn01 ] , a compilation of the latest measurements available from hera and e665 indicates that the different experiments and techniques agree with a precision of around 5% from the most recent data .
the significant rise of @xmath1 is apparent for @xmath73 gev@xmath61 , a behaviour which is described by the zeus nlo ( pqcd ) fit .
this behaviour is not reproduced in the dl ( donnachie - landshoff regge ) fit .
the zeus collaboration has performed both types of fits to the @xmath1 data exploring this transition region .
@xcite in order to determine the relationship between low-@xmath4 zeus bpc data measured in the range @xmath74 gev@xmath61 and @xmath75 gev@xmath61 data , a generalised vector meson dominance ( gvmd ) approach can be taken .
gvmd relates the virtual - photon cross - section to the real cross - section via @xmath76 for fixed @xmath72 and @xmath77 .
a good description of the data is found with @xmath78 gev@xmath61 .
regge theory then determines the @xmath72 dependence of the data , combined with lower energy photoproduction experiments , as @xmath79 where the reggeon intercept @xmath80 is fixed to 0.5 from hadroproduction data and lower energy photoproduction data also constrain @xmath81 , @xmath82 and @xmath83 . from the bpc data alone , the pomeron intercept value is @xmath84 to be compared with the donnachie - landshoff value @xmath85 . in this @xmath4 range ,
the rise of the cross - section is therefore relatively modest . combining the gvmd and regge approaches ,
the resulting zeusregge fit is used to parameterise the @xmath4 and @xmath72 dependence of the low @xmath86 gev@xmath61 data .
the zeus nlo qcd fit to the zeus 94 and zeus svx data , incorporating nmc and bcdms data , ( discussed below ) is used to determine the behaviour and the uncertainties on the gluon and singlet quark densities at low @xmath0 for @xmath87 gev@xmath61 . to quantify the behaviour of @xmath1
, fits to the e665 and zeus data of the form @xmath88 are performed .
the parameter @xmath89 for @xmath90 is then plotted as a function of @xmath4 in fig .
[ xn02 ] . a relatively slow transition from @xmath91
is observed with increasing @xmath4 . also shown
are fits to the dl and zeusregge parameterisations , fitted over the same @xmath0 range as the data , for @xmath92 gev@xmath61 .
these describe the data reasonably but are systematically lower . for @xmath93 gev@xmath61 , the data are compared to the grv94 prediction where the starting scale for the evolution of the parton densities is rather low ( @xmath94 gev@xmath61 ) and pqcd evolution generates the rise at small @xmath0 : this approach is observed to reasonably describe the data and the description is improved using the grv98 pdf ( not shown ) .
@xcite the rise at small @xmath0 is also described by the zeusqcd fit , where the zeus data is used as an input .
this rise of @xmath1 with decreasing @xmath0 is intimately coupled to the scaling violations via the gluon density ( in leading order @xmath95 neglecting sea quark contributions ) . in fig .
[ xn03 ] , fits of the form @xmath96 have been performed to the zeus data and the parameterisations discussed above . for @xmath97 ,
corresponding to @xmath98 gev@xmath61 there is a qualitative change in behaviour where the scaling violations stabilise and then decrease for lower-@xmath0 values , a behaviour which is not reproduced by the grv94 pdf .
the question is whether this scaling violation behaviour and the slow onset of the rise of @xmath1 with decreasing @xmath0 can be simultaneously understood .
a. mueller has discussed the scaling violation behaviour in terms of a geometric model where the spatial extent , @xmath99 , of the @xmath100 fluctuation of the virtual photon in the @xmath101 fixed - target frame is related to the height of the plateau in the scaling violations @xcite @xmath102 for @xmath103 gev@xmath104 and @xmath105 gev@xmath61 , the spatial extent @xmath106 fm .
this appears to be the scale at which a transition takes place and the partons in the proton start to overlap .
@xcite however , perturbative qcd is pushed to its limit and it will be important to test that the parton densities extracted from @xmath1 can be universally applied .
the zeus nlo fit to the @xmath87 gev@xmath61 data describes the data , demonstrating that there is sufficient flexibility in such an approach to go down to relatively low @xmath4 .
however , the relatively stable scaling violations observed around @xmath107 gev@xmath61 in fig .
[ xn03 ] yield a gluon contribution which is rapidly diminishing at small-@xmath0 and which is in fact smaller than the singlet sea quark contribution for small starting scales . for larger @xmath4 values
the gluon dominates the sea and we have an intuitively appealing picture where gluons radiate sea quarks whereas , in this low-@xmath4 region , the sea appears to be driving the gluon at low @xmath0 .
whether such low-@xmath4 partons are universally valid could be tested using e.g. low-@xmath4 diffractive vector meson data .
@xcite an important part of the zeus nlo qcd fit is the determination of the uncertainties on the gluon and singlet quark densities .
these are given for the gluon distribution in fig .
the overall uncertainty is estimated by combining in quadrature : the experimental systematic uncertainties on the zeus as well as the nmc and bcdms data ; the theoretical uncertainties on @xmath108 by @xmath109 , the relative strange quark content by @xmath110 and the charm mass by @xmath111 gev ; and , the parameterisation uncertainties on the starting scale , by varying @xmath112
gev@xmath61 , as well as using a more flexible form of the input in terms of chebycheff polynomials and redefining the @xmath113 including @xmath114 errors .
these variations correspond to a precision on the gluon of @xmath115 at @xmath116 gev@xmath61 where the renormalisation and factorisation scales are set to @xmath117 .
the role of the scale uncertainties is discussed in @xcite with respect to future capabilities to determine @xmath108 at hera .
the theoretical and parameterisation uncertainties are amplified at low @xmath4 such that the gluon is rather poorly determined from the scaling violations of @xmath1 in the transition region discussed above .
( in @xmath118 scheme ) but the physical quantities , @xmath1 , @xmath119 and @xmath3 remain positive within the quoted errors . ]
it is clear , however , that the gluon is significantly suppressed at low @xmath4 . _
@xmath119 determination : _
@xmath120 measurements in dis provide a significant test of the gluon density of the proton determined from the scaling violations of @xmath1 .
they also help to constrain theoretical uncertainties in the fits to @xmath1 where different prescriptions for charm production are adopted .
the zeus preliminary cross - section @xmath121 nb is measured in the range @xmath122 gev@xmath61 , @xmath123 gev , and @xmath124 . in general ,
the h1 @xcite and zeus @xcite data agree with the harris - smith nlo calculations @xcite where the fraction @xmath125 is determined from lep data @xcite , the peterson fragmentation function is characterised by @xmath126 and the renormalisation and factorisation scales are set to @xmath127 .
there is , however , a small discrepancy at lower @xmath128 and higher @xmath129 , corresponding to the proton direction , where the data lie above the prediction .
a similar discrepancy is also observed in the first analysis in the semi - leptonic decay channel .
@xcite together these results indicate that the fragmentation of charm in @xmath58 processes is worthy of further investigation . at this stage
, it is reasonable to extrapolate the measured cross - section to the full @xmath130 range to determine @xmath131 via the expression @xmath132\ ] ] where the @xmath133 contribution can be estimated as a qcd correction . in fig .
[ ch02 ] , the hera @xmath131 data mirror the rise of @xmath1 at small @xmath0 .
the data are in agreement with the grv94 pdf , where the band represents an estimated theoretical uncertainty due to the effective charm mass ( @xmath134 gev ) .
this comparison verifies the steep rise of the gluon density at low @xmath0 with a precision of @xmath135 . _
@xmath3 determination : _ the contribution of @xmath3 enters as a qcd correction to the total dis cross - section where @xmath136 . as such it provides an additional method to calibrate the gluon at low @xmath0 .
@xcite have used two methods to extract @xmath3 from the reduced cross - section @xmath137 at high @xmath15 .
this is the region where the scattered electron energy is low : in the h1 analysis scattered positrons are measured down to @xmath138 gev and backgrounds reduced by requiring the associated track to have correct charge .
@xmath3 is determined as a function of @xmath139 gev@xmath61 by measuring local derivatives of @xmath140 and observing deviations from a straight line at high @xmath15 .
these data are denoted by the stars in fig .
[ defl ] . here , the data are compared to an earlier extrapolation method @xcite applied to the same data , which yields consistent results , as well as to the h1 nlo qcd fit to h1 , nmc and bcdms @xmath1 data .
the data are in agreement with the qcd expectations although there is an indication of a relatively large @xmath3 contribution at the highest @xmath15 ( corresponding to lowest @xmath0 ) values . in conclusion
, a consistent value for the gluon density at low @xmath141 may be extracted from the data on @xmath1 , @xmath119 and @xmath3 with a precision of @xmath115 at @xmath116 gev@xmath61 .
the hera collider provides a unique window to explore @xmath58 interactions at the highest energies , extending the range of momentum transfer @xmath4 by about two orders of magnitude compared to fixed - target experiments .
as the hera luminosity increases we explore the region of @xmath142 gev@xmath61 , where electroweak effects play a rle .
it is in this unexplored kinematic region that we are sensitive to deviations from the standard model ( sm ) . in 1997 , h1 @xcite and zeus @xcite reported an excess of events compared to the sm predictions from the neutral current ( nc ) data taken during 1994 to 1996 .
for the h1 analysis an accumulation of 7 events in a reconstructed @xmath143 mass window of 200@xmath14412 gev was found , compared to an expectation of @xmath145 from 15 pb@xmath60 of data .
one further event was found from the 1997 data corresponding to a further 22 pb@xmath60 , yielding 8 events compared to an expectation of @xmath146 . for the zeus analysis the observed rates agreed with expectations except for an excess at the highest @xmath4 where two outstanding events with @xmath147 gev@xmath61 were observed from a luminosity of 20 pb@xmath60 .
these events still clearly stand out but no new nc outstanding events are observed in the 1997 data , corresponding to a further 26.5 pb@xmath60 . similarly in the charged current ( cc ) channel , the number of events is higher than expectations but is consistent with the standard model .
attention has therefore focussed on measuring the cross - sections at the highest accessible @xmath4 values .
the theoretical uncertainty on the cross - sections was determined as discussed w.r.t .
the zeus low-@xmath4 nlo qcd fit using high-@xmath0 @xmath148 and @xmath149 data to yield sm cross - section uncertainties of @xmath150 6 - 8% on the nc cross - section and @xmath150 6 - 12% on the cc cross - section at the highest accessible @xmath4 values .
these cross - sections therefore represent a benchmark for the standard model .
the cross - sections , discussed below , are corrected to the electroweak born level and integrated over the measured range of @xmath15 for h1 @xcite and corrected to the complete @xmath15 range for zeus .
@xcite _ neutral current cross - sections : _ high-@xmath4 neutral current events are easily identified from the high - energy scattered positron .
the cross - section is particularly sensitive to the valence @xmath51-quark distribution in the proton @xmath151.\ ] ] here , @xmath1 is the generalised structure function , incorporating @xmath152 and @xmath68 terms , which is sensitive to the singlet sum of the quark distributions @xmath153 and @xmath16 is the parity - violating ( @xmath68-contribution ) term which is sensitive to the non - singlet difference of the quark distributions @xmath154 .
the data are now becoming sensitive to electroweak @xmath67 interference effects , which suppress the nc cross - section by @xmath155 30% for @xmath156 gev@xmath61 .
@xcite in the upper plot of fig .
[ ncxs ] , the h1 cross - section is observed to fall over more than six orders of magnitude .
the ratio of the data to the sm , adopting the h1 nlo qcd fit , is shown in the lower plot of fig .
[ ncxs ] where agreement is observed up to @xmath4 @xmath150 30,000 gev@xmath61 .
comparison of the the data uncertainties with those from theory ( not shown ) indicates that the data will constrain the parton densities of the proton at large @xmath0 . _
charged current cross - sections : _ charged current events are identified by their missing transverse momentum ( @xmath157 ) due to the escaping neutrino .
the cross - section is sensitive to the valence @xmath52-quark distribution in the proton @xmath158.\ ] ] in the zeus analysis , @xmath159 was measured for @xmath160 gev@xmath61 using the jacquet - blondel method where @xmath161 , with an rms resolution on @xmath4 @xmath162 , reflecting the @xmath163 hadronic energy resolution .
the systematic uncertainties , mainly due to the hadronic energy scale uncertainty of @xmath164 , correspond to @xmath165 uncertainties on the cross - section at lower @xmath4 but increase at larger @xmath0 and @xmath4 . in the upper plot of fig .
[ ccxs ] the cross - section is observed to fall over more than four orders of magnitude .
the ratio of the data to the sm , adopting the cteq4d pdf , is shown in the lower plot of fig .
[ ccxs ] where good agreement is observed up to @xmath4 of @xmath150 10,000 gev@xmath61 .
comparison of the the data uncertainties with those from theory ( shaded band ) indicates that the data will help to constrain the @xmath52-quark densities at large-@xmath0 .
the cross - section is suppressed relative to nc @xmath152 exchange due to the propagator term : this characteristic dependence on @xmath4 has been fitted to yield a value for the mass of the exchanged ( space - like ) @xmath72-boson of @xmath166 with an additional pdf uncertainty of @xmath167 gev estimated in the zeus analysis and an uncertainty of @xmath168 gev due to electroweak radiative corrections estimated in the h1 analysis .
returning to the bodek - yang analysis , @xcite discussed in relation to the fixed - target results , the zeus cc cross - section is plotted as function of @xmath0 for @xmath160 gev@xmath61 in fig .
the increase in the ratio of @xmath169 corresponds to an increase of the cc cross - section at high @xmath0 .
the uncertainties on the data are large in this region , but this modification does result in better agreement with the data than the standard pdf s at large @xmath0 .
a comparison of the h1 and zeus nc and cc cross - sections for @xmath170 gev@xmath61 is given in fig .
the nc and cc data are in agreement and both cross - sections agree with the sm prediction over a broad range range of @xmath4 . at high @xmath171 gev@xmath61 ,
the cc cross - section is suppressed relative to the nc cross - section due to the @xmath169 ratio being less than unity .
the measurement of the hera cc / nc ratio of cross - sections will therefore provide a direct determination of this ratio , free from the uncertainties due to nuclear binding effects .
there were many highlights in our deepening understanding of nucleon structure presented at the ichep98 conference .
ccfr have performed final analyses of their structure function data which lead to precise tests of qcd .
nutev already provide electroweak input and aim to reduce the uncertainty on @xmath108 to @xmath172 .
important developments have been made in the understanding of higher twist effects in terms of renormalon theory .
input from various experiments and a reassessment of the importance of nuclear binding effects in the deuteron lead to the conclusion that the ratio of @xmath169 parton densities increases , whilst the ratio of @xmath173 decreases compared to earlier determinations . the second phase of spin experiments are now exploring spin structure via nlo qcd fits .
the first observations of scaling violations from combined fits to the world data indicate that the source of the spin puzzle lies with the gluon spin density , @xmath174 , although large uncertainties remain .
semi - inclusive measurements now provide input on the quark composition of the spin .
the first discovery at hera was the rise of @xmath1 at low @xmath0 .
precise data now enable the rise of the associated scaling violations with decreasing @xmath0 and hence the gluon to be determined .
the determinations of @xmath119 and @xmath3 enable this gluon distribution to be calibrated .
the fall of these scaling violations at low @xmath2 enables the region where parton confinement effects take place to be explored at high energies .
the large @xmath59 data sample enables cross - sections to be measured at very large @xmath4 where electroweak effects start to play a rle .
the hera data are consistent with the standard model and place constraints on the parton densities at large @xmath0 .
the outlook is for a similar sample of @xmath175pb@xmath60 of @xmath176 data in the next two years , prior to the hera upgrade where luminosities will be increased five - fold . deep inelastic scattering has historically led to the discovery of the nucleus , quarks and electroweak neutral currents .
the discovery potential of current and future experiments is high and the field continues to provide important input to our understanding of sub - nuclear structure .
it is a pleasure to thank alan astbury and co - organisers for an excellent conference .
many thanks to halina abramowicz , arie bodek , antje brull , allen caldwell , abe deshpande , john dainton , robin devenish , martin erdmann , laurent favart , thomas gehrmann , tim greenshaw , beate heinemann , peppe iacobucci , robert klanner , max klein , masahiro kuze , ludger lindemann , alan martin , gavin mccance , jan okrasiski , jen - chieh peng , alex prinias , robert waugh , arnulf quadt , david saxon , stefan schlenstedt , mike vetterli , manuella vincter , bryan webber , jim whitmore , un - ki yang , rik yoshida , jaehoon yu and all the speakers in the structure functions parallel session for their help , inspiration and advice .
i am grateful to the alexander von humboldt foundation , desy and pparc for their financial support .
99 v. fadin and l. lipatov , hep - ph/9802290 .
forshaw , g.p .
salam and r.s .
thorne , hep - ph/9812304 .
gribov , e.m . levin and m.g .
ryskin , ; e.m . levin and m.g .
j. huston , ichep98 proceedings .
y. dokshitzer , ichep98 proceedings .
m. erdmann , ichep98 proceedings .
martin , r.g .
roberts , w.j . stirling and r.s .
thorne , hep - ph/9803445 .
m. glck , e. reya and a. vogt , hep - p/9806404 .
cteq collab . , h.l .
lai et al . , .
s.a . larin and j.a.m .
vermarseren , .
a.l . kataev and v.v .
starshenko , .
kim , d.a .
harris et al . , hep - ex/9808015 , submitted to .
kulagin , hep - ph/9809219 .
kataev , g. parente and a.v .
sidorov , hep - ph/9809500 .
ccfr collab . ,
w. seligman et al .
, hep - ph/9701017 . m. virchaux and a.m. milsztajn , . c. caso et al . , . f. eisele , physics in collision proceedings , hep - ph/9807028 .
u. yang and a. bodek , ichep98 proceedings , hep - ph/9809480 .
l. frankfurt and m. strikman , .
m. dasgupta and b.r .
ccfr collab . , u.k.yang et al . , dis98 proceedings , hep - ex/9806023 ;
j. yu , ichep98 proceedings .
cdf collaboration , f. abe et al .
, hep - ex/9809001 , submitted to ; w. sakumoto , ichep98 proceedings . nmc collab . , m. arneodo et al . , . s. kumano and k. umekawa , hep - ph/9803359 . e866 collab . ,
hawker et al . , .
e866 collab . , j.c .
peng et al . , hep - ex 9804288 , submitted to .
hermes collab .
, k. ackerstaff et al .
, hep - ex/9807013 ; m. vincter , ichep98 proceedings . c. boros , j. londergan and a. thomas , hep - ph/9804410 .
c. boros , ichep98 proceedings .
ccfr collab . , a. bazarko et al . , .
d. karlen , ichep98 proceedings .
smc collab . , b. adeva et al . , cern - ep/98 - 86 ,
submitted to ; a. deshpande , ichep98 proceedings . g. altarelli , r.d .
ball , s. forte and g. ridolfi , .
l. sorrell , ichep98 proceedings .
e. rondio , ichep98 proceedings .
f.e . close and r.g .
roberts , .
j. ellis , dis98 proceedings ( world scientific ) in press ; j. ellis and m. karliner , .
b. lampe and e. reya , hep - ph/9810270 .
smc collab . , d. adeva et al . , hep - ex/9711008 , submitted to . c.a .
miller , ichep98 proceedings .
zeus collab . , j. breitweg et al . , desy 98 - 121 , submitted to ; a. quadt , ichep98 proceedings .
h1 collab .
, c. adloff et al .
, ichep98 contributed papers 534 and 535 ; m. klein , ichep98 proceedings .
mueller , dis98 proceedings ( world scientific ) in press .
m. botje , dis97 proceedings , .
h1 collab .
, c. adloff et al . , ichep98 contributed papers 538 and 540 ; e. tzamariudaki , ichep98 proceedings .
zeus collab . , j. breitweg et al . , ichep98 contributed papers 768 and 772 ; w. verkerke , ichep98 proceedings .
harris and j. smith , ; b.w .
harris and j. smith , ; b.w .
harris and j. smith , hep - ph/9706334 .
opal collab .
, k. ackerstaff et al .
h1 collab .
, s. aid et al .
h1 collab .
, c. adloff et al . , .
zeus collab . , j. breitweg et al . , .
h1 collab . , c. adloff et al . , ichep98 contributed paper 533 ; m. fleischer , ichep98 proceedings .
zeus collab . , j. breitweg et al . ,
ichep98 contributed papers 751 and 752 ; j. grosse - knetter , ichep98 proceedings . | the latest structure function results , as presented at the ichep98 conference , are reviewed .
a brief introduction to the formalism and the status of global analyses of parton distributions is given .
the review focuses on three experimental areas : fixed - target results and their constraints on the parton densities at high @xmath0 ; spin structure and spin parton densities as well as the status of the associated sum rules ; hera results on the dynamics of @xmath1 at low @xmath2 , charm and @xmath3 as well as the measurement and interpretation of the high-@xmath4 neutral and charged current cross - sections . _
plenary talk presented at the xxix ichep98 conference , _
_ vancouver , july 1998 .
_
_ slides are available from _ _ http://www-zeus.desy.de/conferences98/#ichep98 _ 17.0 cm -1.0 in -42pt plus 2 mm minus 2 mm epsf = cmr8 1.5pt # 1#2#3#4#1 * # 2 * , # 3 ( # 4 ) glas - ppe/980716@xmath5 december 1998 |
one of the major discoveries in physics during the last two decades was the accelerated expansion of the universe .
general relativity and the standard model of particle physics fail to explain this phenomenon .
this situation calls for new alternative ideas able to give a satisfactory explanation of the cosmological observations .
one of the possibilities is to go beyond general relativity and to consider more general theories of gravity . among the most natural generalizations of the original einstein theory are the scalar - tensor theories @xcite .
these theories are viable gravitational theories and can pass all known experimental and observational constraints .
in addition , they can explain the accelerated expansion of the universe .
the scalar - tensor generalizations of the original einstein theory naturally arise in the context of the modern unifying theories as string theory and kaluza - klein theories . in scalar - tensor theories the gravitational interaction
is mediated not only by the spacetime metric but also by an additional scalar field . from a physical point of view
this scalar field plays the role of a variable gravitational constant .
general relativity ( gr ) is well - tested in the weak - field regime , whereas the strong - field regime remains largely unexplored and unconstrained . in the strong - field regime one
expects the differences between gr and alternative theories of gravity to be more pronounced .
the natural laboratories for testing the strong - field regime of gravitational theories are compact stars and black holes .
there exist scalar - tensor theories which are indistinguishable from gr in the weak - field regime but which can differ significantly from gr in the strong - field regime .
an example of such a phenomenon is the so - called spontaneous scalarization , observed in a certain class of scalar - tensor theories . when spontaneous scalarization takes place , in addition to the general relativistic solutions with a trivial scalar field , there exist further solutions with a nontrivial scalar field .
in fact , these scalarized solutions are energetically more favorable than their gr counterparts .
spontaneous scalarization was first observed for neutron stars @xcite , where _ spectacular changes _ were seen in static equilibrium configurations for a given nuclear equation of state .
more recently , spontaneous scalarization was also observed in rapidly rotating neutron stars @xcite , where the deviations of the rapidly rotating scalar - tensor neutron stars from the general - relativistic solutions were even significantly larger than in the static case .
spontaneous scalarization was also observed for static uncharged and charged boson stars @xcite .
the first purpose of the present paper is to study rapidly rotating boson stars in scalar - tensor theories , and to establish the phenomenon of spontaneous scalarizarion for these stationary compact objects .
the second purpose of this paper is to address the existence of scalarized hairy black holes . in general relativity ( gr ) rotating vacuum black holes
are described in terms of the kerr solution .
this solution specifies the full spacetime in terms of only two parameters , its mass and its angular momentum .
hairy black holes appear , when suitable matter fields are included .
examples are chiral fields , yang - mills and higgs fields , yielding hairy static black holes @xcite as well as rapidly rotating hairy black holes @xcite .
recently it was noted , that also a single complex scalar field allows for hairy black holes , provided the black holes are rotating @xcite . in fact , these solutions maybe viewed as a generalization of rotating boson stars , that are endowed with a horizon .
the regular boson stars form part of the boundary of the domain of existence of this new type of hairy black holes .
the other parts of the boundary exist of extremal hairy black holes and scalar clouds .
here we show , that besides these rapidly rotating hairy black holes , already present in gr , scalar - tensor theory again allows for the phenomenon of scalarization .
in particular , we study the physical properties of these scalarized hairy black holes , and map their domain of existence .
denoting the gravitational scalar by @xmath0 , the gravitational action of scalar - tensor theories in the physical jordan frame is given by @xmath1 , \end{aligned}\ ] ] where @xmath2 is the bare gravitational constant , @xmath3 is the spacetime metric , @xmath4 is the ricci scalar curvature , and @xmath5 $ ] denotes the action of the matter fields . the functions @xmath6 , @xmath7 and @xmath8 are subject to physical restrictions : we require @xmath9 , since gravitons should carry positive energy , and @xmath10 ^ 2 \ge 0 $ ] , since the kinetic energy of the saclar field should not be negative .
the matter action @xmath11 depends on the matter field @xmath12 and on the space - time metric @xmath3 .
the matter action does not involve the gravitational scalar field @xmath0 in order to satisfy the weak equivalence principle .
variation of the action with respect to the spacetime metric and the gravitational scalar as well as the matter field leads to the field equations in the jordan frame
. however , these field equations are rather involved .
it is therefore easier to consider a mathematically equivalent formulation of scalar - tensor theories in the conformally related einstein frame with metric @xmath13 @xmath14 in the einstein frame the action then becomes ( up to a boundary term ) @xmath15 , \end{aligned}\ ] ] where @xmath16 is the ricci scalar curvature with respect to the einstein metric @xmath13 , @xmath17 represents the new scalar field defined via @xmath18 with the new functions @xmath19 by varying this action with respect to the metric in the einstein frame @xmath13 , the scalar field @xmath17 , and the matter field @xmath20 , we find the set of field equations in the einstein frame . in particular , we here consider a scalar tensor theory with potential @xmath21 and function @xmath22 @xcite @xmath23 we present a systematic study for the parameter value @xmath24 , but we have also done calculations for larger values of @xmath25 . for the matter action
we choose a complex boson field @xmath26 @xmath27 = - \int d^4x \sqrt{-g } \left [ \frac{1}{2 } { \cal a}^2(\varphi ) g^{\mu\nu } \left ( \psi _ { , \ , \mu}^ * \psi _ { , \ , \nu } + \psi _ { , \ , \nu}^ * \psi _ { , \ , \mu }
\right ) + { \cal a}^4(\varphi ) u ( \left| \psi \right| ) \right]\ ] ] with self - interaction potential @xmath28
to obtain stationary hairy black hole and boson star solutions we employ the line element @xmath29 + \bar{r}^2 \sin^2 \theta \left [ d\varphi - f_3 dt \right]^2 \right ) , \label{metric}\ ] ] with the metric functions @xmath30 , @xmath31 , and @xmath32 where @xmath33 denotes the horizon parameter .
for the boson stars we seet @xmath34 , i.e. @xmath35 .
likewise , we parametrize the gravitational scalar field @xmath0 by @xmath36 . for the boson field @xmath26
we adopt the stationary ansatz @xmath37 where @xmath38 is a real function , @xmath39 denotes the boson frequency , and as required by the single - valuedness of the scalar field
@xmath40 is an integer representing a rotational quantum number . in the lagrangian for the boson field
@xmath26 we employ a quartic self - interaction potential ( [ smatu ] ) with coupling constant @xmath41 .
while we present our results for the value @xmath42 , we have also performed calculations for other values of @xmath41 . as in @xcite
we introduce a new radial coordinate @xmath43 such that @xmath44 , and the event horizon is located at @xmath45 .
the functions satisfy the following set of boundary conditions , obtained from the requirements of asymptotic flatness as well as of regularity at the origin and the event horizon in the case of the boson star , resp .
black hole solutions @xmath46 and @xmath47 , resp .
@xmath48 for boson star and black hole solutions .
the mass @xmath49 and the angular momentum @xmath50 of stationary asymptotically flat space - times can be obtained in scalar - tensor theory - analogously to gr - from the asymptotic behavior of the metric functions @xmath51 and @xmath52 , respectively , @xmath53 and @xmath54 since the action is invariant under the global phase transformation @xmath55 , a conserved current arises @xmath56 it is associated with the noether charge @xmath57 representing the particle number , @xmath58 at the event horizon of the hairy black holes the killing vector @xmath59 @xmath60 is null , and @xmath61 represents the horizon angular velocity .
hairy black holes satisfy @xcite @xmath62 we denote the horizon area by @xmath63 in the einstein frame , and define the areal horizon radius by @xmath64 the horizon temperature by @xmath65 is obtained from the surface gravity @xmath66 , @xmath67 and @xmath68 .
we note , that the limit @xmath69 comprises two different types of configurations : * extremal black holes are obtained , when @xmath70 , * globally regular solutions are obtained , when @xmath71 .
the latter correspond to rotating boson stars . for boson stars @xmath72 .
we have obtained rapidly rotating scalarized boson stars and hairy black hole solutions for a sequence of rotational quantum numbers @xmath73 . to this end
we have solved the set of coupled non - linear partial differential equations subject to the appropriate boundary conditions with a numerical algorithm based on the newton - raphson method @xcite . compactifying space by introducing the radial coordinate @xmath74 we have discretized the equations on a non - equidistant grid in @xmath75 and @xmath76 , with typical grid sizes on the order of @xmath77 , covering the integration region @xmath78 and @xmath79 . in the following
we will used scaled quantities @xmath80 , @xmath81 , @xmath82 @xmath83 , and @xmath84 , with @xmath85 , @xmath86 , @xmath87 , @xmath88 , where @xmath89 denotes the plack mass .
the globally regular ordinary boson star solutions form a large part of the boundary of the domain of existence of the hairy black hole solutions @xcite .
let us therefore first consider the boson star solutions . in fig .
[ fig1](a ) the scaled mass @xmath80 of the boson star solutions is exhibited versus the scaled boson frequency @xmath82 for rotational quantum numbers @xmath90 , 2 , 4 , 6 and 8 .
the families of ordinary boson star solutions emerge from the vacuum at @xmath91 .
they form a first ( and at least in part classically stable ) branch , until the mass reaches its maximal value .
we note , that this maximal value of the mass increases rapidly with @xmath40 . beyond the maximal mass the families of ordinary boson star solutions continue in a spiral - like manner for the lowest values of @xmath40 .
they end in a merger solution , where a branch of extremal hairy black hole solutions is encountered @xcite . for the higher @xmath40 ,
however , they feature only a single further branch , their second branch , before they merge with a branch of extremal hairy black hole solutions .
interestingly , each of these higher @xmath40 second branches of boson star solutions ends close to an extremal kerr black hole solution , possessing almost the same mass and the respective horizon angular velocity @xmath92 . when considering the scaled angular momentum @xmath93 of these ordinary boson star solutions versus the mass @xmath80 as exhibited in fig .
[ fig1](b ) together with the line @xmath94 of extremal kerr black holes , the branches form cusps at extremal values of the mass .
the lower @xmath40 solutions therefore feature several cusps , whereas the higher @xmath40 solutions have a single cusp . clearly , for the higher @xmath40 families of boson star solutions the second branches approach the extremal kerr value closely , when they end in a merger solution , where a branch of extremal hairy black hole solutions is encountered .
in addition , fig .
[ fig1 ] exhibits the scalarized boson star solutions associated with the ordinary boson star solutions . for a given value of @xmath40
the scalarization arises at a critical value of the boson frequency , @xmath95 , where a branch of scalarized boson star solutions emerges from the first branch of ordinary boson star solutions .
interestingly , @xmath95 is rather independent of the rotational quantum number @xmath40 .
the families of scalarized boson stars then extend up to a second critical value @xmath96 , where they merge again into the respective second branch of ordinary boson stars . since @xmath96 decreases with @xmath40 , the domain of existence of rapidly rotating scalarized boson stars increases with @xmath40 . for @xmath90 ,
the critical point @xmath95 is close to but slightly below the maximal value of the mass of the ordinary boson stars . since the mass of the scalarized boson stars decreases monotonically until the critical point @xmath96 is reached , and since the same holds for the particle number @xmath57 , the ordinary boson stars are stable with respect to scalarization along their first branch .
indeed , for a given value of @xmath57 along their first branch the mass of the ordinary boson stars is always lower than the mass of the scalarized boson stars , when these exist .
for @xmath97 the situation is analogous . for @xmath98 , however , the scalarized boson stars assume their maximal mass no longer at @xmath95 , but at a smaller value of @xmath39 .
thus they form a ( potentially ) stable branch , starting at the first minimum of the mass and extending until their global maximum . along this branch
, the scalarized boson stars represent the energetically favored solutions .
thus ordinary boson stars will be unstable with respect to scalarization in this range of frequencies .
we note , that the maximal mass of the scalarized boson stars increases with @xmath40 .
however , unlike the case of rapidly rotating neutron stars , where the maximal mass reached for scalarized neutron stars significantly exceeds the maximal mass of ordinary neutron stars , the maximal mass of scalarized boson stars does not deviate too strongly from the one of ordinary boson stars . considering the end point of the families of scalarized boson star solutions
, we note that for the larger values of @xmath40 , with increasing @xmath40 the end point gets closer to the end point of the respective family of ordinary boson star solutions .
in particular , the second branches of the ordinary boson star solutions shorten with increasing @xmath40 . for @xmath99 ,
the end points of both ordinary and scalarized boson star solutions are rather close to each other .
let us now turn to the hairy black holes .
starting from a generic ordinary boson star solution , a sequence of hairy black holes emerges , when the presence of a small horizon is imposed , and the horizon is then increased in size .
the domain of existence of hairy black holes is then mapped by varying the horizon size and the horizon angular velocity . for ordinary hairy black holes ,
the domain of existence has been studied before for @xmath90 and 2 @xcite , employing only a mass term for the boson field .
there it was shown , that the boundary of the domain of existence of these solutions consists of * the family of boson stars , and the associated families of * extremal hairy black holes and * scalar clouds @xcite .
we exhibit in fig .
[ fig2 ] the domain of existence of ordinary hairy black holes for the case of the @xmath100 potential , and rotational quantum numbers @xmath90 , 4 and 8 .
in particular , we here show the scaled mass @xmath80 ( left column ) and the scaled particle number @xmath81 ( right column ) of the hairy black holes versus the scaled boson frequency @xmath82 . we note , that for black holes @xmath101 . here
we have included the case @xmath90 , to allow for direct comparison with the case without self - interaction @xcite .
studies of the rotational quantum numbers @xmath102 were not reported before , neither with nor without self - interaction . in all these figures ,
the beige regions labelled gr represent the domain of existence of the ordinary hairy black holes .
the families of extremal hairy black hole solutions possess two endpoints . at one endpoint
they join the respective branch of globally regular boson stars solutions in a merger solution . at the other endpoint
they join precisely the respective branch of scalar cloud solutions .
these latter endpoints are marked in the figures by an asterisk . in these figures
the sets of extremal hairy black hole solutions have been obtained by extrapolation , where the horizon parameter @xmath33 was decreased towards zero . case . ]
the hair of these extremal black hole solutions becomes evident , for instance , when inspecting the scaled particle number @xmath81 of these extremal solutions : @xmath57 is always finite ( see fig .
[ fig2 ] ) . as seen in fig .
[ fig2 ] for @xmath90 , the extremal hairy black hole solutions still form part of a spiral , whereas for the higher values of @xmath40 this spiralling pattern is lost . for @xmath103 and in particular for @xmath99 ,
the mass of the extremal hairy black hole solutions is close to the mass of the extremal kerr black holes .
the particle number of these extremal hairy black holes is , however , clearly finite and reaches zero only when the scalar cloud solutions are reached .
the study of @xmath57 therefore helps to clarify the domain of existence and its boundaries .
let us now consider the domain of existence of the scalarized hairy black holes .
it is also exhibited in fig .
[ fig2 ] and marked by the green region labelled stt .
the upper boundary of this domain of existence is always given by the regular scalarized boson stars .
the lower boundary is reached somewhere within the domain of existence of ordinary hairy black holes , at the moment that the scalarization disappears .
thus we do not observe extremal scalarized hairy black holes .
there is always a part of the domain of existence of scalarized hairy black holes , where there are no ordinary hairy black holes .
this part increases with increasing @xmath40 . for the lower @xmath40 ,
scalarized black holes exist for smaller boson frequencies , for the higher @xmath40 scalarized hairy black holes reach higher values of the mass and the particle number than ordinary hairy black holes .
kerr black holes satisfy the bound @xmath104 . but this bound may be exceeded by hairy black holes @xcite .
we demonstrate this in fig .
[ fig3 ] for several families of hairy black holes .
in particular , we exhibit the scaled angular momentum @xmath105 versus the areal horizon radius @xmath84 ( [ r_h ] ) for hairy black holes with @xmath106 , @xmath103 and @xmath99 in gr and with scalarization ( stt ) .
we note , that scalarized hairy black holes can also exceed the kerr bound .
we exhibit in fig .
[ fig4 ] contour plots of the component @xmath107 of the energy momentum tensor ( left column ) and of the particle number density @xmath108 ( right column ) of a hairy black hole in gr ( top ) and a scalarized hairy black hole ( bottom ) . for comparison , we have chosen the same rotational quantum number @xmath99 , boson frequency @xmath109 and mass @xmath110 for these black holes .
the central black hole in gr has a horizon area of @xmath111 and is thus much smaller than the scalarized black hole with @xmath112 . both @xmath113 and @xmath114
are concentrated in tori around the central black hole .
the maximal value of @xmath113 is considerably larger for the gr black hole than for the scalarized one , while the maximal value of @xmath114 is only slightly larger . in the figures showing @xmath114
also the ergosurfaces are indicated .
ordinary hairy black holes can feature ergosurfaces consisting of an ergosphere and an ergoring , forming together an ergo - saturn @xcite .
this is also the case for the gr black hole shown .
here we note , that the same phenomenon may hold as well for scalarized hairy black holes .
their ergosurface may also represent an ergo - saturn , as depicted in the figure .
scalar - tensor theories of gravity offer several observable consequences ( see e.g. , @xcite and references therein ) . here
we have concentrated on the effect scalarization .
first , we have shown that scalarization occurs for rapidly rotating boson stars ( with a fourth order self - interaction ) .
rotating boson stars possess a rotational quantum number , an integer @xmath40 . constructing families of boson stars for @xmath115 , we have shown ,
that with increasing @xmath40 , the scalarization becomes more pronounced .
we expect this trend to continue .
subsequently , we have constructed hairy black holes . after mapping out the domain of existence of hairy black holes in gr , which is bounded by boson stars , extremal hairy black holes and scalar clouds , we have surveyed the domain of existence of scalarized black holes .
one boundary of their domain of existence is formed by scalarized boson stars .
the other boundary , however , is formed by ordinary hairy black holes . here the scalar field simply vanishes , thus reducing the solutions to general relativistic solutions with a trivial scalar field . we have shown that the physical properties of the scalarized hairy black holes resemble in many respects those of hairy black holes in gr .
for instance , they may substantially exceed the kerr bound @xmath104 , and they can exhibit ergosurfaces , that consist of two parts , forming an ergo - saturn .
the scalarization of rapidly rotating boson stars and hairy black holes allows their mass and particle number to exceed the maximal values allowed for their general relativistic counterparts .
this effect seen here for the larger values of the rotational quantum number may be viewed as a downscaled version of what has been observed for neutron stars .
it should be interesting to increase the rotational quantum number further , to see how strong the effect of scalarization may become for rotating hairy black holes .
we gratefully acknowledge support by the dfg within the research training group 1620 models of gravity and by fp7 , marie curie actions , people , international research staff exchange scheme ( irses-606096 ) .
we gratefully acknowledge discussions with e.radu .
j.k . gratefully acknowledges discussions with c.lmmerzahl .
99 p. jordan , nature * 164 * ( 1949 ) 637 .
m. fierz , helv .
acta * 29 * ( 1956 ) 128 .
p. jordan , z. phys .
* 157 * ( 1959 ) 112 . c. brans and r. h. dicke , phys .
* 124 * ( 1961 ) 925 .
r. h. dicke , phys .
* 125 * ( 1962 ) 2163 .
t. damour and g. esposito - farese , phys .
lett . *
70 * , 2220 ( 1993 ) .
d. d. doneva , s. s. yazadjiev , n. stergioulas and k. d. kokkotas , phys .
d * 88 * , no .
8 , 084060 ( 2013 ) .
d. d. doneva , s. s. yazadjiev , n. stergioulas , k. d. kokkotas and t. m. athanasiadis , phys .
d * 90 * , no .
4 , 044004 ( 2014 ) .
a. w. whinnett , phys .
d * 61 * , 124014 ( 2000 ) .
m. alcubierre , j. c. degollado , d. nunez , m. ruiz and m. salgado , phys .
d * 81 * , 124018 ( 2010 ) .
m. ruiz , j. c. degollado , m. alcubierre , d. nunez and m. salgado , phys .
d * 86 * , 104044 ( 2012 ) .
c. a. r. herdeiro and e. radu , phys .
lett . * 112 * , 221101 ( 2014 ) .
s. hod , phys .
d * 86 * , 104026 ( 2012 ) [ erratum - ibid .
d * 86 * , 129902 ( 2012 ) ] . c. herdeiro , e. radu and h. runarsson , phys .
b * 739 * ( 2014 ) 302 . c. herdeiro and e. radu , arxiv:1501.04319 [ gr - qc ] .
w. schnauer and r. wei , j. comput .
27 , 279 ( 1989 ) 279 ; + m. schauder , r. wei and w. schnauer , the cadsol program package , universitt karlsruhe , interner bericht nr . 46/92 ( 1992 ) | in the presence of a complex scalar field scalar - tensor theory allows for scalarized rotating hairy black holes .
we exhibit the domain of existence for these scalarized black holes , which is bounded by scalarized rotating boson stars and ordinary hairy black holes .
we discuss the global properties of these solutions . like their counterparts in general relativity , their angular momentum may exceed the kerr bound , and their ergosurfaces may consist of a sphere and a ring , i.e. , form an ergo - saturn . |
electromagnetic momentum density can be decomposed in terms of orbital momentum and spin momentum densities @xcite .
they are respectively responsible for the generation of the orbital angular momentum ( oam ) and spin angular momentum ( sam ) of electromagnetic ( em ) waves . left and right circularly polarized em waves carry sam of @xmath0 that is intrinsic ( origin - independent ) physical quantity . fundamentally different from sam , oam is an extrinsic origin - dependent quantity , which can be carried by vortex beams with a helical wavefront @xcite . on the other hand ,
the unbounded eigenstates of oam could enhance capacities of radio , optical and quantum communications @xcite . additionally , oam has various potential applications involving super - resolution imaging @xcite , optical tweezers @xcite , etc .
there are several approaches to generate oam of em waves .
one common approach is to introduce desired phase retardation by spiral phase plates @xcite , antenna arrays @xcite , holographic plates @xcite , etc .
another way is to harness abrupt phase changes by exploiting pancharatnam - berry phase concept @xcite . using anisotropic scatterers in a metasurface , with spatially varying orientations ,
a vortex beam with the helical phase can be created .
the main pitfall in current oam designs by metasurface is the low conversion efficiency from a plane wave with zero oam to the vortex beam with nonzero oam .
for example , a metasurface composed of v - shaped scatterers with varied geometric parameters @xcite was proposed to generate oam in the cross - polarized component of scattered field under a linearly polarized wave incidence .
the design achieved a polarization conversion of 30% .
another example is to employ the aperture antennas @xcite that act as linear polarizers .
an azimuthally polarized oam beam was generated under a circularly polarized incident wave .
the conversion efficiency limit is bounded by @xmath1 . in this paper , we propose two types of novel pec ( perfect electric conductor)-pmc ( perfect magnetic conductor ) anisotropic metasurfaces to overcome the low efficiency issue existing in current oam designs .
one of proposed metasurface could perfectly convert a left ( right ) circularly polarized plane wave carrying zero oam to a right ( left ) circularly polarized vortex beam carrying arbitrary order oam .
with azimuthally continuous loops , the other proposed metasurface generates much smoother near - field phase pattern than conventional metasurfaces with discrete scatterers .
for an anisotropic scatterer in metasurface , linear polarized jones vectors of the incident and scattered ( transmitted or reflected ) fields can be connected by the jones matrix @xmath2 @xmath3 where @xmath4 and @xmath5 are the @xmath6 and @xmath7 components of the incident electric field . @xmath8 and @xmath9 are the corresponding components of the scattered electric field .
if @xmath10 and @xmath11 , azimuthally rotating the scatterer by an angle of @xmath12 will result in a new jones matrix @xmath13 under circular basis , @xmath14 will convert to @xmath15 where @xmath16 connects the incident circular polarized jones vectors to the scattered circular polarized ones .
when @xmath17 by mirror symmetry @xcite , the scatterer flips the polarization state of an input beam from left ( right ) to right ( left ) circular polarization @xcite .
simultaneously , an additional uniform phase factor @xmath18 called pancharatnam - berry phase @xcite is introduced , which is able to produce an oam value of @xmath19 .
ideally , one can obtain a perfect ( 100% ) conversion if @xmath20 and @xmath21 have the same unit amplitude and 180-degree phase difference @xcite .
it is well known that pec and pmc surfaces reflect em waves perfectly but with a reverse phase .
if a metasurface functions as a pec plane for @xmath6-polarized em waves , then we got @xmath22 . likewise ,
if the metasurface functions as a pmc plane for @xmath7-polarized em waves , then we arrive at @xmath23 . therefore , a mirror - symmetric and anisotropic pec - pmc scatterer will achieve 100% efficiency for the oam conversion .
inspired by this concept , we propose a novel metasurface as shown in fig . [
fig : unit_cell ] .
figure [ fig : unit_cell](b ) presents a scatterer of the metasurface comprising two dielectric layers , two artificial metal surfaces , and one ground plane .
periodic boundaries and floquet ports are imposed respectively at the lateral and longitudinal sides of the scatterer .
the top - right inset in fig .
[ fig : unit_cell](c ) shows the artificial anisotropic pec surface .
each metal strip with a width of @xmath24 is separated by a gap @xmath25 .
the metal strip array behaves like a parallel - plate waveguide .
plane waves polarized along the @xmath7 direction freely pass through the strip array , because there is no cutoff frequency for the excited tem mode .
while for @xmath6-polarized plane waves , te modes need to be considered , which have cutoff frequencies . here
we choose a sufficiently small gap so that the operating frequency is well below the cut - off frequency of the fundamental @xmath26 mode . by employing the artificial pec surface , the @xmath6-polarized plane wave is totally reflected with an offset phase of @xmath27 .
the bottom - right inset in fig .
[ fig : unit_cell](c ) is the artificial pmc surface realized by the mushroom - like high - impedance surface @xcite . a via inductor links the square patch to the ground plane .
the gap capacitance exists between adjacent patches .
when the mushroom structure is on resonance , the formed high - impedance surface can be regarded as a pmc plane . in view of a fact that the pec surface is on top of the pmc surface , the @xmath6 polarized wave
is perfectly reflected back by the pec surface ( @xmath22 ) , and the @xmath7 polarized wave passing through the strip array is then perfectly reflected back by the pmc surface with a zero phase shift ( @xmath23 ) .
all the simulations are conducted by the commercial software cst mws .
figure [ fig : unitcell_result ] shows the simulated reflection coefficients of the scatterer as depicted in fig .
[ fig : unit_cell](b ) . as expected ,
the amplitudes of the co - polarized reflection coefficients are 1 , while the amplitudes of the cross - polarized reflection coefficients are zero due to the symmetric scatterer .
in contrast to the strip array with a constant reflection phase of 180 degree , the mushroom structure behaves like an lc resonator with a varying reflection phase .
the expected 180-degree phase difference can be realized at @xmath28 ghz . .
( a ) magnitude ; ( b ) phase.,width=316 ] in order to generate oam of @xmath29 with a helical wavefront , the reflection phase shall be varied from 0 to 720 degrees , and the reflection amplitude is required to be a constant at different positions of the metasurface . the varying phase , as described in eq . , is fulfilled by azimuthally rotating the scatterers ( or manipulating the orientations of the scatterers ) . considering the mushroom structure functions as an isotropic pmc surface , no rotation implementation is needed . for the pec surface ,
if the length of the strip array along the @xmath6 direction is infinitely small , rotation of the tiny strip scatterers will lead to a series of concentric loops .
the spacing between adjacent loops is just the strip gap @xmath25 as illustrated in fig .
[ fig : unit_cell](c )
. interestingly , the composite pec metasurface ( concentric loops ) does not have any geometric discontinuities along the azimuthal direction .
the whole pec - pmc anisotropic metasurface [ see fig . [
fig : unit_cell](a ) ] is rotationally invariant about the concentric origin . due to the angular momentum conservation by rotational invariance ,
the metasurface allows spin - to - orbit coupling , where a change of the sam will modify the oam . in other words
, the metasurface perfectly converts a left ( right ) circularly polarized plane wave carrying @xmath30 sam to a right ( left ) circularly polarized vortex beam carrying @xmath31 sam and @xmath29 oam . for simplicity and a comparative study ,
we first simulate concentric loops on top of an ideal pmc plane .
the incident wave is a right circularly polarized plane wave at the frequency of @xmath32 ghz .
figure [ fig : pecpmc ] shows the amplitude and phase distributions of electric fields at a transverse plane of @xmath33 mm .
the incident and reflected fields are extracted by projecting the total electric field onto two orthogonal circular - polarized jones vectors .
a clear electromagnetic vortex ( phase singularity ) is observed at the origin of the reflected field as shown in fig .
[ fig : pecpmc](b ) .
the field amplitudes for both incident and reflected waves are around @xmath34 v / m , indicating that the reflection amplitude is @xmath34 ( where the dielectric losses of substrates are ignored ) . for the reflected field
, phase continuously increases from @xmath35 to @xmath36 around the vortex . integrating the phase of the reflected field around a path enclosing the vortex yields an integer multiple of @xmath37 .
this integer @xmath38 is known as the topological charge of the oam beam , which is generated by the pec - pmc metasurface with the topological charge of @xmath34 .
figure [ fig : pecmushroom ] shows the position - dependent amplitude and phase distributions of reflected electric fields away from the proposed metasurface ( see fig . [ fig : unit_cell ] ) .
the observation planes are placed at @xmath33 mm and @xmath39 mm above the metasurface .
after comparing fig . [
fig : pecmushroom ] to fig .
[ fig : pecpmc ] , the maximum amplitude , electromagnetic vortex and helical phase front obtained by the proposed metasurface are almost identical to those by the ideal metasurface .
this strongly confirms that the proposed design generates oam with a nearly perfect conversion efficiency .
moreover , for the @xmath33 mm case , the abnormal phase distribution near the vortex is caused by undamped evanescent modes excited by the mushroom structure .
when the distance between the observation plane and metasurface becomes large , the evanescent modes are significantly decayed . under this situation ,
the reflected field from the mushroom structure is almost same as that from the ideal pmc plane .
figure [ fig : pecmushroom_f ] illustrates frequency - dependent phase distribution along the azimuthal coordinate at a constant radius of @xmath40 mm .
the observation plane is placed at 10 mm above the proposed metasurface .
a desired linear dependence between the spatial phase and azimuthal angle is achieved at @xmath32 ghz .
the operating frequency is shifted from @xmath28 ghz for the scatterer as depicted in fig .
[ fig : unitcell_result ] to @xmath32 ghz for the whole pec - pmc metasurface .
different from the scatterer configuration , in the composite metasurface , the concentric loops in the circular cylindrical space are not align to the periodic mushroom structure in the cartesian space ( see fig . [ fig : unit_cell ] ) .
hence , original floquet modes in the scatterer with lateral periodic boundary conditions are broken , which is responsible for the frequency shifting .
moreover , the frequency - dependent phase - azimuthal angle diagram offers a powerful tool to examine the generated oam pattern .
mm above the proposed metasurface at fig .
[ fig : unit_cell ] .
( a , b ) amplitude ; ( c , d ) phase ; ( a , c ) @xmath33 mm ; ( b , d ) @xmath39 mm.,width=288 ] mm .
the observation plane is 10 mm above the proposed metasurface at fig .
[ fig : unit_cell].,width=288 ] for comparison , an azimuthally discontinuous metasurface is constructed by similar pec - pmc scatterers .
the dipole pec scatterers are depicted in fig .
[ fig : l1](a ) ; and the reflected field is recorded at a transverse plane of @xmath41 mm . to make a fair comparison
, the distribution of the dipole scatterers satisfies the same rotational symmetry as the concentric loops ( see fig .
[ fig : unit_cell ] ) . in this simulation
, we make use of an ideal pmc at the bottom to better clarify the influence of the discontinuous dipole scatterers . the operating frequency of @xmath32 ghz is identical to the concentric loops ( see fig . [
fig : pecpmc ] ) .
the field pattern shows the azimuthally discontinuous loops also obtain the vortex beam carrying the oam of @xmath42 .
however , due to the influence of the discontinuous scatterers , ripples are observed in the phase pattern .
the ripples are caused by the @xmath43 component of electric field generated from the edges of the discontinuous scatterers .
the distortions will become more serious when the observation plane moves closer to the discontinuous metasurface .
particularly , the @xmath44 component of near - field can be completely eliminated by using the continuous loops . for the metasurface with the continuous loops ,
no resonant scatterers exists and thus smooth phase distribution can be achieved at both near- and far - field regions . by using the discontinuous dipole scatterers , vortex beams with high - order oam can be generated .
figure [ fig : l2](a ) shows the schematic pattern .
obviously , the topological charge of the metasurface is @xmath38 and that of corresponding oam beam is @xmath45 . in fig .
[ fig : l2](b ) , the spatial phase experiences a linear increase from @xmath35 to @xmath46 along a full circular path . mm .
the operating frequency is 6.2 ghz .
the metasurface is constructed with azimuthally discontinuous dipole scatterers ( with a topological charge of @xmath47 ) placed above an ideal pmc plane .
( a ) amplitude ; ( b ) phase.,width=336 ] mm .
the operating frequency is 6.2 ghz .
the metasurface is constructed with azimuthally discontinuous dipole scatterers ( with a topological charge of @xmath48 ) placed above an ideal pmc plane .
( a ) amplitude ; ( b ) phase.,width=336 ]
in conclusion , we propose two types of pec - pmc anisotropic metasurfaces for generating vortex beams carrying oam at the microwave frequency . both types of metasurfaces achieve nearly @xmath49 conversion efficiency .
one metasurface is composed of azimuthally continuous loops , which achieves smoother phase distribution than the other metasurface with azimuthally discontinuous dipole scatterers .
the continuous metasurface converts a left ( right ) circularly polarized incident plane wave with zero oam to a right ( left ) circularly polarized reflected vortex beam with @xmath29 oam .
furthermore , the discontinuous metasurface generates the vortex beam with arbitrary - order oam .
our work provides a great convenience to high - efficiency oam generation . in future
, we could explore emerging digital metamaterials @xcite to generate vortex beam with arbitrary topological charges .
this work was supported by the research grants council of hong kong ( grf 712612 and 711511 ) , national natural science foundation of china ( nos . 61271158 and 61201122 ) , seed fund of university of hong kong ( nos .
201309160052 and 201311159108 ) , and university grants council of hong kong ( no .
aoe / p-04/08 ) .
the authors also thank the collaborator huawei technologies co. ltd . for helping the simulation . | orbital angular momentum ( oam ) is a promising degree of freedom for fundamental studies in electromagnetics and quantum mechanics .
the unlimited state space of oam shows a great potential to enhance channel capacities of classical and quantum communications . by exploring the pancharatnam - berry phase concept and engineering anisotropic scatterers in a metasurface with spatially varying orientations , a plane wave with zero oam
can be converted to a vortex beam carrying nonzero oam . in this paper , we proposed two types of novel pec ( perfect electric conductor)-pmc ( perfect magnetic conductor ) anisotropic metasurfaces .
one is composed of azimuthally continuous loops and the other is constructed by azimuthally discontinuous dipole scatterers .
both types of metasurfaces are mounted on a mushroom - type high impedance surface .
compared to previous metasurface designs for generating oam , the proposed ones achieve nearly perfect conversion efficiency . in view of the eliminated vertical component of electric field ,
the continuous metasurface shows very smooth phase pattern at the near - field region , which can not be achieved by convectional metasurfaces composed of discrete scatterers . on the other hand
, the metasurface with discrete dipole scatterers shows a great flexibility to generate oam with arbitrary topological charges .
our work is fundamentally and practically important to high - performance oam generation . |
the galaxy classification scheme of @xcite has proved durably useful . as modified and extended by @xcite , it is still the standard method for classifying low - redshift galaxies with high surface brightness .
the hubble classification scheme was originally based on the appearance of galaxies on photographic plates .
elliptical galaxies have smooth elliptical isophotes ; spiral galaxies have spiral arms that wind outward from a central bulge or bar .
it was later discovered that for luminous galaxies , the surface brightness profile is strongly correlated with the hubble type .
if the surface brightness @xmath16 is measured along the major axis of a galaxy s image , it is found that bright elliptical galaxies have surface brightness profiles that are well fit by a de vaucouleurs , or @xmath17 law , for which @xmath18 @xcite .
by contrast , the azimuthally averaged surface brightness profile of a spiral galaxy , outside its central bulge , is typically well fit by an exponential law , @xmath19 @xcite .
it was also eventually realized that galaxies of different hubble type have different kinematic properties .
the disks of spiral galaxies are rotationally flattened , with stars and gas on nearly circular orbits with little random motion .
bright elliptical galaxies ( @xmath20 ) , by contrast , are slowly rotating and are supported mainly by their anisotropic velocity dispersion .
one shortcoming of the hubble classification scheme , imposed by necessity , is that elliptical galaxies are classified by their apparent two - dimensional shape , seen in projection on the sky , rather than their intrinsic three - dimensional shape .
consider an idealized galaxy whose surfaces of constant luminosity density are concentric , coaxial , similar ellipsoids , with principal axes of lengths @xmath21 ; the shape of the galaxy can then be described by the two axis ratios @xmath22 and @xmath23 .
equivalently , the shape can be described by the two numbers @xmath8 and @xmath9 , where the triaxiality parameter @xmath9 is given by the relation @xmath24 .
if the ellipsoidal galaxy is seen in projection , though , its isophotes will be concentric , coaxial , similar ellipses .
the shape of the projected image can then be described by the single axis ratio @xmath25 , where @xmath26 and @xmath27 are the major and minor axis length of any isophote .
although knowing the apparent axis ratio @xmath28 is not , by itself , sufficient to determine the intrinsic axis ratios @xmath29 and @xmath8 , the three - dimensional shape of galaxies is not beyond all conjecture .
two approaches to determining the three - dimensional shape of galaxies have been used .
first , the intrinsic shape of an individual galaxy can be modeled if kinematic data are available in addition to photometric data @xcite .
however , accurate galaxy modeling requires either multiple long - slit position angles @xcite or integral - field spectroscopy @xcite , and is best applied to systems with high surface brightness and large angular size .
the second approach , which can be used in the absence of kinematic information , is to make statistical estimates of the distribution of @xmath29 and @xmath8 for a large sample of galaxies .
early estimates of the intrinsic shape distribution made the assumption that elliptical galaxies were oblate spheroids , with @xmath30 @xcite .
more recent studies , using accurate surface photometry , reveal that the distribution of apparent shapes for ellipticals can not be explained by a population of randomly oriented oblate spheroids ; it can easily be explained , however , by a population of triaxial ellipsoids @xcite . in this paper , we use the photometry - only approach to constraining the intrinsic shapes of galaxies , using the sloan digital sky survey data release 3 ( sdss dr3 ) as our source of galaxy photometry .
previous studies using the sdss early data release and data release 1 studied the axis ratios of galaxies best fit by de vaucouleurs profiles @xcite and those best fit by exponential profiles @xcite . in this paper
, we more closely examine the relation between surface brightness profile and intrinsic shape .
in addition , we determine , for each profile type , the dependence of intrinsic shape on galaxy luminosity . for elliptical galaxies , the distribution of apparent shapes is known to depend on absolute magnitude .
elliptical galaxies with @xmath20 are rounder on average than fainter ellipticals @xcite ; for a typical elliptical galaxy color of @xmath31 @xcite , the dividing luminosity corresponds to @xmath32 . in this paper ,
given the large sample size provided by the sdss dr3 , we can look for a similar shape dichotomy not only among galaxies with de vaucouleurs profiles , but among galaxies with a variety of surface brightness profile types . in [ sec - data ] of this paper
, we describe the sdss and the methods by which we determine the apparent axis ratio of the galaxies in our sample . in
[ sec - app ] , we examine how the apparent axis ratios depend on the galaxy profile type and the galaxy luminosity , then use nonparametric kernel estimators to determine the distribution of apparent axis ratios for different samples of galaxies , subdivided by luminosity and profile type . in
[ sec - int ] , we invert the distribution of apparent axis ratios to find the distribution of intrinsic axis ratios , assuming galaxies all have the same trixiality parameter @xmath9 . in addition to looking at purely oblate galaxies ( @xmath33 ) and purely prolate galaxies ( @xmath14 ) , we also examine the results for triaxial galaxies with @xmath34 , @xmath35 , @xmath36 , and @xmath37 . in [ sec - dis ] , we discuss the implications of the observed galaxy shapes for different scenarios of galaxy formation and evolution .
the sloan digital sky survey @xcite will , when complete , provide a map of nearly one - fourth of the celestial sphere . a ccd mosaic
camera @xcite images the sky in five photometric bands ( @xmath38 ; @xcite )
. the main galaxy sample ( mgs ) of the sdss @xcite will eventually contain @xmath39 galaxies with @xmath40 ; the mean redshift of galaxies in the mgs is @xmath41 , determined from a follow - up spectroscopic survey .
the sdss data release 3 , issued to the astronomical community in 2004 october , contains 5282 square degrees of imaging data and 4188 square degrees of spectroscopic data ( @xcite ; see also @xcite , @xcite , and @xcite ) .
the sdss dr3 data processing pipeline provides a morphological star / galaxy separation , with extended objects being classified as ` galaxies ' and point - like objects being classified as ` stars ' . for each galaxy , in each photometric band , a pair of models are fitted to the two - dimensional galaxy image .
the first model has a de vaucouleurs surface profile @xcite : @xmath42 \right ) \ , \ ] ] truncated beyond @xmath43 to go smoothly to zero at @xmath44 .
the second model has an exponential profile : @xmath45 \right ) \ , \ ] ] truncated beyond @xmath46 to go smoothly to zero at @xmath47 . for each model , the apparent axis ratio @xmath48 and the phase angle @xmath49
are assumed to be constant with radius .
the parameters @xmath48 , @xmath49 , @xmath50 , and @xmath51 are varied to give the best @xmath52 fit to the galaxy image , after convolution with a double - gaussian fit to the point spread function .
the sdss dr3 pipeline also takes the best - fit exponential model and the best - fit de vaucouleurs model and finds the linear combination of the two that best fits the galaxy image .
the fraction of the total flux contributed by the de vaucouleurs component is the parameter ` fracdev ` , which is constrained to lie in the interval @xmath53 .
the ` fracdev ` parameter delineates a continuum of surface brightness profile types , from the pure de vaucouleurs ( @xmath1 ) to the pure exponential ( @xmath2 ) .
the sdss dr3 databases provide many different measures of the apparent axis ratio @xmath28 of each galaxy in each of the five photometric bands . in this paper
, we will use the @xmath0 band data , at an effective wavelength of 6165 .
( we also repeated our analysis at @xmath54 and @xmath55 , the other two bands with high signal - to - noise , and did not find significantly different results from our @xmath0 band analysis . ) two measures of the apparent axis ratio are given by the best - fitting axis ratios @xmath48 for the de vaucouleurs and exponential models .
however , the algorithm for fitting the models introduces quantization in the distribution of @xmath48 .
because of this artificial quantization , we do not use the model axis ratios as estimates of the true apparent shapes of galaxies .
a useful measure of the apparent shape in the outer regions of galaxies is the axis ratio of the 25 mag arcsec@xmath56 isophote .
the sdss dr3 data pipeline finds the best fitting ellipse to the 25 mag arcsec@xmath56 isophote in each band ; the semimajor axis and semiminor axis of this isophotal ellipse are @xmath57 and @xmath58 .
the isophotal axis ratio @xmath59 then provides a measure of the apparent galaxy shape at a few times the effective radius . for galaxies in our sample with @xmath1 ,
the mean and standard deviation of @xmath60 are @xmath61 ; for galaxies with @xmath62 , the mean and standard deviation are @xmath63 .
another measure of the apparent shape is @xmath3 , the axis ratio determined by the use of adaptive moments of the galaxy s light .
the method of adaptive moments determines the @xmath64th order moments of a galaxy image , using an elliptical weight function whose shape matches that of the image @xcite . the sdss dr3 adaptive moments use a weight function @xmath65 that is a gaussian matched to the size and ellipticity of the galaxy image @xmath66 . the adaptive first order moments , @xmath67 and @xmath68
tell us the ` center of light ' of the galaxy s image . with this knowledge , we can compute the adaptive second order moments : @xmath69 and so forth .
the sdss dr3 provides for each image the values of the parameters @xmath70 , @xmath71 , and @xmath72 .
the adaptive second moments can be converted into an axis ratio using the relation @xmath73 where @xmath74 .
the adaptive moments axis ratio @xmath3 , computed in the manner given above , is not corrected for the effects of seeing .
the sdss dr3 also provides the fourth order adaptive moments of the galaxy image , and the adaptive moments @xmath75 , @xmath76 , and @xmath77 of the point spread function at the galaxy s location .
these moments can be used to correct for the smearing and shearing due to seeing ; such corrections are essential for studying the small shape changes resulting from weak lensing @xcite . however , in this paper we will only look at well - resolved galaxies for which the seeing corrections are negligibly small .
our full sample of galaxies consists of those objects in the sdss dr3 spectroscopic sample which are flagged as galaxies and which have spectroscopic redshifts @xmath78 ( to eliminate contaminating foreground objects ) and @xmath79 ( to reduce the possibility of weak lensing significantly distorting the observed shape ) . to ensure that galaxies were well resolved
, we also required @xmath80 .
the absolute magnitude @xmath81 of each galaxy was computed from the petrosian @xmath0 magnitude and the spectroscopic redshift , assuming a hubble constant @xmath82 in a flat universe with @xmath83 and @xmath84 .
no k - correction was applied to the data ; for galaxies with normal colors at low redshift the k - correction in the @xmath0 band is small less than 0.2 mag for an average elliptical galaxy at @xmath85 , and even less for galaxies at smaller redshifts or with bluer colors @xcite .
the total number of galaxies in our full sample , selected in this way , is @xmath86 .
of the full sample , only 919 galaxies have @xmath1 , and 26,994 have @xmath2 .
the remainder , constituting 71% of the full sample , are best fit by a combination of a de vaucouleurs and exponential model , with @xmath87 .
it remains to be demonstrated that ` fracdev ` is a useful parameter for describing the surface brightness profile type of galaxies .
after all , a linear combination of the best de vaucouleurs fit and the best exponential fit will not , in general , be the best possible de vaucouleurs plus exponential fit .
an alternative method of describing surface brightness profiles is by fitting a srsic profile @xcite : @xmath88 where @xmath64 can be an arbitrary number .
this fitting method was used , for instance , by @xcite in their study of sdss galaxies .
if , in fact , galaxies are well described by srsic profiles , then ` fracdev ` is a useful surrogate for the srsic index @xmath64 .
consider , for instance , a galaxy whose surface brightness is perfectly described by a srsic profile of index @xmath89 and effective radius @xmath90 .
the best - fitting exponential model for this galaxy ( fitting in the radial region @xmath91 ) has @xmath92 ; the best - fitting de vaucouleurs model has @xmath93 .
combining the models gives @xmath94 for this @xmath89 srsic galaxy .
a similar fit to an @xmath95 srsic galaxy yields @xmath96 . in general ,
if galaxies have srsic profiles with @xmath97 , then ` fracdev ` , as computed by the sdss dr3 pipeline , is a monotonically increasing function of the srsic index @xmath64 , and thus can be used as a surrogate for @xmath64 . a plot of the mean axis ratio @xmath98 as a function of the parameter `
fracdev ` is shown in figure [ fig : fdev ] .
the mean adaptive moments axis ratio @xmath99 , indicated by the filled circles , shows an increasing trend with ` fracdev ` , from @xmath100 for galaxies with purely exponential profiles ( @xmath2 ) to @xmath101 for galaxies with pure de vaucouleurs profiles ( @xmath1 ) .
however , the trend in @xmath98 is not a linear one ; for galaxies with @xmath102 , the value of @xmath98 is nearly constant at @xmath103 ; it is only at @xmath104 that @xmath98 shows an increasing trend with ` fracdev ` .
the mean isophotal axis ratio @xmath105 , indicated by the open circles in figure [ fig : fdev ] , shows less of a trend with ` fracdev ` . except in the case of nearly pure de vaucouleurs profiles ( @xmath106 )
, the axis ratio of a galaxy in its outer regions does nt seem to depend on its surface brightness profile . for convenience in analysis ,
we have divided our galaxy sample into four classes , based on the value of ` fracdev ` .
galaxies with @xmath107 , corresponding to a srsic index @xmath108 , are called ` ex ' galaxies ; there are @xmath109 ` ex ' galaxies in our sample .
galaxies with @xmath110 , corresponding to @xmath111 , are labeled ` ex / de ' galaxies ( @xmath112 ) .
galaxies with @xmath113 , corresponding to @xmath114 , are labeled ` de / ex ' galaxies ( @xmath115 ) .
finally , galaxies with @xmath116 , corresponding to @xmath117 , are called ` de ' galaxies .
the small fraction of ` de ' galaxies in our sample ( @xmath118 ) is partly due to the fact that the centrally concentrated de vaucouleurs galaxies are less likely to satisfy our resolution criterion , and partly due to the fact that galaxies with high srsic indices are intrinsically rare ; @xcite estimated that only @xmath119 of the sdss galaxies in their sample had srsic index @xmath120 .
the ` de ' galaxies are rounder in their central regions than in their outer regions : @xmath121 .
this is consistent with the ` de ' galaxies being relatively bright elliptical galaxies , for which the isophotal axis ratios tend to decrease with increasing semimajor axis length @xcite .
by contrast , the ` ex ' galaxies are actually slightly flatter in their central regions than in their outer regions : @xmath122 . for a given surface brightness profile type , there exists a relation between absolute magnitude and apparent axis ratio .
contour plots of mean apparent axis ratio as a function of both ` fracdev ` and @xmath81 are given in figure [ fig : fdevmag ] .
( to give a feel for the absolute magnitude scale , fitting a schechter function to the luminosity function of sdss galaxies yields @xmath123 @xcite . )
the upper panel of the figure shows that the trend in @xmath99 runs from the flattest galaxies at @xmath124 and @xmath125 , where @xmath126 , to the roundest galaxies at @xmath127 and @xmath128 , where @xmath129 .
this result is not surprising , since moderately bright galaxies with exponential profiles are intrinsically flattened disk galaxies , while extremely bright galaxies with de vaucouleurs profiles are intrinsically nearly spherical giant elliptical galaxies @xcite . more surprising are the results shown in the bottom panel of figure [ fig : fdevmag ] , which shows @xmath105 , the mean isophotal axis ratio . here ,
we find that the apparently flattest galaxies , as measured by @xmath6 , are not exponential galaxies , but galaxies with @xmath130 ( corresponding to srsic index @xmath131 ) and @xmath132 ; these galaxies have @xmath133 .
the roundest galaxies , measured by @xmath105 , are not bright de vaucouleurs galaxies , but bright exponential galaxies ; the maximum value of @xmath105 is @xmath134 , at @xmath124 , @xmath135 .
note also that for bright galaxies ( @xmath136 ) , the contours of constant @xmath105 in figure [ fig : fdevmag ] are nearly horizontal .
that is , among bright galaxies , the flattening of the outer isophotes does nt depend strongly on the surface brightness profile type .
a view of the dependence of @xmath98 on absolute magnitude for each of our four profile types , ` ex ' , ` ex / de ' , de / ex , and ` de ' , is given in figure [ fig : mag ] .
in the upper panel , which shows @xmath99 , note that for each profile type , there is a critical absolute magnitude @xmath137 at which @xmath99 is at a minimum .
this critical absolute magnitude ranges from @xmath138 for the ` de ' galaxies to @xmath139 for the ` ex ' galaxies . at @xmath140 ,
the value of @xmath99 increases relatively rapidly with increasing luminosity ; at @xmath141 , the value of @xmath99 increases less rapidly with decreasing luminosity . at a fixed absolute magnitude ,
the average axis ratio of ` de ' galaxies is always greater than that of ` ex ' galaxies ; however , for @xmath142 , the flattest galaxies , on average , at a given absolute magnitude are not the ` ex ' galaxies , but those with the mixed ` de / ex ' and ` ex / de ' profile types .
the bottom panel of figure [ fig : mag ] shows @xmath105 versus @xmath81 for the different profile types . in the interval @xmath143
, we see that ` ex ' galaxies have _ larger _ values of @xmath144 than galaxies of other profile types .
@xcite divided elliptical galaxies into two classes ; galaxies brighter than @xmath145 are rounder on average than fainter ellipticals . using a typical color for elliptical galaxies of @xmath31 ,
this corresponds to an absolute magnitude in the @xmath0 band of @xmath146 , using the transformation @xmath147 @xcite . in our data
, we do not see an abrupt jump in @xmath148 for ` de ' galaxies at @xmath149 , but rather a more gradual increase between @xmath150 and @xmath151 .
in addition , it is true for any profile type that highly luminous galaxies are rounder , on average , than less luminous galaxies . knowing the mean shape as a function of @xmath81 for each profile type ( ` de ' , ` de / ex ' , ` ex / de ' , ex ) does not tell us the complete distribution of axis ratios , @xmath152 . in order to compare the distribution of axis ratios for highly luminous galaxies with that for less luminous galaxies
, we determined , for each profile type , the dividing absolute magnitude @xmath153 such that the distribution @xmath154 for galaxies with @xmath155 is maximally different from the distribution @xmath154 for galaxies with @xmath156 .
we measure the difference between the distributions by applying a kolmogorov - smirnov ( k - s ) test , and finding the value of @xmath153 that minimizes the k - s probability @xmath157 .
the values of the dividing magnitude @xmath153 for each profile type , and the associated probability @xmath157 , are given in table [ tab : bridim ] .
( using a student s t - test for the significance of the difference of the means yields similar values of @xmath153 , differing by less than 0.2 mag from the values found using the k - s test . )
to estimate the distribution of apparent axis ratios @xmath152 for each of our galaxy subsamples , we use a nonparametric kernel technique @xcite . given a sample of @xmath158 axis ratios , @xmath159 , @xmath160 , @xmath161 , @xmath162 , the kernel estimate of the frequency distribution @xmath152 is @xmath163 where @xmath164 is the kernel function , normalized so that @xmath165 to ensure that our estimate @xmath166 is smooth , we use a gaussian kernel , with @xmath167 .
the parameter @xmath168 in equation ( [ eq : kernel ] ) is the kernel width ; too small a width introduces noise into the estimate , while too great a kernel width produces excessive smoothing .
we choose the kernel width @xmath168 by using the standard adaptive two - stage estimator of @xcite . in this technique ,
an initial estimate @xmath169 is made using a global fixed kernel width @xmath170 with @xmath171 , where @xmath172 is the standard deviation and @xmath173 is the interquartile range of the axis ratios . for samples that are not strongly skewed ,
this formula minimizes the mean square error @xcite .
the final estimate @xmath166 is then computed , using at each data point @xmath174 the kernel width @xmath175^{1/2 } \ , \ ] ] where @xmath176 is the geometric mean of @xmath169 over all values of @xmath55 . the axis ratio for a galaxy must lie in the range @xmath177 . to ensure that @xmath178 for @xmath179 and @xmath180
, we impose reflective boundary conditions at @xmath181 and @xmath182 . in practice , this means replacing the simple gaussian kernel with the kernel @xmath183 although use of reflective boundary conditions ensures that @xmath166 remains within bounds , it imposes the possibly unphysical constraint that @xmath184 at @xmath181 and @xmath182 .
the estimated distributions @xmath185 are shown in the left column of figure [ fig : apparent ] , for the ` de ' , ` de / ex ' , ` ex / de ' , and ` ex ' galaxies ; the equivalent distributions @xmath186 for the isophotal shapes are shown in the right column . in each panel of figure [ fig : apparent ] , the heavier line indicates the shape distribution for the bright galaxies , and the lighter line indicates the distribution for the fainter galaxies .
the dotted lines indicate the 98% error intervals found by bootstrap resampling . in our bootstrap analysis , we randomly selected @xmath158 data points , with substitution , from the original sample of @xmath158 axis ratios .
we then created a new estimate @xmath166 from the bootstrapped data . after doing 1000 bootstrap estimates
, we then determined the 98% error intervals : in figure [ fig : apparent ] , 1% of the bootstrap estimates lie above the upper dotted line , and 1% of the bootstrap estimates lie below the lower dotted line .
note from the upper left panel of figure [ fig : apparent ] that the bright ` de ' galaxies have a distribution that peaks at @xmath187 , similar to the result found for relatively bright elliptical galaxies @xcite .
the faint ` de ' galaxies have a flatter modal shape ( @xmath188 ) and a broader distribution of shapes .
both the bright and faint ` de ' galaxies have a relative scarcity of galaxies with @xmath189 ; this is the usual sign that the galaxies in the population can not be purely oblate . for
` de ' galaxies , the difference between the adaptive moments shapes and the isophotal shapes is intriguing .
note that @xmath186 for the ` de ' galaxies , shown in the upper right panel of figure [ fig : apparent ] , peaks at @xmath190 for the bright galaxies .
for the fainter ` de ' galaxies , the distribution of isophotal axis ratios , though it peaks at @xmath191 , is very broad .
an axis ratio distribution @xmath166 that is nearly constant over a broad range of @xmath28 is the signature of a population of flattened disks .
the distribution @xmath186 drops at @xmath192 , indicating that the disks are not infinitesimally thin , and at @xmath193 , indicating that the disks are not perfectly circular . the distribution of shapes for ` de / ex ' galaxies are seen in the second row of figure [ fig : apparent ] .
as with all profile types , the bright galaxies are rounder on average than the fainter galaxies .
both bright and faint ` de / ex ' galaxies have a scarcity of nearly circular galaxies . particularly noteworthy is the distinct difference between @xmath185 for the bright ` de ' galaxies , very strongly peaked at @xmath187 , and the equivalent distribution @xmath185 for the bright ` de / ex ' galaxies , which is much broader and peaks at the flatter shape of @xmath194 .
the ` ex / de ' and ` ex ' galaxies in our sample have , by definition , @xmath195 and hence srsic index @xmath196 .
we expect these galaxies to be predominantly disk - dominated spiral galaxies .
determining the true distribution of intrinsic axis ratios for spiral galaxies is complicated by the presence of dust .
a magnitude - limited sample of dusty spiral galaxies will show a deficit of galaxies with high inclination and low apparent axis ratio .
this deficit is due to internal extinction by dust ; in the @xmath27 band , for instance , a spiral galaxy of type sc will be up to 1.5 mag fainter when seen edge - on than when seen face - on @xcite . in the @xmath0 band ,
the inclination - dependent dimming is not as great , but will still affect the observed distribution of axis ratios in a magnitude - limited sample such as the sdss spectroscopic galaxy sample @xcite . a dusty disk galaxy which would qualify as a bright ` ex ' galaxy if seen face - on might only qualify as a faint ` ex ' galaxy when seen edge - on .
similarly , a less luminous galaxy , which would qualify only as a faint ` ex ' galaxy when seen face - on , might drop out of our sample entirely when seen edge - on .
the main effect of dust on the distribution of @xmath28 for a magnitude - limited survey of spiral galaxies is to reduce the number of galaxies with @xmath197 ; the distribution for @xmath198 is not strongly affected @xcite . with the caveat that we are undercounting edge - on , low-@xmath28 galaxies , the observed axis ratios for ` ex / de ' and ` ex ' galaxies
are shown in the bottom two rows of figure [ fig : apparent ] .
the shapes of ` ex / de ' galaxies , as measured by either @xmath3 or @xmath6 , spread over a wide range of axis ratios , with bright ` ex / de ' galaxies being slightly rounder , on average , than the fainter galaxies .
the shapes of ` ex ' galaxies are very similar to the shapes of ` ex / de ' galaxies . for the ` ex / de ' and ` ex ' galaxies ,
the dip in @xmath166 at @xmath193 is a consequence of the fact that the disks of these galaxies are not perfectly circular .
having a smooth kernel estimate @xmath166 for the distribution of apparent axis ratios @xmath28 permits us to estimate the distribution of intrinsic axis ratios @xmath8 , given the assumption that all galaxies have the same value of the triaxiality parameter @xmath9 .
we do nt necessarily expect all the galaxies of a given profile type to have the same triaxiality ; cosmological simulations , as well as simulations of merging galaxies , generally give a range of @xmath9 ( see @xcite and references therein ) for bright galaxies .
however , by assuming uniform triaxiality and seeing which values of @xmath9 yield physically plausible results , we can get a feel for which values of @xmath9 are data - friendly , and which are incompatible with the data .
let @xmath199 be the estimated distribution of intrinsic axis ratios , given an assumed value for @xmath9 .
the relation between @xmath200 and @xmath199 is @xcite : @xmath201 where @xmath202 is the probability that a galaxy with intrinsic short - to - long axis ratio @xmath8 and triaxiality @xmath9 has an apparent axis ratio in the range @xmath203 , when viewed from a random orientation .
equation ( [ eq : volterra ] ) represents a volterra equation of the first kind which can be solved by standard numerical techniques .
we first compute @xmath166 on 200 points equally spaced in @xmath28 , using the kernel technique previously described .
we then compute @xmath204 on an @xmath205 grid in @xmath206 space .
for each value of @xmath8 , we compute an adaptive kernel estimate for @xmath204 by randomly choosing a large number of viewing angles ( @xmath207 ) , then computing what the apparent axis ratio @xmath28 would be for each viewing angle , given the assumed values of @xmath8 and @xmath9 @xcite .
the kernel estimate of the distribution of @xmath28 is then our estimate of @xmath208 .
equation ( [ eq : volterra ] ) , in its discretized form , can be written @xmath209 since @xmath210 for @xmath211 ( the apparent axis ratio of a galaxy ca nt be less than its intrinsic axis ratio ) , the matrix @xmath212 is lower triangular , and equation ( [ eq : matrix ] ) can be inverted simply by forward substitution @xcite .
we performed the inversion to find @xmath213 using six different values of the trixiality : @xmath33 ( oblate ) , @xmath34 , @xmath35 , @xmath36 , @xmath37 , and @xmath14 ( prolate ) .
when our ultimate goal is finding @xmath213 , determining the optimal kernel width @xmath214 for @xmath166 is a vexed question .
it is generally true that the best value of @xmath214 for estimating @xmath213 is greater than the best value of @xmath214 for purpose of estimating @xmath166 @xcite ; @xmath215 is a deconvolution of @xmath216 , and the deconvolution process increases the noise present .
@xcite proposed , for the purposes of estimating @xmath215 , using the value of @xmath214 that minimized @xmath217 ^ 2 d\gamma \ , \ ] ] where @xmath218 is an estimate of @xmath215 from a bootstrap resampling of the original data .
unfortunately , we found , as @xcite did , that @xmath219 returns badly oversmoothed estimates of @xmath215 .
thus , we fell back on using our subjective impressions of the smoothness of @xmath213 , and ended by taking the value of @xmath214 given by equation ( [ eq : hwidth ] ) and multiplying by a factor of 1.5 . to perform the inversion and find the estimated distribution of @xmath8
, we assumed that all galaxies have the same triaxiality @xmath9 .
if this assumption is incorrect , then the inversion of equation ( [ eq : matrix ] ) may result in @xmath220 for some range of @xmath8 , which is physically impossible . to exclude our hypothesis ( that all galaxies have triaxiality @xmath9 ) at some fixed statistical confidence level
, we can repeat the inversion for a large number of bootstrap resamplings of the original data set . in practice , we did 1000 resamplings , and used them to place 98% confidence intervals on @xmath213 .
if the confidence interval falls below zero for some valus of @xmath8 , then the hypothesized value of @xmath9 can be rejected at the 99% ( one - sided ) confidence level .
figure [ fig : de_adapt ] shows the distribution of intrinsic shapes for ` de ' galaxies ( @xmath11 ) , when the adaptive moments axis ratio @xmath3 is used as the estimator of the apparent shape .
the heavy line is the distribution for the bright ` de ' galaxies ( @xmath221 ) and the light line is the distribution for the fainter ` de ' galaxies .
the assumed value of @xmath9 is shown in each panel , ranging from purely oblate shapes , with @xmath222 , in the upper left corner to purely prolate shapes , with @xmath14 , in the lower right corner .
an eyecatching feature of figure [ fig : de_adapt ] is the oscillatory nature of @xmath213 in the triaxial cases .
the multiple peaks in @xmath213 at high @xmath8 result from the shape of the conditional probability function @xmath223 for highly triaxial galaxies . when galaxies are axisymmetric ( @xmath33 or @xmath224 ) , the conditional probability peaks at @xmath225 , the minimum possible apparent axis ratio . for triaxial galaxies , however , the conditional probability has one or more local maxima at @xmath226 . to see why this can result in oscillatory solutions consider @xmath213 for faint ` de ' galaxies when @xmath37 ;
this is shown by the light line in the lower left panel of figure [ fig : de_adapt ] .
the distribution @xmath213 has a local maximum at @xmath227 . for @xmath37 , the conditional probability @xmath228 peaks at @xmath229 .
thus , the large number of @xmath230 galaxies result in many galaxies with an apparent axis ratio @xmath231 ; so many , in fact , that @xmath213 must be made very small in the range @xmath232 in order to avoid overproducing @xmath231 galaxies . in fact , as seen in figure [ fig : de_adapt ] , @xmath213 has a local minimum at @xmath233 , at which @xmath213 falls slightly below zero . since @xmath234 peaks at @xmath235 , this produces a deficit of @xmath236 galaxies , which must be compensated for by making @xmath213 very large in the range @xmath237 .
in fact , @xmath213 has another local maximum at @xmath238 . and
so the oscillations continue , with decreasing wavelength , until @xmath239 is reached .
@xcite found that @xmath240 for bright elliptical galaxies was significantly bimodal if the galaxies were assumed to be highly triaxial ; this bimodality also had its origin in the shape of the conditional probability function for triaxial galaxies . for both bright and faint ` de ' galaxies , using @xmath3 as the apparent axis ratio , oblate fits are statistically unacceptable , producing a negative number of galaxies with @xmath241 ( see the upper left panel of figure [ fig : de_adapt ] ) . for bright ` de ' galaxies , statistically acceptable fits are found for @xmath35 , @xmath242 , and @xmath243 ; that is , their 98% confidence intervals never fall entirely below zero .
the mean intrinsic axis ratio for the acceptable fits ranges from @xmath244 for @xmath35 to @xmath245 for @xmath37 .
the permissible fits for the bright ` de ' galaxies are consistent with the deduced triaxial shape of the nearby bright ( @xmath246 ) elliptical galaxy ngc 4365 , for which a combination of photometric and kinematic data yields @xmath247 and @xmath248 @xcite . although the bright ` de ' galaxies are best fit by highly triaxial shapes , the faint ` de ' galaxies are best fit by nearly prolate shapes , with @xmath37 and @xmath14 both giving statistically acceptable fits .
( the highly oscillatory solution for @xmath37 may be physically dubious
why should galaxy shapes be quantized ? but it is statistically acceptable . ) the mean intrinsic shape of faint ` de ' galaxies is @xmath249 if @xmath37 and @xmath250 if @xmath14 .
figure [ fig : de_iso ] shows the deduced distribution of intrinsic shapes for ` de ' galaxies when the isophotal axis ratio @xmath6 is used , rather than the adaptive moments axis ratio . the bright ` de ' galaxies have a statistically acceptable fit when @xmath35 , @xmath36 , @xmath37 , and @xmath14 .
the mean intrinsic shape ranges from @xmath251 when @xmath35 to @xmath252 when @xmath14 . the fainter ` de ' galaxies are acceptably fit , assuming constant triaxiality , only when @xmath14 , which results in @xmath253 .
the @xmath33 case , which can be rejected at the 99% confidence level but not at the 99.9% level , would produce @xmath254 ; this axis ratio is flatter than that of an ice hockey puck .
( although the @xmath33 and @xmath34 fits are statistically unacceptable at the 99% confidence level , the data are consistent with a population of nearly oblate shapes if we relax the assumption of uniform @xmath9 . for instance , the faint ` de ' galaxies can be fit at the 99% confidence level , with a gaussian distribution of @xmath9 peaking at @xmath33 and with @xmath255 . )
the shapes of ` de / ex ' galaxies can be analyzed in the same way as the shapes of ` de ' galaxies .
for instance , the distributions of intrinsic shapes for ` de / ex ' galaxies , using @xmath3 as the apparent axis ratio , are shown in figure [ fig : deex_adapt ] .
for the brighter galaxies with ` de / ex ' profiles , acceptable fits are found when @xmath37 , yielding @xmath256 , and when @xmath14 , yielding @xmath257 .
for the fainter ` de / ex ' galaxies , only the @xmath14 fit is statistically acceptable , yielding a mean intrinsic axis ratio of @xmath258 for the prolate galaxies .
notice that bright ` de / ex ' galaxies , faint ` de / ex ' galaxies , and faint ` de ' galaxies all are consistent with having shapes , at least in their inner regions , which are prolate , or nearly prolate , with a typical axis ratio of @xmath259 .
bright ` de ' galaxies are distinctly different , with shapes that are consistent with their being highly triaxial , with a typical axis ratio of @xmath260 . the intrinsic shape distributions for ` de / ex ' galaxies , using @xmath6 as the apparent shape measure , are shown in figure [ fig : deex_iso ] .
for both the bright and faint ` de / ex ' galaxies , the @xmath14 fit is statistically acceptable .
for the prolate fits , the average shape of the bright ` de / ex ' galaxies is @xmath261 , and the average shape of the faint ` de / ex ' galaxies is @xmath262 .
thus , given the prolate hypothesis , ` de / ex ' galaxies are only slightly flatter in their outer regions , whose shape is measured by @xmath6 , than in their inner regions , whose shape is measured by @xmath3 . with the repeated caveat that we are undercounting edge - on , low-@xmath28 galaxies ,
the results for the ` ex / de ' galaxies are presented in figures [ fig : exde_adapt ] and [ fig : exde_iso ] .
when @xmath3 is used as the measure of apparent shape ( figure [ fig : exde_adapt ] ) , the bright ` ex / de ' galaxies have a statistically acceptable fit when @xmath37 , which results in @xmath263 ; the @xmath14 fit , which goes slightly negative at @xmath264 , yields a mean intrinsic axis ratio of @xmath261 .
none of the six tested triaxialities gave an acceptable fit , at the 99% confidence level , to the faint ` ex / de ' galaxies .
the best ( or the ` least bad ' ) of the fits , with @xmath14 , yields @xmath263 .
when @xmath6 is used as the measure of apparent shape ( figure [ fig : exde_iso ] ) , none of the tested values of @xmath9 gives a fit acceptable at the 99% confidence level .
the best fit , for both bright and faint galaxies , is given by @xmath14 , for which @xmath261 for the bright ` ex / de ' galaxies and @xmath265 for the faint ` ex / de ' galaxies .
as with the ` ex / de ' galaxies , the sample of ` ex ' galaxies is affected by an undercount of low-@xmath28 galaxies . with this caveat ,
the results for the ` ex ' galaxies are shown in figures [ fig : ex_adapt ] and [ fig : ex_iso ] .
in general , the results are similar to those for ` ex / de ' galaxies . when the axis ratios of ` ex ' galaxies are estimated using @xmath3 ( figure [ fig : ex_adapt ] ) , the bright ` ex ' galaxies are acceptably fit when @xmath37 , yielding @xmath263 ; the @xmath14 fit for the bright ` ex ' galaxies , which goes slightly negative at high @xmath8 , yields @xmath266 . the best fit to the faint `
ex ' galaxies is provided by the @xmath14 inversion , which yields @xmath267 .
when @xmath6 is used as the estimate of the apparent axis ratio ( figure [ fig : ex_iso ] ) , the results are extremely similar . for bright
` ex ' galaxies , only @xmath37 , of the tested triaxialities , gave a fit acceptable at the 99% confidence level , yielding @xmath268 ; the not - as - good @xmath14 fit gave @xmath269 .
the best fit for faint ` ex ' galaxies had @xmath14 , @xmath263 .
although the best fits to the ` ex / de ' and ` ex ' galaxies , under the assumption of constant triaxiality , were prolate or nearly prolate , we have an a priori knowledge that bright galaxies with srsic index @xmath196 are generally rotationally supported disks .
the sdss dr3 photometry of ` ex / de ' and ` ex ' galaxies is consistent with their being nearly oblate objects only if they have a range of @xmath9 , instead of being forced into a straitjacket of uniform @xmath9 .
constraining the flattening @xmath8 and triaxiality @xmath9 of galaxies with different luminosity and profile type provides potentially useful clues for the study of galaxy formation and evolution . for example , the dissipationless merger of two equal - mass disk galaxies embedded within dark halos produces a merger remnant whose luminous component has an approximate de vaucouleurs profile @xcite .
the values of @xmath8 and @xmath9 of the remnant depend on the initial orbital parameters of the merging galaxies , but the remnant is typically quite flat and prolate . for a set of eight simulations with different initial orbits , @xcite found @xmath270 and @xmath271 for the most tightly bound quartile of the luminous particles ( the inner region of the galaxy ) .
this is significantly flatter and more prolate than the shape we deduced from the distribution of @xmath3 for bright ` de ' galaxies .
the oblateness of the merger remnant is increased , however , if the initial merging galaxies are unequal in mass @xcite . including gaseous dissipation into a simulated merger
can strongly affect @xmath8 and @xmath9 . as the gas dissipates and falls toward the center of the merger remnant , the central mass concentration will scatter stars on low angular momentum box orbits , thus acting to increase @xmath8 and decrease @xmath9 . for instance ,
a simulated remnant that has @xmath272 and @xmath273 in the absence of dissipation will have @xmath274 and @xmath275 if gas dissipation is added @xcite .
the deduced shapes of bright ` de ' galaxies are consistent with their being merger remnants , as long as the mergers were dissipational . in the standard hierarchical clustering model for structure formation , however , we do not expect most elliptical galaxies to have formed by the relatively recent merger of a pair of spiral galaxies .
more typically , an elliptical galaxy will have formed by the successive merger of a number of smaller galaxies or subgalactic `` clumps '' .
simulations of multiple mergers in small groups reveal that the final merger remnant generally has a surface brightness profile similar to a de vaucouleurs law , but that the shape of the merger remnant depends on the assumed initial conditions @xcite . for instance , @xcite simulated dissipationless mergers of dense clumps embedded in a smooth dark halo .
the profile of the merger remnant , in all cases , was well fit by a de vaucouleurs law .
the triaxiality of the remnant depended primarily on whether the system was initially in a state of virial equilibrium or in a state of non - virial collapse .
the virial case produced more oblate remnants ( @xmath276 ) , while the collapse case produced more prolate remnants ( @xmath277 ) .
the virial case also produced less flattened remnants : @xmath278 for the merger remnants of initially virialized systems versus @xmath279 for initially collapsing systems .
the value of @xmath8 also depended on the initial clump - to - halo mass ratio ; if the clumps contributed only a small fraction of the total mass , they produced a flatter remnant than if they contributed all the mass , with no dark halo .
the best fitting parameters for the bright ` de ' galaxies , @xmath280 and @xmath281 , suggest that they are too flattened and too triaxial to have been formed primarily by mergers within a virialized group , which produce @xmath282 and @xmath276 .
( remember , the effects of dissipation will only act to increase @xmath8 and decrease @xmath9 . )
another influence that is capable of increasing @xmath8 and decreasing @xmath9 is the presence of a central supermassive black hole .
a central mass concentration affects the structure of the surrounding galaxy by disrupting box orbits @xcite .
the resulting chaotic orbits lead to an altered shape for the galaxy ; the trend is toward shapes that are nearly spherical in the inner regions and nearly oblate in the outer regions @xcite . in the inner regions ,
the orbits can support a shape that is nearly oblate , @xmath283 , or strongly triaxial , @xmath284 , but not nearly prolate , @xmath285 @xcite .
the time scale for the shape evolution depends on the mass of the central black hole relative to the total stellar mass of the galaxy .
when the central black hole has a mass equal to 1% of the mass in stars , its shape at the half - mass radius evolves in @xmath286 times the half - mass orbital period ; when the mass is only 0.3% of the stellar mass , the evolution time scale increases to @xmath287 times the orbital period .
the observed relation between central black hole mass and bulge mass @xcite predicts that an elliptical galaxy will typically have a central black hole with a mass @xmath288 of the total mass in stars . for such relatively low black hole masses ,
the shape evolution time scale at the half - mass radius and beyond will exceed the age of the galaxy .
thus , a central black hole will not generally affect the isophotal axis ratio @xmath6 . the isophotal shape , @xmath6 , measures the apparent shape of a galaxy at a few times the effective radius . converted to physical units , the average semimajor axis @xmath57 of the isophotal ellipse ranges from @xmath289 at @xmath290 to @xmath291 at @xmath292 .
at relatively high luminosities ( @xmath293 ) , the mean value of @xmath6 increases with luminosity ; bright galaxies are rounder than fainter galaxies at their outer regions .
we have also uncovered the puzzling result that at a given luminosity , the mean value of @xmath6 is larger for ` ex ' galaxies than for ` de ' galaxies .
why should ` ex ' galaxies , which are flattened disks in their central regions , be rounder than ` de ' galaxies in their outskirts ? at such great distances from the galaxy s center , we expect that tidal distortions from neighboring galaxies will significantly affect the isophotal shape .
galaxies in dense regions , such as the cores of clusters , are subject to more interactions with neighboring galaxies , and may well differ systematically in shape from field galaxies , which are less frequently harassed . since
elliptical galaxies are found preferentially in dense environments , this conjectured environment shape relation would then translate into a relation between profile type and shape . a future paper , studying the relation between apparent axis ratio and environment for
sdss dr3 galaxies ( kuehn & ryden , 2005 , in preparation ) , will test this conjecture .
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 website 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 , university of pittsburgh , princeton university , the united states naval observatory , and the university of washington . | for a sample of 96,951 galaxies from the sloan digital sky survey data release 3 , we study the distribution of apparent axis ratios as a function of @xmath0-band absolute magnitude and surface brightness profile type .
we use the parameter ` fracdev ` to quantify the profile type ( @xmath1 for a pure de vaucouleurs profile ; @xmath2 for a pure exponential profile ) . when the apparent axis ratio @xmath3 is estimated from the moments of the light distribution , the roundest galaxies are very bright ( @xmath4 ) de vaucouleurs galaxies and the flattest are modestly bright ( @xmath5 ) exponential galaxies .
when the axis ratio @xmath6 is estimated from the axis ratio of the 25 mag / arcsec@xmath7 isophote , we find that de vaucouleurs galaxies , at this low surface brightness , are flatter than exponential galaxies of the same absolute magnitude .
for a given surface brightness profile type , very bright galaxies are rounder , on average , than fainter galaxies .
we deconvolve the distributions of apparent axis ratios to find the distribution of the intrinsic short - to - long axis ratio @xmath8 , making the assumption of constant triaxiality @xmath9 . for all profile types and luminosities ,
the distribution of axis ratios is inconsistent with a population of oblate spheroids , but is usually consistent with a population of prolate spheroids .
bright galaxies with a de vaucouleurs profile ( @xmath10 , @xmath11 ) have a distribution of @xmath3 that is consistent with triaxiality in the range @xmath12 , with mean axis ratio @xmath13 .
the fainter de vaucouleurs galaxies are best fit with prolate spheroids ( @xmath14 ) with mean axis ratio @xmath15 . |
the interest in continuum lepton - pair production in high - energy collisions dates back to the seventies , triggered by the detection of the drell - yan process @xcite and the @xmath3 .
the latter , in particular , sharpened the attention to anything which might still have escaped detection , and a flood of new experimental findings on lepton pairs appeared , both for low masses ( lmr , m@xmath41 gev ) and for intermediate masses ( imr , 1@[email protected] gev ) .
the results were usually compared to expectations from an `` hadron - decay cocktail '' , containing all contributions known at that time .
an excess of single leptons and lepton pairs above the known sources was indeed found , coined `` anomalous '' pairs , and created great excitement . a review of the situation in the lmr region as of 1984 is contained in @xcite .
unfortunately , the results did not survive critical reassessment in later years , and they were finally recognized by helios-1 @xcite and , with higher precision , by ceres @xcite as due to a severe underestimate of the contribution from @xmath5 dalitz decays .
only one result , obtained at the isr at @xmath6=63 gev @xcite , survived as non - trivial up to today . in the imr region ,
some excess - pair production was also suspected for a long time , due to insufficient knowledge of the contribution from open charm decays on top of drell - yan .
any significant anomaly in this region was only ruled out much later , see e.g. @xcite .
ironically , these dubious @xmath0 results led already in the late seventies to two seminal theoretical papers , which had an enormous influence on the nascent field of high - energy @xmath2 collisions .
bjorken and weisberg @xcite were the first to propose partons _
produced _ in the collision to be a potential further source of continuum lepton pairs , beyond the intrinsic partons in the collision partners responsible for drell - yan ; they estimated the resulting excess above the latter to be a factor of 10 - 100 in the lmr region .
shuryak @xcite proposed the production of deconfined partons in thermal equilibrium during the collision and phrased the terms `` quark gluon plasma '' for the created medium and `` thermal radiation '' for the emitted lepton pairs in the imr region @xcite . the first systematic discussion , including both particle and nuclear physicist , on the experimental and theoretical aspect of qgp formation in ultra - relativistic nuclear collisions took place at the quark matter conference 1982 in bielefeld @xcite .
the basic instrumental elements of the first - generation experiments at the cern sps as well as the basic theoretical ideas on all observables were addressed .
the principal conclusions for lepton pairs were as follows .
( i ) the physics of dileptons ( virtual photons ) may be both more rich and more rigorous than that of real photons , due to the existence of two independent variables instead of one ( @xmath7 , @xmath8 vs. @xmath8 ) , and due to the simpler lowest - order rates ( @xmath9 vs. @xmath10@xmath11@xmath12 ) .
( ii ) thermal dilepton production in the lmr region may be dominantly hadronic , mediated by the broad vector meson @xmath1 ( 770 ) in the form @xmath13 ; due to its short lifetime of only 1.3 fm , the observation of a `` melting '' ( broadening ) and/or mass shift may serve as a prime probe of _ chiral symmetry restoration _ @xcite .
( iii ) thermal dilepton production in the imr region may be dominantly partonic , mediated in the form @xmath14 , and may serve as a prime probe of _ deconfinement _ ( the idea of @xmath3 suppression was not yet born ) .
classical theoretical papers on continuum lepton pairs with a broad view appeared soon after @xcite .
the first generation of sps experiments sensitive to continuum lepton pairs , helios-2 and na38 , found one anomaly @xcite , but did not follow up its significance at that time . only in the next generation , with ceres , helios-3 and na50 ,
clear signs of new physics appeared in a broader way , 13 years or more after the bielefeld workshop .
the experimental results from ceres for s - au @xcite in the lmr region are shown in fig .
[ fig1 ] ( left ) .
a large excess of pairs above the known hadron decays is seen .
this gave an enormous boost to theory , with hundreds of publications .
a small fraction of those , relying on @xmath1 production without in - medium effects , is contained in the figure .
the pole region is enhanced because of regeneration via @xmath15 during the fireball expansion , but the bulk of the excess residing below the pole is not at all described . only switching - on in - medium effects ,
e.g. mass shifts , based on a direct connection to the restoration of chiral symmetry @xcite , or broadening , based on a hadronic many - body approach @xcite , leads to a satisfactory description , while not discriminating between the two .
this ambivalent situation also persisted into the pb - beam era , as illustrated in fig .
[ fig1 ] ( right ) for the ceres 1995/96 data @xcite ( and still valid for the 2000 data @xcite ) : the main two scenarios @xcite fit the data equally well , and the true in - medium properties of the @xmath1 could unfortunately not be clarified , due to insufficient data quality .
the excess of pairs observed by helios-3 @xcite for s - w with respect to p - w is shown in fig .
[ fig2 ] ( left ) .
it is seen here to also occur in the imr region , not only at low masses , and led to a further important theoretical step : the recognition of the role of chiral ( v - a ) mixing with possibly sizable contributions from @xmath16 processes @xcite .
a strong excess of pairs was finally also reported by na50 @xcite , fig .
[ fig2 ] ( right ) , attributed at that time to either enhanced charm production or thermal radiation .
the former was never followed up theoretically , while the latter received a quantitative description in terms of hadron - parton duality , not explicitly specifying the sources @xcite ( see plot ) .
experimentally , the role of open charm and the nature of the thermal sources dominating the imr region remained open .
na60 is a third - generation experiment , built specifically to follow up the open issues addressed in the previous section .
it combines the muon spectrometer previously used by na50 with a novel radiation - hard silicon pixel vertex tracker , embedded in a 2.5 @xmath17 dipole magnet in the target region @xcite .
track matching between the two spectrometers , both in coordinate and momentum space , improves the dimuon mass resolution by a factor of 4 relative to na50 and allows to distinguish prompt from decay dimuons , while the radiation hardness of the si tracker together with a very high readout speed allow to maintain the high luminosity of common dimuon experiments .
the enormous jump in technology is responsible for the corresponding jump in data quality now achieved by na60 .
the results reported here were obtained from the analysis of data taken in 2003 for in - in at 158 agev .
[ fig3 ] ( left ) illustrates the data sample obtained in the lmr region @xcite . after subtracting the combinatorial background , determined by event mixing , and the signal fake matches between the tracks of the two spectrometers , determined by overlay mc ,
the resulting net spectrum contains about 440000 muon pairs .
the vector mesons @xmath18 and @xmath19 are completely resolved ; the mass resolution at the @xmath18 is 20 mev .
the improvements over ceres are a factor of @xmath201000 in statistics and a factor of 2 - 5 in mass resolution .
the centrality of the events is tagged by determining the associated charged - particle multiplicity density @xmath21 from the tracks of the si telescope .
the peripheral data ( 4 - 30 ) can essentially be described by the expected electromagnetic decays of the neutral mesons , i.e. the 2-body decays of the @xmath5 , @xmath1 , @xmath18 and @xmath19 resonances and the dalitz decays of the @xmath5 , @xmath22 and @xmath18 @xcite .
this is not true for the total data as plotted in fig .
[ fig3 ] ( right ) , due to the existence of a strong excess of pairs . to isolate this excess with _ a priori unknown characteristics _ without any fits ,
the cocktail of the decay sources is subtracted from the total data using _ local _ criteria , which are solely based on the mass distribution itself .
the @xmath1 is not subtracted .
the excess resulting from this difference formation is illustrated in the same figure . in practice ,
the procedure is done separately for each centrality window and , in connection with the @xmath8-spectra , for each @xmath8 bin ( see @xcite for details and error discussion ) .
the evolution of the spectral shape of the excess vs. centrality @xcite is most remarkable : a peaked structure is always seen , residing on a broad continuum with a yield strongly increasing with centrality , but remaining essentially centered around the position of the nominal @xmath1 pole .
these qualitative features are consistent with an interpretation of the excess as dominantly due to @xmath23 annihilation . fig .
[ fig4 ] shows the data for 110@xmath4@xmath21@xmath4170 in comparison to the two main theoretical scenarios discussed before : broadening of the @xmath1 ( rapp / wambach @xcite ) , and dropping mass of the @xmath1 ( brown / rho @xcite ) , both evaluated for in - in at @xmath21=140 for the same fireball evolution @xcite .
since agreement between modeling and data would imply agreement both in shape and yield , the model results are normalized to the data in the mass interval [email protected] gev , just to be independent of the uncertainties of the fireball evolution .
the unmodified @xmath1 , also shown in fig .
[ fig4 ] ( vacuum @xmath1 ) , is clearly ruled out .
the dropping - mass scenario is also ruled out , showing the much improved discrimination power of na60 .
the broadening scenario , on the other hand , fits perfectly well for [email protected] gev .
it is important to note that the data as shown have not been corrected for the mass- and @xmath8-dependent acceptance of the na60 setup , requiring the model results to be propagated through the acceptance filter for fair comparison to the data . by _
pure coincidence _ , as long as no @xmath8 cut is applied , that filtering nearly compensates for all the phase space factors associated with the thermal dilepton radiation , and just leaves a mass spectrum equivalent to the spectral function of the @xmath1 , averaged over momenta and the complete space - time evolution of the fireball , within an accuracy of about 10% @xcite .
the flat part of the measured spectrum may thus reflect the early history close to the qcd phase boundary , while the narrow peak on top may be due to the late part close to thermal freeze - out . by now
, several new sets of model descriptions have appeared @xcite .
all of them are based on broadening of the @xmath1 , and all of them describe the data reasonably well , even in absolute terms .
the fireball evolution , quite different in these sets , takes explicit account both of temperature and of baryon density , and the latter seems clearly required to describe the low - mass tail of the spectral function @xcite , in accord with conclusions reached previously by ceres from an increase of the excess at the lower beam energy of 40 agev @xcite .
the mass region [email protected] gev is now also addressed and again reasonably well described , in terms of either hadronic processes @xcite ( 4@xmath24 ... ) , or partonic processes @xcite ( @xmath25 ) , reflecting the traditional ambivalence of hadron - parton duality in the mass domain in this region .
how could one distinguish ?
as already emphasized above , lepton pairs are characterized by _ two _ variables , @xmath7 and @xmath8 .
the latter contains not only contributions from the spectral function , but encodes in fact the key properties of the expanding fireball , temperature and transverse ( radial ) flow .
in contrast to hadrons , however , which always receive the full asymptotic flow at the moment of decoupling , lepton pairs are continuously emitted during the evolution , sensing small flow and high temperatures at early times , and increasingly larger flow and smaller temperatures at later times . potentially therefore , the resulting space - time folding over the temperature - flow history offers access , through the measurement of @xmath8 spectra , to the emission region of the dileptons and may thus differentiate between a hadronic and a partonic nature of the emitting source @xcite .
@xmath26-spectra of the excess radiation in the lmr region , fully corrected for acceptance , have recently become available @xcite .
the acceptance correction is done in a fine grid in the @xmath7-@xmath8 plane to be independent of the properties of the source ; details can be found in @xcite .
[ fig5 ] ( left ) displays the centrality - integrated data in the form of invariant @xmath27-spectra , where @xmath28 , for four mass windows . at very low @xmath27
, a steepening is observed in all cases , contrary to the expectation for radial flow at masses above the pion mass ; the @xmath19 , plotted for comparison , does not show that ( while pions themselves do ) .
for very peripheral data , shown in fig .
[ fig5 ] ( right ) , the phenomenon disappears , and the @xmath1-like window and the @xmath18 become identical . at higher
@xmath27 , the spectra monotonically flatten with mass , reminiscent of radial flow , but then steepen again above the @xmath1 ; this striking feature will be addressed below .
the central na60 results for the imr region @xcite are illustrated in fig .
[ fig6 ] . by measuring the offset between the muon tracks and the main interaction vertex , the contributions to the prompt and to the offset part ( from @xmath29 decays )
can be disentangled . as shown in the left panel ,
the offset distribution is perfectly consistent with no charm enhancement , expressed by a fraction of 0.9 of the canonical level , while the enhancement seen before by na50 @xcite and now reconfirmed by na60 has to be solely attributed to the prompt part , expressed by a factor of 2.4 of the drell - yan level : a further milestone in experimental results .
the right panel contains the decomposition of the total data into drell - yan , open charm and a prompt excess obtained as the difference with respect to the total .
the excess part is seen to exhibit the same fall - off vs. mass as charm , steeper than dy .
it increases with centrality , and its transverse momentum spectra are also much steeper than dy @xcite .
the fit temperatures of the @xmath27-spectra associated with 3 mass windows are indicated on the bottom of the figure .
there are all indications that this is the thermal radiation addressed already by na50 as one the options ( see fig . [ fig2 ] , right ) . vs. dimuon mass @xmath7 for the combined lmr / imr regions @xcite.,width=238,height=219 ] the central information extracted from the @xmath27 spectra , expressed as the inverse slope parameter @xmath30 vs. dimuon mass ,
is shown in fig .
[ fig7 ] @xcite .
this unifies the data from the lmr and imr regions , including a common fit range ( 0.4@xmath4@[email protected] gev ) and subtraction of charm throughout .
the hadron data for @xmath5 , @xmath18 and @xmath19 obtained as a by - product of the cocktail subtraction are also plotted .
the parameter @xmath30 is seen to rise nearly linearly with mass up to the pole position of the @xmath1 , followed by a sudden decline to a rather constant level of 190 mev above .
the excess data along the rise are quite close to the hadron data .
however , the peak of the @xmath1 , residing on the broad underlying continuum , can be interpreted as the freeze - out ( vacuum ) @xmath1 without in - medium effects @xcite and analyzed separately by a side - window subtraction method , resulting in a value for @xmath30 of about 300 mev ( also contained in fig .
[ fig7 ] ) .
the observed linear rise of @xmath30 with mass of the excess can therefore be considered as qualitatively _ consistent _ with the expectations for radial flow of an in - medium _ hadronic _ source ( here @xmath31 ) decaying into lepton pairs .
the absolute values of @xmath30 are well below the hadron line defined by the vacuum @xmath1 , as expected , and the other hadrons freeze - out earlier , due to their smaller coupling to the pions .
it is interesting to note that the large difference of @xmath2050 mev in @xmath30 between the vacuum @xmath1 and the @xmath18 ( same mass ) quantitatively disappears for the lowest peripheral `` @xmath0-like '' selection @xmath21@xmath410 , as visible in the right panel of fig .
[ fig5 ] . the sudden decline of @xmath30 at masses
@xmath201 gev is _ the other _ most remarkable feature in fig .
[ fig7 ] . _
if _ the rise is due to flow , the sudden transition to a seeming low - flow scenario is hard to reconcile with emission sources which continue to be of dominantly hadronic origin in this region .
a more natural explanation would then be a transition to a qualitatively different source , implying dominantly early , i.e. _ partonic _ processes like @xmath32 for which flow has not yet built up @xcite . while still controversial @xcite , this may well be the first direct hint for thermal radiation of partonic origin , breaking hadron - parton duality for the _ yield _ description in the mass domain .
there are altogether four options in sight which will have the chance to advance the field further : at rhic , lhc , sps and fair . at rhic , phenix is the only experiment capable to measure lepton pairs both in the lmr and the imr region .
first results on electron pairs , showing an excess , have already appeared @xcite , but proper tools for the rejection of low - mass dalitz and conversion pairs have not yet been available , and the present data quality is therefore severely limited by s / b ratios of @xmath4 1/1000 .
a second - generation upgrade at phenix will soon become available to cope with rejection , relying on a `` hadron - blind - detector '' ( a rich ) within a field - free region including the interaction vertex @xcite ; the required magnet geometry , based on a double coil arrangement , was proposed in the course of the ceres r&d @xcite . at the lhc ,
the lmr region is probably unaccessible , due to prohibitive charged particle multiplicity densities , but the ( even extended ) imr region is covered both by alice and cms and may lead to exciting new results . on top , both colliders offer an excellent chance for a revisit of @xmath0 collisions also in the lmr region , made attractive through record - like bjorken energy densities . at the sps ,
a continuation of na60 is conceivable , both for pb - beams and for lower beam energies .
the existence proof for success exists ; the rest is a ( possibly unsurmountable ) mixture of elements which are beyond pure science . at fair , finally , a multi - purpose detector ( cbs ) is in an early stage of development , meant to also cover lepton pairs . in all probability ,
the time horizon is here beyond a decade .
since qm 1982 in bielefeld , 25 years have passed ( a whole generation ) , and only now initial central issues seem to be clearing up .
the first accurate measurement of the @xmath1 spectral function in the hot and dense environment of the qcd phase transition exists , but a solid theoretical link to chiral restoration is still missing . the first hint for thermal radiation of mostly partonic origin exists , but it surely requires theoretical consolidation .
the field takes a long breath indeed ... 00 s. d. drell and t. m. yan , phys .
lett . * 25 * , 316 ( 1970 ) .
h. j. specht , quark matter helsinki 84 , 221 ( 1984 ) ( springer , heidelberg ) .
t. akesson _ et al . _
( helios-1 collaboration ) , z. phys .
c * 68 * , 47 ( 1995 ) .
g. agakishiev _ et al . _ ( ceres collaboration ) , eur .
j. c * 4 * , 249 ( 1998 ) .
t. akesson _ et al . _
( axial field spectrometer collaboration ) , phys .
b * 152 * , 411 ( 1985 ) .
m. c. abreu _ et al . _
( na38 collaboration ) , nucl .
a * 566 * ( 1994 ) 77c .
j. d. bjorken and h. weisberg , phys .
d * 13 * ( 1976 ) 1405 .
e. v. shuryak , phys .
b * 78 * , 150 ( 1978 ) .
j. g. branson _ et al .
_ , phys .
* 38 * ( 1977 ) 1334 .
quark matter bielefeld 82 , ( world scientific , singapore ) .
r. d. pisarski , phys .
110b * , 155(1982 ) .
l. d. mclerran and t. toimela , phys .
d * 31 * ( 1985 ) 545 .
k. kajantie , m. kataja , l. d. mclerran and p. v. ruuskanen , phys .
d * 34 * ( 1986 ) 811 .
g. agakichiev _ et al . _
( ceres collaboration ) , phys .
lett . * 75 * , 1272 ( 1995 ) .
g. q. li , c. m. ko and g. e. brown , phys . rev .
lett . * 75 * , 4007 ( 1995 ) .
g. chanfray , r. rapp and j. wambach , phys .
lett . * 76 * , 368 ( 1996 ) .
g. agakichiev _ et al . _
( ceres collaboration ) , phys .
* b422 * , 405 ( 1998 ) ; b. lenkeit _ et al .
_ , nucl . phys . *
a661 * , 23c ( 1999 ) ; g. agakichiev _ et al .
_ , eur .
j. * c41 * , 475 ( 2005 ) .
d. miskowiec ( ceres collaboration ) , nucl .
a * 774 * , 43 ( 2006 ) .
r. rapp and j. wambach , adv .
. phys . * 25 * , 1 ( 2000 ) .
g. e. brown and m. rho , phys
. rept . * 363 * , 85 ( 2002 ) .
m. masera ( helios-3 collaboration ) , nucl . phys . a * 590 * ( 1995 ) 93c ; a. l. s. angelis _ et al . _ , eur .
j. c * 13 * ( 2000 ) 433 .
g. q. li and c. gale , phys .
* 81 * , 1572 ( 1998 ) .
m. c. abreu _ et al . _
( na38/na50 collaboration ) , nucl .
phys . a * 698 * ( 2002 ) 539 ; l. capelli , phd thesis , universit claude bernard lyon - i , lyon , france , 2001 .
r. rapp and e. v. shuryak , phys .
b * 473 * , 13 ( 2000 ) .
g. usai _ et al . _
( na60 collaboration ) , eur .
j. * c43 * , 415 ( 2005 ) .
r. arnaldi _ et al . _
( na60 collaboration ) , phys .
* 96 * , 162302 ( 2006 ) .
s. damjanovic _ et al . _
( na60 collaboration ) , eur .
j. c * 49 * , 235 ( 2007 ) .
s. damjanovic _ et al . _
( na60 collaboration ) , nucl .
a * 783 * , 327 ( 2007 ) .
r. rapp , private communication ( 2005 ) .
h. van hees and r. rapp , phys .
* 97 * , 102301 ( 2006 ) , r. rapp , j. phys .
g * 34 * ( 2007 ) s405 , h. van hees and r. rapp , j. phys .
g * 34 * ( 2007 ) s1051 .
j. ruppert and t. renk , hep - ph/0612113 ; j. phys .
g * 34 * ( 2007 ) s1047 ; j. ruppert , c. gale , t. renk , p. lichard and j. i. kapusta , hep - ph/0706.1934 . k. dusling , d. teaney and i. zahed , phys .
c * 75 * ( 2007 ) 024908 ; k. dusling and i. zahed , hep - ph/0701253 .
d. adamova _ et al . _ ( ceres / na45 collaboration ) , phys . rev . lett .
* 91 * ( 2003 ) 042301 .
m. asakawa , c. m. ko and p. levai , phys .
* 70 * ( 1993 ) 398
. g. usai _ et al . _
( na60 collaboration ) , j. phys .
g * 34 * ( 2007 ) s233 .
r. arnaldi _ et al . _
( na60 collaboration ) , submitted for publ . to phys
. rev .
r. shahoyan _ et al .
_ , j. phys .
g * 34 * ( 2007 ) s1029 ; r. shahoyan _ et al .
_ , eur .
j. c43 ( 2005 ) 209 .
r. arnaldi _ et al . _
( na60 collaboration ) , proceedings of the inpc07 , tokyo ( 2007 ) .
a. kozlov _ et al .
_ ( phenix collaboration ) , eur .
j. a * 31 * ( 2007 ) 836 .
s. afanasiev _ et al . _
( phenix collaboration ) , arxiv:0706.3034 [ nucl - ex ] .
p. glassel and h. j. specht , berkeley 1987 , proceedings , experiments and detectors for a relativistic heavy ion collider ( rhic ) , 106 . | the key results on lepton - pair production in ultra - relativistic nuclear collisions are shortly reviewed , starting at the roots of @xmath0 collisions in the seventies , and ending at the perspectives of the colliders rhic and lhc .
the presence is dominated by the recent precision results from na60 at the cern sps , culminating in the first measurement of the in - medium @xmath1 spectral function and the transverse flow of the associated thermal radiation .
the seeming cut - off of the flow above the @xmath1 may well be the first direct hint for thermal radiation of partonic origin in nuclear collisions .
the major milestones in the theoretical developments are also covered .
relativistic heavy - ion collisions , quark - gluon plasma , lepton pairs 25.75.-q , 12.38.mh , 13.85.qk |
cooperation has played a fundamental role in the early evolution of our societies@xcite and continues playing a major role still nowadays . from the individual level , where we cooperate with our romantic partner , friends , and co - workers in order to handle our individual problems , up to the global level where countries cooperate with other countries in order to handle global problems ,
our entire life is based on cooperation .
given its importance , it is not surprising that cooperation has inspired an enormous amount of research across all biological and social sciences , spanning from theoretical accounts @xcite to experimental studies @xcite and numerical simulations@xcite .
since the resolution of many pressing global issues , such as global climate change and depletion of natural resources , requires cooperation among many actors , one of the most relevant questions about cooperation regards the effect of the size of the group on cooperative behavior . indeed , since the influential work by olson @xcite
, scholars have recognized that the size of a group can have an effect on cooperative decision - making .
however , the nature of this effect remains one of the most mysterious areas in the literature , with some scholars arguing that it is negative @xcite , others that it is positive @xcite , and yet others that it is ambiguous @xcite or non - significant @xcite .
interestingly , the majority of field experiments seem to agree on yet another possibility , that is , that group size has a curvilinear effect on cooperative behavior , according to which intermediate - size groups cooperate more than smaller groups and more than larger groups @xcite .
the emergence of a curvilinear effect of the group size on cooperation in real life situations is also supported by data concerning academic research , which in fact support the hypothesis that research quality of a research group is optimized for medium - sized groups @xcite . here
we aim at shedding light on this debate , by providing evidence that a single parameter can be responsible for all the different and apparently contradictory effects that have been reported in the literature .
specifically , we show that the effect of the size of the group on cooperative decision - making depends critically on a parameter taking into account different ways in which the notion of cooperation itself can be defined when there are more than two agents . indeed , while in case of only two agents a cooperator can be simply defined as a person willing to pay a cost @xmath0 to give a greater benefit @xmath1 to the other person @xcite , the same definition , when transferred to situations where there are more than two agents , is subject to multiple interpretations .
if cooperation , from the point of view of the cooperator , means paying a cost @xmath0 to create a benefit @xmath1 , what does it mean from the point of view of the _ other _
player__s _ _ ? does @xmath1 get earned by each of the other players or does it get shared among all other players , or none of them ?
in other words , what is the marginal return for cooperation ? of course , there is no general answer and , in fact , previous studies have considered different possibilities .
for instance , in the standard public goods game it is assumed that @xmath1 gets earned by each player ( including the cooperator ) ; instead , in the n - person prisoner s dilemma ( as defined in @xcite ) it is assumed that @xmath1 gets shared among all players ; yet , the volunteer s dilemma @xcite and its variants using critical mass @xcite rest somehow in between : one or more cooperators are needed to generate a benefit that gets earned by each player , but , after the critical mass is reached , new cooperators do not generate any more benefit ; finally , it has been pointed out @xcite that a number of realistic situations can be characterized by a marginal return which increases linearly for early contributions and then decelerates , reflecting the natural decrease of marginal returns that occurs when output limits are approached . in order to take into account this variety of possibilities , we consider a class of _ social dilemmas _ parametrized by a function @xmath2 describing the marginal return for cooperation when @xmath3 people cooperate in a group of size @xmath4 .
more precisely , our _ general public goods game _ is the n - person game in which n people have to simultaneously decide whether to cooperate ( c ) or defect ( d ) . in presence of a total of @xmath3 cooperators ,
the payoff of a cooperator is defined as @xmath5 ( @xmath6 represents the cost of cooperation ) and the payoff of a defector is defined as @xmath7 . in order to have a social dilemma ( i.e. , a tension between individual benefit and the benefit of the group as a whole ) we require that : * full cooperation pays more than full defection , that is , @xmath8 , for all @xmath4 ; * defecting is individually optimal , regardless of the number of cooperators , that is , for all @xmath9 , one has @xmath10 . the aim of this paper is to provide further evidence that the function @xmath11 might be responsible for the confusion in the literature about group size effect on cooperation . in particular , we focus on the situation , inspired from realistic scenarios , in which the natural output limits of the public good imply that @xmath7 increases fast for small @xmath3 s and then stabilizes . indeed , in our previous work @xcite , we have shown that the size of the group has a positive effect on cooperation in the standard public goods game and has a negative effect on cooperation in the n - person prisoner s dilemma .
a reinterpretation of these results is that , if @xmath12 increases linearly with @xmath4 ( standard public goods game ) , then the size of the group has a positive effect on cooperation ; and , if @xmath12 is constant with @xmath4 ( n - person prisoner s dilemma ) , then the size of the group has a negative effect on cooperation .
this reinterpretation suggests that , in the more realistic situations in which the benefit for full cooperation increases fast for early contributions and then decelerates once the output limits of the public good are approached , we may observe a curvilinear effect of the group size , according to which intermediate - size groups cooperate more than smaller groups and more than larger groups .
to test this hypothesis , we have conducted a lab experiment using a general public goods game with a piecewise function @xmath11 , which increases linearly up to a certain number of cooperators , after which it remains constant . while it is likely that realistic scenarios would be better described by a smoother function , this is a good approximation of all those situations
in which the natural output limits of a public good imply that the increase in the marginal return for cooperation tends to zero as the number of contributors grows very large .
the upside of choosing a piecewise function @xmath11 is that , in this way , we could present the instructions of the experiment in a very simple way , thus minimizing random noise due to participants not understanding the decision problem at hand ( see method ) .
our results support indeed the hypothesis of a curvilinear effect of the size of the group on cooperative decision - making .
taken together with our previous work @xcite , our findings thus ( i ) shed light on the confusion regarding the group size effect on cooperation , by pointing out that different values of a single parameter might give rise to qualitatively different group size effects , including positive , negative , and even curvilinear ; and ( ii ) they help fill the gap between lab experiments and field experiments . indeed , while lab experiments use either the standard public goods game or the n - person prisoner s dilemma , _ real _ public goods game are mostly characterized by a marginal return of cooperation that increases fast for early contributions and then approaches a constant function as the number of cooperators grows very large - and our results provide evidence that these three situations give rise to three different group size effects .
we have recruited participants through the online labour market amazon mechanical turk ( amt ) @xcite . after entering their turkid ,
participants were directed to the following instruction screen . _ welcome to this hit . _
_ this hit will take about 5 minutes and you will earn 20c for participating . _ _ this hit consists of a decision problem followed by a few demographic questions . _
_ you can earn an additional bonus depending on the decisions that you and the participants in your cohort will make .
_ _ we will tell you the exact number of participants in your cohort later .
_ _ each one of you will have to decide to join either group a or group b. _ _ your bonus depends on the group you decide to join and on the size of the two groups , a and b , as follows : _ * _ if the size of group a is 0 ( that is , everybody chooses to join group b ) , then everybody gets 10c _ * _ if the size of group a is 1 , then the person in group a gets 5c and each person in group b gets 15c _ * _ if the size of group a is 2 , then each person in group a gets 10c and each person in group b gets 20c _ * _ if the size of group a is 3 , then each person in group a gets 15c and each person in group b gets 25c _ * _ if the size of group a is 4 , then each person in group a gets 20c and each person in group b gets 30c _ * _ and so on , up to 10 : if the size of group a is 10 , then each person in group a gets 50c and each person in group b gets 60c _ * _ however , if the size of group a is larger than 10 , then , independently of the size of the two groups , each person in group a will still get 50c and each person in group b will still get 60c . _ after reading the instructions , participants were randomly assigned to one of 12 conditions , differing only on the size of the cohort ( @xmath13 ) .
for instance , the decision screen for the participants in the condition where the size of the cohort is 3 was : _ you are part of a cohort of 3 participants
. _ _ which group do you want to join ? _ by using appropriate buttons , participants could select either group a or group b. we opted for not asking any comprehension questions .
we made this choice for two reasons .
first , with the current design , it is impossible to ask general comprehension questions such as `` what is the strategy that benefits the group as a whole '' , since this strategy depends on the strategy played by the other players .
second , we did not want to ask particular questions about the payoff structure since this may anchor the participants reasoning on the examples presented .
of course , a downside of our choice is that we could not avoid random noise .
however , as it will be discussed in the results section , random noise can not be responsible for our findings . instead
, our results would have been even cleaner , if we had not had random noise , since the initial increase of cooperation and its subsequent decline would have been more pronounced ( see results section for more details ) . after making their decisions ,
participants were asked to fill a standard demographic questionnaire ( in which we asked for their age , gender , and level of education ) , after which they received the `` survey code '' needed to claim their payment . after collecting all the results ,
bonuses were computed and paid on top of the participation fee , that was $ 0.20 . in case
the number of participants in a particular condition was not divisible by the size of the cohort ( it is virtually impossible , in amt experiments , to decide the exact number of participants playing a particular condition ) , in order to compute the bonus of the remaining people we formed an additional cohort where these people where grouped with a random choice of people for which the bonus had been already computed .
additionally , we anticipate that only 98 subjects participated in the condition with n=100 . this does not generate deception in the computation of the bonuses since the payoff structure of the game does not depend on @xmath4 ( as long as @xmath14 ) . as a consequence of these observations , no deception was used in our experiment . according to the dutch legislation ,
this is a non - wmo study , that is ( i ) it does not involve medical research and ( ii ) participants are not asked to follow rules of behavior .
see http://www.ccmo.nl / attachments / files / wmo- engelse-vertaling-29-7-2013-afkomstig-van-vws.pdf , section 1 , article 1b , for an english translation of the medical research act .
thus ( see http://www.ccmo.nl / en / non - wmo- research ) the only legislations which apply are the agreement on medical treatment act , from the dutch civil code ( book 7 , title 7 , section 5 ) , and the personal data protection act ( a link to which can be found in the previous webpage ) .
the current study conforms to both . in particular , anonymity was preserved because amt `` requesters '' ( i.e. , the experimenters ) have access only to the so - called turkid of a participant , an anonymous i d that amt assigns to a subject when he or she registers to amt . additionally , as demographic questions we only asked for age , gender , and level of education .
a total of 1.195 _ distinct _ subjects located in the us participated in our experiment . _
distinct _ subjects means that , in case two or more subjects were characterized by either the same turkid or the same ip address , we kept only the first decision made by the corresponding participant and eliminated the rest .
these multiple identities represent usually a minor problem in amt experiments ( only 2% of the participants in the current dataset ) .
participants were distributed across conditions as follows : 101 participants played with @xmath15 , 99 with @xmath16 , 102 with @xmath17 , 101 with @xmath18 , 98 with @xmath19 , 103 with @xmath20 , 97 with @xmath21 , 99 with @xmath22 , 97 with @xmath23 , 101 with @xmath24 , 99 with @xmath25 , 98 with @xmath26 .
1 summarizes the main result .
the rate of cooperation , that is the proportion of people opting for joining group a , first increases as the size of the group increases from @xmath15 to @xmath18 , then it starts decreasing .
the figure suggests that the relation between the size of the group and the rate of cooperation is _ not _ quadratic : while the initial increase of cooperation is relatively fast , the subsequent decrease of cooperation seems extremely slow .
this is confirmed by linear regression predicting rate of cooperation as a function of @xmath4 and @xmath27 , which shows that neither the coefficient of @xmath4 nor that of @xmath27 are significant ( @xmath28 , resp . ) .
for this reason we use a more flexible econometric model than the quadratic model , consisting of two linear regressions , one with a positive slope ( for small @xmath4 s ) and the other one with a negative slope ( for large @xmath4 s ) . as a switching point
, we use the @xmath18 , corresponding to the size of the group which reached maximum cooperation .
doing so , we find that both the initial increase of cooperation and its subsequent decline are highly significant ( from @xmath15 to @xmath18 : coeff @xmath29 , @xmath30 ; from @xmath18 to @xmath26 : coeff @xmath31 , @xmath32 ) . to @xmath18 : coeff @xmath29 , @xmath30 ; from @xmath18 to @xmath26 : coeff @xmath31 , @xmath32)._,title="fig : " ] [ fig : intermediate ] we conclude by observing that not only random noise can not explain our results , but , without random noise , the effect would have been even stronger . indeed , first we observe that there is no a priori worry that random noise would interact with any condition and so we can assume that it is randomly distributed across conditions .
then we observe that subtracting a binary distribution with average @xmath33 from a binary distribution with average @xmath34 , one would obtain a distribution with average @xmath35 .
similarly , subtracting a binary distribution with average @xmath33 from a binary distribution with average @xmath36 one would obtain a distribution with average @xmath37 .
thus , if the @xmath38 s are the averages that we have found ( containing random noise ) and the @xmath39 s are the _ true _ averages ( without random noise ) , the previous inequalities allow us to conclude that the initial increase of cooperation and its following decrease would have been stronger in absence of random noise .
here we have reported on a lab experiment providing evidence that the size of a group can have a curvilinear effect on cooperation in one - shot social dilemmas , with intermediate - size groups cooperating more than smaller groups and more than larger groups . joining the current results with those of a previously published study of us @xcite
, we can conclude that group size can have qualitatively different effects on cooperation , ranging from positive , to negative and curvilinear , depending on the particular decision problem at hand .
interestingly , our findings suggest that different group size effects might be ultimately due to different values of a single parameter , the number @xmath12 , describing the benefit for full cooperation .
if @xmath12 is constant in @xmath4 , then group size has a negative effect on cooperation ; if @xmath12 increases linearly with @xmath4 , then group size has a positive effect on cooperation ; in the _ middle _ , all sorts of things may a priori happen .
in particular , in the realistic situation in which @xmath12 is a piecewise function that increases linearly with @xmath4 up to a certain @xmath40 and then remains constant , then group size has a curvilinear effect , according to which intermediate - size groups cooperate more than smaller groups and more than larger groups .
see table 1 .
.summary of the different group size effects on cooperation depending on how the benefit for full cooperation varies as a function of the group size .
[ cols="<,^,^",options="header " , ] to the best of our knowledge , ours is the first study reporting a curvilinear effect of the group size on cooperation in an experiment conducted in the ideal setting of a lab , in which confounding factors are minimized .
previous studies reporting a qualitatively similar effect @xcite used field experiments , in which it is difficult to isolate the effect of the group size from possibly confounding effects . in our case , the only possibly confounding factor is random noise due to a proportion of people that may have not understood the rules of the decision problem .
as we have shown , our results can not be driven by random noise and , in fact , the curvilinear effect would have been even stronger , without random noise .
moreover , since our experimental design was inspired by a tentative to mimic all those _ real _ public goods games in which the natural output limits of the public good imply that the increase of the marginal return for cooperation , when the number of cooperators diverges , tends to zero , our results might explain the apparent contradiction that field experiments tend to converge on the fact that the effect of the group size is curvilinear , while lab experiments tend to converge on either of the two linear effects .
our contribution is also conceptual , since we have provided evidence that a single parameter might be responsible for different group size effects : the parameter @xmath12 , describing the way the benefit for full cooperation varies as a function of the size of the group . of course
, we do not pretend to say that this is the only ultimate explanation of why different group size effects have been reported in experimental studies .
in particular , in real - life situations , which are typically repeated and in which communication among players is allowed , other factors , such as within - group enforcement , may favor the emergence of a curvilinear effect of the group size on cooperation , as highlighted in @xcite . if anything , our results provide evidence that the curvilinear effect on cooperation goes beyond contingent factors and can be found also in the ideal setting of a lab experiment using one - shot anonymous games .
we believe that this is a relevant contribution in light of possible applications of our work .
indeed , the difference between @xmath12 and the total cost of full cooperation @xmath41 can be interpreted has the incentive that an institution needs to pay to the contributors in order to make them cooperate .
since institutions are interested in minimizing their costs and , at the same time , maximizing the number of cooperators , it is crucial to understand what is the `` lowest '' @xmath11 such that the resulting effect of the group size on cooperation is positive .
this seems to be an non - trivial question .
for instance , does @xmath42 give rise to a positive effect or is it still curvilinear or even negative ?
the technical difficulty here is that it is hard to design an experiment to test people s behavior in these situations , since one can not expect that an average person would understand the rules of the game when presented using a logarithmic functions . in terms of economic models , our results are consistent with utilitarian models such as the charness & rabin model @xcite and the novel cooperative equilibrium model @xcite .
both these models indeed predict that , in our experiment , cooperation initially ( i.e. , for @xmath43 ) increases with @xmath4 ( see @xcite for the details ) , and then starts decreasing .
this behavioral transition follows from the simple observation that free riding when there are more than 10 cooperators costs zero to each of the other players and benefits the free - rider .
thus , cooperation in larger groups is not supported by utilitarian models , which then predict a decrease in cooperative behavior whose speed depends on the particular parameters of the model , such as the extent to which people care about the group payoff versus their individual payoff , and people s beliefs about the behavior of the other players .
thus our results add to the growing body of literature showing that utilitarian models are qualitatively good descriptors of cooperative behavior in social dilemmas .
however , we note that while theoretical models predict that the rate of cooperation should start decreasing at @xmath17 , our results show that the rate of cooperation for @xmath18 is marginally significantly higher than the rate of cooperation for @xmath17 ( rank sum , @xmath44 ) .
although ours is a between - subjects experiment , this finding seems to hint at the fact that there is a proportion of subjects who would defect for @xmath17 and cooperate for @xmath18 .
this is not easy to explain : why should a subject cooperate with @xmath18 and defect with @xmath17 ?
one possibility is that there is a proportion of `` inverse conditional cooperators '' , who cooperate only if a small percentage of people cooperate : if these subjects believe that the rate of cooperation decreases quickly after @xmath17 , they would be more motivated to cooperate for @xmath18 than for @xmath17 .
another possibility , of course , is that this discrepancy is just a false positive . in any case , unfortunately our experiment is not powerful enough to detect the reason of this discrepancy between theoretical predictions and experimental results and thus we leave this interesting question for future research .
v.c . is supported by the dutch research organization ( nwo ) grant no .
this material is based upon work supported by the national science foundation under grant no .
0932078000 while the first author was in residence at the mathematical science research institute in berkeley , california , during the spring 2015 semester . 1 kaplan h , gurven m. the natural history of human food sharing and cooperation : a review and a new multi - individual approach to the negotiation of norms .
in : gintis h , bowles s , boyd r , fehr e , editors .
moral sentiments and material interests : the foundations of cooperation in economic life .
cambridge , ma : mit press ; 2005 .
tomasello m. a natural history of human thinking .
cambridge , ma : harvard university press ; 2014 .
trivers r. the evolution of reciprocal altruism .
q rev biol .
1971 ; 46 : 35 - 57 .
axelrod r , hamilton wd .
the evolution of cooperation . science .
1981 ; 211 : 1390 - 1396 .
fehr e , fischbacher u. the nature of human altruism . nature .
2003 ; 425 : 785 - 791 .
five rules for the evolution of cooperation . science . 2006 ; 314 , 1560 - 1563 .
perc m , szolnoki a. coevolutionary games - a mini review .
biosystems 2010 ; 99 : 109 - 125 .
press wh , dyson fj .
iterated prisoner s dilemma contains strategies that dominate any evolutionary opponent .
proc natl acad sci usa .
2012 ; 109 : 10409 - 10413 .
perc m , gmez - gardees j , szolnoki a , flora lm , moreno y. evolutionary dynamics of group interactions on structured populations : a review .
j roy soc interface .
2013 ; 10 : 20120997 .
capraro v. a model of human cooperation in social dilemmas .
plos one 2013 ; 8 : e72427 .
hilbe c , nowak ma , sigmund k. the evolution of extortion in iterated prisoners dilemma games .
proc natl acad sci usa .
2013 ; 110 : 6913 - 6918 .
rand dg , nowak ma .
human cooperation .
trends cogn sci .
2013 ; 17 : 413 - 425 .
capraro v , halpern jy .
translucent players : explaining cooperative behavior in social dilemmas . 2014 .
available : http://ssrn.com/abstract=2509678 .
andreoni j. why free ride ?
j public econ .
1988 ; 37 : 291 - 304 .
fischbacher u , gchter s , fehr e. are people conditionally cooperative ?
evidence from a public goods experiment .
econ lett .
2001 ; 71 : 397 - 404 .
milinski m , semmann d , kranbeck hj .
reputation helps solve the ` tragedy of the commons ' .
2002 ; 415 : 424 - 426 .
frey bs , meier s. social comparisons and pro - social behavior .
testing ` conditional cooperation ' in a field experiment .
am econ rev . 2004 ; 94 : 1717 - 1722 .
fischbacher u , gchter s. social preferences , beliefs , and the dynamics of free riding in public goods experiments .
am econ rev . 2010 ; 100 : 541 - 556 .
traulsen a , semman d , sommerfeld rd , krambeck h - j , milinski m. human strategy updating in evolutionary games .
proc natl acad sci usa .
2010 ; 107 : 2962 - 2966 .
apicella cl , marlowe fw , fowler jh , christakis na . social networks and cooperation in hunter - gatherers . nature .
2012 ; 481 : 497 - 501 .
capraro v , jordan jj , rand dg .
heuristics guide the implementation of social preferences in one - shot prisoner s dilemma experiments .
2014 ; 4 : 6790 .
capraro v , smyth c , mylona k , niblo ga .
benevolent characteristics promote cooperative behaviour among humans .
plos one . 2014 ; 9 : e102881 .
capraro v , marcelletti a. do good actions inspire good actions in others ?
2014 ; 4 : 7470 .
hauser op , rand dg , peysakhovich a , nowak ma . cooperating with the future .
2014 ; 511 : 220 - 223 .
gallo e , yan c. the effects of reputational and social knowledge on cooperation .
proc natl acad sci usa .
doi : 10.1073/pnas.1415883112 .
nowak ma , may rm .
evolutionary games and spatial chaos . nature .
1992 ; 359 : 826 - 829 .
boyd r , gintis h , bowles s , richerson pj .
the evolution of altruistic punishment .
proc natl acad sci usa .
2003 ; 100 : 3531 - 3535 .
santos fc , pacheco jm .
scale - free networks provide a unifying framework for the emergence of cooperation .
phys rev lett .
2005 ; 95 : 098104 .
perc m , szolnoki a. social diversity and promotion of cooperation in the spatial prisoner s dilemma game .
phys rev e. 2008 ; 77 : 011904 .
roca cp , cuesta ja , snchez a. evolutionary game theory : temporal and spatial effects beyond replicator dynamics .
phys life rev .
2009 ; 6 : 208 - 249 .
gmez - gardees j , reinares i , arenas a , flora lm . evolution of cooperation in multiplex networks .
2012 ; 2 : 620 .
jiang l - l , perc m. spreading of cooperative behaviour across interdependent groups .
2013 ; 3 : 2483 .
olson m. the logic of collective action : public goods and the theory of groups .
cambridge : harvard university press ; 1965 .
dawes r , mctavish j , shaklee h. behavior , communication , and the assumptions about other people s behavior in a commons dilemma situation .
j pers soc psychol .
1977 ; 35 : 1 - 11 .
komorita ss , lapwortb wc .
cooperative choice among individuals versus groups in an n - person dilemma situation .
j pers soc psychol .
1982 ; 42 : 487 - 496 .
baland jm , platteau jp .
the ambiguous impact of inequality on local resource management .
world dev .
1999 ; 27 : 773 - 788 . ostrom e. understanding institutional diversity .
princeton : princeton university press ; 2005 gruji j , eke b , cabrales a , cuesta ja , snchez a. three is a crowd in iterated prisoner s dilemmas : experimental evidence on reciprocal behavior .
sci rep . 2012
; 2 : 638 .
vilone d , giardini f , paolucci m. partner selection supports reputation - based cooperation in a public goods game .
preprint/ available : arxiv:1410.6625 .
nosenzo d , quercia s , sefton m. cooperation in small groups : the effect of group size .
2015 ; 18 : 4 - 14 .
mcguire mc .
group size , group homogeneity , and the aggregate provision of a pure public good under cournot behavior .
public choice .
1974 ; 18 : 107 - 126 .
isaac rm , walker jm , williams aw .
group size and the voluntary provision of public goods : experimental evidence utilizing large groups .
j public econ .
1994 ; 54 : 1 - 36 .
haan m , kooreman p. free riding and the provision of candy bars .
j public econ . 2002 ; 83 : 277 - 291 . agrawal a , chhatre a. explaining success on the commons : community forest governance in the indian himalaya .
world dev .
2006 ; 34 : 149 - 166 .
masel j. a bayesian model of quasi - magical thinking can explain observed cooperation in the public good game .
j econ behav organ . 2007 ; 64 : 216 - 231 .
zhang xq , zhu f. group size and incentives to contribute : a natural experiment at chinese wikipedia .
am econ rev .
2001 ; 101 : 1601 - 1615 .
szolnoki a , perc m. group - size effects on the evolution of cooperation in the spatial public goods game .
phys rev e. 2011 ; 84 : 047102 .
esteban j , ray d. collective action and the group size paradox .
am polit sci rev .
2001 ; 95 : 663 - 672 .
pecorino p , temimi a. the group size paradox revisited .
j public econ theory .
2008 ; 10 : 785 - 799 . oliver pe , marwell g. the paradox of group - size in collective action - a theory of the critical mass .
ii . am sociol rev .
1988 ; 53 : 1 - 8 .
chamberlin jr .
provision of collective goods as a function of group size .
am polit sci rev .
1974 ; 68 : 707 - 716 .
todd s. collective action : theory and applications .
ann arbor : university of michigan press ; 1992 gautam ap .
group size , heterogeneity and collective action outcomes : evidence from community forestry in nepal .
int j sustain dev world ecol . 2007 ;
14 : 574 - 583 .
rustagi d , engel s , kosfeld m. conditional cooperation and costly monitoring explain success in forest commons management . science .
2010 ; 330 : 961 - 965 .
poteete ar , ostrom e. heterogeneity , group size and collective action : the role of institutions in forest management .
dev change .
2004 ; 35 : 435 - 461 .
agrawal a , goyal s. group size and collective action - third - party monitoring in common - pool resources .
comp polit stud .
2001 ; 34 : 63 - 93 .
agrawal a. small is beautiful , but is larger better ?
forest management institutions in the kumaon himalaya , india . in : gibson c , mckean ma , ostrom e , editors . people and forests : communities , institutions , and governance .
cambridge , ma : mit press ; 2000 .
yang w , liu w , via a , tuanmu
m - n , he g , dietz t , et al .
nonlinear effects of group size on collective action and resource outcomes .
proc natl acad sci usa .
2013 ; 110 : 10916 - 10921 .
je , macneil ma , basurto x , gelcich s. looking beyond the fisheries crisis : cumulative learning from small - scale fisheries through diagnostic approaches . global environ change .
kenna r , berche b. critical mass and the dependency of research quality on group size . scientometrics .
2011 ; 86 : 527 - 540 .
kenna r , berche b. critical masses for academic research groups and consequences for higher education research policy and management .
high educ manag pol .
2011 ; 23 : 9 - 29 .
kenna r , berche b. managing research quality : critical mass and academic research group size .
j manag math . 2012 ; 23 : 195 - 207 .
barcelo h , capraro v. group size effect on cooperation in one- shot social dilemmas .
2015 ; 5 : 7937 .
diekmann a. volunteer s dilemma .
j confl resolut .
1985 ; 29 : 605 - 610 .
szolnoki a , perc m. impact of critical mass on the evolution of cooperation in spatial public goods games .
2010 ; 81 : 057101 .
marwell g , oliver p. the critical mass in collective action : a micro - social theory .
cambridge , england : cambridge university press ; 1993 .
heckathorn dd .
the dynamics and dilemmas of collective action .
am soc rev .
1996 ; 6 : 250 - 277 .
paolacci g , chandler j ipeirotis pg .
running experiments on amazon mechanical turk .
judgm decis mak .
2010 ; 5 : 411 - 419 .
horton jj , rand dg , zeckhauser rj .
the online laboratory : conducting experiments in a real labor market .
2011 ; 14 : 399 - 425 .
mason w , suri s. conducting behavioral research on amazon s mechanical turk .
behav res meth .
2012 ; 44 : 1 - 23 .
charness g , rabin m. understanding social preferences with simple tests .
q j econ . 2002 ; 117 : 817 - 869 .
capraro v , venanzi m , polukarov m , jennings nr .
cooperative equilibria in iterated social dilemmas . in : proceedings of the 6th international symposium on algorithmic game theory .
lecture notes in computer science ; 2013 .
146 - 158 . | in a world in which many pressing global issues require large scale cooperation , understanding the group size effect on cooperative behavior is a topic of central importance . yet
, the nature of this effect remains largely unknown , with lab experiments insisting that it is either positive or negative or null , and field experiments suggesting that it is instead curvilinear .
here we shed light on this apparent contradiction by considering a novel class of public goods games inspired to the realistic scenario in which the natural output limits of the public good imply that the benefit of cooperation increases fast for early contributions and then decelerates .
we report on a large lab experiment providing evidence that , in this case , group size has a curvilinear effect on cooperation , according to which intermediate - size groups cooperate more than smaller groups and more than larger groups . in doing so ,
our findings help fill the gap between lab experiments and field experiments and suggest concrete ways to promote large scale cooperation among people . |
control of the double dot using dc gate voltages is demonstrated in fig . 1 ( b d ) .
figure 1(b ) shows @xmath20/@xmath28 ( numerically differentiated ) as a function of @xmath12 and @xmath11 . when an electron enters or leaves the double dot , or moves from one dot to the other , the qpc conductance changes .
gate voltage derivatives of @xmath18 and @xmath19 clearly show these changes and map out the double dot charge stability diagram @xcite .
the nearly horizontal lines are due to charge transitions in the left dot , while the nearly vertical lines correspond to charge transitions in the right dot .
for very negative values of @xmath11 and @xmath12 ( see the lower left corner of the charge stability diagram ) charge transitions no longer occur , indicating that the double dot is completely empty , denoted ( 0,0 ) .
transport through the double dot can be correlated with simultaneous charge sensing measurements .
figure 1(c ) shows a color scale plot of @xmath29 near the ( 1,0 ) to ( 0,1 ) charge transition . a charge stability diagram , simultaneously acquired , is shown in fig . 1(d ) . near the ( 1,0 ) to ( 0,1 ) charge transition the system behaves as an effective two - level system .
crossing this transition by making @xmath11 more positive transfers a single electron from the right dot to the left dot .
near the ( 0,1 ) to ( 1,0 ) interdot transition , the double dot forms a two - level charge system that can be characterized by the detuning parameter , @xmath30 , and the tunnel coupling , @xmath31 ( see inset of fig .
2(b ) ) @xcite .
we have used microwave spectroscopy to characterize this two - level system @xcite .
microwaves drive transitions in the double dot when the photon frequency is equal to the energy separation between the ( 1,0 ) and ( 0,1 ) charge states @xcite .
this microwave - induced charge state repopulation can be directly measured using the qpc charge sensors @xcite . the black curve in fig .
2(a ) shows the measured charge on the left dot , @xmath25 , as a function of @xmath30 , in the absence of microwave excitation . as expected , increasing @xmath30 transfers a single charge from the left dot to the right dot .
application of microwaves to gate a results in resonant peaks in @xmath25 vs. @xmath30 that move to larger @xmath32 with increasing frequency .
this resonant peak corresponds to a single photon process that drives an electron from the ( 1,0 ) ground state ( for negative @xmath30 ) into the ( 0,1 ) excited state , or vice versa .
the frequency dependence of the resonance condition can be used to map out the energetics of the charge two - level system .
detailed measurements of the resonant peak position as a function of microwave frequency , @xmath33 , are used to extract @xmath31 for various @xmath13 ( see fig .
2(b ) ) @xcite . at high frequencies the peak positions move linearly with @xmath33 . for small frequencies , probing the region near the (
0,1)-(1,0 ) charge transition , the interdot tunnel coupling modifies the linear dependence .
changing the interdot tunnel coupling modifies the frequency dependence of the resonant peak position . for each value of @xmath13 ,
the experimental data have been fit using @xmath34= @xmath35 , where @xmath36 is the lever arm . @xmath36 and
@xmath31 were used as free parameters for each curve . the lever arm
@xmath36 changes by @xmath120@xmath37 over the range of @xmath13 used in fig . 2 .
the experimental data are well fit by theory and show that the tunnel coupling varies by roughly a factor of 6 when @xmath13 is changed by 70 mv .
measurements of @xmath31 from microwave spectroscopy are consistent with values obtained by measuring the width of the interdot charge transition using charge sensing @xcite .
, as a function of @xmath30 for several microwave frequencies .
( b ) one - half of the resonance peak splitting as a function of @xmath33 for several values of @xmath13 .
solid lines are best fits to the experimental data using the theory outlined in the text .
inset : two - level system energy level diagram .
( c ) amplitude of the resonance , expressed as @xmath38(@xmath39)/@xmath38(@xmath39=5 ns ) , as a function of chopped cw period , @xmath39 , with @xmath33=19 ghz .
theory gives a best fit @xmath0=16 ns ( solid line , see text ) .
inset : single photon peak shown in a plot of @xmath25 as a function of @xmath30 for @xmath39=5 ns and 1 @xmath40s .
( d ) power dependence of the resonance for @xmath33=24 ghz .
widths are used to extract the ensemble - averaged charge dephasing time @xmath41 . at higher microwave powers
multiple photon processes occur .
curves are offset by 0.3 for clarity . ] charge relaxation and decoherence times can be extracted by analyzing the resonant response of the two - level system , as used in the analysis of the cooper pair box @xcite .
the charge relaxation time @xmath0 is determined by measuring the resonance peak height as microwaves are chopped at varying periods , @xmath39 , with a 50% duty cycle @xcite .
the system response is modelled with a saturated signal while microwaves are present , followed by an exponential decay with a characteristic time scale @xmath0 when the microwaves are turned off . calculating the time averaged occupation , we expect : @xmath42 with long periods ( @xmath39@xmath43@xmath0 ) , the exponential tail due to the finite relaxation time is an insignificant part of the duty cycle , and the charge detector measures the time average of the on / off signal , giving a resonant feature with half the height found in the limit @xmath39@xmath440
when the period is very short , such that @xmath39@xmath45@xmath0 , the charge has little time to relax , and the charge detector response is close to saturation ( saturation is defined as @xmath38=0.5 on resonance ) . in the intermediate regime where @xmath39@xmath1@xmath0 , the qpc signal is strongly dependent on @xmath39 .
we present data for @xmath465 ns to avoid artifacts due to the finite rise time of the mixer circuit . in fig .
2(c ) , we plot @xmath38(@xmath39)/@xmath38(@xmath39=5 ns ) as a function of @xmath39 . agreement between experiment and theory is good and gives a best fit @xmath0=16 ns .
the resonance peak width gives a direct measure of the inhomogeneous charge decoherence time , @xmath41 @xcite . in fig .
2(d ) we plot @xmath26 as a function of @xmath30 for several microwave powers . at low power , only the single - photon ( 1@xmath47 ) peak is visible . as the power
is increased the 1@xmath47 peak approaches saturation and a two - photon peak develops @xcite .
a fit to the low power 1@xmath47 peak using a gaussian function is shown in red in fig .
2(d ) . the best fit half - width of 0.077 mv corresponds to an energy of 10.2 @xmath40ev when taking into account the lever arm .
converting this into a time results in a lower bound @xmath41=400 ps .
this measurement of @xmath41 gives a worst - case estimate since charge fluctuations will broaden the resonant feature , resulting in a shorter @xmath41 value .
our measurements of @xmath48 and @xmath41 using charge sensing can be compared with other recent experiments @xcite . in a pulsed - gate experiment , fujisawa _
_ @xcite have measured the energy relaxation time in a vertical quantum dot . from a measurement of the transient current as a function of pulse time
they extract @xmath0=10 ns , which is limited by spontaneous emission of a phonon .
direct observation of coherent charge oscillations has been reported by hayashi _ _ et al.__@xcite . from the decay envelope of the rabi oscillations
hayashi _ et al .
_ extract a @xmath49 time of @xmath11 ns , which serves as an upper bound estimate for @xmath50 .
the @xmath48 and @xmath41 values that we obtain from charge sensing are in good agreement with the results of these previous experiments .
spin physics can be studied in the one - electron regime at high fields , where the spin - up and spin - down states are separated by the zeeman splitting , or in the two - electron regime with singlet and triplet spin states .
we focus on the two - electron regime , where differences in the singlet - triplet splittings in the ( 1,1 ) and ( 0,2 ) charge states can be put to use for spin state readout and initialization .
we show that singlet - triplet relaxation times can be measured by implementing a charge pump experiment in the two - electron regime .
this measurement technique can be used to measure the singlet - triplet relaxation time , @xmath51 , for nearly degenerate singlet and triplet states , a regime in which hyperfine mediated relaxation process are expected to be important . in the two - electron regime , charge transport in a double dot shows a striking asymmetry in bias voltage due to spin selection rules ( pauli blocking ) @xcite . the asymmetry in charge transport is due to the large difference in the singlet - triplet splittings for the ( 1,1 ) and ( 0,2 ) charge states . in the weakly - coupled ( 1,1 ) charge configuration
the singlet and triplet states are nearly degenerate .
however , two tightly confined electrons in the ( 0,2 ) charge state result in a singlet - triplet splitting @xmath52@xmath1400 @xmath40ev . at forward bias ,
transitions from the ( 0,2 ) singlet state , ( 0,2)@xmath53 , to the ( 1,1 ) singlet state , ( 1,1)@xmath53 , are allowed . however , reverse bias ( 1,1 ) to ( 0,2 ) charge transitions are blocked if the ( 1,1 ) state forms a triplet ( 1,1)@xmath54 because the ( 0,2)@xmath54 state resides outside the transport window due to the large singlet - triplet splitting in ( 0,2 ) .
this asymmetry results in current rectification , which can be used for spin - to - charge conversion and spin state readout .
charge transitions are driven by applying pulses to gates l and r. experimental details concerning pulse calibration have been previously published @xcite . in double dots , charge can be pumped by pulsing gates around a triple point , e.g. ( 0,1)@xmath44(1,1)@xmath44(0,2)@xmath44(0,1 ) .
our spin relaxation measurement technique relies on the fact that ( 1,1)@xmath55 to ( 0,2)@xmath56 transitions are spin blocked . measuring this charge transition probability as a function of time using charge sensing allows a measurement of the spin relaxation time .
we demonstrate that the observed time dependence of the charge sensing signal is due to spin blocked transitions .
, as a function of @xmath11 and @xmath12 , measured while applying the forward pulse sequence ( 0,1)@xmath44(1,1)@xmath44(0,2)@xmath44(0,1 ) . the pulse period @xmath39=10 @xmath40s and the perpendicular field @xmath57=100 mt .
we observe a charge sensing signal in the pulse triangle ( bounded by the red lines ) that takes on a value between the raw ( 1,1 ) and ( 0,2 ) signal levels .
this signal is indicative of spin - blocked charge transitions . outside of the pulse triangle ,
transitions through the ( 1,2 ) and ( 0,1 ) charge states are possible and relax the spin blockade .
( b ) energy level diagram illustrating the possible transitions from ( 1,1 ) to ( 0,2 ) .
fast transitions are indicated with black arrows , spin blocked transitions with grey arrows .
( c ) @xmath19 as a function of @xmath11 and @xmath12 while applying the reverse control " pulse sequence,(0,1)@xmath44(0,2)@xmath44(1,1)@xmath44(0,1 ) . the pulse period @xmath39=10 @xmath40s and the perpendicular field @xmath57=100 mt .
the ( 0,2)@xmath53 to ( 1,1)@xmath53 transition is not spin blocked and as a result , there is no detectable pulse signal in the pulse triangle ( bounded by the red lines ) .
( d ) level diagram illustrating the ( 0,2 ) to ( 1,1 ) transition .
a best - fit plane has been subtracted from the data in ( a),(c ) to remove signal from direct gate to qpc coupling . ]
the pulse sequence used to measure @xmath51 is shown in [ fig.3(a ) ] .
the gates are held at point e for 10% of the period , emptying the second electron from the double dot , leaving the ( 0,1 ) charge state .
a pulse shifts the gates to point r ( reset point ) for the next 10% of the period .
this initializes the system into the ( 1,1 ) configuration . since the singlet and triplet states are nearly degenerate in ( 1,1 )
we expect to load into ( 1,1)@xmath53 or any of the three ( 1,1)@xmath54 states with equal probability . for the final 80% of the period ,
the gates are held at the measurement point m where ( 0,2)@xmath53 is the ground state .
the energetics of the spin states at the measurement point are shown in fig .
a ( 1,1)@xmath53 state prepared in the r step will tunnel to ( 0,2)@xmath53 on a timescale set by the interdot tunneling rate , @xmath58 .
the m@xmath59=0 ( 1,1 ) triplet state , ( 1,1)@xmath60 , will dephase into ( 1,1)@xmath53 on a timescale set by @xmath61 ( expected to be @xmath62100 ns @xcite ) followed by a direct transition to ( 0,2)@xmath53 . roughly half of the time the r step will load the m@xmath59=1 ( 1,1 ) triplet state , ( 1,1)@xmath63 , or the m@xmath59=-1 ( 1,1 ) triplet state , ( 1,1)@xmath64 . since ( 0,2)@xmath54 is inaccessible a transition from ( 1,1)@xmath63 or ( 1,1)@xmath64 to ( 0,2 ) requires a spin flip and will be blocked for times shorter than the singlet - triplet relaxation time @xmath51 .
figure 3(a ) shows the time - averaged charge sensor conductance @xmath19 , measured as a function of the dc gate voltages @xmath11 and @xmath12 , while the forward pulse sequence is repeated .
the conductance @xmath19 maps out the ground state population at point m since 80% of the duty cycle is spent there .
the plateaus in @xmath19 at @xmath650.0 , 6.0 , 16 , and 23 @xmath510@xmath66e@xmath67/h indicate full population of the ( 1,2 ) , ( 0,2 ) , ( 1,1 ) , and ( 0,1 ) charge states respectively .
the pulse data differs from ground state data only when point m resides in the triangle defined by the ( 1,1 ) to ( 0,2 ) ground state transition and the extensions of the ( 1,1 ) to ( 0,1 ) and ( 1,1 ) to ( 1,2 ) ground state transitions ( bounded by the red marks in fig .
inside of the pulse triangle " transitions from ( 1,1 ) to ( 0,2 ) may be spin - blocked and the charge sensor registers a conductance intermediate between the ( 1,1 ) and ( 0,2 ) plateaus . outside of the pulse triangle it is possible to access ( 0,1 ) or ( 1,2 ) , which relaxes the spin blockade .
figure 3(a ) shows a signal of 11@xmath510@xmath66e@xmath67/h in the pulse triangle for @xmath39=10 @xmath40s , indicating that approximately 50% of the time the dots remain in ( 1,1 ) even though ( 0,2 ) is the ground state .
this is direct evidence of spin - blocked ( 1,1 ) to ( 0,2 ) transitions . to check that the pulse signal is due to spin - blocked transitions and not just a slow interdot tunnel rate we compare the forward pulse sequence with a reverse pulse sequence that does not involve spin selective transitions [ ( 0,1)@xmath44(0,2)@xmath44(1,1)@xmath44(0,1 ) ]
in the reverse pulse sequence the reset position r occurs in ( 0,2 ) where only the singlet state is accessible , and
m occurs in ( 1,1 ) . now tunneling from r to m
should always proceed on a time scale set by the interdot tunnel coupling , since the ( 0,2)@xmath53 to ( 1,1)@xmath53 transition is not spin blocked .
no signal is seen in the pulse triangle for this reversed control " sequence ( fig .
. measured as a function of @xmath39 . in the ( 0,2 ) charge state @xmath19=10@xmath510@xmath66e@xmath67/h , while @xmath19=20@xmath510@xmath66e@xmath67/h in the ( 1,1 ) charge state . for short @xmath39 ,
the ( 1,1 ) to ( 0,2 ) transition is blocked approximately half of the time , resulting in a pulse signal of 15@xmath510@xmath66e@xmath67/h in the ( 0,2 ) pulse triangle .
the ( 1,1 ) to ( 0,2 ) transition probability increases ( sensor signal decreases ) with @xmath39 due to spin - relaxation with a characteristic time scale of 70@xmath6810 @xmath40s .
a best - fit plane has been subtracted from the data . ] the singlet - triplet relaxation time can be determined by measuring the time dependence of the charge sensing signal inside of the pulse triangle .
we extract @xmath51 by measuring @xmath19 as a function of the pulse train period , @xmath39 .
@xmath19 is measured inside the pulse triangle ( @xmath12,@xmath11 held fixed at -403,-523.8 mv , respectively ) and is plotted as a function of @xmath39 in fig.4 . in ( 1,1 ) , @xmath19@xmath6920@xmath510@xmath66e@xmath67/h , whereas outside the pulse triangle in ( 0,2 ) , @xmath19@xmath7010@xmath510@xmath66e@xmath67/h .
for small @xmath39 , @xmath71@xmath7015@xmath510@xmath66e@xmath67/h in the pulse triangle . at long @xmath39 , @xmath71 approaches 10@xmath510@xmath66e@xmath67/h in the pulse triangle , which indicates complete transfer from the ( 1,1 ) to ( 0,2 ) charge state .
we fit these experimental data assuming exponential singlet - triplet relaxation and find a best fit @xmath51@xmath7270@xmath6810 @xmath40s .
the dependence of the singlet - triplet relaxation time on detuning and magnetic field has been presented in @xcite .
a speed up of spin relaxation near zero field is observed and is consistent with a hyperfine mediated spin relaxation process .
with an external magnetic field applied so that the @xmath73 triplet states are split off , the @xmath74 and @xmath75 states form a logical qubit . due to the large singlet triplet
splitting in ( 0,2 ) we can easily initialize the system in @xmath76 .
@xmath76 can transferred to @xmath74 by sweeping the detuning adiabatically with respect to the interdot tunnel coupling .
the @xmath74 and @xmath76 states are hybridized at zero detuning due to the interdot tunnel coupling .
this hybridization results in an exchange splitting between the @xmath74 and @xmath75 states , @xmath77 . for large negative detunings , @xmath78 .
the exchange @xmath77 can be tuned on ns timescales by applying pulses to the gates defining the double dot . using this fast control of the exchange energy
we have recently implemented a spin swap operation @xcite .
+ + we acknowledge useful discussions with sankar das sarma , hans - andreas engel , xuedong hu , daniel loss , emmanuel rashba , and peter zoller .
funding was provided through the aro under daad55 - 98 - 1 - 0270 and daad19 - 02 - 1 - 0070 , the darpa - quist program , and the nsf under dmr-0072777 , the harvard center for nanoscale systems , and the sloan and packard foundations . | we demonstrate high speed manipulation of a few - electron double quantum dot . in the one - electron regime ,
the double dot forms a charge qubit .
microwaves are used drive transitions between the ( 1,0 ) and ( 0,1 ) charge states of the double dot .
a local quantum point contact charge detector measures the photon - induced change in occupancy of the charge states .
charge detection is used to measure @xmath0@xmath116 ns and also provides a lower bound estimate for @xmath2 of 400 ps for the charge qubit . in the two - electron regime
we use pulsed - gate techniques to measure the singlet - triplet relaxation time for nearly - degenerate spin states .
these experiments demonstrate that the hyperfine interaction leads to fast spin relaxation at low magnetic fields .
finally , we discuss how two - electron spin states can be used to form a logical spin qubit . , , , , ,
, , and , charge qubit , spin qubit , rabi oscillation , coherent manipulation 03.67.mn , 72.25.rb , 85.35.gv semiconducting quantum dots can be used to confine single electrons in an electrically controllable potential @xcite .
coupled quantum dots , containing a single electron , create a tunable two - level system for the manipulation of single charges @xcite . by a similar approach , when two electrons are confined to a double dot the relaxation and dephasing of singlet and triplet spin states can be studied @xcite .
recently , we have demonstrated coherent control of two - electron spin states by using high speed pulsed gate techniques @xcite . in this paper , we review recent experiments performed by our group on few - electron quantum dots that demonstrate quantum control of just one or two electrons @xcite .
samples are fabricated from a gaas / al@xmath3ga@xmath4as heterostructure grown by molecular beam epitaxy ( fig .
1(a ) ) .
electron beam lithography and liftoff techniques are used to fabricate ti / au gates , which deplete the two - dimensional electron gas with electron density 2@xmath510@xmath6@xmath7 and mobility 2@xmath510@xmath8@xmath9/v@xmath10s . when the gates are appropriately biased a double well potential is formed . the electron number in the left(right ) dot is varied by tuning @xmath11(@xmath12 ) .
interdot tunnel coupling is tuned by changing the voltage @xmath13 .
standard lock - in techniques are used to measure the double dot conductance , @xmath14 , and the quantum point contact ( qpc ) conductances , @xmath15 .
the sample is cooled to base temperature in a dilution refrigerator with an electron temperature , @xmath16@xmath1135 mk , as determined from coulomb blockade peak widths .
depending on the experimental arrangement continuous - wave ( cw ) microwaves are applied to gate a , or high speed pulses are applied to gates l and r using bias tees @xcite that are thermally anchored to the mixing chamber .
denotes an ohmic contact .
the double dot conductance , @xmath17 , and the qpc conductances , @xmath18 and @xmath19 , are measured using standard lock - in techniques .
( b ) large scale plot of @xmath20/@xmath21 as a function of @xmath22 and @xmath23 .
charge states are labelled @xmath24@xmath25,@xmath26@xmath27 , where @xmath25@xmath24@xmath26@xmath27 is the time averaged number of electrons on the left ( right ) dot .
@xmath17 , in ( c ) , and @xmath20/@xmath21 , in ( d ) , as a function of @xmath22 and @xmath23 near the ( 1,0 ) to ( 0,1 ) transition .
the gates have been slightly adjusted in ( c d ) relative to ( b ) to allow simultaneous transport and sensing .
identical color - scales are used in ( b ) and ( d ) . ] |
over the past 30 years or so the coherent potential approximation @xcite ( cpa ) has proved to be a generally reliable method for dealing with disordered systems .
@xcite however , being in essence a single - site mean - field theory , @xcite the cpa fails to take into account the effect of non - local potential correlations due to the disorder in the environment of each site and hence leaves much important physics out of consideration .
consequently , considerable effort has been spent in trying to find a way of improving it systematically by a multi - site or cluster generalization .
surprisingly this has turned out to a very difficult problem @xcite and a viable solution has been proposed only recently .
the new method has emerged from the dynamical cluster approximation @xcite ( dca ) which was directed originally at describing dynamical spin and charge fluctuations in simple hubbard models of strongly - correlated electron systems .
recently its static limit has been adapted by jarrell and krishnamurthy for a simple tight - binding model of electrons moving in a disordered potential .
@xcite the same problem was investigated by moradian et al .
@xcite in this paper we develop the ideas behind this approximation further and demonstrate how they can be combined with realistic , ab - initio descriptions of systems of interacting electrons in disordered systems . because the language of our multiple scattering theory is so different from that of the context in which the dca is usually deployed
, we elaborate on this relationship .
firstly , we note that the dca was invented to describe short - range correlations within the framework of the dynamical mean field theory @xcite ( dmft ) of spin and charge fluctuations in many - electron systems .
secondly , we recall that that the dmft can be regarded as the dynamical generalization of the cpa for the hubbard ` alloy analogy ' problem .
@xcite thus , in the light of these remarks , it is natural to investigate the static version of the dca as a generalization of the cpa which includes a description of short - range order .
indeed , jarrell and krishnamurthy @xcite already studied the problem of electronic structure in random alloys from this point of view . in this paper
we tackle the same problem using an identical conceptual framework but a very different description of the electrons afforded by multiple scattering theory .
@xcite to make the above remarks more specific we would like to highlight two of the principal differences between our treatment of the problem and that of ref . .
firstly , we do not make use of a tight - binding model hamiltonian but solve , numerically , a schrdinger equation in each unit cell and match the ` out - going wave ' solution to the incoming waves from all the other unit cells .
this is known as the multiple scattering approach @xcite to the problem of electronic structure in solids and is the foundation of the korringa - kohn and rostoker ( kkr ) band theory method .
consequently , the principal virtue of our formalism , as opposed to those based on tight - binding model hamiltonians is that it prepares the ground for first - principles calculations based on density - functional theories .
@xcite the second difference is a formal consequence of the first . in multiple scattering theories
the object of interest is not the self - energy and the diagrammatic language of perturbation theory is not used .
we will show that the quantities that play the role of the self - energy in multiple scattering theory are the effective scattering amplitudes @xmath0 and effective structure constants @xmath1 which are also the natural concepts in effective medium theories .
@xcite in short , these formal reasons fully account for the fact that we do not base our arguments on ` restoring momentum conservation ' and introducing approximate laue functions to renormalize diagrams but construct our theory in terms of real and reciprocal space clusters .
nevertheless , we believe that our final algorithm described in sec .
[ algorithm ] is equivalent to those investigated by jarrell and krishnamurthy .
@xcite our aim in reformulating the problem is to facilitate the deployment of the method as a first - principles calculation , in other words to develop a non - local kkr - cpa .
@xcite in brief , our kkr - nlcpa method introduces new effective structure constants and this enables us to define an effective medium which includes non - local potential correlations over all length scales . using a ` coarse - graining ' procedure inspired by the dca we can then derive a self - consistent ` cluster generalization ' of the kkr - cpa @xcite which determines an approximation to this effective medium by including non - local correlations up to the range of the cluster size .
the kkr - nlcpa satisfies all of the requirements for a successful cluster generalization of the kkr - cpa as listed by gonis .
@xcite in particular , the kkr - nlcpa becomes exact in the limit of large cluster sizes where it includes non - local correlations over all length scales , and recovers the kkr - cpa for a single - site cluster .
the method is fully causal , allows the effects of short - range order to be modelled , and can be implemented numerically for realistic systems .
the outline of this paper is as follows . in the next section
we describe the formalism for the kkr - nlcpa .
we explain our kkr - nlcpa algorithm and show how to include short - range order .
we describe in more detail how to carry out the coarse - graining with reference to simple cubic , body - centered cubic and face - centered cubic lattices .
finally we explain how to use the kkr - nlcpa formalism to calculate observable quantities such as the configurationally - averaged density of states in preparation for dft calculations . in order to illustrate the improvements over the conventional kkr - cpa , in sec .
[ results ] we present results ( configurationally - averaged density of states ) for the application of the formalism to a one - dimensional model .
however we emphasise that the formalism presented is fully tractable for realistic three - dimensional systems .
for the sake of clarity , we begin by briefly summarising the idea of the conventional kkr - cpa @xcite method .
the path operator equation describing the scattering of an electron in a general array of non - overlapping potentials , centered on site positions @xmath2 , is given by @xmath3 where the underscore denotes a matrix in angular momentum space with the usual angular momentum indices @xmath4 and @xmath5 is the position vector connecting sites @xmath6 and @xmath7 . for spherically symmetric scatterers
the individual t - matrices are diagonal i.e. @xmath8 but the structure constants @xmath9 are not
. we shall be interested in situations where the lattice sites labeled by @xmath6 and @xmath7 form a regular array , an infinite lattice , but the single - site t - matrices vary from site to site in a random fashion .
if we consider the solution of eq .
( 1 ) for all possible disorder configurations and then take its average , we arrive at an averaged path operator @xmath10 which is translationally - invariant . in practice
it is not computationally feasible to average over every possible disorder configuration nor is an equation which determines the averaged path operator @xmath10 readily available in the form of eq . 1 .
under these circumstances a way forward is to follow the strategy of ` effective medium theories ' .
@xcite in the present context a useful ` effective medium ' is that provided by an ordered array of effective scatterers described by the same t - matrix @xmath0 . in the kkr - cpa , @xcite the scattering amplitude describing these effective scatterers
is determined using the self - consistency condition that excess scattering off a single - site impurity embedded in such a medium be zero on the average . as mentioned in the introduction , whilst very successful in many applications , being a single - site approximation the cpa fails to take into account the effect of non - local potential correlations due to the disorder in the environment of each site .
the first step in going beyond the kkr - cpa is to define what we will call the non - local cpa ( nlcpa ) effective medium by the following equation : @xmath11 where a circumflex symbol denotes an nlcpa effective medium quantity .
here we have defined nlcpa effective local t - matrices @xmath0 and a new _ effective propagator _ by @xmath12 this is composed of the usual free - space kkr structure constants @xmath13 which account for the lattice structure plus a _ translationally - invariant _ effective disorder term @xmath14 .
the matrix @xmath15 takes into account , in an averaged manner , the non - local correlations due to the disorder configurations . clearly
a cluster generalization of the kkr - cpa would involve determining the above nlcpa effective medium by requiring excess scattering off an impurity _ cluster _ of real potentials embedded in the effective medium to be zero on the average , and indeed this is the strategy followed in the next section .
however the fact that @xmath15 is a translationally - invariant effective medium quantity means that it is diagonal in @xmath16 , reciprocal space . in the present problem it is important to treat the theory in both real lattice and reciprocal lattice space consistently and crucially
we shall deal with this by a coarse - graining procedure .
the first step is to consider a cluster of sites in the nlcpa effective medium . denoting the sites within the cluster by capital letters
, it can be shown ( see appendix a ) that eq .
( 2 ) can be rewritten as a cluster equation given by @xmath17 with the effective _
cluster t - matrix _ defined as @xmath18 eq .
( 4 ) is simply a re - arrangement of eq .
( 2 ) so that the site matrix elements of all matrices involve the cluster sites only .
the cluster t - matrix @xmath19 describes the scattering within a cluster and @xmath20 takes account of all scatterings outside of the cluster via the effective _ cluster renormalised interactor _
@xcite @xmath21 . nevertheless , since @xmath22 , it is clear that the cluster t - matrix also includes non - local correlations between the cluster sites .
it should be stressed that @xmath21 describes all paths from the cluster site @xmath23 to the cluster site @xmath24 which only involve intermediate sites outside of the cluster .
it is independent of the contents of the cluster and can be viewed as describing the effective medium from which the cluster has been removed i.e. replaced by a _
cavity_. we may now define an ` impurity cluster ' embedded in the effective medium simply by filling up this cavity with a particular configuration of site potentials .
clearly , the path operator for sites @xmath25 belonging to such an impurity cluster is given by @xmath26 with the impurity cluster t - matrix defined by @xmath27 for a cluster containing @xmath28 sites each scattering according to @xmath29 or @xmath30 , there are @xmath31 possible impurity cluster configurations .
we are now in a position to generalize the usual cpa self - consistency condition to determine our approximation to the exact configurationally - averaged medium described by @xmath32 , namely the nlcpa effective medium described by @xmath20 .
this follows by considering for each configuration the impurity cluster path operator for paths starting and ending on the impurity cluster sites and demanding that the average over all configurations be equal to the path operator for the nlcpa effective medium itself i.e. @xmath33 the important point to note is that unlike many other self - consistent cluster theories such as the molecular coherent potential approximation @xcite ( mcpa ) and extensions , here @xmath20 and @xmath0 correspond to an effective medium which is invariant under translation from site to site and hence does not yield spurious gaps in the band structure which could affect the calculation of transport properties etc .
@xcite this is a consequence of our definition of the non - local correlation terms @xmath34 as translationally - invariant . to preserve this periodicity ,
during each self - consistency cycle we must have @xmath35 this can be seen by applying the usual lattice fourier transform to eq .
( 2 ) and then considering the cluster sites @xmath25 . clearly , to carry out the above integration numerically we must have a specific representation of the function @xmath36 .
following the central idea of the dca , we proceed by coarse - graining @xmath36 over the brillouin zone . unlike many other attempts ,
@xcite this approach has been shown to be fully causal .
due to the finite size @xmath28 of the cluster in real - space , the differences in distance between the cluster sites correspond to a set of @xmath28 cluster momenta @xmath37 in reciprocal space according to the relation @xmath38 these @xmath37 are at the centers of a set of @xmath28 reciprocal - space patches which coarse - grain the first brillouin zone of the lattice .
earlier it was noted that the effective cluster t - matrix also includes non - local correlation terms @xmath39 acting between the cluster sites .
we use the above relation to coarse - grain @xmath36 as follows : @xmath40 @xmath41 since the nlcpa maps the effective lattice problem to that of a self - consistently embedded impurity cluster problem , the fundamental assumption we make is that provided we are aiming to reproduce these correlations of finite range in the effective lattice , we may take the above coarse - grained values @xmath42 to be a good approximation to @xmath36 for the effective lattice . in short ,
@xmath43 if @xmath44 is in the @xmath45 reciprocal - space patch .
this is discussed more formally in appendix b. the next step is to define the ` coarse - grained averaged ' reciprocal - space matrix elements for the effective medium path operator by @xmath46 where each integral is over the reciprocal - space patch @xmath47 of volume @xmath48 surrounding the point @xmath49 .
this is straightforward since each @xmath42 is constant within its coarse - graining patch . defining transformations relating the real - space and coarse - grained reciprocal - space matrix elements by @xmath50 @xmath51 means that the real - space effective medium path operator for sites @xmath25 within the cluster is now given by @xmath52 which we use to replace eq .
( 9 ) , and we can now iterate to self - consistency until eq . ( 8) and eq . (
16 ) are satisfied .
finally , note that the nlcpa reduces to the cpa for @xmath53 . in this limit the non - local correlation terms @xmath39 vanish and the nlcpa effective t - matrix becomes equal to the usual cpa effective t - matrix . the nlcpa becomes exact as @xmath54 where @xmath55 and non - local correlations over all length scales are treated . for clarity the full kkr - nlcpa algorithm is now summarised below .
all real - space matrices in the algorithm are super - matrices ( denoted by double underscores ) in cluster - site and angular momentum space . for a particular energy @xmath56 , 1 .
make a guess for the effective cluster t - matrix @xmath19 for the first iteration .
do this by placing an average t - matrix ( ata ) , @xmath57 ( where p(a ) is the probability of a site being occupied by an @xmath58 atom ) , on each cluster site , and for the site to site propagation terms in the cluster use the free - space structure constants i.e. set @xmath59 .
2 . calculate @xmath60 using eq .
( 5 ) i.e. @xmath61 where @xmath62 is the diagonal part of @xmath63 . for the first iteration @xmath60 will of course be zero .
convert the matrix elements @xmath60 to coarse - grained reciprocal space using eq .
4 . calculate the coarse - grained matrix elements @xmath64 using eq .
( 13 ) and convert them to real space using eq .
5 . calculate @xmath21 by solving eq .
( 4 ) i.e. @xmath65 6 .
calculate @xmath66 for each impurity cluster configuration using eq .
( 6 ) and average over all @xmath31 configurations to obtain a new effective path operator at the cluster sites @xmath67 .
calculate the new cluster t - matrix @xmath19 by solving eq .
( 4 ) using @xmath67 above and @xmath21 from step 5 i.e. @xmath68 8 . compare the new cluster t - matrix elements @xmath19 with those in step 1 .
if they are not equal to within the desired accuracy , repeat as necessary steps 2 @xmath69 8 using the new cluster t - matrix until convergence within the desired accuracy is achieved .
note that the integrations over the reciprocal space patches in step 4 only involve the inversion of a matrix in angular momentum space and therefore computational time is not significantly increased over the conventional kkr - cpa regardless of cluster size .
this is in contrast to many other attempts such as the mcpa where the integration over the brillouin zone requires the inversion of a super - matrix in cluster - site and angular momentum space for each value of @xmath44 .
we also draw attention to the recent work of maier and jarrell @xcite where the dca algorithm has been shown to converge more quickly than the mcpa algorithm , with corrections of order @xmath70 where @xmath71 is the linear size of the cluster .
since the nlcpa maps the impurity cluster problem to the effective lattice problem in reciprocal space , it is important to realize that the real - space cluster must have periodic boundary conditions i.e. must preserve the translational symmetry of the lattice .
moreover , as explained in ref . , in order to obtain suitable reciprocal space patches centered at the cluster momenta @xmath37 , we must select the real - space cluster sites by surrounding them with a ` tile ' of size @xmath72 ( where @xmath73 is the dimension ) which preserves the full point - group symmetry of the lattice and only clusters which satisfy this requirement are allowed . the method for finding the corresponding cluster momenta @xmath37 satisfying eq .
( 10 ) for a simple two - dimensional square lattice has been described in ref . .
here we generalize this method to the case of three - dimensional simple cubic ( sc ) , body - centered cubic ( bcc ) , and face - centered cubic ( fcc ) lattices commonly found in real disordered alloys .
for the trivial case of @xmath53 , the real - space ` tiles ' with the required symmetry would simply be wigner - seitz cells surrounding each lattice point . for larger cluster sizes
we may take the ` tiles ' to be simple cubes of volume @xmath74 for each of these lattices .
the smallest possible cluster sizes are given by considering @xmath75 ( where @xmath76 is the lattice constant ) and this yields @xmath53 for sc ( trivial ) , @xmath77 for bcc , and @xmath78 for fcc lattices .
the next set of allowed cluster sizes is given by considering @xmath79 and this yields @xmath80 , @xmath81 , and @xmath82 for sc , bcc and fcc lattices respectively .
the next step is to consider the ` principal tiling vectors ' @xmath83 where @xmath84 for integers @xmath85 map out the centers of the real - space ` tiles ' . applying the usual reciprocal space transformations of the form @xmath86 etc .
gives us principal ` coarse - grained ' reciprocal space vectors @xmath87 .
we take @xmath28 non - equivalent vectors ( i.e. do not differ by a reciprocal lattice vector ) of the form @xmath88 for integers @xmath85 to be our set of cluster momenta @xmath37 and these will satisfy eq . (
the reciprocal space patches surrounding these cluster momenta will be simple cubes of equal volume @xmath89 which together will fill out a volume the size of the first brillouin zone i.e. @xmath90 for sc , @xmath91 for bcc , and @xmath92 for fcc lattices respectively .
integrating over these patches is equivalent to integrating over the first brillouin zone of the lattice since @xmath49 values and parts of patches lying outside of the first brillouin zone can be translated through reciprocal lattice vectors ( of the sc , bcc , or fcc lattice as appropriate ) to lie within the first brillouin zone . in tab .
[ table ] we give some examples of sets of @xmath93 and corresponding @xmath49 values obtained using the above method . [ cols="^,^,^,^,^",options="header " , ] the principal advantage of the kkr - nlcpa over the conventional kkr - cpa is that it can be implemented for alloys in which short - range ordering or clustering is present . to deal with this situation
one must include an appropriate weighting for each of the @xmath31 impurity cluster configurations in step 6 of the algorithm .
note that such short - range order will not destroy the translational invariance because it will be restored by the configurational averaging .
however when using some method to weight the configurations it is important to bear in mind the periodic boundary conditions imposed on the cluster as explained in the previous section . to illustrate the above feature of the theory we will describe short - range order with a one - dimensional example later in this paper . in order to calculate observables such as the configurationally - averaged densities and density of states , we need to know the configurationally - averaged green s function calculated within the nlcpa .
the expression for the green s function before averaging is given by @xmath94 where @xmath95 is an angular momentum index and @xmath96 lies within the unit cell centered at site @xmath97 .
@xmath98 and @xmath99 are the regular and irregular solutions @xcite respectively of the single - site problem at site @xmath6 . in the next section
we show calculations for the configurationally - averaged density of states for a one - dimensional model and so here we demonstrate explicitly how to take the configurational average of the site - diagonal greens function .
it should be stressed that unlike in calculations based on tight - binding models , in the present multiple scattering theory we solve for and describe the site to site fluctuations of the ` orbitals ' @xmath98 . in short
we can calculate the density @xmath100 point by @xmath100 point .
it is this feature of the present theory which requires the following careful averaging procedure .
the first step is to consider @xmath100 and @xmath101 in the neighborhood of a cluster site @xmath23 .
denoting a configuration of the remaining cluster sites by @xmath102 , as a generalization of ref .
we first average over the subset of possible lattice structures that leave the potential in site @xmath23 fixed : @xmath103 where @xmath104 is the path operator for paths starting and ending at site @xmath23 conditionally averaged so that the potential on site @xmath23 is known ( to either be of type @xmath58 or @xmath105 ) and the configuration of the remaining sites in the cluster is known to be @xmath102 .
@xmath106 is the probability that the configuration @xmath102 of the remaining cluster sites occurs given the type of potential at site @xmath23 .
the final step is to average over the possible occupants of site @xmath23 itself : @xmath107 \nonumber \\
-\sum_l\left[p(a)z_l^a(e,\mathbf{r}_i)j_l^a(e,\mathbf{r}'_i)+p(b)z_l^b(e,\mathbf{r}_i)j_l^b(e,\mathbf{r}'_i)\right ] \end{aligned}\ ] ] where @xmath108 and @xmath109 are the probablities that site @xmath23 is an @xmath58 atom or @xmath105 atom respectively ( i.e. the concentrations of @xmath58 and @xmath105 atoms in the material ) .
this can be rewritten as @xmath110 \nonumber \\
-\sum_l\left[p(a)z_l^a(e,\mathbf{r}_i)j_l^a(e,\mathbf{r}'_i)+p(b)z_l^b(e,\mathbf{r}_i)j_l^b(e,\mathbf{r}'_i)\right ] \end{aligned}\ ] ] where @xmath111 and @xmath112 denote cluster configurations with an @xmath58 atom and @xmath105 atom on site @xmath23 respectively .
the above expression is still exact at this stage , however it can be simplified using the nlcpa . in this approximation
, @xmath104 is constructed using an ` impurity ' cluster of configuration @xmath113 embedded in the nlcpa effective medium . by using eq .
( 4 ) to eliminate the cluster renormalised interactor from eq .
( 6 ) , this is given by @xmath114_{ll'}^{ii}\ ] ] where the impurity cluster t - matrix @xmath115 has configuration @xmath113 and the notation implies taking the @xmath116 site and @xmath117 angular momentum element of the super - matrix on the right hand side . it does not matter which cluster site is chosen to be site @xmath23 in all the above formulae as @xmath118 , the resulting approximation to @xmath119 , is a translationally - invariant quantity .
the density of states per site is given by @xmath120 where the integral is over @xmath121 , the volume of site @xmath23 .
the multiple scattering theory in three dimensions envisions three - dimensional potential wells of finite range surrounding the atomic nuclei , the famous muffin - tin potential .
a very useful caricature of this realistic situation can be constructed in one dimension . evidently in this case
the unit cell is a line segment and the potential wells are described by a function @xmath122 of one spatial variable only .
interestingly , one may regard the sign of @xmath123 as an angle and develop an analogue of the three - dimensional angular momentum expansion of the usual kkr theory . since it was first formulated by butler @xcite it has been made good use of by a number of authors , for example see refs . . while it is computationally simple , as one might expect , it is formally identical to kkr in three dimensions .
for example there are two ` angular momentum ' values @xmath124 ( and hence all nlcpa super - matrices in the algorithm have dimension @xmath125 ) and there is an explicit expression for the structure constants . for a detailed description see refs . .
for a cluster of size @xmath28 , the one - dimensional reciprocal space patches are simply defined by the points @xmath126 where @xmath76 is the lattice constant and @xmath127 . as an example , the real space ` one - dimensional tiles ' and reciprocal space patches for a four site @xmath128 cluster are shown in fig .
[ fig1 ] . we have carried out extensive numerical calculations of the configurationally - averaged density of states for a one - dimensional alloy over a wide range of parameters . in all cases
we have found that the kkr - nlcpa systematically improves the density of states with increasing cluster size compared to the conventional kkr - cpa . as a simple illustration we set the lattice constant to be @xmath129 a.u . and potential - well radius to be @xmath130 a.u .
for each of the constituent potentials for @xmath58 and @xmath105 sites which are square potential wells of depth @xmath131 ry and @xmath132 ry respectively .
the concentration of @xmath58 sites is taken to be @xmath133 .
the density of states for the electrons in lattices of purely a sites and purely b sites together with the cpa result for the @xmath134 alloy is shown in fig .
[ fig2 ] . in fig .
[ fig3 ] we show for comparison the results for a four site supercell calculation .
this is obtained by considering an infinite periodic supercell containing four sites of a particular configuration and then averaging over all @xmath135 possible configurations .
the supercell calculation is not the exact result due to the finite size of the supercell but it gives an indication of the type of structure to expect if we are to improve upon the cpa .
a kkr - nlcpa calculation with a cluster size of four @xmath128 is shown in fig .
it is evident that much of the structure missing from the cpa calculation which can be associated with energy bands of particular configurations of the supercell is reproduced here .
states also appear in the band gaps either side of the impurity band centered at @xmath136 ry which are absent in the cpa calculation .
this is because the states near the band edges are the contributions of large clusters of like atoms and these can not be dealt with by a single - site theory such as the cpa . to investigate this further
, an eight site @xmath137 calculation is shown in fig .
[ fig5 ] . clearly with increasing cluster size more and more states
enter the band gaps .
next we illustrate the ability of the kkr - nlcpa to take into account the effects of short - range order .
as mentioned earlier , the kkr - nlcpa can be implemented for arbitrary ensembles including those in which the occupancy of a site by an @xmath58 or @xmath105 atom is correlated to that of neighbouring sites . as a simple example we increase the probablity of an atom occupying a site by a factor @xmath138
if it follows a like atom and decrease its probability by the same factor if it follows an unlike atom in the one - dimensional cluster .
thus positive values of @xmath138 correspond to short - range clustering and negative values of @xmath138 correspond to short - range ordering . as an example , for a four site impurity cluster of configuration @xmath139 we have @xmath140 where we have made use of the periodic boundary conditions imposed on the cluster . in fig .
[ fig6 ] we show a four site @xmath128 kkr - nlcpa calculation for the density of states using the same parameters as before along with short - range order parameter values @xmath141 , @xmath142 and @xmath143 .
peaks which increase or decrease can be identified with specific cluster configurations and the increases or decreases in the amplitude of the peaks are consistent with the increased or decreased cluster probabilities .
we have presented the formalism for the kkr - nlcpa method which systematically improves upon the conventional kkr - cpa for describing disordered systems on the basis of a first - principles description of the crystal potential .
we have demonstrated its use on a one - dimensional model and illustrated in detail the necessary coarse - graining procedure for real three - dimensional lattices .
we have also shown how to calculate observable quantities with a view to combining the kkr - nlcpa with density functional theory . in the explicit calculations the emphasis was on the improved structure in the density of states with increasing cluster size due to non - local correlations , and a simple example of the ability of the kkr - nlpca to model the effects of compositional short - range order . in order for disordered systems to receive a ` first - principles ' description ,
electronic density functional theory @xcite ( dft ) needs to be combined with treatments of disorder .
the self - consistent - field korringa - kohn - rostocker coherent potential approximation ( scf - kkr - cpa ) @xcite approach is a ` first - pass ' way at doing this and has been applied to a wide range of disordered alloys .
@xcite it has also been adapted for the problem of itinerant magnets at finite temperatures whose ` disordered local moment ' spin fluctuations are handled using the cpa .
@xcite in principle a dft calculation of a disordered system should mean that separate scf calculations are carried out to minimise the total energy for each disorder configuration individually and then an average taken over all disorder configurations .
this is , of course , intractable and the strategy behind scf - kkr - cpa calculations has been to minimise a functional for the averaged energy in terms of partially averaged charge ( and spin ) densities , i.e. the average of charge densities arising from all configurations which have either an a or a b atom on one site . from this approach , however , it is not straightforward to include local environment effects such as electrostatic , ` charge - correlation ' @xcite madelung and local lattice displacement effects in alloys or indeed short - range order effects in general . for disordered alloys ,
the implementation requires the solution of the kohn - sham equations for electrons moving in the disordered arrays of the effective potentials associated with a and b sites which are insensitive to their environments . in the context of finite temperature magnetism
it means that the thermally induced spin fluctuations must be characterised as ` local moments ' associated with single sites and not larger clusters . in this paper
we have taken the first steps in showing how these serious omissions can be rectified by presenting a scheme within the kkr framework which goes systematically beyond the cpa .
consider a finite - sized cluster @xmath144 of sites in the effective medium .
( 2 ) for sites @xmath145 belonging to the cluster can be written in the form @xmath146 where the sum over all sites @xmath147 has been split into those involving sites @xmath147 within the cluster and sites @xmath147 outside of the cluster .
it can be shown @xcite that @xmath148 with the effective cluster renormalised interactor @xmath149 given by the locator expansion @xmath150 inserting ( a2 ) into ( a1 ) and using the notation that cluster sites are denoted by capital letters gives @xmath151 which can be re - arranged as @xmath152 to include the effective cluster t - matrix given by eq .
( 5 ) .
firstly we note that the correlations between the cluster sites which we are aiming to reproduce in the effective lattice are of a finite range , say @xmath153 , where @xmath71 is the linear size of the ` tile ' surrounding the cluster sites ( see sec .
[ kpoints ] ) .
the next step is to divide or ` coarse - grain ' the first brillouin zone of the lattice of @xmath154 sites into @xmath155 patches of size @xmath156 ( where @xmath73 is the dimension ) centered at the cluster momenta @xmath37 .
we now consider each coarse - grained value @xmath42 to be the average of @xmath36 over the @xmath157 lattice momenta @xmath158 within the patch surrounding the point @xmath49 : @xmath159 the fundamental assumption we make is to take these coarse - grained values @xmath42 to be a good approximation to @xmath36 for the effective lattice .
@xcite this is because according to nyquist s sampling theorem @xcite in order to reproduce correlations of a finite range @xmath160 in real space , we only need to sample the first brillouin zone of the lattice at intervals of @xmath161 i.e. at the cluster momenta @xmath37 .
the real space correlation terms @xmath162 are cut off if the distance between @xmath6 and @xmath7 is outside the range of the cluster size . as @xmath54 correlations over all length scales are treated since the maximum correlation length is proportional to the cluster size @xmath28 . | we introduce the korringa - kohn - rostocker non - local coherent potential approximation ( kkr - nlcpa ) for describing the electronic structure of disordered systems .
the kkr - nlcpa systematically provides a hierarchy of improvements upon the widely used kkr - cpa approach and includes non - local correlations in the disorder configurations by means of a self - consistently embedded cluster .
the kkr - nlcpa method satisfies all of the requirements for a successful cluster generalization of the kkr - cpa ; it remains fully causal , becomes exact in the limit of large cluster sizes , reduces to the kkr - cpa for a single - site cluster , is straightforward to implement numerically , and enables the effects of short - range order upon the electronic structure to be investigated .
in particular , it is suitable for combination with electronic density functional theory to give an ab - initio description of disordered systems .
future applications to charge correlation and lattice displacement effects in alloys and spin fluctuations in magnets amongst others are very promising .
we illustrate the method by application to a simple one - dimensional model . |
semiclassical properties of systems with mixed classical dynamics is a reach rapidly developing field of research .
one of interesting results obtained in this direction is a chaos assisted tunneling .
it was shown that the structure of the classical phase space of hamiltonian systems can influence such purely quantum processes as the tunneling @xcite .
it was demonstrated in numerical simulations that existence of chaotic motion region in the classical phase space of the system can increase or decrease tunneling rate by several orders of magnitude @xcite .
typically one considers tunneling between kam - tori embedded into the `` chaotic sea '' .
the region of chaotic motion affects tunneling rate because compared to direct tunneling between tori it is easier for the system to penetrate primarily into the chaotic region , to travel then along some classically allowed path and to tunnel finally to another kam - torus @xcite .
chaos assisted tunneling phenomenon as well as the closely related coherent destruction of tunneling were experimentally observed in a number of real physical systems .
the observation of the chaos assisted tunneling between whispering gallery - type modes of microwave cavity having the form of the annular billiard was reported in the ref .
the study of the dynamical tunneling in the samples of cold cesium atoms placed in an amplitude - modulated standing wave of light provided evidences for chaos - assisted ( three - state ) tunneling as well @xcite .
recently , the coherent destruction of tunneling was visualized in the system of two coupled periodically curved optical waveguides @xcite .
the most popular methods which are used to investigate the chaos assisted tunneling are numerical methods based on floquet theory @xcite . among other approaches to chaos - assisted tunneling we would like to mention the path integral approach for billiard systems @xcite and quantum mechanical amplitudes in complex configuration space @xcite . in this paper
we will consider the original approach based on instanton technique , which was proposed in @xcite and numerically tested in @xcite .
instanton is a universal term to describe quantum transition between two topologically distinct vacuum states of quantum system . in classical theory
the system can not penetrate potential or dynamical barrier , but in quantum theory such transitions may occur due to tunneling effect .
it is known that tunneling processes can be described semiclassically using path integrals in euclidean ( imaginary ) time . in this case
instantons are soliton - like solutions of the euclidean equations of motion with a finite action .
for example , in euclidean yang - mills theory distinct vacuum states are the states of different chern - symons classes , and instanton solutions emerge due to topologically nontrivial boundary conditions at infinity . in this paper
we will use a much simpler system to investigate the connection between the tunneling , instantons and chaos , namely the kicked system with double well potential .
our attention will be focused on the behavior of the quasienergy spectrum when perturbation of the kicked type is added to the system .
hamiltonian of the particle in the double - well potential can be written in the following form : @xmath0 where @xmath1 - mass of the particle , @xmath2 - parameters of the potential .
we consider the perturbation @xmath3 where @xmath4 - value of the perturbation , @xmath5 - period of the perturbation , @xmath6 - time .
full hamiltonian of the system is the following : @xmath7 euclidean equations of motion of the particle in the double - well potential have a solution - instanton . in phase space of nonperturbed system instanton solution lies on the separatrix .
perturbation destroys the separatrix forming stochastic layer . in this layer
a number of chaotic instantons appears .
chaotic instanton can be written in the following form : @xmath8 it is a solution of the euclidean equations of motion . here
@xmath9 and @xmath10 - chaotic and nonperturbed instanton solutions , respectively , @xmath11 - stochastic correction .
it is convenient to work in the action - angle variables . using standart technique @xcite
we obtain expressions for this variables in the following form : @xmath12 where @xmath13 and @xmath14 - action and angle variables , @xmath15 - energy of the particle , @xmath16 and @xmath17 - full elliptic integrals of the first and second kinds , respectively , @xmath18 - elliptic integral of the first kind , where @xmath19 is integral argument , @xmath20 - modulus which is in the following way expresses through energy @xmath15
@xmath21 we will use in our analytical estimations expression for the action of the chaotic instanton .
we assume this action is equal to the nonperturbed instanton action corresponding to the euclidean energy which is less than one of the nonperturbed instanton solution .
we expand expression ( [ action ] ) in powers of the ( euclidean ) energy difference from the separatrix @xmath22 , where @xmath15 - energy , @xmath23 - energy on separatrix , and neglect terms higher than linear one . as a result we has the following linear expression @xmath24 = \pi j(e_{sep } - \xi ) = s[x_{inst}(\tau ,
0 ) ] - \alpha \ , \sqrt{\frac{m}{a_2 } } \ ; \xi,\ ] ] where @xmath25 = 2 \sqrt{m } \ , a^{3/2}_2 /(3 \ , a_0)$ ] - nonperturbed instanton action , @xmath26 - numerical coefficient .
when stochastic layer in euclidean phase space is formed under the action of the perturbation a number of chaotic instanton solutions appear , see fig .
[ fig : strob0p01 ] ( b ) .
these solutions implement the saddle points of euclidean action functional and thus give significant contribution to the tunneling amplitude . in
the some sense their role is analogous to the role of the multiinstanton configurations in the case of the nonperturbed system . in the framework of the chaotic instanton approach the width of the stochastic layer determines the contribution of these chaotic solutions of euclidean equations of motion to the tunneling amplitude .
the stochastic layer width estimated using resonances overlap criterion .
for this purpose we should expand the expression for the perturbation ( [ eq : v ] ) in the series of exponents .
at first we evaluate the coordinate @xmath27 of the particle using expression ( [ angle ] ) in the following way : @xmath28 and rewrite it as a series of sines @xmath29,\ ] ] where @xmath30 $ ] - parameter ( @xmath31 ) , @xmath32 - the frequency of the particle oscillations . leaving only the first term in expression ( [ eq : x ] ) and performing simplifications of this relation
we obtain the following form for the coordinate @xmath33.\ ] ] using this relation and performing the fourier transformation of the series of delta functions in the perturbation ( [ eq : v ] ) one can obtain the following equations of motion in action - angle variables @xmath34,\\ \dot{\theta } = \omega(j),\label{dott}\end{aligned}\ ] ] where @xmath14 - angle variable for the particle in the system , @xmath35 is a perturbation frequency . to obtain width of stochastic layer
we calculated the parameter of the resonances overlap .
it is equal to relation of the resonance width in the frequency scale ( @xmath36 ) and the distance between resonances ( @xmath37 ) .
first one is estimated as a frequency of oscillations of the resonance angle variable @xmath38 .
the result is obtained from the equations ( [ dotj ] ) and ( [ dott ] ) using standart technique ( see chapter 1 and 5 in @xcite ) .
the resonance width is the following @xmath39 where @xmath40 - distance from the separatrix .
the distance between resonances is calculating using the expression for the resonance levels @xmath41 .
thus one can obtain @xmath42 using two last expressions ( [ eq : dw ] ) and ( [ eq : dw ] ) the parameter of the resonances overlap can be written in the following form @xmath43^{1/2 } \gtrsim \left[\frac{\epsilon \ , \nu}{\delta h } \right]^{1/2}.\ ] ] overlap parameter is equal to unity on the boundary of the stochastic layer . using equation ( [ eq : k ] ) we can write the expression for the width of the stochastic layer in the following way @xmath44 where @xmath45 is the energy on the border between stochastic and regular regions , @xmath46 - some numerical parameter which can not be obtained in the framework of the criterion used .
the tunneling amplitude for the perturbed system is a sum of the amplitude in the nonperturbed case and the amplitude of tunneling via chaotic instantons .
the later can be evaluated by integration over action of the tunneling amplitude in nonperturbed system . using expression ( [ eq : s ] ) this integral can be transformed to the integral over the energy difference from zero up to the width of the stochastic layer ( [ eq : dh ] ) : @xmath47 } \ , exp\ , ( - s[x_{chaos}(\tau , \xi)]),\ ] ] where @xmath48 is a normalize factor . to calculate contribution of chaotic instantons we use approximate expression for the chaotic instanton action ( [ eq : s ] ) and listed above assumptions .
integration over @xmath49 gives the contribution of zero modes @xcite . as the result
we get the following expression for the amplitude : @xmath50 where @xmath51 is tunneling amplitude in the nonperturbed system , @xmath52 - a time of the tunneling which is put to infinity at the end .
the last exponential factor in the expression ( [ eq : a ] ) is responsible for the tunneling enhancement in the perturbed system . in the nonperturbed case
the width of the stochastic layer is equal to zero and the expression ( [ eq : a ] ) coincides with the known expression describing the ordinary tunneling . substituting @xmath53 in formula ( [ eq : a ] ) by the expression ( [ eq : dh ] ) and expanding it in powers of the perturbation strength we can write phenomenological formula for the quasienergy splitting @xmath54 where @xmath55 we fix this phenomenological parameter value using the results of numerical simulations . for this purpose we perform the linear fitting of the numerical data for the dependencies on the perturbation strength and take average value of the parameter over these dependencies . as the result we have the _ single _ numerical parameter @xmath56 for our phenomenological formula explaining all these dependencies .
for the computational purposes it is convenient to choose as basis vectors the eigenvectors of harmonic oscillator . in this representation matrix elements of hamiltonian ( [ eq : h ] ) and
the perturbation ( [ eq : v ] ) are real and symmetric .
they have the following forms ( @xmath57 ) : @xmath58,\\ v_{m\ , n } & = \epsilon \ ; \delta_{m + 1 \ ; n } \;\sqrt{\frac g2 } \
; \sqrt{m + 1},\end{aligned}\ ] ] where @xmath59 and @xmath60 , @xmath61 - planck constant which we put equal to @xmath62 , @xmath32 - frequency of harmonic oscillator which is arbitrary , and so may be adjusted to optimize the computation .
we used the value @xmath63 with parameters @xmath64 @xmath65 @xmath66 . in numerical simulations size of matrices
was chosen to be equal to @xmath67 .
simulations with larger matrices give the same results .
we calculate eigenvalues of the one period evolution operator @xmath68 and obtain quasienergy levels which are related with the evolution operator eigenvalues through the expression @xmath69 .
then we get the ten levels with the lowest one period average energy which is calculated using the formula @xmath70 ( @xmath71 are the eigenvectors of the one period evolution operator ) .
the dependence of quasienergies of this ten levels on the strength of the perturbation is shown in the figure [ fig : qes ] .
quasienergies of two levels with the minimal average energy ( thick lines in the figure [ fig : qes ] ) has a linear dependence on the strength of the perturbation in the considered region .
they are strongly influenced by the perturbation while some of the quasienergy states are not .
performed numerical calculations give the dependence of the quasienergy splitting on the strength ( fig.[fig : de](a ) ) and the frequency ( fig.[fig : de](b ) ) of the perturbation .
these dependencies are linear as it predicted by the obtained analytical formula ( [ eq : de ] ) .
using least square technique we calculate the _ single _ numerical parameter @xmath56 which describes all numerical dependencies demonstrated in the figures [ fig : de](a ) and [ fig : de](b ) .
relative error in determining of the parameter @xmath56 from numerical results is less than @xmath72 .
analytical results are plotted in the figures [ fig : de ] ( a ) and ( b ) by lines .
numerical points lie close to this lines .
the agreement between numerical simulations and analytical expression is good in the parametric region considered .
double - well system is investigated in presence of external kick perturbation .
analytic chaotic instanton approach is applied for this system in order to obtain the phenomenological formula for the ground quasienergy splitting .
the formula has the single numerical parameter which is determined from the numerical results .
this formula describes ground quasienergy splitting as a function of strength and frequency of perturbation .
it predicts linear dependence of the ground quasienergy splitting on these parameters .
numerical results for the quasienergy splitting as a function of the perturbation frequency and strength demonstrate linear dependence as well .
they are in a good agreement with the analytical formula .
analytical determination of the parameter @xmath56 at the moment seems to be impossible due to estimative character of the analytical method used .
however , it is not a large problem since this parameter can be evaluated with the high accuracy using the single dependence of the ground quasienergy splitting for any particular system parameters values .
parameter @xmath56 determined in such a way allows to describe correctly all other dependencies of the ground quasienergy splitting on the perturbation strength and frequency for any other values of the system parameters .
the only restriction is that the perturbation strength has to be sufficiently small ( in the case considered @xmath73 ) in order the chaotic instanton approach used to be valid . | kicked double - well system is investigated both analytically and numerically .
phenomenological formula for ground quasienergy splitting is obtained using resonances overlap criterion in the framework of chaotic instanton approach .
results of numerical calculations of quasienergy spectrum are in good agreement with the phenomenological formula . |
quantum key distribution is the art of distilling a secret key between two distant parties , alice and bob , who have access to an untrusted quantum channel @xcite . in this scenario ,
one typically assumes that the equipment in alice and bob s labs can be trusted , and moreover , that its behavior is accurately described by a given theoretical model .
unfortunately , this often turns out to be a very strong assumption which is not justified in practice @xcite .
in particular , many loopholes can be exploited by an eavesdropper to get around the usual security proofs : for instance , the state preparation might be imperfect @xcite , or the eavesdropper might perform a blinding attack to take control of the detectors at a distance @xcite . one way around such problems consists in exhaustively listing all the potential mismatches between the theoretical model and the real implementation and taking care of each one of them individually .
however , this approach is dubious as it is impossible to be sure that all loopholes have really been addressed .
another , more promising , approach is inspired by the recent framework of device - independent quantum information processing @xcite . here
, the idea is that if alice and bob are able to experimentally violate a bell inequality @xcite , it means that their data exhibit intrinsic randomness as well as secrecy @xcite , independently of the internal operation of the devices @xcite . in the recent years
, this framework has been used to prove the security of device - independent key distribution @xcite , to certify randomness expansion @xcite , self - testing of quantum computers @xcite and states @xcite , and guarantee the presence of entanglement @xcite . in the present work ,
we focus on the cryptographic task of key distribution , which has been the subject of many very recent developments . until recently
, security proofs were restricted to scenarios where alice and bob have access to a pair of memoryless devices or @xmath0 independent pairs of devices , thus ensuring that the measurements inside their own labs were causally disconnected @xcite or commuting @xcite .
this is reminiscent of the notion of collective attacks in standard qkd , where some independence assumption is required . ideally , one would like a protocol where only one device is required per party , and for which no assumption is needed for the device .
this is indeed the motivation for doing device - independent cryptography in the first place .
recent works have been able to get rid of this assumption . in ref .
@xcite , the authors introduced a protocol based on the chained bell inequality @xcite and established its security against arbitrary adversaries .
the protocol , however , only produces a single secret bit and does not tolerate any noise . in ref .
@xcite , the authors proved a strong converse of tsirelson s optimality result for the clauser - horne - shimony - holt ( chsh ) game , based on the chsh inequality @xcite : the only way using quantum resources to win the game as predicted by tsirelson s bound is to use a strategy close to the optimal one for independent and identically distributed states , that is , applying the optimal measurements on copies of a two - qubit maximally entangled state .
this theorem provides a security proof for diqkd based on the chsh inequality .
unfortunately , the security proof does not resistant any constant amount of noise .
while this work was completed , vazirani and vidick gave a universally composable security proof of diqkd against arbitrary attacks @xcite .
their protocol , based again on the chsh inequality , is both reasonably efficient ( the key length scales linearly with the number of measurements ) and tolerant to a constant fraction of noise .
a drawback , however , is that the maximum amount of noise tolerated is of the order of @xmath1 , significantly lower than the bounds obtained for protocols using @xmath0 pairs of devices . in the present paper ,
we present a security proof that ( i ) works for only two devices , that is , does not require commuting measurements or memoryless devices , ( ii ) can be applied to generic diqkd protocols based on arbitrary bell inequalities , ( iii ) has the same efficiency and tolerance to noise than previous proofs using memoryless devices .
all these nice properties , however , come at the price of assuming that the adversary only holds classical information .
while this may seem a strong requirement , it can be easily enforced in any realistic implementation by delaying the reconciliation process , since the best existing quantum memories have very short coherence times @xcite .
another advantage of our general framework is that it can also provide security beyond quantum theory , that is , against eavesdroppers that are only limited by the no - signalling principle .
the outline of the paper is the following .
we first give a brief reminder of the relation between non locality , that is , violation of a bell inequality , and randomness .
we then describe the quantum key distribution protocol and present its secret key rate .
we prove the security of the protocol under the assumption that the eavesdropper does not have access to a long - term quantum memory .
we conclude by briefly comparing our results with the existing security proofs , and discussing some rather natural follow - up questions .
in the following , we consider a bipartite scenario where alice and bob input random variables @xmath2 and @xmath3 in their respective devices and obtain classical outputs @xmath4 and @xmath5 , respectively .
we denote @xmath6 the sizes of the alphabets of @xmath7 , respectively .
moreover , we denote by @xmath8 the probability of getting the specific results @xmath9 when the inputs are @xmath10 , and @xmath11 the vector with components @xmath8 .
a bell inequality can be written as @xmath12 : = \sum_{a , b , x , y } \beta ( a , b , x , y)\ , p(a , b|x , y ) \leq i_\mathrm{cl}\ , \ ] ] where @xmath13 is the classical upper - bound . to any such bell inequality
, one can associate a bound on the randomness of the output @xmath4 given the input @xmath14 through a function @xmath15 such that @xmath16)\quad \mbox { for all } a\in\lambda_a\ .\ ] ] such a function can be computed using the techniques given in @xcite , as explained in @xcite . without loss of generality , this function can be assumed to be monotonically non - increasing and such that @xmath17 is convex . for simplicity , we consider the case where there exist an input - independent bound , i.e. a function @xmath18 such that @xmath19 for all @xmath20 .
examples of bell inequalities satisfying this property are : the chsh inequality @xcite , the chained inequality @xcite , and the collins - gisin - linden - massar - popescu ( cglmp ) inequality @xcite . our results , however ,
can easily be generalised to cover the case of input - dependent bounds .
the diqkd protocol that we consider in this paper is very general in the sense that it is compatible with arbitrary bell inequalities , in particular with the various examples of bell inequalities mentioned above .
our protocol consists of four steps : measurements , estimation of the bell violation , error correction and privacy amplification .
we note @xmath0 the number of times each device is used during the protocol
. 1 . * measurements . *
alice and bob respectively generate the random variables @xmath21 with distribution @xmath22 for @xmath23 . if @xmath24 then alice measures round @xmath25 with input @xmath26 obtaining outcome @xmath27 . if @xmath28 then alice generates @xmath29 with uniform distribution @xmath30 and measures round @xmath25 with input @xmath29 obtaining outcome @xmath27
bob does the analog with @xmath31 , input @xmath32 , and outcome @xmath33 .
in other words , events where @xmath34 are used to establish a raw key , while events where @xmath35 are used to test the bell inequality and guarantee that a secret key can indeed be extracted from the raw key .
* estimation . * alice and bob publish @xmath36 for all @xmath25 , and discard the data corresponding to the rounds with @xmath37 .
the data corresponding to the @xmath38 post - selected rounds @xmath39 with @xmath40 is relabeled with the index @xmath41 keeping the time order .
the data corresponding to the rounds of the set @xmath42 is also published and used to estimate the bell - inequality violation .
more specifically , alice and bob can use the public data to compute the following quantity : @xmath43 the data of the rounds not in @xmath44 constitutes the raw key of alice @xmath45 and bob @xmath46 .
* error correction .
* alice and bob publish @xmath47 bits in order to correct bobs errors @xmath48 . for sufficiently large @xmath47 , all errors are corrected @xmath49 with high probability .
note that some of the published bits are used to estimate how many more bits need to be publish for a successful error correction . for large @xmath0 ,
publishing @xmath50 bits is enough .
for more details about the functioning of error correction , we refer to @xcite .
* privacy amplification .
* alice generates and publishes a two - universal @xcite random function @xmath51 which maps @xmath52 to an @xmath53-bit string @xmath54 .
the number @xmath53 depends on the published information as @xmath55 where @xmath56 is the largest integer not bigger than @xmath57 .
alice and bob then compute @xmath58 , obtaining two copies of the secret key .
note that if the adversary holds a quantum memory , but can not keep it for an arbitrary long time , the honest parties should implement the protocol in two steps : ( i ) they receive the quantum systems from the source and perform the measurements , ( ii ) some time @xmath59 later they perform the rest of the protocol involving the public communication for the estimation , error correction , and privacy amplification .
we show security under the assumption that the adversary can not keep a quantum memory for a time @xmath59 . according to current and near - future technology
, this assumption can be enforced by taking @xmath59 of the order of a few minutes @xcite .
to prove security , we will not make any assumption on the behaviour of the devices of alice and bob , except that they do not broadcast information about the inputs and outputs towards the adversary ( a condition without which there is no hope of ever establishing any secret ) .
modulo this requirement , we can even assume for simplicity that the devices have been built by the adversary .
the eavesdropper could in particular hold quantum systems that are entangled with the systems in the users devices .
however , our proof of security only holds under the condition that the eavesdropper can not store this quantum information past the measurement step of the protocol .
after this step , she should thus perform a measurement @xmath60 on his quantum system , which would give him some classical information @xmath61 about the behaviour of alice s and bob s devices .
but since until this point no public communication has been exchanged between alice and bob , we can as well assume that the eavesdropper has performed his measurement before the users received their devices from the source .
the fact that our proof of security holds independently of the behaviour of the devices , then implies that it holds independently of the prior classical information @xmath61 that eve holds on the devices , and we can thus forget @xmath61 in the following . at the end of the protocol
, alice holds the secret key @xmath62 , and eve holds the information published in the estimation step @xmath63 $ ] , in the error correction step @xmath64 , and in the privacy amplification step @xmath51 . note that here we consider the worst case , where all the messages published within the error - correction step are a function @xmath65 of alices raw key @xmath52 .
let @xmath66 be the probability distribution for these random variables .
we say that @xmath62 is an ideal secret key if it is uniformly distributed and uncorrelated with all the rest : @xmath67 note that since @xmath44 and @xmath68 are functions of @xmath69 , so is @xmath53 .
it is unrealistic to expect that a protocol can generate an ideal secret key .
instead , what we demand is that the distribution generated by the above protocol is indistinguishable from an ideal secret key .
it is known that the optimal success probability when discriminating the two distributions is @xcite @xmath70 the main result of this work ( see the theorem below ) is to shows that @xmath71 where @xmath57 is a constant and @xmath72 . for large @xmath0
, the success probability ( [ main result ] ) tends to 1/2 , which makes the optimal discriminating strategy not better than a random guess .
let us now discuss the efficiency of the protocol in the asymptotic limit where @xmath0 tends to infinity . for large @xmath0
one expects , @xmath73 with high probability .
this gives an asymptotic secret key rate of @xmath74 this is the same rate as the one given in @xcite for memoryless devices but with security against full quantum adversaries . in the case of the chsh inequality , @xmath75
, we define @xmath76 and @xmath77 such that @xmath78)$ ] holds against an adversary limited by quantum theory and @xmath79)$ ] holds against an adversary limited by the no - signalling principle .
the specific values of these functions was derived in @xcite : @xmath80 in fig .
[ key ] , we plot the asymptotic secret key rate as a function of the visibility of the state @xmath81 shared by alice and bob .
_ vs _ noise @xmath82 for the chsh protocol and a state @xmath81 , where @xmath83 is maximally entangled .
the upper curve corresponds to a quantum adversary while the lower one considers an adversary only limited by the no - signalling principle . ]
we now proceed with a detailed security proof for the protocol described above . before we present and prove our main result which is an explicit bound on @xmath84 , we need three technical lemmas .
let us introduce a more compact notation @xmath85 for @xmath86 .
variables with super - index @xmath87 represent the chain of variables associated to time steps equal or earlier than @xmath87 , that is @xmath88 .
recall that the information made public in the estimation step is @xmath89 $ ] and that the raw key is @xmath90 .
let @xmath91 and note that @xmath92 and @xmath93 .
* lemma 1 . *
the no - signaling constraints imposed by the causal structure of the protocol imply @xmath94 \right)\ , \ ] ] for all @xmath95 , where @xmath96 : = \frac{1}{m } \sum_{i=1}^m i\ ! \left [ p(a_i , b_i|x_i , y_i , t^{i-1},z^{i-1 } ) \right]\ .\ ] ] note that above , in @xmath97 , the symbols @xmath98 are upper - case while @xmath99 are lower - case , meaning that @xmath97 is the vector with components @xmath100 for all values of @xmath101 but fixed @xmath99 .
_ this proof is based on an argument introduced in @xcite .
a useful observation is that bound ( [ tau i ] ) implies @xmath102)\quad \mbox { for all } a , b , x , y\ .\ ] ] the following chain of equalities and inequalities follows from : bayes rule , no - signaling to the future , bounds ( [ tau i ] ) and ( [ extra ] ) , and the concavity of the function @xmath103 .
@xmath104 \right ) \\ & \leq & \tau^{m}\ ! \left ( \bar i [ t^m , z^{m } ] \right)\end{aligned}\]]@xmath105 * lemma 2 . * the numbers @xmath106 , @xmath68 , @xmath107 are functions of the random variable @xmath108 , and satisfy @xmath109 where @xmath72 .
( here a comment is in order .
actually , @xmath107 is not only a function of @xmath108 but also depends on the global probability distribution @xmath110 .
but we think of this distribution as given , fixed and unknown .
this dependence prevents the straight generalization of the results in this paper to a quantum adversary . )
_ the function @xmath111 satisfies @xmath112 \ = \ \frac{i_{\rm est } [ t^m , z^m]\ , |{\cal e}|}{\pr\{u=1|u = v\}}\ , \ ] ] and @xmath113 = i[p(a_i , b_i|x_i , y_i , t^{i-1 } , z^{i-1})]\ , \ ] ] for all @xmath87 .
consider the sequence of functions of @xmath114 defined by @xmath115\ , \ ] ] for @xmath116 .
the fact that @xmath117 = \alpha_{l-1 } ( t^{l-1 } , z^{l-1})\ ] ] implies that the sequence of random variables @xmath118 is a martingale @xcite with respect to the sequence @xmath119 . also , using the fact that @xmath120 and @xmath121 \geq q^2 $ ] , the differences @xmath122 are bounded for all values of @xmath114
. constraints ( [ martingale ] ) and ( [ bound dif ] ) constitute the premises for azumas inequality @xcite @xmath123 for any @xmath124 . using ( [ eta xi ] ) , ( [ new ] ) and ( [ martingale ] )
we obtain @xmath125 & = & \frac{1}{m } \sum_{i=1}^m i\ ! \left
[ p(a_i , b_i|x_i , y_i , z^{i-1 } , t^{i-1 } ) \right ] \\ & = & \frac{1}{m } \left ( \sum_{i=1}^{m } \eta[t_i , z_i ] - \alpha_{m } ( t^{m } , z^m ) \right ) \\ & = & \frac{1}{m } \left ( \frac{|{\cal e}|\ , i_{\rm est}}{\pr\{u=1|u = v\ } } - \alpha_{m } ( t^{m } , z^m ) \right)\ , \end{aligned}\ ] ] and setting @xmath126 gives ( [ l2 ] ) . @xmath127 * lemma 3 .
* there is a good event @xmath128 with probability @xmath129 such that @xmath130 for all @xmath69 such that @xmath131 .
_ this proof uses a trick introduced in @xcite .
the values of @xmath114 in the set @xmath132 are the good ones , since alice and bob correctly lower - bound @xmath133 ( and hence @xmath53 ) from the values @xmath106 and @xmath134 determined in the estimation step . in the condition defining @xmath135 above , every symbol is a constant except for latexmath:[$\bar i , @xmath114 . note that @xmath107 also depends on the global distribution @xmath137 , which prevents the generalization of this results to the case of quantum adversary .
fortunately , according to lemma 2 , the probability of @xmath135 is large @xmath138 note the abuse of notation @xmath139 .
define the set @xmath140 and note that @xmath141 . using this and @xmath142
we obtain @xmath143 .
recall @xmath144 and note that @xmath145 and @xmath146 . define the set @xmath147 and note that @xmath148 where we have used @xmath149 .
the good event mentioned in the statement of this lemma is @xmath150@xmath151 , and has probability @xmath152 , as in ( [ g ] ) .
we assume @xmath153 , since it is a premise of the lemma . if @xmath154 then @xmath155 .
hence , the non - trivial case happens for @xmath156 , which we assume in what follows .
using bayes rule , the definition of @xmath157 and @xmath158 , lemma 1 , and ( [ g1 ] ) , we obtain @xmath159 \right ) \\ & \leq & 2\ , ( \lambda_a \lambda_b)^{2|{\cal e}|}\ , \tau^m \!\ ! \left ( \frac{|{\cal e}|\ , i_{\rm est}(g , z^m)}{m\pr\{u=1|u = v\ } } -n^{-1/8 } \right)\ , \end{aligned}\ ] ] which shows the lemma .
@xmath105 * theorem . *
the distance between the secret key generated by the protocol and an ideal key is @xmath160 _ proof . _ using definitions ( [ n_k ] ) and ( [ pguess ] ) , lemma 3 , and @xmath161 , we obtain : @xmath162 the symbol @xmath163 denotes the knowledge of @xmath52 with respect to @xmath164 ( see appendix ) when the statistics is conditioned on the events @xmath165 and @xmath128 .
next , we use the identity @xmath166 with the event @xmath128 introduced in lemma 3 . noticing that @xmath167 is a function of @xmath168 , using ( [ hg ] ) , the triangular inequality , and lemma 4
, we see that @xmath169 which concludes the proof .
in this work , we provide a novel security proof for diqkd .
contrary to most of the existing proofs , it applies to the situation in which alice and bob generate the raw key using two devices . in particular
, it does not need to assume that the devices are memoryless or , equivalently , that each raw - key symbol is generated using a different device . while there exist other recent proofs that also work without this assumption , they tolerate zero @xcite or rather small amounts of noise @xcite .
another important feature of our proof is that it can also be applied to non - signalling supra - quantum eavesdroppers .
all these advantages come at the price of making an extra assumption on eve : she does not have access to a long - term quantum memory and , therefore , effectively she can not store quantum information .
while this may at first be considered a strong assumption ( and is actually not needed in new security proofs for diqkd @xcite ) , it is a very realistic assumption taking into account current technology .
the natural open question is to understand how the assumption on the memory can be removed within the framework presented here , or how the other proofs @xcite could be improved to tolerate realistic noise rates . in the case of no - signalling eavesdroppers
, there is some evidence suggesting that the fact that eve can store information and delay her measurement prevents any form of privacy amplification between the honest parties @xcite .
however , the recent results of @xcite imply that privacy amplification is indeed possible against quantum eavesdroppers .
a good understanding of privacy amplification in the device - independent quantum scenario is probably the missing ingredient to get robust and practical fully device - independent security proofs .
we acknowledge useful discussion with serge massar . this work is supported by the erc sg percent , by the eu projects q - essence and qcs , by the chist - era diqip project , by the spanish fis2010 - 14830 projects , by the snf through the national centre of competence in research `` quantum science and technology '' , by catalunyacaixa , by the interuniversity attraction poles photonics@be programme ( belgian science policy ) , by the brussels - capital region through a bb2b grant , and from the frs - fnrs under project diqip .
a random function @xmath170 is two - universal @xcite if @xmath171 for all @xmath172 with @xmath173 .
the following is a simple extension of the main result in @xcite .
* lemma 4 .
* let @xmath174 be two ( possibly correlated ) random variables where @xmath52 takes values in the set @xmath175 , and let @xmath170 be a two - universal random function @xcite .
the random variable @xmath54 satisfies @xmath176 where @xmath177 the two main approaches for quantum memories are based on ensemble of atoms or on crystals . to our knowledge ,
the best existing quantum memories with ensemble of atoms have coherence times of the order of 100 milliseconds , a. g. radnaev _
et al . _ , nature phys . * 6 * , 894 ( 2010 ) .
moving to crystals , coherence times of the order of a few seconds have been reported for classical light , see j. j. longdell , e. fraval , m. j. sellars and n. b. manson , phys .
95 , 063601 ( 2005 ) . while in principle
the method should be scalable to light at the quantum level , this has not been demonstrated yet .
of course , improvements on these coherence times may be expected in the foreseeable future , however there is no evidence that these improvements will be significant .
s. pironio , a. acn , s. massar , a. boyer de la giroday , d. n. matsukevich , p. maunz , s. olmschenk , d. hayes , l. luo , t. a. manning and c. monroe , _ random numbers certified by bell s theorem _ nature * 464 * 1021 ( 2010 ) .
f. magniez , d. mayers , m. mosca and h. ollivier , _ self - testing of quantum circuits _ , proceedings of 33rd international colloquium on automata , languages and programming , volume 4051 , series lecture notes in computer science , 72 ( 2006 ) .
m. navascus , s. pironio and a. acn , _ bounding the set of quantum correlations _ _ phys
lett . _ * 98 * 010401 ( 2007 ) ; _ a convergent hierarchy of semidefinite programs characterizing the set of quantum correlations _ , new j. phys . * 11 * , 045021 ( 2009 ) .
r. arnon - friedman , e. hnggi and a. ta - shma , _ towards the impossibility of non - signalling privacy amplification from time - like ordering constraints _
, arxiv:1205.3736 ; r. arnon - friedman and a. ta - shma , _ on the limits of privacy amplification against non - signalling memory attacks _ , arxiv:1211.1125 . | device - independent quantum key distribution ( diqkd ) is a formalism that supersedes traditional quantum key distribution , as its security does not rely on any detailed modelling of the internal working of the devices .
this strong form of security is possible only using devices producing correlations that violate a bell inequality .
full security proofs of diqkd have been recently reported , but they tolerate zero or small amounts of noise and are restricted to protocols based on specific bell inequalities . here , we provide a security proof of diqkd that is both more efficient and noise resistant , and also more general as it applies to protocols based on arbitrary bell inequalities and can be adapted to cover supra - quantum eavesdroppers limited by the no - signalling principle only .
it requires , however , the extra assumption that the adversary does not have a long - term quantum memory , a condition that is not a limitation at present since the best existing quantum memories have very short coherence times . |
strong interaction phases play an important role in the decays of heavy mesons . they produce visible effects in many nonleptonic decays and could be important for the proper interpretation of effects of underlying fundamental physics . for example
, strong phases between isospin amplitudes , @xmath1 and @xmath2 , @xmath3 affect branching ratios of individual decays , as well as ratios of rates of isospin - related transitions .
more importantly , they complicate interpretations of cp - violating phases from @xmath4 transitions observed in the so - called direct cp - violating asymmetries .
provided that b - decay amplitude depends on at least two amplitudes with different weak and strong phases ( for example , tree @xmath5 and penguin @xmath6 amplitudes ) , @xmath7 a cp - violating asymmetry can be formed , @xmath8 which clearly depends on both , cp - conserving @xmath9 and cp - violating phase @xmath10 differences . cp - conserving phase difference is associated with strong interactions .
there are , of course , many more examples .
it is threfore important to have a way of computing those phases , which in general would depend on the meson system under consideration .
the difference of the physical picture at the energy scales relevant to @xmath11 , @xmath12 and @xmath0 decays calls for a specific descriptions for each class of decays .
for instance , the relevant energy scale in @xmath11 decays is @xmath13 gev .
with such a low energy release only a few final state channels are available .
this significantly simplifies the theoretical understanding of fsi in kaon decays .
in addition , chiral symmetry can also be employed to assist the theoretical description of fsi in @xmath11 decays . in @xmath12
decays , the relevant scale is @xmath14 gev .
this region is populated by the light quark resonances , so one might expect their significant influence on the decay rates and @xmath15-violating asymmetries .
no model - independent description of fsi is available , but it is hinted at experimentally that the number of available channels is still limited , allowing for a modeling of the relevant qcd dynamics .
finally , in @xmath0 decays , where the relevant energy scale @xmath16 gev is well above the resonance region , the heavy quark limit might prove useful .
final state interactions in @xmath17 arise as a consequence of the unitarity of the @xmath18-matrix , @xmath19 , and involve the rescattering of physical particles in the final state .
the @xmath20-matrix , @xmath21 , obeys the optical theorem : @xmath22 = { 1 \over 2 } \sum_{i } \langle f | { \cal t}^\dagger |
i \rangle \langle i | { \cal t } | a \rangle \ \ , \label{unit}\ ] ] where @xmath23 denotes the discontinuity across physical cut . using @xmath24 in the form @xmath25
this can be transformed into @xmath26 here , the states @xmath27 represent all possible final states ( including @xmath28 ) which can be reached from the state @xmath29 by the weak transition matrix @xmath20 .
the right hand side of eq .
( [ opt ] ) can then be viewed as a weak decay of @xmath29 into @xmath27 followed by a strong rescattering of @xmath27 into @xmath30 .
thus , we identify @xmath31 as a fsi rescattering of particles .
notice that if @xmath27 is an eigenstate of @xmath18 with a phase @xmath32 , we have @xmath33 which implies equal rates for the charge conjugated decays - violating asymmetries as no @xmath15 asymmetry is generated in this case . ] .
also @xmath34 the matrix elements @xmath35 are assumed to be the `` bare '' decay amplitudes and have no rescattering phases .
this implies that these transition matrix elements between charge conjugated states are just the complex conjugated ones of each other .
( [ watson ] ) is known as watson s theorem @xcite .
note that the problem of unambiguous separation of `` true bare '' amplitudes from the `` true fsi '' ones ( known as omns problem ) was solved only for a limited number of cases .
while the above discussion gives the most general way of determining strong phases , especially if an s - matrix is easily diagonalized , it might not be the most practical way of dealing with strong phases in @xmath0-decays due to the large number of available channels .
theoretical analysis of decays of heavy - flavored mesons , in particular @xmath0-mesons , simplifies in the limit @xmath36 . in this limit , @xmath37-pair produced in the weak decay of a @xmath38-quark , emerges as a small color dipole .
this is a reasonable assumption , as the length scale of @xmath37 production is set by the inverse heavy quark mass , while soft qcd interactions are governed by the length scale associated with @xmath39 , and so their effects will be suppressed by powers of @xmath40 .
then , if @xmath41 decay amplitude is dominated by this two - body - like configuration with small invariant mass , a factorization theorem can be written @xcite @xmath42 all corrections to eq .
( [ facttheor ] ) should be suppressed by either @xmath43 or @xmath44 .
in fact , one can perform phenomenological analysis of @xmath12 and @xmath0-decays to show that amplitude behavior in the large @xmath45 limit is respected @xcite .
one can use perturbative arguments to calculate final state phases for charmless @xmath0 decays using perturbative qcd @xcite . indeed ,
@xmath46 process , with subsequent final state rescattering of the two charmed quarks into the final state ( penguin diagram ) does the job , as for the energy release of the order @xmath47 available in @xmath38 decay , the rescattered @xmath48-quarks can go on - shell generating a perturbative cp - conserving phase and thus @xmath49 , which is usually quite small for the experimentally feasible decays , @xmath50
. one might be tempted to conclude that all strong phases in @xmath0-decays should be dominated by perturbative phases from @xmath51 and @xmath52 and therefore be small .
this conclusion , however , will not be correct if , for instance , real part of the decay amplitude happened to be small .
since @xmath53 the smallness of @xmath43 can be easily overpowered by large @xmath54 factor . in addition , @xmath44 corrections could not be computed in at this time . yet , in this framework , it is there non - perturbative contributions to strong phases are introduced
. those contributions could be quite large .
a multitude of @xmath55 and @xmath56 transitions were evaluated in qcd factorization ( qcdf ) @xcite . however , some predictions appear in disagreement with experiemental data @xcite . among those ,
is the recent observation of @xmath57 branching ratio , @xmath58 , which appears to be quite larger than predicted in qcd factorizaton , @xmath59 , inconsistensies in @xmath60 transitions which rule out small phases predicted by qcdf @xcite , as well as others @xcite .
a way to estimate non - perturbative contributions is to model them using approach final state interactions ( fsi ) . in this approach
, a transition @xmath61 is divided into @xmath62 where @xmath63 are the final and intermediate meson states .
an important point is then how to model the rescattering amplitude @xmath64 .
most recent studies @xcite use a simple t - channel resonance exchange to model the rescattering .
for example , @xmath65 transition with experimental branching ratio of @xmath66 can be affected by a decay @xmath67 with a much larger branching ratios of , say , @xmath68 and subsequent rescattering of @xmath69 into the @xmath70 final state via t - channel @xmath71-resonance exchange . in principle , however , other resonances , such as spin-2 @xmath72 and others should be taken into account .
a rescattering amplitude takes a very simple form , @xmath73 where @xmath74 is a coupling constant ( in practice , a @xmath75-dependent form - factor ) , and @xmath76 is a spin of an exchanged particle . a big problem with this form of exchange amplitude
is that it violates unitarity for @xmath77 and thus is not appropriate for @xmath36 limit calculations .
a solution to this problem is well - known and requires reggeization of the scattering amplitudes @xcite , @xmath78 with @xmath79 being a regge trajectory .
one consequence of this is the fact that high energy fsis are dominated by the multiparticle intermediate states @xcite . @xmath80
it is known that scattering of high energy particles may be divided into ` soft ' and ` hard ' parts .
soft scattering occurs primarily in the forward direction with limited transverse momentum distribution which falls exponentially with a scale of order @xmath81 gev . at higher transverse momentum one
encounters the region of hard scattering , which can be described by perturbative qcd . in exclusive @xmath0 decay one
faces the difficulty of separating the two .
it might prove useful to employ unitarity in trying to describe fsi in exclusive @xmath0 decays .
it is easy to investigate first the
_ elastic _ channel .
the inelastic channels have to share a similar asymptotic behavior in the heavy quark limit due to the unitarity of the @xmath18-matrix .
the choice of elastic channel is convenient because of the optical theorem which connects the forward ( imaginary ) invariant amplitude @xmath82 to the total cross section , @xmath83 where @xmath84 and @xmath75 are the usual mandelstam variables .
the asymptotic total cross sections are known experimentally to rise slowly with energy and can be parameterized by the form @xcite , @xmath85 where @xmath86 gev is a typical hadronic scale . considering only the imaginary part of the amplitude , and building in the known exponential fall - off of the elastic cross section in @xmath75 ( @xmath87 ) @xcite by writing @xmath88
one can calculate its contribution to the unitarity relation for a final state @xmath89 with kinematics @xmath90 and @xmath91 : @xmath92 where @xmath93 , and @xmath94 .
one can refine the argument further , since the phenomenology of high energy scattering is well accounted for by the regge theory @xcite . in the regge model , scattering amplitudes are described by the exchanges of regge trajectories ( families of particles of differing spin ) with the leading contribution given by the pomeron exchange . calculating the pomeron contribution to the elastic final state rescattering in @xmath95 one finds @xcite @xmath96 it is important that the pomeron - exchange amplitude is seen to be almost purely imaginary .
however , of chief significance is the identified weak dependence of @xmath97 on @xmath98 the @xmath99 factor in the numerator is attenuated by the @xmath100 dependence in the effective value of @xmath38 .
the analysis of the elastic channel suggests that , at high energies , fsi phases are _ mainly generated by inelastic effects _
, which follows from the fact that the high energy cross section is mostly inelastic .
this also follows from the fact that the pomeron elastic amplitude is almost purely imaginary . since the study of elastic rescattering has yielded a @xmath20-matrix element @xmath101 , i.e. @xmath102 , and since the constraint of unitarity of the @xmath18-matrix implies that the off - diagonal elements are @xmath103 , with @xmath97 approximately @xmath104 in powers of @xmath98 and numerically @xmath105 , then the inelastic amplitude must also be @xmath106 with @xmath107 .
similar conclusions follow from the consideration of the final state unitarity relations .
this complements the old bjorken picture of heavy meson decay ( the dominance of the matrix element by the formation of the small hadronic configuration which grows into the final state pion `` far away '' from the point it was produced and does not interact with the soft gluon fields present in the decay , see also @xcite for the discussion ) by allowing for the rescattering of multiparticle states , production of whose is favorable in the @xmath36 limit , into the two body final state .
analysis of the final - state unitarity relations in their general form is complicated due to the many contributing intermediate states , but we can illustrate the systematics of inelastic scattering in a two - channel model .
it involves a two - body final state @xmath108 undergoing elastic scattering and a final state @xmath109 which represents ` everything else ' .
as before , the elastic amplitude is purely _ imaginary _ , which dictates the following _ one - parameter _ form for the @xmath110 matrix @xmath111 where we identify @xmath112 .
the unitarity relations become @xmath113 denoting @xmath114 and @xmath115 to be the decay amplitudes in the limit @xmath116 , an exact solution to eq .
( [ big ] ) is given by @xmath117 thus , the phase is given by the inelastic scattering with a result of order @xmath118 clearly , for physical @xmath0 decay , we no longer have a simple one - parameter @xmath18 matrix , and , with many channels , cancellations or enhancements are possible for the sum of many contributions .
however , the main feature of the above result is expected to remain : inelastic channels can not vanish and provide the non - perturbative fsi phase . further studies of flavor - nondiagonal transitions showed that fsis are suppressed by powers of @xmath44 @xcite and thus can be used to estimate nonperturbative strong phases in the heavy quark limit .
i reviewed the physics of final state interactions in meson decays .
i showed that the use of pole exchange graphs to model fsi amplitudes leads to violation of unitarity in the large @xmath119 limit , so consistent studies of high - energy rescattering amplitudes must involve reggiazed rescattering amplitudes .
several examples were provided .
one of the main goals of physics of @xmath15 violation and meson decay is to correctly extract the underlying parameters of the fundamental lagrangian that are responsible for these phenomena .
the understanding of final state interactions is very important for the success of this program .
this work was supported in part by the u.s .
national science foundation under career award phy0547794 , and by the u.s .
department of energy under contract de - fg02 - 96er41005 .
m. beneke , g. buchalla , m. neubert and c. t. sachrajda , phys .
lett . * 83 * , 1914 ( 1999 ) [ arxiv : hep - ph/9905312 ] . m. beneke , g. buchalla , m. neubert and c. t. sachrajda , nucl .
b * 591 * , 313 ( 2000 ) [ arxiv : hep - ph/0006124 ] . c. w. bauer , d. pirjol and i. w. stewart , phys .
lett . * 87 * , 201806 ( 2001 ) [ arxiv : hep - ph/0107002 ] .
h. d. politzer and m. b. wise , phys .
b * 257 * , 399 ( 1991 ) .
m. neubert and a. a. petrov , phys .
b * 519 * , 50 ( 2001 ) [ arxiv : hep - ph/0108103 ] .
m. beneke and m. neubert , nucl . phys .
b * 675 * , 333 ( 2003 ) [ arxiv : hep - ph/0308039 ] .
m. bander , d. silverman , and a. soni , phys .
lett . * 43 * , 242 ( 1979 ) ; j .-
gerard and w .- s .
hou , phys . rev .
d * 43 * , 2909 ( 1991 ) ; yu .
dokshitser , n. uraltsev , jetp lett .
* 52 * ( 10 ) , 1109 ( 1990 ) .
d. s. du , x. q. li , z. t. wei and b. s. zou , eur .
j. a * 4 * , 91 ( 1999 ) [ arxiv : hep - ph/9805260 ] . h. y. cheng , c. k. chua and a. soni , phys .
d * 72 * , 014006 ( 2005 ) [ arxiv : hep - ph/0502235 ] . c. d. lu , y. l. shen and w. wang , phys
. rev .
d * 73 * , 034005 ( 2006 ) [ arxiv : hep - ph/0511255 ] .
a. f. falk , a. l. kagan , y. nir and a. a. petrov , phys .
d * 57 * , 4290 ( 1998 ) [ arxiv : hep - ph/9712225 ] . c. k. chua , w. s. hou and k. c. yang , mod . phys .
lett . a * 18 * , 1763 ( 2003 ) [ arxiv : hep - ph/0210002 ] .
a. b. kaidalov and m. i. vysotsky , arxiv : hep - ph/0603013 .
donoghue and a.a .
petrov , phys . lett . *
b393 * , 149 ( 1997 ) . | i give a brief review of final state interactions ( fsi ) in heavy meson decays , paying particular attention to @xmath0-meson physics .
available theoretical methods for dealing with the effects of fsi are discussed . |
detailed investigation of geophysical flows involves experimental campaigns in which buoys , in the ocean , or balloons , in the atmosphere , are released in order to collect lagrangian data against which theories and models can be tested .
questions concerning oil spill fate , fish larvae distribution or search and rescue operations are only a few examples that make the study of advection and diffusion properties not only a challenging scientific task , but also a matter of general interest . in the past years , an amount of lagrangian data about the south atlantic ocean ( sao ) was collected thanks to the first global atmospheric research program ( garp ) global experiment ( fgge ) drifters , released following the major shipping lines , the southern ocean studies ( sos ) drifters , deployed in the brazil - malvinas confluence ( bmc ) and the programa nacional de bias ( pnboia ) drifters [ brazilian contribution to the global oceans observing system ( goos ) ] , released in the southeastern brazilian bight ( sbb ) .
these data allowed estimates of eddy kinetic energy ( eke ) , integral time scales and diffusivities ( piola et al .
1987 ; figueroa and olson 1989 ; schfer and krauss 1995 ) . despite the relatively uniform coverage , the boundary currents resulted poorly populated by buoys ; furthermore , all previous studies about drifters in the south atlantic have concerned one - particle statistics only . in this regard , in the framework of monitoring by ocean drifters ( mondo ) project , a recent lagrangian experiment , consisting in the release of a set of 39 world ocean circulation experiment ( woce ) surface velocity program ( svp ) drifters , was planned in relationship with an oil drilling operation in proximity of the coast of brazil , around ( @xmath1s , @xmath2w ) .
part of the drifters were deployed in 5-element clusters , some of them with initial drifter separations smaller than 1 km .
this set of satellite - tracked lagrangian trajectories offers , now , the opportunity to revisit advective and diffusive properties characterizing the current systems explored by the drifters . from the analysis of trajectory pair dispersion
we can extract , in principle , information about the dominant physical mechanism acting at a certain scale of motion ( e.g. chaotic advection , turbulence , diffusion ) . a thorough description of the oceanography of the south atlantic ocean , particularly of the main circulation patterns and of the mass transport properties , can be found in peterson and stramma ( 1991 ) ; campos et al .
( 1995 ) ; stramma and england ( 1999 ) .
the major feature characterizing the central region of the sao is the large anticyclonic ( anticlockwise ) circulation known as subtropical gyre ( sg ) .
other relevant surface current systems are : south equatorial current ( sec ) , brazil current ( bc ) , malvinas current ( mc ) , south atlantic current ( sac ) and benguela current ( bgc ) , as shown in fig .
[ fig : sao ] . in the quasigeostrophic ( qg ) approximation ,
valid for relative vorticities much smaller than the ambient vorticity because of the earth s rotation , some theoretical arguments would predict that , from the scale of the forcing at which eddies are efficiently generated by instability , e.g. the rossby radius of deformation , both a down - scale enstrophy cascade and an up - scale energy cascade take place , corresponding to energy spectra @xmath3 and @xmath4 , respectively ( kraichnan 1967 ; charney 1971 ) . from a relative dispersion perspective , in the forward cascade range , the mean square relative displacement between two trajectories grows exponentially fast in time ( lin 1972 ) , whereas , in the inverse cascade range , it follows the @xmath5 power law ( obhukov 1941 ; batchelor 1950 ) .
possible deviations from this ideal picture may reasonably come from the non homogeneous and non stationary characteristics of the velocity field : for example , in the case of boundary currents , as well as from ageostrophic effects . at this regard ,
one presently debated issue is the role of submesoscale vortices ( mcwilliams 1985 ) [ velocity field features of size @xmath0 km ] in determining the shape of the energy spectrum at intermediate scales between the rossby deformation radius [ in the ocean typically @xmath6 km ] and the dissipative scales ( much smaller than 1 km ) .
a thorough discussion about submesoscale processes and dynamics can be found in thomas et al .
recent high - resolution 3d simulations of upper layer turbulence ( capet et al .
2008a , b ; klein et al .
2008 ) have shown that the direct cascade energy spectrum flattens from @xmath7 to @xmath8 for order @xmath9 rossby number @xmath10 , where @xmath11 is the typical velocity difference on a characteristic length @xmath12 of the flow and @xmath13 @xmath14 is the coriolis parameter .
our main purpose is to exploit the mondo drifter trajectories , shown in fig .
[ fig : drifters ] , to examine relative dispersion by means of several indicators , and discuss the consistency of our data analysis in comparison with classical turbulence theory predictions , model simulations and previous drifter studies available for different regions of the ocean .
this paper is organized as follows : in section [ sec : diffusion ] we recall the definitions of the major indicators of the lagrangian dispersion process ; in section [ sec : data ] we give a description of the mondo drifter lagrangian data ; in section [ sec : results ] the outcome of the data analysis is presented ; in section [ sec : model ] , we report the analysis of the ocean model lagrangian simulations in comparison with the observative data ; and , in section [ sec : conclusions ] , we outline and discuss the main results we have obtained in the present work .
let @xmath15 be the position vector of a lagrangian particle , in a 3d space , evolving according to the equation @xmath16 , where @xmath17 is a 3d eulerian velocity field , and let us indicate with @xmath18 the lagrangian velocity along the trajectory @xmath19 .
let us imagine , then , a large ensemble of @xmath20 lagrangian particles , passively advected by the given velocity field , and refer , for every statistically averaged quantity , to the mean over the ensemble .
the autocorrelation function of a lagrangian velocity component @xmath21 can be defined , for @xmath22 , as : @xmath23 / \left [ \langle v(t_0)^2 \rangle - \langle v(t_0 ) \rangle^2 \right ] \label{eq : autocorr}\ ] ] in case of stationary statistics , @xmath24 depends only on the time lag @xmath25 .
the integral lagrangian time @xmath26 is the time scale after which the autocorrelation has nearly relaxed to zero .
typically @xmath26 can be estimated as the time of the first zero crossing of @xmath24 , or , alternatively , as the time after which @xmath27 remains smaller than a given threshold .
absolute dispersion can be defined as the variance of the particle displacement relatively to the mean position at time @xmath28 : @xmath29 ^ 2 \rangle - \langle \left [ \mathbf{r}(t)-\mathbf{r}(0 ) \right ] \rangle^2 \label{eq : absdisp}\ ] ] in the limit of very small times , absolute dispersion is expected to behave as follows : @xmath30 where @xmath26 is the lagrangian autocorrelation time and @xmath31 is the total lagrangian velocity variance .
the ballistic regime ( [ eq : ballistic ] ) lasts as long as the trajectories save some memory of their initial conditions . in the opposite limit of very large times ,
when the autocorrelations have relaxed to zero and the memory of the initial conditions is lost , we have : @xmath32 where @xmath33 is the absolute diffusion coefficient ( taylor 1921 ) . although single particle statistics give information about the advective transport , mostly because of the largest and most energetic scales of motion , two ( or more ) particle statistics give information about the physical mechanisms acting at any scale of motion , compatibly with the available resolution .
let us indicate with @xmath34 the distance between two trajectories at time @xmath28 .
relative dispersion is defined as the second order moment of @xmath35 : @xmath36 where the average is over all the available trajectory pairs @xmath37 . in the small scale range , the velocity field between two sufficiently close trajectories
is reasonably assumed to vary smoothly .
this means that , in nonlinear flows , the particle pair separation typically evolves following an exponential law : @xmath38 where , from the theory of dynamical systems , @xmath39 is the generalized lyapunov exponent of order 2 ( bohr et al .
1998 ) . when fluctuations of the finite - time exponential growth rate around its mean value are weak , one has @xmath40 , where @xmath41 is the ( lagrangian ) maximum lyapunov exponent ( mle ; boffetta et al .
notice that for ergodic trajectory evolutions the lyapunov exponents do not depend on the initial conditions .
if @xmath42 ( except for a set of zero probability measure ) we speak of lagrangian chaos . the chaotic regime holds as long as the trajectory separation remains sufficiently smaller than the characteristic scales of motion . in the opposite limit of large particle separations , when two trajectories are sufficiently distant from each other to be considered uncorrelated , the mean square relative displacement behaves as : @xmath43 where we indicate with @xmath44 the asymptotic eddy - diffusion coefficient ( richardson 1926 ) . at any time @xmath28
, the diffusivity @xmath45 can be defined as : @xmath46 with @xmath47 for @xmath48 . if the velocity field is characterized by several scales of motion , relative dispersion in the intermediate range , i.e. between the smallest and the largest characteristic length , depends on the properties of local velocity differences , i.e. the mean gradients on finite scale .
for instance , in 3d fully developed turbulence ( frisch 1995 ) , relative dispersion follows the so - called richardson s law : @xmath49 with @xmath50 , as long as the trajectory separation lies in the inertial range of the energy cascade ( richardson 1926 ) from large to small scales .
it is worth to remark that richardson s law also holds in the inverse cascade range ( from small to large scales ) of 2d turbulence because , in that case as well , the energy spectrum follows kolmogorov s @xmath51 scaling , exactly as in the inertial range of 3d turbulence ( kraichnan 1967 ) .
any power law of the type ( [ eq : richardson ] ) for @xmath52 with @xmath53 is known as super - diffusion .
the finite - scale lyapunov exponent ( fsle ) has been formerly introduced as the generalization of the mle @xmath54 for non - infinitesimal perturbations ( aurell et al .
if @xmath55 is the size of the perturbation on a trajectory in the phase space of a system , and @xmath56 is the phase space averaged time that @xmath55 takes to be amplified by a factor @xmath57 , then the fsle is defined as @xmath58 the quantity @xmath59 is the exit time of the perturbation size from the scale @xmath55 , and it is defined as the first arrival time to the scale @xmath60 , with @xmath61 .
the computation of the expectation value of the growth rate at a fixed scale , which justifies the definition ( [ eq : fsle ] ) , is described in boffetta et al .
as far as lagrangian dynamics are concerned , the evolution equations of the lagrangian trajectories form a dynamical system whose phase space is the physical space spanned by the trajectories . in this context , the analysis of relative dispersion can be treated as a problem of finite - size perturbation evolution , with scale - dependent growth rate measured by the fsle .
the first who had the idea to measure the relative dispersion , or , equivalently , the diffusivity , as a function of the trajectory separation was richardson ( 1926 ) .
the fsle is fundamentally based on the same principle .
recently , the use of fixed - time and fixed - scale averaged indicators of relative dispersion in various contexts , from dynamical systems to observative data in ocean and atmosphere , have been reviewed and discussed in several works ( artale et al .
1997 ; boffetta et al . 2000
; lacorata et al . 2001 , 2004 ; lacasce and ohlmann 2003 , lacasce 2008 ) . by a dimensional argument ,
if relative dispersion follows a @xmath62 scaling law , then the fsle is expected to scale as @xmath63 .
for example , in the case of standard diffusion we expect @xmath64 ; in richardson s super - diffusion , @xmath65 ; in ballistic or shear dispersion we have @xmath66 .
chaotic advection means exponential separation between trajectories . in terms of fsle
this amounts to a scale - independent @xmath67constant : that is , @xmath68 . in the limit of infinitesimal separation ,
the fsle is nothing but the mle , i.e. @xmath69 ( aurell et al . 1996 ) . under these conditions ,
relative dispersion is said to be a non - local process , since it is determined by velocity field structures with a characteristic scale much larger than the particle separation .
on the contrary , when the growth of the distance between two particles is mainly driven by velocity field structures of the same scale as the particle separation , relative dispersion is said to be a local process .
the super - diffusive processes occurring in 2d and 3d turbulence are phenomena of this type .
an indicator related to the fsle is the mean square velocity difference between two trajectories as function of their separation .
indicating with @xmath70 , @xmath71 , @xmath72 , @xmath73 the positions and the lagrangian velocities , respectively , of two particles @xmath74 and @xmath75 at a given time , we define the finite - scale relative velocity ( fsrv ) at scale @xmath55 : @xmath76 ^ 2 \rangle = \langle \left [ { \mathbf v}^{(1 ) } - { \mathbf v}^{(2 ) } \right]^2 \rangle \label{eq : fsrv}\ ] ] where the average is over all trajectory pairs fulfilling the condition @xmath77 at some time @xmath28 . from the fsrv
a scale - dependent diffusivity can be formed as @xmath78 ^ 2 \rangle^{1/2}$ ] and compared to the classical time - dependent diffusivity @xmath45 defined in ( [ eq : diffusivity ] ) .
we would like to end this section with a remark .
it is physically reasonable to assume that , for statistically homogeneous flows , the behavior of the relative dispersion changes in correspondence of certain characteristic lengths of the system , rather than after certain time intervals . the major shortcoming when measuring the mean square relative displacement at fixed time consists in averaging together particle pairs that may have very different separations , i.e. may belong to very different dispersion regimes . on the contrary , the mean dispersion rate at fixed scale , or other scale - dependent related quantities ,
generally have the benefit of being weakly affected by possible overlap effects .
this is of particular relevance since the presence of scaling laws in the pre - asymptotic regime raises questions about the characteristics ( e.g. , the kinetic energy spectrum ) of the underlying velocity field .
the mondo project is a campaign planned by prooceano for eni oil do brasil to provide unrestricted access to oceanographic data and subsidize not only the companies operational goals but also the scientific community .
the lagrangian experiment consisted in the deployment of 39 woce svp drifters ( sybrandy and niiler 1991 ) in the surroundings of an oil drilling operation ( @xmath1 24 50 `` s and @xmath79 46 22 '' w ) in proximity of the bc .
the drifters are of the holey - sock type , composed by a surface floater , a tether and a submerged drogue , following the design criteria proposed by sybrandy and niiler ( 1991 ) .
the drogue length is 6.44 m , centered at 15 m deep , in order to represent the first 20 m current information .
the drifters are equipped with a gps device and iridium transmission , providing a positioning accuracy of 7 meters and an acquiring rate of 3 hours . from the 39 drifters ,
14 were deployed individually in a 3-day frequency and 25 were deployed in clusters of 5 , every 12 days .
justification for the 3-day frequency deployment is based upon the 6.5 days energy peak in the wind variability in the region ( stech and lorenzetti 1992 ) .
data used in this study comprehend the period within 21 september 2007 - 14 november 2008 and passed the quality control proposed by hansen and poulain ( 1996 ) in the global drifter program ( gdp ) database .
high frequency components have been removed by applying a blackman low - pass filter with a window of 15 points ( 45 hours ) .
the domain explored by the mondo project drifters mainly corresponds to that of the bc and to the area where the southward flowing bc meets the northward flowing mc , forming the bmc , which results in an eastward current feeding the sac .
bc is a warm western boundary current , flowing southward and meandering over the 200 m isobath ( peterson and stramma 1991 ; lima et al .
as reported by de castro ( 2006 ) and garfield ( 1990 ) the velocities inside the bc have values between 25 and 80 @xmath80 . bc flux increases with latitude ( assireu 2003 ; mller et al .
1998 ; gordon and greengrove 1986 ) and veers towards east when meeting the colder malvinas current around 40@xmath81s .
the bmc is a highly energetic zone with strong thermal gradient and great variability , playing an important role in weather and climate of south america ( legeckis and gordon 1982 ; pezzi et al . 2005 ; piola et al .
both bc and bmc show intense mesoscale activity with eddies detaching from both sides of the flow .
lentini et al .
( 2002 ) used satellite derived sea surface temperature ( sst ) to estimate an average of 7 rings per year , with lifetimes ranging from 11 to 95 days with major and minor radii of 126 @xmath82 50 km and 65 @xmath82 22 km respectively .
assireu ( 2003 ) estimates a diffusion coefficient between @xmath83 and @xmath84 @xmath85 , a lagrangian time scale between 1 and 5 days and a characteristic ( eulerian ) length scale varying from 19 to 42 km , which agrees with the meridional variation of the 1@xmath86 internal rossby radius of deformation in the region ( houry et al .
studies addressing the energetics in bc and bmc flows report great spatial and temporal variability ( see , e.g. , assireu et al . 2003 ; piola et al .
1987 ; stevenson 1996 ) with eke varying from 200 to 1200 @xmath87 and representing 70 - 90% of the total kinetic energy ( tke ) .
recently , oliveira et al .
( 2009 ) extended the analysis made by piola et al .
( 1987 ) and suggested that eke is usually lower than mean kinetic energy ( mke ) in the bc and always greater in the bmc .
and latitude @xmath88 , averaged over all drifters , as function of the time lag since the release .
the time sampling is @xmath89 day .
error bars are the standard deviations on the mean values . ] day .
, scaledwidth=70.0% ] from a pre - analysis we eliminated two drifters that very soon lost their drogue , so the set of active drifters is reduced to 37 units .
the lagrangian velocity statistics , reconstructed along the drifter trajectories by means of a finite difference scheme , provides standard deviations , in the zonal and meridional directions , of order @xmath90 cm s@xmath91 , compatible with estimates obtained from other observative data available for the same ocean region ( figueroa and olson 1989 ; schfer and krauss 1995 ) .
the drifters initially tend to move southward , driven by the bc ; then they are nearly stopped by the bmc , and , last , they tend to be transported eastward by the sac .
the time evolution of the drifter mean position coordinates in longitude and latitude is shown in fig .
[ fig : lonlat ] . during the first phase of dispersion , the mean meridional displacement grows almost linearly southward ( bc ) , while the mean zonal displacement is almost constant ; after about 120 days from the release time , the mean meridional displacement tends to saturate ( bmc ) , while the mean zonal displacement tends to grow almost linearly eastward ( sac ) .
given the differences between bc , bmc and sac in terms of energy and circulation patterns , and , in general , between the boundary of the subtropical gyre and its interior , the drifters do not sample a statistically homogeneous and stationary domain . in order to evaluate the lagrangian autocorrelations
, we divide the trajectory lifetime into time windows of @xmath92 days width .
then we compute ( @xmath93 ) where mean and variance of the lagrangian velocity components are recalculated for each window , and then we take the average , for a fixed time lag @xmath25 , over all the windows for all trajectories .
the resulting functions are plotted in fig .
[ fig : autocorr ] .
we notice that , since most of the statistics regards trajectory segments within or near the bc , the motion is slightly more autocorrelated in the meridional than in the zonal direction .
however , the order of the integral time scale is the same for both components . without discussing the various estimates that can be evaluated using different criteria
, we identify the two integral time scales around a value @xmath94 days , with a difference between zonal and meridional components of order @xmath9 days .
this implies that , for time lags significantly larger than @xmath26 , the lagrangian velocities along a trajectory become practically uncorrelated .
absolute dispersion , i.e. the mean square fluctuation around the mean position , for zonal and meridional coordinates is reported in fig .
[ fig : absdisp ] . , in zonal and meridional components .
the time sampling is @xmath89 day.,scaledwidth=70.0% ] in the present work we are not interested in reproducing estimates of small scale diffusivity , obtained by reconstructing a mean velocity field , in general not spatially uniform , and calculating the turbulent components from the difference between the drifter local velocities and the local mean field .
wherever we speak of diffusion coefficients , referring to ( [ eq : absdisp ] ) or ( [ eq : reldisp ] ) , we mean the _ effective _ diffusion coefficients of order @xmath95 where @xmath12 and @xmath11 are determined , respectively , by the characteristic size and rotation velocity of the largest eddies .
absolute dispersion , being dominated by the large scale features of the velocity field , reflects the anisotropy of the currents encountered by the drifters .
the ballistic dispersion @xmath96 scaling is plotted as reference for autocorrelated motions on time lags smaller than 10 days .
the diffusion @xmath97 scaling is a good approximation for meridional dispersion on time lags within @xmath98 days , but it is not good for zonal dispersion on any time lag .
after about @xmath99 days , meridional dispersion tends to saturate , while zonal dispersion starts growing again as @xmath96 , likely because , on those time lags , on average , the drifters have reached the bmc , have stopped their southward meridional run along the bc and start moving eastward along the sac , the natural consequence of which is the saturation of the meridional component and a second ballistic ( or shear ) regime for the zonal component . relative dispersion for three different initial separations is shown in fig .
[ fig : reldisp ] .
distances between two points on the ocean surface are calculated as great circle arcs , according to the spherical geometry standard approximation . for initial separations
@xmath100 km , @xmath101 km and @xmath102 km . the quantity @xmath103 day@xmath91 corresponds to @xmath104 day@xmath91 .
the time sampling is @xmath89 day .
error bars are the standard deviations on the mean values.,scaledwidth=70.0% ] , for initial separations @xmath100 , @xmath105 and @xmath102 km , with @xmath106 day@xmath91 .
the time sampling is @xmath89 day .
error bars are the standard deviations on the mean values.,scaledwidth=70.0% ] a way to increase relative dispersion statistics is to measure ( [ eq : reldisp ] ) also for so - called chance pairs , i.e. , particle pairs which may be initially very distant from each other but that , following the flow , happen to get sufficiently close to each other at some later time after their release , in some random point of the domain ( lacasce 2008 ) .
the maximum number of pairs counted in the statistics depends on the initial threshold : 24 pairs for @xmath100 km , 30 pairs for @xmath101 km and 39 pairs for @xmath102 km . the dependence of @xmath52 on @xmath107 is well evident .
the early regime of relative dispersion is reported in fig .
[ fig : lyapexp ] .
the value of the mle @xmath104 day@xmath91 is of the same order as other estimates obtained in various ocean regions , as discussed later .
the exponential growth has been adapted to the @xmath101 km curve which has a fair compromise between having a sufficiently small initial separation and , at the same time , an acceptable pair statistics .
relative diffusivity in the zonal and meridional directions , for initial separations @xmath101 km , is plotted in fig .
[ fig : timediff ] .
although the curves do not display a clean scaling behavior , in the time range @xmath108 days the diffusivities show some resemblance with the @xmath109 law expected in the richardson dispersion regime . in zonal and meridional components , for initial separations @xmath101 km .
the @xmath109 scaling is plotted as reference to the richardson regime .
the time sampling is @xmath89 day.,scaledwidth=70.0% ] we will examine now the results obtained with scale - dependent indicators .
the fsle has been evaluated for the same initial thresholds 1 , 5 and 10 km ; the results are reported in fig .
[ fig : fsle ] . , @xmath105 and @xmath110 km and amplification factor @xmath111 .
the small scale plateau corresponds to exponential separation with rate @xmath104 day@xmath91 .
the @xmath112 scaling corresponds to the richardson law .
the @xmath113 scaling corresponds to a @xmath8 spectrum .
error bars are the standard deviations on the mean values.,scaledwidth=70.0% ] particle pairs are counted in the statistics if , at a given time , they happen to to be separated by a distance smaller than the fixed threshold .
the density of scales @xmath55 is fixed by setting @xmath111 .
the maximum number of pairs considered varies with the initial threshold : 24 pairs for @xmath100 km , 30 pairs for @xmath101 km and 39 pairs for @xmath102 km .
larger initial thresholds ( viz . , @xmath114 km ) could increase the statistics of an order @xmath115 , but we will not consider this opportunity since , with too large initial thresholds , mesoscale dispersion rates may be affected by the inhomogeneity of the flow .
in other words , at a certain separation scale two drifters might be still spatially correlated in some regions of the domain , but totally uncorrelated in some other ones , depending on the structure of the flow .
the parameter @xmath116 can not be chosen arbitrarily close to 1 for obvious limitations because of finite resolution of the data .
we have checked that , with @xmath111 , `` clipping events '' , i.e. , suspect transitions between two neighbouring scales in only one time step @xmath89 days , are not significant . taking a value @xmath117 km for the rossby radius , we can see that mesoscale dispersion ( scales @xmath118 ) is compatible with a @xmath119 richardson regime , as it occurs in the inverse cascade range of qg turbulence .
cut - off due to finite lifetime of the trajectories occurs before reaching @xmath120 km separations .
the range @xmath121 km is characterized by a step - like behavior of the fsle .
the first plateau in the submesoscale range @xmath0 km corresponds to exponential separation with @xmath122 day@xmath91 , in agreement with the result obtained before for @xmath52 .
the second plateau at @xmath123 day@xmath91 on @xmath124 scales is compatible with exponential separation due to a @xmath7 direct cascade of qg turbulence .
it is unclear why there exist two quasi - constant levels for @xmath125 , which produce a significant variation of the exponential growth rate when @xmath55 decreases from meso to submesoscales .
possible interpretations of this behavior may be found in the action of submesoscale vortices , even though the fsle does not display a continuous cascade connecting these two scale ranges .
the @xmath126 scaling is plotted as a reference to a @xmath8 energy spectrum .
another remark concerns the gap at @xmath127 km between the end of the plateau and the beginning of the @xmath112 regime .
inverse cascade implies that smaller eddies , formed by instability at the rossby scale , tend to merge together in larger eddies up to scales in the range @xmath128 km .
assuming this actually occurs , we also have to consider that the bc and bmc generate rings of size @xmath129 km which detach from the main currents .
such mesoscale rings can have trapping effects on the drifters , at least to some extent , which would determine an abrupt fall of the dispersion rate ( fsle ) .
the presence of the bmc can also have a role in temporarily inhibiting the growth of the two particle separation , until the drifters find their way eastward along the sac .
let us now consider the finite - scale ( mean square ) velocity difference , fig .
[ fig : fsrv ] .
the lagrangian velocity components are reconstructed from the trajectories by means of a finite difference scheme . .
the scalings @xmath130 , @xmath131 and @xmath55 correspond to @xmath7 , @xmath51 and @xmath8 energy spectra , respectively .
error bars are the standard deviations on the mean values .
, scaledwidth=70.0% ] we notice , first , that the fsrv returns back a slightly cleaner picture than other indicators .
this may depend , for instance , on the fact that the amplification time of the separation between two drifters passing from @xmath55 to @xmath60 is affected by large fluctuations due to the non homogeneous characteristics of the flow , while the ( mean square ) relative velocity at scale @xmath55 is less affected by the history of the dispersion between two scales .
the two `` valleys '' at nearly 10 and 100 km leave room to hypothesis of trapping events by structures of size comparable with those scales .
the fsrv is consistent with richardson dispersion on scales larger than the rossby radius ( except for the `` valley '' at 100 km ) ; with exponential separation for two subranges @xmath6 km and @xmath132 km ; with a @xmath55 scaling in the subrange @xmath133 km , in turn consistent with a @xmath8 energy spectrum . the @xmath6 km exponential regime
can be associated to a @xmath7 direct cascade , while the @xmath132 km exponential regime can be associated to the action of submesoscale vortices . from the fsrv
an `` equivalent lagrangian spectrum '' @xmath134 ^ 2 \rangle / k$ ] can be defined , by dimensional arguments , replacing @xmath55 with @xmath135 , fig .
[ fig : spectrum ] . the same scenario formerly indicated by the fsrv is reproduced , in @xmath136 space , by @xmath137 as well . ) , direct cascade ( @xmath7 ) and the submesoscale ( @xmath8 ) spectra are plotted as reference .
the rossby radius @xmath138 30 km corresponds to a wavenumber @xmath139 .
, scaledwidth=70.0% ] we compare now the diffusivity ( see fig . [
fig : diff ] ) computed in both ways : as a fixed - time average ( [ eq : diffusivity ] ) and as a fixed - scale average from the fsrv .
vs @xmath140 and fixed - scale average @xmath141 vs @xmath55 .
the @xmath142 and @xmath143 correspond to @xmath51 and @xmath7 spectra , respectively .
the @xmath144 scaling corresponds to a @xmath8 spectrum.,scaledwidth=70.0% ] both quantities are plotted as functions of the separation between two drifters : @xmath145 ^ 2 \rangle^{1/2}$ ] with @xmath55 as the independent variable , and @xmath45 versus @xmath140 where the independent variable is the time @xmath28 .
the diffusivity is less affected by distortions present in the signal than other indicators .
the two regimes @xmath146 at mesoscales and @xmath147 at scales @xmath124 are consistent with the inverse cascade @xmath51 and direct cascade @xmath7 , respectively , as predicted by qg turbulence theory .
a scaling @xmath148 , corresponding to a @xmath8 spectrum , connects scales of the order of the rossby radius to the submesoscale , and a scaling @xmath147 is the signature of exponential separation driven by velocity field features on scales @xmath0 km .
to check the consistency of the results obtained from the data analyis of the mondo drifters , we have performed two numerical experiments in which @xmath129 virtual drifters are `` deployed '' in a surface current field generated by a global operational forecast system using the hybrid coordinates ocean model ( hycom ; bleck and benjamin 1993 ) and the navy coupled ocean data assimilation ( ncoda ; cummings 2005 ) .
the model grid step is @xmath149 ( approximately @xmath150 km ) and outputs are available with a 1 day time step . from the 32 layer global grid ,
the surface layer is extracted in the area within @xmath151-@xmath152 and @xmath153-@xmath154 for the same period of drifters trajectories ( 20 september 2007 - 21 october 2008 ) .
conversion from earth coordinates to meters is accomplished following the great circle distance approximation .
the integration algorithm of the lagrangian trajectories uses a fixed timestep @xmath155 day and tri - linear interpolations .
the maximum integration time is @xmath156 days but particles reaching the shoreline are stopped and removed from the subsequent integration .
we indicate the numerical experiments with e1 and e2 . in the first one ( e1 )
the drifters are uniformly deployed in an area of about @xmath157 km@xmath158 centered around a position corresponding to the mean initial location of mondo drifters .
the average initial distance between particle pairs is @xmath159 km .
the lifetime of trajectories is between @xmath160 days and @xmath161 days . in the second experiment ( e2 ) the initial distribution of the drifters
is characterized by larger separations , namely comparable to the spacing of the numerical grid ( @xmath162 km ) ; the average initial distance between particle pairs is @xmath163 km and the duration of trajectories is @xmath164 days . here
we present some comparisons between the analysis of data from model trajectories and those from the actual drifters , focusing on the relative dispersion process .
first we consider a fixed - time indicator , namely relative dispersion @xmath52 , for trajectories selected with small enough initial separation , that is for @xmath165 km . as it can be seen in fig .
[ fig : reldisp_vd ] , the model outcome is in reasonable agreement with the behaviour of actual drifters . in particular , experiment e1 is capable of reproducing the early growth of relative dispersion but it can not catch the late time behaviour due to too short trajectories . on the other hand , with experiment
e2 we obtain a nicer agreement at late times , thanks to the longer lifetime of virtual trajectories in this case , but we loose the agreement at early times because now the initial separation between particle pairs is on average much larger . for initial separations
@xmath102 km for mondo drifters and virtual drifters from numerical experiments e1 and e2 . for virtual drifters errors are of the order of point size ; time sampling is @xmath89 day for mondo drifters and @xmath155 day for virtual drifters.,scaledwidth=70.0% ] for what concerns fixed - scale indicators , we consider the fsle and the fsrv ( figures [ fig : fsle_vd ] , [ fig : fsrv_vd ] ) . to increase the statistics we now select trajectories with a larger initial separation , that is for @xmath166 km ( similar results are found for smaller values of @xmath107 , though they are more noisy ) .
the behaviours of both fsle and fsrv support a double cascade scenario on the same scales as those found with mondo drifters .
the plateau value of fsle at scales @xmath167 is very close to the one found in the real experiment ( @xmath168 day@xmath91 ) . at larger scales , for both numerical experiments
e1 , e2 the behaviour of fsle is compatible with a scaling law @xmath169 supporting an inverse energy cascade process .
experiment e2 , which is characterized by longer trajectories , shows a clearer scaling , thanks to a larger number of pairs reaching this range of large scales .
mean square velocity differences in the same range of scales display a reasonably clear @xmath131 scaling , also supporting an inverse energy cascade , with values close to those found with mondo drifters . at separations smaller than the rossby deformation radius ,
both indicators point to the presence of a direct enstrophy cascade : the fsle is constant and the fsrv behaves as @xmath170 ^ 2\rangle \sim \delta^2 $ ] .
this only partially agrees with the results found for real drifters , namely only in the scale range @xmath92 km @xmath171 km . at subgrid scales , velocity field features
are not resolved and relative dispersion is necessarily a nonlocal exponential process driven by structures of size of the order of ( at least ) the rossby radius .
correspondingly , the fsle computed on model trajectories does not display the higher plateau level at scales smaller than @xmath92 km .
km and amplification factor @xmath111 for mondo drifters and virtual drifters from numerical experiments e1 and e2 . for virtual drifters errors
are of the order of point size .
the large - scale saturation ( e1 ) depends on the value of the trajectory integration time.,scaledwidth=70.0% ] km for mondo drifters and virtual drifters from numerical experiments e1 and e2 . for virtual drifters errors
are of the order of point size.,scaledwidth=70.0% ] finally , the behaviour of the relative diffusivity @xmath141 as a function of the separation @xmath55 is shown in fig .
[ fig : diffu_vd ] for both mondo and virtual drifters . here
@xmath141 is computed from the mean square velocity difference , as described in sections [ sec : diffusion]c and [ sec : results]b .
model and experimental data again are in agreement at scales larger than the numerical space resolution ( @xmath162 km ) .
indeed , in both numerical experiments e1 and e2 we find scaling behaviours compatible with a qg double cascade : @xmath172 ( corresponding to richardson s super - diffusion in an inverse cascade regime ) for @xmath173 , and @xmath174 ( corresponding to a direct cascade smooth flow ) for @xmath175 . at variance with the outcome of the real experiment with mondo drifters , here we are unable to detect any significant deviations from the qg turbulence scenario at small scales . as function of the separation @xmath55 for mondo drifters and virtual drifters from numerical experiments
e1 and e2 ; here @xmath166 km . , scaledwidth=70.0% ]
the failure of the model to reproduce flow features at very small scale is of course due to its finite spatial resolution , which is of the order of @xmath92 km . below this length scale , the velocity field computed in the model is smooth , while the one measured in the real ocean clearly displays active scales also in the range @xmath176 km .
nevertheless , the overall conclusion we can draw from the above comparisons is that the characteristics of the relative dispersion process found with mondo drifters agree with those obtained with an ocean general circulation model ( ogcm ) for scales @xmath177 km .
lagrangian dispersion properties of drifters launched in the southwestern corner of the south atlantic subtropical gyre have been analyzed through the computation of time - dependent and scale - dependent indicators .
the data come from a set of 37 woce svp drifters deployed in the brazil current around ( @xmath1s , @xmath2w ) during monitoring by ocean drifters ( mondo ) project , an oceanographic campaign planned by prooceano and supported by eni oil do brasil in relationship with an oil drilling operation .
the experimental strategy of deploying part of the drifters in 5-element clusters , with initial separation between drifters smaller than 1 km , allows to study relative dispersion on a wide range of scales .
single - particle analysis has been performed by computing classic quantities like lagrangian autocorrelation functions and absolute dispersion , defined as the variance around the drifter mean position , as a function of the time lag from the release .
velocity variances ( @xmath178 cm@xmath158 s@xmath180 ) and integral time scales ( @xmath181 days ) are compatible with the estimates obtained in the analysis of the fgge drifters ( figueroa and olson 1989 ; schfer and krauss 1995 ) .
anisotropy of the flow is reflected in the different behavior of the zonal and meridional components of the absolute dispersion .
being the mondo drifters advected mostly by the boundary currents surrounding the subtropical gyre , the brazil current first and the south atlantic current later , the meridional component of the absolute dispersion is dominant as long as the mean drifter direction is nearly southward ( bc ) , while the zonal component dominates ( see the appearance of the late ballistic regime ) as the mean drifter direction is nearly eastward ( sac ) .
early time advection is modulated , also , by the response of the currents to a wind forcing of period @xmath182 days , a characteristic meteorological feature of the bc dynamics .
two - particle analysis has been performed by means of both fixed - time and fixed - scale averaged quantities .
classic indicators like the mean square relative displacement and the relative diffusivities between two drifters as functions of the time lag from the release give loose information about early phase exponential separation , characterized by a mean rate @xmath104 day@xmath91 , and long time dispersion approximated , to some extent , by richardson super - diffusion before the cut - off due to the finite lifetime of the trajectories .
evidence of a small scale exponential regime for relative dispersion is common to other drifter studies for different ocean regions ( lacasce and ohlmann 2003 ; ollitrault et al .
2005 , koszalka et al .
2009 , lumpkin and elipot 2010 ) .
scale - dependent indicators return back a cleaner picture , compatibly with the limited statistics allowed by the experimental data and the non homogeneous and non stationary characteristics of the flow .
the fsle displays a mesoscale [ @xmath183 km ] regime compatible with richardson super - diffusion , lagrangian counterpart of the 2d inverse cascade scenario characterized by a @xmath51 energy spectrum . at scales smaller than 100 km
the fsle has a step - like shape , with a first plateau at level @xmath184 day@xmath91 in the submesoscale range @xmath0 km , and a second plateau at level @xmath123 day@xmath91 for scales comparable with the rossby radius of deformation ( @xmath117 km ) .
constant fsle in a range of scales corresponds to exponential separation .
the @xmath124 plateau could be related to the 2d direct cascade characterized by a @xmath7 energy spectrum , while the origin of the @xmath0 km plateau is likely related to the existence of submesoscale features of the velocity field , the role of which has been recently assessed by means of high - resolution 3d simulations of upper ocean turbulence at rossby numbers @xmath185 ( capet et al . 2008a , b ; klein et al .
the fsle does not display a clean continuous cascade scaling @xmath126 , corresponding to a @xmath8 energy spectrum , from sub to mesoscales ; however , it highlights the existence of scales of motion hardly reconcilable with the qg turbulence scenario , as analogously assessed also by lumpkin and elipot ( 2010 ) for drifter dispersion in the north atlantic .
the fsrv measures the mean square velocity difference at scale @xmath55 .
the scaling of the fsrv is related to the turbulent characteristics of relative dispersion .
the fsrv behavior is , under a certain point of view , cleaner than that of the fsle , but it is affected by the presence of two `` valleys '' , roughly at @xmath186 km and at @xmath187 km , likely associated to trapping events ( the same features are present also in the fsle behavior ) .
coherent structures on scales @xmath162 and @xmath188 km may be responsible of the `` fall '' in the relative dispersion rate .
these scales are of the order of the rossby radius and of the mesoscale rings that detach from the current systems , respectively .
we must also consider the role of the brazil - malvinas confluence which tends to inhibit the growth of the dispersion as the drifters , initially flowing southwestward along the brazil current , encounter the northeastward flowing malvinas current before being , eventually , transported eastward along the south atlantic current .
the fsrv displays a mesoscale @xmath189 scaling ( except for the `` valley '' ) , compatible with a @xmath51 inverse cascade ; a @xmath190 scaling in the two subranges @xmath6 km and @xmath132 km , which correspond to the step - like shape of the fsle ; a @xmath191 scaling which , to some extent , is compatible with a @xmath8 regime connecting scales of the order of the rossby radius to the submesoscale ( below which the velocity field becomes approximately smooth ) .
the equivalent lagrangian spectrum @xmath137 , formed by dividing the fsrv ( essentially the relative kinetic energy at scale @xmath55 ) by the wavenumber @xmath192 , returns back the same scenario in @xmath136 space .
last , we have compared the scale - dependent relative diffusivity @xmath141 constructed from the fsrv , for which @xmath55 is the independent variable , with the classic relative diffusivity @xmath45 seen as a function of the mean separation between two drifters @xmath193 , for which @xmath28 is the independent variable .
what emerges from the analysis of the diffusivities is that , in the mesoscale range , loosely from the rossby radius up to scales @xmath129 km , @xmath194 , corresponding to the @xmath51 inverse cascade ; in the submesoscale range @xmath132 km , and in a limited subrange from about the rossby radius down to @xmath162 km , @xmath195 , corresponding to exponential separation ; in the subrange @xmath133 km , the scaling @xmath196 is compatible with a @xmath8 regime .
the @xmath8 spectrum is a characteristic of upper ocean turbulence when the rossby number is order @xmath0 , as recently assessed with high - resolution 3d model simulations ( capet et al .
2008a , b ; klein et al . 2008 ) , which connects the mesoscale to the submesoscale @xmath197 km ( mcwilliams 1985 ) .
below the submesoscale the velocity field is reasonably assumed to vary smoothly
. a rough estimate of the rossby number associated to the mondo drifter dynamics gives a value ( at least ) @xmath198 , taking @xmath199 m / s , @xmath200 km and @xmath201 1/s , which is , nonetheless , considerably larger than the typical rossby number in open ocean .
although the drifter data analysis does not show a clear evidence of a relative dispersion regime corresponding to the @xmath8 spectrum , the presence of velocity field features of size comparable to submesoscale vortices is reflected , to some extent , by the behavior of the small scale relative dispersion process .
numerical simulations of the lagrangian dynamics have been performed with an ogcm of the south atlantic ( huntley et al .
the results concerning the relative dispersion essentially agree with the data analysis of the mondo drifters , within the limits of the available numerical resolution .
in particular , two - particle statistical indicators such as the fsle , the fsrv and the scale - dependent relative diffusivity @xmath141 , computed on the trajectories of virtual drifters , display the same behaviour found for mondo drifters for scales larger than approximatley @xmath92 km , that is larger than the numerical grid spacing .
below this length scale , evidently , the model flow field is smooth , hence relevant departures from qg turbulence and the role of submesoscale structures can not be assessed .
further investigations on the modeling of submesoscale processes would provide extremely useful in order to make a clearer picture of the small scale dynamics of the surface ocean circulation in the region .
+ fds thanks eni oil do brasil s.a . in the person of ms .
tatiana mafra for being the financial promoter of mondo project and for making the experimental data available to the scientific community .
sb acknowledges financial support from cnrs .
we thank d. iudicone and e. zambianchi for useful discussions and suggestions .
the authors are grateful , also , to three anonymous reviewers who have helped to improve the substance and the form of this work with their critical remarks .
v. artale , g. boffetta , a. celani , m. cencini , a. vulpiani , physics of fluids * 9 * , 3162 ( 1997 ) .
a. t. assireu , _ estudo das caractersticas cinemticas e dinmicas das guas de superfcie do atlntico sul ocidental a partir de derivadores rastreados por satlite _ ( phd thesis , instituto nacional de pesquisas espaciais , 2003 )
. g. k. batchelor , quarterly journal of the royal meteorological society * 76 * , 133 ( 1950 ) .
r. bleck , s. benjamin , monthly weather reviews * 121 * , 1770 ( 1993 ) .
g. boffetta , a. celani , m. cencini , g. lacorata , a. vulpiani , chaos * 10 * ( 1 ) , 50 ( 2000 ) . t. bohr , m. h. jensen , g. paladin , a. vulpiani , _ dynamical systems approach to turbulence _ ( cambridge university press , 1998 )
. e. j. d. campos , j. l. miller , t. j. mller , r. g. peterson , oceanography * 8 * , 87 ( 1995 ) .
x. capet , j. c. mcwilliams , m. j. molemaker , a. f. shchepetkin , journal of physical oceanography * 38 * , 29 ( 2008 ) .
x. capet , j. c. mcwilliams , m. j. molemaker , a. f. shchepetkin , journal of physical oceanography * 38 * , 44 ( 2008 ) .
j. g. charney , journal of atmospheric sciences * 28 * , 1087 ( 1971 ) .
j. a. cummings , quarterly journal of the royal meteorological society * 131 * , 3583 ( 2005 ) .
b. m. de castro , j. a. lorenzetti , i. c. a. da silveira , l. b. de miranda , _ estrutura termohalina e circulao na regio entre o cabo de so tom ( rj ) e o chu ( rs ) _ , in _ o ambiente oceanogrfico da plataforma continental e do talude na regio sudeste - sul do brasil _ , edited by c. l. d. b. rossi - wongtschowski and l. s .- p .
madureira ( edusp , 2006 ) .
a. gordon , c. greengrove , deep sea research part a * 33 * ( 5 ) , 573 ( 1986 ) .
d. v. hansen , p. m. poulain , journal of atmospheric and oceanic technology * 13 * ( 4 ) , 900 ( 1996 ) .
s. houry , e. dombrowsky , p. de mey , j. minster , journal of physical oceanography * 17 * ( 10 ) , 1619 ( 1987 ) .
h. s. huntley , b. l. lipphardt , a. d. kirwan , ocean modelling , doi : 10.1016/j.ocemod.2010.11.001 ( 2010 ) .
p. klein , b. l. hua , g. lapeyre , x. capet , s. le gentil , h. sasaki , journal of physical oceanography * 38 * , 1748 ( 2008 ) .
i. koszalka , j. h. lacasce , k. a. orvik , journal of marine research * 67 * , 411 ( 2009 ) .
r. h. kraichnan , physics of fluids * 10 * , 1417 ( 1967 ) . j. h. lacasce , progress in oceanography * 77 * , 1 ( 2008 ) .
j. h. lacasce , c. ohlmann , journal of marine research * 61 * , 285 ( 2003 ) .
g. lacorata , e. aurell , b. legras , a. vulpiani , journal of the atmospheric sciences * 61 * , 2936 ( 2004 ) .
g. lacorata , e. aurell , a. vulpiani , annales geophysicae * 19 * , 121 ( 2001 ) .
r. legeckis , a. gordon , deep sea research * 29 * , 375 ( 1982 ) . c. lentini , d. olson , g. podest , geophysical research letters * 29 * , 1811 ( 2002 ) .
i. d. lima , c. a. e. garcia , o. o. mller , continental shelf research * 16 * ( 10 ) , 1307 ( 1996 ) .
j. t. lin , journal of atmospheric sciences * 29 * , 394 ( 1972 ) .
r. lumpkin , s. elipot , journal of geophysical research * 115 * , c12017 ( 2010 ) .
j. c. mcwilliams , reviews of geophysics * 23 * , 165 ( 1985 ) .
t. j. mller , y. ikeda , n. zangenberg , l. v. nonato , journal of geophysical research * 103 * ( c3 ) , 5429 ( 1998 ) .
a. m. obhukov , izvestiya akademii nauk sssr , seriya geograficheskaya i geofizichaskaya * 5 * , 453 ( 1941 ) .
l. r. oliveira , a. r. piola , m. m. mata , i. d. soares , journal of geophysical research * 114 * ( c10 ) , c10006 ( 2009 ) .
m. ollitrault , c. gabillet , a. c. d. verdiere , journal of fluid mechanichs * 533 * , 381 ( 2005 ) .
r. g. peterson , l. stramma , progress in oceanography * 26 * ( 1 ) , 1 ( 1991 ) .
l. p. pezzi , r. b. souza , m. s. dourado , c. a. e. garcia , m. m. mata , m. a. f. silva - dias , geophysical research letters , * 32 * ( 22 ) ( 2005 ) .
a. piola , h. figueroa , a. bianchi , journal of geophysical research * 92 * ( c5 ) , 5101 ( 1987 ) .
a. piola , o. o. mller , r. a. guerrero , e. j. d. campos , continental shelf research * 28 * ( 13 ) , 1639 ( 2008 ) .
l. f. richardson , proceedings of the royal society a * 110 * , 709 ( 1926 ) .
h. schfer , w. krauss , journal of marine research * 53 * , 403 ( 1995 ) .
j. stech , j. lorenzetti , journal of geophysical research * 97 * ( c6 ) , 9507 ( 1992 ) . m. stevenson , woce newsletter * 22 * , 1 ( 1996 ) .
l. stramma , m. england , journal of geophysical research * 104 * ( c9 ) , 20863 ( 1999 ) .
a. sybrandy , p. p. niiler , _ woce / toga lagrangian drifter construction manual _ ( university of california , 1991 ) .
g. i. taylor , proceedings of the london mathematical society * 20 * , 196 ( 1921 ) .
l. n. thomas , a. tandon , a. mahadevan , _ submesoscale processes and dynamics _ , in _
eddy resolving ocean modeling _ , edited by m. w. hecht , h. hasumi , amer .
union , 17 ( 2008 ) . | in the framework of monitoring by ocean drifters ( mondo ) project , a set of lagrangian drifters were released in proximity of the brazil current , the western branch of the subtropical gyre in the south atlantic ocean .
the experimental strategy of deploying part of the buoys in clusters offers the opportunity to examine relative dispersion on a wide range of scales . adopting a dynamical systems approach , we focus our attention on scale - dependent indicators , like the finite - scale lyapunov exponent ( fsle ) and the finite - scale ( mean square ) relative velocity ( fsrv ) between two drifters as function of their separation , and compare them with classic time - dependent statistical quantities like the mean square relative displacement between two drifters and the effective diffusivity as functions of the time lag from the release .
we find that , dependently on the given observable , the quasigeostrophic turbulence scenario is overall compatible with our data analysis , with discrepancies from the expected behavior of 2d turbulent trajectories likely to be ascribed to the non stationary and non homogeneous characteristics of the flow , as well as to possible ageostrophic effects .
submesoscale features of @xmath0 km are considered to play a role , to some extent , in determining the properties of relative dispersion as well as the shape of the energy spectrum .
we present , also , numerical simulations of an ogcm of the south atlantic , and discuss the comparison between experimental and model data about mesoscale dispersion . |
galactic cosmic rays ( gcr ) beyond the heliosphere region are considered to be temporally and spatially isotropic at least over large timescales .
they enter into the heliosphere due to random motions and they are subjected to a number of transport effects such as diffusion , adiabatic cooling , convection and drift @xcite .
the relative importance of these processes varies with particle properties , such as energy , timescales and of course the position in space .
the size and structure of the modulation volume is assumed to be approximately a sphere with a radius of @xmath3 au . on the other hand ,
the term `` solar energetic particles '' ( sep ) include all particles in the heliosphere accelerated by processes related to solar activity , such as the anomalous cosmic ray ( acr ) , particles accelerated in corotating interaction region ( cir ) , as well as particles accelerated in solar flares and coronal mass ejection ( cme ) . while , it is believed that particles , continuously expelled by the sun , such as the solar wind plasma , have energies up to several gevs in the tail of the solar energy spectrum and only during transient solar events ( i.e. solar flares ) their energies can reach dozens of gevs .
so far , the particle acceleration mechanism by the sun in a wide band of energies , specially the high energy region , are still poorly understood .
the survey on modulation of cosmic ray particles by the interplanetary magnetic field and the correlation with solar activity began using ground based cosmic ray detectors in the thirties . due to the complexity
, details of the phenomena are still subjected to studies .
interaction of the primary cosmic rays with the atmosphere produce , among other things , a lower energy secondary nucleons , in particular , neutrons that are not slowed by ionization loss .
these secondaries fall in the energy range of a few hundred mev up to about @xmath4 gev .
these nucleons in turn produce further nuclear interaction , either in the atmosphere or in lead target materials surrounding the detectors , in most of cases the so called neutron monitors ( nms ) .
the interaction rate may be measured most conveniently and reliably by detecting the reaction products in neutrons rather than by detecting the charged fragments directly .
the nms worldwide network starting from 1954 by simpson @xcite has shown excellent performances because the intensities are recorder to several geomagnetic cutoffs and anisotropies and other characteristic can be better known .
one of the main obtained results using nms is the long term variation , the cosmic ray intensity increase and decrease with the solar cycle .
they are in anti - correlation with the number of solar spots . on average , every eleven years , solar activity is high and therefor cosmic rays are deflected stronger than during minimum solar activity .
consequently , when the sun is active , fewer galactic cosmic rays reach earth s atmosphere .
the anti - coincidence guard counting rate of gcrs in spacecrafts ( 1972 - 2002 ) have confirmed this long term variation @xcite .
on the other hand , temporal variations of the cosmic ray intensity as an abrupt intensity decrease at ground level were observed already in the thirties by forbush @xcite .
these `` forbush '' events are associated to the passage at the earth s vicinity of a disturbance ( shock and plasma ) causing a shielding effect . at least in the case of large forbuch events ,
the disturbance `` eject '' is emitted by the sun during the coronal mass ejection process .
in addition , near the sun s equatorial plane high and low speed solar wind flows interact . this interaction is known as corotating interaction region ( cir ) .
there are forward and reverse shocks bounding these regions , which are known to modulate galactic cosmic rays .
abrupt depressions in the gcr intensity were also observed in the vicinity of the maximum solar wind speed by iucii and coworkers @xcite analyzing neutron monitor data and later confirmed by spacecraft experiments .
these cosmic ray modulations are associated with corotating high - speed streams and cirs .
another important result obtained by ground based cosmic ray detectors , nms , as well as underground muon telescopes is the short term variation , known as the solar diurnal variation or daily anisotropy of the cosmic ray intensity .
it has been observed in detectors located at different global sites and in a wide range of the cosmic ray spectrum , rigidities between 1 gv to 500 gv .
the solar diurnal variation is attributed to the bulk streaming of the cosmic ray gas caused by the corotating interplanetary magnetic field that is rigidly attached to the sun and it is related in terms of diffusion , convection , and drift of gcr in the imf @xcite . however , at high rigidities ( above 10 - 20 gv ) the standard convection - diffusion picture for gcr is inaccurate and probably inapplicable , because the concept of a diffusive streaming breaks down @xcite .
consequently , the exact nature of galactic cosmic ray contribution to the solar daily anisotropy is not yet clear .
a completely different sort of information on primary cosmic ray comes from the tupi experiment , located at sea level .
starting from april of 2007 this experiment consists of two identical directional muon telescopes , constructed on the basis of plastic scintillators .
one of them with vertical orientation and another one with an orientation of 45 in relationship of the vertical ( zenith ) and pointing at the west . both with an effective aperture of @xmath5 .
the rigidity of response of these detectors to cosmic proton spectrum is above 0.4 gv , allowing registration of muons with @xmath0 .
the daily variation of muon intensity , in most cases , consists of a high muon intensity observed between 9 hours and 18 hours ( local time ) , and a low muon intensity , up to ten times smaller than the high muon intensity , is observed in the remaining 13 hours .
this behavior on the daily variation is subjected to several fluctuations of different sort , such as magnetic disturbances in the heliosphere inside the interplanetary space and near the earth . due to the abrupt transition between the low to high muon intensity and vice - verse , as well as
, the region of high muon intensity @xmath6 ( local time ) coincides with the period where the imf lines overtake the earth , we argue that the high muon intensity is due to a contribution of solar protons ( ions ) on the muon intensity and the galactic cosmic rays are responsible of the low muon intensity . if this picture is right the solar particle spectrum extend to energies beyond 1 gev , and probably up to @xmath7 gv in the tail of the spectrum , because the daily variation has been observed besides in this high rigidity range .
this scenario can explain in a natural way the solar daily anisotropy , without the co - rotating assumption of galactic cosmic rays with the imf inside the standard convection - diffusion picture , hard to be applied in the high rigidity region .
our conclusion is on the basis of the average daily muon intensity with @xmath0 produced in the earth s atmosphere by protons ( ions ) with rigidity above 0.4 gv over a period of two months , and coincides with the actual period of minimum solar activity .
obtained results are in agreement with the data reported in others papers .
the tupi experiment is a muon tracking telescope located at sea level , and whose coordinates are : @xmath8 latitude and @xmath9 longitude .
observations on the muon flux and muon enhancements in association with solar transient events have been reported @xcite .
starting from april of 2007 we have initiated the phase ii of the tupi experiment with two identical muon telescopes on the basis of plastic scintillators , one of them with vertical orientation and another one with an orientation of 45 in relationship of the vertical ( zenith ) and pointing at the west , both with an effective aperture of @xmath5 .
the telescopes are inside a building under two flagstones of concrete and allowing registration of muons with @xmath0 required to penetrate the two flagstones , fig.1 summarized the situation .
the rigidity of response of these detectors to cosmic proton ( ions ) spectrum is given by the local geomagnetic cutoff 0.4 gv .
this low value is due to the brazilian magnetic anomaly ( see section 4 ) .
the directionality of the muon telescopes is guaranteed by a veto or anti - coincidence guard as is shown in fig.2 .
the vertical telescope has a veto or anti - coincidence guard , using a detector of the inclined telescope and vice - verse .
therefore , only muons with trajectories close to the telescope axis are registered .
the data acquisition is made on the basis of the advantech pci-1711/73 card with an analogical to digital conversion at a rate of up to 100 khz .
the primary cosmic ray particles ( i.e. protons and nuclei ) can be inferred through the detection of muons by telescopes at ground and underground levels .
the upper layers of the earth s atmosphere is bombarded by a flux of cosmic primary particles .
the chemical composition of this primary cosmic particles depend on the energy region . in the low energy region ( above 1 gev to several tev ) the dominant particles are protons ( @xmath10 ) .
the primary cosmic rays collide with the nuclei of air molecules and produce an air shower of particles that include nucleons , charged and neutral pions , kaons etc .
these secondary particles then undergo electromagnetic and nuclear interactions to produce yet additional particles in a cascade process
. of particular interest is the fate of charged pions , @xmath11 , produced in the cascade .
some of these will interact via the strong interaction with air molecule nuclei but others will spontaneously decay via the weak interaction into a muon , @xmath12 , plus a neutrino or anti - neutrino , @xmath13 , following the scheme @xmath14 muons are quite penetrating and they can reach the ground , enter the laboratory through the walls or roof of the building , and be detected with a suitable apparatus and very high energy muons reaches underground levels .
the expected energy spectra of muons @xcite at sea level at two zenith angles , @xmath15 and @xmath16 are shown in fig.3 .
for muons with energies around @xmath17 the vertical intensity @xmath15 is twice of the intensity at @xmath18 .
this behavior is a consequence of the zenith angle distribution of muons in the gev region , to be related as @xmath19 with the value of @xmath20 for @xmath16 .
exactly , this is the reason for the tupi experiment using two identical telescopes , a vertical one and another one inclined in an angle of @xmath21 .
we have observed that in average the counting rate in the vertical telescope is twice the counting rate in the inclined telescope as is shows in fig.4 ( upper panel ) , where the 5 min muon counting rate , observed in both telescopes vertical and inclined , before and after pressure correction and for two consecutive days 3th and 4th april 2007 , is presented . on the average ,
the counting rate on the vertical direction is two times higher than the counting rate on the inclines ( @xmath21 ) direction , as the two telescopes have the same aperture @xmath22 , the intensity in the vertical direction is twice than intensity in the inclined direction .
this behavior is used as a `` quality control '' in the output time series . from this figure , it is possible to see that the influence of the atmospheric pressure variation ( see fig.4 lower panel ) on the muon intensity at sea level is not relevant .
ground - level solar cosmic ray events are usually observed by high latitude neutron monitors at relatively low rigidities ( @xmath23 gv ) and in most cases the ground - level events are linked with solar flares of high intensity whose prompt x - ray emission is cataloged as x - class ( above @xmath24 ) evidently , the solar flare detection at ground depends on several aspects such as a good magnetic connection between the sun and earth . in the new array ( phase ii ) of the tupi experiment ,
constituted by two directional muon telescopes and during the present survey for diurnal anisotropies , it were found several muon enhancements .
the event presented here is clearly distinguishable above the muon background , with statistically significance of @xmath25 in temporal coincidence with a flare of small scale whose prompt x - ray emission is cataloged by the goes11 as c3 class on april 16 , 2007 at 22.8 h ut ( see fig.5 ) .
this is the first detection of a ground - level solar cosmic ray event linked with the impulsive or prompt emission of energetic particles emitted during a very small solar flare ( it is the record in the detection at ground of the smallest solar flare ) .
the tupi telescope had already detected before events tied with the gradual emission via cme of a flare of small scale , see ref .
@xcite .
the high sensitivity of the tupi telescopes can be in part a consequence of the brazilian magnetic anomaly , with a minimum at 26s , 53w , which is very close to the position 22s , 43w where the tupi experiment is located .
the lowest magnetic field in the region of the anomaly at ground is three times lower than the magnetic field at polar regions .
then , the area of the magnetic anomaly is like a funnel for incoming charged particles from space .
fig.6 summarized the situation , where the contours of the omni - directional proton intensity ( @xmath26 mev ) around the anomaly region is plotted @xcite .
the brazilian magnetic anomaly is responsible for the low rigidity response ( @xmath27 gv ) of the tupi telescopes to incoming protons ( ions ) from space . without the magnetic anomaly , the geomagnetic cutoff at latitude where the tupi telescopes are located is around @xmath28 .
we believe that the presence of the brazilian magnetic anomaly , on the place of the tupi experiment is responsible at least partly for the results obtained about the daily variation on the muon intensity and that will be presented in the section 6 .
a more accurate conclusion on this subject require further study .
the solar daily variations known also as the diurnal solar anisotropy of the cosmic ray intensity have been observed by ground and underground based detectors , covering a wide range of the cosmic ray spectrum , rigidities between 1 gv to 400 gv .
the anisotropy reflects the local interplanetary cosmic ray distribution .
it is widely believed that the anisotropy arises when the galactic cosmic ray particles ( gcr ) co - rotate with solar wind stream following the interplanetary magnetic field ( imf ) lines , and it is related in terms of diffusion , convection and drift of galactic cosmic ray in the imf . according to forman and gleeson @xcite at 1 au the co - rotating stream ( solar wind )
has a speed of order 450 @xmath29 ( in average ) and at @xmath30 ( local time ) approximately in the same direction as the earth s orbital motion ( of 30 @xmath29 ) .
in other words , the co - rotating stream will overtake the earth `` almost in the vertical '' from the direction of @xmath31 , and it is known as the phase of the anisotropy .
this phase have shifted toward earlier hours @xmath32 during the lower solar cycle due to drift process .
so far , the drift process is still the most likely and accepted explanation of the phase shift .
however , the solar daily variation specially observed in the high rigidity region have shown remarkable changes in phase and amplitude during long period of observation .
a review of characteristics of the observed cosmic ray diurnal variation over three decades has been reported by ahluwalia and fikani @xcite .
they conclude that the maxima and minima in the amplitude and phases exhibit features related to the solar activity , and the amplitude decrease systematically with increasing value of rigidity .
we have extracted from this survey the maxima and minima in the phases obtained with the deep river ( @xmath33 ) and huancayo ( @xmath34 ) neutron monitors ( nms ) and the embudo ( @xmath35 ) underground muon telescope ( mt ) for the 1965 to 1994 period , as well as , we have included the results of the nagoya ( @xmath36 ) muon telescope ( mt ) reporter in the mori s review paper @xcite , for the 1971 to 1994 period . from these results
we conclude that the low rigidity diffusive - convective stream picture predictions as @xmath37 and @xmath32 for the maximum and minimum phase of the anisotropy becomes inaccurate in the high rigidity region .
the situation becomes critical analyzing the data of the nagoya muon telescope since both values , minimum and maximum of the phase are completely out of two theoretical prediction .
these results strongly constrain the assumption of corotating galactic cosmic ray as the origin of the anisotropy . or at least the exact nature of galactic cosmic ray contribution to the solar daily anisotropy is not yet clear .
the above results show that the phase of the solar daily anisotropy is distributed in a wide range from @xmath38 up to @xmath39 .
this behavior of the phase distribution agree with our results on the basis of muon intensity registered at sea level by two directional muon telescopes and they are presented in the next section .
starting from april of 2007 we have initiated the phase ii of the tupi experiment , with a survey on the daily variation of the muon intensity at sea level using two identical muon telescopes at sea level as described in section 2 .
the method applied here to study the cosmic ray anisotropy is based on the idea that a fixed detector scans the sky due to the earth s rotation . in each 0ne of these figures ,
fig.7 , fig.8 , fig.9 and fig.10 , we show representative data on the daily variation of the muon intensity , observed in five consecutive days . for the particular period shown in fig.7
, we can see in the first three days on 2 , 3 , and 4 of may 2007 , there are two levels in the muon intensity : a high muon intensity level between the @xmath40 ut and @xmath41 ut and correspond to 9h lt and 18h lt and a low muon intensity in the remaining hours .
hereafter , the period of high intensity 12h to 18h will be called with the jargon `` solar window to muons '' the blue and red lines marked as tupi 4 and tupi 2 respectively are the muon intensities registered in the vertical direction , before and after atmospheric pressure correction and the black and green lines marked as tupi 1 and tupi 3 are the muon intensities registered on the inclined direction ( @xmath21 ) , before and after pressure correction .
this figure caption is also valid for fig.8 and fig.10 . in all cases
the tupi muon intensities are compared with the solar proton intensity in the mev energy region , registered by the sis detector on board of the ace spacecraft and located in the l1 ( lagrange ) point @xcite .
the orange and magenta lines marked as sis 1 and sis 2 represent the sis integral solar proton flux for energies above 10 gev and 30 gev respectively .
some important features observed in these figures are : \a ) as already commented , the influence of the atmospheric pressure variation in the muon intensity at sea level is not relevant .
+ b ) in both muon intensity levels , high and low , the vertical muon intensity is around twice higher than the muon intensity impacting with @xmath21 .
this behavior is in agreement with the expected zenith dependence like @xmath19 for the muon intensity in the gev region . + c ) the transition between the low to high muon intensity and vice - verse is abrupt . in most cases ,
the high level muon intensity is up to 10 times higher than the low level muon intensity .
+ d ) there are days where the muon intensities remain in the low level during the 24h .
the opposed situation is also observed , the muon intensity remain in the high level .
however , in this case , it is observed muon intensities fluctuate exactly in the `` solar window to muons '' limits ( 9h and 18h local time , see fig.8 and fig.10 ) .
these changes , probably are associated to transient changes on the mechanism of solar energetic particle emission , modifying the solar particle spectrum , as well as due to magnetic field fluctuations in the interplanetary space that include the magnetic disturbances on the earth s magnetic field , increasing the geomagnetic cutoff ( shielding effect ) or decreasing the geomagnetic cutoff ( enhancement of particles with low rigidity ) .
+ e ) in fig.9 , we have included also for comparison the solar proton flux in the kev energy region as observed by the epam detector on board of the ace spacecraft and located at l1 . on 2007
april 30 , the tupi telescopes have detected a muon enhancement in coincidence with the arrival of a large kev proton burst in the epam detector .
this means that the energy spectrum of the proton burst extend beyond 1 gev , because they can produce muons in the earth s atmosphere .
however , this proton enhancement was not registered by sis detector ( mev region ) located in the same ace spacecraft . in addition
, the solar proton intensity in the mev region as is observed by ace - sis detector has been very stable , at least during the period the observation of this survey and its small fluctuations happen in coincidence3 with the tupi muon fluctuations as is shown in fig.10 on 2007 april 26 .
\f ) the solar proton intensity as detected on ace sis detector ( above 10 - 30 mev region ) is around 100 times higher than the tupi high level muon intensity and correspond to primary particles with rigidities above 0.4 gv ( 0.4 gev for protons ) under the assumption that the high intensity muons are produced by solar protons ( ions ) , we have estimated the intensity of solar protons ( ions ) with rigidity above 1 gv . a typical representation ( under quiet conditions ) of the intensities ( epam , sis and tupi )
is presented in fig.9 on 2007/04/27 and the integral energy spectrum is shown in fig.11 . as expected
its spectrum is close to a power law , with an integral spectral index of @xmath42 .
this behavior of the energy spectrum strongly suggest that the tupi high level muons are produced by solar particles from the tail of the energy spectrum .
+ g)in order to compact all observation , the hourly muon intensity at sea level for @xmath0 , averaged over two months is shown in the upper panel of fig.12 , as well as , the daily variation of the amplitude of the first harmonic for the muon intensity defined as @xmath43 averaged over two months is shown in the lower panel . in the `` solar window to muons '' region ( @xmath1 until @xmath2 ) there are two peaks . the first peak is linked to the amplitude and phase of the first harmonic of the diurnal anisotropy as 20% and 13h respectively .
the second peak at 18h is linked to the east - west anisotropy due to the earth s magnetic field .
the high value for the amplitude obtained here , corresponds to the minimum solar activity period and must be smaller when data obtained on the maximum solar activity period are included .
the `` solar window to muons '' region include all the values of the maxima and minima phase observed by neutron monitors and muon telescopes ( see table i ) . another important feature , and crucial for the formulation of a new picture for the origin of the solar daily variation , is the verification that in the region coverted by the `` solar window to muons '' the interplanetary magnetic lines overtake the earth . the situation is summarized in fig.13 .
all imf lines inside of shaded region overtake the earth . on the other hand
, we have used the fast fourier transformation ( fft ) for studying periodicities and scaling properties that might be present in time series constructed using the hourly muon intensities .
the power spectra of the hourly muon intensities for two months are show in fig.14 . in this case
there is a series of peaks , such as at 0.99 days ( daily anisotropy ) and a small peak also can be observed at 0.58 days known as the semi - diurnal anisotropy .
the harmonic @xmath44 , with @xmath45 giving a peak at @xmath46 days and correspond to the quasi - periodic corotating streams that occurs due to solar rotation period of 27 days .
we would like point to out that the harmonics like @xmath44 has been observed in the power spectra of solar wind speed measurements reported by burlaga and lazarus @xcite .
these results strongly suggest that the solar wind and the protons producing the high muon intensity have a common origin , the sun .
finally , in a previous article @xcite results obtained with the tupi telescope working on raster scan regime ( telescope always pointed to the imf lines ) were presented , where sudden depressions in the muon intensity were observed .
initially , the events were interpreted as mini - forbush events , a shielding effect due to the passage for the earth vicinity of a small interplanetary disturbance .
the data shows that the depressions in most of the cases began at @xmath41 ut and that corresponds to at 18h hours ( local time ) .
now it is clear , the depressions are the exit of the `` solar windows for muons '' .
the observations correspond to 2004 and 2005 periods .
now , 2007 in the minimum of solar activity , the abrupt transitions between high and low levels of the muon intensity , are more frequent .
the observation of the solar daily variation of the cosmic ray intensity in a wide region of rigidities is a challenge to phenomenological models to explain the phenomenon . despite there are some mechanisms to explain the effect especially in the low rigidity band
, none of the possibilities can explain the daily variation in all the cases .
therefore , the cosmic ray daily variation is still an open question .
the solar daily variation of the cosmic ray particle intensity is believed to be due to the corotating galactic cosmic rays with the interplanetary magnetic field lines .
however , the data of solar daily variation , especially in the high rigidity region obtained by the underground secondary cosmic ray intensity has shown remarkable changes in phase and amplitude during several decades of observation @xcite and @xcite , hard to explain under the assumption of the standard diffusion - convection picture of galactic cosmic rays . in this paper
we have presented new results on the daily variations of cosmic ray muons with @xmath0 based on the data of two directional muon telescopes at sea level and with a rigidity of response to cosmic proton spectrum above 0.4 gv .
abrupt transition between a low to a high muon intensity ( up to @xmath47 times higher ) and vice - verse were observed , the period of high muon intensity is from @xmath1 until @xmath2 ( local time ) and coincides with the period in which the solar magnetic lines intercept the surface of the earth .
this behavior strongly suggest that the high muon intensity level is produced by protons whose origin is the sun and implying that the solar proton energy spectrum extends beyond 1 gev .
this assumption is reinforced by some important experimental features such as the solar proton energy spectrum obtained by direct observations by detectors on board of the ace spacecraft , epam protons(kev band ) and sis protons ( mev band ) and which can be extended to higher energies following a power law by the tupi protons ( gev band ) on the basis of the high muon intensity .
the fft power spectra of the hourly muon intensities have a series of peaks , such as the harmonic @xmath44 , with @xmath45 giving a peak at @xmath46 days and correspond to the quasi - periodic corotating streams that occurs due to solar rotation period of 27 days .
these same harmonics have been found in the power spectra of solar wind speed measurements , suggesting a common origin by the solar wind and the protons producing the high muon intensity .
the assumption of a solar origin for energetic particles , as responsible for the diurnal intensity variation , takes into account also the phase shift of diurnal variation at reversal of the solar magnetic field .
if the origin of the energetic charged particle is the sun , their propagation follows the interplanetary magnetic field lines .
in addition , the structure of the interplanetary magnetic field depends on the expansion speed of the plasma ( solar wind ) expelled by the sun .
fig.13 summarized the situation , where the solar magnetic field lines between the sun and the earth for several values of the speed of solar wind is shown @xcite . in the period of low solar activity
, the average speed of the solar wind increases , and the magnetic lines and consequently the energetic charged solar particle propagation is close to the earth - sun line .
the result is observed as a shift of the phase of the anisotropy to early hours @xcite .
the so called `` solar window to muons '' ( @xmath48 until @xmath49 ) includes all values of the anisotropy phase , observed in a wide range of rigidity ( see table i ) .
however , some results @xcite indicate that the maximum amplitude in the range of ( 1 - 100)gv , observed with narrow - angle detectors at the equator , would be 0.4 per cent of the average primary flux and independent of the rigidity .
this disagrees with our results where an amplitude of @xmath50 is observed ( see fig.11 ) .
even so , it must be smaller when results obtained in the high solar activity period are included . the high amplitude observed in this work
can be a consequence of the brazilian magnetic anomaly region , where the lowest magnetic region located at 26s , 53w , is very close to the position 22s , 43w where the tupi experiment is located .
the contours of the omnidirectional proton intensity ( @xmath26 mev ) around the anomaly region is ploted in fig.14 @xcite .
the lowest magnetic field in the region of the anomaly is three times lower than the magnetic field at polar regions .
our data also suggested that the amplitude of the diurnal variation would be dependent on the rigidity .
this conclusion is in accordance with previous results where the amplitudes decrease systematically with increasing rigidity @xcite .
the second peak in the amplitude at @xmath30 ( see fig.12 lower panel ) is linked with the east - west anisotropy due to earth s magnetic field , because particles of rigidity below 60 gv are subject to the geomagnetic effect .
we would like to point out , that the tupi results are on the basis of two muon telescopes with a high sensitivity , capable to observer , under certain conditions , muon enhancement in association with solar flares of small scale ( type c ) and muon enhancements in coincidence with direct observation of solar protons epam and sis on board of the ace spacecraft .
this work is supported by the national council for research ( cnpq ) in brazil , under grant no .
the authors wish to express their thanks to dr .
a. ohsawa from tokyo university for help in the first stage of the experiment .
we are also grateful to the various catalogs available on the web and to their open data police , especially to the ace real - time solar wind ( rtsw ) data .
e. n. parker , interplanetary dynamical processes , monographs and texts in phys . and astro . , 8 , inter - science publishers : new york , new york ( 1963 )
. j. simpson , phys .
94 , 426 ( 1954 ) .
i. a. richardson , space science rev .
, 111 , 267 ( 2004 ) .
s. c. forbush , phys .
51 , 1108 ( 1937 ) .
n. iucci , m. parise , m. storine and g. villoresi , il nuovo cimento , 2c , 1 ( 1979 ) .
a. forman , l. j. gleeson , astrophys .
space sci .
j. kota , proc .
cosmic ray conf . 2 , 145 ( 1997 ) .
c. r. augusto , c. e. navia and m. robba , nucl .
methods phys .
503 , 554 ( 2003 ) .
c. e. navia , c. r. a. augusto , m. b. robba , m. malheiro and h. shigueoka , astrophys
. journal .
621 , 1137 ( 2005 ) . c. r. a. augusto ,
c. e. navia and m. b. robba , phys .
d71 , 103011 ( 2005 ) .
p. j. green et al .
d20 , 1598 ( 1979 ) .
p. lipari , astr .
, 1 , 195 ( 1993 ) .
f. nichitiu , available in + @xmath52 h. s. ahluwalia and m. m. fikani , proc .
cosmic ray conf . 2 , 125 ( 1997 ) .
s. mori , il nuovo cimento , vol.19 , 791 ( 1996 ) .
ace real - time solar wind data website , world wide wide : http://www.sec.noaa.gov/ace/ l. f. burlaga and a. j. lazarus , j. geophys .
research , v.105 , 2357 ( 2000 ) . c. e. navia , c. r. a. augusto , k. h. tsui and m. b. robba , phys .
d72 , 103001 ( 2005 ) .
d. s. peacock and t. thanbyahpillai , nature 215 , 146 ( 1967 ) .
helios vocca group , solar magnetic field through lisa and earth , available in + @xmath53 .
r. m jacklyn , nature 200 , 1306 ( 1963 ) .
.the times for the maxima and minima of the diurnal variation ( phase ) observed in four global sites .
for the 1965 to 1994 period ( deep river , huancayo and embudo ) and for the 1971 to 1994 period ( nagoya ) [ cols="^,^,^,^,^",options="header " , ] | analysis on the daily variations of cosmic ray muons with @xmath0 based on the data of two directional muon telescopes at sea level and with a rigidity of response to cosmic proton spectrum above 0.4 gv is presented .
the analysis covers two months of observations and in 60% of days , abrupt transitions between a low to a high muon intensity and vice - verse is observed , the period of high muon intensity is from @xmath1 up to @xmath2 ( local time ) and coincides with the period when the interplanetary magnetic field ( imf ) lines overtake the earth .
this behavior strongly suggest that the high muon intensity is due to a contribution of solar protons ( ions ) on the muon intensity produced by the galactic cosmic rays , responsible for the low muon intensity .
this implies that the solar particle spectrum extends to energies beyond 1 gev .
we show that this picture can explain the solar daily variation origin , and it is a most accurate scenario than the assumption of corotating galactic cosmic ray with the imf lines , specially in the high rigidity region .
obtained results are consistent with the data reported in others papers . some aspects on the sensitivity of our muon telescopes are also presented . |
topological open strings and topological d - branes have recently been enjoying the attention of both physicists and mathematicians .
the most obvious physical motivation for studying topological string theory is that it is a toy - model for `` physical '' string theory .
thus a better understanding of topological d - branes could shed light on the general definition of a boundary condition for a two - dimensional conformal field theory ( 2d cft ) , something which is not known at present .
further , if a 2d topological field theory ( 2d tft ) is obtained by twisting a 2d supersymmetric field theory , then it is possible to regard topological d - branes as a special class of `` physical '' d - branes ( bps d - branes ) .
in fact , much of recent progress in string theory has resulted from studying bps d - branes . from the mathematical viewpoint ,
topological string theory is an alternative way of describing certain important geometric categories , such as the category of coherent sheaves on a calabi - yau manifold , and can serve as a powerful source of intuition .
an outstanding example of such intuition is the homological mirror symmetry conjecture @xcite .
most works on topological string theory considered the case of topologically twisted @xmath0 sigma - models @xcite with a calabi - yau target space .
this is the case when the world - sheet theory is conformal , and topological correlators can also be interpreted in terms of a physical string theory @xcite .
however , one can also consider more general topologically twisted @xmath0 field theories and the corresponding d - branes .
one class of such theories is given by sigma - models whose target is a fano variety ( say , a complex projective space , or a complex grassmannian ) .
such qfts , although conformally - invariant on the classical level , have non - trivial renormalization - group flow once quantum effects are taken into account .
another set of examples is provided by @xmath0 landau - ginzburg models ( lg models ) @xcite .
in fact , in many cases these two classes of @xmath0 theories are related by mirror symmetry @xcite .
for example , the sigma - model with target @xmath1 is mirror to a landau - ginzburg model with @xmath2 fields @xmath3 taking values in @xmath4 , and a superpotential @xmath5 thus if one wants to extend the homological mirror symmetry conjecture to the non - calabi - yau case , one needs to understand d - branes in topologically twisted lg models .
note that all critical points of this superpotential are isolated and non - degenerate ; this means that all the vacua have a mass gap , and the infrared limit of this lg model is trivial . in what follows we will call such lg models _
massive_. despite the triviality of the infrared limit , the homological mirror symmetry conjecture remains meaningful and non - trivial in this case .
very recently it has been proposed that massive @xmath0 @xmath6 qfts can be used to describe certain non - standard superstring backgrounds with ramond - ramond flux @xcite .
thus a study of d - branes in massive qfts could be useful for understanding open strings in such ramond - ramond backgrounds . in order to formulate our problem more concretely , let us first summarize the situation in the calabi - yau case , where the @xmath0 field theory is conformal .
@xmath0 superconformal field theories have two topologically twisted versions : a - model and b - model @xcite .
the corresponding d - branes are called a - branes and b - branes .
mirror symmetry exchanges a - branes and b - branes .
tree - level topological correlators give the set of either a - branes or b - branes the structure of an @xmath7-category ; gauge - invariant information is encoded by the corresponding derived categories .
it has been argued that the derived category of b - branes is equivalent to the derived category of coherent sheaves @xcite .
a detailed check of this proposal has been performed in ref .
@xcite .
for a - branes on calabi - yau manifolds , it has been proposed that the relevant @xmath7-category is the so - called fukaya category , whose objects are ( roughly ) lagrangian submanifolds carrying vector bundles with flat connections @xcite . recently it has been shown that the derived fukaya category is too small and does not accommodate certain physically acceptable a - branes @xcite . in particular ,
if we want the homological mirror symmetry conjecture to be true for tori , then the fukaya category must be enlarged with non - lagrangian ( more specifically , coisotropic ) branes . in the case of fano varieties ,
the sigma - model is not conformal .
what is more important , the axial @xmath8 r - current is anomalous , and therefore one can not define the b - twist @xcite .
one _ can _ consider d - branes which preserve b - type supersymmetry , but the relation with the derived category of coherent sheaves is less straightforward @xcite .
mirror symmetry relates b - branes on fano varieties with a - branes in lg models .
the latter have been studied from a variety of viewpoints in refs .
@xcite . in the case
when the fano variety is @xmath1 , the prediction of mirror symmetry has been tested in ref .
in particular , the mirrors of `` exceptional '' bundles on @xmath1 have been identified , and in the case of @xmath9 and @xmath10 it has been checked that morphisms between these bundles in the derived category of coherent sheaves on @xmath1 agree with the floer homology between their mirror a - branes .
one can also consider the category of a - branes on a fano manifold . since the vector @xmath8 r - current is not anomalous ,
the a - twist is well - defined , and a - branes can be regarded as topological boundary conditions for the a - model .
presumably , the category of a - branes contains the derived fukaya category as a subcategory , but other than that little is known about it , even in the case of @xmath1 .
if we assume mirror symmetry , we can learn about the category of a - branes on @xmath1 by studying b - branes in the mirror lg model .
the b - twist is well - defined for any lg model whose target has a trivial canonical bundle , thus b - branes in such a lg model can be regarded as topological boundary conditions for the b - model .
an obvious question is how the introduction of the superpotential deforms the relation between the category of b - branes and the derived category of coherent sheaves .
important steps towards understanding b - branes in lg models have been taken in refs .
@xcite ( see also refs .
@xcite for a related work ) . in these papers
general properties of b - branes have been studied , and several concrete examples have been discussed .
a somewhat surprising lesson from these works is that the category of b - branes remains non - trivial even in a massive lg model , where the bulk 2d tft is trivial .
for example , if we take the superpotential @xmath11 to be a non - degenerate quadratic function on @xmath12 , the graded algebra of endomorphisms of a d0-brane sitting at the critical point of @xmath11 is isomorphic to a clifford algebra with @xmath2 generators @xcite .
this raises the question if one can determine the category of b - branes in any massive lg model . a proposal which accomplishes
this has been put forward by m. kontsevich .
roughly speaking , the proposal is that the superpotential @xmath11 deforms the derived category of coherent sheaves by replacing complexes of locally free sheaves with `` twisted '' complexes . here
`` twisted '' means that compositions of successive morphisms in a complex are equal to @xmath11 , instead of zero .
one also needs to switch from @xmath13-graded complexes to @xmath14-graded ones .
kontsevich s proposal is supposed to describe b - branes in any lg model such that the critical set of @xmath11 is compact ; in particular , it does not require the critical points of @xmath11 to be non - degenerate .
the main goal of this paper is to provide evidence for kontsevich s proposal .
our evidence is of two kinds .
first , we argue on physical grounds that twisted complexes arise as a consequence of brst - invariance . more precisely , while in the presence of the superpotential a holomorphic vector bundle or a complex of vector bundles
does not correspond to a b - type boundary condition , we show that any twisted complex of holomorphic vector bundles is a valid b - brane .
second , we test the proposal in some specific cases where morphisms between branes ( i.e. spectra of topological open strings ) can be easily computed .
we focus on the massive case , where the proposal simplifies considerably .
namely , the category of @xmath14-graded twisted complexes can be related to the sum of several copies of the category of finite - dimensional @xmath14-graded modules over a clifford algebra .
we will denote the latter category @xmath15 in what follows .
this reformulation is helpful , because the functor from the category of b - branes in a massive lg model to @xmath15 is very simple to describe . in this paper
we perform some checks that this functor is an embedding of graded categories . in view of the above - mentioned `` duality '' between @xmath14-graded twisted complexes and @xmath15 , this provides a test of kontsevich s proposal .
assuming the validity of the proposal , we infer that the category of b - branes in an arbitrary massive lg model is a full sub - category of @xmath15 .
the latter has a very simple and explicit description .
since for many fano varieties the mirror lg model is known , our results allow one to effectively compute the category of a - branes for such varieties . from the mathematical viewpoint
, it is an interesting challenge to reproduce such results using methods of symplectic geometry .
axiomatic definitions of topological d - branes for 2d topological field theories ( 2d tfts ) have been recently proposed by g. moore and g. segal @xcite and c. i. lazaroiu @xcite .
one of the main unresolved problems in the axiomatic approach is whether these axioms determine unambiguously the category of topological d - branes associated to a given tft .
we show that the category of b - branes for a massive lg model with a quadratic superpotential provides a counter - example to uniqueness .
in fact , this example shows that uniqueness , if understood naively , fails also for ordinary ( i.e. non - topological ) d - branes in any closed superstring background .
however , this particular failure is rather mild , i.e. it does not seem to have serious physical consequences .
now let us describe the content of the paper in more detail . in section [ sec :
a ] we recall some basic facts about mirror symmetry between fano varieties and lg models . in section [ sec : b ] we review general properties of b - branes in lg models .
we argue that if the superpotential has only isolated critical points , then it is sufficient to study b - branes in the infinitesimal neighborhood of each critical point .
for example , if all critical points of @xmath11 are non - degenerate , one does not lose anything if one replaces the superpotential by its quadratic approximation near each critical point .
the material in this section is not new and has been previously discussed in refs .
@xcite . in sections
[ sec : c ] and [ sec : cp ] we study b - branes in the lg model with the superpotential @xmath16 this is the simplest lg model where b - branes of dimension larger than @xmath17 are present . in section
[ sec : d ] we discuss b - branes in more general lg models with the superpotential @xmath18 in section [ sec : e ] we explain kontsevich s proposal and show that our results are consistent with it .
we also explain why brst invariance of boundary conditions requires twisted complexes of vector bundles instead of ordinary complexes .
this provides a physical explanation of kontsevich s proposal .
we also relate b - branes in massive lg models to @xmath14-graded clifford modules . in section [ sec : f ] we use homological mirror symmetry to compute the category of a - branes for @xmath10 and @xmath19 .
section [ sec : g ] contains concluding remarks .
a fano variety is a compact complex manifold whose anti - canonical line bundle is ample .
this is equivalent to saying that the first chern class of the canonical line bundle is negative - definite .
an @xmath0 sigma - model whose target space is a fano variety describes an @xmath0 @xmath6 field theory which is free in the ultraviolet . in the infrared
, it can either flow to a massive vacuum , or to a non - trivial @xmath0 scft .
note that classically @xmath0 sigma - models have both vector and axial @xmath8 r - symmetries , but for fano varieties quantum anomalies break the axial r - symmetry down to a discrete subgroup .
generically , this subgroup is @xmath14 , but in special cases it can be larger . in the calabi - yau case
the full axial r - symmetry is non - anomalous , and it is this fact that makes calabi - yau target spaces so special .
the simplest examples of fano varieties are complex projective spaces @xmath1 .
the corresponding @xmath0 field theories are well studied ; in fact , these models are integrable , in the sense that the exact s - matrix is known @xcite .
these theories have only massive vacua . a more general set of examples is given by grassmann varieties @xmath20 , which are defined as spaces of complex @xmath21-planes in an @xmath2-dimensional complex vector space .
the corresponding @xmath0 field theories are also integrable @xcite .
an @xmath0 field theory which has a conserved vector ( resp .
axial ) r - current admits a topological a - twist ( resp .
b - twist ) , which yields a 2d topological field theory called the a - model ( resp .
b - model ) .
@xmath0 superconformal field theories have both axial and vector r - symmetries , and therefore admit both kinds of twisting .
a - branes and b - branes are `` defined '' as boundary conditions which are consistent with a - twist and b - twist , respectively .
these two sets of branes have the structure of a category .
the space of morphisms is defined as the state space of topological open strings stretched between pairs of branes .
composition of morphisms is defined by means of 3-point correlators in topological open string theory .
since state spaces of open strings are graded vector spaces , brane categories are graded categories . in the case of a - branes
, spaces of morphisms are graded by the axial r - charge ; in the case of b - branes , by the vector r - charge . for fano varieties , the axial r - symmetry is generically @xmath14 , so spaces of morphisms in the category of a - branes are @xmath14-graded vector spaces . in the calabi - yau case ,
the full axial @xmath8 r - symmetry is non - anomalous , and therefore the category of a - branes is @xmath13-graded .
this is the reason one has to work with @xmath13-graded lagrangian submanifolds in the calabi - yau case @xcite . in the fano case ,
one only needs to require that lagrangian submanifolds be oriented . in the calabi - yau case
, we can also consider the category of b - branes . since the vector r - current is non - anomalous , this category is @xmath13-graded . in the fano case ,
the b - twist is not defined , and there is no obvious way to define the category of b - branes .
given a graded category , one can enlarge it by adding for any object @xmath22 its shifts @xmath23 $ ] , where @xmath24 or @xmath25 , and defining morphisms as follows : @xmath26,y_2[j]\right)=mor^{k+j - i}(y_1,y_2).\ ] ] in string theory , the r - charge of strings connecting two different branes is defined only up to an integer constant ; changing this constant by @xmath21 shifts the degree of all morphisms by @xmath21 .
the effect of this arbitrariness is that for any brane @xmath22 its shifts @xmath23 $ ] are automatically included .
this implies that no information is lost if we replace groups of morphisms with their degree-0 components .
this is what one usually does when working with categories of complexes , such as the derived category .
nevertheless , in this paper we will keep morphisms of all degrees , since this conforms better to physical conventions . from this viewpoint ,
the mathematical counterpart of the category of b - branes on a calabi - yau @xmath27 is not @xmath28 , but a @xmath13-graded category which is called the completion of @xmath28 with respect to the shift functor . for a sigma - model on a calabi - yau manifold which is a complete intersection in a toric variety ,
the mirror theory is again a sigma - model of the same kind . for fano varieties which are complete intersections in a toric variety ,
the mirror theory is a lg model whose target is a non - compact calabi - yau @xcite .
a general definition of a lg model involves , besides a choice of a target manifold , a choice of a holomorphic function @xmath11 on this manifold ( the superpotential ) .
thus non - trivial lg models require non - compact target spaces .
this non - compactness usually does not cause trouble : the important thing is for the critical set of @xmath11 to be compact . in general , superpotential breaks vector r - symmetry down to @xmath14 .
thus the a - twist is not defined , in general . on the other hand , since the canonical bundle of the target manifold is trivial , the axial r - symmetry is not anomalous , and the b - twist is well - defined .
we expect that the category of a - branes on a fano variety is equivalent to the category of b - branes on the mirror lg model . for example
, the mirror of @xmath1 is a lg model whose target is @xmath29 with the superpotential eq .
( [ wcpn ] ) .
this superpotential has @xmath30 non - degenerate critical points given by @xmath31 the physical interpretation is that the theory has @xmath30 massive vacua .
this agrees with the count of vacua in the @xmath1 model .
furthermore , the superpotential breaks vector @xmath32 symmetry down to @xmath14 given by @xmath33 where @xmath34 are the usual odd coordinates on the chiral @xmath35 superspace . as a consequence
, spaces of morphisms in the category of b - branes in this lg model are @xmath14-graded .
this is mirror to the fact that morphisms in the category of a - branes on @xmath1 are @xmath14-graded .
r - symmetry ; the @xmath14 symmetry discussed in the text is its subgroup .
this is mirror to the fact that the @xmath1 sigma - model has non - anomalous axial @xmath36 r - symmetry . in this paper
we will only keep track of @xmath14-gradings . ] as a rule , it is easier to understand b - branes , rather than a - branes .
therefore , we now turn to the study of b - branes in lg models , in the hope that it will illuminate the properties of a - branes on fano varieties .
the classical geometry of b - branes was described in refs .
@xcite ( see also ref . @xcite ) . in this section
we summarize the results of refs .
@xcite which are relevant for us and discuss some simple consequences .
let @xmath27 be the target space of a lg model . on general grounds ,
it must be khler manifold ( possibly non - compact ) .
let the @xmath11 be a fixed holomorphic function on @xmath27 ( the superpotential ) .
let @xmath22 be a submanifold of @xmath27 , and let @xmath37 be a hermitian vector bundle over @xmath22 with a unitary connection @xmath38 .
the rank of @xmath37 will be called the multiplicity of the corresponding d - brane .
it is shown in ref .
@xcite that the triple @xmath39 defines a classical b - type boundary condition if and only if @xmath22 is a complex submanifold of @xmath27 , @xmath11 is constant on @xmath22 , and the pair @xmath40 , where @xmath41 is the anti - holomorphic part of @xmath38 , is a holomorphic vector bundle .
for example , a point on @xmath27 together with a choice of multiplicity @xmath42 defines a b - type boundary condition .
the class of b - branes described in the previous paragraph does not exhaust all possible b - branes .
but it appears plausible that all b - branes can be obtained as bound states of the branes described above .
it was noticed in ref .
@xcite ( see also ref .
@xcite ) that most of the `` classical '' b - branes should be regarded as zero objects in the category of b - branes .
a classical b - brane is isomorphic to the zero object if and only if the space of its endomorphisms is zero - dimensional , i.e. when there are no supersymmetric open string states connecting the brane with itself . in this case
one says that world - sheet supersymmetry is spontaneously broken .
for example , it is explained in refs .
@xcite that if @xmath22 is a point on @xmath27 which is not a critical point of @xmath11 , then b - type supersymmetry is spontaneously broken , and therefore @xmath22 is isomorphic to the zero object in the category of b - branes .
this phenomenon reduces enormously the number of b - branes that one needs to consider , and makes it plausible that the whole category can be described combinatorially , using only the number and type of critical points of @xmath11 . to substantiate this claim
, we first notice that to any b - brane in the class described above one can assign a complex number , the value of @xmath11 on this brane .
further , there are no non - zero morphisms between branes with different values of @xmath11 , because any string connecting such branes will have non - zero energy and will not be supersymmetric @xcite .
( unlike in the case of a - branes , there is no central charge in the supersymmetry algebra , and supersymmetric states must have zero energy . )
thus the category of b - branes can be regarded as a family of categories parametrized by @xmath43 , and the categories at different points in @xmath43 do not `` talk '' to each other .
second , zero - energy classical configurations of an open string must be constant maps from the interval to @xmath22 , such that the potential energy @xmath44 vanishes .
this implies that unless a b - brane passes through a critical point of @xmath11 , there are no supersymmetric states for strings connecting this b - brane to any other b - brane ( including itself ) .
it follows that categories corresponding to non - critical values of @xmath11 are trivial ( contain only the zero object ) .
now let us assume that all critical points of @xmath11 are isolated . by scaling up the khler form
, we can make the semi - classical approximation arbitrarily good .
this means that wave - functions of all string states will be arbitrarily well localized near a particular critical point of @xmath11 , and the overlap between wave - functions associated to different critical points will be arbitrarily small . since topological correlators do not depend on the khler form , it is clear that morphisms between b - branes can be computed using only the leading terms in the taylor expansion of @xmath11 around the critical points .. ] to be more precise , we can attach a category to each isolated critical point of @xmath11 as follows : we replace @xmath11 by a polynomial which has the same singularity , and consider the category of b - branes on an affine space with this polynomial superpotential .
now let us form the direct sum of such categories over all critical points of @xmath11 and call it @xmath45 .
there is an obvious map which associates to any b - brane an object of @xmath15 .
invariance of topological correlators under variations of the khler form means that this map extends to a functor , and this functor is full and faithful . in other words ,
the category of b - branes is a full sub - category of @xmath15 . in particular , when all critical points of @xmath11 are non - degenerate ( i.e. when all vacua are massive ) , the problem reduces to understanding b - branes in the lg model with target @xmath12 and superpotential @xmath46 the corresponding bulk theory is free , but since the boundary conditions need not be linear , the problem of determining all b - branes is far from trivial . in this paper
we will study b - branes which correspond to linear boundary conditions .
some such branes have been considered in refs .
we will see below that if kontsevich s conjecture is true , then these branes generate the whole category of b - branes . note that the lg superpotential eq .
( [ wcpn ] ) satisfies the conditions stated above .
thus , assuming mirror symmetry , we can gain information about a - branes on @xmath1 by studying b - branes in the free lg model with the superpotential eq .
( [ wfreen ] ) . in the case @xmath47
this has been done in ref .
@xcite ; in that case the category of a - branes is independently known , and one can see that the mirror conjecture holds true . in section [ sec :
f ] we discuss the less trivial case @xmath48 . before continuing ,
let us make some further comments on the relation between @xmath15 and the category of b - branes .
we do not claim that the two categories are equivalent , only that the latter is a full sub - category of the former .
this means that each b - brane can be regarded as a direct sum of `` local '' b - branes attached to critical points , but not every direct sum is a valid b - brane .
we will see in section [ sec : f ] some examples where the category of b - branes is strictly smaller than @xmath15 .
nevertheless , since each b - brane behaves as a composite of `` local '' b - branes , it is reasonable to enlarge the category of b - branes by allowing arbitrary sums of `` local '' b - branes .
then the category of b - branes becomes equivalent to @xmath15 . from a purely algebraic standpoint ,
this is a very natural procedure ( c.f . a discussion in ref .
@xcite concerning reducible and irreducible branes ) , but the drawback is that the new branes lack a clear geometric interpretation .
section [ sec : f ] contains a further discussion of this issue .
we begin by recalling the results of ref .
@xcite concerning b - branes in the simplest lg model with the superpotential @xmath50 . in this case , the only allowed b - branes are d0-branes located at @xmath51 .
it has been shown in ref .
@xcite that the space of endomorphisms of a single d0-brane is two - dimensional , with one - dimensional even subspace and one - dimensional odd subspace . as a graded algebra , it is generated over @xmath43 by the identity and an odd element @xmath52 with the relation @xmath53 this is a clifford algebra @xmath54 .
if we take @xmath55 d0-branes , then the algebra of endomorphisms becomes @xmath56 where @xmath57 is the algebra of @xmath58 complex matrices .
we are interested in the next simplest lg model with the superpotential @xmath49 .
again , d0-branes must be localized at @xmath59 , and it has been shown in ref .
@xcite that the algebra of endomorphisms of a single d0-brane is generated by two odd elements @xmath60 with the relations @xmath61 this is a clifford algebra @xmath62 .
more invariantly , if we denote by @xmath63 the complex vector space which is the target space of our lg model , we can say that fermion zero modes for the open string take values in @xmath63 .
the hessian of the superpotential @xmath64 defines a non - degenerate symmetric bilinear form on @xmath63 , and the endomorphism algebra of the d0-brane is the clifford algebra associated to the pair @xmath65 .
( some standard facts about clifford algebras and their modules are described in the appendix .
we will freely use these facts in what follows . )
but in this case there can also be b - branes of higher dimension , namely d2-branes .
( this is briefly discussed in ref .
irreducible d2-branes are irreducible components of the critical level set @xmath66 , which is a singular quadric @xmath67 thus there are two candidate irreducible d2-branes , given by @xmath68 and @xmath69 , respectively .
our immediate goal is to compute their endomorphisms , as well as morphisms between d2-branes and d0-branes . in physical terms
, we will compute the spectrum and disk correlators of the topological open string with appropriate boundary conditions .
we consider the landau - ginzburg model on @xmath70 $ ] with two chiral superfields @xmath71 and @xmath72 . the superpotential assumes the following form @xmath73 we include a positive factor @xmath74 in the superpotential in order to keep track of dimensions of topological correlators later . in physical terms
, @xmath75 is a measure of the mass gap in the landau - ginzburg model . assuming the standard khler potential @xmath76 the world - sheet action reads @xmath77}\!\!{{\rm d}}^2x { \left}\{\sum_{{{\alpha}}=1}^{2}{\left}(|\partial_t\phi^{{{\alpha}}}|^2- & & { \left}.-\,|m\phi^1|^2 -|m\phi^2|^2- m{\left}(\psi_+^1\psi_-^2+\psi_+^2\psi_-^1{\right } ) - \bar{m}{\left}(\psi_-^{\bar1}\psi_+^{\bar2}+\psi_-^{\bar2}\psi_+^{\bar1}{\right}){\right}\},{\nonumber}\end{aligned}\ ] ] where @xmath78 are the bosonic components of @xmath79 , and @xmath80 are their fermionic partners . the world - sheet parametrization @xmath81 is such that @xmath82 is the world - sheet time
. from the bosonic lagrangian density @xmath83 one readily obtains the eom s for @xmath84 @xmath85 similarly , in terms of new variables @xmath86 the fermionic lagrangian density can be written as @xmath87 the fermionic eom s are given by @xmath88 b - type supersymmetry transformations are well known ( see , for example , ref .
@xcite ) and look as follows : @xmath89 now we will perform canonical quantization of this system with various boundary conditions which correspond to d0-d0 strings , d2-d2 strings , and d0-d2 strings .
we would like to find supersymmetric states of d0-d0 strings , since these correspond to endomorphisms of the d0-brane .
the relevant boundary conditions are @xmath90 first consider the bosonic degrees of freedom .
the boundary conditions give the following mode expansions for @xmath84 and its conjugate momentum @xmath91 : @xmath92 where @xmath93 to quantize , we impose the following commutation relations : @xmath94=\big[{{\tilde{a}}}_{{{\alpha}},n } , { { \tilde{a}}}_{{{\beta}},m}^\dagger\big]=\delta_{{{\alpha}}{{\beta}}}\delta_{mn},\ ] ] with all other commutators vanishing .
it is easy to check that ( [ eq : comm_a ] ) is compatible with the following canonical commutation relations @xmath95 = i\delta_{{{\alpha}}{{\beta}}}\cdot \frac2{\pi}\sum_{n=1}^{\infty } \sin n{\sigma}\sin
n{\sigma } ' = i\,\delta_{{{\alpha}}{{\beta}}}\,\delta({\sigma}-{\sigma } ' ) \label{eq : can_phi}\ ] ] the bosonic hamiltonian is given by @xmath96 where the additive constant `` @xmath97 '' in the sum is the bosonic zero point energy .
we shall see later that it is exactly canceled by the fermionic zero point energy .
next we consider the fermionic degrees of freedom .
it will turn out convenient to use the following combinations as new dynamical variables : @xmath98 the main advantage of using @xmath99 and @xmath100 is that the eom s for the unbarred quantities are decoupled from those for the barred quantities .
the mode expansions for these fields have the following form : @xmath101 to fix the commutation rules for the oscillators we impose the canonical commutation relations for @xmath102 : @xmath103 a convenient choice of compatible commutation rules for oscillators is @xmath104 with all others vanishing .
one can easily check that the canonical commutation relations for @xmath105 fields are also respected , provided that the following relations are imposed : @xmath106 where @xmath107 and @xmath108 are defined by @xmath109 the fermionic hamiltonian is @xmath110{\nonumber}\\ & = \int_0^\pi\!{{\rm d}}{\sigma}\ , { \left}[\ , i(c_+'-mc_+)\bar{b}_+ + i(c_-'+mc_-)\bar{b}_- + { \rm h.c . } { \right}]{\nonumber}\\ & = \sum_{n=1}^\infty \ ; { \omega}_n { \left}({{\alpha}}^\dagger_{1,n}{{\alpha}}_{1,n}+ { { \alpha}}^\dagger_{2,n}{{\alpha}}_{2,n}+{{\tilde{{{\alpha}}}}}^\dagger_{1,n}{{\tilde{{{\alpha}}}}}_{1,n}+{{\tilde{{{\alpha}}}}}^\dagger_{2,n}{{\tilde{{{\alpha}}}}}_{2,n}-2{\right}).{\nonumber}\end{aligned}\ ] ] note that the additive constant @xmath111 cancels the bosonic zero point energy .
the hamiltonian is diagonalized in the fock basis , and the zero - energy states are @xmath112 the supercharge @xmath113 can also be expanded in terms of oscillators .
it can be shown that each term in the expansion contains an annihilation operator for non - zero modes , and that @xmath113 does not depend on zero - mode oscillators @xmath108 and @xmath107 .
therefore @xmath113 annihilates all four ground states .
since we have two different d2-branes related by a symmetry , there are two inequivalent possibilities : either our string begins and ends on the same d2-brane , or it begins on one d2-brane , and ends on the other d2-brane .
the first situation corresponds to endomorphisms of a d2-brane , while the second one corresponds to morphisms from one d2-brane to the other one .
first we consider the case when both boundaries end on the same brane , say , the one given by the equation @xmath114 . the relevant boundary conditions are @xmath115 first let us look at the bosons .
the mode expansions for @xmath116 and its conjugate momentum are the same as before , while for @xmath117 and its conjugate momentum they are given by @xmath118 canonical commutation relations for bosonic fields imply the commutation relations eq .
( [ eq : comm_a ] ) for the oscillators . in terms of oscillators
the bosonic hamiltonian is @xmath119 for the fermions , the mode expansions are given by @xmath120\sin n{\sigma},{\nonumber}\\ b_2 & = & \frac{-1}{\sqrt{2\pi } } { \left } ( { { \tilde{{{\alpha}}}}}_{0 } - { { \alpha}}_0^\dagger{\right}){\nonumber}\\ & + & \frac{-i}{\sqrt{2\pi}}\sum_{n=1}^\infty { \left}[-\frac{n+im}{{\omega}_n}{\left}({{\tilde{{{\alpha}}}}}_{1,n}-{{\alpha}}_{1,n}^\dagger{\right } ) + \frac{n - im}{{\omega}_n}{\left}({{\tilde{{{\alpha}}}}}_{2,n}-{{\alpha}}_{2,n}^\dagger{\right}){\right}]\cos n{\sigma},{\nonumber}\\ c_1 & = & \frac{1}{\sqrt{2\pi}}{\left}({{\alpha}}_{0}+{{\tilde{{{\alpha}}}}}_{0}^\dagger{\right } ) + \frac{1}{\sqrt{2\pi}}\sum_{n=1}^\infty { \left}({{\alpha}}_{1,n}+ { { \alpha}}_{2,n}+{{\tilde{{{\alpha}}}}}_{1,n}^\dagger+{{\tilde{{{\alpha}}}}}_{2,n}^\dagger{\right})\cos n{\sigma},{\nonumber}\\ c_2 & = & \frac{1}{\sqrt{2\pi}}\sum_{n=1}^\infty { \left}({{\tilde{{{\alpha}}}}}_{2,n}-{{\tilde{{{\alpha}}}}}_{1,n}+{{\alpha}}_{2,n}^\dagger-{{\alpha}}_{1,n}^\dagger{\right})\sin n{\sigma}.{\nonumber}\end{aligned}\ ] ] the canonical commutation relations for the fields @xmath121 and @xmath122 are equivalent to the following commutation relations for the oscillators : @xmath123 with all others vanishing .
the fermionic hamiltonian can be shown to be @xmath124{\nonumber}\\ & = & \sum_{n=1}^\infty \ ; { \omega}_n { \left}({{\alpha}}^\dagger_{1,n}{{\alpha}}_{1,n}+ { { \alpha}}^\dagger_{2,n}{{\alpha}}_{2,n}+{{\tilde{{{\alpha}}}}}^\dagger_{1,n}{{\tilde{{{\alpha}}}}}_{1,n}+ { { \tilde{{{\alpha}}}}}^\dagger_{2,n}{{\tilde{{{\alpha}}}}}_{2,n}-2{\right})\nonumber\\ & & + \ ; m{\left}({{\alpha}}_0^\dagger{{\alpha}}_0 + { { \tilde{{{\alpha}}}}}_0^\dagger{{\tilde{{{\alpha}}}}}_0 - 1{\right}).{\nonumber}\end{aligned}\ ] ] the fermionic zero point energy cancels the bosonic zero point energy , and we see that there is a unique state with zero energy : the fock vacuum . for the same reason as in the d0-d0 case , this state is supersymmetric ( is annihilated by the supercharge ) .
now consider the case when one end of the string ( @xmath125 ) is attached to @xmath114 , and the other one ( @xmath126 ) is attached to @xmath127 .
the boundary conditions are @xmath128 the mode expansions for the bosons are @xmath129 while for the fermions they are @xmath130\sin k_n{\sigma},{\nonumber}\\ b_2 & = & \frac{1}{\sqrt{2\pi}}\sum_{n=1}^\infty { \left}[\frac{k_n+im}{{\omega}_n}{\left}({{\tilde{{{\alpha}}}}}_{1,n}-{{\alpha}}_{1,n}^\dagger{\right } ) - \frac{k_n - im}{{\omega}_n}{\left}({{\tilde{{{\alpha}}}}}_{2,n}-{{\alpha}}_{2,n}^\dagger{\right}){\right}]\cos k_n{\sigma},{\nonumber}\\ c_1 & = & \frac{1}{\sqrt{2\pi}}\sum_{n=1}^\infty { \left}({{\alpha}}_{1,n}+{{\alpha}}_{2,n}+{{\tilde{{{\alpha}}}}}_{1,n}^\dagger+{{\tilde{{{\alpha}}}}}_{2,n}^\dagger{\right})\cos k_n{\sigma},{\nonumber}\\ c_2 & = & \frac{i}{\sqrt{2\pi}}\sum_{n=1}^\infty { \left}({{\tilde{{{\alpha}}}}}_{1,n}-{{\tilde{{{\alpha}}}}}_{2,n}+{{\alpha}}_{1,n}^\dagger-{{\alpha}}_{2,n}^\dagger{\right})\sin k_n{\sigma},{\nonumber}\end{aligned}\ ] ] where @xmath131 one can show as before that commutation relations ( [ eq : comm_a ] ) and ( [ eq : comm_alpha ] ) yield all the canonical commutation relations , and the total hamiltonian is diagonalized in the fock basis as follows : @xmath132 again there is a single ground state which is annihilated by the supercharge .
the boundary conditions for the bosons are @xmath133 the corresponding mode expansions are @xmath134 where @xmath135 imposing ( [ eq : can_phi ] ) , we infer that the bosonic oscillators obey ( [ eq : comm_a ] ) .
the bosonic hamiltonian is given by @xmath136 now let us consider fermions .
the boundary conditions imposed by supersymmetry are @xmath137 which , when combined with the eom s , give the following boundary conditions for @xmath102 and @xmath105 fields separately : @xmath138 @xmath139 the mode expansions are @xmath140 where @xmath141 imposing canonical commutation relations on the fields implies the following commutation relations for the oscillators : @xmath142 @xmath143 all other anti - commutators vanish . in view of the commutation relations for @xmath144 and @xmath145 , we set @xmath146 so that @xmath147 the fermionic hamiltonian has the form @xmath148{\nonumber}\\ & = & \sum_{n=1}^\infty { \omega}_{1,n } { \left}({{\alpha}}^\dagger_{n}{{\alpha}}_{n}+{{\tilde{{{\alpha}}}}}^\dagger_{n}{{\tilde{{{\alpha}}}}}_{n}-1{\right } ) + \sum_{n=1}^\infty { \omega}_{2,n } { \left } ( { { \beta}}^\dagger_{n}{{\beta}}_{n } + \tbb^\dagger_{n}\tbb_{n}-1{\right}).{\nonumber}\end{aligned}\ ] ] as before the bosonic zero - point energy is canceled by the fermionic zero - point energy .
there are two zero - energy states : @xmath149 again it can be shown that they are annihilated by the supercharge
. therefore both ground states are supersymmetric .
the landau - ginzburg model admits a topological twist to yield the so - called b - model . this topological twist turns the world - sheet spinor fields @xmath150 and @xmath151 of the original lg model into a pair of sections of the pullback bundle @xmath152 , which we denote by @xmath153 , and a world - sheet one - form @xmath154 with values in @xmath155 .
the brst transformations of the twisted fields are @xmath156 to make connection with the fields in the original landau - ginzburg theory , we note that @xmath107 and @xmath52 are the twisted versions of @xmath157 and @xmath158 respectively , while @xmath154 comes from @xmath102 and @xmath105 .
we also adopt the common notation @xmath159 .
the local physical observables are in one - to - one correspondence with the brst - cohomology , i.e. local quantities which are brst invariant but not brst exact .
it is easy to see from the above brst transformations that in the bulk the physical observables correspond to holomorphic functions of @xmath84 modulo @xmath160 , where @xmath161 is an arbitrary holomorphic vector field .
there are no additional local observables from the fermionic fields , as long as @xmath11 is nontrivial . in particular , when @xmath162 , the space of bulk observables is @xmath163/i$ ] , where @xmath164 is the ideal generated by the first partial derivatives of @xmath11 . in the boundary sector , where @xmath11 is constrained to be constant , additional observables will arise from the @xmath52 fields .
we now specialize to the d0 and d2 branes studied above .
we consider the boundary component which is mapped to the d0 brane located at @xmath165 .
the boundary conditions require , among other things , that @xmath166 therefore boundary observables can only come from the @xmath52 fields . from the brst transformation @xmath167 one sees immediately that both @xmath168 and @xmath169 are brst invariant on the boundary .
let us denote the restriction of @xmath170 to the boundary by the same letter @xmath170 .
thus the ring of boundary observables is generated by @xmath168 and @xmath169 . without loss of generality
, we may assume that the d2 brane sits at the locus @xmath171 .
the relevant boundary conditions read @xmath172 from this one sees that @xmath169 no longer gives rise to a boundary degree of freedom .
also , since @xmath117 is not constrained to vanish on the boundary , @xmath168 is no longer brst invariant .
thus there are no boundary observables ( except for the identity operator ) associated with the d2 brane .
first let us compute topological correlators for strings connecting a brane with itself . since in the d2-d2 case there is only the vacuum state , the problem is non - trivial only in the d0-d0 case .
in order to compute disk correlators with products of @xmath168 and @xmath169 inserted on the circumference , we proceed as in ref .
we start with a world - sheet diagram which has the topology of a cylinder , with the d0 and d2 boundary conditions imposed on the two boundary circles ( c.f .
[ fig : cyl ] ) . on the d0 boundary
there can be operator insertions .
viewed in the open - string channel , this world - sheet diagram computes the one - loop amplitude @xmath173\ ] ] where the @xmath174 s are operators inserted at the d0 boundary .
the trace on the rhs can be reduced to the hilbert space of open - string zero modes by the standard argument . in the specific case at hand ,
the zero mode space is spanned by @xmath175 and @xmath176 as described in sec .
[ sec : d0d2 ] , and one easily obtains @xmath177 more generally , one can compute @xmath178 from ( [ eq : cyl_1 ] ) and ( [ eq : cyl_2 ] ) one deduces the relations in the boundary operator product algebra for the d0-brane : @xmath179 this is the clifford algebra with two generators corresponding to the quadratic form @xmath180 up to a numerical factor , this matrix is the hessian of @xmath11 in the basis @xmath60 .
thus one can state the result more invariantly by saying that the boundary operator product algebra for the d0-brane is the clifford algebra @xmath181 , where @xmath63 is target space of our lg model , and @xmath113 is the quadratic form given by the hessian of @xmath11 .
topological correlators on the disk can be inferred from the computation on the cylinder using factorization in the closed string channel @xcite .
namely , we insert a complete set of states in the closed string channel ( cf . fig .
[ fig : cyl]b ) and rewrite the cylinder amplitude as @xmath182 where @xmath183 s form a complete set of bulk operators , and @xmath107 is the ( inverse ) metric on the space of bulk operators defined via topological correlators on the sphere .
the relative normalization is fixed by demanding the following relation for the d2-d2 cylinder amplitude @xmath184 in our case , the only bulk operator is the identity , therefore all disk correlators for the d0-brane simply coincide with the cylinder correlators .
besides the algebra structure , another important datum is a non - degenerate inner product on the space of endomorphisms .
this inner product is determined by the two - point disk correlator and makes the endomorphism algebra into a ( non - commutative ) graded frobenius algebra . in our case
the only non - vanishing inner products are @xmath185 note that the bilinear form corresponding to this product is even .
in contrast , in the model @xmath50 the bilinear form is odd and given by @xmath186 so far we have determined the endomorphism algebra of the d0-brane ( it is isomorphic to @xmath62 ) and the d2-brane ( it is isomorphic to @xmath43 ) .
now we turn to the computation of compositions of morphisms between different branes .
we begin with the case when both branes are d2-branes .
let us denote the d2-brane given by the equation @xmath114 ( resp .
@xmath127 ) by @xmath187 ( resp .
@xmath188 ) .
it was shown in the previous section that the vector space @xmath189 is one - dimensional for all @xmath190 and @xmath191 .
when @xmath192 , this space is even , but for @xmath193 there is no canonical choice for the r - charge . in other words , for @xmath193 the `` vacuum '' vector spanning @xmath189 can equally well be regarded as even or odd . for reasons which will become clear later
, we define @xmath194 to be purely odd ; since @xmath195 is dual to @xmath194 , it is also purely odd , while @xmath196)$ ] is purely even . here
@xmath197 $ ] denotes the shift of @xmath187 .
let @xmath198 and @xmath199 be generators of @xmath194 and @xmath200 , respectively . since the endomorphism algebra of a d2-brane
is spanned by the identity morphism , we only need to determine if @xmath201 is zero or not .
this product is evaluated by the disk amplitude with two insertions of boundary - changing operators . by conformal
invariance , this is the same as the vacuum - vacuum transition amplitude for open strings stretched between @xmath188 and @xmath187 . since there are no fermionic zero modes in this case , this amplitude is non - zero .
this means that @xmath202 with @xmath203 .
this trivial computation implies that the even generator of @xmath196)$ ] is an isomorphism . in physical language
, @xmath188 is isomorphic to the anti - brane of @xmath187 .
there remain compositions of morphisms involving both d2-branes and the d0-brane . in view of the previous paragraph
, it is sufficient to consider morphisms between @xmath188 and the d0-brane .
no new computations are actually required , the result being fixed by general properties of topological string theory @xcite .
first of all , we note that by cyclic symmetry of topological correlators computing compositions of morphisms from d2 to d0 and back ( or the other way around ) is equivalent to computing how the endomorphism algebra of the d0-brane acts on the space of morphisms from d2 to d0 . in more detail
, we have a non - degenerate pairing @xmath204 given by the path - integral on an infinite strip .
( this paring is odd in our case , because there is a single fermionic zero mode . )
similarly , we have an even non - degenerate pairing @xmath205 thus computing the product map @xmath206 is the same as computing the map @xmath207 furthermore , in our case @xmath208 is isomorphic to @xmath62 , and we know from the previous section that @xmath209 is two - dimensional .
the @xmath14 graded algebra @xmath62 has a unique representation on @xmath210 , up to a flip of parity ( up to isomorphism , it is given by any two pauli matrices ) . since in string theory
the parity of morphisms is not canonically fixed anyway , we conclude that the module structure of @xmath209 is completely determined , up to the unavoidable ambiguity in the overall parity .
this in turn determines the composition of morphisms going from d0 to d2 and back .
we now turn to massive landau - ginzburg models which involve more than two fields . without loss of generality
, we may assume that the superpotential on @xmath12 is given by @xmath211 we can construct examples of b - branes in this lg model for any @xmath2 , using the results of the previous section . for @xmath212 , @xmath213
, we consider an equivalent superpotential @xmath214 since it is a sum of @xmath21 copies of the superpotential @xmath49 , we can construct a b - type boundary condition by picking @xmath21 arbitrary b - type boundary conditions for the latter model , and tensoring them .
for example , if we take all boundary conditions to be d0-branes , the tensor product state will also be a d0-brane , and its endomorphism algebra will be @xmath215 . if we take all boundary conditions to be d2-branes , the tensor product boundary state will be a d(@xmath216)-brane , and its endomorphism algebra will be @xmath43 .
similarly , for @xmath217 we consider an equivalent superpotential @xmath218 clearly , b - branes for this lg model can be constructed by taking tensor product of @xmath21 boundary states for the lg model with @xmath49 and a boundary state for the lg model with @xmath50 . in this way one obtains b - branes of dimension up to @xmath216 .
it is easy to see that the endomorphism algebra of the d0-brane will be isomorphic to @xmath219 , while the endomorphism algebra of the d(@xmath216)-brane will be isomorphic to @xmath54 .
more generally , one can explicitly construct all b - branes which correspond to linear subspaces of the critical level set @xmath66 .
since @xmath11 is quadratic , these are the same as linear subspaces isotropic with respect to the bilinear form @xmath113 .
classification of such isotropic subspaces is well known @xcite .
the maximal dimension of an isotropic subspace is @xmath220.$ ] for @xmath2 odd , there is a single irreducible family of isotropic subspaces of maximal dimension parametrized by @xmath221 parameters . for @xmath2 even , there are two irreducible families of isotropic subspaces of maximal dimension parametrized by @xmath222 parameters .
any isotropic subspace lies in one of the maximal isotropic subspaces .
it is straightforward to compute morphisms and their compositions ( i.e the spectrum and topological correlators ) between all linear b - branes . in the next subsection we discuss in some detail the results for the case @xmath223 , when the lg superpotential has the form @xmath224 .
then we will describe the general case . note that in principle there could also be b - branes corresponding to non - linear boundary conditions ( e.g. non - linear submanifolds of the quadric @xmath66 ) .
such b - branes are hard to study directly . in what follows
we shall focus on linear boundary conditions .
maximal isotropic linear subspaces on the quadric surface @xmath66 are complex lines , and there is a single irreducible family of them .
this family is parametrized by @xmath9 as follows : @xmath225 where @xmath226 $ ] are homogeneous coordinates on @xmath9 .
any two distinct lines in the family intersect at a single point ( @xmath227 ) . using a linear change of basis in the target space which preserves @xmath11 , one can always map any line in the above family to the line @xmath228 . for the brane @xmath228
we already know that the endomorphism algebra is isomorphic to @xmath54 , and since linear changes of variables preserving the superpotential are invariances of the topological lg model , we conclude that the same is true for any d2-brane in the above family .
next we consider morphisms between different lines in the family .
clearly , there are no bosonic zero modes , so the space of morphisms will be spanned by the `` vacuum '' state and its fermionic excitations with zero energy . as remarked in the previous section , only some components of @xmath52 have a chance to be brst - non - trivial boundary observables . thus all brst - invariant states can be obtained by acting by some components of @xmath52 on the vacuum state .
let @xmath229 be the target space of our lg model , and let @xmath230 and @xmath231 be two distinct lines in @xmath63 isotropic with respect to the quadratic form @xmath11 .
the corresponding b - branes will be denoted @xmath188 and @xmath187 .
let us look at the @xmath52-field restricted to the boundary of the world - sheet which is mapped to @xmath230 .
we can regard @xmath232 as basis elements of @xmath63 .
brst transformations are @xmath233 on the boundary the vector with components @xmath234 can be an arbitrary element of @xmath230 .
it follows that brst - invariant components of @xmath52 must be orthogonal to @xmath230 with respect to the form @xmath113 .
we denote the orthogonal subspace by @xmath235 . of course
, since @xmath230 is isotropic , we have an inclusion @xmath236 .
similarly , brst - invariant fermionic fields on the @xmath231-boundary are parametrized by elements of @xmath237 .
the total space of brst - invariant fermionic fields is @xmath238 however , not all of these are non - zero .
neumann boundary conditions plus supersymmetry imply that the components of @xmath52 along @xmath239 vanish .
thus non - trivial brst invariant fermionic zero modes are parametrized by elements of the quotient space @xmath240 this space is one - dimensional .
thus there is a single fermionic zero mode , and the space of morphisms between two different lines is isomorphic to its exterior algebra ( as a @xmath14-graded vector space ) .
that is , @xmath194 has one - dimensional even subspace , and one - dimensional odd subspace .
composition of morphisms between two distinct lines is fixed by consistency considerations .
if @xmath188 and @xmath187 are any two lines , then @xmath194 must be a left module over @xmath241 and right module over @xmath242 .
there is only one such module of dimension two : the clifford algebra itself , regarded as a bi - module over itself .
together with various parings given by the 2-point correlators , this fixes the structure of correlators involving any two d2-branes .
in particular , it is easy to see that the element in @xmath194 corresponding to the identity element in @xmath54 is invertible .
this means that any two lines give isomorphic objects in the category of b - branes .
similar arguments can be used to determine boundary correlators involving both d2 and d0 . as explained above
, the endomorphism algebra of the d0-brane is isomorphic to @xmath243 . as for the space of morphisms between a d2-brane and d0-brane , it is 4-dimensional , with two - dimensional even subspace and two - dimensional odd subspace . indeed , since all d2-branes are isomorphic , it is sufficient to consider morphisms from the d2-brane @xmath228 .
this d2-brane is the tensor product of the d2-brane in the lg model @xmath49 and the d0-brane in the lg model @xmath50 .
hence the computation of the space of morphisms and their compositions is reduced to the one we have performed in the previous section .
the above arguments can be easily generalized to arbitrary @xmath2 .
we shall consider only linear boundary conditions .
let the target space be @xmath245 , let @xmath11 be a non - degenerate quadratic function on @xmath63 , and let @xmath246 be its hessian .
a b - brane is a linear subspace @xmath247 which is isotropic with respect to @xmath113 . as mentioned above
, @xmath248 is less or equal to @xmath220 $ ] . using linear changes of variables
, we can bring @xmath11 to the standard form , and @xmath249 to the subspace given by @xmath250 .
such a d(2k)-brane is a tensor product of @xmath21 copies of d2-branes in the model @xmath49 , @xmath220-k$ ] copies of the d0-brane in the model @xmath49 , and , for @xmath2 odd , one copy of a d0-brane in the model @xmath50 .
it follows that the space of endomorphisms has dimension @xmath251 and is isomorphic as a @xmath14-graded algebra to the clifford algebra with @xmath252 generators .
in particular , the algebra of endomorphisms of a d - brane of maximal possible dimension is isomorphic to @xmath43 or @xmath54 depending on whether @xmath2 is even or odd , while the the endomorphism algebra of the d0-brane is isomorphic to @xmath253 .
next let us discuss morphisms between two different b - branes .
let @xmath230 and @xmath231 be isotropic linear subspaces corresponding to b - branes @xmath188 and @xmath187 .
the same arguments as in the previous subsection tell us that the space of fermionic zero modes can be identified with @xmath254 the space of morphisms is isomorphic as a graded vector space to the exterior algebra of this vector space ( up to an overall flip of parity ) .
it is easy to see that the dimension of the space of zero modes is given by @xmath255 , where @xmath256 .
therefore the dimension of the space of morphisms is given by @xmath257 in particular , in the case @xmath258 we recover the result eq .
( [ dimend ] ) obtained by other means .
let us give a few examples .
first , let @xmath2 be even , and @xmath230 and @xmath231 be distinct maximal isotropic subspaces .
then @xmath259 for @xmath260 , and there are no zero modes .
this means that the space of morphisms between any two maximal isotropic subspaces is one - dimensional . as usual , the r - charge assignment is ambiguous , but it is natural to require the r - charge to vary continuously as one varies @xmath261 .
since in the case @xmath258 the space of endomorphisms is even and isomorphic to @xmath43 , this implies that for any two maximal isotropic subspaces in the same irreducible family the space of morphisms is even and isomorphic to @xmath43 .
we will fix the remaining ambiguity by saying that the space of morphisms between two maximal isotropic subspaces in _ different _ irreducible families is odd .
the reason for such a convention will be explained in the next section .
the fact that for even @xmath2 there are no fermionic zero modes for open strings connecting two maximal isotropic subspaces implies that the vacuum - vacuum transition amplitude is non - zero in this sector .
this is equivalent to saying that the composition of non - zero morphisms between two maximal b - branes is a non - zero multiple of the identity endomorphism .
if these two b - branes are in the same irreducible family , this means that they represent isomorphic objects in the category ; if they are in different irreducible families , then the interpretation is that they are isomorphic up to a shift .
if @xmath2 is odd , and @xmath230 and @xmath231 are maximal linear subspaces , then there is a single fermionic zero mode .
thus @xmath194 is two - dimensional , with one - dimensional even and one - dimensional odd subspaces .
the composition of morphisms going between two maximal b - branes is fixed by consistency requirements .
namely , the space of morphisms must be a graded bi - module over @xmath54 ( the endomorphism algebra of a single b - brane ) , and there is only one such graded bi - module of dimension two : @xmath54 itself .
furthermore , there is an odd non - degenerate pairing @xmath262 which is invariant with respect to both actions of @xmath54 in an obvious sense . up to isomorphism
, there is only one such pairing , namely @xmath263 where @xmath264 is defined by @xmath265 together with the module structure of @xmath194 , this pairing determines the composition of morphisms going between any two lines . as in the case @xmath224 ,
it is easy to see that the morphism corresponding to the identity element of @xmath54 is invertible , and therefore any two maximal b - branes are isomorphic .
our third example is the case @xmath266 , that is , the case of the d0-brane .
the space of zero modes coincides with @xmath63 , and the space of endomorphism is isomorphic to @xmath267 as a @xmath14-graded vector space ( @xmath63 is regarded as odd ) .
this agrees with an independent argument of subsection [ secdgen ] .
there we also showed that the algebra of endomorphisms is isomorphic to @xmath181 .
our fourth and final example is the case when @xmath230 is an arbitrary isotropic subspace of dimension @xmath268 $ ] , and @xmath269 . in other words ,
the second brane is the d0-brane .
then the space of zero modes is @xmath270 .
its dimension is @xmath271 , and therefore the space of open strings stretched between a maximal linear subspace and the d0-brane has dimension @xmath272 the space of morphisms has the structure of a graded module over the endomorphism algebra of the d0-brane , which is isomorphic to @xmath181 .
if we neglect the grading , there is a unique such module , which is a sum of @xmath273-k}$ ] irreducible ( spinor ) modules .
let us recall how to define the derived category of coherent sheaves on a smooth affine variety @xmath27 following ref .
let @xmath274 be the category of coherent sheaves on @xmath27 , or equivalently the category of finite modules over the coordinate ring @xmath275 of @xmath27 .
we define @xmath276 to be a category whose objects are bounded @xmath13-graded complexes of projective objects of @xmath274 .
equivalently , we can think about the coordinate ring of @xmath27 as a differential graded algebra ( dg - algebra ) which is concentrated in degree zero and has a trivial differential ; then objects of @xmath276 are differential graded modules ( dg - modules ) over this dg - algebra such that all homogeneous components are projective @xmath275-modules , and all but a finite number of homogeneous components are trivial .
morphisms in @xmath276 are morphisms of these dg - modules regarded simply as @xmath275-modules ( i.e. morphisms do not necessarily preserve the grading or respect the differentials ) .
groups of morphisms in the category @xmath276 are naturally @xmath13-graded and have a natural differential of degree @xmath277 .
for example , closed morphisms of degree 0 in the category @xmath276 are ordinary morphisms of complexes ( the ones which preserve the grading and commute with the differentials ) , while exact morphisms of degree 0 are morphisms of complexes which are homotopic to zero .
thus @xmath276 is a dg - category .
there is a general way to make a triangulated category out of any dg - category @xcite .
one takes the category of `` twisted objects '' of the dg - category , which is again a dg - category , and then passes to degree-0 homology , i.e. replaces groups of morphisms with their degree-0 homology . in the present case ,
since we are working with complexes of projective modules , it is not necessary to consider twisted objects , and one can simply apply the functor @xmath278 to @xmath276 .
the resulting triangulated category is simply the homotopy category of bounded complexes of projective @xmath275 modules , and it is well known that it is equivalent to the bounded derived category of @xmath274 ( see e.g. ref .
alternatively , one can apply to @xmath276 the functor @xmath279 .
this gives a graded category which is the completion of @xmath28 with respect to the shift functor .
as discussed in section [ sec : a ] , the latter alternative conforms better to physical conventions .
now we can formulate kontsevich s proposal rather simply .
let @xmath27 be a smooth affine variety , and @xmath11 be a holomorphic function on @xmath27 ( the superpotential ) , whose critical set is compact .
let @xmath280 be a critical value of @xmath11 .
first , since in the presence of the superpotential morphisms between b - branes are @xmath14-graded , we will have to use @xmath14-graded complexes in order to construct the analogue of @xmath276 .
second , we deform our @xmath14-graded complexes of projective modules by asking that the composition of two successive morphisms be equal to @xmath281 , instead of zero .
thus objects of the deformed category @xmath282 are pairs of finitely generated projective @xmath275-modules @xmath283 and morphisms @xmath284 and @xmath285 such that @xmath286 we can regard the pair @xmath287 as a @xmath14-graded @xmath275-module , and @xmath288 as an odd endomorphism @xmath289 of this module whose square is @xmath281 ( `` twisted differential '' ) .
morphisms in this category are defined as ( ungraded ) morphisms of the corresponding @xmath275-modules .
they have a natural @xmath14-grading , and a natural differential .
the differential on @xmath290 is defined as @xmath291 here @xmath292 acts as @xmath277 on the even component and as @xmath293 on the odd component .
it is easy to see that @xmath294 is an odd operator whose square is zero .
thus @xmath282 is a differential @xmath14-graded category . in
what follows the term `` dg - category '' ( resp .
`` graded category '' ) will refer to a differential @xmath14-graded category ( resp .
@xmath14-graded category ) , unless specified otherwise .
applying to @xmath282 the functor @xmath279 , we obtain a graded category , which is proposed to be equivalent to the category of b - branes corresponding to the critical value @xmath295 .
one can show that all spaces of morphisms in this category are finite - dimensional , provided the critical set of @xmath11 is compact .
an unsatisfactory feature of this construction is that one needs to use complexes of projective modules , instead of general complexes .
this causes problems if one tries to extend the definition from affine varieties to algebraic ones .
there is a way to repair this defect @xcite , but we will not try to explain this more complicated definition in this paper . in this subsection
we give a physical argument supporting the identification of b - branes with objects of the category @xmath282 .
our argument is modelled on those in refs .
@xcite , where it was explained why complexes of locally free sheaves on a calabi - yau manifold can be thought of as b - branes . for our purposes , it is sufficient to consider @xmath14-graded complexes
. then the argument of refs .
@xcite can be summarized as follows .
consider a pair of locally free sheaves ( i.e. holomorphic vector bundles ) @xmath296 and @xmath297 .
we already know that @xmath296 and @xmath297 can be thought of as b - branes , i.e. as topological boundary conditions for a topological sigma - model .
the same goes for @xmath298 and @xmath299 $ ] .
( in the physical setting , one has additional data , such as hermitian metrics on @xmath296 and @xmath297 and compatible connections . )
now we can deform the boundary condition corresponding to @xmath299 $ ] by adding a boundary term to the action which depends on a pair of sections @xmath300 and @xmath301 ( the `` tachyons '' ) . in order to preserve brst invariance
, one has to require that @xmath302 and @xmath303 be holomorphic , and @xmath304 , @xmath305 .
this deformed boundary condition corresponds to a @xmath14-graded complex @xmath306 one expects ( although there is no iron - clad argument ) that any b - brane is isomorphic to a b - brane of this kind .
this provides a physical explanation for the relation between complexes of locally free sheaves on a calabi - yau and b - branes . in the case of a lg model ,
locally free coherent sheaves on @xmath27 are not valid b - branes , because their support is the whole @xmath27 , and @xmath11 is not constant on @xmath27 .
technically , the problem occurs because the brst variation of the bulk action contains a non - vanishing @xmath11-dependent term which is a total derivative on the world - sheet .
this is the so - called warner problem @xcite .
the sum @xmath299 $ ] is not a b - brane either .
however , we can try to add a boundary term to the action of the sigma - model so that brst invariance is restored .
we take the same term as in the case @xmath66 . as in refs .
@xcite , we have two holomorphic sections @xmath302 and @xmath303 to play with .
as we show below , the condition of brst - invariance is modified to @xmath307 , @xmath308 , where @xmath295 is a constant .
this shows that any object of @xmath282 corresponds to a b - brane .
now let us work out the brst - invariance conditions and show that @xmath302 and @xmath303 must satisfy the constraints stated above .
for simplicity , let us assume at first that both @xmath296 and @xmath297 are line bundles .
the boundary lagrangian is taken to be @xmath309 here @xmath310 is a complex fermion living on the boundary , @xmath311 is the restriction of a bulk fermionic field to the boundary , and @xmath312 are holomorphic sections of @xmath313 and @xmath314 , respectively .
they depend on the fields @xmath315 restricted to the boundary .
the fermion @xmath310 takes values in @xmath314 , and the covariant derivative @xmath316 along the boundary makes use of the unitary connections on @xmath296 and @xmath297 .
the boundary lagrangian is manifestly gauge - invariant .
if we set @xmath317 , we get the usual path - integral representation of the parallel transport operator in the bundle @xmath318 @xcite . for non - zero @xmath302 or @xmath303
we get a deformation of the usual boundary condition . in the special case
@xmath319 we get the boundary lagrangian used in ref .
@xcite .
we postulate the following supersymmetry transformations for @xmath310 : @xmath320 here @xmath321 and @xmath322 are regarded as independent complex grassmann variables .
we also note that @xmath302 and @xmath303 transform as follows : @xmath323 brst transformations are obtained by setting @xmath324 .
one can check that the brst variation of the boundary lagrangian is given , up to a total derivative , by @xmath325 on the other hand , the brst variation of the bulk action is a boundary term given by @xmath326.{\nonumber}\end{aligned}\ ] ] the first two terms in the bulk variation are standard and vanish when the standard neumann boundary conditions are imposed on @xmath315 and @xmath327 .
the last term is the warner term @xcite . obviously , in order for the variation of the boundary lagrangian to cancel the warner term , we need to require @xmath328 one can get rid of the factor @xmath190 by redefining @xmath329 .
this implies that instead of an ordinary complex of holomorphic vector bundles we are dealing with an object of @xmath330 it is straightforward to generalize the construction to the higher - rank case .
the fermion @xmath310 still takes values in @xmath314 , which means that it is a matrix of size @xmath331 . in order for the path - integral over @xmath332 to reproduce a path - ordered exponential in the representation of the gauge group of dimension @xmath333
, one needs to insert a projector onto the sector where the total fermion number ( including the boundary contribution from @xmath310 ) is equal to @xmath17 or @xmath277 @xcite .
the rest of the argument is unchanged .
the conditions of brst - invariance now read @xmath334 the two conditions arise by requiring that the brst variation of the bulk term be cancelled on both boundaries of the world - sheet . by taking the trace of these two equations and comparing them ,
one infers that the ranks of @xmath296 and @xmath297 are in fact the same , and the constant terms in the equations are also the same .
it follows that any object of @xmath282 is a b - brane .
one can also check that the total brst charge is nilpotent . indeed , it is easy to see that the square of the bulk contribution to the brst charge is equal to @xmath335 on the other hand , the boundary supercharge coming from one of the two boundaries is given by @xmath336 canonical quantization yields @xmath337 and therefore @xmath338 it is also easy to check that @xmath339 and @xmath340 anti - commute ( the holomorphicity of @xmath302 and @xmath303 is important here . )
hence the sum of @xmath340 and the two boundary supercharges is nilpotent .
we start with the case @xmath341 , @xmath50 , where there is only a d0-brane to worry about . in the absence of the superpotential , d0-brane on @xmath43
is associated with the structure sheaf of a point , which has a two - term projective resolution @xmath342 if we pass to @xmath14-graded complexes , we obtain the following dg - module : @xmath343 where @xmath344 and it is understood that the module @xmath345 has the obvious grading .
if we turn on the superpotential , we need to deform the differential so that its square be equal to @xmath50 .
it is clear how to do this : simply consider the object @xmath346 where @xmath347 this is our candidate object for the d0-brane at @xmath51 . as a check ,
let us compute its endomorphism algebra , following kontsevich s prescription .
in the category @xmath282 , the algebra of endomorphisms is @xmath348).\ ] ] the @xmath14-grading is the natural grading on @xmath349 matrices ( diagonal elements are even , off - diagonal elements are odd ) .
the differential acts on this graded vector space as follows : @xmath350 here @xmath351 are elements of @xmath352 i.e. simply polynomials in @xmath353 .
computing @xmath279 , we find that this abelian group is isomorphic to the group of complex matrices of the form @xmath354 multiplicative structure is given by matrix multiplication . clearly , this algebra is generated over @xmath43 by the identity and an odd matrix @xmath355 which squares to identity .
this agrees with the endomorphism algebra of the d0-brane in the lg model @xmath50 @xcite .
now let us discuss b - branes in the lg model @xmath49 . using the same reasoning as above ,
it is easy to guess that the d2-brane given by the equation @xmath68 should correspond to the object @xmath356 where the map is defined as @xmath357 similarly , the d2-brane given by the equation @xmath69 should correspond to @xmath358 with the twisted differential @xmath359 the group of endomorphisms of the former object in the category @xmath282 is @xmath360),\ ] ] with the differential which acts as follows : @xmath361 the homology is readily computed ; the result is that it is spanned by the identity matrix .
thus the algebra of endomorphisms in the derived category is isomorphic to @xmath43 .
this agrees with the computation in section [ sec : cp ] .
of course , for the other d2-brane we get the same result . finally , in order to compute morphisms between the two d2-branes , we note that one is a shift of the other .
is isomorphic to the anti - brane for the d2-brane with the equation @xmath69 . ]
thus the space of morphisms is the space of endomorphisms with gradings reversed , i.e. it is spanned by the identity matrix regarded as odd .
composing two such odd morphisms going in the opposite directions we get the identity endomorphism .
this agrees with the computations in section [ sec : cp ] and explains why we declared the space of morphisms between two different d2-branes to be odd .
now let us discuss the d0-brane .
consider the direct sum of objects corresponding to d2-branes with equations @xmath68 and @xmath69 .
it is easy to see that its algebra of endomorphism is the clifford algebra @xmath62 , so we propose that this object corresponds to the d0-brane .
it is easy to check that morphisms to and from other objects agree with our computations in section [ sec : cp ] .
finally , we propose an object of @xmath362 corresponding to the d0-brane in the free massive lg model with @xmath2 fields . if we bring the superpotential to the standard form eq .
( [ generalw ] ) , then we can simply tensor @xmath2 copies of the object eq .
( [ d0object ] ) .
consequently , the endomorphism algebra will also be the graded tensor product of @xmath2 copies of @xmath54 , which is isomorphic to @xmath363 .
more invariantly , let @xmath364 be the complex vector space whose coordinates we denoted by @xmath365 , let @xmath366 be the corresponding basis in @xmath63 , let @xmath367 be the dual basis in @xmath368 , and let @xmath246 be the hessian of @xmath11 .
we start with the @xmath14-graded version of the koszul resolution of the point at the origin : @xmath369 where @xmath370 , and the differential is induced by the wedge product with @xmath371 ( we use einstein s convention of summing over repeating indices ) .
now we modify the differential so that its square be @xmath11 instead of zero .
the obvious guess is @xmath372 where @xmath373 , denotes the interior product with an element of @xmath368 . using the identity
@xmath374 one can easily check that @xmath375 .
note that @xmath376 is essentially the fourier transform of the dirac operator , if we identify @xmath377 and @xmath378 with spinor bundles .
we regard the above computations as a convincing check of kontsevich s proposal for massive lg models . in this subsection
, we would like to address the following three questions .
first , how do we match b - branes with objects in the category @xmath362 if @xmath379 ? ( as explained in section [ sec : b ] , it is sufficient to consider the case @xmath380 , @xmath11 quadratic non - degenerate , and @xmath381 . )
second , is there an efficient method to compute morphisms in the category @xmath362 ?
third , assuming the validity of kontsevich s proposal , what do we learn about b - branes in massive lg models ?
that is , is there a simpler way to describe @xmath362 ? to answer these questions , we will define a functor from the category of b - branes to the category of finite - dimensional @xmath14-graded modules over the clifford algebra @xmath363 .
let us denote this category @xmath382 .
( as usual , we allow both even and odd morphisms ; thus @xmath382 is a @xmath14-graded category . )
since we set @xmath380 , @xmath381 , and @xmath11 is quadratic and non - degenerate , the category @xmath362 really depends only on @xmath2 ; we will denote this category @xmath383 for short .
there is a further functor from @xmath382 to @xmath383 .
composing these two functors gives a way to associate objects of @xmath383 to b - branes .
in fact , as explained below , the second functor is an equivalence of categories which implies that we can calculate morphisms in @xmath382 instead of @xmath383 .
this equivalence is a cousin of the much - studied koszul duality for quadratic algebras ( see below ) .
the structure of @xmath382 is quite simple : any object is a direct sum of irreducible objects ( spinor modules ) , and there is one or two non - isomorphic irreducible objects , depending on whether @xmath2 is odd or even .
thus we have a completely explicit description of @xmath382 , and therefore , by koszul duality , of @xmath383 . assuming the validity of kontsevich s conjecture , this amounts to a solution of topological open string theory for any massive lg model .
one can associate a @xmath14-graded clifford module to a b - brane as follows . for any b - brane @xmath22 ,
consider the graded vector space @xmath384 .
since @xmath208 is isomorphic to @xmath363 as a graded algebra , @xmath385 is a functor from the category of b - branes to the category of left @xmath14-graded modules over @xmath363 .
since spaces of open strings are expected to be finite - dimensional , @xmath386 is expected to be a finite - dimensional vector space .
thus @xmath385 is a graded functor from the graded category of b - branes to @xmath382 . the results of section [ sec : d ]
( see also the appendix ) imply that the functor @xmath385 maps the maximal linear b - brane to an irreducible clifford module ; for even @xmath2 maximal isotropic subspaces which belong to different irreducible families are mapped to non - isomorphic clifford modules related by parity reversal .
the d0-brane is mapped to the free module of rank one .
a linear b - brane of complex dimension @xmath387 $ ] is mapped to a module which is a direct sum of @xmath273-\ell}$ ] irreducible modules .
next we would like to explain why @xmath382 is equivalent to @xmath383 .
the relation between these two rather different - looking categories is a generalization of the so - called koszul duality for quadratic algebras @xcite .
any serious attempt to discuss koszul duality would take us out of our depth , so we will just make a few remarks which may help to orient the reader who would like to study these questions deeper .
classical koszul duality applies to quadratic algebras , i.e. @xmath13-graded algebras generated by degree-1 elements , such that all relations between generators are homogeneous quadratic .
the basic example of a dual pair is the pair @xmath388 , where @xmath389 is the symmetric algebra of a finite - dimensional vector space @xmath368 , and @xmath267 is the exterior algebra of the dual vector space .
the statement of koszul duality is that their derived categories of finitely - generated @xmath13-graded modules are equivalent .
there is a generalization of koszul duality to the case where the relations are non - homogeneous quadratic @xcite .
but the dual object in this case is not a graded algebra , but a quadratic cdg algebra .
a cdg algebra is a triple @xmath390 , where @xmath391 is a graded algebra , @xmath376 is a degree-1 derivation , and @xmath392 is a degree-2 element @xmath392 such that @xmath393 $ ] for any @xmath394 .
cdg means `` curved differential graded '' ; another name for a cdg algebra is a `` q - algebra '' @xcite .
a module over a cdg algebra @xmath390 is a graded module @xmath395 over @xmath391 equipped with a degree-1 derivation @xmath396 such that @xmath397 for any @xmath398 .
what we need is a @xmath14-graded version of non - homogeneous koszul duality .
indeed , on one hand , the clifford algebra is a @xmath14-graded quadratic algebra , while on the other hand , the category @xmath282 can be regarded as a category of modules over a certain @xmath14-graded cdg algebra .
this cdg algebra is purely even and isomorphic to @xmath275 as an algebra .
the derivation @xmath376 is identically zero , but the even element @xmath392 is not : it is given by @xmath11 . the category @xmath383 can be regarded as the derived category of the category of finitely generated cdg modules over the cdg algebra @xmath399 .
this cdg algebra is koszul - dual to the clifford algebra in the sense of refs .
@xcite , and we expect that the corresponding derived categories of modules are equivalent . more precisely , we expect that the derived category of finite - dimensional @xmath14-graded clifford modules is equivalent to the derived category of finitely - generated modules over the cdg algebra @xmath399 .
since the clifford algebra can be regarded as a deformation of the exterior algebra , and the cdg algebra @xmath399 is a deformation of the polynomial algebra , this claim looks like a generalization of the classic result of ref .
in fact , the deformed duality is in some sense simpler than the classic one , since the category @xmath382 is semi - simple and `` deriving '' it is a trivial operation ( gives us back the same category ) .
it is also more useful : while the classic duality of ref .
@xcite reduced the problem of classifying coherent sheaves on @xmath1 to a _ very difficult _ problem in linear algebra , the deformed duality reduces the problem of classifying b - branes in the free massive lg model to a _ very simple _ problem in linear algebra ( classification of finite - dimensional graded modules over a clifford algebra . ) let us describe the functors which establish the equivalence of @xmath383 and @xmath382 .
the first one , from @xmath383 to @xmath382 , is obvious : it takes an object @xmath22 of @xmath383 to @xmath400 , where @xmath401 is the object of @xmath383 described in the last paragraph of subsection [ sec : konts ] .
the mapping of morphisms is the obvious one . to define the functor acting in the opposite direction ,
let us consider for any object @xmath395 of @xmath382 the vector space @xmath402 where @xmath275 is simply the algebra of polynomial functions on @xmath12 .
since @xmath395 is @xmath14-graded , this vector space is also @xmath14-graded .
it is also an @xmath275 module , for obvious reasons .
it remains to define the twisted differential @xmath403 , i.e. an odd endomorphism of @xmath55 which squares to @xmath11 .
let @xmath63 be the vector space which appears in the definition of the clifford algebra ; we will also identify the target space @xmath27 of the lg model with @xmath63 . the twisted differential will be @xmath404 where @xmath405 is a basis in @xmath63 , @xmath365 are the corresponding linear coordinates , and the dot denotes the clifford algebra action .
it is easy to check that @xmath403 is odd , and that @xmath406
. thus we defined a map which sends an object of @xmath382 to an object of @xmath383 .
the mapping of morphisms is the obvious one : if @xmath407 is a morphism of clifford modules @xmath395 and @xmath408 , then the corresponding element of @xmath409 is @xmath410 .
it is easy to check that @xmath410 is closed , and thus is a well - defined morphism in the category @xmath383 .
the claim is that compositions of these two functors in any order are isomorphic to identity functors .
we will not try to prove this claim here , but to make it more plausible note that the mapping of objects is given by essentially the same formulas as in the classic case @xcite .
the mirror of the the nonlinear sigma model with target @xmath10 is the affine @xmath411 toda model @xcite .
the affine @xmath411 toda model is an @xmath412 landau - ginzburg theory of two chiral superfields @xmath413 and @xmath414 taking values in @xmath4 and a rational superpotential @xmath415 we can test the homological mirror symmetry conjecture by comparing the fukaya category of @xmath10 with the category of b - branes in the toda model . as discussed above , every b - brane in the toda model lies on some holomorphic curve @xmath416 .
in addition , in order for open strings to have a supersymmetric ground state , we require this curve to pass through a critical point of @xmath11 . in the @xmath411 theory
there are three distinct critical points : @xmath417 the values of @xmath11 corresponding to these critical points are pairwise distinct : @xmath418 .
there is an obvious @xmath419 symmetry which permutes the critical points .
this implies that the categories @xmath420 are all equivalent . from now on we will focus on one of them , say , the one corresponding to @xmath421 . all b - branes associated to this critical point must be complex submanifolds of the holomorphic curve in @xmath422 given by @xmath423 this curve is a singular cubic with a single node ( see fig . [
fig : cubic ] ) .
thus the category of b - branes is a full sub - category of the category of b - branes in the lg model @xmath49 .
we have seen that the latter is equivalent to the category @xmath424 .
it remains to understand which objects in the latter category correspond to b - branes .
clearly , the d0-brane sitting at the critical point @xmath425 is a valid b - brane . as for d2-branes
, they must be ( desingularizations of the ) irreducible components of the curve eq .
( [ curve ] ) .
but it is easy to see that the singular cubic is irreducible .
thus there is only one d2-brane of type b : the one which corresponds to the structure sheaf of the desingularized cubic .
the corresponding object in the `` local '' category associated to the critical point is the direct sum of the d2-brane @xmath68 and the d2-brane @xmath69 .
this direct sum is isomorphic to the d0-brane ( see section ( [ sec : cp ] ) ) .
we conclude that the basic b - brane in the toda model is the d0-brane , all other branes being direct sums of several copies of the d0-brane .
the endomorphism algebra of the d0-brane is isomorphic to @xmath62 .
we see that the category of b - branes in this case is strictly smaller than the `` local '' category @xmath426 which is equivalent to @xmath424 . as discussed in section [ sec : b ] , since the d0-brane looks like a composite of two d2-branes , one can formally add these missing d2-branes to the category of b - branes for the toda model .
the enlarged category is equivalent to the category @xmath424 .
now let us interpret these results from the point of view of homological mirror symmetry .
the mirror of the d0-brane has been identified in ref .
@xcite using the dualization argument of ref .
the mirror is a certain lagrangian 2-torus in @xmath10 equipped with a rank one trivial vector bundle and a certain flat connection .
let us be more specific .
consider the unit 5-sphere in @xmath427 , i.e. a hypersurface defined by the equation @xmath428 the quotient of this 5-sphere by a free @xmath429 action @xmath430 is diffeomorphic to @xmath10 .
in fact , the standard symplectic form on @xmath10 is obtained by restricting to @xmath431 the standard khler form on @xmath427 and then pushing it down to the quotient .
now consider a 3-torus in @xmath427 defined by the equations latexmath:[\[\label{3torus } and invariant with respect to the @xmath429 action .
hence by passing to the quotient , we obtain a 2-torus embedded in @xmath10 .
it is trivial to check that this 2-torus is lagrangian with respect to the standard symplectic form on @xmath10 .
the flat connection can be specified by its monodromy representation .
let @xmath433 and @xmath434 be the loops on the 3-torus ( [ 3torus ] ) defined by @xmath435 their images under the quotient map generate the fundamental group of our lagrangian 2-torus . according to ref .
@xcite , the mirror of the d0-brane sitting at the point @xmath436 @xmath437 corresponds to the monodromy representation which maps both generators to @xmath438 . in particular ,
the d0-brane which sits at the point @xmath425 is mirror to the trivial flat connection on the lagrangian 2-torus . as a simple check of this claim , note that the algebra of endomorphisms of a d0-brane in the model @xmath49 has euler characteristic zero . in the mirror picture ,
the corresponding object is the euler characteristic of the floer complex , which coincides with the classical euler characteristic of the 2-torus .
thus the euler characteristics match
. it would be nice to compute the floer homology groups as well and to check that they agree with the predictions of mirror symmetry .
namely , we expect that 1 .
the floer homology of the lagrangian 2-torus equipped with a rank - one flat connection is non - vanishing only for the three special flat connections defined above ; 2 .
for these choices of the flat connection , the floer homology is isomorphic to the classical cohomology of the torus as a @xmath14-graded vector space ; 3 . as a @xmath14-graded algebra ,
the floer homology is isomorphic to the clifford algebra with two generators , i.e. it is a quantum deformation of the classical cohomology ring ; 4 .
floer homology groups which compute morphisms between different flat connections of rank one vanish .
it was argued above that we can formally add d2-branes to the category of b - branes .
it is reasonable to ask if this procedure is consistent with or perhaps even forced on us by homological mirror symmetry . to answer this question
we need to identify the mirrors of the added d2-branes .
there are two such d2-branes for each critical level set . for each of them
the euler characteristic of the endomorphism algebra is @xmath277 .
if we assume that the mirror of a d2-brane is a lagrangian submanifold , then it must be homeomorphic to a real projective plane @xmath439 .
but since @xmath439 is not orientable , it is not an admissible object of the fukaya category ( one needs orientability in order to define @xmath14-graded maslov index and @xmath14-grading on the floer complex ) . we conclude that the mirrors of the added d2-brane can not be lagrangian submanifolds , and therefore homological mirror symmetry does not force us to include them on the b - side . on the other other hand , if we added d2-branes on the b - side , we can maintain homological mirror symmetry by adding certain objects on the a - side .
in other words , we would like to regard the lagrangian 2-torus with a trivial flat connection , which is mirror to the d0-brane , as a direct sum of two irreducible objects , which are mirror to the d2-branes .
but since there are no such objects in the fukaya category , we simply add these direct summands `` by hand . '' let us clarify what we mean by adding direct summands `` by hand . ''
let @xmath37 be an object of an additive category @xmath440 .
a projector is an element of @xmath441 which satisfies @xmath442 .
given any projector , we would like to have the corresponding direct summand , i.e. an object @xmath443 and a pair of morphisms @xmath444 and @xmath445 such that @xmath446 and @xmath447 .
if @xmath443 does not exist for all projectors and for all @xmath37 , then we look for the smallest additive category which contains @xmath440 as a full subcategory and in which every projector has a direct summand . to summarize , to maintain homological mirror symmetry , we must either add formal direct summands on both a and b sides , or on neither side .
the mirror in this case is the lg model with target @xmath448 and the superpotential @xmath449 here @xmath450 and @xmath451 are nonzero complex numbers whose logarithms are mirror to the periods of the complexified khler form on the two @xmath9 s .
this superpotential has four non - degenerate critical points . for generic @xmath452
there are four critical level sets all of which look like a cubic with a node .
thus we are in exactly the same situation as in the previous subsection , and the only b - branes are d0-branes sitting at the critical points .
another way to see these d0-branes is to note that the lg model is a product of two lg models with the superpotential @xmath453 .
this model is mirror to @xmath9 and has been studied in ref .
its only b - branes are d0-branes sitting at the two critical points of the superpotential .
taking tensor products of pairs of such b - branes gives us four d0-branes discussed above .
the mirror of each d0-brane is a lagrangian 2-torus with some flat connection .
indeed , the mirror of a d0-brane in the model @xmath453 is the equatorial circle on @xmath9 @xcite , therefore the mirror of a d0-brane in the product model is the product of two equatorial circles .
the monodromy around the two generators of the fundamental group is @xmath454 . for @xmath455
something special happens both on the a and b sides . on the a side
, we get a new lagrangian submanifold which is homeomorphic to a 2-sphere . to see this ,
let @xmath353 and @xmath456 be coordinates on the standard affine patches on the two @xmath9 s .
consider the `` anti - diagonal '' 2-sphere given by @xmath457 .
let @xmath458 be the fubini - study form on @xmath9 , @xmath459 be the projection maps from @xmath19 to the two factors , and @xmath460 be complex numbers .
it is trivial to check the restriction of @xmath461 to the `` anti - diagonal ''
2-sphere vanishes if and only if @xmath462 .
thus the 2-sphere is lagrangian if and only if @xmath455 . on the b side ,
the critical level set @xmath66 now contains two critical points .
the equation of this critical level set @xmath463 shows that it is reducible .
the irreducible components are a line and a non - singular quadric which intersect transversally at two points ; these are the two critical points mentioned above .
we have two irreducible d2-branes of type b corresponding to the two irreducible components of the critical level set .
it is easy to see that one is isomorphic to the shift of the other , while their sum is isomorphic to the sum of two d0-branes sitting at the two critical points .
note that this is another example where the category of b - branes is strictly smaller than the sum of `` local '' categories associated to critical points .
this happens because all d2-branes pass through both critical points in the set @xmath66 .
thus a single d0-brane sitting at a critical point is irreducible . of course , if we only look at the infinitesimal neighborhood of one of the critical points , then we are in the same situation as in the model @xmath49 , and the d0-brane appears to be composite .
if desired , we can enlarge the category of b - branes by adding all formal direct summands .
then it will become equivalent to the sum of categories attached to the two critical points ( each of which is equivalent to @xmath424 ) , and each d0-brane will be the sum of two irreducible objects .
now let us match the objects on a and b sides .
d0-branes correspond to `` equatorial '' lagrangian tori , as before .
the mirror of a d2-brane must be a lagrangian 2-sphere .
indeed , each d2-brane passes through two critical points , each of which contributes @xmath277 to the euler characteristic of the endomorphism algebra .
an obvious conjecture is that the two d2-branes are mirror to the lagrangian 2-sphere discussed above and its shift ( i.e. orientation - reversal ) . if this is true , then the sum of the lagrangian 2-sphere and its shift must be isomorphic ( in the fukaya category ) to the sum of two `` equatorial '' lagrangian tori with monodromies @xmath464 and @xmath465
. it would be interesting to check this by computing the floer homology between all the objects involved .
in this paper we have described the category of b - branes for the free massive lg model with @xmath2 chiral fields .
we also argued that this allows one to determine the category of b - branes for an arbitrary massive lg model .
the most striking feature of our results is their simplicity .
for example , if we consider the free massive lg model , there is a multi - parameter family of maximal isotropic subspaces of the quadric @xmath66 , but they are all isomorphic as objects of the category of b - branes ( up to a shift ) .
moreover , b - branes of lower dimension , including the d0-brane , are isomorphic to direct sums of b - branes of maximal dimension .
these rather counter - intuitive observations solve the problem of computing tree - level topological open string correlators in these models .
it is interesting to compare our results with those of refs .
@xcite , where a general framework for classifying d - branes in 2d topological field theories has been proposed . in our case , the 2d tft in the bulk is rather trivial : it is isomorphic , as a frobenius algebra , to @xmath43 with its unique frobenius structure . the theory of ref .
@xcite ( generalized to the @xmath14-graded case ) tells us that the algebra of open strings connecting a brane with itself must be simple .
we saw that in our case endomorphism algebras of b - branes are all isomorphic to clifford algebras , and these are indeed simple ( as @xmath14-graded algebras ) .
however , unlike in the purely bosonic case , in the @xmath14-graded case not every two simple finite - dimensional algebras are morita equivalent .
in fact , there are two morita - equivalence classes of such algebras , represented by @xmath43 and @xmath54 .
a clifford algebra with @xmath21 generators is morita - equivalent to @xmath43 or @xmath54 depending on whether @xmath21 is even or odd .
we have seen that when the number of fields @xmath2 is even ( resp .
odd ) only even ( resp .
odd ) values of @xmath21 occur .
this suggests that it is impossible to have a topological open string theory which includes d - branes of both kinds .
indeed , we have seen that all pairings between spaces of morphisms induced by 2-point correlators are either even or odd , depending on whether @xmath2 is even or odd . on the other hand ,
one of the basic axioms of topological open string theory is that all pairings must have the same parity @xcite .
this observation provides a simple counter - example to the belief that a 2d tft determines uniquely the associated category of topological boundary conditions .
in fact , we can make a stronger statement . given any 2d scft representing a superstring background , we can tensor it with the topological lg model @xmath50 . since the latter theory is trivial
, this does not change the closed string sector .
but the open string sector does change : one has to tensor every `` physical '' d - brane with the d0-brane of the lg model , and this results in tensoring the open string spectrum of each d - brane with @xmath54 .
this is equivalent to the introduction of an odd chan - paton label .
thus we have two inequivalent open string theories for a given closed string theory .
this particular ambiguity is rather mild and can be easily eliminated .
the difference between odd and even @xmath2 comes from the number of fermionic zero modes on a disk , or equivalently from the parity of the bilinear forms computed by the 2-point disk correlators .
thus to specify completely the open string theory we are dealing with , it is sufficient to fix the parity of all bilinear forms
. it would be interesting to extend the considerations of this paper to lg models which flow to non - trivial scfts in the infrared limit .
for example , one could study landau - ginzburg realizations of @xmath0 minimal models .
for these theories much information about b - branes is available from the boundary state formalism , and it would interesting to see if it is consistent with kontsevich s proposal . by analogy with the massive case , one expects that the category of b - branes will be describable in terms of modules over the algebra of endomorphisms of a d0-brane .
from the mathematical viewpoint , this algebra must be related by a koszul - like duality to the cdg algebra which appears in kontsevich s proposal .
it appears that for @xmath11 of degree
higher than two koszul duality relates kontsevich s cdg algebra to a finite - dimensional @xmath7-algebra @xcite .
b - branes should correspond to finite - dimensional @xmath7-modules over this @xmath7-algebra . in this way solving topological open string theory is reduced to a problem in linear algebra .
hopefully , the latter problem is manageable . in the axiomatic approach of refs .
@xcite , topologically twisted @xmath0 minimal models correspond to non - semi - simple frobenius algebras
. it would be interesting to explore the uniqueness of the open string sector in such models .
in this appendix we collect some well - known facts about complex clifford algebras and their modules .
let @xmath63 be a complex vector space of dimension @xmath2 , and @xmath113 be a non - degenerate symmetric bilinear form on @xmath63 .
clifford algebra @xmath181 has @xmath63 and the identity as its set of generators , and the following relations : @xmath467 as a vector space , @xmath181 is isomorphic to @xmath267 and therefore has dimension @xmath468 .
we can regard @xmath181 either as an ordinary associative algebra , or as a @xmath14-graded algebra , such that all the generators are odd . in the latter case
, the grading corresponds to the decomposition of @xmath267 into polyvectors of even and odd degree .
since the isomorphism class of @xmath181 depends only on the dimension @xmath2 of @xmath63 , we will also use the notation @xmath363 to denote this isomorphism class .
if @xmath469 and @xmath470 are complex vector spaces with non - degenerate bilinear forms @xmath471 and @xmath472 , then @xmath473 here all clifford algebras are regarded as @xmath14-graded algebras , and @xmath474 denotes their @xmath14-graded tensor product .
if @xmath2 is even , then @xmath181 regarded as an ungraded algebra is isomorphic to the algebra of complex @xmath475 matrices , which we will denote @xmath476 .
if @xmath2 is odd , then @xmath181 regarded as an ungraded algebra is isomorphic to @xmath477},{{\mathbb c}}){\oplus}mat(2^{[n/2]},{{\mathbb c}})$ ] . in particular
, @xmath54 is isomorphic to @xmath478 .
we see that @xmath181 is a simple algebra only for even @xmath2 .
however , if we regard it as a @xmath14-graded algebra , then it is simple for all @xmath2 .
now let us discuss finite - dimensional modules over @xmath181 .
the category of clifford modules is semi - simple , i.e. every exact sequence splits .
thus every clifford module is a direct sum of irreducible modules .
the number and properties of irreducible modules depend on the parity of @xmath2 , as well as whether we regard @xmath181 as a @xmath14-graded algebra .
if we neglect the grading , then for even @xmath2 we have a unique irreducible module @xmath479 of dimension @xmath480 .
it is called the spinor module and can be constructed as follows . pick a pair of subspaces @xmath481 of @xmath63 such that both @xmath249 and @xmath11 are isotropic with respect to @xmath113 , @xmath482 , and @xmath483 .
one can easily see that @xmath113 gives a non - degenerate pairing between @xmath249 and @xmath11 and thus we may identify @xmath11 with @xmath484 .
set @xmath485 , and define the action of clifford algebra on @xmath479 as follows : if @xmath486 where @xmath487 and @xmath488 , then for any @xmath489 we let @xmath490 here we used the identification of @xmath11 with @xmath484 mentioned above . for odd @xmath2 clifford algebra is a sum of two matrix algebras , and therefore there are two non - isomorphic irreducible modules of dimension @xmath273}$ ] ( two spinor modules ) .
for example , for @xmath47 the algebra is generated by the identity and an odd element @xmath491 with a single relation @xmath492 ; the two irreducible modules are one - dimensional , with the action of @xmath491 given by @xmath493 . for general @xmath2 one can use the property eq .
( [ tensorcl ] ) to reduce the problem to the cases already considered .
if we regard @xmath181 as a @xmath14-graded algebra , then we should look for @xmath14-graded irreducible modules . for
even @xmath2 there are two inequivalent choices of grading on the spinor module related by parity reversal .
therefore we have two non - isomorphic irreducible spinor modules @xmath479 and @xmath494 . for odd @xmath2 the `` minimal '' @xmath14-graded module
has dimension @xmath495 ; as an ungraded module , it is isomorphic to the direct sum of two inequivalent irreducible ungraded modules . furthermore , the choice of grading is unique up to isomorphism .
we will denote this unique spinor module by @xmath479 .
for example , for @xmath47 @xmath496 , and @xmath491 acts as any of the three pauli matrices , say @xmath497 .
then the parity operator can be chosen to be @xmath498 .
to summarize , for even @xmath2 any @xmath14-graded clifford module is a direct sum of several copies of two inequivalent spinor modules @xmath479 and @xmath494 . for odd @xmath2
the situation is the same , except that @xmath479 is isomorphic to @xmath494 .
the dimension of the spinor module is given by @xmath499}$ ] for any @xmath500 .
in particular , @xmath181 regarded as a left module over itself is a direct sum of @xmath273}$ ] copies of spinor modules . for @xmath2
even half of them are @xmath479 s , and the other half are @xmath494 s .
we are deeply grateful to maxim kontsevich for sharing with us his ideas about b - branes in landau - ginzburg models , and to alexander polishchuk for pointing out the relevance of non - homogeneous koszul duality .
we also thank kentaro hori and dmitri orlov for reading a preliminary draft of the paper and making a number of valuable comments .
the first author would like to thank institut des hautes tudes scientifiques for hospitality during the writing of this paper .
this work was supported in part by the doe grant de - fg03 - 92-er40701 . | we study topological d - branes of type b in @xmath0 landau - ginzburg models , focusing on the case where all vacua have a mass gap . in general ,
tree - level topological string theory in the presence of topological d - branes is described mathematically in terms of a triangulated category .
for example , it has been argued that b - branes for an @xmath0 sigma - model with a calabi - yau target space are described by the derived category of coherent sheaves on this space .
m. kontsevich previously proposed a candidate category for b - branes in @xmath0 landau - ginzburg models , and our computations confirm this proposal .
we also give a heuristic physical derivation of the proposal .
assuming its validity , we can completely describe the category of b - branes in an arbitrary massive landau - ginzburg model in terms of modules over a clifford algebra .
assuming in addition homological mirror symmetry , our results enable one to compute the fukaya category for a large class of fano varieties .
we also provide a ( somewhat trivial ) counter - example to the hypothesis that given a closed string background there is a unique set of d - branes consistent with it . |
according to standard solar models , the solar luminosity increased from about 70% of the present solar luminosity at 4.5 gyr before present ( bp ) to its present value .
if earth were a black body , its temperature would have been @xmath1k lower , enough to have kept large parts of it frozen until about 1 - 2 gyr bp . besides however the past eon , and the eon between 2 and 3 gyr bp
, it appears that glaciations were altogether absent from the global surface .
this is the crux of the faint sun paradox [ _ sagan & mullen _ , 1972 ; _ pollack _ , 1991 ; _ sagan & chyba _ , 1997 ] .
a common solution to this apparent paradox is that larger amounts of the greenhouse gas ( ghg ) co@xmath2 compensated for the cooler sun [ _ kuhn & kasting _ , 1983 ; _ kasting _ , 1993 ] .
however , some upper limits on the atmospheric partial pressure of co@xmath2 ( @xmath3co@xmath2 ) suggest that it was at most modest . for example , _ rye et al . _ [ 1995 ] find @xmath3co@xmath4 bar between 2.2 and 2.7 gyr bp , with indications that it could not have been much higher earlier [ _ sleep & zahnle _ , 2001 ] . at these levels , the ghg warming could at most compensate for about half the reduction in the solar flux in the global radiation budget corresponds to the same global temperature change irrespective of whether it is due to extra solar flux in optical and near - ir or the blocking of far - ir due to a higher level of @xmath3co@xmath2 .
this assumption is reasonable but not necessary . ] .
thus , it appears that co@xmath2 could have been important for warming the early earth , but not sufficient to resolve the faint sun paradox by itself .
note however that theoretical considerations do suggest that @xmath3co@xmath2 was higher in the early terrestrial atmosphere [ _ kasting _ , 1987 ] , such that it is likely that it did contribute at least some ghg warming .
because of these problems , it was suggested that other ghgs could have been important for warming the young earth . in particular , it was suggested that small amounts of nh@xmath5 could have supplied the required ghg warming [ sagan & mullen , 1972 ; _ sagan & chyba _ , 1997 ] . although not impossible , it is not easy to keep nh@xmath5 from irreversibly photolyzing into h@xmath2 and n@xmath2 .
another suggestion was that ch@xmath6 played the major ghg warmer [ _ pavlov et al .
_ , 2000 ] .
this solution requires a long residency time of methane in the atmosphere , and probably dominance of methanogenic bacteria .
unfortunately , there are currently no reasonable bounds on the amounts of methane or ammonia in the early atmosphere .
thus , this type of solutions can neither be ruled out , nor proven at his point .
other effects are more certain but probably not large enough to solve the faint sun paradox .
in particular , the young earth rotated faster than today and it may have had less land mass . employing a typical general circulation model ( gcm ) , it was found that removing the land mass altogether could increase the temperature by 4@xmath7k , while a 14 hr day would further increase the temperature by @xmath8k [ _ jenkins _ , 1993 ] .
another solution to the faint sun paradox involves strong solar winds , such that the earlier solar mass was higher .
a 10% more massive sun would have had a larger luminosity , enough to more than compensate for the early faint sun and also explain the observed lithium and beryllium depletion [ _ graedel et al .
_ , 1991 ] .
however , solutions with a relatively _ gradual _ loss of mass from the sun , with a typical time scale of @xmath9gyr ( as opposed to a mass loss time scale of a few 100 myr ) are inconsistent with helioseismology , rendering this solution less probable [ _ guzik & cox _ , 1995 ] .
moreover , a solar wind history can be reasonably reconstructed if the sun is compared to similar nearby stars at various ages .
this gives a benign mass loss which is insufficient to resolve the paradox at any time of the terrestrial history [ _ wood et al .
_ , 2002 ] .
nevertheless , even a modest mass loss could potentially have interesting implications .
the stronger wind was more effective at lowering the galactic cosmic ray flux ( crf ) .
this , as we shall soon see , can present yet another possibility for helping resolve the faint sun paradox .
it should however be stressed that it is not unlikely that the true solution is a combination of several of the above factors . various empirical evidence [ _ tinsley & deen _ , 1991 ; _ svensmark _ , 1998 ; _ egorova et al .
_ , 2000 ; _ marsh & svensmark _ , 2000 ; _ todd & kniveton _ , 2001 ; _ shaviv _ , 2002a , b ; _ shaviv & veizer _ , 2003 ] , theoretical analyses [ _ ney _ , 1959 ; _ dickinson _ , 1975 ; _ harrison & aplin _ , 2001 ; _ yu _ , 2002 ] , and experimental results [ _ harrison & aplin _ , 2001 ; _ eichkorn et al . _ , 2002 ]
link cosmic rays ( crs ) to climate on earth
. the apparent crf / climate link could originate through an effect that atmospheric ionization , which is governed by the crf , may have on cloud formation . the empirical evidence is based on several significant correlations between the crf and climate on earth over times scales ranging from days to geological time scales .
the theoretical and experimental evidence describe various elements of an emerging physical picture relating atmospheric ionization , through the formation of condensation nuclei by charged ion clusters , to the formation of clouds .
even with the above evidence , this crf / climate link is still a highly debated topic .
this is primarily because a robust physical picture is still missing . for an objective review ,
the reader is encouraged to read _ carslaw et al . _ [ 2002 ] . it should be emphasized that the analysis which follows rests on the assumption that this link indeed exists an issue which we will address again in list of caveats in [ sec : discussion ] .
irrespective of the outcome of this debate , since there is now enough evidence to seriously consider the possibility of crs as a variable affecting the climate , we should consider the various implications of this link while bearing in mind that the underlying basis deserves extensive study .
we begin in [ sec : estimate ] with an estimate of the total warming effect that the strong early solar wind has had , and compare it to the cooling induced by the fainter sun . in [ sec : evolution ] , we use the actual solar wind evolution ( as deduced from nearby stars ) combined with the variable star formation rate ( sfr ) in the milky way ( mw ) to calculate the evolution of the terrestrial temperature .
we summarize our results in [ sec : results ] and discuss them in [ sec : discussion ] .
we assume henceforth that cosmic rays do have a cooling effect .
thus , the positive crf / low altitude cloud cover correlation [ _ svensmark _ , 1998 ) implies that a higher crf will have a net cooling effect under the suspected crf / climate link .
second , the long term _
positive _ correlation between crf variations and ice - age epochs suggests that there is a _ negative _ correlation between crf and temperature [ _ shaviv _ , 2002a , _ shaviv & veizer _ , 2003 ] . ] . under this assumption , the stronger wind of the young sun had a warming effect .
this is because the stronger wind necessarily lowered the galactic crf reaching the inner solar system .
a lower crf would in turn translate into higher temperatures through the crf / climate link .
thus , a stonger solar wind should have had a warming effect that acted to compensate the solar dimness . to quantify this effect ,
two basic numbers are required .
first , it is necessary to know the ratio between changes in the crf and changes in the global temperature .
this will allow us to translate solar wind variations into temperature change on earth .
second , if we wish to compare this effect with the lower temperature associated with the faint sun , and also with possible ghg warming , we need to know how changes in the radiation budget translate into a global temperature change .
the relation between radiative forcing and temperature change is still an open and highly discussed question in climatology . for a blackbody earth , this number is 0.30@xmath7k/(w
/ m@xmath10 ) .
earth however does not behave as a black body , and various positive and negative feedbacks act to either amplify or decrease this number by a factor @xmath11 .
values obtained in gcms [ _ ipcc _ , 1995 ] range between @xmath12 and @xmath13 .
there are however debated claims that various negative feedbacks should be considered , yielding a much smaller @xmath11 of @xmath14 [ _ lindzen et al .
_ , 2001 ] , or even @xmath15 [ _ ou _ , 2001 ]
. this number was also obtained from analyses of actual temperature variations .
for example , from the rather quick return of the global temperature to its average " after a series of volcanic erruptions , it was deduced that @xmath16 with @xmath17 being inconsistent with the effects of volcanic driving [ _ lindzen & giannitsis _ , 1998 ] . on the other hand , _ gregory et al . _ [ 2002 ] find that a lower limit of @xmath18 ( at 95% confidence ) can be placed by considering the increase of ocean heat content .
another constraint was obtained by _
covey et al . _
[ 1996 ] who compared @xmath3co@xmath2 changes to temperature variations over geological time scales ( in particular , the last glacial maximum , the crataceous and early eocene ) .
the authors of this study found that @xmath11 is consistent with the range of values obtained in gcms .
however , the latter two constraints were obtained while neglecting the possible forcing of the crf and the nonthermal solar activity .
thus , if crf / climate link is real , these two constraints are not applicable anymore . once crf variations are considered over geological time scales , a markable correlation is obtained between crf as a driver and temperature change .
this correlation , combined with the lack of correlation with @xmath3co@xmath2 variations , yields @xmath19 ( or @xmath20 at 90% confidence and @xmath21 at 99% confidence ) [ _ shaviv & veizer _ , 2003 ] .
thus , _ if the crf / climate connection is valid _
, which we assume here to be the case , we should consider @xmath19 as the most probable value , but also @xmath22 as an extreme upper limit .
the same combined geological and astronomical analysis [ _ shaviv & veizer _ , 2003 ] could also measure the relation between temperature change and relative crf variations , which was parameterized through the form @xmath23,\ ] ] where @xmath24 is the energy flux reaching the troposphere ( to which the ionization is proportional ) , and @xmath25 is the flux reaching today .
the power @xmath26 relates changes in the ionization rate to changes in global temperature .
since the expected route is most likely through the effect that atmospheric ionization has on the formation of condensation nuclei , we should have @xmath27 [ _ yu _ , 2002 ; _ harrison & aplin _ , 2001 ] .
it was also found that @xmath28c , where most of the uncertainty is due to the inaccurate determination of the crf variations from meteorites .
we reduced here the upper range .
although a larger @xmath29 is consistent with the meteoritic data , it is physically inconsistent as it requires too large an albedo variation of earth to be physically reasonable , as will soon be evident . when the solar system was young , the solar wind is expected to have been significantly stronger .
this implies that most of the crf responsible for tropospheric ionization was blocked ( i.e. , the energy flux satisfied @xmath30 ) . today , for comparison
, the flux variations are only @xmath31 .
thus , we should expect an average global surface temperature increase of @xmath32k , when the wind was significantly stronger than today .
next , the decrease in solar luminosity as we go back to the end of the heavy bombardment era ( at about 4 gyr bp ) , is about @xmath925% .
this is an inevitable result of all standard models .
the only astrophysical remedy is having a heavier younger sun .
however , as mentioned in the introduction , a strong and slowly decaying wind that would allow the young sun to be bright for a long enough duration , is inconsistent with the winds of nearby stars [ _ wood et al .
_ , 2002 ] .
the @xmath33 reduction in solar flux corresponds to a decrease of @xmath960 w m@xmath34 relative to the @xmath35 w m@xmath34 reaching the surface today . for @xmath36 or
@xmath37 , the average global surface temperature decrease @xmath38 associated with the fainter sun is @xmath39k , @xmath40k or @xmath41k respectively .
thus , for climate sensitivities consistent with the crf / climate link ( @xmath42 ) , we find that the solar dimness can in principle be completely compensated for . nevertheless , to actually obtain @xmath43
, we require values of @xmath11 and @xmath29 which are consistent with observations but are theoretically uncomfortable .
for example , @xmath44k with @xmath45 implies that crs change the albedo by 28% , which is nearly impossible .
the more reasonable values of @xmath46k and @xmath47 yield a more realistic albedo change of 12.5% . in this case
however , @xmath48k relative to today .
this implies that the average hadean and archean temperatures would have been well above freezing .
however , a full resolution of the faint sun paradox requires the temperatures at these epochs to have been higher than today , since current day earth has glaciations , while glaciations where absent from the young earth ( e.g. , _ crowell _ [ 1999 ] ) . with @xmath49 ,
the missing @xmath50k can be compensated with @xmath51 w m@xmath34 .
this could arise from a modest " ghg warming .
for example , @xmath52 bar of co@xmath2 , which is about half of the present upper limits at @xmath53 gyr bp [ _ rye et al .
_ , 1995 ; _ sleep & zahnle _ , 2001 ] would provide @xmath54 w m@xmath34 of ghg heating on the atmosphere is @xmath55 $ ] where @xmath56 is the pre - industrial level of atmospheric co@xmath2 [ _ hansen et al .
_ , 1998 ] ] , ensuring that the early climate was warm enough to inhibit the appearance of glaciations .
to estimate the secular change in the solar wind modulation of the crf , we first require the evolution of the solar wind parameters ( i.e. , mass loss rate and velocity ) .
we will then translate this evolution to variations in the crf reaching earth . for the evolution of the mass loss rate @xmath57
, we use the observational results of _ wood et al . _
[ 2002 ] who found that the mass loss rate of nearby solar - like stars ( main sequence stars of spectral type g and k ) evolves as : @xmath58 with @xmath59 .
next , we take the velocity of the wind to be constant .
this is because the terminal wind velocity depends primarily on the escape velocity from the sonic radius of the wind , which is expected to change only logarithmically with the mass loss rate ( e.g. , _ lamers & cassinelli _ [ 1999 ] ) .
this implies that to first approximation we can take @xmath60 in a relation of the form @xmath61 , but we leave @xmath62 as a free parameter for generality .
the energy loss suffered by crs as they journey from the heliopause to earth is through multiple scattering off the magnetic field of the solar wind .
the average energy lost , for crs with energies significantly larger than their rest mass , can be approximated with @xmath63 [ _ perko _ , 1987 ] , where @xmath64 is the radius of the heliopause and @xmath65 is a constant proportional to @xmath66 , the diffusion coefficient of crs in the solar wind .
this expression is valid _ only if _ @xmath67
, @xmath68 and @xmath69 . thus , for the calculation of the energy loss we further require the scaling of both @xmath64 and @xmath66 with @xmath57
. the value of @xmath64 can be obtained by equilibrating the solar wind ram pressure at its termination @xmath70 with the average pressure of the interstellar medium ( ism ) , which we assume to remain constant on average .
this yields @xmath71 .
the diffusion coefficient @xmath66 depends on the effective mean free path @xmath72 arising from the larmour gyration of the crs in the disordered magnetic field of the solar wind . since @xmath73
, we have @xmath74 .
thus , the last scaling we seek is in the form @xmath75 . it would allow us to relate @xmath57 to changes in @xmath76 . at the radius of earth s orbit ,
the magnetic field energy density in the solar wind is typically 50 times smaller than the kinetic energy density of the wind .
however , magnetic field fluctuations should decay as @xmath77 [ _ jokipii_,1973 ] , ( whether frozen - in or alfvnic wave fluctuations ) , implying that at a few solar radii , where the solar wind is accelerated , the magnetic field energy density ( @xmath78 ) is comparable to the kinetic energy density ( @xmath79 ) .
although the current mass loss is probably not unique , we find equipartition between the two energy densities at the acceleration region of the wind .
thus , it is reasonable to assume that this tendency for equipartition remains also for different @xmath57 s .
in such a case , @xmath80 at any given radius .
this implies that @xmath81 if equipartition remains . the combination of these scalings yields @xmath82 , with @xmath83 . with nominal mass loss evolution ( @xmath84 ) , constant wind velocity ( @xmath85 ) and aforementioned equipartition ( @xmath86 )
, we have @xmath87 . the crf spectrum ( i.e. , the differential number flux ) in the ism can be written as @xmath88 for energies greater than about @xmath89 gev , where @xmath90 is the cr number flux
, @xmath91 is a normalization constant , and @xmath92 .
furthermore , the differential number flux @xmath93 reaching earth is related to @xmath94 through @xmath95 . since only crs with an energy larger than the cutoff energy @xmath96 gev
can actually reach the troposphere , we find that the total energy @xmath97 deposited in the troposphere is @xmath98 more interesting is the energy flux at a time @xmath99 ( after the formation of the solar system ) relative to the flux today , at @xmath100 .
we define @xmath101 .
since @xmath102 gev [ _ perko _ , 1987 ] , we have @xmath103 , and also @xmath104 eq . [ eq : crft ] for the temperature change induced by the wind evolution now becomes : @xmath105 d,\ ] ] where @xmath106 is the nearby sfr to which the crf is proportional myr .
because this time scale is practically instantaneous ( relative to 4 gyrs ) and because crs ( at 10 gev ) originate from supernova explosions , the crf will be proportional to the sfr ( see references in _ shaviv _ [ 2002b ] ) . ] .
for the sfr , we take the results of _ rocha pinto et al . _
[ 2000 ] who derived the mw sfr by analyzing the chromospheric ages of hundreds of late type dwarfs .
the results of this comprehensive study are consistent with previous analyses , such as _ barry _ [ 1988 ] who too used chromospheric ages , or _ scalo _ [ 1987 ] who studied the mass distribution of nearby stars . in particular
, it appears that the sfr was higher than average between 2 and 3 gyr before present , while it was below average between 1 and 2 gyr bp .
interestingly , the mw sfr activity is correlated with sfr activity in the large magellanic cloud ( lmc ) .
it appears that there was a significant increase in the sfr in the lmc around 2 gyr bp [ _ gallagher et al .
_ , 1996 ; _ vallenari et al . _ , 1996 ; _ dopita et al .
_ , 1997 ] and that between 0.7 to 2 gyr bp , the sfr in the lmc was below its average [ _ westerlund_,1990 ]
. this correlation could be coincidental , but may reflect the tidal interactions between the two galaxies , during their perigalactica sometime around 0.2 - 0.5 gyr bp and around 1.6 - 2.6 gyr bp [ _ gardiner et al .
_ , 1994 ; _ lin et al .
_ , 1995 ] .
irrespective of the origin of the sfr variability , it modulated the crf reaching the solar system , and presumably also climate on earth .
the fact that the moon was closer to earth , and both were rotating faster [ _ zharkov _ , 2000 ] and that earth probably had a smaller land mass are not large effects by themselves [ _ jenkins _ , 1993 ] .
we will not try to model them accurately . nevertheless , _
jenkins _ [ 1993 ] has shown that no land mass and a 14 hr day will correspond to an average global temperature increase of typically 5.5@xmath7k .
this however was obtained with gcm models which typically have @xmath107 .
if , on the other hand , negative feedbacks exist to reduce @xmath11 , the increase in temperature will be smaller accordingly . for @xmath108
, we should expect that the whole effect of no land mass and a faster rotation would contribute only a @xmath109k warming that gradually decreases with time between the formation of earth and today . to mimic these two effects ,
we approximate the temperature correction using @xmath110 2.2^\circ$]k where @xmath111 gyr is the time today .
this is not a large correction at all , but unlike effects such as ghg warming by nh@xmath5 or ch@xmath6 , which can potentially be much larger , the effects of land mass and rotation can be considered somewhat more quantitatively .
using the above physical components , we proceed to calculate the evolution of the average terrestrial temperature .
the results are depicted in fig . 1 for various models which either include or exclude the solar wind / crf effect using nominal values for the parameters , the effects of radiative forcing , and also the possibility of ghg warming by modest levels of @xmath3co@xmath2
. the figure recapitulates the main conclusion of [ sec : estimate ] .
namely , coupling the crf / climate effect to the stronger wind of the young sun , results with a higher temperature .
the effect is large , but if the nominal values are used it compensates only @xmath112 s of the temperature reduction associated with the dim sun .
thus , either the crf effect is larger than inferred from the nominal parameters , or more likely , there is some ghg warming . by estimating the actual evolution of the temperature
, we find additional conclusions to those obtained in [ sec : estimate ] .
first , the effects of sfr variability , spiral arm passages and ghg warming are all comparable in amplitude .
second , because of the different time scales associated with the slowly varying sfr and the fast " oscillations associated with spiral arm passages ( over @xmath113 yr ) , we find that at any epoch with a given sfr , a range of temperatures should exist from spiral arm variability .
in other words , in addition to the statistical error on the average sfr [ _ rochapinto et al .
_ , 2000 ] , we should consider that the location within the mw gives rise to additional variability of the current crf [ _ lavielle et al .
_ , 1999 ; _ shaviv _ , 2002b ] but periods during which the solar system spent near spiral arms , the solar system was bathed with a flux higher than today , which presumably reduced the global temperatures enough to allow glaciations . ] .
thus , not only is the _ average _ important , but also the expected lower limit of the temperature range if we wish to compare with geological evidence for the presence of glaciations . on the other hand ,
if we had a good record for the presence of extremely high temperatures , then the expected upper limit of the temperature range would have been the important variable . if we study fig . 1 and consider models with the nominal parameters of @xmath108 and @xmath114k , we find that only during the past eon ( neo - proterozoic phanerozoic period ) , and the period of 2 - 3 gyr bp ( late - archean
huronian ) could there have been climatic periods during which it was cold enough as to produce temperatures lower than today and explain glaciations .
this is consistent with the actual geological evidence , which indicates that only during the neo - proterozoic and phanerozoic ( past eon ) , during the huronean ( 2.2 - 2.4 gyr bp ) and late archean ( 2.9 - 3.0 gyr bp ) were glaciations present on earth ( e.g. , _
crowell _ [ 1999 ] ) .
the exception is the first eon , which for the nominal parameters should have been cold enough to have had glaciations ( though not as cold as 2.5 gyr bp ) .
since there is no evidence to support their existence , we return back to the point that even a modest ghg contribution , for example , by a modest 0.01 bar co@xmath2 atmosphere , can raise the temperature of the very young earth such that with the reduced crf , the temperature is kept above today s values .
the solid line is the nominal model with @xmath3co@xmath2 kept at todays value while the dashed line has a saturation @xmath3co@xmath2 of 0.02 bar before 3 gyr bp .
_ panel ( b ) _ describes the effect of changing the mass loss evolution power law @xmath115 ( within observed error range ) .
_ panel ( c ) _ describes the effect of the power @xmath26 which relates the atmospheric ionization rate change to temperature change .
_ panel ( d ) _ describes the effect of changing the ratio between the average sfr in the past few 100 myr and the current rate ( nominal value is @xmath116 , _ lavielle _ [ 1999 ] ) . _ panel ( e ) _ describes the effect of a changed climatic sensitivity to radiation forcing .
_ panel ( f ) _ describes the effect of a changed climatic sensitivity to crf changes .
the normalization is such that for today s flux , which is about @xmath117 higher than the average over the past eon , the temperature is @xmath118c . ]
2 is a parameter study showing the effects of changing various model parameters from their nominal values .
it is apparent from the figure that there are two types of parameters .
the first type are those which affect the total crf / solar wind effect when the solar wind is completely switched off , or affect the way in which the effect of the dim sun translates into a temperature change .
these parameters determine the gross temperature behavior .
for example , @xmath11 , @xmath29 and maximum level of @xmath3co@xmath2 determine the temperature on the very young earth . the second type of parameters are secondary in the sense that they only affect the evolution of the temperature between the young earth and today .
a good example is the power @xmath115 determining the temporal evolution of @xmath57 .
it does not affect the temperature in the extremely young atmosphere nor the temperature today , since the former is obtained in the limit of a very strong wind ( irrespective of how strong it actually is ) , and the latter is simply given by the current @xmath57 . among the secondary parameters ,
the most important is the power @xmath26 which relates the ionization rate to temperature fluctuations .
the value @xmath119 is theoretically preferred [ _ harrison & aplin _
, 2001 ; _ yu _ , 2002 ] .
a higher @xmath26 would imply that crf variability arising from a variable sfr and passages through spiral arms becomes more prominent .
nevertheless , all the secondary parameters are not particularly important .
although future variability of the mw sfr can not be predicted , we can predict the future evolution of the solar luminosity and the decrease in strength of the solar wind . since the solar wind currently filters out only about 10% of the cosmic rays responsible for tropospheric ionization ( with energies @xmath120 gev ) , the cooling effect of the solar - wind / crf coupling is almost at its fullest .
this implies that from now on , in the coming few billion years , the global temperature is expected only to increase on average .
this is exemplified in the fig .
this is not to say that bursts of star formation could reduce the temperature temporarily " , for example , when the lmc will be swallowed by the mw .
we expect that within about 1.5 gyr , the long term average global temperature will increase to values higher than anytime previously ( i.e. , @xmath121c ) , at which point the negative temperature excursions arising from spiral arm passages and a higher sfr will not cool earth down to current day temperatures , and no more ice - age epochs will ever appear . from about 2 gyr into the future , even spiral arm passages and a higher sfr will not be able to reduce the global temperature to below the warmest temperature experienced in the past 4 gyr .
we combined here the proposition that crf variations affect the global climate with the prediction that the early solar wind should have been stronger than today .
this was shown to result with the following conclusions : 1 .
the stronger wind of the young sun was more effective at blocking the crf reaching earth such that the crf was significantly reduced when compared with the current flux .
when coupled to the suggested crf / temperature link , the lower crf reaching earth should have translated into a higher global temperature .
3 . using the apparent correlation between crf variability and global temperature over the past 550 myr [ _ shaviv & veizer _ , 2003 ] , it is possible to quantify this effect .
it was found that the associated warming was enough to significantly compensate for the _ faint sun _ and explain about @xmath122 to @xmath123 s of the temperature increase required to warm the young earth to above the average temperature today .
the remaining @xmath124 to @xmath125 could come from one of the previously suggested factors , such as ghg warming by _ modest levels _ of co@xmath2 ( @xmath3co@xmath2 of order 0.01 bar ) , which are consistent with the geological constraints .
besides explaining the secular climate trend , the evolution of the solar wind also reproduces , essentially without any free parameters , long term temperature trends on earth once the variable sfr in the mw is considered . in particular , the crf evolution explains why earth has had glaciations in the past eon and between 2 and 3 gyr bp .
although the fine " details of a variable sfr and the passage through spiral arms can not be predicted to more than several 100 myr into the future , it is possible to extrapolate the decrease in solar wind and the increase in solar flux . it is found that from about 1.5 billion years into the future , earth is expected to be warmer on average than anytime in its previous history .
this is because the crf reaching earth is currently about as large as it can be . 1 .
first and foremost , the aforementioned conclusions rest on the assumption that the crf / climate connection is indeed real .
since this link still lacks a robust physical picture to support it ( and is therefore still a subject of debate ) , it clearly deserves extensive study , especially in view of the repercussions that it has .
for example , it bears consequences on the value of the radiative forcing , since only a low value is consistent with the crf / climate link .
2 . even if we were safe to assume that the crf climate connection is real , since the process is not fully understood , it is not trivial to guaranty that it also operated before the past eon , during which periodic ice - age epochs existed . in particular , the crf / climate link could depend on a biogenic agent which may have been absent during the archean or early proterozoic earth . the link could also depend on atmospheric composition which undergone a few notable changes over the eons .
we have neglected in the present analysis additional effects that may be important but hard to quantify .
these include for example effects of atmospheric composition , land mass characteristics ( area and albedo ) , and even galactic migration of the solar system
. , m. a. , e. vassiliadis , p. r. wood , s. j. meatheringham , j. p. harrington , r. c. bohlin , h.c .
ford , t. p. stecher , and s. p. maran , hubble space telescope observations of planetary nebulae in the magellanic clouds . v. mass dependence of dredge - up and the chemical history of the large magellanic cloud , , 188 , 1997 . , s. , s. wilhelm , h. aufmhoff , k. h. wohlfrom , and f. arnold , cosmic ray - induced aerosol formation : first observational evidence from aircraft based ion mass spectrometer measurements in the upper troposphere , , doi:10.1029/2002gl015044 , 2002 . | standard solar models predict a solar luminosity that gradually increased by about 30% over the past 4.5 billion years . under the faint sun ,
earth should have been frozen solid for most of its existence . yet , running water is observed to have been present since very early in earth s history .
this enigma is known as the faint sun paradox .
we show here that it can be partially resolved once we consider the cooling effect that cosmic rays are suspected to have on the global climate and that the younger sun must have had a stronger solar wind , such that it was more effective at stopping cosmic rays from reaching earth .
the paradox can then be completely resolved with the further contribution of modest greenhouse gas warming .
when we add the cosmic ray flux modulation by a variable star formation rate in the milky way , we recover the long term glacial activity on earth . as to the future
, we find that the average global temperature will increase by typically @xmath0k in the coming 2 gyr . |
for a low enough electron density , wigner@xcite predicted that electrons should localize creating an ordered spatial structure , thenceforth named a wigner crystal , that breaks the complete translational symmetry of the homogeneous electron gas ( also see ref . ) .
indeed , the formation of the wigner crystal was observed in two - dimensional ( 2d ) distributions of electrons on the surface of liquid helium.@xcite a phase transition , induced by the magnetic field , from an electron liquid to a crystalline structure has also been reported for a 2d electron plasma at a gaas / algaas heterojunction.@xcite the existence of different phases in quantum dots , where a few electrons are confined into a small space , has become a topical subject in mesoscopic physics ( see , for a recent review , ref . ) .
in fact , the high controllability of quantum dots suggests that these systems could provide an attractive opportunity to achieve the appropriate conditions for localized states .
it is precisely to stress this controllability that the names _ artificial atoms _ and _ quantum dots _ have been coined .
there is a general persuasion that the wigner crystallization in quantum dots , whose localized states are referred to as wigner molecules , should occur at significantly larger densities than in the 2d bulk .
it is based on the argument that in quantum dots potential - energy contributions can easily exceed the kinetic terms and , therefore , electronic motion can be effectively quenched by manipulating the external confinement and/or an applied magnetic field .
as for the homogeneous gas , one would expect that in crystallized states the kinetic energy is solely that of the vibrational zero - point motion of the electrons around their mean positions , much smaller than the interaction ( potential ) energy .
various approaches including ab initio calculations within diffusion and path integral monte carlo methods , hartree - fock and spin - density functional methods _ etc _ have been applied to analyze the onset of the crystallization.@xcite however , a nonambiguous theoretical result that would justify the above conjecture for a zero magnetic field is lacking . the case with an intense magnetic field is better understood since the magnetic field induces an edge reconstruction , beginning with the appearance of localized vortices on the outer region , that ultimately propagates to all the dot for very high @xmath0s.@xcite in the simpler case of a two - electron 2d quantum dot at zero magnetic field , yannouleas and landman@xcite pointed out that the excited - state energies of this system closely follow the rotor sequence when the repulsion - to - confinement ratio , as given by the wigner parameter @xmath1 , is large enough ( @xmath2 ) .
this was shown to be a proof of the crystallization of the two electrons on fixed positions in a reference frame which is rotating .
quite remarkably , the hypothesized _ rotating wigner molecule _ fulfills at the same time the strict symmetry conditions of quantum mechanics circularity in this case and the obvious preference for opposite positions when repulsion is large enough .
this is a major difference from the above mentioned bulk case where a hamiltonian symmetry ( translation ) is broken by the crystallized state . for wigner molecules ,
symmetries are preserved in the laboratory frame and one must consider an intrinsic ( rotating ) frame to see the underlying deformation .
a similar situation is found for particular states of two - electron atoms that have been much investigated in physical chemistry ( we address the reader to the review paper by berry@xcite ) .
for the two - electron quantum dot , however , the crystallization condition from ref . , @xmath3 , looks disappointing since it seems unrealistic to achieve such a value experimentally .
although the exact ground - state wave function of the two - electron artificial atom can be obtained , at least numerically , it may seem paradoxical that one also needs the excited states in order to ascertain the existence of a crystallization .
in fact , this inability to disentangle the system s intrinsic structure from its full wave function in a clear way can be taken as a weakness of the ab initio , symmetry preserving , approaches .
in general , even in cases when the exact ground- and excited - state wave functions and energies are known , an intrinsic deformation can only be inferred by comparing with the result of simpler models in which either symmetries are relaxed or the intrinsic structure is imposed .
a clear example of the former approach is given by the unrestricted hartree - fock ( hf ) method for the ground state@xcite followed by the random - phase approximation ( rpa ) for excitations.@xcite
conversely , the roto - vibrational model of wendler _
et al._@xcite for two electrons in a ring could be included in the latter category .
one should be aware that when symmetries are relaxed , as in the hartree - fock approach , artifacts or unphysical properties may appear .
in a recent contribution reusch and grabert@xcite discussed the validity of the latter , drawing special attention to the caution with which one must take hartree - fock predictions on symmetry breaking , in agreement with the results presented below .
therefore , a complete physical understanding requires both exact results and model solutions . this way
the system s intrinsic deformations are physically understood while , at the same time , artifacts can be safely discarded . a paradigmatic case where the proposed analysis can be performed
is given by the two - electron 2d parabolic quantum dot .
the separation of center - of - mass and relative coordinates along with the circular symmetry restriction allows the reduction of the multidimensional schrdinger equation to just a radial one , easily solvable numerically . on the other hand , the hartree - fock and rpa solutions without any symmetry restriction
can also be obtained . a most convenient basis for this latter calculations is given by the fock - darwin orbitals in terms of which one can analytically develop much of the required algebra .
it is our aim in this work to determine the crystallization onset of two - electron parabolic dots by recourse to the three different approaches referred to above ; namely , ( a ) an analytical roto - vibrational model , ( b ) a numerical solution of the schrdinger equation , and ( c ) symmetry unrestricted hartree - fock and random - phase approximations . hereafter , we refer to the solution of the schrdinger equation for the two - electron parabolic dot as the _ exact _ solution .
it should be pointed out that , as shown by taut,@xcite this schrdinger equation is analytically solvable only for particular confinement / interaction strengths . for general values of this quantity a numerical treatment is required . as mentioned above ,
the most straightforward one is an integration of the radial equation@xcite but , nevertheless , other methods such as diagonalization in a basis@xcite and the monte carlo method @xcite have also been applied .
one of us has used the so - called oscillator representation method , perturbatively treating the residual interaction , to derive analytical expressions for the energy levels.@xcite the paper is organized as follows .
section ii introduces the magneto - parabolic units that allow one to trace the evolution of ground and excited states of artificial atoms at various conditions . an analytical roto - vibrational model for the two - electron parabolic quantum dot
is described in sec .
section iv provides details of our numerical calculation of exact solutions and compares these solutions with those of the roto - vibrational model .
section v analyzes the reliability of the hartree - fock and rpa results for the present system .
a short summary is finally drawn in sec .
we consider two electrons with a coulomb interaction between them .
the electrons move in the @xmath4 plane where a circular parabolic confinement induces the formation of an electron island .
the system is also subject to an external magnetic field applied in the vertical direction ( @xmath5 ) .
the full hamiltonian thus reads @xmath6_i}+ { e^2\over \kappa\ , r_{12}}\nonumber\\ & + & g^ * \mu_b b s_z\ ; .\end{aligned}\ ] ] in eq .
( [ eqh ] ) , @xmath7 , @xmath8 , and @xmath9 are the electron s effective mass , the dielectric constant , and the effective gyromagnetic factor , respectively , and we have used planar polar coordinates ( @xmath10 ) .
the two contributions within the square brackets are , respectively , the generalized kinetic energy in terms of the vector potential @xmath11 , and the external confinement . within the so - called symmetric
gauge one has @xmath12 .
the next contribution is the coulomb repulsion and , finally , the last term is the zeeman energy involving the total spin operator @xmath13 and the universal bohr s magneton @xmath14 .
it is well known that in the chosen gauge the magnetic field contributions can be recast into the form of an effective parabolic confinement of frequency @xmath15 , where @xmath16 is the cyclotron frequency , and an additional angular - momentum - dependent term @xmath17 ( cf
. refs . ) . the magneto - parabolic units ( mpu ) we shall use consist of taking @xmath18 as the energy unit and @xmath19 as the length unit . in addition , one also imposes @xmath20 as angular momentum unit which
, in turn , fixes the time unit @xmath21 . summarizing in the standard abuse of notation we may write @xmath22 .
this is a natural choice for magneto - parabolic confinements and it allows one to express the spatial part of the hamiltonian in terms of only two adimensional parameters , namely , @xmath23 note that @xmath24 and @xmath25 give , respectively , the ratios of coulomb interaction strength and cyclotron frequency to effective confinement . in the absence of a magnetic field @xmath24 coincides with the so - called wigner parameter @xmath1 of ref . .
also note that @xmath25 has a maximal value @xmath26 that corresponds to a zero confinement @xmath27 .
we also mention that reusch and grabert used these parameters in their recent hartree - fock calculations.@xcite within the mpu system the hamiltonian reads @xmath28_i}+ \frac{r_{\it mp}}{r_{12}}\nonumber\\ & + & \frac{g^ * m^*}{2 } w_{\it mp } s_z\ ; , \end{aligned}\ ] ] where @xmath29 is the adimensional ratio of effective to bare electron mass , i.e. , @xmath30 .
the passage from the mpu system , with a given @xmath24 and @xmath25 , to physical units requires the knowledge of the effective mass @xmath7 and the dielectric constant @xmath8 .
more specifically , with fixed @xmath7 and @xmath8 one can invert eqs .
( [ eq2 ] ) and ( [ eq3 ] ) for the effective confinement @xmath31 and cyclotron @xmath32 frequencies or , equivalently , for the external confinement @xmath33 and the magnetic field @xmath0 . in terms of these results
the physical values of the energy ( @xmath18 ) and length ( @xmath34 ) units are readily found .
figure [ fig1 ] shows the equivalence between the adimensional parameters @xmath35 of the mpu hamiltonian and the physical values of @xmath36 for the case of a gaas host semiconductor having @xmath37 and @xmath38 .
the advantage of working with the mpu system becomes obvious when realizing that , with unit redefinition , the same numerical results can be equally applied to a variety of confinements , magnetic fields , and material parameters ( effective mass and dielectric constants ) .
therefore , the model acquires a certain degree of universality .
we also expect that quantum dot properties such as scaling laws or phase diagrams will be better displayed in terms of the adimensional mpu parameters .
taking eq .
[ eqmph ] as a starting point and introducing the standard center of mass @xmath39 and relative @xmath40 coordinates it is well known that the hamiltonian separates and , therefore , that the wave function factorizes .
the center of mass ( cm ) problem is that of a single particle in a harmonic potential and magnetic field , having an analytical solution in terms of fock - darwin orbitals and energies @xmath41 , with @xmath42 the radial and angular momentum center - of - mass quantum numbers .
focusing next on the relative problem , one introduces the wave function @xmath43 having good @xmath44 angular momentum ( @xmath7 ) , and an additional quantum number @xmath45 whose meaning will be clarified below .
the equation for the unknown @xmath46 reads @xmath47 u_{nm } = 0 \ ; , \ ] ] where we have defined @xmath48 in terms of the relative - motion energy @xmath49 and the mpu parameter @xmath25 . equation ( [ eqrel ] ) will be the basis of our roto - vibrational model .
note that it resembles a schrdinger one - dimensional equation with an effective potential @xmath50 that includes the rotational motion term @xmath51 characterized by the angular momentum quantum number @xmath7 .
we can expect a rigid - rotor behaviour if @xmath52 has a deep minimum at a particular value @xmath53 .
when this occurs the situation resembles that of diatomic molecules like h@xmath54 , where the potential well for nuclear motion is described by the morse potential ( see ref . ) . in the present case
the effective potential indeed has a minimum although it is in general rather shallow .
this property is responsible for the coupling between rotation and vibration or , equivalently , for the _ floppiness _ of the rotating molecule mentioned in ref . . the minimum condition on @xmath52 yields the rotor radius from @xmath55 neglecting the third contribution on the left - hand - side , an assumption that will be valid for large enough @xmath56
, one finds the asymptotic law @xmath57 .
now , expanding to second order around @xmath56 we approximate @xmath58 a result that , when substituted into eq .
( [ eqrel ] ) for the round parentheses , leads immediately to the analytical prediction@xcite @xmath59 the result embodied by eq .
( [ eq9 ] ) has a clear physical interpretation .
it contains a rotor - like contribution , @xmath60 , with a moment of inertia given by @xmath61 mpu , and a vibrational one characterized by a quantum number @xmath45 .
the vibrational frequency @xmath62 ( @xmath63 ) is given by the last square - root factor .
similar to atomic molecules , there is roto - vibrational coupling , since the vibration frequency depends on @xmath7 and , in addition , centrifugal distortion since @xmath56 also depends on @xmath7 . for large enough values of @xmath24 , implying large @xmath56 and therefore small average densities , the centrifugal distortion disappears and one has @xmath64 for all @xmath7 s .
in this limit rotational terms become negligible , as well as roto - vibrational ones .
thus , eq . ( [ eq9 ] ) reduces to a simple @xmath7-independent asymptotic expression @xmath65 when adding the magnetic field , the roto - vibrational energy becomes @xmath66 in agreement with the expectations from ref . for two interacting electrons in a quantum ring .
@xcite it is worth stressing that since the @xmath25 dependence only amounts to an energy shift of the eq .
( [ eqrel ] ) eigenvalue , the radial function @xmath46 does not depend on @xmath25 .
therefore , one may conclude that the roto - vibrational properties are not affected by magnetic fields , for a fixed @xmath67 state .
of course , since the energy shift varies for different states , the magnetic field will modify the ordering of energy levels . for instance , the level crossings as a function of @xmath25 will cause the ground state to have a nonvanishing @xmath7-value .
this actually explains the buildup of increasing permanent currents in the dot s ground state .
the results from this section will be validated by comparing with the exact ones below .
the roto - vibrational model presented here allows one to determine the crystallization onset from the criterion that rotation and vibration motions decouple when intrinsic - frame electron localization sets in .
conversely , when the coupling is strong the system could be represented by either a vibrating rotor or a rotating vibrator and , therefore , the situation can not be clearly resolved .
it is also worth stressing that the roto - vibrational model describes all possible excitations of the relative - motion problem .
for this particular system , this amounts to a description of all the excitations since the center - of - mass and spin degrees of freedom can be analytically integrated out .
we have solved the radial equation for the relative problem , eq .
( [ eqrel ] ) , numerically .
several standard approaches to eigenvalue problems with boundary conditions can be used for this purpose .
specifically , we have applied the so - called matching method where one integrates with the numerov or runge - kutta algorithms from the origin @xmath68 outwards assuming an @xmath69 behaviour . additionally , imposing an exponential decay law @xmath70 for large @xmath71 ,
inwards integration is performed and the two solutions are required to match at an intermediate @xmath71-point . to ascertain the numerical result for the eigenvalue , as a control we also used the method of node - counting , where only outwards integration is performed and the eigenvalue is found from the condition that @xmath46 increases by one the number of radial nodes from the required value when the energy exceeds the correct eigenvalue by an infinitesimal amount . in principle , the node - counting method assures the correct boundary condition at @xmath72 automatically .
however , in practice , there always remains a small difference between the numerical and exact eigenvalues , responsible for a deviation from the exponential decay from some ( large ) @xmath71 onwards . in spite of this possible difference in asymptotic behavior ,
the two methods ( matching and nodecounting ) provide to a high accuracy the same eigenvalue . with the above numerical methods an exploration of a part of the @xmath73 plane has been performed .
figure [ fig2 ] summarizes the results for ground - state energy and angular momentum .
note that the ground state always has @xmath74 for the cm quantum numbers and that , because of symmetry , even ( odd ) @xmath7 states are associated with singlet ( triplet ) total spin .
clear singlet - triplet oscillations , as studied in the literature ( cf . ref . ) , are seen in the lower panel of fig .
it is also worth mentioning that the ground - state energy contour lines are piece - wise linear as a result of the simple dependence on the @xmath25 parameter .
indeed , the energy is totally @xmath25-independent for @xmath75 and a fixed @xmath24 .
figure [ fig3 ] displays the radial wave functions @xmath76 for two different values of @xmath24 as well as the corresponding effective potentials @xmath77 defined by eq .
( [ eq7 ] ) .
we note that , in agreement with the discussion of the preceding section , when increasing @xmath24 the effective - potential minimum moves outwards and it effectively binds the lower states to its neighborhood . when this occurs the radial probability is strongly quenched at small @xmath71 ( lower panel ) and the scenario indeed resembles the familiar one from the physics of diatomic molecules .
a more detailed comparison of the roto - vibrational model with the exact results will be done in the next subsection . the validity of the roto - vibrational model presented in sec .
iii is proved by the results of fig .
as shown in the upper panel , the error of the analytical prediction for @xmath78 is important only when @xmath79 is small . as a matter of fact , for @xmath80 the discrepancy for @xmath81 is always below 2% , even with the asymptotic expression eq .
( [ asympt ] ) .
note that the relative errors slightly increase with increasing @xmath45 ( dashed curves ) , although the analytical approximation can still be considered quite good . the analogous comparison for @xmath82 ( not shown ) yields smaller relative errors than those for @xmath75 .
the lower panel of fig .
[ fig4 ] analyzes the excitation energies as measured from @xmath81 , again showing an excellent agreement between the exact results and the ones obtained with the analytical model , with small deviations only at small @xmath24 .
based on the asymptotic expression , eq .
( [ asympt ] ) , we have chosen @xmath83 as an independent variable in order to better display the linear behavior associated with the rigid rotor at large @xmath79 .
we shall rely on the high accuracy of the roto - vibrational model to provide a quantitative measure of the crystallization onset
. our criterion will be the following : the two - electron parabolic system is assumed to be crystallized as a rotating wigner molecule when the roto - vibrational coupling falls below a given percentage ( typically chosen as 2 or 3 % ) .
the roto - vibrational coupling is defined as @xmath84 where @xmath85 is the square - root factor in eq .
( [ eq9 ] ) and @xmath86 is the limit of this quantity for @xmath87 . using a 3% condition the crystallization onset for each angular momentum is given by the @xmath24 value where each @xmath7 curve of fig .
[ fig5 ] enters the shaded region .
note that the crystallization onset moves towards higher values as the angular momentum is increased , reflecting the property that roto - vibrational coupling is stronger for high-@xmath7 states .
it is also worth mentioning that , since not all @xmath7-states are simultaneously crystallized , in practice the _ rotational bands _ will gradually degrade for increasing @xmath7 s as a consequence of the roto - vibrational coupling . in agreement with the bulk gas situation ,
the crystallized states with the proposed criterion are characterized by having a potential energy that largely exceeds the kinetic one , as can be easily checked from eq .
( [ eq9 ] ) .
the crystallization properties of @xmath7 and @xmath88 states are identical since one can easily check that @xmath89 . therefore , taking into account the variations in ground state angular momentum we can draw the boundaries for wigner crystallization in the @xmath73 plane , i.e. , the crystallization phase diagram ( see the lower panel of fig . [ fig2 ] ) .
of course , if instead of a 3% threshold for roto - vibrational decoupling one chooses a different value the crystallization onset will vary , although as shown in fig .
[ fig5 ] , for low @xmath7 s the crystallization is not crucially dependent on the precise percentage in the range 2 to 4% .
actually , it should be more appropriate to speak of crystallization onset for a given percentage of roto - vibrational decoupling than of an absolute value .
it should be noted that the onset of wigner crystallization may be studied by means of the conditional probability distribution ( cpd , cf .
refs . ) for finding one electron at positon @xmath90 given that the second electron is at position @xmath91 .
as soon as the interaction is switched on , the cpd , which for a system with only two electrons is just the modulus squared of the wave - function @xmath92 , exhibits the formation of a molecular - like state ( see fig . [ fig6 ] ) .
it is difficult , however , to use this measure alone as a conclusive evidence of the formation of the wigner molecule in two - electron dots .
the difficulty arises from the fact that even weak interactions , for which we should not expect a crystallization , yield the formation of a hole around the electron at @xmath91 ( the correlation hole ) and a maximum at @xmath93 . indeed , as seen from fig .
[ fig6 ] , the results for @xmath94 hint at a molecular state even though according to our analysis the roto - vibrational coupling is still strong and this state is not associated with a crystallized phase ( see fig . [ fig5 ] ) .
when @xmath79 increases the depletion of the cpd around the fixed electron becomes much stronger , which is in qualitative agreement with the crystallization trend from our analysis .
in this section we discuss the results obtained within the symmetry unrestricted hartree - fock method for the ground state and the corresponding rpa for excitations .
we recall that the hf and rpa approaches were devised for the analysis of many - body systems .
therefore , their application to a two - electron quantum dot is merely an exploratory approach to the qualitative features of the wigner crystallization rather than a quantitative description of the above exact results .
we have solved the hf problem in the fock - darwin basis which diagonalizes the square bracket in eq .
( [ eqh ] ) , namely @xmath95 , where @xmath96 labels the orbital part and @xmath97 the spin .
our basis has been optimized such a way that we consider the 70 lowest fock - darwin states , of the non - interacting energy level scheme , for a chosen value of the magnetic field .
an arbitrary single - particle orbital @xmath98 is then expanded as @xmath99 . in the chosen basis ,
the hf equations are written as a system of nonlinear eigenvalue equations for the matrix of @xmath0 coefficients ( see , for instance , details in ref . ) .
we have imposed good @xmath100 hf orbitals leaving totally unspecified the remaining spatial symmetries .
note that the slater determinant built with these single - particle orbitals will be an eigenstate of the total @xmath13 operator but not , in general , of @xmath101 .
this , as we shall see in the results , is intimately connected with the prediction of broken circular symmetry . in fig .
[ fig7 ] we show the hf phase diagram in the @xmath73 plane ( lower panel ) and the corresponding total energies ( upper panel ) .
the total energies resemble those obtained in the exact treatment , with approximate piecewise linear regions between orbital angular momentum transition lines .
as expected the actual values at a given point in the diagram lie slightly above the corresponding exact results . in the lower panel
different gray regions reflect a measure@xcite of the deviation from circularity of the ground state density , with the lightest intensity corresponding to a circular ( nonbroken symmetry ) solution and more intense gray levels to noncircular ( broken symmetry ) results .
the contour lines show the total orbital angular momentum @xmath102 .
regions i and iii are of circular symmetry and for them @xmath102 has a good quantum number , taking the values 0 and -1 , respectively . for the rest of the diagram ( regions ii , iv and v ) the contour lines only indicate the expectation value of @xmath102 but , since the mean field in not circularly symmetric , this is no longer a good quantum number .
the dotted curve separates the states having total spin projection @xmath103 ( below ) from @xmath104 ( above ) . for @xmath103 configurations ,
the hf method predicts a broken - symmetry solution when @xmath24 exceeds a value @xmath105 , somehow below the onset of crystallization obtained in sec .
iv ( @xmath106 ) . in this region ( ii )
the hf solution is indeed a mixture of singlet and triplet states , as can be verified by computing the total spin dispersion @xmath107 .
the corresponding spatial density is built from two opposite and localized single electron orbitals .
it is instructive to compare the hf solutions in region ii with those obtained using a total - spin conserving ansatz @xmath108 \nonumber\\ & & \times \chi_{\it trp}(\eta_1,\eta_2 ) \label{trp}\end{aligned}\ ] ] for singlet ( @xmath109 ) and triplet ( @xmath110 ) states , where @xmath111 $ ] denotes the antisymmetrized product of the two orbitals @xmath112 and @xmath113 while the @xmath114 s are the well known singlet and triplet spin states . as a sample
result the total energies obtained for @xmath115 are @xmath116 , @xmath117 , @xmath118 and @xmath119 ; i.e. , by requiring total @xmath101 conservation the mean field energy raises considerably . in addition , while the hf solution breaks circular symmetry , both @xmath109 and @xmath110 yield circular densities because of the spatial dependence of the @xmath120 s in eqs .
( [ sng ] ) and ( [ trp ] ) .
we have also checked that these results are equivalent to those obtained using the lipkin - nogami projection method @xcite for the effective hamiltonian @xmath121 in order to restore @xmath101 symmetry approximately .
the above ansatz for states with good total spin are examples of the use of _ constraints _ in mean field approaches , which necessarily raise the energy above the mean field minimum .
one could also project the symmetry - unrestricted hf orbitals as discussed in ref . .
in the latter case , however , the wave function is no longer a single slater determinant but rather a sum of few determinants of the corresponding symmetry operator .
therefore , the ground state energy with the restored symmetry is no longer bound by the mean - field minimum and thus can be closer to the absolute minimum imposed by the variational principle . with the above results ,
we conclude that the lowest hf solution in region ii requires a simultaneous breaking of the spin and space symmetries .
taking into account the results of the preceding sections we can say that singlet - triplet mixing in region ii is an artifact of the hf solution .
however , space symmetry breaking in this region is a true physical mechanism indicating an intrinsic structural change in the exact wave function , as supported by the roto - vibrational model discussed above .
we believe this peculiar combination of artifact and physics is due to the smallness of the configuration space for a two - electron system .
we must also point out that at high @xmath24 and/or @xmath25 ( regions iv and v ) the hf prediction fails to match the results of the analysis given in sec .
in contrast to the exact results of fig . [ fig2 ]
where a region ( although narrow ) with @xmath122 corresponding to a singlet state appears at large @xmath25 , the hf approximation predicts only one @xmath123 transition as @xmath25 increases .
this can be understood as an overestimation of the exchange energy in the hf model which tends to favor spin alignment whenever orbital overlapping occurs , as it does in region iii with circular orbitals , as well as in regions iv and v with two - lobed orbitals .
we consider next the results obtained by solving the rpa for excitations . as discussed in ref .
the rpa determines the moment of inertia associated with the collective rotation of a deformed hf structure ( see eq . ( 18 ) in ref . ) .
figure [ fig8 ] presents the evolution with @xmath24 , at @xmath124 , of the rpa moment of inertia @xmath125 ( circles ) . for comparison , the values computed through the solution of eq .
( [ eq8 ] ) of the roto - vibrational model , @xmath61 , are also shown ( solid lines ) .
each @xmath126-line starts at the crystallization onset for the corresponding angular momentum using the criterion of sec .
note that @xmath125 remains null until the hf solution breaks rotational symmetry at @xmath127 , from there on it reasonably agrees with the exact values , somehow averaging the exact results for different @xmath7 s .
the molecule stretching , yielding larger @xmath56 s ( @xmath126 s ) as @xmath7 increases , is obviously outside the rpa .
all @xmath126 values slowly converge to a common result with increasing @xmath24 , i.e. , to an exact rigid - rotor behavior .
for comparison , the dashed line represents the asymptotic value corresponding to @xmath64 .
the nice qualitative agreement between @xmath125 and @xmath126 is giving additional support to the above conclusion that space symmetry breaking in region ii of the hf phase diagram indicates a genuine physical effect and , thus , it also supports the overall picture of a rotating wigner molecule .
although the rpa restores circularity on the deformed hf mean field,@xcite associated with the @xmath102 operator , it is not able to restore the symmetry related to @xmath101 , since the latter one is a two - body operator which is beyond the rpa treatment of broken symmetries .
a side effect of the hf spin artifact in region ii is that when spin - flip bosonic pairs are included in the rpa , generalizing our previous calculation,@xcite the correlation energy in this symmetry - broken phase is badly overestimated , i.e. , it is between two and three times larger than the exact value .
this does not occur , however , in the circularly symmetric regions .
focusing now on the rpa vibron states , we must also distinguish excitations associated with spin and space ( charge ) degrees of freedom .
while the rpa qualitatively describes all excitations in regions i and iii , it fails for spin excitations in phase ii . obviously , this is due to the hf spin artifact in this region .
it also fails for space excitations of phases iv and v. finally , we end this section by pointing out that , in spite of the spin artifact of phase ii , in pure triplet phases rpa reproduces the exact spin precession frequencies known from the theory of magnetic resonance.@xcite that is , a pure spin - flip state ( precessional mode ) is expected at the larmor energy @xmath128 mpu .
indeed , within the rpa , i.e. , in the quasi - boson approximation,@xcite one finds @xmath129 = \hbar\omega_{\cal
l } o^+_{\cal l}\ ; , \ ] ] with the vibron operator for the larmor mode @xmath130 in eq .
( [ eq15 ] ) @xmath131 is the hf expectation value of the @xmath13 operator .
in fact , the larmor mode at @xmath132 appears whenever the ground state has @xmath133 and it is normally the lowest excitation of the system .
we performed a systematic study of the evolution of ground and excited states of two - electron quantum dots subject to an external magnetic field .
the analysis has been done in terms of magneto - parabolic units and the associated parameters @xmath35 that give , respectively , the ratios of coulomb interaction strength and cyclotron frequency to effective confinement .
the ground state calculations are summarized in a phase diagram that can be equally applied to a variety of confinements , magnetic fields and material parameters .
we suggested an analytical model for the interpretation of the exact results , including roto - vibrational coupling and centrifugal distortion ( molecule stretching ) . within this roto - vibrational model we proposed a criterion to determine the onset of wigner crystallization based on the decoupling of rotational and vibrational motions . for a 3%-decoupling threshold we found that wigner crystallization appears , for zero - angular - momentum states , when @xmath24 exceeds a value @xmath134 .
states with larger @xmath7 s crystallize at higher @xmath79 values . in agreement with the homogeneous gas situation
the potential energy of the crystallized states is much larger than the kinetic energy , the latter one being solely due to the vibrational zero - point motion of the electrons .
the hf calculations predict that crystallization for @xmath103 occurs when @xmath135 , the new phase ( ii ) being in an artificial mixture of singlet and triplet spin states .
the space symmetry breaking in phase ii is a genuine physical effect but the spin mixture is an artifact due to the smallness of the configuration space for a two - electron system .
other hf symmetry - broken phases ( iv and v ) do not agree with the exact results .
the rpa moment of inertia qualitatively agrees with the result from the roto - vibrational model , although the molecule centrifugal distortion is missed . on the other hand ,
the rpa produces reliable results for space ( charge ) excitations in regions i , ii and iii , as well as for spin excitations when the hf solutions possess good @xmath101 and @xmath13 quantum numbers ( regions i and iii ) .
we would expect a broader applicability of the many - body theories ( hf+rpa ) for larger systems .
work along this line is in progress . in conclusion ,
the combined use of exact and model calculations allowed us to ascertain the existence of a rotating wigner molecule in a two - electron dot for relatively large electron densities or , equivalently , small @xmath24 parameters .
l. jacak , p. hawrylak , and a. wojs , _ quantum dots _
( springer , berlin , 1998 ) .
t. chakraborty , _ quantum dots : a survey of the properties of artificial atoms _ ( north - holland , amsterdam , 1999 ) .
b. h. bransden and c. j. joachain , _ physics of atoms and molecules _
( longman scientific & technical , new york , 1994 ) strictly speaking , this expression is not fully analytical since @xmath56 is determined by the algebraic eq .
( [ eq8 ] ) that , in general , we solve numerically .
nevertheless , to simplify the discussion we shall continue to call this result and also , by extension , the complete roto - vibrational model , analytical .
after the submission of the paper , we became aware that in the limit of strong magnetic fields and strong interaction , when @xmath57 , similar results for the energy levels have been obtained in a. matulis and f. m. peeters , solid state commun .
* 117 * , 655 ( 2001 ) .
s. a. mikhailov , phys .
b * 66 * , 153313 ( 2002 ) . we have used two alternative orbital symmetry - breaking measures , @xmath136 and @xmath137 with @xmath138 . h. j. lipkin , ann .
, * 12 * , 425 ( 1960 ) . | we provide a quantitative determination of the crystallization onset for two electrons in a parabolic two - dimensional confinement .
this system is shown to be well described by a roto - vibrational model , wigner crystallization occurring when the rotational motion gets decoupled from the vibrational one .
the wigner molecule thus formed is characterized by its moment of inertia and by the corresponding sequence of rotational excited states .
the role of a vertical magnetic field is also considered .
additional support to the analysis is given by the hartree - fock phase diagram for the ground state and by the random - phase approximation for the moment of inertia and vibron excitations . |
the magellanic stream ( ms ) , a 10 wide tail of neutral hydrogen ( hi ) emanating from the magellanic clouds and trailing for almost 100 on the sky ( dec @xmath3 to @xmath4 ) @xcite , is the only clear example of a gaseous halo stream in the milky way s close proximity . while it is well accepted that the ms is the result of interactions between the milky way ( mw ) and the magellanic clouds , the relative importance of tidal stripping and various kinds of gasdynamical interactions is still very much under debate .
most recently several new attempts were made to model the observed hi column density and velocity distribution as being due to purely tidal stripping @xcite , or gravitational @xmath5 hydrodynamical interactions @xcite .
these models focus on reproducing general features in the magellanic system , and gradients in hi column density and velocity along the ms . to add more excitement to this topic , recent estimates of the proper motion of the small magellanic cloud ( smc ) by @xcite and the most recent calculations of the magellanic cloud s orbits by @xcite suggest , contrary to all previous studies , that the clouds are only on their first passage around the mw .
the new orbits imply that neither tidal nor ram pressure stripping would have had enough time to produce the ms , calling for alternative formation mechanisms .
the distance to the ms , especially to its tip , or the region the farthest away from the magellanic clouds , is another contentious question and varies greatly between models . under the ram pressure hypothesis
the tip has fallen the farthest toward the mw and is at a distance of only 25 kpc .
early tidal models @xcite place the tip at a distance of 60 - 70 kpc .
the latest tidal simulations @xcite find an even more distant component extending from 170 to 200 kpc .
the latest orbit calculations @xcite would also imply a significantly large distance to the ms , @xmath6 kpc , although it is not clear where exactly the ms is relative to the clouds in this framework . for many years
the ms was viewed as a complex of six discrete concentrations ( labeled as ms i to vi ) .
new hi parkes surveys by @xcite and @xcite , with an angular resolution of 15.5@xmath0 , revealed a more complex nature of the ms gas , with a fascinating network of filaments and clumps .
two large spatial filaments were found to run in parallel over most of the ms length . around dec @xmath7
the dual filaments disperse into many small clumps and filaments culminating in a chaotic appearance at the tip .
the only high resolution view of two selected regions at the tip of the ms was by stanimirovic et al .
( 2002 ) who used the arecibo telescope to image two small regions in ms v ( dec @xmath8 ) and ms vi ( dec @xmath9 ) .
this work showed that the ms clumps have a complex morphology at 3.5@xmath0 resolution , strongly suggestive of interactions between the ms and an external medium . while most previous studies thought that the ms dissipates at
its tip ( dec @xmath7 ) , very sensitive westerbork observations by @xcite suggested that the ms remarkably extends further to the north all the way to dec @xmath10 .
another interesting phenomenon brought to light by the parkes surveys is the presence of numerous small hi clumps which surround the main ms filaments in position and velocity @xcite .
while the origin of these clumps is still unclear , several possible mechanisms have been invoked : the clumpiness in the original gas drawn out of the magellanic clouds , the instabilities along the ms s edge , or dense condensations within an extended mainly ionized ms component .
one of the crucial issues about the origin and structure of the ms in general , is to what extent interactions with the mw halo determine or influence the ms gas .
this problem becomes particularly important at the extreme northern end of the ms , because this portion of the ms is considered to be the oldest and has been immersed in the hot mw halo for a long time . in this paper we present results from the recent hi observations of the tip of the ms obtained with the arecibo telescope as part of the on - going survey by consortium for galactic studies with the arecibo l - band feed array ( galfa ) . in section 2
we briefly outline our observing and data processing methods .
section 3 describes several new filaments discovered at the tip of the ms , while in section 4 we present a statistical summary of properties of numerous hi clouds found in this region . in section 5
we discuss the origin of the ms and its clumpy structure based on our observations and results .
the observations were conducted with the arecibo telescope .
galfa hi survey consists of many individual projects .
the data presented in this paper represent a combination of three galfa projects : togs ( or ` turn - on galfa spectrometer ' being undertaken in parallel with the alfalfa extragalactic survey , pi : putman ) , a2172 ( pis : heiles & peek , whose original target was a shell - like structure at galactic velocities ) , and a2032 ( pi : stanimirovic , which observed a small region at the ms tip ) .
as general galfa observing and data reduction strategies are summarized in @xcite , we emphasize here only a few important points .
all observations were obtained with the dedicated spectrometer galspect which has a fixed velocity resolution of 0.18 . for a2032 and a2172 ,
the telescope was driven in the basket - weave mode producing inter - woven zig - zag scans in the ra
dec coordinate frame . for togs , numerous drift scans along right ascension were made , as this is the preferred observing strategy for the alfalfa survey @xcite . before each observing scan about 5 minutes
were spent running `` the least - squares frequency switching '' calibration procedure @xcite which allows us to derive a reference spectrum to be applied to all spectra within the given scan .
we have then combined scans from togs ( the ` fall ' session ) and a2172 to obtain the data cube that covers a region @xmath11@xmath12 in size .
various zig - zag and drift scans cross in many places .
the crossing points were used to compare spectra and fine - tune our calibration .
this procedure removes small gain offsets between adjucent drift scans and produces a resultant image mostly free of stripping artifacts . after this fine - tuning , spectra from the two projects
were gridded into a data cube . in the case of a2032 observations
our strategy was different .
after basic calibration , we have produced an a2032 data cube , and then added it linearly ( in the image domain ) to the combined togs@xmath5a2172 cube .
both combination processes worked very well .
the images presented in this paper are the first demonstration of how successfully we can piece together data from different galfa projects .
the only problem encountered during the data reduction is radio interference present is several togs scans at high negative lsr velocities .
this interference affects all alfa beams , varies across a given scan , and is centered around 1422 mhz .
it is caused by a still unidentified source .
we have edited affected scans where possible , however some artifacts are still noticeable in several images .
our future software developments will be better able to deal with this , and similar problems . as the integration time was different for the three observing projects ,
our sensitivity varies across the image .
alfalfa and togs cover the sky twice with about 15 sec of integration time per beam , while a2172 and a2032 observations were obtained in a fast - scanning mode with an integration time of only about 5 sec per beam .
declination bands from 11 to 16 and 24 to 28 have the lowest rms noise level , 0.03 k per 1.4 wide channels .
the rest of the image is noisier , with about 0.06 - 0.07 k per 1.4 wide channels .
our best 3-@xmath13 sensitivity is @xmath14 @xmath15 over 20 .
figure [ f : chans ] shows hi images from the final data cube at several selected velocities . in figure
[ f : moments ] we show the first moment image , or the intensity - weighted velocity along the line - of - sight .
the area covered in our observations is huge about 870 square degrees ( @xmath16@xmath17 105 , @xmath18@xmath19 ) , focusing on the very tip of the ms where the two traditional ms filaments were found to be significantly spatially bifurcated ( one is located at ra 23@xmath20 30@xmath21 , and the other one at ra @xmath2223@xmath20 00@xmath21 ) .
the only previous high resolution observations of about 10 square degrees within this region were obtained by stanimirovic et al .
( 2002 ) . as figures [ f : chans ] and
[ f : moments ] show , several coherent hi filaments extend all the way to dec @xmath23 .
@xcite reported an extension of the two ms filaments , in the form of diffuse gas , from dec = 20 to 40 .
we confirm the continuous extension of the ms from dec = 5 to @xmath2225 . beyond this point
our observations may not be sensitive enough to detect a very low - column - density emission found by @xcite .
our sensitivity of @xmath24 @xmath15 ( over 20 ) is significantly worse than @xcite s lowest contour ( in their figure 5 ) shown at @xmath25 @xmath15 .
however , the picture of the two interwoven filaments changes dramatically in our observations : instead of two traditional filaments we find several !
while filaments often break into numerous branches and small clouds when viewed with arecibo s resolution , there are at least four prominent and coherent large - scale structures noticeable across the whole velocity range . to facilitate further discussion
we define and label filaments in the following way . starting from the lowest negative velocities ( @xmath26 ) in figure [ f : chans ] : 1 . `
s1 ' is centered at ra 23@xmath20 35@xmath21 , this is a portion of one of the two traditional ms filaments ; 2 . `
s2 ' is centered at ra 23@xmath20 10@xmath21 and can be traced through the whole velocity range ; 3 . `
s3 ' is centered at ra 23@xmath20 05@xmath21 and starts at a higher negative velocity of @xmath27 ; 4 . `
s4 ' starts at ra 23@xmath20 00@xmath21 and extends to ra 22@xmath20 35@xmath21 .
figure [ f : intensity_ann ] ( left ) shows a schematic representation of the four streams .
both s1 and s4 can be seen from the lowest negative velocities up to about @xmath28 .
s2 , s3 and s4 seem to have spatially a common point of origin , and are most likely related to the second traditional ms filament .
the morphology of the four streams is significantly different : s2 , s3 and s4 consist of and are surrounded by numerous small , compact clumps , having a beads - on - string appearance .
the number of compact clumps peaks at velocities @xmath29 to @xmath28 .
s1 on the other hand has a more diffuse morphology and contains the smallest number of compact clumps .
this may indicate that s1 has a different origin relative to the other three streams , or that it is younger and therefore did not have enough time to fragment .
we get back to this issue in section [ s : discussion ] . while generally clumpy morphology is prominent , the four defined streams are coherent structures , which can be seen most clearly in the first moment image , figure [ f : moments ] .
all four streams show clear velocity gradients , although not all gradients are similarly steep . in figure
[ f : cuts ] we show position
velocity images along the streams .
s1 and s3 have a velocity gradient of about 5 deg@xmath30 over a length of 15 degrees , in agreement with the overall velocity gradient along most of the ms @xcite
. s2 has the steepest gradient of almost 10 deg@xmath30 along the first 10 degrees , and then it almost turns over in velocity roughly staying around @xmath31 to @xmath32 .
s4 has almost a constant velocity of @xmath33 .
all streams have a similar velocity dispersion , @xmath34 - 20 , see figure [ f : intensity_ann ] ( right ) .
this is in agreement with @xcite who measured a dispersion of 15 along most of the ms .
a striking result from the above figures is the presence of numerous small hi clouds .
the smallest clouds have a size down to our resolution limit of 3.5@xmath0 and are unresolved .
most of the clouds are compact and round , head - tail morphology is surprisingly uncommon .
previously , @xcite noticed two types of clouds which surround the ms along its southern portion : head - tail clouds with tails pointing along the major axis of the ms ( typically , the head has a column density of @xmath35 @xmath15 while the tail is 3 - 4 times fainter ) , and ubiquitous dense hi clouds which follow the main ms filaments in position and velocity , some again with a head - tail morphology . we have compiled a catalog of 180 hi clouds .
the positions of all clouds are shown in figure [ f : all_clouds ] .
clouds were selected by eye from the spectral line data cube .
the integrated intensity image was used as a guide for cloud selection , local peaks in this image were considered as potential cloud candidates .
we then used the full data cube to follow potential clouds spatially and in velocity , and measure cloud properties .
special care was taken to avoid breaking larger clouds into separate entities , however the catalog may still be slightly biased toward smaller clouds .
we have made several checks to ensure cloud reliability and to exclude clouds potentially affected by scanning artifacts or rfi .
there were several cases ( mainly at low dec ) where multiple clouds are present along the line - of - sight at clearly different velocities .
these cases were treated as individual clouds . for each cloud
we have measured the cloud size , the central velocity , and the peak hi column density .
in addition , at the center of each cloud we have fitted the hi velocity profile with one or two gaussian functions .
although the cloud catalog is not 100% complete , as it is hard to select clouds in the regions with higher noise , it certainly represents well the cloud population whose physical properties we want to study .
basic cloud properties are presented and discussed in the following sub - sections . to ensure that we are dealing only with clouds of magellanic origin ,
our cloud catalog was cross - correlated with the catalog of hvcs by @xcite , the catalog of hvcs and chvcs by @xcite , and the catalog of recently discovered mini - hvcs by @xcite .
hvcs come with a wide range of sizes and properties . while some represent large complexes , like the ms , most are smaller with sizes of a few tens of degrees .
on the other hand , chvcs represent a class of hvcs which are compact ( with sharp edges in the hi column density distribution ) , apparently isolated from larger complexes , and with angular size less that 2 .
one interpretation is that chvcs are essentially the same as hvcs but located at much greater distances .
recently discovered mini - hvcs are even smaller and have diameters of 9@xmath0 to 35@xmath0 .
these clouds are also more diffuse than typical hvcs , having the peak hi column density of only @xmath24 @xmath15 .
several of mini - hvcs were found to be superimposed on the northern extension ( dec @xmath36 ) of the ms but at velocities distinct from those of the ms gas .
one suggestion is that these clouds represent outliers of the ms complex @xcite .
there are 8 hvcs from @xcite in our field of view .
three hvcs are clearly detected and identified as clouds in our catalog , these are wvw413 , wvw491 and wvw485 .
additional three hvcs ( wvw508 , wvw492 and wvw472 ) have clouds nearby but at an angular separation larger than the beam of the leiden - dwingeloo survey .
there are 19 hvcs and 3 chvcs from @xcite in our field of view .
all of these chvcs were detected .
as chvcs were classified as isolated clouds it will be surprising to find a large population of chvcs in our cloud catalog . only 3 chvcs from @xcite are in our field of view .
chvc229 ( from de heij et al .
2002 ) is extended and was found to be associated with several of our clouds at very low negative velocities ( @xmath37 ) .
we have excluded those hi clouds from our further analyses as at @xmath37 these clouds are clearly spatially located far away from the ms emission .
three mini - hvcs are in our field of view : lmpu 387 - 230 , lmpu 387 - 275 , and lmpu 387 - 297 .
lmpu 387 - 275 and lmpu 387 - 297 are at velocities ( @xmath38 and @xmath39 , respectively ) where we detect several small clumps , the closest one being about 24@xmath0 away .
we did not detect lmpu 387 - 230 however , although @xcite estimated its hi column density to be @xmath40 @xmath15 , clearly above our detection limit . in summary
, we do not find that our catalog is contaminated by a significant population of chvcs , it therefore contains primarily clouds associated with the ms .
[ table1 ] lcccc distance & radius & hi mass & hi volume density & pressure + ( kpc ) & ( pc ) & ( m@xmath41 ) & ( @xmath42 ) & ( k @xmath42 ) + 20 & 30 & @xmath43 & 0.4 & 40 + 60 & 90 & @xmath44 & 0.1 & 10 + 150 & 220 & @xmath45 & 0.05 & 5 + figure [ f : cloud_histos ] shows histograms of cloud basic properties : angular size , hi column density , and central ( lsr ) velocity .
the observed cloud angular size distribution strongly peaks around 10@xmath0 .
90% of clouds have a size in the range 3@xmath0 to 35@xmath0 .
the abundance of very small , compact - looking clouds is significant , more than half of clouds in the catalog have an angular size less than 10@xmath0 .
obviously , these clouds would have been missed in previous large - scale surveys such as the leiden - argentine - bonn ( lab ) survey @xcite , or even parkes surveys of the ms .
the low - size cut - off of the histogram is clearly affected by our resolution limit of 3.5@xmath0 , and higher resolution observations will probably find many smaller clouds .
the cloud central hi column density distribution peaks around @xmath46 @xmath15 , 90% range is @xmath47 to @xmath48 @xmath15 .
this distribution function is affected by our sensitivity limit , deeper observations will probably find many clouds with lower hi column density . for comparison , the hi column density distribution of mini - hvcs peaks at @xmath24 @xmath15 .
figure [ f : dec_vel_more ] ( left ) shows that the number of clouds decreases steeply along the ms , in the direction toward the tip .
the number of clouds in each latitude bin has been divided by the observed angular area to account for different areal coverage across the image .
figure [ f : dec_vel_more ] ( middle ) shows a tendency for clouds with angular size larger than 20@xmath0 to be found closer to the magellanic clouds ( at @xmath49 ) , while clouds with angular size less than 20@xmath0 appear along the whole latitude range .
this could be interpreted as evidence that part of the ms with @xmath49 is at a larger distance , in agreement with tidal models .
please note that the gap around @xmath50 on this plot is due to the lack of clouds , not due to the lack of data in this region . to provide an example of cloud linear size , hi mass , volume density , and thermal pressure
, we use an angular diameter of 10@xmath0 and a median peak hi column density of @xmath46 @xmath15 , for three representative distance values , 20 , 60 , and 150 kpc , see table [ table1 ] .
interestingly , the cloud central velocity , shown in figure [ f : cloud_histos ] ( right ) , shows a clear dichotomy .
there are two peaks centered at @xmath51 and @xmath52 , with the velocity fwhm of 10 and 45 , respectively .
as shown in figure [ f : dec_vel_more ] ( right ) , the velocity dichotomy manifests in the way that there are two main groups of clouds , one around @xmath53 to @xmath54 ( or dec = 10 - 15 ) and @xmath55 , and the other one around @xmath56 to @xmath57 ( dec = 18 - 25 ) and @xmath58 .
this figure also shows an almost continuous linear decrease in velocity ( from @xmath59 to @xmath60 ) with galactic latitude .
this is in agreement with the general velocity trend along the ms .
the observed velocity dichotomy may indicate a kinematic bifurcation of the ms tip .
such bifurcation would not be surprising .
n - body simulations by @xcite found two kinematically separate components , one that follows closely the orbit of the lmc and has a lower negative velocity , and the other one with a more negative velocity . while separated kinematically , both components occupy the same @xmath61 range . about 20% of the clouds in our catalog have velocity profiles whose fitting requires two gaussian functions .
this is indicative of the existence of a multi - phase medium .
however , some of these velocity profiles could be due to the existence of multiple clouds along the line of sight , whose velocity profiles partially overlap . to account for this effect ,
we accept as multi - phase clouds only those which have a subsonicly moving core and a warmer envelope .
this criterion translates into the requirement that the absolute difference between the centers of the two fitted gaussian functions does not exceed @xmath62 the fwhm of the broad component .
our final sample of multi - phase clouds contains 21 entries . as an example , in figure [ f : some_clouds ] we show velocity profiles of ten multi - phase ( the first two rows ) , and five clouds with a single gaussian profile ( the last row ) . in most multi - phase cases ,
a narrow gaussian component fits the line center , while a broader gaussian function is needed to fit the line wings .
narrow line profiles were scarcely found in the ms in previous studies .
@xcite found one cloud with a velocity fwhm of only 4 , at @xmath63 , @xmath64 .
@xcite used the lab survey and decomposed all hi velocity profiles into gaussian components .
they found that the portion of the ms close to the magellanic clouds ( and with positive velocities ) has 27% of the sight lines with a narrow velocity component , while the portion further away from the clouds ( with negative velocities ) has a significantly lower fraction , only about 10% .
the lower number of narrow components was attributed to either a lower halo pressure and/or a larger distance of the northern ms extension .
our final fraction of multi - phase clouds ( 12% ) is similar to kalberla & haud s estimate for the negative velocity portion of the ms .
figure [ f : cloud_histos1 ] ( left ) shows the fwhm vs central velocity of all fitted gaussian functions : crosses correspond to clouds with a single gaussian component , while diamonds and circles represent the secondary ( or narrow ) and the primary ( or broad ) velocity component for clouds with multi - phase medium .
the fwhm of single - component clouds is predominantly in the range of 2030 . for multi - phase clouds ,
the fwhm of the narrow component has a broad , almost continuous , range from 3 to 20 , with a median value of 13 .
this would imply an upper limit on the kinetic temperature in the range 2009000 k , mostly being in the thermally unstable regime .
for comparison , @xcite find the fwhm of the narrow component in the range 528 .
the fwhm of the broad component of multi - phase clouds is around 25 , in agreement with what is found for clouds with only a single gaussian component .
the right panel in figure [ f : cloud_histos1 ] shows the central velocities for both types of clouds .
the velocity dichotomy is still noticeable with two peaks around @xmath65 and @xmath29 .
this shows that the velocity bifurcation affects both clouds with and without multi - phase structure .
we return to this point in section [ s : distances ] .
there are several important observational points presented in the previous sections : * there are four coherent streams at the tip of the ms ; * s1 is diffuse , while s2 , s3 and s4 are clumpy ; * spatially , s2 , s3 and s4 appear to originate from a similar location ; * the velocity gradient along streams gradually decreases from s2 to s4 ; * there is a lack of large clouds at low galactic latitudes ; * central velocities of clouds with and without multi - phase structure show velocity bifurcation at @xmath29 and @xmath65 . traditionally , it was assumed that the ms originated purely from the smc gas .
@xcite suggested that the two double - helix - like ms filaments could arise from gas stripped from the magellanic bridge and the smc . by using the lab survey of galactic hi ( angular resolution of 0.6 ) ,
@xcite were able to trace one of the two filaments all the way to an over - density region in the lmc .
the only numerical study that addressed spatial bifurcation of the ms is by @xcite .
these authors showed that gross features of the ms and the leading arm , as well as the spatial bifurcation of the main ms filament , can be reproduced purely by tidal interactions between the mw and the magellanic clouds .
they also suggested that the main ms filament is broken into two kinematic components .
several distinct features are found in their simulation . 1 .
the main ms filament was created 1.5 gyr ago in a major encounter between the smc , the lmc , and the mw .
further tidal ` kicks ' from encounters with the lmc 1.05 and 0.55 gyrs ago resulted in its spatial , and then kinematic bifurcation .
while one of the bifurcated components follows the orbit of the magellanic clouds and has a lower negative velocity ( @xmath66 ) , the other possesses a higher negative velocity ( @xmath67 to @xmath32 ) .
in addition to the main stream , which extends in distance from 60 to @xmath68 kpc , there is the third stream right at the tip which spatially coincides with the main stream but has distinct kinematics and is significantly further away ( distance of 170 to 220 kpc ) .
this component was created in an encounter between the smc and the lmc 2.2 gyrs ago , and is the oldest and the most distant feature in the magellanic system .
two tidal tails were drawn @xmath69 myr ago from the magellanic bridge and follow the main filament along most of its length . by drawing a parallel with this simulation ,
streams s2 , s3 and s4 could represent a 3-way splitting of the main ms filament . as we show in the next section , this gas has had enough time to cool and fragment and is therefore very clumpy .
as suggested by the simulation , most of the spatial 3-way splitting could have occurred in the more recent past , 1 gyr to 500 myr ago , during close encounters between the ms and the lmc .
the fact that s2 to s4 show a decreasing steepness in their velocity gradients may be suggesting that these filaments were separated in subsequent encounters .
s1 on the other hand , may represent one of the more recently formed tidal tails , drawn out of the magellanic bridge @xmath69 myrs ago .
the fact that it is significantly less clumpy than the other three streams suggests that s1 may be significantly younger , and have not had enough time to cool and fragment .
the lack of large hi clumps at the higher galactic latitude suggests that the far tip of the ms may be more distant .
however , the fact that we find clumps with multi - phase structure almost along the whole latitude range implies that the far tip is probably not too distant .
as discussed in the next section , it would be hard to justify the existence of multi - phase medium at a distance of @xmath70 kpc .
another disagreement with the simulation concerns the observed lsr velocity at the far tip , we find @xmath71 , while the simulation predicts velocities down to @xmath67 .
however , the observed large - scale filaments at the tip of the ms , as well as the evidence for kinematic bifurcation , give strong support to the tidal formation scenario for the ms .
while the observations are only qualitatively in agreement with the simulations , and many details still have to be worked out , it is extremely encouraging to see that detailed spatial and velocity structure of the ms can be reproduced _
purely _ by tidal interactions .
better observed coverage of this region , as well as high resolution observations of the southern ms extension , will certainly help to constrain this and other possibilities in the future .
at certain velocities the ms has a very clumpy morphology .
one of the obvious questions is : what physical processes are responsible for such a clumpy structure ? in trying to understand the cloud origin we look briefly into three important physical processes that would affect a tidal stream of warm gas moving through a hot ambient medium .
all three processes have their characteristic time and spatial scales .
thermal instability arises due to gas cooling and results in spatial fragments of the wnm , whose size is defined by the cooling length @xmath72 .
this process operates on the cooling timescale , defined by @xmath73 , with @xmath74 being the cooling function , @xmath75 being the boltzmann constant , and @xmath76 and @xmath77 being the temperature and volume density of the wnm gas .
if we consider a wnm gas that originated in the smc , then @xmath78 @xmath42 ( based on the size and total mass of the smc , also measurements in @xcite ) , then for @xmath79 k we estimate @xmath80 pc and the cooling timescale of @xmath81 myr . for a lower value of @xmath77 , which is not unreasonable for the gas drawn out of the smc outskirts , both @xmath72 and @xmath82 will increase . considering that most models suggest the ms age of 1 - 1.5 gyr , thermal instability , with a typical timescale of a few tens to a few hundreds of myrs ,
most likely have had a significant effect on the evolution of the ms gas .
recent numerical simulations by @xcite provide an example of thermal condensation of warm gas .
these simulations focus on a collision of incoming turbulent wnm flows and thermal condensation of wnm into colder gas .
while not exactly applicable in our case , it is still very instructive to notice the properties of typical thermal fragments .
a collision of incoming wnm streams creates a thermally unstable region , which partially condenses into cold clouds .
the thermally unstable warm gas ( with temperature of 200 5000 k ) has a filamentary morphology and its fragmentation into cold clouds ( with temperature @xmath69 k ) is heavily controlled by turbulence .
cold clouds formed in simulations are thermally stable and long - lived , in the case of stronger turbulence they have round morphology , while a weaker turbulence is responsible for more elongated morphology of cold clouds .
the simulations show that the fraction of cold gas ( with temperature @xmath69 k ) produced in this process ranges from 10% in a strong turbulent case , to about 30% in a weak turbulent case .
it is interesting to note that the lower bound on this cnm - to - wnm fraction agrees with the fraction of hi clouds with cool cores we find in our observations ( 12% ) .
kelvin - helmholtz ( kh ) instability occurs at the interface of the moving warm stream of gas and the ambient medium and also results in the formation of small fragments with a power - law size distribution .
the timescale on which this instability develops , @xmath83 ( given in gyr ) , is given by : @xmath84 @xcite , where @xmath85 would be the initial size ( radius ) of the ms gas pulled out of the magellanic clouds , @xmath86 corresponds to the density contrast of the warm kh fragments and the ambient hot medium , and @xmath87 is the stream velocity .
we use @xmath88 @xcite , assuming that the velocity of the original ms gas is equal to its present day velocity .
we also assume that @xmath89 , based on the size of the msvi portion of the ms , and that this material was initially at a distance of 60 kpc .
if @xmath90 @xmath42 , and @xmath91 @xmath42 , then @xmath92 myr . as
this timescale is almost comparable to the age of the ms ( in the framework of tidal and ram - pressure models ) , we can conclude that the kh instability has had a negligible effect on the evolution of the ms gas .
once the warm stream has fragmented due to thermal / kh instability into smaller clouds , the third process becomes important this is conductive heat transfer between the warm stream fragments and the hot ambient medium .
@xcite show that the size of the smallest fragments is set by the cloud ability to radiate away the energy flux due to conduction .
the warm clouds reside within warmer phases , hot halo gas with @xmath93 k and warm ionized gas with @xmath94 k. detections of ovi absorption from gas associated with the ms @xcite give strong support for the existence of an ionized component around the ms with @xmath95 k. it is expected that different phases are separated by an interface region through which heat flows between the two phases .
the interface is a surface of transition where , for small clouds , the wnm evaporates into the warmer medium . the classical evaporation theory
@xcite predicts a critical radius for clouds at which radiative losses balance the conductive heat input , and at which the cloud neither evaporates nor condenses . clouds smaller than the critical radius evaporate , while clouds larger than the critical radius accrete by condensation of the surrounding gas . in the case of an interface between the wnm and the him , the cloud critical radius ( @xmath96 ) times
the density of the surrounding medium , @xmath97 , is @xmath98 @xmath15 for @xmath99 k , or @xmath100 @xmath15 for @xmath101 k ( from fig . 2 in mckee & cowie 1977 ) .
generally , the expected critical radius decreases with increasing @xmath97 .
however , for @xmath102 @xmath42 , the critical radius is large , @xmath103 pc .
clouds smaller than @xmath70 pc would then evaporate , while wnm clouds larger than @xmath70 pc would accrete the him by condensation .
thermal fragments we considered in section 5.3 with a size @xmath104 pc will be predominantly smaller than the critical radius and therefore undergoing evaporation .
these small ( @xmath70 pc ) , evaporating clouds are surprisingly long - lived though . for mainly neutral clouds surrounded by the hot gas , the classical mass - loss rate is @xmath105 gr yr@xmath30 for @xmath99 k , or @xmath106 gr yr@xmath30 for @xmath101 k @xcite . in the case of saturated evaporation
the mass - loss rates are even lower . for @xmath78
@xmath42 the evaporation timescale is @xmath107 gyr for @xmath99 k , or @xmath108 gyr for @xmath101 k. but the timescales are in general long , approaching or being longer than 1 gyr .
we hence conclude that small fragments would be able to survive for a long time , at least a gyr , most likely throughout the whole lifetime of the ms .
as thermal instability must have played a very important role in the evolution of the oldest ms gas , we suggest that this could be the dominant process responsible for the clumpy structure found in our observations . the importance of thermal instability is further supported by the large abundance of small hi clumps ( see section 4.2.1 ) . the kh instability is important as well , however it has a significantly longer timescale .
as we have shown , it is reasonable to expect that thermal and/or kh fragments can survive for a long time , especially if surrounded by the ambient gas with @xmath95 k. it is interesting that cloud morphology in our observations is dominated by round shapes .
elongated or head - tail clouds are surprisingly absent in our hi images .
head - tail morphology is commonly observed for chvcs and is often seen as a signature of ram - pressure stripping , or other types of interactions between a cloud and the ambient halo medium .
we therefore conclude that ram pressure effects probably have a secondary role , with gravitational forces being the dominant gas removal process .
if we assume that thermal fragmentation is the dominant process responsible for the clumpy structure , we can estimate a distance to the ms .
if we assume that the median value of the cloud angular size distribution ( 10@xmath0 ) corresponds to the linear size of ` typical ' thermal fragments , @xmath109 pc ( derived assuming @xmath78 @xmath42 ) , we estimate a distance of @xmath2 kpc
. please note that these numbers are only representative .
the cloud angular size varies and is also affected by our resolution limit , the volume density of the warm ms gas is also not a constant number .
however , this exercise is very instructive and demonstrates that thermal fragments can reasonably well explain at least some small - scale structure in the ms .
the ms distance of @xmath2 kpc is in agreement with the predictions from tidal models .
even more impressively , this distance agrees well with the recent estimate of 75 kpc by @xcite , who used a geometrodynamical model and assumed that the ms follows a planar orbit around the galactic center .
also , the existence of multi - phase hi clouds can be used to place an upper limit on the ms distance .
it is important to stress that the existence of multi - phase clouds along the tip of the ms , and especially at a distance of @xmath110 kpc , is extremely surprising from a theoretical point of view .
based on the consideration of cooling and heating processes in an isothermal 10@xmath111 k halo , @xcite predicted that multi - phase clouds , pressure confined by the hot halo , should exist in the ms only at distances less than 20 kpc from the galactic plane .
more recently , considering chvcs @xcite showed that dark matter can provide an additional confinement mechanism , allowing the existence of multi - phase structure at distances @xmath112 kpc .
the multi - phase clouds at the tip of the ms therefore suggest that the tip is likely to be at a distance @xmath113 kpc .
this argues against the existence of the third , very distant stream found in the simulations by @xcite , at least within the area covered by our observations .
also , as pointed out by @xcite , the multi - phase structure at the ms distance suggests a lower confining halo gas pressure than what is expected in the theoretical framework .
we have obtained hi observations of a region , @xmath114 square degrees large , at the tip of the ms .
the major results are as follows . 1 .
we find four coherent hi streams at the tip of the ms , extending continuously from dec 5 to @xmath115 .
the streams have different morphology and velocity gradients .
three streams are very clumpy , while one stream is mainly made up of large diffuse structures .
the comparison of observations with the tidal simulations by @xcite suggests that the three clumpy streams could be a result of a 3-way tidal splitting of the main ms filament , induced by encounters between the lmc and the ms .
the fourth stream is probably much younger , and could have been tidally drawn from the magellanic bridge in the more recent past , @xmath116 myr .
this stream is more diffuse because it has not had enough time to cool and fragment .
we find an extensive population of hi clouds at the tip of the ms .
the smallest clouds have a size reaching down to our resolution limit of 3.5@xmath0 .
70% of clouds have an angular size in the range 3.5@xmath010@xmath0 .
about 12% of the ms clouds show multi - phase velocity profiles , indicating the co - existence of cooler and warmer gas .
cloud central velocities show two peaks at @xmath29 and @xmath32 .
we show that for warm gas trailing through the galactic halo for the past gyr or so , the effects of kh instability are relatively unimportant .
however , thermal instability , with a typical timescale of a few myrs to a few hundred myrs , is significantly important and will result in the fragmentation of gas and the formation of smaller clumps .
the observed clumpiness at the tip of the ms could be fully or partially due to this mechanism .
we show that thermal / kh fragments can survive in the hot halo for a long time ( @xmath117 gyr ) , especially if surrounded by a @xmath1 k halo gas .
if the observed angular size of hi clumps is due to thermal instability , then the tip of the ms is at a distance of @xmath2 kpc .
we are grateful to everyone at the arecibo observatory for their help in conducting these observations .
it is a pleasure to acknowledge stimulating discussions with fabian heitsch , bruce elmegreen , bob benjamin , and jay gallagher .
we thank an anonymous referee for constructive suggestions .
support by nsf grants ast-0097417 and ast-0707679 is gratefully acknowledged . , r. , haynes , m. p. , kent , b. r. , perillat , p. , saintonge , a. , brosch , n. , catinella , b. , hoffman , g. l. , stierwalt , s. , spekkens , k. , lerner , m. s. , masters , k. l. , momjian , e. , rosenberg , j. l. , springob , c. m. , boselli , a. , charmandaris , v. , darling , j. k. , davies , j. , lambas , d. g. , gavazzi , g. , giovanardi , c. , hardy , e. , hunt , l. k. , iovino , a. , karachentsev , i. d. , karachentseva , v. e. , koopmann , r. a. , marinoni , c. , minchin , r. , muller , e. , putman , m. , pantoja , c. , salzer , j. j. , scodeggio , m. , skillman , e. , solanes , j. m. , valotto , c. , van driel , w. , & van zee , l. 2005 , , 130 , 2598 , d. & burton , w. b. 1997 , atlas of galactic neutral hydrogen ( atlas of galactic neutral hydrogen , by dap hartmann and w. butler burton , pp . 243 .
isbn 0521471117 .
cambridge , uk : cambridge university press , february 1997 . ) | we present results from neutral hydrogen ( hi ) observations of the tip of the magellanic stream ( ms ) , obtained with the arecibo telescope as a part of the on - going survey by the consortium for galactic studies with the arecibo l - band feed array .
we find four large - scale , coherent hi streams , extending continously over a length of 20 degrees , each stream possessing different morphology and velocity gradients .
the newly discovered streams provide strong support for the tidal model of the ms formation by @xcite , which suggested a spatial and kinematic bifurcation of the ms .
the observed morphology and kinematics suggest that three of these streams could be interpreted as a 3-way splitting of the main ms filament , while the fourth stream appears much younger and may have originated from the magellanic bridge .
we find an extensive population of hi clouds at the tip of the ms .
two thirds of clouds have an angular size in the range 3.5@xmath010@xmath0 . we interpret this as being due to thermal instability , which would affect a warm tail of gas trailing through the galactic halo over a characteristic timescale of a few myrs to a few hundred myrs .
we show that thermal fragments can survive in the hot halo for a long time , especially if surrounded by a @xmath1 k halo gas .
if the observed clumpy structure is mainly due to thermal instability , then the tip of the ms is at a distance of @xmath2 kpc .
a significant fraction of hi clouds at the tip of the ms show multi - phase velocity profiles , indicating the co - existence of cooler and warmer gas . |
late - type dwarf galaxies are commonly thought to have slowly rising rotation curves and to be dominated by dark matter at all radii ( e.g. , carignan & beaulieu 1989 ; persic et al .
1996 ; ct et al .
however , in a recent study of a large sample of late - type dwarf galaxies for which the rotation curves were derived in a uniform way , taking the effects of beam smearing into account , swaters et al .
( 2009 ) found that the rotation curves of late - type dwarf galaxies have shapes similar to those of late - type spiral galaxies .
for the dwarf galaxies in their sample , the rotation curves , when expressed in units of disk scale lengths , rise steeply in the inner parts and start to flatten at two disk scale lengths , as is usually seen in spiral galaxies ( e.g. , broeils 1992a ; verheijen & sancisi 2001 ) .
such a difference in rotation curve shapes may have implications for the dark matter properties for late - type dwarf galaxies .
we will investigate the implications for the swaters et al .
( 2009 ) sample here . for spiral galaxies , mass models based on the extended rotation curves
indicate that large amounts of dark matter are required to explain the outer parts of observed rotation curves ( e.g. , van albada et al .
1985 ; begeman 1987 ; broeils 1992a ) . in most of the galaxies in these studies , the inner parts of the observed rotation curves ( out to two or three disk scale lengths ) could be explained by scaling up the contribution of the stellar disk to the rotation curve , in agreement with findings based on optical rotation curves ( kalnajs 1983 ; kent 1986 ) .
the same scaling , however , leaves large discrepancies in the outer parts of galaxies with rotation curves ( van albada & sancisi 1986 ) .
this discrepancy is interpreted as evidence for the existence of large amounts of dark matter in galaxies .
alternatively , the observed discrepancy could be explained by a different theory of gravitaty , such as mond ( modified newtonian dynamics ; milgrom 1983 ; sanders 1996 ) . the dark matter properties of galaxies are usually based on mass modeling of the rotation curves
if the contributions of the visible components are fixed , then whatever remains is the dark matter .
a major obstacle is that the precise contribution of the stars to the rotation curve is not known , because the mass - to - light ratio of the stars is unknown .
upper limits to the mass - to - light ratios have been obtained by assuming that the contribution of the stellar disk is maximal ( kalnajs 1983 ; kent 1986 , 1987 ; van albada and sancisi 1986 ) .
this ` maximum disk ' solution minimizes the amount of dark matter required to explain the observed rotation curves . at the same time , as shown e.g. , by van albada & sancisi ( 1986 ) , the uncertainties in the stellar mass - to - light ratios allow for a range in mass models with different dark matter distributions .
rotation curve studies of the dwarf galaxy ddo 154 ( carignan & freeman 1988 ; carignan & beaulieu 1989 ) indicated , however , that this galaxy is dominated by dark matter at all radii , including the region well within the optical disk . even when the contribution of the stellar disk is scaled as high as is allowed by the observed rotation curve ( i.e. , the maximum disk solution ) , the stellar disk could not be scaled to explain the observed rotation curves out to two or three disk scale lengths .
the observations of ddo 154 , along with studies of scaling relations based on relatively few well - studied dwarf galaxies ( e.g. , casertano & van gorkom 1991 ; broeils 1992a ; persic et al .
1996 ) , led to the generally accepted picture that dwarf galaxies have slowly rising rotation curves and are dominated by dark matter at all radii .
there are , however , also studies that provide a different picture , in which the stellar disks could be scaled to explain all of the inner rise of the rotation curves ( e.g. , carignan 1985 ; carignan et al .
1988 ; lake et al .
1990 ; broeils 1992b ; kim et al .
1998 ) , suggesting that the dark matter properties may be similar to those of spiral galaxies .
a major problem is that in studies to date the galaxies have been observed with very different instrumental setups , and that the rotation curves were derived using different procedures , some of which may have been prone to introducing systematic errors ( see e.g. , swaters et al .
2002 ; de blok et al . 2008 ) .
furthermore , the effects of beam smearing were not taken into account , even though these can be important ( see e.g. , begeman 1987 ; swaters et al .
2009 ) .
in order to improve this situation we have obtained observations for a sample of 73 dwarf galaxies with a single instrument ( swaters 1999 , hereafter s99 ; swaters et al .
2002 , hereafter paper i ) , as well as @xmath0-band observations ( swaters & balcells 2002 , hereafter paper ii ) . from the observations , we derived rotation curves in a uniform way , taking into account the effects of beam smearing ( s99 ; swaters et al .
2009 , hereafter paper iii ) . from this sample
we have selected 18 high quality rotation curves for a detailed mass model analysis which we report in this paper .
the layout of this paper is as follows . in the next section
we will describe the sample and the rotation curves . in section [ themodels ]
the different components that are used in the mass models and the fitting of these mass models to the rotation curves are described .
section [ themodres ] presents the results of the mass modeling . in section [ thedisc ]
the results are discussed , and we present our conclusions in section [ theconclusions ] .
the late - type dwarf galaxies in this sample have been observed as part of the whisp project ( westerbork hi survey of spiral and irregular galaxies ; for a more detailed description of the whisp project and its goals , see paper i ) .
the galaxies in the whisp sample have been selected from the ugc catalog ( nilson 1973 ) , taking all galaxies with declinations north of @xmath1 , blue major axis diameters larger than @xmath2 and measured flux densities larger than 100 mjy . from this list
we selected the late - type dwarf galaxies , defined as galaxies with hubble types later than sd , supplemented with spiral galaxies of earlier hubble types but with absolute @xmath3-band magnitudes fainter than @xmath4 .
for a detailed description of the selection criteria , see paper i. optical @xmath0-band data for all these galaxies are presented in paper ii , which also includes a discussion of the distance uncertainties for the galaxies presented here .
rrrrrrrrrrrrrrrrr 731 & 8.0 & -16.6 & 1.65 & 23.0 & 0.34 & 57 & 15.1 & 1.60.3 & @xmath5 & 0.45 & 17030 & 0.80.1 & 0.23 & 15030 & 0.810.09 & 0.22 + 3371 & 12.8 & -17.7 & 3.09 & 23.3 & 0.18 & 49 & 12.5 & 1.10.2 & @xmath5 & 0.73 & 376 & 2.10.2 & 0.85 & 326 & 2.20.3 & 0.82 + 4325 & 10.1 & -18.1 & 1.63 & 21.6 & 0.08 & 41 & 9.1 & & & 1.81 & 33080 & 0.80.1 & 1.77 & 30070 & 0.750.10 & 1.49 + 4499 & 13.0 & -17.8 & 1.49 & 21.5 & 0.30 & 50 & 2.3 & 145 & 2.80.7&0.58 & 6515 & 1.30.2 & 0.15 & 3710 & 1.70.3 & 0.22 + 5414 & 10.0 & -17.6 & 1.49 & 21.8 & & 55 & 4.6 & & & 1.13 & 4412 & 1.50.3 & 0.14 & 308 & 1.70.4 & 0.18 + 6446 & 12.0 & -18.4 & 1.87 & 21.4 & 0.17 & 52 & 4.0 & 5.31.5 & 6.21.6&0.39 & 20040 & 0.80.1 & 0.16 & 12020 & 0.950.10 & 0.05 + 7323 & 8.1 & -18.9 & 2.20 & 21.2 & & 47 & 3.0 & 2.60.8 & @xmath5 & 0.39 & 5012 & 2.00.3 & 0.66 & 278 & 2.70.7 & 0.30 + 7399 & 8.4 & -17.1 & 0.79 & 20.7 & 0.27 & 55 & 7.7 & 314 & 3.00.3&1.27 & 51080 & 0.60.1 & 0.98 & 37060 & 0.760.08 & 0.97 + 7524 & 3.5 & -18.1 & 2.58 & 22.2 & 0.24 & 46 & 7.2 & 1.40.3 & @xmath5 & 0.27 & 738 & 1.40.1 & 0.44 & 597 & 1.50.1 & 0.29 + 7559 & 3.2 & -13.7 & 0.67 & 23.8 & 0.15 & 61 & 13.1 & 6.12.0 & @xmath5 & 0.22 & 10040 & 0.50.2 & 0.09 & 9040 & 0.550.21 & 0.09 + 7577 & 3.5 & -15.6 & 0.84 & 22.5 & & 63 & 0.9 & & & 0.40 & 84 & 0.80.3 & 0.15 & 0 & & 0.63 + 7603 & 6.8 & -16.9 & 0.90 & 20.8 & 0.54 & 78 & 4.1 & 7.92.5 & 4.91.8&0.84 & 10020 & 0.90.1 & 0.30 & 6418 & 1.20.2 & 0.17 + 8490 & 4.9 & -17.3 & 0.66 & 20.5 & 0.29 & 50 & 4.4 & 3510 & 1.90.3&0.39&1100200&0.330.04&0.34&570150&0.440.05 & 0.20 + 9211 & 12.6 & -16.2 & 1.32 & 22.6 & 0.41 & 44 & 11.2 & 5.22.2 & 5.12.1&0.58 & 8626 & 1.00.2 & 0.31 & 7224 & 1.10.2 & 0.24 + 11707 & 15.9 & -18.6 & 4.30 & 23.1 & 0.14 & 68 & 9.3 & 0.70.1 & @xmath5 & 0.83 & 6121 & 1.70.3 & 0.42 & 4916 & 1.80.4 & 0.42 + 12060 & 15.7 & -17.9 & 1.76 & 21.6 & 0.07 & 40 & 8.3 & 0.50.2 & @xmath5 & 0.18&500200&0.450.14&0.55 & 400160&0.480.15 & 0.36 + 12632 & 6.9 & -17.1 & 2.57 & 23.5 & 0.16 & 46 & 15.1 & 1.00.2 & @xmath5 & 1.06 & 11020 & 1.00.2 & 0.12 & 10020 & 1.00.2 & 0.13 + 12732 & 13.2 & -18.0 & 2.21 & 22.4 & 0.27 & 39 & 7.5 & 2.00.4 & 167 & 0.37 & 7418 & 1.40.2 & 0.17 & 5316 & 1.60.3 & 0.24 + the sample of late - type dwarf galaxies used here is selected from the sample of 62 late - type dwarf galaxies for which rotation curves could be derived in paper iii . here , we only selected galaxies with high quality rotation curves ( classified as such in paper i ) and with inclinations in the range @xmath6 ( but mass models for galaxies with lower - quality rotation curves galaxies were presented by s99 ) .
the lower limit of @xmath7 was chosen to also include ugc 12732 .
note that edge - on galaxies have been excluded , because their rotation curves are difficult to determine , and also because the radial distributions of the gas and especially that of the light from the stellar disk are difficult to measure .
we excluded ugc 11861 because of its high luminosity .
the resulting sample , listed in table [ tabfits ] , contains 18 galaxies .
it is possible that some of the galaxies in our sample are affected by noncircular motions .
however , because the galaxies were selected from the larger sample presented in paper iii to be symmetric in nature , it is unlikely noncircular motions play an important role .
moreover , because the rotation curves have been derived by averaging over annuli , effects from noncircular motions are minimized . as can be seen in table [ tabfits ] , most of the galaxies in our sample have absolute magnitudes fainter than @xmath8 , as expected for dwarf galaxies . some are a little brighter than @xmath8 , because the initial sample selection , described in detail in paper i , was in part based on morphological type , and because of uncertainties in the determination of the absolute magnitudes that were used in the initial sample selection .
most of the late - type dwarf galaxies have small exponential disk scale lengths .
the median value for the scale length is 1.6 kpc , and 5 galaxies have scale lengths of less than 1 kpc .
however , 7 galaxies have scale lengths larger than 2 kpc , and 5 of these have surface brightnesses lower than @xmath9 mag .
these galaxies have properties that are similar to those of the class of low surface brightness ( lsb ) galaxies studied by e.g. , de blok et al .
( 1995 ) .
the rotation curves have been derived with an interactive procedure , as described in detail in paper iii .
first an estimate of each rotation curve was made based on a simultaneous fit by eye to six position - velocity diagrams with different position angles .
next , each rotation curve was refined by constructing a model data cube , based on the input rotation curve , and then adjusting the rotation curve to find the best match between the model cube and the observed data cube . as the comparison with high - resolution h@xmath10 data in paper iii showed , this procedure made it possible to correct for the effects of beam smearing to a large degree
the circular velocity is a direct reflection of the gravitational potential of a galaxy , assuming it is axially symmetric and in equilibrium : @xmath11 where @xmath12 is the radial force , @xmath13 the gravitational potential , @xmath14 the radius and @xmath15 the circular velocity . as was discussed in paper iii
, the corrections for asymmetric drift tend to be uncertain because the gas pressure usually can not be determined accurately . because the corrections are expected to be small ( less than 1 km s@xmath16 in the inner parts of the rotation curves and less than 3 km s@xmath16 at all radii for 95% of the galaxies presented here )
, the observed rotation curves were used to represent the circular velocities .
the gravitational potential is the sum of the gravitational potentials of the individual mass components in each galaxy .
expressed in velocities , this sum becomes @xmath17 with @xmath18 the contribution of the stellar disk to the rotation curve , @xmath19 the contribution of the gas and @xmath20 that of the dark halo . in equation
[ eqprefit ] the contribution of a bulge component is left out because the galaxies in this sample have little or no bulge . since we have no prior knowledge of the stellar mass - to - light ratio , some value has to be assumed .
the contribution of the gas to the rotation curve includes the contribution of helium .
this scaling of the is represented by @xmath21 .
making this explicit , equation [ eqprefit ] becomes @xmath22 where @xmath23 is the contribution of the stellar disk for a stellar mass - to - light ratio of unity , and @xmath24 is that of the only .
each of the individual components in this equation is described below .
the @xmath0-band luminosity profiles presented in paper ii have been used to calculate the contribution of the stellar disk to the observed rotation curve , assuming the mass - to - light ratio is independent of radius . because most of the galaxies in this sample have light profiles that are well represented by an exponential disk , and generally have little or no bulge , the light profiles were not decomposed into a disk and a bulge component .
the contribution of the stellar disk to the rotation curve was calculated using the prescription given in casertano ( 1983 ) , assuming that the galaxies are optically thin and assuming an intrinsic thickness of @xmath25 .
this value was chosen because it appears to be a suitable value for the average of the intrinsic thickness over the range of galaxy types included in this sample .
late - type disk galaxies , of morphological types around sd , are reported to have intrinsic thicknesses closer to @xmath26 ( e.g. , de grijs 1997 ; schwarzkopf & dettmar 2000 ; bizyaev & kajsin 2004 ) , but for late - type dwarf galaxies higher values may be found , up to @xmath27 , especially towards the lowest luminosity systems ( van den bergh 1988 ; binggeli & popescu 1995 ; sung et al .
we estimate the uncertainty in the amplitude of the rotation curve introduced by the variations in @xmath28 from galaxy to galaxy to be @xmath29 ( see s99 for more details ) .
the major uncertainty in determining a mass model for a galaxy is the mass to light ratio , which is not known a priori and can not be derived from the rotation curve fits alone .
equally well fitting mass models can be obtained with a range in ( e.g. van albada et al .
1985 ; swaters et al . 2000 , 2003 ; dutton et al . 2005
; see also section [ themodres ] ) . however , it is possible to obtain limits on by scaling up the contribution of the stellar disk to explain most of the observed rotation curve in the inner parts ( the so - called maximum disk hypothesis ) , and by reducing the contribution of the stellar disk to its minimum while still obtaining a good fit to the rotation curve .
the procedure for deriving the contribution of the gas to the rotation curve is very similar to that of the stellar disk .
the radial profiles presented in paper i were used to calculate this contribution .
we assumed that the gas layer has an intrinsic thickness of @xmath25 , the same value that was assumed for the thickness of the stellar disk .
little is known about the vertical distribution of the gas in late - type dwarf galaxies .
there is some evidence that the gas and the stars have the same vertical distribution ( bottema et al .
fortunately the precise choice of the thickness of the layer has very little influence on the shape or amplitude of the rotation curve and the uncertainty is smaller that that introduced by assuming a constant thickness of @xmath25 for the optical disk . in order to derive the contribution of the gas to the rotation curve ,
the was assumed to be optically thin . to correct for the mass fraction of helium ,
the mass was scaled by a factor @xmath30 , derived assuming a primordial helium fraction of 24% ( steigman 2007 ) .
other gas components that might contribute to the rotation curve , such as molecular hydrogen , have been ignored .
to date , most evidence suggests there is much less molecular hydrogen in late - type dwarf galaxies than there is , even when taking into account that the conversion factor to convert the observed co flux into a molecular gas column density may be substantially higher in late - type dwarf galaxies ( e.g. , israel et al .
1995 ; boselli et al .
2002 ; buyle et al .
2006 ) . moreover , if molecular gas were to have the same radial distribution as the stellar disk , its contribution is implicitly included in the maximum disk fits .
several different halo radial mass profiles have been proposed over the years , most of which produce good fits to the observed rotation curves .
halo mass - density profiles as predicted by cosmological simulations have received a great deal of attention , because a confrontation of these profiles against the observed rotation curves provides a powerful test of these cosmological models . for most of the sample presented here , such tests have been presented in van den bosch & swaters ( 2001 ) and swaters et al .
( 2003 ) . in this paper
, we focus on an empirical description of the properties of the dark matter halos . for this purpose
, we have used the density profile of an isothermal sphere that has often been used in the literature to represent the dark halo .
the advantage is that it will be easier to compare our results with those of other studies ( e.g. , broeils 1992a ; verheijen 1997 ; noordermeer 2006 ; see also section [ subcomp ] ) .
the radial density profile of an isothermal sphere can be approximated by @xmath31^{-1 } , \label{eqisorad}\ ] ] where @xmath32 is the central density of the halo , and @xmath33 the core radius .
this density profile gives rise to a rotation curve of the following form @xmath34}. \label{eqisovrot}\ ] ] the relative contribution of each of the three components described above is determined by a simultaneous fit of the right - hand side of equation [ eqfit ] to the observed rotation curve . for an isothermal halo , there are four free parameters in equation [ eqfit ] .
however , the contribution of the atomic gas to the rotation curve is known , and therefore , with the correction for the mass fraction of helium , @xmath21 is fixed to 1.32 . on the other hand ,
the value of is not known beforehand , and a range of and halo parameters exists which produce a good fit to the data ( van albada et al .
mathematically , it is possible to derive the parameters for the isothermal halo and the value of from a best fit to the observed rotation curve .
however , the physical meaning of such a best fit is unclear , as the values found from such a fit depend on the small scale variations and uncertainties in the rotation curves . because of these uncertainties in ,
two limiting cases are presented . in one case , the contribution of the stellar disk
is scaled up to explain most of the rotation curve out to two or three disk scale lengths .
this is the maximum disk solution . in the other limiting case
, the contribution of the stellar disk is completely ignored , and the rotation curve is fitted with the contributions of dark halo and gas only .
the maximum disk fits were done by eye , ensuring that the rotation curve of the stellar disk contributes maximally to the observed rotation curve within its observational uncertainties , following e.g. , broeils ( 1992a ) , verheijen ( 1997 ) , and noordermeer ( 2006 ) .
we have refit several mass models from the literature using our methods .
we were able to reproduce the original results to better than 10 percent .
a mass model is only considered to be compatible with a maximum disk if the stellar disk can be scaled up to explain at least @xmath35% of the rotation velocity at 2.2 disk scale lengths ( e.g. , sackett 1997 ; see also section [ themain ] ) .
there has been considerable effort over the past decade to relate the broad - band colors of galaxies to their through stellar population synthesis modeling ( e.g. , bell & de jong 2001 ; bell et al .
2003 ; zibetti et al .
the advantage of using such models is that it provides an independent estimate of and possibly a more realistic value of .
however , there is some uncertainty in as derived from population synthesis modeling , because of , for example , uncertainties in the star formation history , the initial stellar mass function , metallicity , and the details of the late - phases of stellar evolution .
the combined uncertainty may be a factor 4 or even larger if some of uncertainties are systematic ( see e.g. , bershady et al .
2010 ) .
the six galaxies in our sample with both @xmath3 and @xmath0-band observations presented in paper ii have an average @xmath36 color of 0.8 mag . following bell et al .
( 2003 ) , this corresponds in @xmath37 , although this value of is uncertain , as discussed above .
however , for any near unity , all but one of the galaxies in our sample are completely dominated by dark matter , so the exact choice of has little influence on the derived dark halo parameters ( see also section [ themodres ] ) . given these considerations , and because we had already calculated the contribution of the stellar disk for @xmath38 for the input of the mass modeling , we used these to determine the dark halo properties in these population synthesis - based restricted- fits .
the rotation curve fits are shown in figures [ figallisomax ] and [ figallisoml1 ] .
the minimum disk fits are not shown separately because in most cases the mass models are very similar to those for @xmath38 .
the fit parameters are given in table [ tabfits ] .
the uncertainties on the fitted parameters have been derived from the 68% confidence levels and include the covariance between @xmath32 and @xmath33 .
the uncertainties on the derived rotation velocities are non - gaussian , as the points in the rotation curves are correlated , and the rotation curves and their uncertainties can be affected by systematic effects .
the confidence levels and the corresponding uncertainties should be considered estimates .
in this section we will present the results from fitting the different mass models : maximum disk , minimum disk , and fixed .
the most striking result from the maximum disk fits , presented in figure [ figallisomax ] , is that in all 18 galaxies the contribution of the stellar disk can be scaled up to explain most of the inner parts of the rotation curves out to two or three disk scale lengths , with the exception of ugc 4499 .
thus , it seems that the fact that the stellar disk can be scaled to explain most of the inner parts of the rotation curves of late - type dwarf galaxies is rule rather than exception .
this is in contrast with the results from several previous studies of dwarf galaxies , in which the stellar disk only can explain a small fraction of the inner rise ; this is examined in more detail in section [ subcomp ] . to illustrate how the maximum disk mass models depend on the optical galaxy properties ,
we have plotted the mass models against surface brightness and absolute magnitude in figure [ figisomaxplot ] . in this figure ,
the origin of each model indicates the @xmath39 and @xmath40 for each galaxy .
the rotation curves are represented by filled circles for radii smaller than three disk scale lengths , and by open circles for larger radii . from an inspection of figure [ figisomaxplot ]
it appears that there is no clear trend between the relative contribution of the stellar disk to the rotation curve on the one hand , and rotation curve shape , luminosity or surface brightness on the other .
maximum disks are seen at all surface brightnesses and all absolute magnitudes .
this point is demonstrated in more detail in figures [ figrelvds23h ] and [ figrelvdmu ] .
figure [ figrelvds23h ] shows that there is no clear correlation between the rotation curve shape as determined by the logarithmic slope between two and three disk scale lengths @xmath41 , and the relative contribution of the stellar disk to the rotation curve at 2.2 disk scale lengths ( see paper iii for details ) .
independent of the rotation curve shape , the contribution of the stellar disk to the rotation curve is about 90% at 2.2 disk scale lengths .
figure [ figrelvds23h ] also shows the correlation between rotation curve shape and the relative contribution of the stellar disk at four disk scale lengths .
the contribution of the stellar disk to the rotation curve has dropped to about 70% .
there is one galaxy , ugc 12060 , where even at four disk scale lengths the stellar disk can still explain most of the observed rotation curve .
this galaxy has excess light between three and four disk scale lengths , when compared to an exponential fall - off . from figure [ figrelvdmu ]
it is clear that the maximum disk hypothesis works equally well for all surface brightnesses . from the results presented here ,
it is apparent that for the vast majority of these late - type dwarf galaxies the contributions of their stellar disks to the rotation curves can be scaled to explain most of the inner parts of the observed rotation curves .
therefore , these dwarf galaxies do not necessarily have insignificant mass in their stellar disks and dominant dark halos , as concluded in earlier work . for the maximum disk fits ,
dark matter only becomes important at radii larger than three or four disk scale lengths .
these maximum disk mass models for late - type dwarf galaxies are similar to those seen in spiral galaxies ( e.g. , begeman 1987 ; broeils 1992a ; verheijen 1997 ) .
although for the majority of the late - type dwarf galaxies in our sample the contribution of the stellar disk can explain most of the rotation curve in the inner parts , the stellar mass - to - light ratios may be high . in figure [ fighistml ]
a histogram is presented of the derived values of for the maximum disk fits .
the average value for is 7.7 , but ranges from 1 to 15 , with the high end being well outside the range predicted by current population synthesis models .
these high values of will be discussed in more detail in section [ thedisc ] .
figure [ figmlvsmuabs ] shows plotted against the central surface brightness @xmath39 and the absolute magnitude @xmath40 .
galaxies with lower surface brightnesses have higher values for under the assumption of maximum disk ( see also section [ secmls ] ) . in three cases ( ugc 4325 , ugc 5414 and ugc 7577 ) , the rotation curve can be explained by the stellar disk and the alone ; see figure [ figallisomax ] .
hence , the fitted dark halos will have central densities close to zero , and the core radii are not constrained . for several galaxies with more extended rotation curves ( see figure [ figallisoml1 ] ) , the rotation curves of the dark halos have a solid body appearance , indicating that the dark matter halos have constant density cores . in these cases ,
the core radius is also unconstrained and only the halo density can be determined . only for eight galaxies with sufficiently extended rotation curves both the core radius and the central density
can be determined reliably .
figure [ figrcvsh ] shows that the core radius @xmath33 and the disk scale length @xmath42 are correlated .
the halo core radius is on average @xmath43 times as large as the disk scale length
. a similar ratio of @xmath44 is found for the spiral galaxies presented in broeils ( 1992a ) , demonstrating the similarity between the maximum disk fits of the late - type dwarf galaxies in our sample and those of spiral galaxies .
such a ratio of core radius to disk scale length of about three is expected for galaxies with a more or less flat rotation curve , as was also mentioned by van albada & sancisi ( 1986 ) , who found that the expected ratio is about 3.5 . for scaling relations between the optical properties of the galaxies in our sample and the relative fractions of mass in stars , gas , and dark matter
, we refer to s99 .
note from figure [ figallisomax ] that in the case of maximum disk models the shapes of the halo rotation curves often are similar to the shapes of the gas rotation curves .
this is remarkable and indicates that the ratio of dark matter surface density to gas surface density is constant in the outer parts .
this was noted for spiral galaxies by bosma ( 1981 ) and further investigated by hoekstra et al .
. a detailed study of the apparent link between the and dark matter densities will be the subject of an upcoming paper .
a lower limit to the contribution of the stellar disk to the rotation curve can be obtained by decreasing to its lowest possible value while the mass model still produces a good fit to the rotation curve . for the late - type dwarf galaxies in this sample
, the rotation curves can be well fitted using only a dark halo and the ( i.e. , @xmath45 ) .
a minimum disk fit thus does not provide a useful lower limit on , but it does provide an upper limit on the halo mass within a given radius and on the central density , and a lower limit on the halo core radius .
the results for the minimum disk fits are given in table [ tabfits ] .
the mass models themselves are not shown because the dark halos are very similar to those for @xmath38 , shown in figure [ figallisoml1 ] .
details of the correlations between the minimum - disk dark halo properties and optical properties were presented in s99 . as was shown above , and as will be discussed in more detail in section [ thedisc ] , the maximum - disk values are higher than expected from population synthesis models . to investigate the dark halo properties for values of that are in agreement with population synthesis models ( as was described in section [ subrotcurfits ] )
, we also fit mass models for @xmath38 .
the results of the fits are presented in table [ tabfits ] , and the fits themselves are shown in figure [ figallisoml1 ] . with @xmath38 , the relative contribution of the stellar disk to the rotation curve
is tightly correlated with surface brightness , as can be seen in figure [ figrelvdmu ] .
such a tight correlation is expected for models with a fixed @xmath46 , because in that case the surface brightness is directly linked to the surface mass density and hence to the relative contribution of the stellar disk .
the main result of the present study is that the contribution of the stellar disk can be scaled up ( `` maximum disk '' ) to explain the inner parts of the rotation curve of all but one of the 18 late - type galaxies in the present sample .
these maximum disk models are similar to those for late - type spiral galaxies in which the stellar disks can also be scaled to explain the inner rotation curves ( e.g. , kalnajs 1983 ; kent 1987 ; begeman 1987 ; broeils 1992a ; verheijen 1997 ) and also those of the lsb galaxies presented in swaters et al .
( 2000 ; 2003 ) .
how well maximum disk works is described by @xmath47 the fractional contribution of the maximum stellar disk to the rotation curve at two disk scale lengths .
a low value for @xmath47 means that even when scaled as high as possible , the stellar disk can only explain a small fraction of the observed rotation curve .
a high value of @xmath47 ( near unity ) means that the stellar disk can be scaled to explain most or all of the observed rotation curve . for the galaxies in our sample @xmath48 , with a dispersion around the mean of @xmath49 .
this is nearly identical to the average for the spiral galaxies in the sample of broeils ( 1992a ) , for which @xmath50 , and to the average for spiral galaxies in the verheijen ( 1997 ) sample , with @xmath51 ( see figure [ figmaxdiskfrac ] ) .
if we assume that the values for @xmath47 near unity mean that these dwarf galaxies physically have maximal disks ( despite the high required mass - to - light ratios ) , then the stellar disk can explain the mass distribution over the optical parts of the galaxy , and dark matter only becomes relevant at large radii .
under the same assumption , the average ratio of stellar - to - dark mass within four disk scale lengths is @xmath52 for the dwarf galaxies in our sample . for the spiral galaxies in the sample presented in broeils ( 1992a ) the average stellar - to - dark mass ratio within four disk scale lengths is @xmath53 .
thus , the maximum disk fits for dwarf galaxies and spiral galaxies are very similar , with similar relative amounts of mass in the dark and luminous components within similar numbers of disk scale lengths . in these maximum disk fits ,
dwarf galaxies are not dominated by dark matter at all radii .
dark matter is only needed to explain the outer part of the rotation curve .
the conclusion reached here contrasts with the general notion from previous work that in late - type dwarf galaxies dark matter dominated everywhere . the average value of @xmath47 in the comparison samples of broeils ( 1992a ) and verheijen ( 1997 ) when including only late - type dwarf galaxies is lower than what we find here . for the late - type dwarf galaxies in the sample of broeils ( 1992a ) ,
@xmath54 , and for the verheijen ( 1997 ) sample , @xmath55 .
such lower values of @xmath47 are found when a rotation curve rises slowly .
indeed , the conclusion that the stellar disk can not be scaled to explain the inner rise of the observed rotation curve is reached in many of the dark matter studies of late - type dwarf galaxies , e.g. ddo 154 ( carignan & freeman 1988 , carignan & beaulieu 1989 ) , ngc 300 ( puche et al . 1990 ) , ngc 55 ( puche et al . 1991 ) , ngc 3109 ( jobin & carignan 1990 ) , ngc 5585 ( ct et al . 1991 ) , ddo 168 ( broeils 1992a ) , ic 2574 ( martimbeau et al . ) , ngc 2915 ( meurer et al .
1996 ) , five dwarfs in the sculptor and centaurus a groups ( ct et al . 1997
; see also ct 1995 ) , ngc 5204 ( sicotte & carignan 1997 ) , ngc 2976 ( simon et al .
2003 ) , and ngc 6822 ( weldrake et al .
many of these studies have led to the notion that dwarf galaxies are dominated by dark matter at all radii . there are , however , also studies that provide a different picture . some early mass models for late - dwarf galaxies showed that the stellar disk could be scaled to explain all of the inner rise of the rotation curves , suggesting that the dark matter properties of these galaxies were similar to those of spiral galaxies ( carignan 1985 ; carignan , et al .
similar results were found for the four late - type dwarf galaxies in the virgo clusters studied by skillman et al .
( 1987 ) , ddo 170 ( lake et al .
1990 ) , ddo 105 ( broeils 1992a ) , ngc 1560 ( broeils 1992b ) , most of the late - type dwarf galaxies presented in van zee et al . ( 1997 ) , the large magellanic cloud ( kim et al .
1998 ; bekki & stanimirovic 2009 ) , and most of the galaxies presented in s99 and swaters et al .
( 2003 ) .
the spread in results seen in earlier studies raises the question whether the spread reflects an intrinsic spread in the rotation curve shapes of dwarf galaxies , which then results in a range of different mass models , or whether this observed spread is due to other factors that affect the derived rotation curves .
there are well - known systematic effects that can affect the shape of the derived rotation curve and can make it appear shallower .
for example : ( i ) beam smearing can wash out steep gradients and make the inner rotation curve appear more shallow ( e.g. , begeman 1987 ; paper iii ) ; ( ii ) rotation curves derived from high - inclination galaxies are susceptible to underestimated velocities in the inner parts ( e.g. , sancisi & allen 1979 ; swaters et al .
2003 ) ; ( iii ) some methods of deriving the velocity fields , ( e.g. , adopting the intensity - weighted mean velocity of a profile ) can lead to an underestimate of the derived rotation curve ( e.g. , bosma 1978 ; paper i ; de blok et al .
in addition , in some cases derived rotation curves may have been affected by noncircular motions ( e.g. , rhee et al .
2004 ; spekkens & sellwood 2007 ; valenzuela et al .
2007 ) . given that beam smearing has been ignored in many of the previous studies , that some of the galaxies in those studies have high inclinations , and that the majority of those studies used intensity - weighted mean velocity fields , it is likely that the inner slopes of the rotation curves in those previous studies have been underestimated . because such an underestimate of the inner slope limits how much the contribution of the stellar disk can be scaled in a maximum disk fit , an underestimate of the inner slope directly results in an underestimate of @xmath47 .
it is beyond the scope of this paper to review each rotation curve in the literature in detail and assess whether systematic effects may have played a role .
however , it is useful to consider as an example the well - studied prototype dwarf galaxy ddo 154 . from synthesis observations of this galaxy a slowly rising rotation curve was derived , and mass modeling indicated that ddo 154 was almost entirely dominated by dark matter ( carignan & freeman 1988 ; carignan & beaulieu 1989 ) .
based on the published mass model , we estimate that @xmath56 , much lower than the values found for our sample .
those early observations had a resolution of @xmath57 , and the derivation of the rotation curve was certainly affected by beam smearing , especially given the galaxy s inclination of approximately @xmath58 ( de blok et al .
2008 ) .
recently , de blok et al .
( 2008 ) presented higher resolution data for this galaxy . because of the @xmath59 resolution , and
the fact that the velocity field was derived using skewed gaussians , which can account for asymmetries often observed in line profiles , the derived rotation curve is less prone to the systematic effects discussed here . the rotation curve derived by de blok et al .
( 2008 ) rises somewhat more steeply than the one derived earlier by carignan & beaulieu ( 1989 ) , and de blok et al .
( 2008 ) find @xmath60 .
however , the position - velocity diagram shown in figure 81 of de blok et al .
( 2008 ) suggests that the rotation curve in the inner @xmath61 could be even steeper .
indeed , a new study of the kinematics of ddo 154 in the central regions based on long - slit spectroscopy and integral - field spectroscopy ( swaters et al .
, in preparation ) shows a rotation curve that rises somewhat more steeply than the one of de blok et al .
( 2008 ) , resulting in a value of @xmath62 .
a key feature of the sample presented in this paper is that the systematic effects mentioned above were avoided as much as possible .
the sample galaxies have inclinations between @xmath7 and @xmath63 and galaxies with large asymmetries were excluded .
the rotation curves were derived with an interactive technique using modeling of the observed data cube to correct for the effects of beam smearing ( paper iii ) . for the maximum disk models , the derived mass to light ratio @xmath46
is tightly coupled to surface brightness , with lower surface brightness galaxies having higher @xmath46 .
this is expected in order to explain the high mass density needed for the maximum disk despite the lsb nature of the stellar disks .
the values for @xmath46 required in the maximum disk models , reach as high as @xmath64 , well outside the range of what population synthesis models predict ( e.g. , bell et al .
2003 ; zibetti et al .
the precise value of @xmath46 as derived from such population synthesis may be uncertain by a factor of 2 to 4 ( portinari et al . 2004 ) , and possibly even higher ( bershady et al .
however , even considering these uncertainties , values as high as @xmath64 appear unlikely , as they would require a strong bias in the initial mass function towards low - mass stars , or a large fraction of the stellar mass being locked up in stellar remnants . the presence of a nonstellar component of mass with a distribution similar to that of the stars , e.g. , in the form of cold molecular gas as suggested by pfenniger & combes ( 1994 ) and revaz et al .
( 2009 ) , could provide an explanation .
alternatively , the stellar disks may be somewhat less than maximal .
independent measurements of from stellar velocity dispersions in spiral galaxies suggest that stellar disks have a submaximum contribution to the rotation curve , with a peak contribution of the stellar disk to the rotation curve of around 65% ( bottema 1993 ; westfall et al . , in preparation ) . at this level
, the stellar disk still determines the inner shape of the rotation curve , i.e. , one would still expect to find @xmath65 .
if the stellar disks would also contribute around 65% in late - type dwarf galaxies , this would reduce @xmath46 by almost a factor of 2.5 .
this would bring the mean and maximum @xmath46 of the galaxies in our sample to 3 and 6 , respectively , still above what population synthesis models predict .
if @xmath46 is assumed to be near unity , as predicted by stellar population synthesis models ( see section [ subrotcurfits ] ) , then well - fitting mass models are found as well . in this case , all galaxies except ugc 7577 are dominated by dark matter .
moreover , the relative contribution of dark matter increases towards lower surface brightness , and the contribution of the disk decreases correspondingly .
thus , for models with @xmath66 , we reach the same conclusions as are usually found for dwarf and lsb galaxies ( e.g. , carignan & beaulieu 1989 ; broeils 1992a ; de blok & mcgaugh 1997 ) .
there is , however , a critical difference . in many of those previous studies , as described above , low values for @xmath47 were derived . for our sample ,
however , we find a value for @xmath47 that is near unity .
this not only means that the stellar disk can be scaled to explain most of the inner parts of the observed rotation curve . if @xmath67 , this also means that the rotation curve of the stellar disk alone must have the same shape as that of the dark matter that dominates the potential . otherwise , @xmath47 could not be near unity . in spite of the uncertainties in @xmath46 , one result from this analysis is firm : the fact that the contribution of the stellar disk can be scaled up to explain the inner parts of the rotation curve indicates that the shapes of the rotation curve of the stellar disk and of the observed rotation curve are similar
. there appears to be a strong coupling between the distribution of the stars and that of the total mass ( see also sancisi 2004 ; paper iii ) .
we have presented mass models for a sample of 18 late - type dwarf galaxies . for this sample , the systematic effects due to beam smearing , intensity - weighted - mean velocity fields , high inclination , and noncircular motions
were avoided as much as possible .
we have analyzed the rotation curves of the galaxies in our sample by fitting mass models consisting of a stellar disk , a gaseous disk and the contribution of an isothermal halo , and we have reached the following conclusions : \(1 ) dwarf galaxies are not necessarily dominated by dark matter within their optical radii .
the rotation curve of the stellar disk can be scaled up to explain most of the inner parts of the observed rotation curves .
the maximum disk mass models are similar to those for spiral galaxies , with similar fraction of dark matter within a radius of four disk scale lengths . however , the required stellar mass - to - light ratios are high , up to about 15 in the @xmath0-band .
\(2 ) for the maximum disk models , the contribution of the stellar disk to the rotation curve is about 90% at two disk scale lengths , independent of surface brightness or luminosity .
the maximum - disk is correlated with surface brightness .
\(3 ) well fitting mass models can be obtained by assuming that the contribution of the stellar disk to the rotation is maximal , but also by assuming it is negligible .
this demonstrates that mass models for late - type dwarf galaxies suffer from the same indeterminacies as those for spiral galaxies .
\(4 ) if the stellar mass - to - light ratios are set to the values predicted by stellar population synthesis modeling , the relative contribution of the stellar disk decreases with surface brightness , while the relative fraction of dark matter increases .
\(5 ) whatever the contribution of the stellar disk to the rotation curve , the similarity in shapes between the rotation curve expected for the stellar disk and the observed rotation curve implies that in the inner parts of late - type dwarf galaxies the total mass density distribution is closely coupled to the luminous mass density distribution . | mass models for a sample of 18 late - type dwarf and low surface brightness galaxies show that in almost all cases the contribution of the stellar disks to the rotation curves can be scaled to explain most of the observed rotation curves out to two or three disk scale lengths .
the concept of a maximum disk , therefore , appears to work as well for these late - type dwarf galaxies as it does for spiral galaxies .
some of the mass - to - light ratios required in our maximum disk fits are high , however , up to about 15 in the @xmath0-band , with the highest values occurring in galaxies with the lowest surface brightnesses .
equally well - fitting mass models can be obtained with much lower mass - to - light ratios .
regardless of the actual contribution of the stellar disk , the fact that the maximum disk can explain the inner parts of the observed rotation curves highlights the similarity in shapes of the rotation curve of the stellar disk and the observed rotation curve .
this similarity implies that the distribution of the total mass density is closely coupled to that of the luminous mass density in the inner parts of late - type dwarf galaxies . |
flavor symmetry plays essential role in classification of the hadrons .
the light hadronic states are successfully described by using su(3 ) flavor symmetry . in the case
this symmetry is exact , hadrons belonging to the same representation of su(3 ) flavor group could be degenerate . experimentally it is known that the hadrons belonging to the same representation have different masses , which leads to su(3 ) flavor symmetry breaking . at quark level
, this symmetry is broken due to the mass difference of the light u , d and s quarks .
the breaking of the su(3 ) flavor symmetry might lead to mixing of hadrons .
in other words , the definite flavor eigenstates can mix to form the physically observed states .
long time ago , it is observed that the lowest lying hadrons @xmath1 and @xmath0 can be represented as the combination of the su(3 ) octet , pure isospin @xmath4 , and @xmath5 baryons in the following form @xcite , [ emix01 ] ^0 - ^0 , + ^0 + ^0 .
the @xmath6 mixing angle is estimated in framework of the quark model whose value is predicted to be @xmath7 @xcite .
( see also @xcite ) .
very recently , the lattice qcd ( lqcd ) group presented the latest estimate on @xmath6 mixing angle to have the value @xmath8 @xcite , which is approximately two times smaller compared to the prediction of the quark model .
the aim of the present note is to determine the @xmath0@xmath1 mixing angle within the qcd sum rules , and compare this mixing angle with the predictions of the quark model and lqcd . in determination of the @xmath0@xmath1 mixing angle within the qcd sum rules we follow the method suggested in @xcite , and for this goal we start by considering the following correlation function , [ emix02 ] = i d^x e^ipx 0 \{_h(x ) |_h(0 ) } 0 , where t is the time ordering operator , @xmath9 is the interpolating current , carrying the same quantum numbers as the corresponding hadron .
if the bare @xmath10 and @xmath11 states are mixed , the corresponding physical states with definite mass should be the linear combinations of these bare states . in this case , the interpolating currents corresponding to the physical states could be represented as the superposition of the interpolating currents corresponding to the bare states , i.e. , [ emix03 ] _ _ ^0 + _ ^0 , + _ _ ^0 - _ ^0 , where @xmath12 is the mixing angle between @xmath13 and @xmath14 states . in presence of only two physical states , eq .
( [ emix02 ] ) can be written as , [ emix04 ] = i d^4x e^ipx 0 \{_(x ) | _ } 0 .
it should be remembered that the general form of the correlator function is , [ nolabel01 ] ( p ) = _ 1(p^2 ) + _ 2(p^2 ) i , and coefficients of the @xmath15 and @xmath16 ( unit operator ) structures , i.e. , @xmath17 and @xmath18 can both be used in determining the mixing angle . in order to construct the sum rules for the mixing angle @xmath12 , the correlation function ( [ emix04 ] ) is calculated in terms of hadrons , quarks and gluons .
using the duality ansatz these two representations are matched and the sum rules for the corresponding physical quantity is obtained .
the hadronic representation of the correlation function is obtained by saturating it with the full set of baryons having the same quantum numbers as the corresponding interpolating current .
since @xmath19 and @xmath20 can create only the states @xmath21 and @xmath22 , correspondingly , the hadronic part of the correlation function is obviously zero . using eq .
( [ emix03 ] ) in eq .
( [ emix04 ] ) , one can easily obtain the expression for the mixing angle for both structures , [ emix05 ] 2= 2 _ ^0 _ ^0 - _
^0 , where @xmath23 are the correlation functions corresponding to the unmixed states , i.e. , [ emix06 ] _
ij^0 = i d^4x e^ipx 0 \ { _ i^0 ( x ) _ j^0 } 0 , where @xmath24 or @xmath25 .
so the problem of determination of the mixing angle requires the calculation of the theoretical part of the correlation function , for which the expressions of the interpolating currents are needed . according to the @xmath26 classification
the interpolating currents for the unmixed @xmath13 and @xmath14 are chosen as , [ emix07 ] _ ^0 2 ^abc \{2 ( u^at c d^b ) _ 5 s^c + 2 ( u^at c _ 5 d^b ) s^c + ( u^at c s^b)_5 d^c + ( u^at c _ 5 s^b ) d^c + ( d^at c s^b)_5 u^c - ( d^at c _ 5 s^b ) u^c } , + _
^0 ^abc \{(u^at c s^b ) _ 5 d^c + ( u^at c _ 5 s^b ) d^c + ( d^at c s^b ) _ 5 u^c + ( d^at c _ 5 s^b ) u^c } , where @xmath27 are the color indices , @xmath28 is the charge conjugation operator , and @xmath29 is the arbitrary constant with @xmath30 corresponding to the so called ioffe current . using the operator product expansion at @xmath31 , one can easily obtain the expressions for the correlation functions @xmath32 , @xmath33 , and @xmath34 from eq .
( [ emix03 ] ) from the qcd side for the @xmath35 and @xmath16 structures .
the expressions of these correlation functions are presented in the appendix . in
order proceed for the numerical calculations we need the values of the input parameters that are given as : @xmath36 @xcite , @xmath37 , @xmath38 , @xmath39 @xcite . for the masses of the light
quarks we use their @xmath40 values given as : @xmath41 , @xmath42 , @xmath43 @xcite .
it follows from the expressions of the invariant functions that in order to determine the @xmath0@xmath1 mixing angle three arbitrary parameters are involved , namely , the continuum threshold @xmath44 , the borel mass parameter @xmath45 , and the parameter @xmath29 ( see the expressions of the interpolating currents ) ; and of course the mixing angle should be independent of them all . as is well known , the continuum threshold is related to the energy of the first excited state .
the difference @xmath46 , where @xmath47 is the mass of the ground state , is equal to the energy needed to excite the particle to its first energy state .
this difference usually changes in the range between @xmath48@xmath49 .
it follows from the analysis of the mass sum rules that in order to reproduce the experimental values of the masses of the @xmath0 and @xmath1 baryons , the continuum threshold @xmath44 should lie in the range @xmath50 @xcite .
moreover , the working region of the borel mass parameter should be such that , the results for the @xmath0@xmath1 mixing angle should exhibit good stability with respect to the variation of @xmath45 at fixed values of @xmath44 .
the upper bound of @xmath45 is obtained by demanding that the higher states and continuum contributions should be less than 30% of the total result .
the lower bound of @xmath45 is determined from the condition that the contributions of higher dimensional operators should be less than the perturbative one . from these conditions
the working region of @xmath45 is determined to be @xmath51 . in figs .
( 1 ) and ( 2 ) , we present the dependence of the mixing angle @xmath12 on @xmath45 at the value of the continuum threshold @xmath52 and , at several fixed values of the auxiliary parameter @xmath29 , for the coefficients of the structures @xmath15 and @xmath16 , respectively . we observe from fig .
( 1 ) that in the range @xmath53 of the borel parameter , the mixing angle @xmath12 exhibits good stability for the values of the auxiliary parameter @xmath54 for the structure @xmath15 . as can be traced from fig .
( 2 ) , the mixing angle @xmath12 seems to be rather stable at all considered values of the auxiliary parameter @xmath29 for the structure @xmath16 at the fixed value of the continuum threshold @xmath52 .
our final attempt for determination of the mixing angle is to find the region of @xmath29 where the mixing angle exhibits insensitivity to its variation .
for this aim we study the dependence of the mixing angle @xmath12 on @xmath55 where @xmath56 , at several fixed values of @xmath45 and at @xmath52 , and presented them in figs .
( 3 ) and ( 4 ) for the coefficients of the structures @xmath15 and @xmath16 , respectively . in this respect ,
the results of our numerical analysis depicted in figs . ( 3 ) and ( 4 ) can be summarized as follows : * for the structure @xmath15 , in the above determined working regions of @xmath45 and @xmath44 , the best stability for the mixing angle is achieved when @xmath57 , and the mixing angle is found to have the value @xmath58 . * for the structure @xmath16 not only there is no stability region for the mixing angle , but also the mixing angle changes its sign .
therefore prediction for the value of the mixing angle from the structure @xmath16 is not reliable .
therefore we conclude that , the final result for the mixing angle is @xmath59 which is obtained from the @xmath15 structure . the error in determination of the mixing angle can be attributed to the uncertainties in the value of the continuum threshold @xmath44 , the quark condensates , and the scale parameter @xmath1 .
the results presented in this work can further be improved by taking @xmath60 corrections in to account .
finally , we compare our result on the @xmath0@xmath1 mixing with the calculations of the quark and lattice qcd models , whose predictions are , [ nolabel02 ] 0.86 ^ 0 , @xcite + 0.40 ^ 0 0.026 ^ 0 , @xcite respectively . from these results we observe that , our prediction on the @xmath0@xmath1 mixing angle is very close to the result obtained in the quark model , and more than two times larger compared to that of the result obtained in the lattice qcd model .
a reliable lattice qcd determination of the @xmath0@xmath1 mixing angle requires an equally highly accurate reproduction of the octet baryon mass differences , which has not yet been established . in conclusion , the mixing angle between the @xmath0 and @xmath1 baryons is estimated within the framework of the light cone sum rules method .
a comparison of our result with the predictions of the quark and lattice qcd models is presented .
99 r. h. dalitz , and f. von hippel , phys .
b * 10 * , 153 ( 1984 ) .
a. gal , and f. schreck , nucl .
b * 2 * , 110 ( 1967 ) ; a. gal , phys .
d * 92 * , 018501 ( 2015 ) .
keith a. olive _
( particle data group ) , chin .
c * 38 * , 090001 ( 2014 ) .
r. horsley _
( qcdsf ukqcd collaboration ) , phys .
d * 91 * , 074512 ( 2015 ) ; _ ibid _ d * 92 * , 018502 ( 2015 ) .
t. m. aliev , a. zpineci , v. i. zamiralov , phys .
d * 83 * , 016008 ( 2011 ) .
b. l. ioffe , prog .
. phys . * 56 * , 232 ( 2006 ) .
v. m. belyaev , b. l. ioffe , sov .
jetp * 57 * , 716 ( 1983 ) .
t. m. aliev , a. zpineci , m. savci , phys .
d * 66 * , 016002 ( 2002 ) .
f. x. lee , w. wang , phys .
d * 83 * , 094008 ( 2011 ) .
i. i. balitsky , v. m. braun , and a. v. kolesnichenko , nucl .
b * 312 * , 509 ( 1989 ) ; k. g. chetyrkin , a. khodjamirian , and a. a. pivovarov , phys . lett .
b 651 , 250 ( 2008 ) .
& & e^m_^0 ^ 2/2 m^2 e^m_^0 ^ 2/2 m^2 _ ^0 ^ 0 ^ 0 ( u , d , s ) = + & & - 13072 ^2 m^4 ( 1 - ) m_0 ^ 2 + \ { + ( 1 - ) ( m_d - m_u ) + 12 ^2 ( 1 - ) ( 7 + 5 ) m_0 ^ 2 - + 3 ( 1 - ^2 ) } + ( 1 - ^2 ) ( _ e - ) m_0 ^ 2 ( m_s + m_d- m_u- m_s ) + m^2 ( 2 - - ^2 ) ( m_s + m_d- m_u- m_s ) + \ { 3 m_0 ^ 2 + 32 ^2 ( 2 - - ^2 ) ( - ) + 3 m_0 ^ 2 } .
& & e^m_^0 ^ 2/2 m^2 e^m_^0 ^ 2/2 m^2 ( _ ^0 ^ 0 ^ 0 - _ ^0 ^ 0 ^ 0 ) ( u , d , s ) = + & & 14608 ^2 m^4 ( 1 - ) \ { + ( 4 + 5 ) ( m_d + m_u ) - } m_0 ^ 2 + \ { + \ { + 12 ^2 ( 1 - ) ( 7 + 5 ) m_0 ^ 2 ( - 2 ) } - \{(1 - ) + 3 ( 1 + + ^2 ) m_u + 12 ^2 ( 1 - ) ( 7 + 5 ) m_0 ^ 2 } + 9 ( 1 - ^2 ) } + ( 1 - ^2 ) ( _ e - ) m_0 ^ 2 + m^2 ( 2 - - ^2 ) + \{+ 32 ^2 ( 2 - - ^2 ) + 3 m_0 ^ 2 } .
& & e^m_^0 ^ 2/2 m^2 e^m_^0 ^ 2/2 m^2 _ ^0 ^ 0 ^ 0 ( u , d , s ) = + m^6 ( 2 - - ^2 ) ( m_d - m_u ) + m^4 ( 2 - - ^2 ) ( - ) + m^2 ( 1 - ) ( 1 + 2 ) ( _ e - ) ( m_d - m_u ) + m^2 ( 1 - ) + + ( 1 - ) \ { + [ ( 1 + 2 ) + 16 ^2 ( 2 + ) ( m_s - ( m_d - m_u ) ) } .
& & e^m_^0 ^ 2/2 m^2 e^m_^0 ^ 2/2 m^2 ( _ ^0 ^ 0 ^ 0 - _ ^0 ^ 0 ^ 0 ) ( u , d , s ) = + m^6 ( 2 - - ^2 ) ( m_d - 2 m_s + m_u ) + m^4 ( 2 - - ^2 ) ( - 2 + ) + m^2 ( 1 - ) ( 1 + 2 ) ( _ e - ) ( m_d - 2 m_s + m_u ) + m^2 ( 1 - ) + \ { ( 1 - ) ( 1 + 2 ) ( m_d - 2 m_s + m_u ) ^2 + 512 ^4 ( 1
+ + ^2 ) ( m_u - 2 m_s + m_d ) m_0 ^ 2 } + ( 1 - ) \{16 ^2 ( 2 + ) ( m_s - 2 m_u ) - ( 1 + 2 ) ( 2- ) + } , where @xmath45 is the borel parameter and @xmath1 is the energy cut off separating perturbative and nonperturbative regimes ; and @xmath63 is the euler constant .
note that the scale parameter @xmath1 is calculated in @xcite whose value is in the range @xmath64 .
* fig . 1 * dependence of the @xmath1@xmath0 mixing angle on the borel mass parameter @xmath45 at the fixed value of the continuum threshold @xmath52 , and at several fixed values of the auxiliary parameter @xmath29 , for the structure @xmath15 . + + * fig .
2 * the same as in fig . 1 , but for the structure @xmath16 . + + * fig .
3 * dependence of the @xmath1@xmath0 mixing angle on @xmath55 at the fixed value of the continuum threshold @xmath52 , and at several fixed values of the borel mass parameter @xmath45 , for the structure @xmath15 . + + * fig .
4 * the same as in fig . 1 , but for the structure @xmath16 . | the @xmath0@xmath1 mixing angle is calculated in framework of the qcd sum rules .
we find that our prediction for the mixing angle is @xmath2 which is in good agreement with the quark model prediction , and more than two times larger than the recent lattice qcd calculations .
# 1#2#3 @xmath3 # 1#2#3 @xmath3 0^*0 5_5 o _ ^0 _ pacs numbers : 11.55.hx , 11.30.hv , 12.38.t , 14.20.jn |
mathematics is the art of abstraction and generalization . historically ,
`` numbers '' were first natural numbers ; then rational , negative , real , and complex numbers were introduced ( in some order ) .
similarly , the concept of taking derivatives has been generalized from first , second , and higher order derivatives to `` fractional calculus '' of noninteger orders ( see for instance @xcite ) , and there is also some work on fractional iteration .
however , when we add some number of terms , this number ( of terms ) is still generally considered a natural number : we can add two , seven , or possibly zero numbers , but what is the sum of the first @xmath0 natural numbers , or the first @xmath1 terms of the harmonic series ? in this note , we show that there is a very natural way of extending summations to the case when the `` number of terms '' is real or even complex .
one would think that this method should have been discovered at least two hundred years ago and that is what we initially suspected as well . to our surprise , this method does not seem to have been investigated in the literature , or to be known by the experts , apart from sporadic remarks even in euler s work @xcite ( see equation ( [ eq : euler ] ) below ) .
of course , one of the standard methods to introduce the @xmath2 function is an example of a summation with a complex number of terms ; we discuss this in section [ secfromaxtodef ] , equation ( [ eqgamma ] ) . since this note is meant to be an introduction to an unusual way of adding , we skip some of the proofs and refer the reader instead to the more formal note @xcite . some of our results were initially announced in @xcite .
we start by giving natural conditions for summations with an arbitrary complex number of terms ; here @xmath3 , @xmath4 , @xmath5 , and @xmath6 are complex numbers and @xmath7 and @xmath8 are complex - valued functions defined on @xmath9 or subsets thereof , subject to some conditions that we specify later : ( s1 ) continued summation : : @xmath10 ( s2 ) translation invariance : : @xmath11 ( s3 ) linearity : : for arbitrary constants @xmath12 , @xmath13 ( s4 ) consistency with classical definition : : @xmath14 ( s5 ) monomials : : for every @xmath15 , the mapping @xmath16 is holomorphic in @xmath9 .
@xmath17 right shift continuity : : if @xmath18 pointwise for every @xmath19 , then @xmath20 more generally , if there is a sequence of polynomials @xmath21 of fixed degree such that , as @xmath22 , @xmath23 for all @xmath19 , we require that @xmath24 the first four axioms ( s1)(s4 ) are so obvious that it is hard to imagine any summation theory that violates these . they easily imply @xmath25 for every @xmath26 , so we are being consistent with the classical definition of summation .
axiom ( s5 ) is motivated by the well - known formulas @xmath27 and similarly for higher powers ; we shall show below that our axioms imply that all those formulas remain valid for arbitrary @xmath28 .
finally , axiom @xmath17 is a natural condition also .
the first case , in ( [ eqs5 ] ) , expresses the view that if @xmath7 tends to zero , then the summation `` on the bounded domain '' @xmath29 $ ] should do the same . in ( [ eqs5b ] ) ,
the same holds , except an approximating polynomial is added ; compare the discussion after proposition [ prop1 ] .
it will turn out that for a large class of functions @xmath7 , there is a unique way to define a sum @xmath30 with @xmath19 that respects all these axioms . in the next section
, we will derive this definition and denote such sums by @xmath31 .
we call them `` fractional sums . '' to see how these conditions determine a summation method uniquely , we start by summing up polynomials .
the simplest such case is the sum @xmath32 with @xmath33 constant .
if axiom ( s1 ) is respected , then @xmath34 applying axioms ( s2 ) on the left and ( s4 ) on the right - hand side , one gets @xmath35 it follows that @xmath36 .
this simple calculation can be extended to cover every sum of polynomials with a rational number of terms .
[ prop1 ] for any polynomial @xmath37 , let @xmath38 be the unique polynomial with @xmath39 and @xmath40 for all @xmath19 . then : * the possible definition @xmath41 satisfies all axioms ( s1 ) to @xmath17 for the case that @xmath7 is a polynomial . *
conversely , every summation theory that satisfies axioms ( s1 ) , ( s2 ) , ( s3 ) , and ( s4 ) also satisfies ( [ eqsumpoly ] ) for every polynomial @xmath42 and all @xmath43 with rational difference @xmath44 .
* every summation theory that satisfies ( s1 ) , ( s2 ) , ( s3 ) , ( s4 ) , and ( s5 ) also satisfies ( [ eqsumpoly ] ) for every polynomial @xmath42 and all @xmath43 . to prove the first statement ,
suppose we use ( [ eqsumpoly ] ) as a definition .
it is trivial to check that this definition satisfies ( s1 ) , ( s3 ) , ( s4 ) , and ( s5 ) . to see that it also satisfies ( s2 ) ,
consider a polynomial @xmath42 and the unique corresponding polynomial @xmath45 with @xmath46 and @xmath39 .
define @xmath47 and @xmath48 .
then @xmath49 , and @xmath50 .
hence @xmath51 to see that ( [ eqsumpoly ] ) also satisfies @xmath17 , let @xmath52 be the linear space of complex polynomials of degree less than or equal to @xmath53 .
the definition @xmath54 for @xmath55 introduces a norm on @xmath52 .
if we define a linear operator @xmath56 via @xmath57 , then this operator is bounded since @xmath58 .
thus , if @xmath59 is a sequence of polynomials with @xmath60 , we have @xmath61 .
axiom @xmath17 then follows from considering the sequence of polynomials @xmath62 with @xmath63 and noting that pointwise convergence to zero implies convergence to zero in the norm @xmath64 of @xmath65 , and thus of @xmath66 . to prove the second statement , we extend the idea that we used above to show that @xmath67 . using ( s1 ) , we write for an integer @xmath68 @xmath69 where the left - hand side has a classical interpretation , using ( s1 ) , ( s2 ) , and ( s4 ) . rewriting the right - hand side according to ( s2 ) and using ( s3 ) ,
we get @xmath70 where the @xmath71 are polynomials of degree @xmath72 ( and all @xmath73 ) .
now we argue by induction .
if @xmath74 , the previous equation clearly determines @xmath75 and by linearity also the corresponding sum over arbitrary constants @xmath33 .
once the value of the sum of any polynomial of degree @xmath72 is determined , the equality also determines the value of @xmath76 , and by linearity , the sum of every polynomial @xmath42 of degree @xmath77 . using ( s2 ) again
, we see that the axioms ( s1 ) to ( s4 ) uniquely determine the sum @xmath78 if @xmath44 .
as we have seen , equation ( [ eqsumpoly ] ) is a possible definition satisfying those axioms ; hence it is the only possible definition for @xmath44 .
finally , it is clear how the restriction @xmath44 can be lifted by additionally assuming ( s5 ) ( equation ( [ eqsumpoly ] ) is already satisfied for @xmath79 by requiring just continuity in ( s5 ) ; holomorphy is required for @xmath80 ) .
consider now an arbitrary function @xmath81 .
if we are interested in @xmath82 for complex @xmath43 , we can write @xmath83 where @xmath26 is an arbitrary natural number .
hence , @xmath84 what have we achieved by this elementary rearrangement ? in the last line , the first sum on the right - hand side involves an integer number of terms , so this can be evaluated classically .
all the problems sit in the last sum on the right - hand side .
the payoff is that we have translated the domain of summation by @xmath85 to the right .
since ( [ eqheuristics ] ) holds for every integer @xmath85 , we can use @xmath17 to evaluate the limit as @xmath86 : if @xmath87 as @xmath86 for all @xmath5 , then @xmath17 implies that the limit as @xmath86 of the last sum should vanish .
we get @xmath88 this is of course a special condition to impose on @xmath7 , but the same idea can be generalized .
for example , if @xmath89 , then for @xmath90\subset{\mathbb{r}}^+$ ] , the values @xmath91 are approximated well by the constant function @xmath92 , with an error that tends to @xmath93 as @xmath86 : we say that @xmath94 is `` approximately constant . '' using ( s3 ) , @xmath95 for every @xmath26 .
but by @xmath17 , the last sum vanishes as @xmath86 , while the first sum on the right - hand side has a constant summand and is evaluated using proposition [ prop1 ] . taking the limit @xmath86 in ( [ eqheuristics ] )
, it follows by necessity that @xmath96 before generalizing our definition further , we take courage by observing that this interpolates the factorial function in the classical way : we define @xmath97 and thus get @xmath98 using a well - known product representation of the @xmath2 function @xcite .
it is now straightforward to use the heuristic calculation in ( [ eqheuristics ] ) together with proposition [ prop1 ] and axiom @xmath17 to derive a general definition : all we need is that the value of @xmath99 can be approximated by some sequence of polynomials @xmath100 of fixed degree for @xmath86 .
some care is needed with the domains of definition : the example of the logarithm shows that it is inconvenient to restrict to functions which are defined on all of @xmath9 .
all we need is a domain of definition @xmath101 with the property that @xmath102 implies @xmath103 .
this leads to the following ( using the convention that the zero polynomial is the unique polynomial of degree @xmath104 ) .
[ deffracsummable ] let @xmath105 and @xmath106 .
a function @xmath107 will be called _ fractional summable of degree @xmath108 _ if the following conditions are satisfied : * @xmath109 for all @xmath110 ; * there exists a sequence of polynomials @xmath21 of fixed degree @xmath108 such that for all @xmath110 @xmath111 * for every @xmath112 , the limit @xmath113 exists , where @xmath114 is defined as in ( [ eqsumpoly ] ) . in this case
, we will use the notation @xmath115 for this limit .
moreover , we can define fractional products by @xmath116 whenever @xmath117 is fractional summable .
note that this definition does not depend on the choice of the approximating polynomials @xmath21 : if @xmath118 is another choice of approximating polynomials , then @xmath119 for all @xmath110 , and hence for all @xmath120 since the set of polynomials of degree at most @xmath108 is a finite - dimensional linear space .
as shown in proposition [ prop1 ] , sums of polynomials satisfy axiom @xmath17 . substituting @xmath93 for @xmath7 and @xmath121 for @xmath122 in @xmath17
proves that @xmath123 .
moreover , this definition is the unique definition that satisfies axioms ( s1 ) to @xmath17 : definition [ deffracsummable ] satisfies all the axioms ( s1 ) to @xmath17 ( for suitable domains of definition ) , and it is the unique definition with this property ( for the class of functions that we are considering ) .
we have already proved uniqueness above , by deriving definition [ deffracsummable ] from the axioms ( s1 ) to @xmath17 .
it remains to prove that this definition indeed satisfies all the axioms .
clearly , ( s3 ) and ( s5 ) are automatically satisfied . substituting the definition into ( s1 ) , ( s2 ) , and ( s4 ) , these axioms can be confirmed by a few lines of direct calculation .
to prove @xmath17 , we use the definition and the other axioms ( in particular continued summation ( s1 ) ) and calculate @xmath124 this proves that definition [ deffracsummable ] satisfies all the axioms .
now that we have a definition of sums with noninteger numbers of terms , it is interesting to find out how many of the properties of classical finite sums remain valid in this more general setting , and what new properties arise that are not visible in the classical case .
one of the most basic identities for finite sums is the geometric series . for simplicity ,
let @xmath125 .
then the function @xmath126 is approximately zero ( we have @xmath127 for every @xmath19 ) , and the definition reads @xmath128 thus , the formula for the geometric series remains valid for every @xmath120 .
a similar calculation shows that the binomial series remains valid in the fractional case : for every @xmath129 and @xmath120 with @xmath130 , we have @xmath131 there are generalizations of ( [ eqgeometric ] ) to the case @xmath132 and of ( [ eqbinomial ] ) to the case @xmath133 : these involve a `` left sum '' as introduced in section [ secleftsummation ] .
an example of a summation identity with more complicated structure is given by the series multiplication formula @xmath134 for every @xmath120 , given that all the three fractional sums exist ( see ( * ? ? ?
* lemma 7 ) ; it generalizes the formula @xmath135 , and similarly for all positive integers @xmath3 .
as shown in section [ secfromaxtodef ] , our definition interpolates the factorial by the @xmath2 function , @xmath136 an amusing consequence is @xmath137 this is because @xmath138 using @xmath139 .
many basic fractional sums are related to special functions . as a first example , consider the harmonic series . since @xmath140 is approximately zero , the definition reads @xmath141 and in particular @xmath142 which was noticed already by euler ( * ? ? ? * , 19 ) . for general @xmath3
, the harmonic series can be expressed in terms of the so - called digamma function @xcite @xmath143 and the euler - mascheroni constant @xmath144 : one obtains @xcite @xmath145 note that the reflection formula @xcite for the digamma function becomes @xmath146 as a further generalization , it is convenient to consider the hurwitz @xmath147 function , traditionally defined by the series @xmath148 by analytic continuation , @xmath149 can be defined for every @xmath150 , except for a pole at @xmath151 . for @xmath152 ,
the hurwitz @xmath147 function equals the well - known riemann @xmath147 function : @xmath153 it turns out that the hurwitz @xmath147 function can be understood as a fractional power sum
. it can be shown ( * ? ? ?
* corollary 14 ) that for every @xmath154 and for all @xmath155 , @xmath156 a useful special case is @xmath157 note that such equations give in many cases intuitive ways to compute properties and special values of special functions .
everybody knows the formula @xmath158 , so @xmath159 , and thus by ( [ eqzeta12 ] ) @xmath160 it follows that @xmath161 .
similarly , we have @xmath162 ( in the second equality , we interchanged differentiation and fractional summation ; it is not hard to check that this is indeed allowed ) . since @xmath163
, this easily implies that @xmath164 .
similarly , differentiating ( [ eqhurwitz ] ) @xmath165 times with respect to @xmath166 and arguing as before ( compare also ( * ? ? ?
* sec . 6 ) ) , we obtain @xmath167 there are some classically unexpected special values like @xmath168 there is an identity for classical sums which is almost never mentioned , because it seems so trivial .
consider the sum @xmath169 obviously , there are two formally correct possibilities to write this sum , @xmath170 classically , it is clear that @xmath171 .
does this carry over to the fractional case ?
there is a fundamental problem : our definition of fractional sums involves @xmath172 , i.e. , @xmath7 is evaluated near @xmath173 , and when @xmath174 is replaced by @xmath175 then @xmath7 would be evaluated near @xmath104 where the values may be unrelated .
this will be discussed in the next section .
looking back at the axioms given in section [ sectheaxioms ] , there is one axiom that could possibly be modified : in @xmath17 , limits as @xmath22 are considered , but one could equally well look at limits as @xmath176 .
this way , one obtains an axiom of `` left shift continuity '' : @xmath177 left shift continuity : : if @xmath178 pointwise for @xmath19 , then @xmath179 more generally , if there is a sequence of polynomials @xmath21 of fixed degree such that @xmath180 for @xmath19 as @xmath86 , we require that @xmath181 repeating the calculations of section [ secfromaxtodef ] , one gets an alternative definition ) can determine a definition uniquely : only adding or subtracting @xmath26 to the upper summation boundary consists of adding or subtracting @xmath85 terms to the series , which can be done classically . ] which we do not state here formally : it is exactly the same as definition [ deffracsummable ] , except that in every limit , @xmath86 is replaced by @xmath176 .
it can be shown that this definition is the unique one that satisfies axioms ( s1 ) , ( s2 ) , ( s3 ) , ( s4 ) , ( s5 ) , and @xmath177 .
note that in general , the existence of @xmath182 and @xmath183 are independent , and if both left and right fractional sums exist , they may have different values .
for example , for every @xmath5 with @xmath184 , we have @xmath185 in contrast to equation ( [ eqzeta12 ] ) .
what we do have is the obvious relation @xmath186
fractional sums are not simply a new world with results that have no meaning in the classical context ; they allow us to derive identities that can be stated entirely in classical terms .
some of these formulas are known and some seem to be new .
of course , all these identities can in principle be computed without fractional sums . but proving them with the help of fractional sums is rather intuitive and simple , since most of the steps use fractional generalizations of basic , very well - known classical summation properties . as a first example
, we show how to compute a closed - form expression for the infinite product @xmath187 for @xmath188 .
it was first considered by borwein and dykshoorn in 1993 ( see @xcite ) . by taking logarithms
, one gets @xmath189 consider the function @xmath190 that tends to @xmath191 as @xmath192 . according to definition [ deffracsummable ] , we have @xmath193.\end{aligned}\ ] ] thus , we get @xmath194 where we have used equation ( [ eqvlnv ] ) , index shifting , equation ( [ eqhurwitzdiff ] ) , equation ( [ eqgamma ] ) , and continued summation . by exponentiating , we finally get @xmath195 ) .
the right - hand side corresponds to the lowest curve , while the other three curves ( from top to bottom ) are plots of approximations to @xmath196 ( finite products as in ( [ eq : defpx ] ) ) for @xmath197 , @xmath198 , and @xmath199 respectively . , scaledwidth=60.0% ] by application of this method , a large class of infinite products can be explicitly computed , which seems to include the class of products considered in @xcite .
here is an example of a new identity : using the same steps as in the calculation above , one easily proves that for @xmath200 , @xmath201 resolving the definition and exponentiating , we get the following classical limit identity : @xmath202\nonumber\\ & = & 2^{-\frac 1 4 \ln 2 + 4 \zeta'\left(-1,\frac{x+1 } 2\right)-4\zeta'\left(-1,\frac x 2 + 1\right ) } \left ( \frac{\gamma\left(\frac{x+1 } 2 \right ) } { \gamma\left(\frac x 2 + 1\right ) } \right)^{-2x\ln 2}\times\nonumber\\ & & \times\quad e^ { 2 \zeta''\left(-1,\frac x 2 + 1\right)-2\zeta''\left(-1,\frac{x+1 } 2\right ) + x\left ( \zeta''\left(0,\frac { x+1 } 2\right)-\zeta''\left(0,\frac x 2 + 1\right ) \right ) } .\label{eqcomplicated}\end{aligned}\ ] ] again , we have used mathematica for a quick numerical check that is shown in figure [ fig : secondzeta ] . ) .
the right - hand side corresponds to the uppermost curve , while the other three curves ( from bottom to top ) are plots of the left - hand side for @xmath198 , @xmath203 , and @xmath204 respectively .
, scaledwidth=60.0% ] in this section , we consider the multiple gamma function @xmath205 , a generalization of the classical gamma function @xmath2 , defined for @xmath26 and @xmath19 by the recurrence formula ( compare @xcite ) @xmath206 these equations do not determine the functions @xmath205 uniquely , so one needs the additional bohr - mollerup - like condition that @xmath207 is positive and @xmath85 times differentiable on @xmath208 , and that @xmath209 is increasing ( see @xcite ) . for @xmath197 ,
this definition reproduces the classical gamma function : @xmath210 .
the ( reciprocal of the ) special case @xmath211 is known as the barnes @xmath212 function @xmath213 by ( [ eqdefgamman ] ) , it satisfies @xmath214 more generally , @xmath215 so @xmath216 is the reciprocal of the product of @xmath205 , which means that @xmath217 is something like an @xmath85-fold product of the first @xmath218 natural numbers : @xmath219 while this equation only makes sense for @xmath220 , one can easily show that the definition of @xmath217 for @xmath19 is compatible with our definition for fractional sums and products , i.e. , that for every @xmath19 ( except for poles ) and @xmath221 , we have @xmath222 we will now show some properties of the multiple gamma function @xmath205 , specifically for the example @xmath211 , simply by using basic fractional sum identities , without using any special function properties of @xmath205 . by the multiplication formula ( [ eqseriesmultiplication ] ) , we have @xmath223 the last equality follows from equation ( [ eqhurwitzdiff ] ) .
thus , we have found an explicit formula for @xmath224 in terms of derivatives of the hurwitz @xmath147 function .
equation ( [ eqvlnv ] ) gives the special value @xmath225 these are of course very well - known results , but the calculations are strikingly simple .
moreover , this example shows that there is a wide variety of interesting `` special functions '' that do not have to be defined separately , but can be treated in a unified manner by our theory of fractional sums .
new generalizations comparable to @xmath226 include @xmath227 @xmath228 here , @xmath229 , @xmath230 , and @xmath231 are the euler - mascheroni , stieltjes , and catalan constants , respectively .
again , these formulas have classical limit representations looking like equation ( [ eqcomplicated ] ) which we do not write down here explicitly . the paper `` on some strange summation formulas '' @xcite contains some formulas like ( [ eqgosper ] ) below .
there might possibly be very short proofs for all these identities using fractional sums .
the only problem is that there is one single step ( indicated by the question mark ) which we are unable to justify : it is basically an interchange of a fractional sum and an infinite series .
we will now use the basic identity @xmath235 for every @xmath120 and @xmath26 , which can be shown in two different ways : the first possibility is to see that for every @xmath236 , @xmath237 since even and odd terms cancel each other . by continued summation and proposition [ prop1 ]
, @xmath238 is a polynomial in @xmath3 , so equation ( [ eqsumxminusx ] ) must be valid for every @xmath120 .
going back to the fractional sum in ( [ eqsb ] ) , the odd function @xmath240 is holomorphic in the entire complex plane , except for a pole at @xmath241 , so we can develop it into a power series .
we get @xmath242 the next step is critical : we apply the fractional sum term - by - term .
unfortunately , it is not clear that this manipulation is justified .
@xmath243 equations ( [ eqreflection ] ) and ( [ eqsumxminusx ] ) yield @xmath244 this method only works for a certain class of functions which obviously contains @xmath92 from ( [ eqfn ] ) and other functions like @xmath245 , but which does _ not _ contain other simple functions like @xmath246 .
it is an open question to give sufficient conditions for the validity of this method , i.e. , for justification of termwise fractional summation as in equation ( [ eqtermwise ] ) .
we would like to thank several colleagues from the community of `` special functions and exotic identities '' for their encouragement and support , especially richard askey and mourad ismail .
moreover , we are grateful to otto forster , irwin kra , armin leutbecher , john milnor , as well as to the seminar audiences in stony brook and mnchen for encouragement , interest , and helpful discussions .
we would also like to thank the referees for helping us improve the exposition .
l. euler , dilucidationes in capita postrema calculi mei differentialis de functionibus inexplicabilibus [ ] , _ memoires de lacademie des sciences de st .- petersbourg * 4 * _ ( 1813 ) , 88119 ; reprinted in _
opera omnia , series 1 _ , * 16 * , 133 ; also available at http://www.math.dartmouth.edu/@xmath247euler/. * markus mller * received his dr .
nat . from technical university of berlin in 2007 , where he is currently working as a postdoc in quantum information theory . after playing around with fractional sums as a high - school student
, he was lured away from math to physics by popular scientific articles on schrdinger s cat and quantum weirdness . since then
, his main motivation has been the idea that the notion of information opens up an unexpected , fresh perspective on foundational problems of physics .
this line of thought led him to work on quantum turing machines and kolmogorov complexity , concentration of measure , and generalized probabilistic theories beyond quantum theory . at the moment
, he is preparing for some tough postdoc years abroad by skiing with friends and enjoying the alternative music occasions in berlin .
* dierk schleicher * studied physics and computer science in hamburg and obtained his ph.d .
in mathematics at cornell university .
he enjoyed longer educational and research visits in princeton , berkeley , stony brook , paris , and toronto .
after many years in mnchen , he became the first professor at the newly - founded jacobs university bremen in 2001 and built up the mathematics program there .
his main research area is dynamical systems , especially complex dynamics : `` real mathematics is difficult , complex mathematics is beautiful . ''
he has always been active in math circles and special programs for talented high school students , which is what lured him away from physics to mathematics .
he was one of the main organizers of the 50th international mathematical olympiad ( imo ) in bremen / germany , in 2009 .
he enjoys outdoor activities : kayaking , paragliding , mountain hiking , and more . | starting from a small number of well - motivated axioms , we derive a unique definition of sums with a noninteger number of addends .
these `` fractional sums '' have properties that generalize well - known classical sum identities in a natural way .
we illustrate how fractional sums can be used to derive infinite sum and special functions identities ; the corresponding proofs turn out to be particularly simple and intuitive .
`` god made the integers ; all else is the work of man . ''
leopold kronecker |
the star formation history ( sfh ) of the universe deduced from optical / uv surveys ( lilly et al .
1996 ; madau et al . 1996 , 1998 ) has recently been revised to account for dust extinction ( steidel et al .
1999 ) , largely motivated by the detection of a population of dusty star - forming objects in deep submm surveys with scuba ( hughes et al . 1998 ; eales et al . 1999 ; smail , ivison & blain 1997 ; scott et al . 2002 ) .
the current best estimate of the sfh is that the global star - formation rate in the universe was fairly constant over the redshift interval @xmath2 , falling by a factor @xmath3 to the present epoch , see fig .
[ sfrdf ] .
( steidel et al . 1999 ; ivison et al . 2002 ) .
a simple integration of this history implies that @xmath4% of all stars and metals today should have formed by a redshift of @xmath5 .
however , the metal content of objects believed to contain most of the baryonic matter at high - z principally the lyman alpha forest and damped lyman alpha absorption systems ( dlas) is only 2.5% of the total amount of metals in the universe today ( pettini 1999 ; pagel 2002 ) .
thus 90% of metals expected to exist at @xmath6 have not yet been seen or recognised
the ` missing metals ' .
the submm surveys revealed a population of extremely dusty and obscured objects with very high bolometric luminosities , probably residing at high redshifts , @xmath7 ( dunlop 2001a ; ivison et al .
if the dust is heated by stars , the star formation rates implied by the submm fluxes are @xmath8 , high enough to produce an entire galaxy worth of stars in @xmath9 gyr . since the properties of low - redshift ellipticals suggest that most of their stars formed quickly at high redshift ( bower et al .
1992 ; renzini 1997 ; jimenez et al .
1999 ) , several authors have argued that the scuba sources may be elliptical galaxies in the process of formation ( lilly et al .
1999 ; eales et al . 2000 ; dunlop 2001b ; granato et al .
it has been claimed ( cen & ostriker 1999 ) that the regions of greatest over - density in the universe at any epoch are the most metal enriched , and that these regions will have reached a given metallicity much earlier than lower density environments .
this argument suggests that scuba galaxies should be a large reservoir for metals at high redshift , since if they are proto - ellipticals they should have formed from the highest density peaks at high redshift when there has been less time for pollution of the low density inter - galactic medium ( igm ) . in this paper
, we wish to investigate the connection between the ` missing metals ' , the formation of the spheroids and the dusty sources seen by scuba .
however , instead of using the bolometric luminosities of the submm sources to infer their physical properties , we interpret the results of the deep submm surveys and our submm survey of the nearby universe in a completely different way , by carrying out the first investigation into the cosmic evolution of the dust mass in galaxies .
this provides a new way of looking at the chemical evolution of the universe , as there is evidence that the fraction of ism metals which are bound up in dust grains is @xmath10 percent , a value which seems remarkably constant from galaxy to galaxy and which also appears to apply at higher redshifts ( issa , maclaren & wolfendale 1990 ; edmunds 2001 ; pei , fall & hauser 1999 ; james et al .
we will determine the dust content of the low and high - z universe by constructing dust mass functions , i.e. the space densities of galaxies as a function of dust mass . at low - z
, we will use the results of the scuba local universe galaxy survey ( dunne et al .
2000 ) , the first attempt at an unbiased submm survey of the local universe . for the high redshift dust mass function
, we will use submm data from the deep scuba submm surveys . in section 4
we will use these dust mass functions , together with a simple chemical evolution model , to estimate the co - moving density of metals and baryons associated with the ism of galaxies at low and high redshift .
finally , we will produce a simple prediction for the dust mass function of the spheroids during the epoch in which they had their maximum dust masses , and compare with our observational estimate of the dust mass function of the scuba sources .
dust mass can be calculated from the submm flux according to @xmath11 where @xmath12 is the distance for the assumed cosmology and @xmath13 is the observed flux .
the dust mass opacity coefficient @xmath14 normalises the submm emission for a given dust mass and is currently measured only at fir wavelengths and then extrapolated to the submm with a @xmath15 dependency .
we have taken the average value from the following sources of @xmath16 , where the @xmath17 is the error on the mean value ( hildebrand 1983 ; draine & lee 1984 ; krgel , steppe & chini 1990 ; sodroski et al .
1997 ; bianchi , davies & alton 1999 ) .
we use @xmath18 ( for which there is compelling evidence from our local survey and elsewhere , dunne & eales 2001 and references within ) to extrapolate @xmath19 to the rest wavelength of the submm emission ( @xmath20 m for an 850@xmath21 m source at @xmath22 ) .
this agrees with the recent estimation of @xmath23 using a different technique ( james et al .
2002 ) .
we also require the value of @xmath24 , which is a mass - weighted dust temperature .
submillimetre and iso observations of many nearby galaxies have shown that , when enough data is present , the sed is best described by dust at more than one temperature ( haas et al .
1998,2000 ; frayer et al .
1999 ; braine et al .
1997 ; dumke et al .
1997 , calzetti et al .
2000 ; gulin et al . 1995
; sievers et al .
1994 ; neininger et al . 1996
; alton et al . 1998 , 2001 ; papadopoulos & seaquist 1999 ; reach et al . 1995 ; trewhella et al .
2000 ) . for our local scuba survey of bright iras galaxies
, we found that in addition to the typical ` iras - detected ' dust at @xmath25 k , a colder dust component at @xmath26 k was required in all cases ( dunne & eales 2001 ) , even for highly luminous galaxies such as arp 220 .
klaas et al .
( 2001 ) also find that the seds of 41 local ulirgs are best represented by a combination of warm and cold dust .
following the notation of the two - component sed model from dunne & eales ( 2001 ) , the value which should enter eqn .
[ mduste ] can be approximated by @xmath27 where @xmath28 is the characteristic temperature of the cold / warm dust , and @xmath29 is the ratio of mass in the cold / warm components . for the local survey ,
the value of @xmath24 was 21.3 k , with a standard error in the mean for the population of 0.5 k. this mass - weighted temperature should not be confused with the dust temperature traditionally used to estimate bolometric luminosity , which should be the temperature defining the peak in the spectrum . since we are not integrating the sed and have observations at wavelengths longer than the peak , our estimation of the dust mass from a single submm flux is not as sensitive to temperature as an estimate of the bolometric luminosity would be . in local galaxies ,
including ulirgs , most dust is contained in the cold @xmath30 k component , which means we do not need to worry about the location of the peak in the sed ; the temperature of the warm dust has a negligible effect on the dust mass estimate , providing @xmath31 . for the local objects , we use @xmath24 for each galaxy in our survey ( dunne & eales 2001 ) .
the biggest uncertainty in our analysis is our lack of knowledge of @xmath24 in the high redshift scuba galaxies .
as discussed above , there is now a lot of evidence that most of the dust in low redshift fir sources , including ulirgs , is at a relatively cold temperature .
it will not be possible to perform a similar analysis of the temperature of dust in high redshift galaxies because of the lack of instruments operating at wavelengths long enough to sample the rest - frame emission from the cold dust ( i.e. 850@xmath21 m in the _ rest - frame _ ) .
it is difficult to determine the existence of a colder dust component without many fluxes at wavelengths at and beyond the peak in the sed .
it is plausible that the mean temperature of dust in galaxies was higher in the past , as star - formation rates and possibly interstellar radiation fields were higher .
however , it is also possible that the mass - weighted dust temperatures in high - z scuba galaxies might be as low , or even lower than in low - z ulirgs .
this is because dust equilibrium temperatures are very sensitive to the relative geometry of dust and heating sources , as well as optical depth effects .
if the clouds in which the stars form are very dense , it is likely that few uv / optical photons escape ( we know this is the case for the high redshift scuba galaxies as a whole ) , and so dust outside these regions will see a much weaker radiation field and hence have a lower temperature .
there is also evidence ( frayer 2001 ; ivison et al .
2001,2002 ; isaak et al .
2002 ; lutz et al .
2001 , eales et al .
2000 ) that the physical sizes of the star forming regions in some submm luminous objects at high redshift are larger than in local ulirgs .
local ulirgs have very compact regions of star formation ( @xmath32 kpc ) while the radio , mm and co sizes of the high - z submm sources have been estimated to be as large as 1030 kpc in some cases .
this could be due to the high - z objects having much greater gas fractions than local ulirgs .
mihos & hernquist ( 1996 ) have argued that it is the presence of large stellar bulges in local ulirgs which provide enough stability to allow the gas to reach very high central concentrations before forming stars .
galaxies at high redshift may have greater gas fractions and small or non - existent stellar bulges compared to galaxies today .
this means that the star formation induced by an interaction or perturbation at high redshift could be much more widespread , as the gas will collapse to form stars before it inflows to the centre of the remnant ( mihos 1999 ; tissera et al .
2002 ) . a given star formation rate which is spread over a larger volume will result in a lower intensity radiation field , possibly leading to lower dust temperatures
two sources at @xmath33 from the 170@xmath21 m firback survey have also been followed up with scuba and appear to have much colder dust for their luminosity than local objects selected by iras ( chapman et al .
such sources may be more representative of objects selected in deep submm surveys , as selection at longer wavelengths is less biased toward objects dominated by warmer dust .
the models of pei , fall & hauser ( 1999 ) however , predict that there should have been a modest amount of positive evolution in the dust equilibrium temperature between @xmath34 and @xmath6 . in the absence of any direct evidence on the amount of cold dust in the high - z scuba galaxies
, we have assumed the equilibrium temperature predicted for dust sources at @xmath35 by pei , fall & hauser ( 1999 ) of 25 k. if temperatures are more akin to local galaxies ( 20 k ) then the high - z dust masses we present will increase by a factor @xmath36 , and more if the equilibrium temperatures at high redshift are actually lower than in local ulirgs .
we estimate the space density of galaxies as a function of dust mass ( the dust mass function dmf ) using standard accessible volume techniques , such that @xmath37 where @xmath38 is the accessible volume in the parent sample of the @xmath39th source and the sum is over all objects in the range @xmath40 . for the local galaxies
@xmath38 is their accessible volume at 60@xmath21 m , which was the wavelength of selection of their parent sample , the iras bgs ( soifer et al .
the local dust mass function ( dmf ) we will use here is described in detail in dunne & eales ( 2001 ) . for the high redshift dust sources , we construct the dmf using submm data from the cudss blank field submm survey ( eales et al .
2000 ; webb et al . 2002 ) , and also the submm number counts at higher and lower flux densities ( smail et al .
2002 ; scott et al . 2002 ; borys et al .
2002 ) . in order to allow for star forming objects which may be too submm faint to appear in the blank field surveys , we use the statistical submm analysis of starburst galaxies in the hdf by peacock et al .
the characteristic shape of the thermal dust spectrum means that in a flat cosmology , an object of a given dust mass will be observed to have virtually the same 850@xmath21 m flux - density over a wide range of redshift ( @xmath41 ) .
therefore the redshift we assume for the sources has very little effect on the derived dust masses , providing that the objects are at @xmath42 ( any objects known to have @xmath43 were excluded from this analysis ) .
spectroscopic redshifts ( eales et al . 2000 ) and estimates using the dunne , clements & eales ( 2000 ) version of the radio - submm redshift indicator ( carilli & yun 1999 ) were used where available , otherwise sources were placed at @xmath44 for the purposes of calculating @xmath45 .
figure [ fluxzf ] shows that due to the flat flux - density redshift relation , a source visible at @xmath46 will still be visible ( in a flux limited sample ) at @xmath47 ( depending on the exact shape of the spectrum ) ; therefore @xmath38 for the submm sources was simply taken to be the volume in the redshift interval @xmath48 .
we fix @xmath49 as there is dramatic evolution in the submm population below this redshift , very few submm sources being found at @xmath43 ( dunlop 2001a ) , while we take @xmath50 .
this is the redshift at which the 850@xmath21 m flux density of a source with an sed similar to arp 220 would start to fall off ( fig .
[ fluxzf ] ) .
the effects of using @xmath51 or @xmath52 are a modest @xmath53% change in the space density .
we have treated the statistical detection of submm flux in the hdf in a similar manner to the blank field sources , using @xmath54 , the average of the photometric redshifts , to estimate the dust mass .
the photometric redshifts show that the galaxies responsible for the submm emission are found between @xmath55 and this redshift range was used to calculate the space density .
there were also 5 optical sources in the hdf which are plausible @xmath56 detections in the submm , these are treated separately and we have used their photometric redshifts to estimate the dust mass . this high - z dust mass function is shown in fig .
[ dmff ] for two cosmologies , @xmath57 ( solid ) and a @xmath58 model with @xmath59 ( blue dashed ) , along with the dmf from our local survey .
a value of @xmath60 is used throughout .
it is immediately obvious that there has been substantial evolution in the dust mass function , particularly at the highest dust masses .
there is growing concern that the metal content of the high - z universe is not observed to be as large as the inferred star formation would suggest ( pettini 1999 ; pagel 2002 ; blain et al .
1999a , b ) .
accounting for the metals observed in dlas and the lyman forest leaves a shortfall of @xmath61% in the cosmic metal budget at @xmath6 .
metals are thought to be produced by supernovae and in the winds of cool evolved stars .
they are then injected into the ism with a fraction of them ( @xmath62 ) bound into , or subsequently accreting onto dust grains . from here
they can be expelled out of the galaxy into the inter - galactic medium ( igm ) by starburst winds , or become part of the next generation of stars .
the fraction of metals found in these different locations today is summarised in fig .
[ histf]a .
these fractions do not remain constant , of course , as the fraction of total metals in stars and the igm will decrease with redshift .
today the fraction of metals in the ism of galaxies is negligible , but how different could this have been in the past , when galaxies were more gas rich ?
it is clear that more metals should be found in the most evolved objects at a given redshift , although their distribution between stars and ism will depend on how far the object has evolved .
this has been presented by cen & ostriker ( 1999 ) as a correlation between over - density and metallicity , the fastest evolving objects being found preferentially in the most over - dense regions .
the monolithic collapse models predict massive episodes of star formation at high redshift , and this should be accompanied by rapid enrichment of the locality .
observations of dlas and the igm at high redshift have failed to find evidence for this level of enrichment , although high - z qsos are often found to have near solar or super - solar metallicities ( warner et al .
this makes sense if qsos represent the accretion phase of the massive black hole , which is presumably forming co - evally with the stellar bulge , as required to produce the bh - bulge mass correlation in the local universe ( magorrian et al .
the igm is a low - density environment and so not expected to be very enriched at high - z .
it has been argued that the damped lyman alpha absorbers represent the co - moving mass density of gas in galaxies at high redshift and that the column density - weighted metallicity of these systems traces the global metallicity evolution of the universe ( pei & fall 1995 ) . however , there are potential biases in the selection of dlas as they are only observed along sight lines to optically bright qsos and thus they are less likely to be observed if they contain dense gas and dust .
the cross - sectional area available at the edges of galaxies is also greater than the denser , more evolved inner regions and therefore a sight - line is more likely to intercept a less dense and less metal enriched part of a galaxy ( mathlin et al .
thus current samples of dlas may underestimate the metal content of the neutral gas at any redshift .
the obvious place to look for the most enriched systems at high - z is to look at the most obscured and actively star forming objects submm sources , especially as they show evidence for being located in over - dense regions ( ivison et al . 2000 ) .
one simple thing we can do with the high - z dmf is to integrate the total amount of dust in galaxies at low and high redshift .
the co - moving density of dust is given by @xmath63 , where @xmath64 and @xmath65 are the schechter parameters fitted to the dust mass functions in fig .
@xmath66 can be converted into @xmath67 , the co - moving density of metals , by dividing by @xmath62 the fraction of ism metals which are in dust .
we have used @xmath68 and there is some evidence that this fraction is fairly constant both locally and at higher redshifts ( pei , fall & hauser 1999 ; issa , maclaren & wolfendale 1990 ; edmunds 2001 ; james et al .
table [ metalst ] lists the ism metals at low and high - z inferred from the dust , along with an inventory of metals and baryons in other locations .
integrating the dmf is currently quite uncertain at low redshift due to the very steep faint end slope fitted to the local dmf .
the minimum and maximum values in table [ metalst ] are obtained by integrating the local dmf down to the lowest observed mass , and to zero mass respectively . to facilitate comparisons with work from the literature ,
we now restrict ourselves to the @xmath69 cosmology , noting that this does not affect any of our conclusions .
lccccc + + & & & + & & & & & + + stars & @xmath70 & @xmath71 & & @xmath72 & @xmath73 + + & & @xmath74 sfh & & @xmath75 & see text + + ism & @xmath76 & @xmath77 & & @xmath78 & @xmath79 + + & @xmath80 & dmf & model & & & @xmath81 k + + dlas & & @xmath82 & & & @xmath83 + + & & ism / igm ? & & & @xmath84 + + igm observed & @xmath85 & @xmath86 & & @xmath87 & @xmath88 + + & cluster gas & ly@xmath89 forest & & @xmath90 & @xmath91 + + igm+ & @xmath92 & @xmath93 & & @xmath94 & @xmath95 + + & group / field gas & extra inferred & & @xmath96 & @xmath91 + & & by cdm & & & + + + total observed & @xmath97 & @xmath98 & & @xmath99 & @xmath100 + + inferred & @xmath101 & @xmath102 & & @xmath103 & @xmath104 + + + table adapted from pettini ( 1999 ) and pagel ( 2002 ) with @xmath105 and @xmath106 for consistency with previous work .
we assume @xmath107 .
the metal content has been estimated from either the baryon content and the metallicity , or from the dust masses .
the ` igm observed ' represents the directly observed gas ( the hot cluster gas at low - z , and the ly@xmath89 clouds at high - z ) .
the ` igm+ ' is the additional igm gas which has been inferred but not directly observed ( the group / field gas locally , and predictions for the total igm from cdm simulations at high - z ) .
the ` total observed ' row has a minimum when counting only ` igm observed ' and a maximum when ` igm+ ' is included .
all values are from the literature except the following : the metal content of the ism at both high and low - z , and the baryon content of the ism at high - z , and the stellar metal content at high - z have been estimated as described in the text .
the metallicity of the postulated group / field gas at low - z has been estimated to be @xmath108 but this has not yet been measured . the bottom line shows the baryon density inferred from the cmb measurements ( netterfield et al .
2002 ) , and the metal content of the universe at high and low - z predicted from the integrated star formation history ( pettini 1999 ; pagel 2002 ) , with a yield of 0.0210.042 .
refs : @xmath109 fukugita , hogan & peebles ( 1998 ) .
@xmath110 pettini ( 1999 ) .
@xmath111 netterfield et al .
@xmath112 pagel ( 2002 ) the baryon values are taken from the literature ( fukugita , hogan & peebles 1998 fhp ) except for that associated with the ism at high - z . for this we have used the metals in dust , plus a simple ` closed - box ' chemical evolution model to predict the baryon content of the ism at high redshift . this chemical evolution model ( fig .
[ maxdmf ] ) , and others which incorporate inflow and outflow of gas , predict that the ratio of dust mass to baryonic mass should change quite slowly over a large part of a galaxy s evolution , and should peak when roughly half the gas has been converted into stars ( eales & edmunds 1996 , edmunds & eales 1998 ) .
we have used the maximum ratio of dust mass to baryonic mass ( we are assuming that the submm sources are observed at their maximum dust mass ) given by the model , and the integral of the dmf , to estimate the total amount of baryons associated with the objects in the dmf .
since the maximum occurs at a gas fraction of @xmath113 , the total baryonic mass is split equally between stars and gas ( ism ) .
the assumption that the submm sources are observed close to their peak dust mass is probably valid for the bright submm selected sources , but can not be presumed to hold for the faintest lensed sources and the statistical detection of optically selected sources in the hdf .
this means that the baryon density we have associated with the ism in table [ metalst ] is really a minimum .
+ + the results in table [ metalst ] and fig .
[ histf ] show a number of different things .
first , there was @xmath114 times more dust in galaxies at @xmath44 than there is today ( assuming the dust properties used to estimate the masses are similar at low and high - z , most of the uncertainty here comes from the integration of the local dmf ) .
secondly , as illustrated in fig .
[ histf ] , the amount of metals traced by dust at high - z is 310 times the total amount of metals in dla and lyman forest systems .
the lyman forest represents the igm at high - z , while the role of the dlas is less clear . they may be the ism of galaxies ( wolfe et al .
1986 ) , either the outer parts of galactic disks or their precursors ( efstathiou 2000 ; boissier , peroux & pettini 2002 ) , or dwarf galaxies ( yanny & york 1992 ) .
some fraction of them may also be linked to outflows from star - forming galaxies and therefore could be classed as igm ( shaye 2001 ) . whichever is the case , at high - z the metal content of the ism is clearly much greater than that of the igm . at low redshift
the situation is reversed : the igm now contains far more metals than the ism of galaxies .
this is due to two processes : the consumption of ism gas and metals by star formation and the expulsion of metal enriched gas from galaxies by winds .
much of the igm today may be contained in groups of galaxies ( this is shown in fig . [ histf ] and table [ metalst ] as ` igm+ ' ) .
the amount of this gas is very uncertain but could be enough to provide the baryonic matter necessary to match that inferred from recent cmb measurements ( netterfield et al .
the amount of metals which may be contained in the postulated group igm gas is even more speculative .
the hot igm gas in rich clusters has a metallicity of @xmath115 ( renzini 1997 ) but the metallicity of gas in less dense environments , such as in groups , may be lower ( as predicted by cen & ostriker 1999 ) .
we have used an estimate of @xmath96 for the ` igm+ ' metallicity but stress the large uncertainty in both the amount of gas and metals in this category .
we have also included in table [ metalst ] an admittedly uncertain estimate of the total amount of metals contained in stars at @xmath116 .
this was estimated from the following three populations : 1 .
lyman break galaxies ( lbg ) we took the @xmath54 lf from shapley et al .
( 2001 ) , corrected to this cosmology ( @xmath69 ) , and integrated to @xmath117 . using @xmath118 and @xmath119 ( pettini et al .
2001 ) we used the integral of the lf to estimate the stellar mass density @xmath120 .
we used a metallicity of @xmath121 ( pettini et al .
2001 ) to produce an estimate of the metal density in stars in bright lbg : @xmath122 .
an obvious uncertainty here is the correction to the luminosity density required for dust extinction , which could be a factor 25 2 .
submm selected sources ( @xmath123 mjy ) the dust mass density ( from integrating the dust mass function down to 3 mjy ) , and the closed box model were used to estimate the mass of metals in stars ( and the mass in stars ) .
as above , we assumed that the galaxies were halfway through their evolution , with a gas fraction of @xmath113 .
this gave a stellar mass density of @xmath124 and a stellar metal density of @xmath125 .
the main uncertainties here are the temperature of the dust ( estimates will increase by a factor 2 if the temperature is 20 k rather than 25 k ) , and the broadness of the maximum in the ratio of dust mass / total mass ( fig .
[ maxdmf ] ) .
this means there is a considerable range in the gas fraction and hence the stellar mass associated with a given dust mass .
using a model with inflow or outflow will lead to one inferring a greater baryonic mass from a given dust mass 3 .
the rest of the stellar mass at @xmath44 predicted from the sfh and not in bright lbg or scuba sources .
we assumed that this is in lower luminosity systems and dlas and has @xmath126 ( the average dla metallicity ( pettini 1999 ) ) .
the sum of these three populations is designed to be equivalent to the total stellar density predicted from the sfh , and has a mean metallicity of @xmath127 .
correction of the lbg values for dust would mean less stars in category ( iii ) and more stellar metals overall ( as lbg have higher metallicities ) . a comparison of the metals traced by dust ( ism ) and those in stars at the two epochs leads to a third conclusion : at low redshift the ratio of metals in the ism to those in stars is only @xmath128 , while at @xmath22 this ratio is @xmath129 .
the chemical evolution model suggests that the ratio will be as high as this when only @xmath130% of stars in the universe have formed , in agreement with the star formation history inferred from observations .
taken together , this means that most metals at @xmath6 are in the ism of galaxies .
this is as expected in the early stages of galaxy evolution ; later as the gas is consumed , the metals become locked into stars and expelled into the igm .
table [ metalst ] shows that we can now account for all the metals predicted to exist ( from integrating the sfh ) at both low and high redshift , and thus there is no longer a ` missing metals ' problem .
the dlas do not seem to trace the bulk of the metals within galaxies at high - z , suggesting that there are indeed biases present in dla samples against regions of high density and obscuration ( pei & fall 1995 ) .
a survey of dla s selected in the radio ( ellison et al .
2001 ) has found that there is no evidence for a large missing population of high column density absorbers .
the data allow for @xmath131 to increase by a factor of 2 at most .
however , the statistics of the survey are still somewhat limited , especially at the highest column densities where one may also expect to find the greatest quantities of metals ( in an unbiased survey ) .
we have also shown that the ` missing metals ' can be accounted for without resorting to a large reservoir of metals in a hot , diffuse and as yet undetected part of the igm at high redshift , as has been suggested by various authors ( pettini 1999 ; pagel 2002 ) .
however , there is still a ` missing baryons ' problem , since the total amount of directly observed baryons is less than the amount inferred from recent microwave background ( cmb ) measurements ( netterfield et al .
this shortfall is currently thought to be made up of inter - galactic gas at both low and high redshift ( fhp ) ( ` igm+ ' in table [ metalst ] and fig .
[ histf ] ) .
the chemical evolution models can also be used to try to explain the evolution in the dust mass function shown in fig .
[ dmff ] . fig .
[ modelf ] shows how the ratio of dust mass at any point in the evolution of a spiral galaxy compares with that today .
present - day spiral galaxies have a gas fraction of @xmath132 , and so in fig .
[ modelf ] we have shown how the mass of dust in a galaxy is predicted to change relative to the dust mass it has at @xmath133 .
the figure shows that there is relatively little scope for the evolution of the dust masses of spirals ; they can have at the most dust masses which are factor of 4 greater in the past than their dust masses today .
the local dmf is composed entirely of spiral galaxies , and using pure luminosity evolution ( or , in this case , dust mass evolution ) requires an increase in the dust masses of the local galaxies of a factor @xmath134 in order to reproduce the high redshift objects with @xmath135 . due to the different shapes of the high and low redshift dmfs ,
the discrepancy is not as large for the lowest mass objects , and so the dust masses of the faintest submm sources ( from the statistical analysis of optically selected starbursts in the hdf ) are consistent with simple evolution from the local dmf .
thus the high redshift dmf can not be reproduced by a chemical evolution model applied to local spiral galaxies , which argues that the high redshift dust sources found in the deep scuba surveys are a population which is not represented in the dust mass function today , i.e. that the high redshift sources are in fact the ancestors of the gas and dust poor elliptical galaxies in the local universe .
present - day elliptical galaxies have a much lower typical gas fraction @xmath136 and if fig .
[ modelf ] were to be drawn for ellipticals ( showing how dust mass in the past changes relative to that at a gas fraction of 0.01 ) , we would see curves of similar shape to those in fig .
[ modelf ] but which peak at a maximum of between 10 and 35 times more dust in the past .
however , we note that because the gas fractions and dust masses of present - day ellipticals are poorly known , and because the ratio in fig . [ modelf ]
becomes very sensitive to @xmath137 at low gas fractions these simple models are unable to predict accurately the factor by which the dust mass of ellipticals could have been greater at their peak dust content . by the time elliptical galaxies have reached their current evolutionary status , there will be competition between grain formation in low - mass giant stellar atmospheres ( @xmath138 ) and the very occasional snae - ia , and long - term erosion of grains by hot x - ray gas .
these are very different conditions from the grain formation in massive and intermediate - mass stellar envelopes , snae - ii , mantle growth in interstellar clouds and ism gas conditions that were present at the peak of their dust content .
until a comprehensive model can be based on reliable measurements of the gas and dust content of present - day ellipticals , it is probably unwise to try to extrapolate the dust masses ellipticals should have possessed at their peak dust content ( robust value ) to the dust masses they should have today ( very uncertain ) , other than to accept that there are good reasons why the dust masses could have been much greater in the past ( by factors 10100 ) .
the highest ratio in fig . [ modelf ] is for an outflow proportional to 4 times the star formation rate .
this is quite extreme , and leads to an original mass 5 times greater than the remnant today , which would be difficult to produce in most hierarchical models of galaxy formation .
outflow episodes are expected in a galaxy s lifetime but at a much lower rate , or for a shorter time .
table [ metalst ] also shows that the change in ism baryons from high redshift to today is of the same order as the increase in stellar baryons .
this means there is not scope for 7080% of the ism to have outflowed between @xmath22 and today .
some level of outflow is probably necessary to produce sufficient evolution in the dust masses of ellipticals , and to provide an obvious source of at least some of the inter - galactic metals observed in the lyman - forest at high redshift , and the inter - cluster medium today .
however , the elementary outflow models can give rise to problems in interpreting the nucleosynthetic yields and metallicities .
we now use the results so far to address the hypothesis that scuba galaxies are the ancestors of today s massive ( @xmath139 ) spheroids .
first , we compare the baryons contained in the two populations . at low - z ,
we construct a @xmath140 optical spheroid luminosity function ( lf ) from the 2df redshift survey ( folkes et al .
1999 ) and type specific bulge / disk ratios ( fhp ) .
we then multiply by the stellar mass - to - light ratio ( fhp ) @xmath141 to give a spheroid baryonic mass function ( assuming that in today s spheroids , the stars dominate the baryonic mass ) .
this is integrated down to @xmath142 to give the mass density in @xmath139 spheroids as @xmath143 . at high - z
, we integrate the dmf down to a mass equivalent to an 850@xmath21 m flux of 3 mjy ( @xmath144 ) .
we then use the ratio of @xmath145 from the closed box model ( fig . [ maxdmf ] ) at a gas fraction of 0.5 to estimate the baryon content of the submm sources , @xmath146 .
since we use the ratio at its maximum ( which is very broad ) we are being conservative in our estimate ; using outflow or inflow would also increase the inferred @xmath147 .
the baryonic content of the scuba sources at high - z is therefore similar to the baryonic content of today s massive spheroids , the distant submm sources simply being observed at a time when a greater fraction of those baryons were in the form of gas rather than stars .
there is a further interesting result , namely that if we compare the baryons at @xmath44 associated with the scuba sources to those in dlas , we get the same ratio ( 1.2 2.4 ) as when we compare the baryons in @xmath139 spheroids and all disk galaxies today ( fhp ) .
this coincidence supports the idea that spiral disks are the descendants of the high - z dla population ( wolfe et al .
1986 ; pettini 1999 ) , and that the spheroids are the descendants of the scuba sources . that there are far more metals in the scuba sources than the dlas is evidence that the bulk of stars in spheroids formed at earlier times than those in disks .
we can also use the chemical evolution model to predict the dust mass function that should have been produced by the spheroids when they had formed @xmath148% of their stars .
we again take the local spheroid lf and multiply by the stellar mass / light ratio to give a spheroid mass function . using the closed - box model from fig .
[ maxdmf ] , we then multiply the spheroid masses by the maximum ratio of dust / baryonic mass to give a dust mass function . this
is shown as the shaded region in fig .
[ dmff ] . including outflow has little effect on this shaded region , as the decrease in the maximum dust / baryonic mass ratio ( fig . [ maxdmf ] ) is offset by the increase in overall baryonic mass at higher gas fractions .
the agreement between this very simple prediction and the observed dmf is rather good , supporting the idea that the descendants of scuba galaxies are not the dusty objects in our low - z dmf , but the now dust - poor spheroids . putting this another way , bright
submm selected sources ( @xmath123 mjy ) can account for all the predicted dust from @xmath139 spheroids .
we have used submillimetre surveys of the local and distant universe to produce dust mass functions . taken together
these show the evolution in dust mass and ism metals from redshift zero to @xmath149 . in conjunction with a simple chemical evolution model we have found that : 1 .
there was much more dust in galaxies in the past .
simple dust mass evolution in local dusty galaxies to match the high - z dust mass function is not feasible as the chemical models do not permit spiral galaxies to have had @xmath150 more dust at an earlier point in their history .
this argues that the local submm sources are not the descendants of the sources in the deep surveys . 2 .
most metals ( @xmath151% ) at @xmath22 are found in the ism of galaxies preferentially in the most obscured and active objects .
this solves the ` missing metals ' problem , which was evident when comparing the metals observed in dlas and the lyman forest with the metal density inferred from integrating the star formation history .
this supports the suggestion ( pei & fall 1995 ) that dla samples are not unbiased tracers of star forming galaxies at high - z , but instead sample the lower density / unobscured ( and therefore un - enriched ) environments .
3 . comparison of baryons in @xmath152 spheroids today , and @xmath153 submm sources at high - z , show good agreement , indicating that the submm sources are plausibly the ancestors of the spheroids . using the local optical spheroid lf and the chemical evolution model ,
we can also predict the dust mass function the spheroids should have had at the time when they had formed @xmath148% of their stars .
this agrees remarkably well with the observed dust mass function at high - z , and again strongly supports the hypothesis that scuba sources ( particularly brighter ones ) are elliptical galaxies in the process of forming most of their stars .
we would like to thank the referee for helpful suggestions on improving the manuscript and also jon davies , sarah ellison and bernard pagel for useful comments .
l. dunne aknowledges the support of a pparc fellowship .
s. eales thanks the leverhulme trust for the award of a research fellowship . | deep surveys in many wavebands have shown that the rate at which stars were forming was at least a factor of 10 higher at redshifts @xmath0 than today .
heavy elements ( ` metals ' ) are produced by stars , and the star formation history deduced by these surveys implies that a significant fraction of all metals in the universe today should already exist at @xmath1 .
however , only 10% of the total metals expected to exist at this redshift have so far been accounted for ( in damped lyman alpha absorbers and the lyman forest ) . in this paper , we use the results of submillimetre surveys of the local and high redshift universe to show that there was much more dust in galaxies in the past .
we find that a large proportion of the missing metals are traced by this dust , bringing the metals implied from the star formation history and observations into agreement .
we also show that the observed distribution of dust masses at high redshift can be reproduced remarkably well by a simple model for the evolution of dust in spheroids , suggesting that the descendants of the dusty galaxies found in deep submm surveys are the relatively dust - free spiral bulges and ellipticals in the universe today .
galaxies : evolution galaxies : ism galaxies : luminosity function , mass function submillimetre
( ism):dust , extinction |
when the nonlocal quantum correlations become important in a many - level system , one may not extract the complete information about individual sub - levels although the total information about the whole system is known .
in such a case , sub - levels become strongly correlated with each other , and this fact may allow the information between two distant points ( such as two qubits separated by large distances ) to be communicated instantly . this phenomenon is called action at a distance , and such a pair of sub - levels is called entangled . formerly ,
einstein and his co - authors @xcite , as well as schrdinger @xcite argued that this `` spooky action at a distance '' is a direct consequence of incomplete nature of quantum mechanics .
however , after three decades , bell @xcite showed that einstein s realist idea based on the locality is wrong and it pioneered consecutive experimental realizations which proved that the predictions of quantum mechanics are true regarding the entanglement phenomenon as a nonlocal property of nature . during the last two decades
, a great many experimental efforts have been devoted to entanglement phenomenon in a wide variety of physical systems including entanglement of many photons , mesoscopic systems , and so on @xcite .
hensen et al .
@xcite very recently produced 245 entangled pairs of electrons ( which were 1.3 kilometers apart from each other ) in nine days .
they reported that their results rule out large classes of local realist theories . on the other hand ,
in the theoretical ground , it was quite a challenge to measure the amount of entanglement between two correlated sub - systems @xcite .
the two distinct measures to distinguish between entangled and separable states are concurrence @xcite and negativity @xcite .
one should notice that concurrence can not be used as a criterion for separability condition for the systems with dimensions larger than @xmath7 in hilbert space . using concurrence as a measure of entanglement between two qubits , models based on localized spins in the form of ising , xy and isotropic , as well as anisotropic heisenberg systems
have been widely investigated in the literature @xcite . in order to observe entanglement phenomenon in such systems , selected hamiltonian should include either off - diagonal terms such as anisotropic exchange coupling and dzyaloshinskii - moriya ( dm ) interaction , and/or inhomogeneous external magnetic fields along the ising axis .
apart from these , pairwise entanglement in the systems with three or more qubits @xcite have also been studied in the forms of xx , xy and heisenberg models , as well as in the forms of their varieties . according to these works ,
some important findings can be listed as follows : under certain conditions , next - nearest - neighbor entanglement may be larger than nearest - neighbor entanglement near zero temperature @xcite . as the number of qubits becomes larger than a larger value of external homogeneous magnetic field
is needed to observe entanglement .
however , entanglement disappears shortly after the field exceeds some critical value @xcite .
moreover , isotropic heisenberg spin chain exhibits non - zero concurrence only when the exchange coupling is of antiferromagnetic type @xcite whereas if one applies a magnetic field then the su(2 ) symmetry is broken and it becomes possible for a ferromagnetic isotropic heisenberg chain to have thermal and ground states which are completely entangled @xcite . for xx qubit rings with periodic boundary conditions , ref.@xcite also realized that pairwise entanglement between the nearest - neighbor qubits is invariant under the magnetic field reversal @xmath8 , and that for the same model containing `` even number of qubits '' , bipartite thermal entanglement between neighboring sites should be independent of both the sign of magnetic fields and exchange constants .
@xcite showed for isotropic heisenberg model that the ground state entanglement becomes enhanced ( diminished ) with increasing number of qubits in odd ( even)-numbered qubit rings .
it is also possible to distinguish between thermally entangled and separable states via examining macroscopic properties such as specific heat and magnetic susceptibility which can play the role of some kind of entanglement witness @xcite .
there are also some other works dealing with entanglement properties of qubit - qutrit and qutrit - qutrit chains @xcite . in an extended work , wang et al .
@xcite studied the entanglement in a spin-1/2 and spin-@xmath9 ferrimagnetic chain in which they reported that as the magnitude of spin-@xmath9 increases then the temperature value at which entanglement vanishes becomes higher whereas the ground state thermal entanglement for small-@xmath9 chains is enhanced .
similarly , ref .
@xcite showed that threshold temperature at which entanglement vanishes increases with increasing spin magnitude . in practice ,
it is a difficult task to control the entanglement in a system by manipulating the exchange interactions , and in some cases , the control of the magnitude and direction of externally applied magnetic field proved useful for tuning the entanglement of a spin chain @xcite . therefore , in the present paper , our aim is to clarify the entanglement phenomena in @xmath6 qubit @xmath1 and @xmath2 chains in the presence of magnetic fields applied in both longitudinal ( i.e. easy axis ) and transverse ( hard axis ) directions .
the outline of the paper can be summarized as follows : in sec .
[ formulation ] we define our model .
numerical results are presented in sec .
[ discussion ] . finally sec .
[ conclusion ] contains our conclusions .
we consider 1d heisenberg @xmath1 and @xmath2 spin chain systems consisting of @xmath6 spin-1/2 particles interacting with nearest neighbor interaction .
each qubit in the system is under the influence of a polarized magnetic field applied in @xmath3 plane . within the open boundary condition
, the hamiltonian of a such system can be described as follows : @xmath10+h_{x}\sum_{i=1}^{n}\sigma_{i}^{x}+h_{z}\sum_{i=1}^{n}\sigma_{i}^{z}\ ] ] where @xmath11 and @xmath12 are the pauli spin operators at the site @xmath13 . @xmath14 and @xmath15 are spin - spin exchange interaction terms of the @xmath16 and @xmath4 components of two neighboring spins which are selected to be as @xmath17 and @xmath18 , i.e. , antiferromagnetic type interaction . throughout the work ,
we have fixed the total number of particle @xmath0 and value of spin - spin interaction @xmath19 .
@xmath20 and @xmath21 are the components of the external polarized field . here ,
@xmath22 and @xmath23 are the amplitude of the field and the angle between the field and its corresponding component of the pauli spin operators , respectively . in eq .
( [ eq1 ] ) , the first summation is over the nearest neighbor qubits while the second and third ones are over all of the qubits .
the hamiltonian considered here covers @xmath1 model with @xmath24 and isotropic heisenberg model with @xmath25 in the limit cases .
there are several studies in the literature regarding the limit cases and some variants of the model hamiltonian @xcite .
the physical properties of the system in thermal equilibrium can be obtained via the density matrix operator , which can be defined as follows : @xmath26 here @xmath27 , @xmath28 and @xmath29 are the temperature and boltzmann constant , respectively .
@xmath30 is the partition function which is defined by the trace of the density matrix as @xmath31 .
the basic strategy for obtaining the concurrence ( as a measure of bipartite entanglement ) for two qubits ( say @xmath32 and @xmath33 ) is , constructing the reduced density matrix by tracing out the other spins , and calculating @xmath34 via @xcite @xmath35 where @xmath36s are the square roots of the eigenvalues of the operator @xmath37 which is obtained by @xmath38 in descending order . here
, @xmath39 is the @xmath40 component of the pauli spin matrix related to the spin @xmath41 , @xmath42 is reduced density matrix , and @xmath43 denotes the complex conjugation . the value of concurrence varies from @xmath44 for a separable state ( no entanglement ) to @xmath45 for a maximally entangled state .
in this section , we will focus our attention on the thermal entanglement properties of the @xmath0 qubits @xmath1 and @xmath2 heisenberg spin chains under the existence of a polarized magnetic field .
we especially study the pairwise entanglement between central qubit and its nearest neighbor entanglement @xmath46 and as well as next - nearest entanglement @xmath47 to show how the applied magnetic fields , interaction constant of the @xmath4 component and the temperature affect the natures of entanglement features of the system . for this aim
, the boundaries in related planes separating entangled and non - entangled regions will be given for the selected hamiltonian parameters .
the physical mechanisms underlying of these types of observations will be discussed in detail .
depending on the considered hamiltonian parameters , in the entangled region , the concurrence has a non - zero value corresponding to the presence of a strongly correlated two qubit pair .
contrary to this , concurrence vanishes in the non - entangled region . keeping these facts in mind , we illustrate the boundaries in the @xmath48 contour map in figs .
1(a - d ) with varying @xmath5 parameters such as @xmath49 ( a ) , 0.5 ( b ) , 1.0 ( c ) and 1.5 ( d ) , respectively .
the figures are presented for the nearest - neighbor concurrence @xmath46 , and for the relatively low value of temperature , i.e. for @xmath50 .
it is clear from figure [ fig1](a ) that along the @xmath51 axis , when the value of @xmath52 increases starting from zero , the concurrence @xmath46 tends to rise until it reaches a certain value for a given temperature .
then , @xmath46 begins to decrease with further increment in value of @xmath52 .
moreover , as the value of @xmath52 is sufficiently large , the entanglement between qubits @xmath53 and @xmath54 disappears , hence , @xmath46 vanishes .
these behaviors appear symmetrically at two sides of @xmath55 . along the @xmath56 axis ,
magnetic field dependency @xmath57 of @xmath46 exhibits a nearly similar character to the previous one mentioned above .
the only difference is that the entanglement between qubits @xmath53 and @xmath54 still lives as the value of @xmath58 is sufficiently too large .
figures [ fig1](b - d ) show effect of the interaction constant @xmath5 on the thermal entanglement character of @xmath46 . for varying values of the @xmath5 ratio such as ( a ) @xmath49 , ( b ) 0.5 , ( c ) 1.0 and ( d ) 1.5 , respectively .
the figures are plotted for the value of @xmath50.,title="fig : " ] for varying values of the @xmath5 ratio such as ( a ) @xmath49 , ( b ) 0.5 , ( c ) 1.0 and ( d ) 1.5 , respectively . the figures are plotted for the value of @xmath50.,title="fig : " ] for varying values of the @xmath5 ratio such as ( a ) @xmath49 , ( b ) 0.5 , ( c ) 1.0 and ( d ) 1.5 , respectively .
the figures are plotted for the value of @xmath50.,title="fig : " ] for varying values of the @xmath5 ratio such as ( a ) @xmath49 , ( b ) 0.5 , ( c ) 1.0 and ( d ) 1.5 , respectively . the figures are plotted for the value of @xmath50.,title="fig : " ] based on our numerical results , it has been found that @xmath5 parameter plays an important role in enhancing the entanglement .
as the value of @xmath5 is too large , much more energy originating from magnetic fields is needed to drive the system from entangled state to non - entangled one .
the aforementioned situations can be easily seen by comparing the figures [ fig1](a - d ) with each other .
the findings also demonstrate that the geometrical shape where the entanglement character exists sensitively depends on the selected @xmath5 parameter .
for instance , figures [ fig1](a ) and ( b ) present an elliptical shape whereas the figure ( c ) shows a circular shape entanglement character .
however , as depicted in figure [ fig1](d ) , the general trend of entanglement character again becomes elliptical with further increment in value of @xmath5 .
for the same hamiltonian parameters with figure ( [ fig1]).,title="fig : " ] for the same hamiltonian parameters with figure ( [ fig1]).,title="fig : " ] for the same hamiltonian parameters with figure ( [ fig1]).,title="fig : " ] for the same hamiltonian parameters with figure ( [ fig1]).,title="fig : " ] in figure [ fig2](a - d ) , we examine thermal entanglement features of the next - nearest neighbor concurrence @xmath47 of nine - qubit system for the same hamiltonian parameters with figure ( [ fig1 ] ) . in contrary to the figure [ fig1 ] , in the absence of polarized magnetic field , there is no entanglement behavior between qubits @xmath53 and @xmath59 .
however , increment in strength of the magnetic field gives rise to the existence of an entanglement character .
next , as the strength of applied magnetic field is increased further , @xmath53 and @xmath59 qubits become non - entangled . the mechanism briefly mentioned here is often called the re - entrant type entanglement behavior . we should indicate that there are several studies in the literature showing a re - entrant type entanglement character @xcite .
in order to elucidate the temperature influences on the entanglement properties of the studied system , we plot figures ( [ fig3 ] ) and ( [ fig4 ] ) corresponding to the concurrences @xmath46 and @xmath47 , respectively .
the figures are demonstrated for the value of @xmath60 .
the raising temperature shows a tendency to reduce and destruct the quantum correlations between the considered qubits for the fixed sets of hamiltonian parameters .
if one compares the figures ( [ fig1 ] ) and ( [ fig3 ] ) with each other , it can be easily seen that the concurrence @xmath46 prominently decreases when the value of temperature is increased because the thermal energy is dominant against the spin - spin interactions .
we should also note that although an increment in the value of temperature causes a quantitative change in concurrence @xmath46 , it does not lead to a change in characteristic behavior .
it is beneficial to notice that similar type observations originating from the thermal agitations have been reported in ref .
@xcite where ground state and finite temperature features of two qubit system under the influence of a polarized magnetic field have been discussed in detail .
furthermore , by comparing the figures ( [ fig2 ] ) and ( [ fig4 ] ) with each other , it is possible to say that when the temperature increases from @xmath50 to @xmath61 , unusual and interesting thermal entanglement behaviors occur in next - nearest neighbor concurrence @xmath47 .
these dramatic changes have emerged at the relatively low values of @xmath5 such as for @xmath49 and @xmath61 .
for example , non - entangled region expands in @xmath62 and @xmath55 plane as well as a re - entrant type entanglement character takes place along the @xmath63 lines .
it means that the considered qubits are dragged from non - entangled region to entangled region with increasing value of magnetic field @xmath55 .
then , as the energy arising from the magnetic field is increased further , the system displays opposite behavior than the previous one discussed above . for varying values of the @xmath5 ratio such as ( a ) @xmath49 , ( b ) 0.5 , ( c ) 1.0 and ( d ) 1.5 , respectively .
the figures are plotted for the value of @xmath60.,title="fig : " ] for varying values of the @xmath5 ratio such as ( a ) @xmath49 , ( b ) 0.5 , ( c ) 1.0 and ( d ) 1.5 , respectively .
the figures are plotted for the value of @xmath60.,title="fig : " ] for varying values of the @xmath5 ratio such as ( a ) @xmath49 , ( b ) 0.5 , ( c ) 1.0 and ( d ) 1.5 , respectively . the figures are plotted for the value of @xmath60.,title="fig : " ] for varying values of the @xmath5 ratio such as ( a ) @xmath49 , ( b ) 0.5 , ( c ) 1.0 and ( d ) 1.5 , respectively .
the figures are plotted for the value of @xmath60.,title="fig : " ] for the same hamiltonian parameters with figure ( [ fig3]).,title="fig : " ] for the same hamiltonian parameters with figure ( [ fig3]).,title="fig : " ] for the same hamiltonian parameters with figure ( [ fig3]).,title="fig : " ] for the same hamiltonian parameters with figure ( [ fig3]).,title="fig : " ]
in conclusion , we have investigated the thermal entanglement properties of @xmath0 qubits @xmath1 and @xmath2 heisenberg spin chains within the open boundary condition by making use of concurrence concept .
we design the system such that each qubit in the chain is exposed to the polarized magnetic field in @xmath3 plane . according to our detailed investigation
, the most conspicuous findings mentioned in the present study can be listed as follows .
the boundaries which separate the entangled and non - entangled regions sensitively depend upon the spin - spin interaction term of the @xmath4 component of two neighboring spins @xmath5 , temperature as well as polarized magnetic field components .
for example , @xmath5 parameter plays a crucial role in improving the entanglement depending on the other hamiltonian parameters .
it allows us to control the amount of the entanglement between the selected pair of qubits by varying the value of @xmath5 .
moreover , it is possible to realize a magnetically stimulated entanglement behavior by selecting the suitable system parameters . by changing the value of polarized magnetic field ,
one may create or destruct the quantum correlations between the pair of qubits .
it has also been found that the present system presents a re - entrant type thermal entanglement character depending on the considered system parameters if one considers the next - nearest neighbor pair of qubits .
finally , we deal with the finite temperature influences on the entanglement character of system , and the raising temperature demonstrate a tendency to shrink the quantum correlations between the considered qubits for the fixed sets of hamiltonian parameters .
the numerical calculations reported in this paper were performed at tbitak ulakbim ( turkish agency ) , high performance and grid computing center ( truba resources ) . 99 a. einstein , b. podolsky , n. rosen , phys .
47 ( 1935 ) 777 .
e. schrdinger , naturwissenschaften 23 ( 1935 ) 807 .
e. schrdinger , math .
cambridge 31 ( 1935 ) 555 .
bell , physics 1 ( 1964 ) 195 .
r. horodecki , p. horodecki , m. horodecki , k. horodecki , rev . mod .
( 2009 ) 865 , and the references therein .
b. hensen , h. bernien , a.e .
drau , a. reiserer , n. kalb , m.s .
blok , j. ruitenberg , r.f.l .
vermeulen , r.n .
schouten , c. abelln , w. amaya , v. pruneri , m.w .
mitchell , m. markham , d.j .
twitchen , d. elkouss , s. wehner , t.h .
taminiau , r. hanson , nature 526 ( 2015 ) 682 .
m. horodecki , p. horodecki , r. horodecki , phys .
a 223 ( 1996 ) 1 .
a. peres , phys .
77 ( 1996 ) 1413 .
s. hill , w.k .
wootters , phys .
78 ( 1997 ) 5022 .
n. gisin , phys .
lett . a 210 ( 1996 ) 151 .
terhal , phys .
lett . a 271 ( 2000 ) 319 .
wootters , phys .
80 ( 1998 ) 2245
. g. vidal , r.f .
werner , phys .
a 65 ( 2002 ) 032314 .
d. gunlycke , v.m .
kendon , v. vedral , phys .
rev . a 64 ( 2001 ) 042302 .
kamta , a.f .
starace , phys .
88 ( 2002 ) 107901 . y. sun , y. chen , h. chen , phys .
a 68 ( 2003 ) 044301 .
a. asoudeh , v. karimipour , phys .
a 71 ( 2005 ) 022308 .
zhang , s.s .
li , phys .
rev . a 72 ( 2005 ) 034302 .
youn , k. kang , c.s .
kim , j. phys . a : math
39 ( 2006 ) 10523 .
f. kheirandish , s.j .
akhtarshenas , h. mohammadi , phys .
a 77 ( 2008 ) 042309 .
wang , z.l .
cao , j. phys .
: condens .
matter 20 ( 2008 ) 325229 . c. akyz , e. aydiner , chin .
( 2008 ) 1557 .
a. abliz , j.t .
cai , g.f .
zhang , g.s .
jin , j. phys .
b : at . mol .
42 ( 2009 ) 215503 .
yang , l. zhou , commun .
( beijing , china ) 49 ( 2008 ) 1635 .
arnesen , s. bose , v. vedral , phys .
( 2001 ) 017901 .
k. m. oconnor , w.k .
wootters , phys .
a 63 ( 2001 ) 052302 .
x. wang , p. zanardi , phys .
lett . a 301 ( 2002 ) 1 .
m. asoudeh , v. karimipour , phys .
a 70 ( 2004 ) 052307 .
x. wang , phys .
a 66 ( 2002 ) 034302 .
chen , s.r .
hao , r.h .
yue , physics letters a 300 ( 2002 ) 567 .
a. sahintas , c. akyz , physica a 448 ( 2016 ) 10 .
u. glaser , h. btner , h. fehske , phys .
a 68 ( 2003 ) 032318 .
lin , y.q .
li , phys .
a 68 ( 2003 ) 042330 .
y. yeo , phys .
a 68 ( 2003 ) 022316 .
l. zhou , h.s .
song , y.q .
guo , c. li , phys .
a 68 ( 2003 ) 024301 .
m. cao , s. zhu , phys .
rev . a 71 ( 2005 ) 034311 .
wang , physics letters a 334 ( 2005 ) 352 .
r. eryigit , y.gndc , r. eryigit , physics letters a 349 ( 2006 ) 37 . c. min , z.s .
qun , chin .
phys . lett .
23 ( 2006 ) 2888 .
s. zhe and w.x .
guang , commun .
45 ( 2006 ) 61 .
qiang , c.w .
xue , l. qi , y.r .
hong , commun .
48 ( 2007 ) 1009
zhao , l. zhou , int . j. theor .
46 ( 2007 ) 2437 .
jia , c.w .
wen , l.t .
kun , h.y .
xia , l. hong , x.y .
jie , chin .
b 17 ( 2008 ) 1674 . c. min , l.y .
sheng and z.s .
qun , commun .
51 ( 2009 ) 811 .
x. wang , phys .
a 66 ( 2002 ) 044305 .
t. vrtesi , e. bene , phys .
b 73 ( 2006 )
134404 . o. osenda , g.a .
raggio , phys .
a 72 ( 2005 ) 064102
. z. sun , x.g .
wang , y.q .
li , new j. phys . 7
( 2005 ) 83 .
x. wang , h.b .
li , z. sun , y.q .
li , j. phys . a math .
( 2005 ) 8703 .
zhang , j.q .
liang , g.e .
zhang , q.w .
yan , eur .
d 32 ( 2005 ) 409 . z. sun , x.g .
wang , a.z .
li , physica a 370 ( 2006 ) 483 .
guo , x.l .
huang , h.s .
song , phys .
76 ( 2007 ) 327 .
z. yan , z. shi - qun , h. xiang , chin .
( 2007 ) 2229 . c. akyz , e. aydiner , . mstecaplioglu , optics commun . 281 ( 2008 ) 5271 .
ma , optics commun . 281 ( 2008 ) 484 .
e. albayrak , chin .
b 19 ( 2010 ) 090319 .
guo , m.c .
liang , h.y .
zhu , j. phys . a : math .
theor 43 ( 2010 ) 505301 .
abgaryan , n.s .
ananikian , l.n .
ananikian , a.n .
kocharian , phys .
83 ( 2011 ) 055702 .
carrillo , r. franco , j.s .
valencia , physica a 390 ( 2011 ) 2208 .
x. wang , z.d .
wang , phys .
a 73 ( 2006 ) 064302 .
su , a.m. wang , phys .
lett . a 369 ( 2007 )
terzis , e. paspalakis , phys .
lett . a 333 ( 2004 ) 438 .
y. wang , j. cao , y. wang , phys .
lett . a 342 ( 2005 ) 375 .
ling , y. dong , g.s .
jian , chin .
lett . 25 ( 2008 ) 832 .
huang , int .
50 ( 2011 ) 70 .
l. y .- sheng , commun . theor . phys . 48
( 2007 ) 1017 .
t. stauber , f. guinea , phys .
a 70 ( 2004 ) 022313 .
n. canosa , r. rossignoli , phys .
rev . a 73 ( 2006 ) 022347 .
n. canosa , r. rossignoli , phys .
a 75 ( 2007 ) 032350 .
alcaraz , a. saguia , m.s .
sarandy , phys .
a 70 ( 2004 ) 032333 . | we elucidate the finite temperature entanglement properties of @xmath0 qubits heisenberg @xmath1 and @xmath2 models under the presence of a polarized magnetic field in @xmath3 plane by means of concurrence concept .
we perform a systematic analysis for a wide range of the system parameters .
our results suggest that the global phase regions which separate the entangled and non - entangled regions sensitively depend upon the spin - spin interaction term of the @xmath4 component of two neighboring spins @xmath5 , temperature as well as polarized magnetic field components .
thereby , we think that polarized magnetic field can be used a control parameter to determine the amount of thermal entanglement between pair of qubits for different temperatures and spin - spin interaction terms .
moreover , it has been found that the nearest - neighbor pair of qubits does not point out a re - entrant type entanglement character when one only deals with the nearest - neighbor pair of qubits .
however , as one considers next - nearest neighbor pair of qubits , it is possible to see the evidences of re - entrant type entanglement behaviors
. thermal entanglement , quantum correlations , @xmath6 qubit heisenberg @xmath1 and @xmath2 models , polarized magnetic field . |
recently , using the radiative return method @xcite , a new state , the @xmath8 ( also named @xmath10 in the literature ) , was observed in @xmath11 with the dipion invariant mass close to the @xmath12 region , explicitly , for @xmath13 mev @xcite . later on this state
was also detected at bes in the @xmath14 reaction @xcite and belle in @xmath15 @xcite . in an update of the analysis of ref .
@xcite , results on the channel @xmath16 were presented in the form of number of events as a function of the dikaon invariant mass @xcite .
indeed , in this work , the cross section for @xmath17 was measured as a function of the center of mass energy up to @xmath18 and it is shown there that this reaction is dominated by events where one kaon pair comes from the decay of a @xmath19 . selecting events with a kaon pair within @xmath20 of the @xmath19 mass , an enhancement in the invariant mass of the other kaon pair close to threshold is observed and suggested to be due to the @xmath12 tail , but the low statistics and uncertainties in the @xmath21 line shape prevent the authors to present a cross section for @xmath22 using the @xmath23 final state
. inspired in the physics behind the radiative @xmath24 decay @xcite , a detailed theoretical study of @xmath25 for pions in @xmath26-wave was performed in @xcite .
the process @xmath27 has been also studied in the context of nambu - jona - lasinio models @xcite . in ref .
@xcite , it was shown that the tree level contributions through the @xmath28 mixing are negligible and @xmath25 proceeds through the production of off - shell @xmath29 and @xmath4 pairs , the successive decay of the off - shell kaon or @xmath30 into an on - shell @xmath19 and an off - shell @xmath31 and subsequent @xmath32 scattering .
the starting point was the @xmath2 lagrangian @xcite supplemented with the anomalous term for the @xmath33 interactions .
the corresponding predictions , valid for low virtualities of the exchanged photon and low dipion invariant mass were improved in two respects .
first , the @xmath26-wave @xmath34 amplitudes entering the loop calculations were replaced by the full @xmath34 isoscalar amplitudes as calculated in @xmath5 , which contain the scalar poles @xcite .
second , the lowest order terms of the kaon form factor were replaced by the full kaon form factor as calculated in @xmath5 @xcite which describes satisfactorily the scarce data for energies around @xmath35 @xcite . likewise , the @xmath4 isoscalar transition form factor arising from @xmath2 and the anomalous term , was replaced by the transition form factors as extracted from data on @xmath36 @xcite . the @xmath12 couples strongly to the @xmath37 system and it should contribute to the mechanisms studied in @xcite in the case of @xmath38 production .
therefore it is worthy to study this channel also .
as discussed above , some experimental data has been released for @xmath16 .
we devote this work to the study of this reaction in the framework developed in ref .
in contrast to the @xmath39 final state , the @xmath23 final state is induced at tree level in this framework .
furthermore , as noticed in @xcite , the @xmath12 pole is close to the threshold for the production of the dikaon system and the loop contributions can be enhanced by this pole thus a complete analysis requires to calculate rescattering contributions . in this concern , we know that the @xmath40 meson couples strongly to the @xmath41 system but not to the @xmath42 system . therefore , in addition to the @xmath12 contributions we also expect contributions from the @xmath40 meson to the @xmath23 final state .
the @xmath12 and @xmath40 poles lie slightly below @xmath43 , hence their complete shapes are not expected to be seen in this reaction but their respective tails could give visible effects close to the reaction threshold .
contributions from intermediate vector mesons , @xmath44 , are forbidden by charge conjugation . an important improvement with respect to the formalism used in @xcite
is the more accurate characterization of the @xmath4 form factors .
indeed , recently , the cross section for @xmath45 was precisely measured in the @xmath46 @xmath47 region @xcite .
we use the @xmath4 isoscalar and isovector transition form factors as extracted from this data in our analysis instead of the old data from @xcite .
the paper is organized as follows : in section ii we outline the calculation .
results for the tree level contributions to @xmath16 are given in section iii .
in section iv we adapt previous calculations for the rescattering effects in the @xmath48 final state to the @xmath23 final state .
section v is devoted to the extraction of the @xmath4 transitions form factors from data .
in section vi we analyze the different contributions .
section vii is devoted to estimate the intermediate @xmath8 contribution .
our summary and conclusions are given in section viii .
the calculation of the cross section for the @xmath49 reaction is a non - trivial task . indeed , the leading electromagnetic contributions to this reaction are due to a single photon exchange in whose case we need to calculate the @xmath50 vertex function for a hard virtual photon ( @xmath51 ) .
this is an energy scale far beyond the well grounded calculations based on @xmath52 , its @xmath53 saturated version @xmath54 or even the unitarized formalism , @xmath5 , which is expected to be valid up to energies of the order of @xmath55 ; therefore it is not evident that one can perform reliable calculations at the energy of the reaction .
however , whatever the responsible mechanisms for the reaction be , they must leave their fingerprint at low energies , i.e. for low photon virtualities in the @xmath50 vertex function , which can be calculated using the effective theory for qcd at low energies .
we use this fact to attempt a reasonable calculation of @xmath49 .
we calculate the @xmath50 vertex function at low photon virtualities using @xmath2 supplemented with the anomalous lagrangian for the @xmath33 interactions . as a result
we obtain the electromagnetic part of the @xmath50 vertex function dominated by form factors while the pure hadronic interactions are within the scope of @xmath2 in the case of tree level contributions and involve the leading order on - shell @xmath52 meson - meson amplitudes in the case of one loop contributions .
these results , valid at low energies are improved using the unitarized kaon form factors ( which account for the scarce experimental data at the energy of the reaction ) and experimental data on the @xmath4 transition form factors .
similarly , the leading order on - shell meson - meson interactions are iterated following @xcite to obtain the unitarized meson - meson amplitudes containing the scalar poles .
we start with the calculation of the tree level contributions to @xmath16 within @xmath2 and consider also intermediate vector meson exchange using the conventional anomalous lagrangian for @xmath33 interactions .
we follow the conventions in @xcite and the relevant interactions are given in ref .
the reaction @xmath16 is induced at tree level by the diagrams shown in fig ( [ tree ] ) .
in addition to the vertices quoted in ref .
@xcite there is a tree level @xmath56 point interaction whose vertex is given by@xmath57 with the labels @xmath58 and all incoming particles .
: a ) pseudoscalar mesons exchange with point - like @xmath59 interaction plus contact term , b ) pseudoscalar mesons exchange with vector meson mediated @xmath59 interaction , c ) intermediate vector mesons.,width=377,height=302 ]
a straightforward calculation of the set of diagrams a ) in fig.([tree ] ) yields @xmath60 \eta^{\alpha\nu } \label{mpcontact}%\ ] ] where @xmath61 , @xmath62 .
the tensor @xmath63 is given by @xmath64 where @xmath65 .
it can be easily shown that this tensor satisfies@xmath66 where @xmath67 denotes the antisymmetric tensor used to describe the @xmath19 field in @xmath2 . the second term in eq . ( [ mpcontact ] )
is explicitly gauge invariant , thus the @xmath50 vertex function in the amplitude ( [ mpcontact ] ) is gauge invariant .
diagrams b ) in fig .
( [ tree ] ) involve the propagation of vector particles . these diagrams yield@xmath68 w_{\mu\nu } q_{\alpha}\eta^{\alpha\nu}.\ ] ] with the gauge invariant tensor @xmath69 using the gauge invariance property in eq .
( [ gi ] ) it is possible to relate this tensor to @xmath70 as@xmath71\ ] ] thus the amplitude can be rewritten as@xmath72
q_{\alpha } \eta^{\alpha\nu}%\ ] ] where@xmath73 adding up contributions from diagrams 1a ) and 1b ) we obtain@xmath74 q_{\alpha}\eta^{\alpha\nu}\label{mptree}\\ & + \frac{e^{2}\sqrt{2}}{f^{2}}\left ( g_{v}-\frac{f_{v}}{2}\right ) \frac{l^{\mu}}{k^{2 } } g_{\mu\nu}k_{\alpha } \eta^{\alpha\nu},\end{aligned}\ ] ] where@xmath75 notice that the second term in eq .
( [ mptree ] ) contains the vector meson contributions to the kaon form factor , but the constant term due to the electric charge is missing .
a similar result was obtained in @xcite .
this term should come from lagrangians with higher derivatives of the fields , specifically from the term @xmath76 , which is absent in our basic interactions .
we will assume in the following that the constant term due to the charge is provided by such missing interactions ; hence , we will write @xmath77 instead of @xmath78 in the second term of eq . ( [ mptree ] ) . also , notice that the function @xmath79 accounts for the lowest order terms of the charged kaon form factor and describes it properly at low @xmath80 .
the high virtuality of the exchanged photon in our process requires to work out the complete @xmath81 vertex function .
the calculation of the kaon form factors has been done in the context of @xmath5 in ref . @xcite .
although this calculation misses the contributions of the first radially excited vector mesons which are important around @xmath82 @xmath47 , the existing experimental data close to the @xmath83 threshold are properly described by the unitarized kaon form factors as shown in fig ( [ kff ] ) where the results for the unitarized charged kaon form factor are plotted together with data from the dm2 coll .
@xcite . thus in the following we will replace the lowest order terms so far obtained , @xmath79 , by the full unitarized form factor @xmath84 .
in addition to the terms associated to the kaon form factor we get a contact term with the combination @xmath85 .
this combination is small and it vanishes in the context of vector meson dominance @xcite .
clearly , this term can not be extrapolated to high photon virtualities without its dressing by a form factor .
the impact of this term on the cross section for the @xmath39 final state upon its dressing with the kaon form factors was found to be very small in @xcite .
its contribution turns out to be small also in this case when it is dressed with the charged kaon form factor , thus we will skip it in the following . with these considerations the amplitude reads
@xmath86 q_{\alpha}\eta^{\alpha\nu}. \label{mptprel}%\ ] ] in terms of the conventional polarization vector @xmath87 satisfying@xmath88 we finally obtain the tree level contribution from the exchange of pseudoscalar mesons as@xmath89
\eta^{\nu}. \label{mpt}%\ ] ] the amplitude for the diagrams c ) in fig .
( [ tree ] ) is@xmath90 here , we get the @xmath91 transition form factor to leading order as@xmath92 and @xmath93 stands for the tensor@xmath94 \epsilon_{\sigma \tau\alpha\nu},\ ] ] with @xmath95 , which is explicitly gauge invariant .
it can be shown that @xmath96 where@xmath97 q^{\alpha}.\ ] ] the amplitude in eq .
( [ mcprel ] ) contains the leading order terms for the @xmath91 transition form factor valid for low photon virtualities .
the photon exchanged in @xmath16 has @xmath98 and we should work out this form factor for such high photon virtualities . in the numerical computations we replace the leading order terms by the complete transition form factor , @xmath99 , extracted from recent data on @xmath36 , @xmath100 to be discussed below . with these considerations@xmath101
the final state @xmath23 can get contributions from the @xmath102 rescattering .
furthermore , this final state can also be reached through the production of @xmath103 and the rescattering @xmath104 .
excitation of higher mass states such as @xmath4 is also possible and the final state @xmath23 can be produced by the initial production of an off - shell @xmath105 ( @xmath91 ) pair , the subsequent decay of the off - shell @xmath106 ( @xmath107 ) into a @xmath108(@xmath109 ) and the rescattering @xmath104 ( @xmath102 ) .
this section is devoted to the study of these contributions .
we calculate the rescattering effects to lowest order in the chiral expansion from @xmath2 supplemented with the anomalous term for the @xmath33 interactions .
these results are improved by the unitarization of meson - meson amplitudes proposed in @xcite which dynamically generates the scalar resonances .
the relevant diagrams in @xmath2 are shown in fig .
( [ loops ] ) , where for simplicity a shaded circle and a dark circle account for the diagrams @xmath110 plus @xmath111 and @xmath112 plus @xmath113 respectively , which differentiate the direct photon coupling from the coupling through an intermediate vector meson . in @xmath114 at one loop level.,width=377,height=377 ]
the factorization of the meson - meson chiral amplitudes on - shell out of the loop integrals has been discussed in @xcite in the case of the @xmath39 final state .
the same considerations apply to the @xmath115 final state and we refer the reader to ref .
@xcite for the details . to account for the whole rescattering effects in @xmath102 and @xmath104 in these diagrams we iterate the lowest order chiral amplitudes in the scalar channel , @xmath116 and @xmath117 to obtain the unitarized scalar amplitudes , @xmath118 and @xmath119 ( in the following we skip the suffix @xmath120 used in @xcite for the scalar meson - meson amplitudes in order to handle isospin labels that will be necessary below and denote @xmath121 the @xmath122 unitarized amplitude in the isospin channel @xmath123 ) .
unlike the @xmath39 case where only the isoscalar meson - meson unitarized amplitude containing the @xmath12 pole is involved , in the case of the @xmath23 final state the necessary @xmath118 and @xmath119 amplitudes are linear combinations of the isoscalar and isovector scalar amplitudes @xmath124 and @xmath125 .
the last one contains the @xmath40 pole .
notice that we are using the scalar amplitudes @xmath118and @xmath119 instead of the full amplitudes which in principle contain higher angular momentum contributions .
this is expected to be a good approximation for energies near the reaction threshold due to the small tri - momentum of the dikaon system .
furthermore , at these energies the dikaon mass is near the @xmath12 and @xmath40 poles and the @xmath126-wave amplitudes ( containing vector poles ) give vanishing contribution to this process due to charge conjugation .
higher @xmath127 ( tensor ) contributions are expected to be important at much higher center of mass energy .
the dominance of s - waves near threshold has been observed in alike processes as @xmath128 @xcite .
the loop calculations go similarly to those of the @xmath39 final state done in ref .
we adopt the same convention of all external particles incoming and refer the reader to that work for the details . for kaons in the loops we obtain @xmath129 \eta^{\nu } \label{mkfinal}%\ ] ] with the lorentz structures@xmath130 and @xmath131 i_{p},\\ b_{p } & = \frac{\sqrt{2}m_{\phi}g_{v}}{2f^{2}}\left [ t_{k^{+}k^{+}}f_{k^{+}% } ( k^{2})+t_{k^{0}k^{+}}f_{k^{0}}(k^{2})\right ] \left ( j_{p}+\frac{m_{k}^{2}% } { 4k^{2}}g_{k}\right ) .\end{aligned}\ ] ] the loop functions are given by@xmath132 with @xmath133 .
notice that the only change from the final @xmath134 state @xcite , to the present case , is the substitution of the factor @xmath135 by @xmath136 .
it is convenient to write the meson - meson scalar amplitudes in isospin basis . with the conventions in refs .
@xcite , we get @xmath137 thus , in terms of the isoscalar ( @xmath138 @xmath139 ) and isovector ( @xmath125 , @xmath140 ) meson - meson amplitudes and kaon form factors respectively we obtain@xmath141 i_{p},\label{ap}\\ b_{p } & = \frac{m_{\phi}g_{v}}{2\sqrt{2}f^{2}}\left [ t_{kk}^{0}f_{k}% ^{0}(k^{2})+t_{kk}^{1}f_{k}^{1}(k^{2})\right ] \left ( j_{p}+\frac{m_{k}^{2}% } { 4k^{2}}g_{k}\right ) , \label{bp}%\end{aligned}\ ] ] with the isoscalar and isovector form factors@xmath142 similarly to the tree level contributions , the calculation of kaon loops yields also a contact term with the combination @xmath85 not shown in eq.([mkfinal ] ) . for the same reasons we neglected the analogous term in the previous section , we will also neglect it here .
the process @xmath143 can also proceed through the diagrams shown in fig .
( [ fdv ] ) .
again , the calculations are similar to the case of @xmath39 in the final state calculated in ref .
@xcite and we refer to this work for details on the notation and conventions . . ,
width=302,height=151 ] the amplitude from the diagram in fig .
( [ fdv ] ) gets contributions from charged and neutral @xmath4 in the loops .
we obtain the amplitude from these diagrams as@xmath144 \eta^{\nu } \label{mvfinal}%\ ] ] with @xmath145 i_{v},\\ b_{v } & = -\frac{gm_{\phi}^{2}}{4\sqrt{2}}\left [ f_{k^{\ast+}k^{-}}% ( k^{2})t_{k^{+}k^{+}}+f_{k^{\ast0}\overline{k^{0}}}(k^{2})t_{k^{0}k^{+}% } \right ] j_{v}\\ i_{v } & = m_{k}^{2}\left ( i_{g}-i_{2}+2+\frac{1}{2}\log\frac{m_{k}^{2}}% { \mu^{2}}\right ) + q\cdot k\ j_{v}.\end{aligned}\ ] ] here @xmath99 stands for the leading order terms in the @xmath91 transition form factor in eq .
( [ tfflo ] ) and @xmath146 denotes the @xmath105 leading order terms of the neutral transition form factor given by @xmath147 the loop functions are given by @xmath148\label{i2}\\ i_{g } & = -2+\log\frac{m_{k}^{2}}{\mu^{2}}+\sigma\log\frac{\sigma+1}% { \sigma-1}.\end{aligned}\ ] ] notice again that the only change from the final @xmath134 state to the present case is the replacement of the factor @xmath149 by @xmath150 . in terms of the isoscalar and isovector amplitudes and transition form factors we get@xmath151 i_{v},\qquad b_{v}=-\frac{gm_{\phi}^{2}% } { 8\sqrt{2}}\left [ t_{kk}^{0}f_{k^{\ast}k}^{0}(k^{2})+t_{kk}^{1}f_{k^{\ast}% k}^{1}(k^{2})\right ] j_{v}. \label{avbv}%\ ] ] with@xmath152 finally , taking into account both pseudoscalars ( eq.([mkfinal ] ) ) and vectors ( eq.([mvfinal ] ) ) in the loops we obtain the total amplitude as@xmath153 \eta^{\nu}%\ ] ] where@xmath154 with the specific functions in eqs.([ap],[bp],[avbv ] ) .
recall these results are valid for ingoing particles .
for the numerical computations in the following section we reverse the momenta of the final particles to obtain@xmath155 \eta^{\nu } \label{finalamp}%\ ] ] with@xmath156 where the functions @xmath157 @xmath158 @xmath159 @xmath160 are obtained from the untilded functions in eqs .
( [ ap],[bp],[avbv ] ) just changing the sign of @xmath161 in the integrals @xmath162 @xmath163 @xmath164 and @xmath165 in eqs .
( [ iab],[jab],[jv],[i2 ] ) .
our results involve the @xmath4 isoscalar and isovector transition form factors at the center of mass energy of the reaction .
our calculation based on lagrangians suitable for low energies yields the lowest order terms in the chiral expansion for these form factors . since a calculation of these form factors at such huge energies as @xmath166
@xmath47 is beyond the scope of the effective theories used here , we consider our calculation as a convenient procedure to identify the main physical effects ( form factors and meson - meson amplitudes ) and must find a way to enlarge the range of validity of the calculation to high energies .
we do this by replacing the leading order terms in the transition form factors by an appropriate characterization of these form factors at the energy of the reaction . in the case of the @xmath39 final state , only the isoscalar transition form factor is required and in @xcite it was extracted from data on @xmath167 at @xmath168 @xmath169 reported in @xcite . in the present case , also the isovector transition form factor is required .
this form factor can not be extracted from this data due to the low statistics . indeed , in @xcite it is assumed that the isovector amplitude is dominated by a @xmath170 with a mass and width fixed to @xmath171 @xmath172 and @xmath173 @xmath172 in the interpretation of the data .
fortunately , high precision measurements for the cross section of @xmath174 , @xmath175 were recently released where the contributions from intermediate @xmath30are identified @xcite .
the first reaction involves the charged @xmath4 transition form factor while the second gets contributions of both charged and neutral transition form factors . using this data , the isoscalar and isovector components of the cross section for @xmath45 at @xmath168 @xmath176
were extracted along with an energy dependent phase which signals another resonance below @xmath177 @xmath47 which could not be resolved @xcite .
we extract the form factors from this data as follows .
first we calculate the @xmath178 vertex function in @xmath2 and replace the lowest order terms in the transition form factor by the full one to obtain @xmath179 with all incoming particles
. a straightforward calculation of the @xmath180 cross section using this effective vertex yields@xmath181 where@xmath182 in ref .
@xcite a fit was done to the isovector and isoscalar components of @xmath183 , @xmath175using also data on the cross section for the production of @xmath184 where the @xmath185 peak is visible .
the analysis below @xmath177 @xmath47 required the introduction of an energy dependent relative phase pointing to the existence of an unresolved resonance in this energy region , in addition to the @xmath170 and the @xmath185 . here , we are interested only in an appropriate characterization of the isoscalar and isovector form factors in the @xmath186 @xmath47 region where the effect of this phase should be small .
the experimental points for the squared absolute value of the form factors are obtained using eq .
( [ sigksk ] ) and tables vi and vii of ref .
physically , these form factors are dominated by the exchange of resonances ; hence in their characterization we complement the lowest order terms obtained in @xmath2 with the exchange of heavy mesons @xmath187 with the energy dependent widths used in @xcite @xmath188 \\ \gamma_{\rho^{\prime}}(s ) & = \gamma_{\rho^{\prime}}\left [ \frac { \mathcal{p}_{4\pi}(s)}{\mathcal{p}_{4\pi}(m_{\rho^{\prime}}^{2})}b_{4\pi } ^{\rho^{\prime}}+(1-b_{4\pi}^{\rho^{\prime}})\right ] , \end{aligned}\ ] ] where@xmath189 the values for the constants appearing here are extracted from the central values of tabel xv in @xcite as @xmath190 , @xmath191 , @xmath192 , @xmath193 @xmath172 , @xmath194 @xmath172 , @xmath195 @xmath172 , @xmath196 @xmath172 .
the parameters @xmath197 , @xmath198 are fitted to the data for the isoscalar and isovector cross sections reported in tables vi and vii of @xcite respectively .
the fit to the isoscalar cross section yields @xmath199@xmath200 .
the fit to the isovector cross section yields @xmath201 @xmath200 . notice that these values are nearly equal , pointing to the @xmath185 and @xmath170 as members of an @xmath202 nonet .
our results for the corresponding cross sections are shown in figs .
( [ csis],[csiv ] ) , where the shadowed bands correspond to the @xmath203 regions for the paramenters @xmath197 and @xmath198 .
a good description of the isoscalar cross section is obtained but more accurate data on the isovector @xmath204 cross section is desirable .
the discrepancy in this case is due to the effects another intermediate resonance around @xmath205 in @xmath204 which however could not be resolved in @xcite . in the isoscalar channel .
the band corresponds to the @xmath206 region of our parametrization of the form factors in eq.([ksktffi0 ] ) .
data points are taken from table vi of ref .
] in the isovector channel .
the band corresponds to the @xmath206 region of the parametrization of the form factors in eq.([ksktffi1 ] ) .
data points are taken from table vii of ref .
] as a cross check , we calculate the full cross section for @xmath183 from the exchange of @xmath30 which involves only the charged transition form factor and obtain a proper description of data in table iii of ref .
@xcite up to @xmath207 .
with respect to the @xmath39 production with @xmath26-wave pions studied in @xcite , there is now the novelty of the tree level contributions in @xmath115 production with @xmath41 pairs in all possible angular momentum states . in the case of loops ,
our calculations consider only s - wave @xmath41 in the final state because we are using the unitarized amplitudes for @xmath208 which per construction are projected onto well defined angular momentum ( @xmath209 ) and isospin ( @xmath210 ) . in the appendix
we give details on the integration of phase space and state our conventions .
the dikaon spectrum turns out to be @xmath211 where @xmath212 and these amplitudes are given in eqs .
( [ mpt],[mvt],[finalamp ] ) respectively .
we refer the reader to the appendix for further details on the notation and integration of phase space .
we evaluate numerically the integrals and the differential cross section using the physical masses and coupling constants .
we remark that all the parameters entering this calculation have been fixed in advance and , in this sense , there are no free parameters . in our numerical analysis
we use @xmath213 , @xmath214 , @xmath215 , @xmath216 , @xmath217 , @xmath218 , and @xmath219 . concerning the rescattering effects we remark that the @xmath220 and @xmath104 scattering between the kaons in the loops and the final kaons takes place at the dikaon invariant mass independently of the value of @xmath221 and of the momenta in the loops .
as a consequence , when replacing the lowest order terms for this amplitude by the unitarized amplitude , we can safely use the results of @xcite and take a renormalization scale @xmath222 @xmath47 for the function @xmath223 @xcite , in spite of the fact that the reaction @xmath224 takes place at a much higher energy @xmath225 .
the unitarized amplitudes naturally contain the scalar poles and there is no need to include explicitly these degrees of freedom in the calculation .
we obtain the dikaon spectrum shown in fig .
( [ dsdm ] ) for @xmath226 @xmath172 . in order to compare with the dikaon spectrum in fig .
( 26 ) of @xcite , we should collect all events from threshold up to @xmath227 @xmath47 as included in the sample for that figure .
this amounts to integrate our differential cross section on @xmath221 in this energy range .
although a precise comparison can not be carried out since data is presented in the form of number of events , our results in fig .
( [ spectrum ] ) follow a similar pattern to those of fig .
( 26 ) of @xcite . in order to check the conjecture in @xcite that the enhancement close to the dikaon threshold in the integrated spectrum is due to the presence of the @xmath12 pole , in this plot we also show the individual contributions coming from tree level and rescattering .
the dashed plots are obtained integrating on @xmath221 the tree level and rescattering individual contributions to the dikaon spectrum at different @xmath221 shown in figs .
( [ dsdmtree],[dsdmloops ] ) . the analogous spectrum for the interference between tree level and rescattering contributions is shown in fig.([dsdmint ] ) . the enhancement near threshold for rescattering effects in fig .
( [ dsdmloops ] ) is due to the tails of the @xmath12 and the @xmath40 whose poles are well reproduced in the unitarization of meson - meson @xmath26-wave amplitudes present in our calculation .
in contrast to the @xmath39 final state studied in @xcite where the complete @xmath12 resonance is visible in the dipion spectrum , here the peaks of both , @xmath228 and @xmath229 , lie below @xmath43 in the dikaon invariant mass , i.e. below the threshold for the production of the dikaon system in the reaction @xmath16 , thus , it is only the tail of these resonances what the kinematics allows us to see .
close to the threshold for the dikaon production , rescattering effects are enhanced by the scalar poles .
however , the integrated spectrum in fig.([spectrum ] ) shows that these contributions are of the same order than tree level ones close to threshold only . beyond this
region tree level contributions become dominant .
we can also see a constructive interference between these contributions close to threshold and a destructive interference beyond @xmath230 in the integrated dikaon spectrum .
a closer analysis of tree level contributions shows that they are dominated by the exchange of @xmath30 , thus the enhancement close to threshold in the integrated dikaon spectrum in fig.([spectrum ] ) is due to both , the scalar poles and tree level @xmath30 exchange . events beyond threshold are mainly due to @xmath30 exchange at tree level .
the interference is more transparent at the level of the dikaon spectrum as a function of @xmath221 shown in fig.([dsdmint ] ) .
the sign depends both on @xmath231 and @xmath221 but for a fixed @xmath221 it is positive close to the dikaon threshold and evolves to a negative interference beyond this threshold .
this interference shows up differently in the total cross section for @xmath16 obtained upon integration of the dikaon invariant mass in fig .
( [ dsdm ] ) .
our result for the cross section is shown in fig .
( [ sig ] ) . here ,
we get a constructive interference close to the reaction threshold which evolve to a destructive interference beyond @xmath232 . in this figure
our results are compared with measurements of the cross section for @xmath17 from ref .
this comparison makes sense since , as shown in fig .
( 25 ) of @xcite , for most of the considered events one of the final @xmath41 pairs comes from a @xmath19 . using the narrow width approximation ( quite appropriate for the @xmath19 ) it can be shown that the measured cross section coincides with the cross section for @xmath16 . in fig .
( [ sig ] ) we see that the interference between tree level and rescattering effects is crucial in the proper description of data . above @xmath233
we expect higher @xmath127 dikaon contributions ( exchange of tensor mesons in the @xmath29 rescattering ) not considered in the unitarized amplitudes which per construction are the @xmath209 projected amplitudes . as a function of the center of mass energy ( solid line ) .
short - dashed lines correspond to the @xmath206 region of the transition form factors in figs .
( [ csis],[csiv ] ) .
long - dashed line corresponds to tree level contributions .
the experimental points are taken from ref .
@xcite and correspond to the cross section for @xmath234 with one of the kaon pairs coming from a @xmath19 ( see the discussion in the body of the paper ) . ] in ref .
@xcite the authors observed a peak , named @xmath8 , around @xmath235 in the cross section of the @xmath236 with the dipion invariant mass in the @xmath12 peak region .
the @xmath237 mode was also observed at bes and belle thus the @xmath8 must be an isoscalar @xmath238 state .
the mass and width measured in the @xmath239 channel , @xmath240 , @xmath241 , disagree with the predictions of quark models for the @xmath242 state ( e.g. @xcite predicts @xmath243 , @xmath244 ) , thus alternative structures have been proposed for this resonance @xcite . in particular , a narrow peak with a mass close to the experimental value has been generated solving the faddeev equations for the three body problem involving chiral interactions of the @xmath245 system @xcite .
the peak appears for a @xmath246 invariant mass close to the @xmath228 pointing to the @xmath8 as a @xmath247 dynamically generated resonance .
clearly , there must exist contributions from the intermediate @xmath8 to @xmath16 . furthermore , since the @xmath41 system in this reaction can also be in an isovector state , in principle this reaction could get contributions from an isovector companion of the @xmath8 if it exists and in this concern it is worth to remark that in the approach of @xcite no peak is generated in the isovector channel . since our calculation accounts for most of the experimental data , we would expect these contributions to be small except perhaps around @xmath248 where there is a point out of region obtained using the characterization of the @xmath4 transition form factors within @xmath206 .
this point is slightly above the @xmath8 peak observed by babar and bes collaborations and could be a signal of intermediate resonances whose shape is distorted by their own interference and interference with the mechanisms studied here . in this concern
it is desirable to have a rough estimate of the contributions of the intermediate @xmath8 via the chain @xmath249 with a slightly off - shell @xmath228 .
we devote the following section to the calculation of these contributions .
@xcite , the product @xmath250 was obtained .
this allows us to estimate the intermediate @xmath8 contributions depicted in fig .
( [ x ] ) . for the @xmath251 and @xmath252 interactions we use the phenomenological lagrangians @xmath253 contributions to @xmath254 . ] from these interactions , in the corresponding center of momentum system ( c.m.s . )
we obtain @xmath255 the amplitude for fig .
( [ x ] ) reads @xmath256 using the babar central values @xmath257 , the product @xmath258 as extracted from eqs.([gammas ] ) , @xmath259 @xcite and the particle data group values @xmath260 @xcite we obtain the cross section shown in fig .
( [ sigx ] ) .
close to the peak of the @xmath8 these contributions are of the same order as the ones previously considered thus they must be included in our calculation .
the relative sign between this amplitude and the previous contributions can not be fixed thus we explore both constructive and destructive interference .
our results are shown in figs .
( [ sigtotal],[sigdest ] ) for constructive and destructive interference respectively .
contributions to @xmath254 as a function of the center of mass energy . ] as a function of the center of mass energy including all contributions(solid line ) and constructive interference .
short - dashed lines correspond to the @xmath206 region of the transition form factors in figs .
( [ csis],[csiv ] ) . ] for destructive interference as a function of the center of mass energy including all contributions(solid line ) .
short - dashed lines correspond to the @xmath206 region of the transition form factors in figs .
( [ csis],[csiv ] ) . ] in the case of constructive interference , our calculation suggests that the effects of the @xmath8 could be seen in @xmath254 over the background generated by the intermediate light vectors and the rescattring effects . in the case of destructive interference ,
a dip is generated instead of a peak and this seems to be the case favoured by data .
however , definite conclusions require a more precise extraction of the isovector @xmath4 transition form factor at @xmath261 ( see fig .
( [ csiv ] ) ) because the tree level intermediate light vectors and rescattering contributions are sensistive to this form factor .
indeed , using only data in the @xmath262 region in the fit of the isovector form factor we obtain an increase of roughly @xmath263 in the cross section shown in figs .
( [ sigtotal],[sigdest ] ) with a similar energy dependence for @xmath264 .
furthermore , intermediate contributions of an hypotetical isovector companion of the @xmath8 can not be excluded at this point .
the prediction of the absence of isovector companions of the @xmath8 in the framework of the dinamically generated resonances @xcite makes interesting to measure more accuratelly the cross section for @xmath254 in the near threshold region .
the possible isovector resonance could be seen cleanly in the pure isovector @xmath265 final state thus we encourage experimentalists to measure also the @xmath266 cross section around @xmath267 .
we study electron - positron annihilation into @xmath23 .
we start with the @xmath2 lagrangian which yields tree level contributions to this reaction involving both pseudoscalar and vector exchange . for the latter we use the conventional anomalous lagrangian .
the dynamics is dominated by the kaon form factors and the @xmath4 transition form factors , whose lowest order terms ( valid for low photon virtualities ) , arise in the calculation with the effective theory .
this result is improved replacing the lowest order terms provided by @xmath2 for the kaon form factors , by the full form factor as calculated in @xmath268@xcite .
the @xmath4 transition form factors are extracted from recent data on @xmath269 @xcite and used in the numerics instead of the lowest order terms arising from the anomalous @xmath3 lagrangian .
we consider also rescattering of the final charged kaons and @xmath115 production through intermediate @xmath103 production and subsequent rescattering @xmath104 .
the corresponding results , valid for low energy dikaon mass are improved using the unitarized meson - meson amplitudes containing the scalar poles @xcite instead of the lowest order terms obtained in the calculation with @xmath52 .
the dynamical generation of the scalar resonances in the meson - meson unitarized amplitudes enhances the rescattering effects close to the reaction threshold .
the poles of the scalar resonances lie below the dikaon threshold energy but the tail of the @xmath12 and @xmath40 mesons are visible in the dikaon spectrum of rescattering contributions . in spite of this , beyond the close to threshold region
, the tree level vector exchange contribution turns out to be dominant .
the calculated cross section is sensitive to the @xmath4 form factors but within @xmath206 in their fit to @xmath269 recent data , we obtain results in agreement with measurements of the cross section for @xmath17 , where one of the kaon pairs comes from the decay of a @xmath19 , except in a narrow region near @xmath270 .
since the @xmath8 has been observed in the @xmath271 reaction , and the @xmath12 couples strongly to the @xmath272 system , the @xmath8 could also show up in the cross section for @xmath16 slightly beyond the threshold energy of the reaction in spite of the fact that it is not possible to see the whole @xmath12 shape in the dikaon invariant mass because it lies below the kinematical threshold for the production of two kaons .
we estimate these contributions using phenomenological lagrangians and extracting the product of the @xmath273 and @xmath274 couplings from the babar value for @xmath7 .
we obtain a sizable contribution of intermediate @xmath8 to the @xmath16 at the @xmath8 peak . including all contributions we get the results shown as shown in figs .
( [ sigtotal],[sigdest ] ) .
the large uncertainties in the extraction of the isovector @xmath4 transition form factor from data do not allow for definite conclusions but the data around @xmath270 hints to a destructive interference between the intermediate @xmath8 contributions and to possible contributions from an isovector companion of the @xmath8 . in the light of a recent calculation of the @xmath8 as a three body structure where no isovector companion is generated @xcite , it is important to have more precise data on the isovector @xmath4 transition form factor and on the cross section for @xmath16 near threshold .
a cleaner signal of an hypothetical isovector companion of the @xmath8 could be detected in a pure isovector final state and in this concern it is worthy to study the @xmath275 channel from both the experimental and theoretical point of view .
this work was partly supported by dgicyt contract number fis2006 - 03438 and the generalitat valenciana .
this research is part of the eu integrated infrastructure initiative hadron physics project under contract number rii3-ct-2004 - 506078 .
the work of m. napsuciale was also supported by dinpo - ug and conacyt - mxico under project conacyt-50471-f .
selim gmez - avila wish to acknowledge support by conacyt - mxico under the mixed grants program .
m.n and s.g .
wish to thank the hospitality of the theory group at ific .
we thank e. p. solodov for useful mail exchange concerning data presented in @xcite .
the differential cross section is given by@xmath276 integrating @xmath277 variables we obtain@xmath278 next we integrate the @xmath279 variables .
we work in the center of momentum system ( cms ) .
the kinematical constriction in this frame reads@xmath280 where @xmath281 @xmath282 and @xmath283 . in order to directly integrate the @xmath279 variables we choose our frame as shown in fig.([frame ] ) . with these conventions the @xmath284 angle
@xmath285 is the @xmath279 polar angle and can be integrated using the @xmath286 function to obtain@xmath287 where we neglected @xmath288 terms in the flux factor .
the squared amplitude @xmath289 is a function of the scalar products of @xmath290 momentum conservation requires @xmath291 furthermore @xmath292 thus@xmath293 hence the squared amplitude for this process in general depends only on the following scalar products@xmath294 the momentum @xmath295 thus @xmath296 integration of @xmath297 requires the explicit form of the average squared amplitude since the dependence on @xmath298 introduces a dependence on @xmath299 .
here we must take into account that this angle is related to the @xmath300beam angle @xmath301 and the @xmath302beam angle @xmath303 as@xmath304 thus there is a non - trivial dependence on @xmath297 .
the integration of the azimuthal angle of the @xmath19 is straightforward .
then we integrate the polar angle of the @xmath19 .
integration on @xmath305 requires to calculate the maximum and minimum kaon energy allowed by the kinematical constraint @xmath306 for @xmath307 respectively .
these are the solutions to@xmath308 which turn out to be @xmath309\ ] ] this way we get the spectrum@xmath310 this spectrum can be written in terms of the dikaon invariant mass related to @xmath311 and @xmath312 as @xmath313 defined in eq .
( [ lambda]).the dikaon spectrum is@xmath314 it is convenient to work with the dimensionless quantities@xmath315 in terms of these quantities@xmath316 in terms of these variables the scalar products read@xmath317 the range of values for @xmath311 are @xmath318 @xmath319 thus@xmath320 m. benayoun , s. i. eidelman , v. n. ivanchenko and z. k. silagadze , mod .
a * 14 * , 2605 ( 1999 ) [ arxiv : hep - ph/9910523 ] .
a. denig , nucl .
suppl . * 162 * , 81 ( 2006 ) [ arxiv : hep - ex/0611024 ] ; d. leone [ kloe collaboration ] , nucl .
* 162 * , 95 ( 2006 ) ; h. czyz , a. grzelinska , j. h. kuhn and g. rodrigo , eur .
j. c * 47 * , 617 ( 2006 ) [ arxiv : hep - ph/0512180 ] ; a. g. denig [ kloe collaboration ] , int .
j. mod .
phys . a * 20 * , 1935 ( 2005 ) .
j. h. kuhn , eur .
j. c * 33 * , s659 ( 2004 ) .
s. e. muller [ kloe collaboration ] , nucl .
suppl . * 126 * , 335 ( 2004 ) ; a. g. denig _ et al . _ [ the kloe collaboration ] , nucl .
suppl . *
116 * , 243 ( 2003 ) [ arxiv : hep - ex/0211024 ] ; b. aubert _ et al . _
[ babar collaboration ] , phys .
d * 70 * , 072004 ( 2004 ) [ arxiv : hep - ex/0408078 ] .
b. aubert _ et al . _ [ babar collaboration ] , phys .
d * 71 * , 052001 ( 2005 ) [ arxiv : hep - ex/0502025 ] .
b. aubert _ et al . _
[ babar collaboration ] , phys .
d * 73 * , 012005 ( 2006 ) [ arxiv : hep - ex/0512023 ] .
b. aubert _ et al .
_ [ babar collaboration ] , phys .
d * 73 * , 052003 ( 2006 ) [ arxiv : hep - ex/0602006 ] .
r. baldini ferroli [ babar collaboration ] , int . j. mod .
a * 21 * , 5565 ( 2006 ) ; b. a. shwartz [ belle collaboration ] , nucl .
suppl . * 144 * , 245 ( 2005 ) ; h. czyz , a. grzelinska , j. h. kuhn and g. rodrigo , eur .
j. c * 39 * , 411 ( 2005 ) [ arxiv : hep - ph/0404078 ] .
h. czyz , j. h. kuhn , e. nowak and g. rodrigo , eur .
j. c * 35 * , 527 ( 2004 ) [ arxiv : hep - ph/0403062 ] .
h. czyz , a. grzelinska , j. h. kuhn and g. rodrigo , eur .
j. c * 27 * , 563 ( 2003 ) [ arxiv : hep - ph/0212225 ] .
v. m. aulchenko _ et al . _
[ snd collaboration ] , phys .
b * 440 * , 442 ( 1998 ) [ arxiv : hep - ex/9807016 ] ; r. r. akhmetshin et al .
[ cmd-2 collaboration ] , phys .
b * 462 * , 380 ( 1999 ) [ arxiv : hep - ex/9907006 ] ; a. aloisio _ et al .
_ [ kloe collaboration ] , phys . lett .
b * 537 * , 21 ( 2002 ) [ arxiv : hep - ex/0204013 ] ; f. ambrosino _ et al .
_ [ kloe collaboration ] , eur .
j. c * 49 * , 473 ( 2007 ) [ arxiv : hep - ex/0609009 ] .
f. ambrosino _ et al .
_ [ kloe collaboration ] , phys .
b * 634 * , 148 ( 2006 ) [ arxiv : hep - ex/0511031 ] .
a. bramon , r. escribano , j. l. lucio m , m. napsuciale and g. pancheri , eur .
j. c * 26 * , 253 ( 2002 ) [ arxiv : hep - ph/0204339 ] ; d. black , m. harada and j. schechter , phys .
d * 73 * , 054017 ( 2006 ) [ arxiv : hep - ph/0601052 ] ; d. black , m. harada and j. schechter , phys .
lett . * 88 * , 181603 ( 2002 ) [ arxiv : hep - ph/0202069 ] .
e. marco , s. hirenzaki , e. oset and h. toki , phys .
b * 470 * , 20 ( 1999 ) [ arxiv : hep - ph/9903217 ] ; v. e. markushin , eur .
j. a * 8 * , 389 ( 2000 ) [ arxiv : hep - ph/0005164 ] ; j. e. palomar , l. roca , e. oset and m. j. vicente vacas , nucl .
a * 729 * , 743 ( 2003 ) [ arxiv : hep - ph/0306249 ] ; j. a. oller , nucl .
a * 714 * , 161 ( 2003 ) [ arxiv : hep - ph/0205121 ] .
m. napsuciale , e. oset , k. sasaki and c. a. vaquera - araujo , phys . rev .
d * 76 * , 074012 ( 2007 ) [ arxiv:0706.2972 [ hep - ph ] ] .
m. bystritskiy , m. k. volkov , e. a. kuraev , e. bartos and m. secansky , phys .
d * 77 * , 054008 ( 2008 ) [ arxiv:0712.0304 [ hep - ph ] ] .
g. ecker , j. gasser , a. pich and e. de rafael , nucl .
b * 321 * , 311 ( 1989 ) .
j. a. oller , e. oset and j. r. pelaez , phys .
* 80 * , 3452 ( 1998 ) [ arxiv : hep - ph/9803242 ] ; j. a. oller , e. oset and j. r. pelaez , phys . rev .
d * 59 * , 074001 ( 1999 ) [ erratum - ibid .
d * 60 * , 099906 ( 1999 ) ] [ arxiv : hep - ph/9804209 ] .
g. ecker , j. gasser , h. leutwyler , a. pich and e. de rafael , phys .
b * 223 * , 425 ( 1989 ) .
u. g. meissner and j. a. oller , nucl . phys .
a * 679 * , 671 ( 2001 ) [ arxiv : hep - ph/0005253 ] ; a. falvard _ et al . _
[ dm2 collaboration ] , phys .
d * 38 * , 2706 ( 1988 ) ; m. ablikim _ et al .
_ [ bes collaboration ] , phys .
b * 607 * , 243 ( 2005 ) [ arxiv : hep - ex/0411001 ] ; j. z. bai _ et al . _
[ bes collaboration ] , phys .
d * 68 * , 052003 ( 2003 ) [ arxiv : hep - ex/0307058 ] .
t. barnes , n. black and p. r. page , phys .
d * 68 * , 054014 ( 2003 ) [ arxiv : nucl - th/0208072 ] .
g. j. ding and m. l. yan , phys .
b * 650 * , 390 ( 2007 ) [ arxiv : hep - ph/0611319 ] ; g. j. ding and m. l. yan , phys .
b * 657 * , 49 ( 2007 ) [ arxiv : hep - ph/0701047 ] ; z. g. wang , nucl .
a * 791 * , 106 ( 2007 ) [ arxiv : hep - ph/0610171 ] ; h. x. chen , x. liu , a. hosaka and s. l. zhu , phys .
d * 78 * , 034012 ( 2008 ) [ arxiv:0801.4603 [ hep - ph ] ] . | in this work we study the @xmath0 reaction .
the leading order electromagnetic contributions to this process involve the @xmath1 vertex function with a highly virtual photon .
we calculate this function at low energies using @xmath2 supplemented with the anomalous term for the @xmath3 interactions .
tree level contributions involve the kaon form factors and the @xmath4 transition form factors .
we improve this result , valid for low photon virtualities , replacing the lowest order terms in the kaon form factors and @xmath4 transition form factors by the form factors as obtained in @xmath5 in the former case and the ones extracted from recent data on @xmath6 in the latter case .
we calculate rescattering effects which involve meson - meson amplitudes .
the corresponding result is improved using the unitarized meson - meson amplitudes containing the scalar poles instead of the lowest order terms . using the babar value for @xmath7
, we calculate the contribution from intermediate @xmath8 .
a good description of data is obtained in the case of destructive interference between this contribution and the previous ones , but more accurate data on the isovector @xmath4 transition form factor is required in order to exclude contributions from an intermediate isovector resonance to @xmath0 around @xmath9 . |
since the beginning of financial science , stock prices , option prices and other quantities have been described by stochastic and partial differential equations . since the 1980s however , the path integral approach , created in the context of quantum mechanics by richard feynman @xcite , has been introduced to the field of finance @xcite .
earlier , norbert wiener @xcite , in his studies on brownian motion and the langevin equation , used a type of functional integral that turns out to be a special case of the feynman path integral ( see also mark kac @xcite , and for a general overview see kleinert @xcite and schulman @xcite ) .
the power of path - integration for finance ( @xcite,@xcite ) lies in its ability to naturally account for payoffs that are path - dependent .
this makes path integration the method of choice to treat one of the most challenging types of derivatives , the path - dependent options .
feynman and kleinert @xcite showed how quantum - mechanical partition functions can be approximated by an effective classical partition function , a technique which has been successfully applied to the pricing of path - dependent options ( see ref . @xcite and references therein , and refs .
@xcite for recent applications ) .
there exist many different types of path - dependent options .
the two types which are considered in this paper are asian and barrier options .
asian options are exotic path - dependent options for which the payoff depends on the average price of the underlying asset during the lifetime of the option @xcite .
one distinguishes between _ average price _ and _ average strike _ asian options . the average price asian option has been treated in the context of path integrals by linetsky @xcite .
the payoff of an average price is given by @xmath0 and @xmath1 for a call and put option respectively . here
@xmath2 is the strike price and @xmath3 denotes the average price of the underlying asset at maturity @xmath4 .
@xmath3 can either be the arithmetical or geometrical average of the asset price .
average price asian options cost less than plain vanilla options .
they are useful in protecting the owner from sudden short - lasting price changes in the market , for example due to order imbalances @xcite .
average strike options are characterized by the following payoffs : @xmath5 and @xmath6 for a call and put option respectively , where @xmath7 is the price of the underlying asset at maturity @xmath4 .
barrier options are options with an extra boundary condition .
if the asset price of such an option reaches the barrier during the lifetime of the option , the option becomes worthless , otherwise the option has the same payoff as the option on which the barrier has been imposed .
( for more information on exit - time problems see ref . @xcite and the references therein ) in section [ average strike option ] we treat the geometrically averaged asian option . in section
[ 1 ] the asset price propagator for this standard asian option is derived within the path integral framework in a similar fashion as in ref .
@xcite for the weighted asian option .
the underlying principle of this derivation is the effective classical partition function technique developed by feynman and kleinert @xcite . in section
[ 2 ] we present an alternative derivation of this propagator using a stochastic calculus approach .
this propagator now allows us to price both the average price and average strike asian option . for both types of options
this results in a pricing formula which is of the same form as the black - scholes formula for the plain vanilla option .
our result for the option price of an average price asian option confirms the result found in the literature @xcite . for the average strike option no formula of this simplicity exists as far as we know .
our derivation and analysis of this formula is presented in section [ 3 ] , where our result is checked with a monte carlo simulation . in section [ 4 ] we impose a boundary condition on the asian option in the form of a barrier on a control process , and check whether the method used in section [ average strike option ] is still valid when this boundary condition is imposed on the propagator for the normal asian option , using the method of images . finally in section [ 5 ] we draw conclusions .
the path integral propagator is used in financial science to track the probability distribution of the logreturn @xmath8 at time @xmath9 , where @xmath10 is the initial value of the underlying asset .
this propagator is calculated as a weighted sum over all paths from the initial value @xmath11 at time @xmath12 to a final value @xmath13 at time @xmath14@xmath15 \,dt\right)\ ] ] the weight of a path , in the black - scholes model , is determined by the lagrangian@xmath16 = \frac{1}{2\sigma^{2}}\left [ \dot { x}-\left ( \mu-\frac{\sigma^{2}}{2}\right ) \right ] ^{2 } \label{black - scholes lagrangiaan}\ ] ] where @xmath17 is the drift and @xmath18 is the volatility appearing in the wiener process for the logreturn @xcite . for asian options ,
the payoff is a function of the average value of the asset .
therefore we introduce @xmath19 as the logreturn corresponding to the average asset price at maturity @xmath4 .
when @xmath3 is the geometric average of the asset price , then @xmath20 is an algebraic average . @xmath21 the key step to treat asian options within the path integral framework is to partition the set of all paths into subsets of paths , where each path in a given subset has the same average @xmath20
. summing over only these paths that have a given average @xmath20 defines the conditional propagator @xmath22 : @xmath23 \,dt\right ) \label{conditionele propagator}\ ] ] this is indeed a partitioning of the sum over all paths:@xmath24 the delta function in the sum @xmath25 over all paths picks out precisely all the paths that will have the same payoff for an asian option .
the calculation of @xmath26 is straightforward ; when the delta function is rewritten as an exponential , @xmath27 + \frac{1}{t}ikx(t)\right ) dt\right ) , \ ] ] the resulting lagrangian is that of a free particle in a constant force field in 1d .
the resulting integration over paths is found by standard procedures @xcite : @xmath28 ^{2}\right .
\nonumber \\ & \left .
-\frac{6}{\sigma^{2}t}\left ( \bar{x}_{t}-\frac{x_{t}}{2}\right ) ^{2}\right \ } , \label{conditionele propagator uitgerekend}\ ] ] and corresponds to the result found by kleinert @xcite and by linetsky @xcite . the conditional propagator @xmath29 is interpreted in the framework of stochastic calculus as the joint propagator @xmath30 of @xmath31 and its average @xmath20 .
the calculation of @xmath32 here is similar to the derivation presented in ref .
@xcite where this joint propagator is calculated for the vasicek model .
the main point is that in a gaussian model the joint distribution of the couple @xmath33 has to be gaussian too . as a consequence this joint distribution
is fully characterized by the expectation values and the variances of @xmath31 and @xmath20 and by the correlation between these two processes .
the expectation value of @xmath34 is given by @xmath35 , its variance by @xmath36 and the correlation between the two processes by @xmath37 .
the density function of such a gaussian process is then known to be@xmath38 ^{2}+3\left [ \bar { x}_{t}-\left ( \mu-\frac{\sigma^{2}}{2}\right ) \frac{t}{2}\right ] ^{2}\right .
\nonumber \\ & \left .
\left . -3\left
[ x_{t}-\left ( \mu-\frac{\sigma^{2}}{2}\right ) t\right ] \left [ \bar{x}_{t}-\left ( \mu-\frac{\sigma^{2}}{2}\right ) \frac{t}{2}\right ] \right \ } \right)\end{aligned}\ ] ] this agrees with eq .
( [ conditionele propagator uitgerekend ] ) for @xmath22 . if the payoff at time @xmath4 of an asian option is written as @xmath39 , then the expected payoff is@xmath40 = { \displaystyle \int \limits_{-\infty}^{\infty } } dx_{t}{\displaystyle \int \limits_{-\infty}^{\infty } } d\bar{x}_{t}\text { } v_{t}^{asian}(x_{t},\bar{x}_{t})\mathcal{k}\left ( x_{t},t\ , \left \vert 0,0\right \vert \bar{x}_{t}\right ) \label{algemene vorm waarde van de aziatische optie}\ ] ] the price of the option , @xmath41 is the discounted expected payoff , @xmath42 \label{6}\ ] ] where @xmath43 is the discount ( risk - free ) interest rate . using expression ( [
algemene vorm waarde van de aziatische optie ] ) the price of any option which is dependent on the average of the underlying asset during the lifetime of the option can be calculated .
we will now derive the price of an average strike geometric asian call option explicitly . in order to do this , expression ( [ algemene vorm waarde van de aziatische optie ] ) has to be evaluated using the payoff : @xmath44 substituting ( [ payoff ] ) in ( [ 6 ] ) yields@xmath45 where the lower boundary of the @xmath31 integration now depends on @xmath46 .
when considering an average price call , the payoff ( for a call option ) is @xmath0 leading to a constant lower boundary @xmath47 for the @xmath20 integration , and the integrals are easily evaluated . in the present case
however , the integration boundary is more complicated and it is more convenient to express this boundary through a heaviside function , written in its integral representation : @xmath48 now the two original integrals have been reduced to gaussians at the cost of inserting a complex term in the exponential .
expression ( [ waarde van de aziatische optie 1 ] ) can be split into two terms denoted @xmath49 and @xmath50 , where@xmath51 ^{2}\right .
\nonumber \\ & \left .
-\frac{6}{\sigma^{2}t}\left ( \bar{x}_{t}-\frac{x_{t}}{2}\right ) ^{2}+i\left ( x_{t}-\bar{x}_{t}\right ) \tau+x_{t}\right \}\end{aligned}\ ] ] and @xmath50 has the same form , except with @xmath20 instead of @xmath31 in the last term of the argument of the exponent . as a first step , the gaussian integrals over @xmath31 and @xmath20 are calculated , yielding@xmath52 with@xmath53\ ] ] now the integral has been reduced to a form which can be rewritten by making use of plemelj s formulae . taking into account symmetry ,
this reduces to@xmath54\ ] ] with@xmath55{l}\smallskip a=\dfrac{\sigma^{2}t}{6}\\ \smallskip b=\left ( \mu+\dfrac{\sigma^{2}}{2}\right ) \dfrac{t}{2}\end{array } \right.\ ] ] the first term thus becomes@xmath56 + 1\right \}\ ] ] the second term , @xmath57 is evaluated similarly , leading to @xmath58 + 1\right \ } \right .
\nonumber \\ & \left .
-\exp \left [ \left ( \mu-\frac{\sigma^{2}}{6}\right ) \frac{t}{2}\right ] \left \
{ \operatorname{erf}\left [ \sqrt{\frac{3t}{8\sigma^{2}}}\left ( \mu-\frac{\sigma^{2}}{6}\right ) \right ] + 1\right \ } \right)\end{aligned}\ ] ] using the cumulative distribution function of the normal distribution @xmath59\ ] ] this can be rewritten in a more compact form as@xmath60 with the following shorthand notations@xmath55{c}\smallskip d_{1}=\sqrt{\dfrac{3t}{4\sigma^{2}}}\left ( \mu+\dfrac{\sigma^{2}}{2}\right ) \\
\smallskip d_{2}=\sqrt{\dfrac{3t}{4\sigma^{2}}}\left ( \mu-\dfrac{\sigma^{2}}{6}\right ) \end{array } \right.\ ] ] expression ( [ priceform ] ) is the analytic pricing formula for an average strike geometric asian call option , obtained in the present work with the path integral formalism . to the best of our knowledge ,
no pricing formula of this simplicity exists . to check this formula
, we compared its results to those of a monte carlo simulation .
the monte carlo scheme used is as follows @xcite : first , the evolution of the logreturn is simulated for a large number of paths .
this evolution is governed by a discrete geometric brownian motion for a number of time steps .
using the value for the logreturn at each time step , the average logreturn can be calculated for every path .
subsequently the payoff per path can be obtained , which is then used to calculate the option price by averaging over all payoffs per path en discounting back in time . the analytical result and
the monte carlo simulation agree to within a relative error of 0.3% when 500 000 samples and 100 time steps are used .
this means that our analytical result lies within the error bars at every point .
we also obtained the result for an average price asian option ; in contrast to the new result for the average strike option this could be compared to the existing formula @xcite , and was found to be the same .
[ h ] figure1.eps using the propagator ( [ propagator voor aziaat met barriere ] ) the price of an asian option with a barrier @xmath78 can be calculated .
the general pricing formula is given by:@xmath79 this calculation was done for an average price option : @xmath80 .
the calculation , though rather cumbersome , is essentially the same as for the asian options in section [ average strike option ] .
the integral over @xmath31 is a gaussian integral , and the remaining two integrals can be transformed into a standard bivariate cumulative normal distribution , defined by:@xmath81=\frac{1}{2\pi \sqrt{1-\chi^{2}}}\int_{-\infty}^{a}\int_{-\infty } ^{b}\exp \left ( -\frac{1}{2\left ( 1-\chi^{2}\right ) } \left ( x^{2}+y^{2}-2\chi xy\right ) \right ) dxdy \label{cumnormdistr}\ ] ] this eventually leads to the following pricing formula for an asian option with a barrier:@xmath82 } \left ( \frac{b}{s_{0y}}\right ) ^{\frac{2\left [ \frac{4}{\xi } \left ( \nu-\frac{\xi^{2}}{2}\right ) -3\frac{\rho}{\sigma}\left ( \mu -\frac{\sigma^{2}}{2}\right ) \right ] } { \xi \left ( 4 - 3\rho^{2}\right ) } } n\left ( d_{5},d_{6},-\sqrt{\frac{3}{4}}\rho \right ) \nonumber \\ & \left .
+ k~e^{\frac{3}{\sigma^{2}}x_{s}\left [ \frac{2x_{s}}{t}+\left ( \mu-\frac{\sigma^{2}}{2}\right ) \right ] } \left ( \frac{b}{s_{0y}}\right ) ^{\frac{2\left [ \frac{4}{\xi}\left ( \nu-\frac{\xi^{2}}{2}\right ) -3\frac{\rho}{\sigma}\left ( \mu-\frac{\sigma^{2}}{2}\right ) \right ] } { \xi \left ( 4 - 3\rho^{2}\right ) } } n\left ( d_{7},d_{8},-\sqrt{\frac{3}{4}}\rho \right ) \right ] \label{optieprijs aziaat met barriere}\ ] ] where the following shorthand notations were used:@xmath83{l}d_{1}=-\dfrac{\ln \left ( \dfrac{k}{s_{0x}}\right ) -\dfrac{t}{2}\left ( \mu+\dfrac{\sigma^{2}}{6}\right ) } { \sqrt{\dfrac{\sigma^{2}t}{3}}}\\ d_{2}=\dfrac{\ln \left ( \dfrac{b}{s_{0y}}\right ) -t\left ( \nu-\dfrac{\xi ^{2}}{2}+\dfrac{\sigma \xi \rho}{2}\right ) } { \sqrt{\xi^{2}t}}\\ d_{3}=-\dfrac{\ln \left ( \dfrac{k}{s_{0x}}\right ) -\dfrac{t}{2}\left ( \mu-\dfrac{\sigma^{2}}{2}\right ) } { \sqrt{\dfrac{\sigma^{2}t}{3}}}\\ d_{4}=\dfrac{\ln \left ( \dfrac{b}{s_{0y}}\right ) -t\left ( \nu-\dfrac{\xi ^{2}}{2}\right ) } { \sqrt{\xi^{2}t}}\\ d_{5}=-\dfrac{\ln \left ( \dfrac{k}{s_{0x}}\right ) -t\left [ 2\dfrac{x_{s}}{t}+\dfrac{1}{2}\left ( \mu+\dfrac{\sigma^{2}}{6}\right ) \right ] } { \sqrt{\dfrac{\sigma^{2}t}{3}}}\\ d_{6}=\dfrac{\ln \left ( \dfrac{b}{s_{0y}}\right ) -\dfrac{t}{\sigma}\left [ 3\xi \rho \dfrac{x_{s}}{t}+\sigma \dfrac{x_{s}}{t}+\sigma \left ( \nu-\dfrac { \xi^{2}}{2}+\dfrac{\sigma \xi \rho}{2}\right ) \right ] } { \sqrt{\xi^{2}t}}\\ d_{7}=-\dfrac{\ln \left ( \dfrac{k}{s_{0x}}\right ) -t\left [ \dfrac{2x_{s}}{t}+\dfrac{1}{2}\left ( \mu-\dfrac{\sigma^{2}}{2}\right ) \right ] } { \sqrt{\dfrac{\sigma^{2}t}{3}}}\\ d_{8}=\dfrac{\ln \left ( \dfrac{b}{s_{0y}}\right ) -\dfrac{t}{2\sigma}\left [ \dfrac{1}{t}\left ( 6\rho x_{s}\xi+2\sigma y_{s}\right ) + 2\sigma \left ( \nu-\dfrac{\xi^{2}}{2}\right ) \right ] } { \sqrt{\xi^{2}t}}\end{array}\ ] ] fig . ( [ versch_corr_artikel ] ) shows the option price for an asian option with a barrier as a function of the initial asset price belonging to the @xmath63 process , defined by : @xmath84 .
[ h ] figure2.eps this figure shows that the analytical result derived in section [ derivation of the option price ] deviates from the monte carlo simulation with increasing correlation .
the approximate nature of our approach can be understood as follows .
the essence of the approach presented here is that to calculate the price of asian barrier options , two steps need to be taken .
first , a partitioning of paths according to the average along the path must be performed , and second , the method of images must be used in order to cancel out paths which have reached the barrier .
the difficulty combining these two steps , is that mirror paths have a different average than the original paths , and thus belong to a different partition .
this difficulty can apparently be overcome by treating the average itself as a separate , correlated process ( as proposed in ref .
this procedure , relating @xmath85 to @xmath32 , leads to the correct propagator ( and price ) in the case of a plain asian option as shown in section [ 2 ] . however , from the results shown in fig .
[ versch_corr_artikel ] it is clear that this is no longer the case for an asian option with a barrier on a correlated control process .
this is because the exact average of the @xmath62 process does not behave as a separate , correlated process ( the average described by this process is henceforth called the approximate average ) .
this approach is exact for a plain asian option , where all paths contribute , but when a barrier is implemented using the method of images , and thus eliminating some of the paths , the following approximation is made .
when the @xmath63 process hits the barrier and is thus eliminated , its corresponding @xmath62 and @xmath86 processes are eliminated as well .
but the @xmath86 process considered in our derivation is only approximate , so the wrong @xmath86 paths are eliminated .
the central question is whether this will lead to a difference between the distribution of contributing paths for the exact averages and the corresponding distribution for the approximate averages , when a barrier has been implemented .
figure 3 shows that this is indeed the case , and that this difference increases when correlation increases .
[ h ] figure3.eps when the correlation is zero , the paths which are eliminated for both the exact and the approximate average are randomly distributed ( because the behavior of @xmath86 has nothing to do with the behavior of @xmath63 ) , which means that both distributions remain the same gaussian as they would be without a barrier .
this is the reason why our result is exact when correlation is zero .
another source of approximation lies in the use of the black - scholes model which has well - known limitations @xcite .
several other types of market models propose to overcome such limitations , for example by introducing additional ad hoc stochastic variables @xcite or by improving the description of the behavior of buyers / sellers @xcite .
the extension of the present work to for example the heston model lies beyond the scope of this article .
in this paper , we derived a closed - form pricing formula for an average price as well as an average strike geometric asian option within the path integral framework .
the result for the average price asian option corresponds to that found by linetsky @xcite , using the effective classical partition function technique developed by feynman and kleinert @xcite .
the result for the average strike asian option was compared to a monte carlo simulation .
we found that the agreement between the numerical simulation and the analytical result for an average strike asian option is such that they coincide to within a relative error of less than 0.3 % for at least 500 000 samples and 100 time steps .
furthermore , a pricing formula for an asian option with a barrier on a control process was developed .
this is an asian option with the additional condition that the payoff is zero whenever the value of @xmath87the control process crosses a certain predetermined barrier .
the pricing of this option was performed by constructing a new propagator which consisted of a linear combination of two propagators for a regular asian option .
the resulting pricing formula is exact when the correlation is zero , and is approximate when the correlation increases .
the central approximation made in our derivation , is that the process for the average logreturn @xmath86 is treated as a stochastic process , which is correlated with the process of the logreturn @xmath62 .
this assumption is correct whenever all price - paths contribute to the total sum , but becomes approximate when a boundary condition is applied .
the authors would like to thank dr .
sven foulon and prof .
karel in t hout for the fruitful discussions .
this work is supported financially by the fund for scientific research - flanders , fwo project g.0125.08 , and by the special research fund of the university of antwerp bof noi ua 2007 . d. lemmens , m. wouters , j. tempere , s. foulon , _ a path integral approach to closed - form option pricing formulas with applications to stochastic volatility and interest rate models , _ physical review .
e 78 , 016101 ( 2008 ) . | we derive a closed - form solution for the price of an average price as well as an average strike geometric asian option , by making use of the path integral formulation .
our results are compared to a numerical monte carlo simulation .
we also develop a pricing formula for an asian option with a barrier on a control process , combining the method of images with a partitioning of the set of paths according to the average along the path .
this formula is exact when the correlation is zero , and is approximate when the correlation increases . |
the generalization of renormalizable relativistic gauge theories to the standard model ( sm ) was very successful and has been confirmed experimentally in an impressive way , including detailed tests of radiative corrections . , @xmath0 and @xmath1 in the neutrino sector may be the first signs of physics beyond the sm @xcite .
] however , it is important to keep in mind that the mechanism of electroweak ( ew ) symmetry breaking is still mostly untested .
the postulated higgs particle has so far not been observed and there is only indirect evidence from quantum corrections that a sm higgs boson should be lighter than about @xmath2 @xcite .
the higgs sector has furthermore well - known theoretical problems , especially the gauge hierarchy problem , which strongly suggest that new physics exists which is connected to the mechanism of ew symmetry breaking .
whatever the correct symmetry breaking mechanism is , it must satisfy by now a number of stringent direct and indirect constraints . given the success of the sm it is , however , immediately clear how an alternative symmetry breaking scenario can be consistent with data . in the limit
where new physics decouples , it just has to reproduce effectively the sm higgs sector with a light higgs particle @xcite .
if the model has such a decoupling limit , as in the case discussed in this paper , then deviations from the sm can be understood as a departure from the decoupling limit . motivated by the evidence for neutrino masses , we discuss the possibility that the ew symmetry is broken dynamically by a neutrino condensate .
this would normally lead to neutrino masses of the order of the symmetry breaking scale , i.e. @xmath3 .
neutrinos may , however , possess both dirac and majorana mass terms and the dynamical generation of large dirac mass terms leads via the see - saw mechanism @xcite to small , phenomenologically acceptable neutrino masses .
a composite higgs particle will emerge that is not affected by the see - saw mechanism , i.e. it will have a mass of the order of the ew symmetry breaking scale .
the low - energy effective lagrangian in the decoupling limit is therefore the sm , with a composite higgs instead of a fundamental scalar .
a heavy dirac neutrino mass is similar to the heavy top mass of the order of the ew scale , which gave rise to speculations that top condensation might be responsible for ew symmetry breaking @xcite .
however , top condensation is not viable in its simplest version , since it predicts too large top and higgs masses .
different non - minimal models are in principle viable @xcite , and the possibility that third - generation neutrinos contribute to top condensation was studied by martin @xcite .
the condensation of a full fourth generation ( including neutrinos of the fourth generation ) @xcite was also discussed in this context .
we study the case where only neutrinos are responsible for the dynamical breakdown of the ew symmetry .
we briefly discuss a mixed case as well where combined neutrino and top condensation leads to an effective two - higgs scenario with a leptonic and a hadronic higgs particle .
the paper is organized as follows : in the next section we discuss the condensation of neutrinos , i.e. we study the relevant system of coupled gap equations in combination with the see - saw mechanism in the proper mass eigenstate basis .
afterwards , we solve the gap numerically .
the following section contains the phenomenology and the predictions arising from the renormalization group improved compositeness conditions .
section [ sec:3neutrinos ] contains a short discussion of the option that all three generations of neutrinos condense simultaneously and in section [ sec:2higgs ] we outline briefly the possibility of a combined neutrino - top condensation scenario .
we assume as mentioned that some physics exists at high energies which yields an effective four - fermion picture similar to weak interactions at low energies .
however , contrary to weak interactions we assume that certain four - fermion couplings become strong enough to trigger the formation of condensates , thus giving masses to some of the fermions via gap equations .
the remaining fermions could e.g. obtain masses from further four - fermion couplings , which only subdominantly contribute to the gap . for top condensation
it was shown how this can be justified in the context of broken renormalizable gauge theories at high energies , for example in the framework of so - called top - color models @xcite or u(1 ) models @xcite . in this spirit , we consider the particle content of the sm extended by three right - handed neutrinos but without a fundamental higgs field . instead of the sm higgs field
we assume four - fermion couplings involving the lepton doublets and the right - handed neutrinos .
in addition , since the right - handed neutrinos are singlets under the sm gauge group and since there is no protective symmetry , we assume them to have huge majorana masses . in order to show the essential aspects of such a scenario , we consider first the case where only one of the four - fermion couplings drives the condensation , while the others vanish .
thus , the four - fermion lagrangian is @xmath4 where we have omitted the @xmath5 indices and where @xmath6 , @xmath7 stand for the relevant neutrino degrees of freedom .
moreover , we assume the majorana mass matrix to be diagonal so that the condensing pair of neutrinos can be studied independently .
therefore , we need to consider only one majorana mass term , @xmath8 we will see that this describes the most interesting features of neutrino condensation .
more general scenarios will be discussed briefly in sections [ sec:3neutrinos ] and [ sec:2higgs ] .
the question whether a non - perturbative solution for the ground state exists in the presence of the huge majorana mass , i.e. if the gap equation has a non - trivial solution , will be studied in section [ sec : gap ] .
if the gap equation produces a fermion condensate which is a doublet under @xmath9 and which carries a suitable @xmath10 charge , then it is immediately clear that the @xmath11 gauge symmetry is broken . for the chosen njl - like
interaction a composite higgs particle emerges .
this is visualized in fig .
[ fig : bubblesumhiggsexchange ] , where a massive scalar pole and three massless goldstone bosons are produced by the summation of a certain class of diagrams with dynamical fermion propagators . the higgs mechanism and the `` eating '' of the goldstone bosons
is illustrated in fig .
[ fig : bugglesumgaugebosonmass ] . for more details see e.g. @xcite .
@xmath12{dynsb56.eps } } } } + { \vcenter{\hbox{\includegraphics[scale=0.7]{dynsb57.eps } } } } + { \vcenter{\hbox{\includegraphics[scale=0.7]{dynsb58.eps } } } } + \dots & = & { \vcenter{\hbox{\includegraphics[scale=0.7]{dynsb59.eps } } } } \end{aligned}\ ] ] [ f1 ] @xmath13{opigaugebosonpropagator.eps } } } } = } \\ & = & { \vcenter{\hbox{\includegraphics[scale=0.7]{dynsb89.eps } } } } + { \vcenter{\hbox{\includegraphics[scale=0.7]{dynsb90.eps } } } } + { \vcenter{\hbox{\includegraphics[scale=0.7]{dynsb91.eps } } } } + \dots \\[2 mm ] & = & { \vcenter{\hbox{\includegraphics[scale=0.7]{dynsb89.eps } } } } + { \vcenter{\hbox{\includegraphics[scale=0.7]{dynsb92.eps}}}}\end{aligned}\ ] ] [ f2 ] we analyze now the gap equations of our model with an explicit majorana mass term for @xmath7 .
if a non - trivial solution produces dynamically a large dirac mass term , it breaks the ew symmetry .
then the presence of the huge singlet majorana mass will lead to a dynamical see - saw mechanism with small neutrino masses .
a computation of the gap equation in the basis of mass eigenstates must therefore include in a self - consistent way the possibility of a dynamically generated dirac mass term , @xmath14 for any value of @xmath15 , the mass eigenstates are two majorana fermions , given by @xmath16 with @xmath17 and @xmath18 and the orthogonal matrix @xmath19 the corresponding mass eigenvalues [ eq : masseigenvalues ] @xmath20 are related to @xmath21 by @xmath22 for convenience , we rewrite the singlet majorana mass term as well as the neutrino part of the four - fermion interaction in terms of majorana fermions , @xmath23 the feynman rules for the interactions of the mass eigenstates are derived from these lagrangians by inserting the relations of eq . .
the gap equations for the masses @xmath24 and @xmath25 , corresponding to fig .
[ fig : gap ] , are [ eq : gapequations ] @xmath26 + s^2\,m\;,\\ m_\rho & = & 2 \,g^{(\nu ) } \,c^2\,s^2\,\left [ m_\rho\,i_\mathrm{gap}(m_\rho ) -m_\lambda\,i_\mathrm{gap}(m_\lambda ) \right ] + c^2\,m \;,\end{aligned}\ ] ] where we have introduced @xmath27 \;.\ ] ] @xmath28 is the condensation scale , which acts as a cutoff . as @xmath29 due to eq .
, the gap equations are linearly dependent , so that it is sufficient to solve one of them .
note that since the mass eigenvalues are given by @xmath24 and @xmath25 , non - trivial solutions also imply a dynamically generated dirac mass @xmath15 .
such solutions are found to exist indeed , as will be shown in section [ sec : numgap ] .
remarkably , in this kind of gap equations , the heavy majorana degree of freedom plays an essential role and does not decouple . @xmath30 { \vcenter{\hbox{\includegraphics{dynsb47.eps } } } } \end{array } + { \vcenter{\hbox{\includegraphics{dynsbs1.eps } } } } + { \vcenter{\hbox{\includegraphics{dynsbs2.eps } } } } $ ] + @xmath31 { \vcenter{\hbox{\includegraphics{dynsb48.eps } } } } \end{array } + { \vcenter{\hbox{\includegraphics{dynsb50.eps } } } } + { \vcenter{\hbox{\includegraphics{dynsb51.eps } } } } $ ] to check the self - consistency of our calculation , we consider the gap equation for a bilinear term that contains the fields @xmath32 and @xmath33 as shown in fig .
[ fig : consistency ] .
as we are working in the mass eigenstate basis , this term , which corresponds to an off - diagonal entry @xmath34 in the mass matrix , has to vanish identically .
we obtain @xmath35\,c^2\,s^2\ , \left [ m_\lambda\,i_\mathrm{gap}(m_\lambda)-m_\rho\,i_\mathrm{gap}(m_\rho ) \right ] + c\,s\,m \nonumber\\ & \ ! = & \tfrac{1}{2}\,\cot\varphi\,\left[m_\lambda - s^2\,m\right ] + \tfrac{1}{2}\,\tan\varphi\,\left[m_\rho - c^2\,m\right ] + c\,s\,m \;.\end{aligned}\ ] ] using the gap equations as well as the relations and for the mass eigenvalues , it follows that @xmath34 vanishes as required .
this confirms that our solution is self - consistent .
@xmath36 { \vcenter{\hbox{\includegraphics{dynsb52.eps } } } } \end{array } + { \vcenter{\hbox{\includegraphics{dynsb43.eps } } } } + { \vcenter{\hbox{\includegraphics{dynsb44.eps } } } } $ ] the gap equation for @xmath24 can be considered as an equation for @xmath37 and @xmath15 , if fixed values are assigned to @xmath38 and @xmath28 . for @xmath39 and @xmath40 ,
the solution is shown in fig .
[ fig : majgapnumsol ] . instead of @xmath37
, we plot the dimensionless coupling constant @xmath41 .
@xmath42{majoranagap.eps}}}}$ ] we find non - trivial solutions for @xmath15 , if the coupling @xmath43 is larger than a critical value .
this result is quite similar to top condensation @xcite , even though the right - handed neutrino has a large majorana mass .
note , however , that in order to obtain correct solutions , the exact relations for the mass eigenvalues have to be used rather than an expansion in powers of @xmath44 . in order to obtain a dirac mass of the order of the ew scale , fine - tuning is required in the bubble sum approximation , as can be seen from the extremely small slope of the graph in fig .
[ fig : majgapnumsol ] . even if the same fine - tuning is present in the exact gap equation , loop corrections , which destabilize the hierarchy in the usual perturbative framework of the sm ,
do not pose an additional problem here .
in other words , the dynamical scenario under consideration can not explain why the hierarchy is large , but it does explain why it remains large . in that sense the hierarchy is stable and hence the model might even be considered to be on equal footing with the solution of the hierarchy problem by supersymmetry .
the above results for @xmath25 and @xmath24 were obtained in the bubble approximation . in order to obtain more reliable low - energy results ,
the renormalization group ( rg ) running of the effective theory has to be taken into account . following the procedure used in @xcite ,
we show first that the low - energy effective theory of the model with four - fermion interactions is just the sm with right - handed neutrinos .
, where all non - critical four - fermion couplings vanish . by including such couplings , the missing yukawa interactions and masses can easily be incorporated . ]
therefore , we look at the original lagrangian at the condensation scale @xmath28 , @xmath45 where @xmath46 contains kinetic terms for the gauge bosons , the usual sm fermions and the right - handed neutrinos .
there is no higgs field present .
we rewrite @xmath47 in terms of a static ( non - propagating ) scalar auxiliary field @xmath48 , @xmath49 this lagrangian is equivalent to the lagrangian in eq . ,
which can be seen by exploiting the equations of motion for the auxiliary field , @xmath50 the same result can be found by integrating out the auxiliary field in the path integral formalism . at scales below @xmath28
, the dynamics of the theory will induce all renormalizable and gauge invariant terms that are allowed by symmetries , including a kinetic term and a quartic self - interaction for @xmath48 .
thus , the lagrangian becomes @xmath51 note that @xmath52 , @xmath53 and @xmath54 are running quantities , even though their dependence on the energy scale @xmath55 is not written explicitly . for @xmath56
the lagrangian has to become identical to the one of eq . , which leads to the following boundary ( compositeness ) conditions for the rg evolution : [ eq : compositenessunnorm ] @xmath57 eq .
is already very similar to the sm lagrangian .
the auxiliary field has acquired a kinetic term , i.e. it has become a propagating composite higgs field , but its lagrangian is not yet written in the usual normalization where @xmath58 . to fix this
we perform the rescaling @xmath59 which leads to the lagrangian @xmath60 where we have defined @xmath61 thus , we have recovered the sm extended by right - handed neutrinos , which proves that the lagrangian yields exactly the same low - energy physics , but with additional constraints on the parameters , namely the compositeness conditions . from eqs .
and we find that the compositeness conditions for the couplings and the mass parameter become [ eq : compositenessnorm ] @xmath62 the low - energy neutrino and higgs masses are now obtained from the rg equations for the sm extended by right - handed neutrinos . for the rg analysis in see - saw models , it is crucial to integrate out the right - handed neutrinos and the corresponding part of the neutrino yukawa coupling matrix at the mass thresholds .
the computation of the rg evolution requires the @xmath63-functions of all gauge and yukawa couplings ( including those of the neutrinos ) , of the quartic higgs coupling and of the dimension 5 neutrino mass operator ( see e.g. @xcite ) . for completeness
, we have listed the relevant @xmath63-functions in the appendix .
when integrating out the right - handed neutrinos , the dimension 5 neutrino mass operator has to be matched at each threshold @xcite . in this framework
, we calculate the low - energy parameters , starting with the compositeness conditions at the condensation scale and solving the relevant systems of coupled differential equations .
the results for the higgs mass and for the mass of the neutrino participating in condensation turn out to be not very sensitive to the exact boundary conditions and due to the quasi - fixed - point behavior that arises once eq .
is satisfied . the rg evolution of the majorana mass of the light neutrino participating in condensation is shown in fig . [ fig : neutrinoyukawacoupling ] .
it can be seen that a large , non - perturbative neutrino yukawa coupling is in agreement with the current limits on neutrino masses , if the see - saw scale @xmath38 is large enough .
since the neutrino yukawa coupling at the condensation scale has to be non - perturbatively large and is thus not a free parameter of the model , an allowed range for the neutrino mass translates into an allowed range for @xmath38 .
for instance , with @xmath64 $ ] , @xmath65 $ ] at the condensation scale , which we have chosen to be @xmath66 , we obtain a range @xmath67 $ ] for the neutrino mass at low energy . except for the neutrino and the top ,
the yukawa couplings of the fermions have been omitted .
@xmath42{letterdynsb1_1.eps}}}}$ ] the higgs mass can be predicted due to the quasi - fixed point in the rg evolution of the higgs self - coupling .
the running of the higgs mass from the condensation scale @xmath28 to the ew scale is shown in fig .
[ fig : higgsmassfp ] for a wide range of parameters @xmath38 , @xmath68 and @xmath32 at the scale @xmath28 . for @xmath40
, we obtain higgs masses in the range @xmath69 .
one should keep in mind that this prediction depends on the condensation scale @xmath28 , which is a free parameter in our model . as in section
[ sec : numass ] , except for the neutrino and the top , the yukawa couplings of the fermions have been omitted .
the model discussed so far can be extended to the case where all three generations of neutrinos participate in the condensation . in general ,
the low - energy theory contains several higgses .
however , if the four - fermion couplings satisfy a `` factorization relation '' @xcite , @xmath70 then by following the steps of eqs .
( [ eq:4flag1h ] ) ( [ eq : hence ] ) , the lagrangian can be rewritten in terms of _ one _ auxiliary field @xmath71 .
hence we obtain again a one - higgs model after condensation analogous to section [ sec : laguebergang ] .
the gap equations can be treated as in section [ sec : neutrinocondensation ] .
if three neutrinos condense , the infrared quasi - fixed point of the rg evolution leads to three heavy dirac mass eigenvalues . the majorana mass term of the right - handed neutrino in eq . must now be generalized to a mass matrix @xmath72 for three right - handed neutrinos , with entries which are unprotected by symmetries , leading to three heavy eigenvalues .
the degeneracy or hierarchy of the see - saw scales is then conveyed rather directly into the full light neutrino mass pattern .
the higgs mass is almost unchanged compared to the results of section [ sec : higgsmass ] , in spite of the contribution from the yukawa couplings of the additional neutrinos .
using the same input parameters as in fig .
[ fig : higgsmassfp ] , now with three equally large dirac masses for the neutrinos , we find @xmath73 .
besides the neutrino condensate discussed in section [ sec : neutrinocondensation ] , there can be a further condensate connected to the top quark .
this means that in addition to the four - fermion coupling of eq . for the neutrino
, there is a corresponding term for the top quark , @xmath74 , and a mixed term @xmath75 . if @xmath76 and @xmath77 become critical , then in general two independent condensates form , similar to the case of combined top and bottom condensation @xcite .
note that for @xmath78 this is a one - higgs scenario , which coincides with the model discussed by martin @xcite .
however , in general we are dealing with an effective two higgs doublet model ( 2hdm ) , where the effective lagrangian does not coincide with the 2hdms usually discussed , which are phenomenologically severely constrained . the scalars which are generated dynamically here are a `` leptonic higgs '' , @xmath79 , and a `` hadronic higgs '' , @xmath80 .
the mixed coupling @xmath77 leads to a term of the form @xmath81 in the effective higgs potential .
besides , four - higgs interactions with odd numbers of @xmath79 and @xmath80 become allowed in the low - energy effective theory .
an interesting feature of 2hdms with a leptonic and a hadronic higgs is that the correct top mass can now be obtained dynamically as a quasi - fixed point of the rg evolution .
this has been found to be impossible in minimal models with pure top condensation , where the predicted top mass is too large . in the two - higgs case ,
the necessary additional degree of freedom is the ratio of the vevs , @xmath82 .
a top mass of approximately @xmath83 is obtained for @xmath84 , as illustrated in fig .
[ fig:2hdmtopyukawacoupling ] .
this also yields large dirac masses for the neutrinos , but due to the see - saw suppression their physical masses remain tiny .
@xmath42{letterdynsb1_2.eps}}}}$ ] another issue that has to be addressed is the higgs mass spectrum , as massless or very light higgses would be in conflict with experiments .
as suggested by the analysis of the case with a top and a bottom condensate @xcite , it should be possible to obtain phenomenologically viable higgs masses by making the parameter @xmath85 sufficiently large .
hence , it seems worthwhile to study 2hdms with neutrino and top condensates in more detail .
in this article we have studied the possibility that the ew symmetry is broken dynamically by the formation of a neutrino condensate .
we have started from the sm with right - handed neutrinos , but without a fundamental higgs field . instead of the higgs field we have postulated strong attractive four - fermion interactions .
in addition , we have included huge majorana masses for the right - handed neutrinos , since there is no protective symmetry .
the analysis of the coupled dirac - majorana gap equations has shown that non - trivial solutions exist , which dynamically produce a large dirac neutrino mass and which break the ew symmetry in the desired way . however , due to the presence of the huge majorana mass term we obtain physical neutrino masses in the phenomenologically allowed region via the see - saw mechanism .
we have illustrated in the auxiliary field formalism that the njl - type interactions lead to a composite scalar sector which resembles a higgs lagrangian with certain boundary conditions , i.e. predictions .
in particular , the compositeness conditions require that the yukawa coupling @xmath68 of the condensing neutrino becomes non - perturbative at high energies . to evaluate the predictions , we have translated these boundary conditions in the framework of the low - energy effective lagrangian into high - energy boundary conditions for the renormalization group running . for our minimal model discussed in sections [ sec : neutrinocondensation ] and [ sec : phenomenology ]
the effective lagrangian is the sm with massive neutrinos .
the boundary conditions and the infrared quasi - fixed - point behavior of the higgs self - coupling and of the relevant yukawa coupling @xmath68 lead to two predictions for a given condensation scale @xmath28 . for @xmath40 , typical values of @xmath86 and @xmath87 consistent with the boundary conditions , and for
@xmath65 $ ] , we have found a higgs mass in the range @xmath69 , which is in agreement with current experimental bounds .
moreover , we have found that an upper bound for the neutrino mass of the order of @xmath88 ( @xmath89 ) translates into a lower bound for the see - saw scale @xmath38 of the order of @xmath90 ( @xmath91 ) .
we have also briefly studied the possibility that more than one neutrino condenses .
this leads in general to several higgses and under a factorization condition to a one - higgs scenario .
furthermore , we have outlined the possibility of a combined neutrino - top condensation , which corresponds without further assumptions to a two - higgs model .
in addition , it might be possible to extend neutrino condensation to a supersymmetric model .
we have not attempted to embed these scenarios into a larger framework where the four - fermion terms are generated in strongly coupled broken gauge theories as it was done in `` top color '' theories @xcite in the case of top condensation .
this should be interesting , since such a framework would , for example , allow to address the question if the gauge couplings of the extended gauge sector unify above the condensation scale .
in addition , threshold effects near the condensation scale ( in combination with extended gauge sectors or by themselves ) might affect unification .
such threshold corrections are generally expected to be large in this non - perturbative scenario . to conclude , we have introduced a dynamical realization of electroweak symmetry breaking with massive neutrinos , where the sm higgs particle emerges from neutrino condensation , leading to predictions for the higgs mass and for the see - saw scale .
this work was supported in part by the `` sonderforschungsbereich 375 fr astro - teilchenphysik der deutschen forschungsgemeinschaft '' .
in the sm extended by right - handed neutrinos one has to consider several effective theories , corresponding to the ranges between the non - degenerate eigenvalues of the majorana mass matrix @xmath38 . at the thresholds ,
the heavy degrees of freedom are successively integrated out .
as introduced in @xcite , a superscript @xmath92 denotes a quantity between the @xmath93th and the @xmath94th mass threshold . with this definition ,
the @xmath63-functions for the yukawa coupling matrices are given by @xmath95 \right\ } \;,\\
16\pi^2 \ , \accentset{(n)}{\beta}_{y_e } & = & y_e \left\ { \vphantom{\operatorname{tr}\left[y_e^\dagger y_e + z_\nu^{(1)}\,\accentset{(n)}{y}_\nu^\dagger \accentset{(n)}{y}_\nu + 3z_d^{(1)}\,y_d^\dagger y_d\ + 3z_u^{(1)}\,y_u^\dagger y_u\right ] } \frac{3}{2 } y_e^\dagger y_e -\frac{3}{2}\ , \accentset{(n)}{y}_\nu^\dagger \accentset{(n)}{y}_\nu - \frac{15}{4 } g_1 ^ 2 - \frac{9}{4 } g_2 ^ 2 \right.\nonumber\\ * & & \hphantom{y_e\left\ { \right . }
\left.{}+ \operatorname{tr}\left[y_e^\dagger y_e + \accentset{(n)}{y}_\nu^\dagger \accentset{(n)}{y}_\nu + 3\,y_d^\dagger y_d\ + 3\,y_u^\dagger y_u\right ] \right\}\ ; , \\ 16\pi^2 \ , \accentset{(n)}{\beta}_{y_d } & = & y_d \left\ { \frac{3}{2 } y_d^\dagger y_d -\frac{3}{2}\ , y_u^\dagger y_u - \frac{5}{12 } g_1 ^ 2 - \frac{9}{4 } g_2 ^ 2 - 8\,g_3 ^ 2 \right.\nonumber\\ * & & \hphantom{y_e\left\ { \right . } \left.{}+ \operatorname{tr}\left[y_e^\dagger y_e + \accentset{(n)}{y}_\nu^\dagger \accentset{(n)}{y}_\nu + 3\,y_d^\dagger y_d + 3\,y_u^\dagger y_u\right ] \right\}\ ; , \\
16\pi^2 \ , \accentset{(n)}{\beta}_{y_u } & = & y_u \left\ { \frac{3}{2 } y_u^\dagger y_u - \frac{3}{2}\ , y_d^\dagger y_d - \frac{17}{12 } g_1 ^ 2 - \frac{9}{4 } g_2 ^ 2 - 8\,g_3 ^ 2 \right.\nonumber\\ * & & \hphantom{y_e\left\ { \right . } \left.{}+ \operatorname{tr}\left[y_e^\dagger y_e + \accentset{(n)}{y}_\nu^\dagger \accentset{(n)}{y}_\nu + 3\,y_d^\dagger y_d + 3\,y_u^\dagger y_u\right ] \right\}\;. \end{aligned}\ ] ] the @xmath63-function for the majorana mass matrix @xmath38 reads @xmath96 and the rg evolution of the quartic higgs self - coupling is determined by - function has not yet been written explicitly in the literature . ]
@xmath97 \\ * & & { } -8\,\operatorname{tr}\left [ y_e^\dagger y_e\,y_e^\dagger y_e + \accentset{(n)}{y}_\nu^\dagger\accentset{(n)}{y}_\nu \,\accentset{(n)}{y}_\nu^\dagger\accentset{(n)}{y}_\nu + 3\,y_d^\dagger y_d\,y_d^\dagger y_d + 3\,y_u^\dagger y_u\,y_u^\dagger y_u \right]\;.\nonumber\end{aligned}\ ] ] finally , the @xmath63-function for the effective neutrino mass operator reads @xmath98 the one - loop @xmath63-functions for the gauge couplings are of course unchanged compared to the sm .
m. lindner , _ the role of a standard model decoupling limit in electro - weak symmetry breaking _ , in _ proceedings of the international workshop on physics beyond the standard model : from theory to experiment ( valencia 97 ) _
( i. antoniadis , l. e. ibanez , and j. w. f. valle , eds . ) , world scientific , singapore ( 1998 ) , 213230 .
m. carena , t. e. clark , c. e. m. wagner , w. a. bardeen , and k. sasaki , _ dynamical symmetry breaking and the top quark mass in the minimal supersymmetric standard model _
, nucl . phys . *
b369 * ( 1992 ) , 3353 . | we show that the electroweak symmetry can be broken in a natural and phenomenologically acceptable way by a neutrino condensate .
therefore , we assume as particle content only the chiral fermions and gauge bosons of the standard model and in addition right - handed neutrinos .
a fundamental higgs field is absent .
we assume instead that new interactions exist that can effectively be described as four - fermion interactions and that can become critical in the neutrino sector .
we discuss in detail the coupled dirac - majorana gap equations which lead to a neutrino condensate , electroweak symmetry breaking and via the dynamical see - saw mechanism to small neutrino masses .
we show that the effective lagrangian is that of the standard model with massive neutrinos and with a composite higgs particle .
the mass predictions are consistent with data . tum - hep-491/02 + desy 02 - 208 * dynamical electroweak symmetry breaking + by a neutrino condensate * + stefan antusch , jrn kersten , manfred lindner + _ physik - department t30 , technische universitt mnchen + james - franck - strae , 85748 garching , germany _
+ michael ratz + _
deutsches elektronensynchrotron desy + 22603 hamburg , germany _ |
damped @xmath9 systems ( dlas ) are fundamentally important , because they contain most of the neutral gas in the universe at all times since cosmological reionization ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
molecular clouds , within which star formation takes place , likely condense out of cold dense neutral atomic gas contained in dlas , evidenced by the fact that the neutral hydrogen ( surface ) density in dlas and molecular hydrogen ( surface ) density in molecular clouds form a continuous sequence ( e.g. , * ? ? ?
* ; * ? ? ?
therefore , dlas likely hold key to understanding the fuel for and ultimately galaxy formation . a substantial amount of theoretical
work has been devoted to studying the nature of dlas ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , since the pioneering investigation of @xcite in the context of the cold dark matter ( cdm ) cosmogony .
a very interesting contrast is drawn between the observationally based inference of simple large disk galaxies possibly giving rise to dlas @xcite and the more naturally expected hierarchical buildup of structures in the cdm cosmogony where galactic subunits may produce some of the observed kinematics of dlas @xcite .
clearly , the implications on the evolution of galaxies in the two scenarios are very different .
we have carried out a set of eulerian adaptive mesh refinement ( amr ) simulations with a resolution of @xmath10kpc proper and a sample size of several thousand galaxies with mass @xmath11 to statistically address the physical nature of dlas in the current standard cosmological constant - dominated cdm model ( lcdm ) @xcite .
mechanical feedback from star formation driven by supernova explosions and stellar winds is modeled by a one - parameter prescription that is physically and energetically sound .
part of the motivation was to complement and cross - check studies to date that are largely based on smooth - particle - hydrodynamics ( sph ) simulations . with the simulation set and detailed analysis performed here this study represents a significant extension of previous works to simultaneously subject the lcdm model to a wider and more complete range of comparisons with observations .
we examine in detail the following properties of dlas in a self - consistent fashion within the same model : dla column density distribution evolution , line density evolution , metallicity distribution evolution , size distribution evolution , velocity width distribution evolution , kinematic structural parameters evolution , neutral mass content evolution and others .
a gallery of dlas is presented to obtain a visual understanding of the physical richness of dla systems , especially the effects of galactic winds and large - scale gaseous structures . in comparison to the recent work of
@xcite we track the metallicity distribution and evolution explicitly and show that the metallicity distribution of dlas is , in good agreement with observations , very wide , which itself calls for a self - consistent treatment of metals transport . in agreement with the conclusions of @xcite ,
although not in the detailed process , we show that galactic winds are _ directly _ responsible for a large fraction of wide dlas at high redshift , by entraining cold clouds to large velocities and causing large kinematic velocity widths .
we find that the simulated @xmath121808 line velocity width , kinematic shape measures and dla metallicity distributions that are all in excellent agreement with observations .
taking all together , we conclude that the standard lcdm model gives a satisfactory account of all properties of dlas . finally , we examine the properties of dla hosts , including their mass , star formation rate , hi content , gas to stellar mass ratio and colors , and show that dlas arise in a variety of galaxies and roughly trace the entire population of galaxies at any redshift .
this may reconcile many apparently conflicting observational evidence of identifying dlas with different galaxy populations . specifically , at @xmath13 we show that @xmath14 of dlas are lyman break galaxies ( lbgs ) @xcite , while the majority arise in smaller galaxies .
the outline of this paper is as follows . in
2 we detail our simulations , method of making galaxy catalogs , method of making dla catalogs and procedure of defining line profile shape measures .
results are presented in 3 . in
3.1 we present a gallery of twelve dlas .
we give line velocity width distribution functions in 3.2 , demonstrating excellent agreement between simulations and observations , particularly at high velocity width end .
3.3 is devoted to the three kinematic measures of the absorption line and we show the model produces results that are consistent with observations . column density distribution and evolution , line density and neutral gas density evolution are described in 3.4 , where , while simulations are consistent with observations , we emphasize large cosmic variance from region to region with different large - scale overdensities .
the focus is shifted to metallicity distribution and evolution in 3.5 and simulations are found to be in excellent agreement with observations where comparisons can be made . the next subsection 3.6 performs a detailed analysis of the size of dlas and finds that the available observed qso pairs with dlas are in accord with the expectation of our model . having found agreement between simulations and observations in all aspects pertinent to dlas , we turn our attention to the properties of dla hosts in 3.7 , where we show that dlas , while slightly favoring galaxies that are more gas rich , less massive and bluer in color , and have higher hi mass and higher gas to stellar mass ratio , roughly trace the entire population of galaxies at all redshifts .
conclusions are given in 4 .
we perform cosmological simulations with the adaptive mesh refinement ( amr ) eulerian hydro code , enzo @xcite .
first we ran a low resolution simulation with a periodic box of @xmath15mpc on a side .
we identified two regions separately , one centered on a cluster of mass of @xmath16 and the other centered on a void region at @xmath6 .
we then resimulate each of the two regions separately with high resolution , but embedded in the outer @xmath17mpc box to properly take into account large - scale tidal field and appropriate boundary conditions at the surface of the refined region .
we name the simulation centered on the cluster
c " run and the one centered on the void v " run .
the refined region for c " run has a size of @xmath18mpc@xmath19 and that for v " run is @xmath20mpc@xmath19 . at their respective volumes , they represent @xmath21 and @xmath22 fluctuations .
the initial condition in the refined region has a mean interparticle - separation of @xmath23kpc comoving , dark matter particle mass of @xmath24 .
the refined region is surrounded by two layers ( each of @xmath25mpc ) of buffer zones with particle masses successively larger by a factor of @xmath26 for each layer , which then connects with the outer root grid that has a dark matter particle mass @xmath27 times that in the refined region . because we still can not run a very large volume simulation with adequate resolution and physics
, we choose these two runs to represent two opposite environments that possibly bracket the average . at redshift @xmath28 , as we will show , the average properties of most quantities concerning dlas in c " and v " runs are not very different , although the abundances of dlas in the two runs are already very different .
it is only at lower redshift where we see significant divergence of some quantities of dlas between the two runs , presumably due to different dynamic evolutions in the two runs .
we choose the mesh refinement criterion such that the resolution is always better than @xmath29pc physical , corresponding to a maximum mesh refinement level of @xmath30 at @xmath6 .
we also ran an additional simulation for c " run with a factor of two lower resolution to assess the convergence of the results which we name
c/2 " run and , as we will show in the appendix , the convergence is excellent for all quantities examined here .
the simulations include a metagalactic uv background @xcite , and a model for shielding of uv radiation by neutral hydrogen @xcite .
they also include metallicity - dependent radiative cooling @xcite .
star particles are created in cells that satisfy a set of criteria for star formation proposed by @xcite .
each star particle is tagged with its initial mass , creation time , and metallicity ; star particles typically have masses of @xmath31@xmath32 .
supernova feedback from star formation is modeled following @xcite .
feedback energy and ejected metal - enriched mass are distributed into 27 local gas cells centered at the star particle in question , weighted by the specific volume of each cell , which is to mimic the physical process of supernova blastwave propagation that tends to channel energy , momentum and mass into the least dense regions ( with the least resistance and cooling ) .
we allow the whole feedback processes to be hydrodynamically coupled to surroundings and subject to relevant physical processes , such as cooling and heating , as in nature .
as we will show later , the extremely inhomogeneous metal enrichment process demands that both metals and energy ( and momentum ) are correctly modeled so that they are transported into right directions in a physically sound ( albeit still approximate at the current resolution ) way .
the primary advantages of this supernova energy based feedback mechanism are three - fold .
first , nature does drive winds in this way and energy input is realistic .
second , it has only one free parameter @xmath33 , namely , the fraction of the rest mass energy of stars formed that is deposited as thermal energy on the cell scale at the location of supernovae .
third , the processes are treated physically , obeying their respective conservation laws ( where they apply ) , allowing transport of metals , mass , energy and momentum to be treated self - consistently and taking into account relevant heating / cooling processes at all times .
we use @xmath34 in these simulations . the total amount of explosion kinetic energy from type ii supernovae with a chabrier imf translates to @xmath35
. observations of local starburst galaxies indicate that nearly all of the star formation produced kinetic energy ( due to type ii supernovae ) is used to power gsw ( e.g. , * ? ? ?
given the uncertainties on the evolution of imf with redshift ( i.e. , possibly more top heavy at higher redshift ) and the fact that newly discovered prompt type i supernovae contribute a comparable amount of energy compared to type ii supernovae , it seems that our adopted value for @xmath33 is consistent with observations and within physical plausibility .
we use the following cosmological parameters that are consistent with the wmap7-normalized @xcite lcdm model : @xmath36 , @xmath37 , @xmath38 , @xmath39 , @xmath40 and @xmath41 .
convergence test of results are presented separately in appendix a in order not to disrupt the flow of the presentation in the results section .
the tests show that our results are quite converged and should be robust at the accuracies concerned here , suggesting that our resolution has reached an adequate level for the present study .
the reader may go to the appendix any time to gauge the convergence of relevant computed quantities .
the fact that most of the contributions to dla incidence come from galaxies of mass @xmath42 that are well above our resolution , the results of our convergence tests are self - consistent .
we identify galaxies in our high resolution simulations using the hop algorithm @xcite , operated on the stellar particles , which is tested to be robust and insensitive to specific choices of concerned parameters within reasonable ranges .
satellites within a galaxy are clearly identified separately .
the luminosity of each stellar particle at each of the sloan digital sky survey ( sdss ) five bands is computed using the gissel stellar synthesis code @xcite , by supplying the formation time , metallicity and stellar mass .
collecting luminosity and other quantities of member stellar particles , gas cells and dark matter particles yields the following physical parameters for each galaxy : position , velocity , total mass , stellar mass , gas mass , mean formation time , mean stellar metallicity , mean gas metallicity , star formation rate , luminosities in five sdss bands ( and various colors ) and others .
while our simulations also solve relevant gas chemistry chains for molecular hydrogen formation @xcite , molecular formation on dust grains @xcite and metal cooling extended down to @xmath43k @xcite , at the resolution of the simulations , molecular clouds are not properly modeled . to correct for that
, we use the @xcite observation that at @xmath44hi / cm@xmath19 2 fraction is about 50% and then implement the following prescription to remove neutral gas in extrapolated high density regions and put it in 2 phase . in detail , we assume that the density profile is isothermal below our resolution , which would translate the fraction of mass in 2 is @xmath46 .
thus , we post - process the neutral neutral density in the simulation by the following transformation : @xmath47 , where @xmath48 is
the hi density directly from the simulation , and @xmath49 is that after this processing step .
a very precise choice of the parameter in the above equation is unimportant ; changing @xmath50 to @xmath51 makes marginally noticeable differences in the results .
the primary effect of doing this is to remove very high hi column dlas and causes the hi column density distribution function to steepen at @xmath52 , in agreement with observations .
in addition , because of that , the total amount of neutral gas in dlas also become convergent and more stable . after the above post - process step ,
we shoot rays through the entire refined region of each simulation along all three orthogonal directions using a cell size of @xmath53kpc comoving . in practice , this is done piece - wise , one small volume of the simulation box at a time , due to limited computer memory .
the spectral bin size is @xmath54km / s .
all physical effects are taken into account , including temperature broadening and peculiar velocities .
both intrinsic lorentzian line profile and doppler broadening are taken into account for both @xmath9 and @xmath121808 line , although , in practice , for dlas , doppler broadening is important for @xmath121808 line and lorentzian profile for @xmath9 line .
all relevant atomic data are taken from @xcite .
a dla is defined , as usual , a system with hi column larger than @xmath55@xmath56 .
we assume that the fractional abundance of is equal to fractional abundance of hi .
since , as we will see later , the hi regions of dlas are peaky " with well - defined line - of - sight boundaries and since dlas are very optically opaque to ionizing photons , any refined treatment of radiative self - shielding etc is unlikely to have any significant effect .
note that we have already included a crude self - shielding method during the simulation , which should work well for optically opaque regions .
as a side , one numerical point to note is that , because of the very large dynamic range of both line cross sections as a function of frequency shift from the line center and the delta function like cross section shapes in the line core regions , the convolution operations involved in the detailed calculations of optical depths require at least @xmath57bits precision for floating point numbers . for each dla , we compute the hi column weighted metallicity , register its position relative to the center of the primary galaxy ( i.e , the impact parameter ) , and for dlas that are physically connected by at least one cell side in projection we merge them and in the end compute projected area @xmath58 of each connected region to define it size @xmath59 . for each galaxy
we also register the maximum velocity width @xmath60 among its associated dlas .
we are able to identify more than one million dlas through ray tracing at each redshift examined in each of the runs .
so the statistical errors are very small for each specific run at any redshift .
but that does not speak to cosmic variance and as we shall show later , cosmic variance is indeed quite large concerning quantities that directly or indirectly pertain to the number density of dlas .
other quantities , such as size , metallicity , kinematic properties , etc .
, however , appear to depend weakly on environments and their variances are small .
a dla belongs " to the largest galaxy in the region , within whose virial radius the dla lies .
for example , a dla that is physically more closely located to a satellite galaxy that in turn is within the virial radius of a larger galaxy is said to belong to that larger galaxy .
we do not add instrumental noise to the simulated spectra , but we adopt the same observational procedure to compute the kinematic measures for the absorption lines . for all relevant measures for the line ,
we follow identically the procedures and definitions in @xcite .
we generate synthetic spectra for both @xmath9 and line with @xmath54km / s pixels and then smooth it with a 9-pixel boxcar averaging procedure .
we define the velocity width of a absorption line associated with a dla to be the velocity interval of 90% of the total optical depth , @xmath61 . for the three kinematic shape measures for line we use all intensity troughs ( optical depth peaks ) without the @xmath62 constraint , where @xmath63 is the continuum flux , as re - emphasized by @xcite .
the kinematic shape measures , @xmath64 , @xmath65 and @xmath66 , are defined exactly the same way as in @xcite .
cm cm cm cm cm cm cm cm we first present a gallery of twelve dlas at @xmath67 ( figure [ fig : pic2]-[fig : pica7 ] ) to show the richness of their physical properties . for each dla
six maps and five quantitative panels are displayed . in each of the six maps the line of sight ( los ) intercepting the dla
is shown as a white horizontal line and the exact location of the primary component of the dla is at the intersection with another , white vertical line . in cases with multiple components along the los ,
the primary component coincides with the highest neutral density .
four of the maps - top left ( temperature in kelvin ) , middle left ( atomic hydrogen density in @xmath68 ) , bottom left ( metallicity in solar units ) and top middle ( pressure in units of kelvin @xmath68 ) - have a physical thickness of @xmath69kpc . also indicated in top middle ( pressure ) is the peculiar velocity field with a scaling of @xmath70kpc corresponding to @xmath71km / s .
the remaining two maps - middle middle ( baryonic overdensity ) and bottom middle ( stellar surface density in @xmath72/kpc@xmath73 ) - are projected over the entire galaxy of depth of order of the virial diameter of the primary galaxy . while these two projected maps give an overall indication of relative projected location of the dla respect to the galaxy , the exact depth of the dla inside the paper is , however ,
not shown . when we quote distance from the galaxy , we mean the projected distance on the paper plane .
the five panels on the right column show various physical quantities along the line of sight ( i.e , along the white horizontal line shown in the maps on the left two columns ) . from top to bottom they are : atomic hydrogen density ( in @xmath68 ; red solid curve with a narrower shape ) along with total hydrogen density ( dotted green curve with a more extended shape ) , gas metallicity ( in solar units ) , line - of - sight proper peculiar velocity ( in km / s ) , @xmath9 flux and flux for @xmath121808 line .
the top three panels are plotted against physical distance , whereas the bottom two panels are plotted versus the los velocity .
also indicated in the @xmath9 flux panel ( second from bottom ) are several quantitative measures of the dla , including the neutral hydrogen column density ( @xmath74 n(hi ) ) , the halo mass of the primary galaxy in the system ( @xmath75 ) , the velocity width of the associated line ( @xmath61 ) and three kinetic measures of the line , @xmath64 , @xmath65 , @xmath66 .
we now describe in turn each of the twelve dla examples .
figure [ fig : pic2 ] shows a dla produced by the los intersecting the tip of a long chimney at a distance of @xmath76kpc , the velocity structure suggests that it is still moving away ( upwards ) from the galaxy at a velocity of @xmath77km / s , likely caused by galactic winds .
the metallicity at the interception is @xmath3\sim -1.5 $ ] but there are large gradients and variations of metallicity ( we found that some other very nearby dla systems intersecting different parts of the chimney have metallicity @xmath3<-3 $ ] , not shown ) , suggesting very inhomogeneous enrichment process by galactic winds .
while the primary galaxy has a mass of @xmath78 , i.e. , a 1-d velocity dispersion of @xmath71km / s , the kinetic width of this line is only @xmath79km / s with @xmath80 .
although the @xmath9 flux appears as a single component , as will be the case in all subsequent examples , the absorption has several separate features , reflecting the two - peak structure of the absorbing column and complex velocity structure within .
we note that the nearby satellite galaxies may have triggered the starburst and the galactic winds . the responsible gas for this dla is probably cooling and confined by external pressure likely due to thermal instability , as seen in the pressure panel .
cm cm cm cm cm cm cm cm in figure [ fig : pic3 ] a dla of width @xmath81km / s is created jointly by two major components along the sightline , one at @xmath82kpc of metallicity of @xmath3\sim [ -1.0 , -0.5]$ ] and size of @xmath83kpc at an impact parameter of @xmath76kpc and the other at @xmath84kpc of metallicity of @xmath3\sim 0.0 $ ] and size of @xmath76kpc .
what is striking is many long gaseous structures in this galaxy . as we will see frequently , there are often long gaseous structures connected with galaxies that seem always coincidental with visible galaxy interactions of multiple galaxies or galaxies and satellites in close proximity .
we shall call these features galactic filaments " hereafter .
it seems likely that some of these galactic filaments are cold streams @xcite . however ,
galactic filaments found in our simulations appear to be very rich in variety and disparate in metallicity ( spanning 3 decades or more in metallicity ) . in other words
, they are not necessarily primordial cold streams . in the case of this dla
, the filaments are likely made of gas pre - enriched , having cooled ( as the pressure panel shows ) and now rotating about the galaxy ( roughly counter - clockwise ) .
in contrast , note that the dla in figure [ fig : pic2 ] was still moving away from the galaxy .
like in figure [ fig : pic2 ] , the rich galactic filaments appear to be associated with significant satellite structures in close proximity .
the absorption has several separate features , reflecting the two separate physical components as well as substructures within each component .
cm cm figure [ fig : pic5 ] shows a dla that is associated with a low metallicity ( @xmath3\sim [ -2.0 , -1.5]$ ] ) filament that is feeding a small satellite , which in turn appears to be interacting and possibly feeding the primary galaxy at a projected distance of @xmath83kpc .
this is yet another example of interacting galaxies producing rich gas - feeding filaments , as already seen in figures [ fig : pic2 ] and [ fig : pic3 ] .
the relatively large width of @xmath85km / s is produced by steep velocity gradient in the region from @xmath86 to @xmath87kpc .
one could see that galactic winds are blowing to the upper left corner by the primary galaxy , whose starburst is likely triggered by the interaction .
cm cm figure [ fig : pic7 ] shows a dla that is made up by several filaments at distances of @xmath88kpc from the galaxy .
the metallicity of all the components is near solar , indicating that these are probably pre - enriched gas cooling due to thermal instability .
the velocity structures show that they are falling back towards the galaxy , in a fashion perhaps similar to galactic fountains @xcite , on somewhat extended scales .
once again , it appears that galaxy - galaxy interactions may be responsible for the rich gas filaments , as seen in figures [ fig : pic2 ] , [ fig : pic3 ] and [ fig : pic5 ] .
there is evidence that winds are blowing upwards from the galaxy .
cm cm figure [ fig : pic8 ] shows another example of a dla arising from a galaxy with a very rich filament system due to galaxy interactions .
the primary galaxy is the same as the one shown in figure [ fig : pic3 ] and we are now looking at its south side .
these filaments that are responsible for the neutral column of the dla appear to have been enriched to a level of @xmath3\sim -1.0 $ ] and have cooled to low temperature .
the large width of @xmath89km / s is due to the multiple components spanning a spatial range of @xmath90kpc each of physical depth of several kpc and individual velocity width @xmath91km / s .
interestingly , for this dla system , while most of the gas filaments are now falling back toward the galaxy ( not necessarily radially ) , galactic winds are still blowing towards the upper right corner .
comparison of figure [ fig : pic8 ] and figure [ fig : pic3 ] indicates that the metallicity in the upper right quadrant of the galaxy is somewhat more metal enriched ( @xmath3\sim 0 $ ] ) than other regions ( @xmath3\sim -1 $ ] or lower ) , consistent with the directions of ongoing galactic winds .
this is strongly suggestive that metallicity enrichment process not only is episodic , multi - generational , anisotropic , but also in general possesses no parity .
cm cm figure [ fig : pic9 ] shows a dla intercepting two filaments at a small inclined angle , giving rise to a broad physical extension of @xmath92kpc .
all the visible filaments appear to run roughly top - left to bottom - right , whereas the metal enriched regions seem to spread out like a butterfly in the direction roughly perpendicular to the filaments .
this is the classic picture that galactic winds tend to blow in directions that are perpendicular to the filaments that are feeding the galaxy .
the inner regions of the filaments appear to be less enriched ( @xmath3\sim -2 $ ] ) than the outer regions of the filaments ( @xmath3\sim [ -0.5 , 0]$ ] ) , strongly indicative of galactic winds tending to circumvent the denser filaments .
the large width of @xmath93km / s appears to be caused by oppositely moving ( i.e. , converging ) flows at @xmath94kpc , probably caused by the bipolar winds interacting with the complex filament structures .
this galaxy has a mass of @xmath95 and we notice that most of its surrounding regions is relatively cold , whereas in figures ( [ fig : pic2]-[fig : pic8 ] ) we consistently see a hot atmosphere permeating the circumgalactic regions .
the galaxies in figures ( [ fig : pic2]-[fig : pic8 ] ) all have mass @xmath96 , consistent with the mass demarcation of cold and hot accretion modes @xcite .
nevertheless , the existence of cold galactic filaments seen in figures ( [ fig : pic2]-[fig : pic8 ] ) is consistent with the suggested cold mode of accretion of massive galaxies at high redshift @xcite .
figure [ fig : pic10 ] shows a normal " dla where a relatively quiet galactic disk is pierced edge - on .
its large width of @xmath97km / s is simply due to the large halo that the galaxy is residing in of mass @xmath98 ( at @xmath67 ) .
the surrounding environment seems relatively pristine " with no widespread metal enrichment at a level of @xmath3\ge -1 $ ] .
however , the temperature panel indicates that there is a hot halo permeating the entire region and embedding and pressure - confining ( see the pressure panel ) the cold neutral clouds .
it seems likely that this hot gaseous halo is produced by gravitational shocks rather than galactic wind shocks .
there are several filaments attached to the galaxy .
this is a good example of cold streams feeding a massive galaxy by penetrating a hot atmosphere .
figure [ fig : pica3 ] shows a dla with a large velocity width arising from a relatively small galaxy of total mass @xmath99 .
the galactic winds are blowing in the north - east direction that entrain cold neutral clouds with it .
the los of the dla intercepts a high velocity component at @xmath100kpc .
the combination of this high velocity component with the low velocity component at @xmath101 produces the relatively large width of @xmath102km / s .
note that an isotropic maxwellian velocity distribution of dispersion equal to that of the halo velocity dispersion would only yield a width of @xmath103km / s .
clearly , galactic winds are directly responsible for the large width of this dla , by entraining cold gas clouds to a high velocity .
figure [ fig : pica4 ] shows another dla produced by a small galaxy of mass @xmath104 with a large velocity width .
the galaxy system have multiple , interaction galaxies at close distances .
the galactic winds are blowing primarily , in a bipolar fashion , in north - east and south - west direction , roughly perpendicular to the galactic disk , that entrain the cold neutral cloud at @xmath105kpc to a broad velocity of @xmath106km / s relative to the galaxy itself .
in combination with another complex structure at @xmath107kpc the galactic winds produce a very large width of @xmath108km / s .
note that for this galaxy @xmath109km / s .
figure [ fig : pica5 ] shows yet another dla arising from a small galaxy but having a large velocity width of @xmath110km / s .
this is an interesting case where the galactic winds are blowing , by the primary galaxy , towards and passing through the satellite galaxy at @xmath111kpc in the north - east direction .
the sdss u band luminosity map suggests that the satellite galaxy itself is experiencing a starburst and likely blowing and enhancing the north - east / north winds .
a very large positive velocity gradient in the positive x - direction ( downstream ) of @xmath112km / s over an los physical interval of @xmath113kpc is produced , resulting in a very large width .
note that @xmath114km / s for this galaxy .
the entrained neutral gas cloud and its downstream appear to have escaped the metal enrichment by ongoing winds and remain at @xmath3\sim -1 $ ] - shadowing " effect due to dense clouds and filaments . figure [ fig : pica6 ] shows another wide dla from a small galaxy . for this dla
the majority of the column is due to the intersection with the disk of the galaxy , which would have produced a velocity width of @xmath115km / s on its own .
the galactic winds blowing at the south - east direction entrain some cold clouds at @xmath116kpc to a velocity up to @xmath71km / s .
together , a large width of @xmath110km / s is produced .
given the small mass of the galaxy the surrounding regions are not embedded in a hot gravitationally shock heated atmosphere .
there are some solid angles with low gas column that have been heated by galactic winds , probably triggered by the binary galaxy interaction , as can be seen by comparing temperature map and the density map .
note that @xmath117km / s for this galaxy . finally , figure [ fig : pica7 ] shows the last example of a wide dla from a small galaxy .
two galactic filaments make up this dla , one at @xmath118kpc and the other at @xmath119kpc .
the galactic system is a primary - satellite binary that is interacting , which has likely caused both to experience starbursts .
the primary galaxy at @xmath120kpc is blowing bipolar galactic winds mainly in the north - south direction , whereas the satellite at @xmath121kpc is blowing bipolar galactic winds in the east - west direction .
together they produce a very complex , multi - stream velocity structure .
the total velocity width of this dla is @xmath71km / s , although each of the two components individually has a velocity width of @xmath122km / s .
note that @xmath123km / s for the primary galaxy . in summary , we see that dlas arise in a wide variety of cold gas clouds , from galactic disks to cold streams to cooling gas from galactic winds to cold clouds entrained by hot galactic winds at a wide range of distances from galaxies , with a wide range of metallicity and in galaxies of all masses from @xmath124 at @xmath13
. inspection of the gallery has already hinted that many large velocity width dlas may be produced directly or indirectly by galactic winds .
that is , directly by entraining cold gas clouds and compressing cold gas clouds with high pressure and indirectly by enhancing cooling and thermal instability with added metals and shock compression .
in addition , the composite nature of many large width dla systems should also help remove the perceived failure of the standard lcdm model with respect to producing large width systems ( e.g. , * ? ? ? * ) .
quantitative results later prove this is indeed the case .
we now present more quantitative statistical results on the velocity width distribution functions of dlas at several redshifts . since the velocity structure in @xmath9 flux of a dla is
damped " and does not provide the kinematic information of the underlying physical cloud . following @xcite , the velocity width , @xmath61 , is defined to be the velocity interval of 90% of the optical depth of the @xmath121808 absorption line associated with the dla .
figure [ fig : v90his ] shows the velocity width distribution at three redshifts ( @xmath125 ) , covering most of the observed redshift range .
we see a factor of @xmath126 variation from c " to v " run , indicating the need to have a larger statistical set of simulations covering , more densely , different environments , before a more precise comparison can be made with observations .
insofaras the observed velocity width distribution function lies inbetween the two bracketing runs , c " and v " , and the shape of the functions are in excellent agreement with observations , including the high velocity tail ( @xmath127km / s ) , this should be considered a success for the lcdm model - there is no lack of large width dlas with @xmath127km / s in the lcdm simulation .
this conclusion is consistent with that of @xcite , who studied this issue with a different code and a different feedback implementation .
there is a significant difference between our results and theirs in the that we find galactic winds are directly responsible for many of the large width dlas , by entraining neutral dense clouds to large velocities .
in addition , they conclude that a large halo mass ( @xmath128 ) is a necessary condition for producing large velocity widths , while we find that a non - negligible fraction of large velocity width dlas arise in halos less massive than @xmath129 . to help understand the large velocity width dlas , we plot in figure [ fig : v90max ] the maximum @xmath61 of all dlas , @xmath60 , associated with each galaxy against the halo mass of the galaxy , @xmath130 .
the black line @xmath131 is what @xmath61 would be if the velocity distribution is an isotropic and maxwellian distribution with its dispersion equal to @xmath132 , and the gas density is constant across the dla .
we see that the maxwellian velocity distribution ( the black line ) approximately provides a lower bound to @xmath61 , although there is , unsurprisingly , some fraction of systems that lie below ( see figure [ fig : pic2 ] for an example ) .
what is very interesting is that at @xmath67 there is a large number of galaxies whose @xmath60 are substantially larger than what @xmath132 could produce , i.e. , super - gravitational motion " in the terminology of @xcite .
this super - gravitational motion is produced by galactic winds , as we have seen clearly in figures [ fig : pica3 ] , [ fig : pica4 ] , [ fig : pica5 ] , [ fig : pica6 ] and [ fig : pica7 ] in 3.1 .
we also note that at @xmath5 for both c " and v " run ( and especially the c " run ) , the correlation between @xmath61 and @xmath132 becomes substantially better with much reduced scatter , and the excess of dlas with large @xmath133 is much removed .
this is circumstantial but strong evidence that galactic winds are responsible for most of the large @xmath133 dlas , because of higher star formation activities hence galactic winds at @xmath67 than @xmath5 .
figure [ fig : v90maxz0 ] below will further strengthen this point .
it appears that the redshift evolution at a fixed environment is relatively mild in the redshift range @xmath134 to @xmath5 .
we speculate that the weak evolution of the velocity width distribution from @xmath134 to @xmath5 may be coincidental and attributable to two countering processes : growth of halo mass hence virial velocity with time and diminution of supergravitational motion produced by galactic winds with time ( due to reduced star formation activities with time at @xmath135 ) .
this prediction of a weak evolution of velocity width distribution with redshift is verifiable with future larger dla sample and is a powerful test for the non - gravitational origin of a large fraction of the large width systems .
-2 in figure [ fig : v90max ] does not , however , fairly characterize the relative contribution of halos of different masses to velocity width distribution function , because it does not specify the number of dlas at a given halo mass . in figure [ fig : v150his ]
we show the halo mass probability distribution function for dlas above three velocity width cuts , @xmath136km / s , respectively .
we see a clear trend that larger halos make larger contribution to larger width dlas , as one would have expected . for example , about one half of all dlas with @xmath137km / s arise in halos of mass greater than @xmath138 at @xmath67 , whereas that division line drops to @xmath95 for @xmath139km / s .
it should be noted that the ratio of the virial velocity of a halo of mass @xmath95 to that of @xmath138 is @xmath140 , significantly greater than @xmath141 , indicating an overweight of dla cross - section by large galaxies . for moderate to large velocity width of @xmath139km / s , halos of mass @xmath142 dominate the contribution to dla incidence , largely in agreement with @xcite .
slightly at odds with @xcite , however , we find a significant fraction of these relative wide systems arising in galaxies of mass less than @xmath142 : @xmath143 of dlas with velocity width larger than @xmath144km / s are due to galaxies with mass less than @xmath142 .
we note that our definition of associating dlas with galaxies biases associating them to larger galaxies ; figure [ fig : pic5 ] gives an example , where the dla is defined to arise from the larger galaxy of mass @xmath145 , even though it is more closely related to a much smaller satellite galaxy that is orbiting around the larger galaxy .
our results are perhaps unsurprising in the sense that one would have expected that galactic winds , when they are blowing , should be stronger , or at least not weaker , in dwarf starburst galaxies than larger galaxies thanks to shallow gravitational potential wells in the former , when cold gas is still abundant at high redshift . both figure [ fig : v150his ] and gallery pictures in 3.1 confirm this point .
galactic winds , however , could be weaker in dwarf galaxies if star formation is inproportionally less vigorous
. this may be the case at lower redshift , as shown in figure [ fig : v90maxz0 ] .
what is interesting , and further evidence , is that at @xmath6 the dwarf galaxies in v " run appear to have more super - gravitational motion than in the c " run , simply because the former are gas richer and have higher star formation rate than the latter .
thus , it seems that galactic winds are a bivariate function of galaxy mass and star formation rate , in a fashion that is consistent with observations ( e.g. , * ? ? ?
-1 cm -0.7 cm -1.8 cm -0.7 cm -1.8 cm -0.7 cm -1.8 cm having found an overall good agreement with observations with respect to the velocity width distribution , we now turn to shape measures of @xmath121808 absorption line profile . before comparing to observational data from @xcite
, we shall first try to understand the relationship among the optical depth of a line , hi column density metallicity and velocity width .
assuming that the optical depth profile of the line is a simple top - hat ( assuming a different profile such as a gaussian makes no material difference for our purpose ) , it can be shown that @xmath146 where @xmath147 is the metallicity of dla in solar units . left panel of figure [ fig : itwopeaks ] shows @xmath61 as a function of @xmath148 for @xmath149 . as expected , an increase in velocity width requires a corresponding increase in column density to produce a same optical depth .
more important is that , quantitatively , in order to achieve an optical depth of @xmath150 , with a width @xmath151km / s it requires a dla column of @xmath152@xmath56 , if the dla is composed of one single component with @xmath3=-1 $ ] . since the abundance of dlas with @xmath153@xmath56 declines
rapidly ( see figure [ fig : colhis ] ) but the abundance of line peaks near @xmath151km / s ( figure [ fig : v90his ] ) , this suggests that a significant fraction of lines must have multiple components . to quantitatively illustrate this , we define a new simple two - component measure as follows .
if there are at least two peaks in the optical depth profile that are separated by more than @xmath154 and the ratio of the peak heights is greater than @xmath155 , we define the dla to be a two - component dla .
the ratio , @xmath155 , comes about such that the lower peak is guaranteed to be included in the accounting of @xmath61 interval , although changing it to say @xmath156 makes no dramatic difference in the results .
note that dlas with more than two components are included as two - component systems .
right panel of figure [ fig : itwopeaks ] shows the percentage of two - component dlas as a function of @xmath61 . in good agreement with the simple expectation
, we see that at @xmath157km / s , about 50% of dlas have more than one component and that number increases to @xmath158 at @xmath159km / s .
this result is also consistent with the anecdotal evidence shown in the gallery examples in 3.1 , where most of large width dlas contain more than one physical component .
we now turn to the three kinematic shape measures defined in @xcite , @xmath64 , @xmath65 , @xmath66 , representing , respectively , measures of the symmetry , leading - edgeness and two - peakness of the profile of @xmath121808 absorption lines associated with dlas ( see the bottom right panels of the gallery pictures in 3.1 ) . figures ( [ fig : fmmhis],[fig : fedgehis],[fig : f2pkhis ] ) show comparisons of simulation results with observations at three redshifts @xmath5 , @xmath67 and @xmath134 .
we see the overall agreement between simulations and observations is excellent , with k - s tests ( indicated in the figures ) for both runs ( c " and v " ) at three compared redshifts ( @xmath125 ) all being at acceptable levels .
our results are in good agreement with one of the models with feedback in @xcite , except for the case of @xmath66 : our simulations find acceptable k - s test values of 26 - 29% , 4 - 34% and 23 - 34% at @xmath5 , @xmath160 and @xmath134 , respectively , whereas they find none of their models have probability higher than 5% at @xmath67 .
we speculate that difference in the detailed treatments of metal transport process as well as feedback prescription between our simulations and theirs may have partly contributed to this difference ; with detailed metal transport we find very inhomogeneous metallicity distributions across space and among dlas in our simulations ( see figure [ fig : mtlhis ] below ) , whereas they assume a constant metallicity of @xmath3=-1 $ ] for all dlas .
it is also noted that the metallicity distributions of our simulations at redshift range @xmath161 are in excellent agreement with observations ( figure [ fig : mtlhis ] ) . at the bottom - right panels of each four - panel set in figures ( [ fig : fmmhis],[fig : fedgehis],[fig : f2pkhis ] ) we show a comparison between @xmath5 and @xmath67 distributions for each of the shape statistics and find that there is significant evolution in all three shape measures .
current small observational sample does not allow for such a test .
our results demonstrate that the standard lcdm model , with a proper modeling of astrophysical processes , including galaxy formation and feedback in the forms of mechanical feedback and metal enrichment , can successfully produce line shapes that are in good agreement with observations .
-0.5 cm let us now address the fundamentally important observable : the column density distribution of dlas and its evolution .
figure [ fig : colhis ] shows the column density distribution at several redshifts from @xmath6 to @xmath163 .
where comparisons can be reliably made with observations , at @xmath164 , @xmath67 and @xmath163 , we see that the overdense run c " and underdense run
v " appropriately bracket the observational data in amplitude .
similar to the situation for the velocity distribution function ( figure [ fig : v90his ] ) , the strong environmental dependence of the column density distribution renders it impractical to make vigorous comparisons between the simulations and observations .
given that the amplitude of observed column density distribution lies between that of c " and that of
v " run , and the shapes of both simulated functions are in reasonable agreement with observations we tentatively conclude that the standard lcdm model can reasonably reproduce the observed the column density distribution . note that the shape at the highest column end depends on the treatment of high density regions , for which we have used an empirical relation .
ultimately , when pc resolution is reached , we can make more definitive tests .
what is also interesting is that the variations between different environments are larger than the redshift evolution of the column density distribution in each run .
it is further noted that , seen in the lower - left panel of figure [ fig : colhis ] , the evolution of the column density distribution in c " and v " run is different . in
c " run , we see weak evolution from @xmath163 to @xmath5 and then a relatively large drop in amplitude at @xmath6 . in
v " run , on the other hand , we see practically little evolution from @xmath67 to @xmath6 .
this likely reflects the dynamical stage of a simulated sample , where the c " run is more dynamically advanced than the v " at a same redshift , consistent with the behavior seen in figure [ fig : dishis ] below .
-1 cm left panel of figure [ fig : omega ] shows the redshift evolution of dla line density , defined to be the number of dlas per unit absorption length .
right panel of figure [ fig : omega ] shows the redshift evolution of neutral gas density in dlas .
inherited from the situation shown in figure [ fig : colhis ] , there is a large variation of both plotted quantities between the two ( c " and v " ) runs .
what is reassuring is that the observed data lie sensibly between results from these two bracketing environments . if one assumes that the cosmic mean of each of the two plotted quantities should lie between c " and v " run , reading the range spanned by the two runs suggests that the lcdm model is likely to agree with observations to within a factor of @xmath165 with respect to both quantities ,
although what the overall temporal shape will look like is difficult to guess . to firmly quantify these important observables and to more precisely assess the agreement / disagreement between the predictions of the lcdm model and observations , a larger set of simulations sampling , more densely , different environments in a statistically correct fashion will be necessary , so is a more accurate treatment of the transition from atomic to molecular hydrogen in very high density regions ( that likely affects the shape at the high column density end ) .
we reserve this for future work .
-1 cm the current set of simulations is vastly superior to those used in our earlier work addressing the observed relatively weak but non - negligible evolution of dla metallicity @xcite and here we return to this critical issue .
figure [ fig : mtlhis ] shows the dla metallicity distributions at four redshift , @xmath166 . for the three redshifts , @xmath125 , where comparisons can be made , we find that the agreement between simulations and observations at @xmath5 to @xmath6 is excellent , as k - s tests show . this is a non - trivial success , given that our feedback prescription has essentially one free parameter , that is , the supernova energy that is driving galactic winds transports energy , metals and mass throughout interstellar ( ism ) , circumgalactic ( cgm ) and intergalactic space ( igm ) .
furthermore , the absolute amount of metals is totally fixed by requiring that 25% of stellar mass with metallicity equal to @xmath167 returning to the ism , cgm and igm .
the agreement indicates that our choices of both the supernova ejecta mass and its metallicity and the explosion energy , which are inspired by theories of stellar interior and direct observations , may provide a reasonable approximation of truth .
we see that the peak of the dla metallicity distribution evolves from @xmath3=-1.5 $ ] at @xmath0 , to @xmath3=-0.75 $ ] at @xmath5 , and to @xmath3=-0.5 $ ] at @xmath6 .
thus , both simulations and nature indicate that there is a weak but real evolution in dla metallicity .
what is also important to note is that , in agreement with observations , simulations indicate that the distribution of metallicity is very wide , spanning three or more decades at @xmath168 .
this wide range reflects the rich variety of neutral gas that composes the dla population , from relatively pristine gas clouds falling onto or feeding galaxies , to metal - enriched cold clouds that are falling back to ( galactic fountain ) or still moving away from ( due to entrainment of galactic winds ) galaxies , to cold neutral gas clouds in galactic disks .
there is a metallicity floor at @xmath3\sim -3 $ ] at @xmath2 and that floor moves up to @xmath3\sim -1.5 $ ] by @xmath6 , consistent with observations @xcite .
the distribution at @xmath6 is significantly narrower , partly reflecting the overall enrichment of the igm and partly due to much reduced variety of dlas with galactic disks becoming a more dominant contributor to dlas ( see discussion below ) .
@xcite find that proximate dlas ( pdlas ) , those within a velocity distance from the qso @xmath169km / s , seem to have metallicity higher than the more widely studied , intervening dlas .
it seems conceivable that the total sample of pdlas plus conventional ( intervening ) dlas may somewhat shift the metallicity distribution to the right , perhaps bringing it to a still better agreement with our simulations .
1.7 cm -2.5 cm observations have found a strong positive correlation between galaxy mass and metallicity ( e.g. , * ? ? ? * ) .
we divide the simulated dla sample at @xmath67 into two subsets , one with metallicity less than @xmath3=-1 $ ] and the other more than @xmath3=-1 $ ] .
we then compute the velocity width functions separately for each subset , which are shown as solid dots ( lower metallicity ) and solid squares ( higher metallicity ) in the left panel of figure [ fig : v90hiszdiv ] .
what we see is that there is a small excess of large velocity width dlas for the higher metallicity subset compared to the lower metallicity one .
this is of course in the sense that is consistent with the observed metallity - mass relation .
however , current observational data sample is consistent with simulations , and the difference between the two simulated subsets and between the two observed subsets is statistically insignificant . a larger sample ( by a factor of 4 ) may allow for a statistically significant test .
do we expect a larger difference in the disk model @xcite ?
we do not have a straight answer to this question , without a very involved modeling .
however , we we suggest that the picture we have presented , where dlas arise from a variety of galactic systems , in a variety of locations of widely varying metallicity ( see the gallery in 3.1 ) , would be consistent with the small difference found , because the velocity widths of large width dlas do not strongly correlate with galaxy mass ( see figure [ fig : v90max ] ) . in other words , the observed correlation between metallicity and galaxy mass is largely washed out by dlas that do not arise in disks and whose metallicity do not strongly correlate with galaxy mass .
if one combines the information provided by figure [ fig : v150his ] and figure [ fig : v90hiszdiv ] , one may reach a similar conclusion . -1 cm the implication may be that dlas do not arise predominantly in gaseous disks of spiral galaxies at high redshift , in agreement with @xcite and @xcite .
we shall elaborate further on this significant point . in figure
[ fig : dishis ] we show the distribution of physical distance of dlas from the galactic center ( i.e. , impact parameter ) at four redshifts , @xmath166 .
since we have shown in figure [ fig : v150his ] that dla incidence contribution peaks at @xmath170 , let us make a simple estimate of their size at @xmath13 . as a reference ,
let us take the radius of milky way ( mw ) stellar disk to be @xmath171kpc .
taking mw to @xmath172 self - similarly would give a radius of @xmath173kpc and for a @xmath174 galaxy the stellar disk radius would be @xmath175kpc at @xmath172 , corresponding to @xmath176 in the shown x - axis . observed large galaxies ( of mass likely in the range @xmath177 ) at @xmath13 appear to have sizes of @xmath178kpc @xcite , roughly consistent with the simple scaling .
the distance distribution peaks at @xmath179kpc at @xmath0 , which is much larger than a few kpc of the observed ( or expected based on @xmath6 galaxies ) stellar disk size at @xmath13 .
it is noted that the virial radius of a milky way size galaxy is about @xmath180kpc at @xmath172 .
thus , these gaseous structures occur at about half the virial radius at @xmath172 , thus , we conclude that at @xmath0 most of the dlas _ do not _ arise from large galactic stellar disks .
they appear to come from regions that are @xmath181 larger than the stellar disks .
the ubiquitous extended structures - galactic filaments - appear to be at the right distances of @xmath179kpc , seen in the gallery examples in 3.1 . while the extremely close association of galactic filaments with galaxy interactions suggest that the host galaxies are likely experiencing starbursts , as seen in the gallery examples , the clouds that give rise to dlas do not appear to have ongoing _ in situ _ star formation .
clearly , most dlas do not arise in disks and most dlas have low metallicities , as we have shown , are self - consistent . in other words , aside from those dlas that arise from galactic disks and are metal rich , the metallicity of the vast majority of more metal poor dlas do not appear to be forming stars .
it may be that , if and when the gas in the galactic filaments forms stars , either they are destroyed by star formation feedback and remove themselves from the dla category or they have already incorporated into disks of galaxies .
we suggest that our model gives a natural explanation to the apparent puzzle of the lack of obvious star formation of gas - rich dlas @xcite . on the other hand ,
the inferred cooling rates of dlas may be provided , in part , by radiative heating from the host galaxy ( see figure [ fig : sfrhis ] below ) and possibly in part by compression heating as we frequently see higher external pressure in 3.1 .
figure [ fig : zratsfrhis ] shows the ratio of gas metallicity for dlas at different subset of dlas with different column density ranges to the mean metallicity of ongoing star - forming gas .
it is clear that only the high end of the high column density range ( @xmath182@xmath56 ) dlas are forming stars ; most of the dlas have little star formation .
-0.7 cm -1.8 cm returning to figure [ fig : dishis ] , at @xmath5 there is a very interesting divergence between the two distributions for c " and v " run , where the distribution for c " run peaks at @xmath183kpc and for v " run at @xmath184kpc .
this is consistent with the expectation that the overdense region in the c " run and the underdense region in the v " run start to feel " the difference in their respective local large - scale density environment and evolve differently dynamically .
that is , in c " run gravitational shock heating due to large - scale structure formation begins to significantly affect the cold gas in galaxies , whereas in
v " run the galaxies have not changed significantly since @xmath0 except that they are now somewhat smaller due to lower gas density at lower redshift . by @xmath6
the two distributions once again become nearly identical ; this is rather intriguing and may reflect the following physical picture : while galaxies in the v " run has by now dynamically caught up " with the field galaxies in the c " run , giving rise to the similar gaussian - like distribution centered at @xmath185 kpc , the original gas - rich galaxies in the c " run have fallen into the cluster , lost gas and disappeared " from the dla population . while there is almost no dla that is further away than @xmath87kpc at @xmath0 , there is a second bump at @xmath186kpc in the distribution for the c " run at @xmath6 .
this bump is likely due to gas rich satellite galaxies orbiting larger galaxies or small groups of mass @xmath187 . beyond @xmath188kpc ,
there is no dla in the c " run , which is due to gas - starvation of galaxies in still larger groups or clusters at @xmath6 . with direct inspection of simulation data
we find that there is virtually no gas rich galaxies within the virial radius of the primary cluster in the c " run .
what is also interesting is that the peak distance of @xmath184kpc at @xmath6 is totally consistent with the notion that gaseous disks of field galaxies , like the one in our own galaxy , significantly contribute to dlas .
right panel of figure [ fig : v90hiszdiv ] shows the velocity distributions of two subsets of dlas , divided at metallicity at @xmath3=-1 $ ] at @xmath6 .
here we see a very clear difference between the two distributions : the higher metallicity subset have large velocity widths , i.e. , there is a strong positive correlation between metallicity and velocity width at @xmath6 .
this supports the picture that a large fraction of dlas arise in gaseous disks of large field galaxies .
most of the dlas at @xmath6 have a higher metallicity of @xmath3\ge -1.0 $ ] with the overall distribution peaking at @xmath3=-0.5 $ ] , also providing support for this picture .
therefore , by @xmath6 the situation appears to have reversed : galactic disks of large galaxies make a major contribution to dlas at @xmath6 .
the fact that the peak distance has dropped from @xmath88kpc at @xmath0 to @xmath189kpc at @xmath6 is physically in partly due to a large decrease ( a factor of @xmath190 ) in the mean gas density of the universe from @xmath0 to @xmath6 .
-0.7 cm -1.8 cm -0.7 cm -1.8 cm binary quasars , physical or lensed , provide an unique tool to probe the size of dlas . here
we present our predictions of size distributions of dlas in the lcdm model . as we have described in 2.3 , any cells ( of size @xmath53kpc comoving )
that are connected by one side in projection are merged into a single isolated " dla .
the area of each isolated " dla , @xmath58 , is then used to define the size ( radius ) of the dla by @xmath59 .
the total area of all isolated dla associated with a galaxy along three orthogonal directions ( @xmath191 ) , @xmath192 , @xmath193 and @xmath194 , are summed to obtain @xmath195 and the total dla size ( radius ) of the galaxy is defined to be @xmath196 .
note that , if dlas arises from a thin disk , the @xmath197 computed this way will be the exact size of the disk face on , regardless of its orientation . on the other hand , if each dla cloud is a sphere , this method overestimate the size ( area ) by a factor of @xmath198 .
-0.7 cm -1.8 cm figure [ fig : sizehis ] shows the size ( radius ) distribution at redshift @xmath166 for individual dla size and total dla size of each galaxy .
figure [ fig : pdfcolrad ] presents the size information in a different way , where we show the probability that , for a random pair of sightlines separated by @xmath199kpc at the redshift in question , one sightline intercepts a dla and the other intercepts a column density lower than shown in the x - axis .
we see that , in the sense that is consistent with distance distributions shown in figure [ fig : dishis ] , on average , individual dla size as well as the total dla size of galaxies are larger at high redshift than at lower redshift .
the individual dla size distribution appears to sharply peak at @xmath200kpc at @xmath163 , then move left to a broader peak at @xmath201kpc at @xmath67 , then sharpen somewhat to peak at @xmath202kpc at @xmath5 , and finally move rightward slightly to peak at @xmath203kpc at @xmath6 for the c " run .
the numbers for the v " run are largely comparable to the
c " run : @xmath204kpc at @xmath205 , respectively .
what is quite remarkable is that , at @xmath5 where comparison can be made , the predicted size distribution and the available observation agree better than anyone would have expected .
this is a testament to the success of the lcdm model and physical treatment of our simulations , especially considering other agreements that we have already found , for example , with respect to metallicity distribution , column density distribution , kinematics .
intriguingly , @xcite find a new population of faint lyman alpha emitters which they advocate may be the host galaxies of dlas at @xmath13 .
their estimated size , based on observed @xmath9 line emission profile , shown as the open circle in @xmath67 panels of figure [ fig : sizehis ] , would be consistent with our model , although the true nature of this population is relatively uncertain . given the reasonable agreement in size , we will make some further speculation to gauge if our model can accommodate or explain this population as dlas .
one possible physical mechanism for the @xmath9 emission of this population , if they are indeed dla hosts , would be fluorescence due to ionizing photons from the host galaxies . assuming that the distance of each dla cloud from the host galaxy is @xmath206kpc ( figure [ fig : dishis ] ) , the size of each dla cloud is @xmath26kpc ( figure [ fig : sizehis ] ) and the star formation rate of host galaxy is @xmath207/yr ( see figure [ fig : sfrhis ] below ) , then a typical expected star formation rate inferred from their fluorescence would be @xmath208/yr .
this very rough estimate curiously falls within the range of @xmath209/yr estimated by @xcite for the observed emitters if there are @xmath9 emitters at @xmath13 .
lcccccccc h1413 + 1143 & 1.440 & b - a & 0.753 & 3.13 & 60 & 9.0 & 0.84 & 0.80 + & & b - d & 0.967 & 4.02 & 60 & 0.25 & 0.11 & 0.13 + & & b - c & 1.359 & 5.65 & 60 & 0.20 & 0.17 & 0.21 + & & a - d & 1.118 & 4.64 & 9.0 & 0.25 & 0.13 & 0.16 + & & a - c & 0.872 & 3.62 & 9.0 & 0.20 & 0.093 & 0.11 + & & d - c & 0.893 & 3.71 & 0.25 & 0.20 & 0.097 & 0.12 + h1413 + 1143 & 1.486 & d - a & 1.118 & 4.32 & 2.0 & @xmath210 & 0.097 & 0.11 + & & d - b & 0.967 & 3.73 & 2.0 & @xmath211 & 0.084 & 0.097 + & & d - c & 0.893 & 3.45 & 2.0 & @xmath210 & 0.070 & 0.079 + h1413 + 1143 & 1.662 & b - a & 0.753 & 2.17 & 6.0 & @xmath212 & 0.92 & 0.89 + & & b - c & 1.359 & 3.91 & 6.0 & @xmath213 & 0.13 & 0.17 + & & b - d & 0.967 & 2.78 & 6.0 & @xmath214 & 0.064 & 0.078 + & & a - c & 0.872 & 2.51 & 1.5 & @xmath213 & 0.058 & 0.074 + & & a - d & 1.118 & 3.22 & 1.5 & @xmath214 & 0.086 & 0.11 + & & c - d & 0.893 & 2.57 & 0.6 & @xmath214 & 0.052 & 0.063 + he1104 - 1805 & 1.662 & a - b & 3.0 & 4.47 & 6.3 & @xmath215 & 0.10 & 0.11 + um673 & 1.6265 & a - b & 2.22 & 2.71 & 6.3 & @xmath216 & 0.036 & 0.036 + lbqs1429 - 0053 & 1.6616 & a - b & 5.1 & 43.9 & 3.0 & @xmath150 & 0.051 & 0.050 now we make more direct comparisons between our model and available observed 18 pairs of qsos ( all lensed pairs / multiples except one real physical binary ) at @xmath217 , given in table 1 .
columns 1 - 7 list parameters of each observed pair and columns 8 - 9 give the probability of each pair occurring the c " [ p(c ) ) ] and v " [ p(v ) ] run respectively at @xmath5 . we have made the following simplification for computing the probability : we treat all dla with @xmath218@xmath56 the same regardless of column density values and for non - dla absorbers the probability that we present is the probability of having a column equal to or less than the indicated value .
we see from table 1 that the the observed qso pairs are all statistically consistent with our model at @xmath219 level or better , entirely consistent with the agreement shown in figure [ fig : sizehis ] between the computed size distribution and observationally inferred size , which is somewhat model dependent .
finally , in figure [ fig : sizemtot ] , we plot the total dla cross section of each galaxy ( @xmath197 ) against its halo mass at @xmath67 ( left panel ) and @xmath6 ( right panel ) .
also shown as the black line is a linear fit assuming that the effective radius @xmath196 is proportional to the virial radius for simplicity .
note that a least - square linear fit is slightly flatter than the black curve .
it is noted that there are a small fraction of galaxies that do not have significant neutral gas thus do not show up in the plotted range .
what is most striking is that , for gas rich galaxies that give rise to dlas at @xmath67 , on average , the effective area that gives rise to dlas occupies about @xmath220 of the total area within the virial radius , i.e. , @xmath221 . by @xmath6 the effective radius has much reduced to @xmath222 , becoming comparable to the size of the stellar disk .
the total dla cross section as a function of halo mass is in agreement with @xcite within a factor of @xmath54 ( ours is higher than theirs ) .
we note that the method we use to compute the total dla cross section gives a correct value for the case of a thin disk but could overestimate it by a factor of @xmath198 in the case of a sphere .
let us fast forward to figure [ fig : mhalohis ] and take a note that at @xmath67 the typical halo mass for dla incidence is @xmath223 having a virial radius of @xmath224kpc , which then gives @xmath225kpc ( see above relation between @xmath226 and @xmath227 ) . from figure
[ fig : dishis ] we see that the typical distance of dla from galaxy center is @xmath228kpc , thus @xmath229 .
this near equality between the distance and size suggests that a large fraction of the sky seen by an object sitting in the galaxy , up to @xmath230 , may be covered by dlas !
if that object is a qso , the situation may be different , because the qso , with its intense radiation ( and possibly other effects , such as winds ) , may have significantly modified its surroundings , including the nearby dlas .
note that a qso proximity radius is about @xmath231mpc .
the evidence we see in the gallery examples that the surge in dla incidence of a galaxy seems always associated with starbursts and in gas - rich ( probably ) spiral like galaxies , whereas qsos tend to reside in gas - poor elliptical galaxies .
these two facts together suggest that it is not straight forward to estimate the fraction of qsos that have this type of proximity dlas .
the situation for gamma - ray bursts ( grbs ) may be relatively more straight forward to analyze , because grbs are associated with star formation hence likely coincides with the surge in dla cross section of interacting galaxies and because radiation from grbs does not significantly affect their surroundings at a distance of @xmath92kpc .
therefore , we suggest that at high redshift @xmath232 , a significant fraction ( up to @xmath233 ) of grbs , which depends on detailed geometry / orientation of dla clouds , have associated dlas with a typical velocity difference from its systemic redshift of @xmath234km / s ( see figure [ fig : dvgaldlahis ] below for @xmath235 ) .
one complication factor is that some ( or possibly most of ) grb - dlas may be due to dense gas in the close vicinity of grbs - nearby dlas .
since grbs are thought to occur in star - forming regions , i.e. , embedded in molecular clouds , one might expect to see a signicant amount of molecules associated with nearby dlas and they should be dusty and relatively more metal enriched . however ,
if progenitors of grbs are biased metal poor , such as preferred in the collapsar model @xcite , then it may become more complicated , although latest observations seem to indicate otherwise ( e.g. , * ? ? ?
thus , we suggest that a subset of grb - dlas _ with little molecular hydrogen column density _ and being relatively metal poor and dust poor , such as those in @xcite , @xcite and @xcite , may be identified with circumgalactic dlas proposed here . this fraction of grbs with dlas is , however , likely to decrease rapidly towards lower redshift because of ( 1 ) smaller @xmath236 ratio and ( 2 ) a smaller fraction of dlas outside stellar disk .
the unknown in the above estimate is the geometry of dlas .
we will reserve a detailed analysis of this issue for a future study .
so far we have presented our simulations and compare to observations of dlas in many different aspects . from these detailed comparisons
we conclude that the model is consistent with all extant observations of dlas , including kinematic properties , metallicity , column density distribution , line density and neutral hydrogen density , size distribution , and their evolution , wherever comparisons could be made .
proper modeling of galactic winds along with high numerical resolution seems to have contributed to the success .
what is clear is that galactic winds play an indispensable role in alleviating previous tension between the lcdm model and observations , especially in producing large velocity width dlas and disparate and low metallicities of dla gas .
we now examine the properties of dla host galaxies .
-0.7 cm -1.8 cm -0.7 cm -1.8 cm figure [ fig : mhalohis ] shows the dla incidence as a function of the halo mass .
we see that at redshift @xmath237 it appears that the dla cross section of each galaxy is roughly proportional to the square of its virial radius , evidenced by the approximate agreement between the red and green histograms at @xmath125 .
thus our model makes this rather simple prediction : dlas closely trace the overall population of galaxies at @xmath237 , with a slight relative bias for lower mass galaxies for dla hosts compared to the general galaxy population .
the bias is more noticeable at @xmath6 .
this is not to say that every galaxy has a dla in it .
rather , this simply says that the portion of galaxies that give rise to dlas at any particular time has statistically the same mass function as the galaxy population as a whole . at @xmath2 ,
the vast majority of dlas arise in halos of mass @xmath1 , as these galaxies dominate the overall population of galaxies .
we expect the clustering of dlas to be nearly identical to the overall population of galaxies at @xmath232 and gradually becomes relatively weaker compared to the galaxy population as a whole with time .
based on clustering analyses , the mass range of the observed lbgs is inferred to be @xmath238 at z=@xmath239 , @xmath240 at @xmath241 , and @xmath242 at z=1.7 @xcite . comparing these
observationally inferred ranges to the histograms in figure [ fig : mhalohis ] we find that @xmath14 of dlas at @xmath13 may be lbgs and the fraction drops to @xmath243 at @xmath244 .
@xcite pointed out that , if dlas have a radius of @xmath245 kpc , lbgs could account for all the dlas at @xmath13 .
working backwards , the estimated @xmath14 of dlas at @xmath13 being lbgs would imply a dla disk radius of @xmath246kpc for lbgs , which is significantly larger than the size of lbg stellar disks , suggesting that the distribution of neutral dla gas is in a much more extended region than stellar disk , likely a combination of some extended gaseous disk and galactic filaments .
while a significant overlap between dlas and lbgs is expected , the overall clustering of dlas is expected to be comparable but slightly weaker than lbgs , since the median halo mass for dla incidence distribution is slightly smaller than the typical lbg halo mass ( more quantitative comparison will be performed in future work ) .
@xcite conclude that the properties of dla hosts , including half - light radius , radial profile , optical to near - infrared color , morphology , @xmath9 emission equivalent width , and @xmath9 emission velocity structure , lie within the measured range for the general population of lbgs .
since those dla hosts that are sufficiently luminous to be detected are large galaxies in the mass range of lbgs and because dlas tend to arise in gaseous galactic filaments of galaxies that experience starbursts ( via galaxy interactions ; see 3.1 ) , like lbgs , it is naturally expected in our model that they find similarity between their dla hosts and lbgs . of the remaining 70 - 80% that are not covered by lbgs at @xmath0 ,
roughly 10 - 20% are due to more massive galaxies and 50 - 70% are from smaller galaxies .
when systems become more dynamically advanced with time and large galaxies are no longer abundant with cold gas , a slightly more significant shift occurs between dla hosts and the overall population of galaxies at @xmath6 . at @xmath6
we see that the contribution to dla population is broadly peaked at @xmath247 , whereas most of the sum of the square of virial radius is also broadly distributed but , unlike for dla host , having a significant contribution from halos of mass @xmath248 , in the c " run ; for the v " run most of dlas are in halos of mass @xmath249 , whereas most of the sum of the square of virial radius come from halos of mass @xmath250 .
figure [ fig : sfrhis ] shows the dla incidence distribution as a function of host star formation rate ( sfr ) ( red histograms ) and the corresponding notional distribution if the total dla cross section of a galaxy is proportional to its stellar mass to the two - third power .
the relative distribution between the two seems complex and environment dependent . to zeroth order
, dla hosts roughly span the same sfr range of the general galaxy population and most of dla hosts have sfr in excess of @xmath251/yr at all redshifts .
to first order , there is a bias for dla hosts to have slighly higher sfr at @xmath0 and slighly lower sfr at @xmath6 than the general galaxy population .
the most likely sfr rate for a typical dla host falls in the range @xmath7/yr at @xmath0 .
lbgs lie at the upper end of this sfr range , consistent with the earlier finding that 20 - 30% of dlas overlap with lbgs .
the peak in the distribution move to @xmath8/yr by @xmath6 .
it is beyond the scope of this study to investigate the cause for the preferred sfr range by dlas at different redshifts .
we conjecture that the dla incidence rate likely surges " when mergers or other significant interactions between galaxies produce the filaments seen in the gallery in 3.1 .
these same interactions also cause the concerned galaxies to experience significant star formation , plausibly falling into the found range above .
the trend of moving to lower sfr galaxies at lower redshift is perhaps due to the transition from major mergers at high redshift ( @xmath0 ) to minor mergers or smooth accretion and lack of cold gas in very large galaxies at @xmath6 .
all evidence - size ( figure [ fig : sizehis ] ) , distance ( figure [ fig : dishis ] ) , metallicity ( figure [ fig : mtlhis ] ) and now sfr ( figure [ fig : sfrhis ] ) - seems to point to the picture that by @xmath6 most of dls arise in disks of large galaxies of mass @xmath252 that are relatively quiet .
-0.7 cm -1.8 cm figure [ fig : mhihis ] shows the dla incidence distribution as a function of host hi mass .
it is interesting that the hi mass appears to peak at @xmath253 at all redshifts with @xmath254 covering nearly all the contributions .
there is a bias for dla hosts to have slightly more gas rich than the general population of galaxies . why the peak position of @xmath255 does not change with time
is intriguing but not fully understood presently .
figure [ fig : gshis ] shows the dla incidence distribution as a function of host hi mass to stellar mass ratio .
we see that at @xmath0 the majority of galaxies are gas - rich ( @xmath256 ) and dla hosts very closely follow the general population of galaxies , consistent with what we saw in figure [ fig : mhalohis ] .
this statement remains true for the v " run until @xmath6 , whereas for the regions that are more dynamically advanced , such as @xmath6 in the c " run , a significant segregation is evident in that most of the galaxies are now gas poor ( @xmath257 ) but majority of dla hosts are still gas rich ( @xmath256 ) .
moreover , at @xmath6 in c " run , there is a population of old , red and dead " galaxies with virtually no cold gas ; these are likely galaxies in clusters of galaxies , which dlas largely avoid .
-0.7 cm -1.8 cm figure [ fig : colorhis ] shows the dla incidence distribution as a function of host sdss g - r color . while the individual color of the simulated galaxies may not be totally consistent with observations due to its sensitivity to a relatively small variation in the amount of recent star formation that the simulation may not necessarily properly capture , a comparative statement is still valid statistically .
we see that dla hosts trace the general population of galaxies at @xmath0 in all environments with a slight bias to a bluer color peaking sharply at @xmath258 . at lower redshift dla hosts
become significantly bluer than average galaxies by roughly @xmath259 and they avoid the reddest galaxies that are presumably elliptical galaxies or galaxies in clusters of galaxies .
finally , in figure [ fig : dvgaldlahis ] we show the distribution of the los velocity difference between the dla and its host .
we find the following exponential function provides an excellent fit , as shown in the figure , while a gaussian fares much worse : @xmath260 we find that the dispersion parameter @xmath261 only increases mildly with decreasing redshift from @xmath262km / s at @xmath0 to @xmath263km / s at @xmath6 . the relatively small velocity dispersion and
a symmetric distribution may make dlas a useful proxy for the systemic velocity of host galaxies .
this property may be utilized to measure galactic wind velocities , using a large statistical sample of dla hosts with measured interstellar absorption lines , such as na i line ( e.g. , * ? ? ?
* ) . in summary
, we show that , at @xmath0 , dla hosts approximately follow the general population of galaxies with respect to halo mass , star formation rate , hi mass , gas mass to stellar mass ratio , and color , with a small but noticeable tendency of being more gas rich , slightly smaller and bluer .
it is seen that lbgs make up about 20 - 30% of the overall dlas at @xmath0 decreasing to @xmath243 by @xmath264 . of the remaining 70 - 80% that are not covered by lbgs at @xmath0 ,
roughly 10 - 20% are due to galaxies more massive than lbgs and 50 - 70% are from smaller galaxies .
the comparisons at lower redshifts become not as clean cut as at high redshift , but in a way that seems physically understandable . specifically , at lower redshifts , while the general population of galaxies gradually become less cold gas - rich , dla hosts seem to pick out those that are gas richer and bluer , but not necessarily of higher star formation rate .
with high resolution , physically sound treatment of all relevant physical processes , our state - of - the - art , adaptive mesh - refinement eulerian cosmological hydrodynamic simulations can reproduce a whole array of observables of damped lyman alpha systems , including the distributions of the following quantities and their evolution with redshift : column density , @xmath121808 absorption line velocity width , kinematic shape measures of the line , metallicity , size , line density , hi mass density . this allows to examine , in addition , with significant confidence , the properties of dla host galaxies .
we are able to reach the following conclusions .
\(1 ) dla hosts roughly trace the overall population of galaxies at all redshifts , with respect to halo mass , star formation rate , hi mass and color , with a noticeable tendency of being smaller and bluer , on average .
the standout selection criterion seems being gas rich .
most of dla hosts appear to have hi mass peaked at @xmath253 at all redshifts with @xmath254 covering nearly all dla hosts .
the majority of dla hosts are gas - rich ( @xmath256 ) closely following the general population of galaxies at high redshift ; by @xmath6 most of the stars are in systems with ( @xmath257 ) , but majority of dla hosts are still gas rich ( @xmath256 ) .
\(2 ) it seems that the history of dla evolution is cosmological in nature and reflects the underlying cosmic density evolution , galaxy evolution and galaxy interactions . with higher density and more interactions at high redshift dlas
are larger in both absolute terms and in relative terms with respect to virial radii of halos . at @xmath0 galactocentric distance of dlas
is , on average , @xmath265 , whereas at @xmath6 it decreases to @xmath266 .
the typical dla impact parameter is @xmath267kpc at @xmath0 and @xmath189kpc at @xmath6 .
the typical size ( radius ) of individual dlas is @xmath126kpc at @xmath268 dropping to @xmath269kpc at @xmath217 , in agreement with observations , then increasing to @xmath270kpc at @xmath6 .
\(3 ) the variety of dlas at high redshift is richer with a large contribution coming from galactic filaments - filamentary gaseous structures sometimes extending to as far as virial radius - that are created through close galaxy interactions .
the portion of gaseous disks of galaxies where most stars reside makes relatively small contribution to dla incidence at @xmath0 . by @xmath6 galaxy interactions
have become rare , cosmic mean density has decreased dramatically , so what is left to contribute to dla incidence is gas rich disks of galaxies in the mass range of @xmath271 .
\(4 ) galactic winds play an indispensable role in shaping the kinematic properties of dlas .
specifically , the high velocity width dlas are a mixture of those arising in high mass , high velocity dispersion halos and those arising in smaller mass systems where cold gas clouds are entrained to high velocities by galactic winds .
closer examination reveals that most of the large width dlas are due to multiple physically distinct components of varying los velocities within a @xmath272kpc separation ( along the los ) .
\(5 ) quantitatively , the vast majority of dlas arise in halos of mass @xmath1 at @xmath2 , as these galaxies dominate the overall population of galaxies then . at @xmath0 , 20 - 30% of dla hosts
are lyman break galaxies ( lbgs ) , 10 - 20% are due to galaxies more massive than lbgs and 50 - 70% are from smaller galaxies .
the fraction of lbg dla hosts drops to @xmath243 by @xmath244 .
\(6 ) in agreement with observations , we see a weak but noticeable evolution in dla metallicity . the metallicity distribution centers at @xmath3=-1.5 $ ] to @xmath4 and spans more than three decades at @xmath0 , with the peak moving to @xmath3=-0.75 $ ] at @xmath5 and @xmath3=-0.5 $ ] by @xmath6 .
the overall metallicity seems floored at @xmath3\sim -3 $ ] at @xmath2 and at @xmath3\sim -1.5 $ ] at @xmath6 . when most of the dla incidence is due to cold gas outside the stellar disks at high redshift , not only the metallicity is relatively low , but also there is little star formation within
this explains self - consistently their low metallicity and lack of star formation activity .
if and when significant internal star formation occurs , there are two possible scenarios . if that takes place outside galactic disk , the dla will be destroyed and remove itself from the dla category . if it takes place on galactic disk , the dla will either be destroyed permanently , or temporarily and then cool back down to become a more metal - enriched one , after star formation has stopped .
the former occurs predominantly at high redshift , which is why most of dlas are metal poor .
the latter occurs at low redshift , which explains the significant increase of metallicity .
\(7 ) the star formation rate of dla hosts is , however , quite strong , heavily concentrated in the range @xmath7/yr at @xmath0 , gradually shifting lower to peak at @xmath8/yr by @xmath6 .
the finding that the typical size and galactocentric distance of dlas are both large and comparable gives a moderate , apparent " star formation rate seen in @xmath9 emission due to fluorescence of ionizing photons from the host galaxies .
both the size of @xmath9 emission and the magnitude of this fluorescence is , curiously , consistent with the population of faint @xmath9 emitters observed by @xcite , if they are at @xmath13 .
\(8 ) finally , we suggest that a significant fraction of sky seen by gamma - ray bursts should be covered by circumgalactic dlas at high redshift ( @xmath232 ) . they may be identified with those grb - dlas that have little h@xmath273 column densities and are relatively metal and dust poor .
i would like to thank dr .
m.k.r . joung for help on generating initial conditions for the simulations and running a portion of the simulations and greg bryan for help with enzo code .
i would like to thank dr .
jason x. prochaska and dr .
andrew pontzen for kindly providing the observational data , dr . sara l. ellison for sending a paper draft before publication and for discussion , dr .
edward jenkins for discussion on atomic data .
computing resources were in part provided by the nasa high- end computing ( hec ) program through the nasa advanced supercomputing ( nas ) division at ames research center .
this work is supported in part by grants nnx08ah31 g and nas8 - 03060 .
the simulation data are available from the author upon request . 70 natexlab#1#1 , t. , anninos , p. , zhang , y. , & norman , m. l. 1997 , new astronomy , 2 , 181 , k. l. , steidel , c. c. , pettini , m. , shapley , a. e. , reddy , n. a. , & erb , d. k. 2005 , , 619 , 697 , g. , & charlot , s. 2003 , , 344 , 1000 , g. l. 1999 , comput .
, vol . 1 , no . 2 , p. 46 - 53 , 1 , 46 bryan , g. l. , & norman , m. l. 1999 , in structured adaptive mesh refinement grid methods , ed .
n. p. c. s. b. baden ( i m a volumes on structured adaptive mesh refinement methods , no .
117 ) , 165 , f. , trujillo , i. , conselice , c. j. , bouwens , r. j. , dickinson , m. , & yan , h. 2008 , , 687 , l61 , r. , kang , h. , ostriker , j. p. , & ryu , d. 1995 , , 451 , 436 , r. , nagamine , k. , & ostriker , j. p. 2005
, , 635 , 86 , r. , & ostriker , j. p. 1992
, , 399 , l113 , r. , ostriker , j. p. , prochaska , j. x. , & wolfe , a. m. 2003 , , 598 , 741 , r. , pettini , m. , steidel , c. c. , king , l. j. , rudie , g. c. , & rakic , o. 2010 , arxiv e - prints , a. , & mccray , r. a. 1972 , , 10 , 375 , a. , & birnboim , y. 2006 , , 368 , 2 , a. , birnboim , y. , engel , g. , freundlich , j. , goerdt , t. , mumcuoglu , m. , neistein , e. , pichon , c. , teyssier , r. , & zinger , e. 2009 , , 457 , 451 eisenstein , d. , & hu , p. 1999 , apj , 511 , 5 , s. l. , prochaska , j. x. , hennawi , j. , lopez , s. , usher , c. , wolfe , a. m. , russell , d. m. , & benn , c. r. 2010 , , 814 , d. k. , shapley , a. e. , pettini , m. , steidel , c. c. , reddy , n. a. , & adelberger , k. l. 2006 , , 644 , 813 ferguson , h. c. , dickinson , m. , giavalisco , m. , kretchmer , c. , ravindranath , s. , idzi , r. , taylor , e. , conselice , c. j. , et al .
2004 , apj , 600 , l107 gardner , j. , katz , n. , hernquist , l. , & weinberg , d. h. 1997a , apj , 484 , 31 . 2001 , apj , 559 , 131 , j. p. , katz , n. , weinberg , d. h. , & hernquist , l. 1997 , , 486 , 42 , f. , & madau , p. 1996
, , 461 , 20 , m. g. , steinmetz , m. , & rauch , m. 1998 , , 495 , 647 , t. m. 2001 , in astronomical society of the pacific conference series , vol .
240 , gas and galaxy evolution , ed .
j. e. hibbard , m. rupen , & j. h. van gorkom , 345 , t. m. , lehnert , m. d. , strickland , d. k. , & armus , l. 2000 , , 129 , 493 , m. , & sofue , y. 2002 , , 54 , 223 , j. , mller , p. , gorosabel , j. , fynbo , j. p. u. , toft , s. , jaunsen , a. o. , kaas , a. a. , pursimo , t. , torii , k. , kato , t. , yamaoka , h. , yoshida , a. , thomsen , b. , andersen , m. i. , burud , i. , castro cern , j. m. , castro - tirado , a. j. , fruchter , a. s. , kaper , l. , kouveliotou , c. , masetti , n. , palazzi , e. , pedersen , h. , pian , e. , rhoads , j. , rol , e. , tanvir , n. r. , vreeswijk , p. m. , wijers , r. a. m. j. , & van den heuvel , e. p. j. 2003 , , 597 , 699 , s. , katz , n. , dav , r. , fardal , m. , kere , d. , & oppenheimer , b. d. 2010 , arxiv e - prints , m. r. , cen , r. , & bryan , g. l. 2009 , , 692 , l1 , n. , weinberg , d. h. , hernquist , l. , & miralda - escude , j. 1996 , , 457 , l57 + kennicutt , r. c. j. 1998 , apj , 498 , 541 , d. , katz , n. , weinberg , d. h. , & dav , r. 2005 , , 363 , 2 , e. , smith , k. m. , dunkley , j. , bennett , c. l. , gold , b. , hinshaw , g. , jarosik , n. , larson , d. , nolta , m. r. , page , l. , spergel , d. n. , halpern , m. , hill , r. s. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2010 , arxiv e - prints , c. , vreeswijk , p. m. , smette , a. , fox , a. j. , petitjean , p. , ellison , s. l. , fynbo , j. p. u. , & savaglio , s. 2009 , , 506 , 661 , e. m. , soderberg , a. m. , kewley , l. j. , & berger , e. 2010 , arxiv e - prints lowenthal , j. d. , koo , d. c. , guzman , r. , gallego , j. , phillips , a. c. , faber , s. m. , vogt , n. p. , illingworth , g. d. , et al .
1997 , apj , 481 , 673 , a. i. , & woosley , s. e. 1999 , , 524 , 262 , a. h. , prochaska , j. x. , somerville , r. s. , & primack , j. r. 2001 , , 326 , 1475 , c. l. 2005 , , 621 , 227 , p. , warren , s. j. , fall , s. m. , fynbo , j. u. , & jakobsen , p. 2002
, , 574 , 51 , e. m. , turnshek , d. a. , & rao , s. 2009 , , 397 , 943 , d. c. 2003 , , 149 , 205 , k. , springel , v. , & hernquist , l. 2004 , , 348 , 421 . 2004 , , 348 , 435 , k. , wolfe , a. m. , hernquist , l. , & springel , v. 2007 , , 660 , 945 , b. w. , bryan , g. , bordner , j. , norman , m. l. , abel , t. , harkness , r. , & kritsuk , a. 2004 , arxiv astrophysics e - prints , c. , mcmahon , r. g. , storrie - lombardi , l. j. , & irwin , m. j. 2003 , , 346 , 1103 , a. , governato , f. , pettini , m. , booth , c. m. , stinson , g. , wadsley , j. , brooks , a. , quinn , t. , & haehnelt , m. 2008 , , 390 , 1349 prochaska , j. x. , gawiser , e. , wolfe , a. m. , castro , s. , & g. , g. d. 2003 , apj , 595 , l9 , j. x. , herbert - fort , s. , & wolfe , a. m. 2005 , apj , 635 , 123 prochaska , j. x. , & wolfe , a. m. 1997 , apj , 487 , 73 , j. x. , & wolfe , a. m. 2009 , , 696 , 1543 . 2010 , arxiv e - prints , s. m. , turnshek , d. a. , & nestor , d. b. 2006 , , 636 , 610 , m. , haehnelt , m. , bunker , a. , becker , g. , marleau , f. , graham , j. , cristiani , s. , jarvis , m. , lacey , c. , morris , s. , peroux , c. , rttgering , h. , & theuns , t. 2008 , , 681 , 856 , a. o. , norman , m. l. , prochaska , j. x. , sommer - larsen , j. , wolfe , a. m. , & yang , y. 2008 , , 683 , 149 , a. o. , norman , m. l. , prochaska , j. x. , & wolfe , a. m. 2006 , , 645 , 55 , j. 2001 , , 559 , l1 , p. r. , & field , g. b. 1976 , , 205 , 762 , c. c. , giavalisco , m. , pettini , m. , dickinson , m. , & adelberger , k. l. 1996 , , 462 , l17 + , l. j. , & wolfe , a. m. 2000 , apj , 543 , 552 , e. , viel , m. , tornatore , l. , & borgani , s. 2009 , , 397 , 411 , s. , van dokkum , p. , franx , m. , labbe , i. , frster schreiber , n. m. , wuyts , s. , webb , t. , rudnick , g. , zirm , a. , kriek , m. , van der werf , p. , blakeslee , j. p. , illingworth , g. , rix , h. , papovich , c. , & moorwood , a. 2007 , , 671 , 285 , i. , frster schreiber , n. m. , rudnick , g. , barden , m. , franx , m. , rix , h. , caldwell , j. a. r. , mcintosh , d. h. , toft , s. , hussler , b. , zirm , a. , van dokkum , p. g. , labb , i. , moorwood , a. , rttgering , h. , van der wel , a. , van der werf , p. , & van starkenburg , l. 2006 , , 650 , 18 , j. , prochaska , j. x. , chen , h. , dessauges - zavadsky , m. , & bloom , j. s. 2007 , , 668 , 667 , a. m. , & chen , h. 2006 , , 652 , 981 , a. m. , turnshek , d. a. , smith , h. e. , & cohen , r. d. 1986 , , 61 , 249 , s. e. , & heger , a. 2006 , , 637 , 914 , a. w. , van der wel , a. , franx , m. , labb , i. , trujillo , i. , van dokkum , p. , toft , s. , daddi , e. , rudnick , g. , rix , h. , rttgering , h. j. a. , & van der werf , p. 2007
, , 656 , 66 , m. a. , & prochaska , j. x. 2006 , , 643 , 675
in figure [ fig : colhisconverge ] we show a comparison of dla column density distributions for the c " and c/2 " run at @xmath164 .
recall that
c/2 " run has a spatial resolution lower than c " run by a factor of 2 . while the agreement is not perfect , it is reasonably good and the difference is at @xmath274 across the dynamic range .
figure [ fig : v90hisconverge ] compares velocity width ( @xmath61 column density distributions for the c " and c/2 " run at @xmath164 , where we see that the level of agreement is comparable to that seen in figure [ fig : colhisconverge ] .
figure [ fig : mtlhisconverge ] shows a comparison of the metallicity distribution for the c " and c/2 " run at @xmath164 .
it it seen that the metallicity distribution agrees to within about @xmath275dex and it appears that a still higer resolution simulation might give a somewhat lower metallicity , perhaps in still better agreement with observations .
figure [ fig : sizehisconverge ] shows a comparison of the dla size distribution for the c " and c/2 " run at @xmath164 . since the spatial resolution more directly affects the size , here we see the worst disagreement between the two runs in terms of the shape of the distribution , but overall still mostly at a level of @xmath276dex .
however , a direct comparison of probability for qso binary sightlines shown in figure [ fig : pdfcolradconverge ] suggests that the overall convergence is quite good . | using adaptive mesh - refinement cosmological hydrodynamic simulations with a physically motivated supernova feedback prescription we show that the standard cold dark matter model can account for extant observed properties of damped lyman alpha systems ( dlas ) .
we then examine the properties of dla host galaxies .
we find : ( 1 ) while dla hosts roughly trace the overall population of galaxies at all redshifts , they are always gas rich and have tendencies of being slightly smaller and bluer . ( 2 ) the history of dla evolution is cosmological in nature and reflects primarily the evolution of the underlying cosmic density , galaxy size and galaxy interactions . with higher density and more interactions at high redshift dlas
are larger in both absolute terms and in relative terms with respect to virial radii of halos .
( 3 ) the variety of dlas at high redshift is richer with a large contribution coming from galactic filaments , created through close galaxy interactions .
the portion of gaseous disks of galaxies where most stars reside makes relatively small contribution to dla incidence at @xmath0 .
( 4 ) the vast majority of dlas arise in halos of mass @xmath1 at @xmath2 , as these galaxies dominate the overall population of galaxies then . at @xmath0 ,
20 - 30% of dla hosts are lyman break galaxies ( lbgs ) , 10 - 20% are due to galaxies more massive than lbgs and 50 - 70% are from smaller galaxies .
( 5 ) galactic winds play an indispensable role in shaping the kinematic properties of dlas . specifically , the high velocity width dlas are a mixture of those arising in high mass , high velocity dispersion halos and those arising in smaller mass systems where cold gas clouds are entrained to high velocities by galactic winds . ( 6 ) in agreement with observations , we see a weak but noticeable evolution in dla metallicity .
the metallicity distribution centers at @xmath3=-1.5 $ ] to @xmath4 and spans more than three decades at @xmath0 , with the peak moving to @xmath3=-0.75 $ ] at @xmath5 and @xmath3=-0.5 $ ] by @xmath6 .
( 7 ) the star formation rate of dla hosts is concentrated in the range @xmath7/yr at @xmath0 , gradually shifting lower to peak at @xmath8/yr by @xmath6 .
= 1 |
recent progress in astroparticle physics has improved our level of understanding of the outstanding unsolved problems concerning the origin , acceleration , and composition of primary cosmic rays ( pcrs ) during over past @xmath0 years since its discovery [ 1 ] at ultra - high energy ( uhe ) range with the continuous progress in experimental techniques and methods of measurements .
nowadays we have relatively more sensitive eas experiments [ 2 - 5 ] consisting of a variety of modern detectors to observe the secondary components in an eas that contribute important results . on the other hand , to arrive at any specific conclusions about cosmic - rays ( crs ) from their indirect investigation it is very important to know how they interact with the atmosphere and how the eas develops .
this knowledge is obtained by monte carlo ( mc ) simulations which are tested against data .
hence air shower simulations are a crucial part of the design of air shower experiments and analysis of their data . but a mc technique relies heavily on high energy hadronic models which suffer to some degree of uncertainties from one model to another and increasing primary energy .
therefore , we are challenged to develop more accurate hadronic interaction models in place to predict robust results .
recently the lhc [ 6 ] data have been tuned in hadronic interaction models namely epos - lhc [ 7 ] and qgsjet-04 [ 8 ] , those are now included in the mc code _ corsika _ version 7.400 [ 9 ] .
these recent efforts have improved the predictive power of eas simulations significantly . to characterize an eas initiated by any type of primary species
, one has to know at least about its shower size ( more specifically total charged particle size or electron size ( @xmath1 ) , muon size ( @xmath2 ) , or hadron size ( @xmath3 ) ) and shower age ( @xmath4 ) [ 10 ] .
near shower maximum , the shower size / electron size is closely related to the energy of the primary particle [ 11 ] . on the other hand ,
eas parameters like @xmath2 , @xmath4 etc . have been used consistently to determine the nature of the shower initiating particle .
measurements of all these parameters of an individual shower are made either by an individual or hybrid detection method .
these detection techniques require an array of scintillation detectors and assembly of proportional counters as muon tracking detectors ( mtd ) respectively . from the measurement , the particle density distribution data ( @xmath5 or @xmath6 ) at the observation level
are obtained .
these data will then be used to reconstruct a shower with the help of a suitable lateral density profile . in the cascade theory ,
such a lateral density profile of cascade particles can be approximated at sea level by the well known nishimura - kamata - greisen ( nkg ) structure function [ 12 ] .
the radial distribution of cascade particles in an average eas is generally assumed to be symmetrical in the plane perpendicular to the shower axis .
but , presence of intrinsic fluctuations ( due to stochastic nature of eas development ) from shower to shower , in addition higher zenith angles ( @xmath7 ) and geomagnetic effects ( ges ) can perturb this axial symmetry noticeably . such effects
may even break the axial symmetry a little to the distribution of eas particles even in vertically incident showers .
inclined showers though experience similar effects as vertical showers but manifest significantly large asymmetries . and @xmath8 generated from a parent particle in an eas by the geomagnetic field in two different situations.,title="fig : " ] and @xmath8 generated from a parent particle in an eas by the geomagnetic field in two different situations.,title="fig : " ] disregarding the ge , in ground array experiments the analysis of the density data is usually performed by assuming axial symmetry .
therefore , asymmetries would come from the polar variation of the charged eas particles and unequal attenuation accounted from different locations of the eas in the ground plane with inclined incidence .
these are known as geometrical and atmospheric attenuation effects to polar asymmetries . to accentuate the ges on the eas charged particle distribution alone ,
the geometric and attenuation effects must be isolated or corrected out in the analysis .
the data analysis technique which we are going to introduce here will remove the polar asymmetry caused by the geometric effect , while the asymmetry resulting from the attenuation effect would be ignored judiciously as muons come across very little attenuation in the atmosphere . in this work , it is shown that the asymmetry arises in polar distribution of @xmath9-s predominantly from the geomagnetic field ( gf ) , may explore a new possibility for the determination of the chemical composition of primary crs . in this paper , we address the influence of the gf on the spatial distribution of @xmath9-s with a general limit on the @xmath7 ( @xmath10 ) , valid for all applications of the technique , choosing the kascade experiment [ 2 ] site and look at the charge separation between @xmath11 and @xmath12 to arrive at a possible mass dependent parameter , called the _ transverse muon barycenter separation _ ( tmbs ) .
this tmbs is expected to be dependent on the nature of the primary particle and hence , in principle , the parameter can be exploited to estimate primary mass .
the method presented here is applied to mc data simulated in three limited primary energy regions : @xmath13 , @xmath14 and @xmath15 pev ( to obtain sufficient number of eas events at these narrow energy ranges by our available computing power ) . the analysis described in this work
is based on mc simulations carried out with the code _ corsika _ ( * co*smic * r*ay * si*mulation for * ka*scade ) [ 9 ] .
we also discuss the practical realization of the proposed method in a real experiment . in this paper , basics of the influence of the gf on the eas cascade , which are of direct relevance to this work ,
is discussed in the following section . in section iii
, we report on the simulation procedure adapted here .
the mc data analysis technique is discussed in section iv . in section
v , we present our results with discussion obtained from ges on eas muons . finally , a possible experimental approach of the present method and concluding remarks are pointed out in section vi .
it is well known that the gf causes the east - west asymmetry on the primary crs according to their rigidity and energy .
the study of crs with primary energies above @xmath16 pev is usually based on the measurements of eass , which are essentially cascades of secondary particles produced by interactions of cr particles with atmospheric nuclei . during the development of a cr cascade in the atmosphere
, the gf affects the propagation of secondary charged particles in the shower : the perpendicular component of this field causes the trajectories of secondary charged particles to become curved , with positive and negative charged particles deflecting to produce a overall transverse barycenter separation .
this aspect was first pointed out by cocconi nearly sixty years back [ 13 ] .
he further suggested that the geomagnetic broadening effect can be non - negligible in compare to the coulomb scattering , particularly for the young showers .
since then , several studies have been carried out to address the influence of the gf on cr eas and some important effects arising out of it were also reported ( see for instance [ 14 - 30 ] and references therein ) .
we should however state that some of these effects of the gf mentioned in the references have no direct relevance to this particular study . for the soft component ( @xmath5 ) ,
the radiation lengths in the atmosphere are short and they suffer many scatterings with @xmath17 ( where , @xmath18 mev in air ) and stronger bremsstrahlung effect ( when @xmath19 ) thereby randomly changing the directions of their momenta relative to the gf . as a result , the lateral spread of electrons is mainly due to the multiple coulomb scattering and bremsstrahlung , and hence the effect of gf is less pronounced [ 31 ] .
in contrast , after their generation from pion and kaon decays ( mainly those in first few generations from their parent particles at great atmospheric heights ) , @xmath9-s travel much longer paths encountering negligible scattering with the medium ( suffer lesser bremsstrahlung also ) , and hence come under the influence of gf noticeably . as a result
, ge should be more pronounced in medium to high momenta @xmath9-s than lower ones , particularly for very large and strongly inclined showers .
such a situation is depicted by the figure 1 , where , * top * : separation of @xmath20 and @xmath8 generated from a parent particle in an eas comes from the west and advances into the east direction , and * bottom * : same as the * top * , but the eas comes from the south and advances into the north direction in the _ corsika _ coordinate system .
besides , in highly inclined showers , a high percentage of eas @xmath5 is absorbed in the atmosphere , may reduce shower - to - shower fluctuations to a great extent . exploiting this feature , some earlier mc simulation studies [ 24 ] reported that heavy nuclei and proton induced showers may be discriminated from the elliptic footprint of lateral muon distribution and the muon charge ratio ( the ratio of @xmath20 to @xmath8 numbers ) at convenient distances from the shower core . through our preliminary study on this aspect of gf ,
their prediction was substantiated and reported in [ 32 ] from mc data .
this paper includes more accurate data analysis technique , and quantifies different new features compared to earlier studies in [ 24,32 ] , in order to obtain the primary mass information by the gf effects alone .
in the framework of the air shower simulation program _ corsika _ of version @xmath21 [ 9 ] , the eas events are simulated by combining the high energy ( above @xmath22 ) hadronic interaction models qgsjet @xmath23 version @xmath24c [ 33 ] and epos @xmath25 [ 34 ] , and the low energy ( below @xmath22 ) hadronic interaction model urqmd [ 35 ] .
the egs@xmath26 program library is chosen for simulation of the electromagnetic ( em ) component of shower that incorporates all the major interactions of electrons and photons [ 36 ] .
we consider the us - standard atmospheric model with planar approximation , which works for @xmath7 of the primary particles being less than @xmath27 [ 37 ] .
events have also been generated at zenith angles @xmath28 with curved option of the atmosphere in the standard _
corsika _ program [ 9 ] .
the eas events have been simulated at the geographical location of the experimental site of kascade ( latitude @xmath29 n , longitude @xmath30 e , @xmath31 m a.s.l . ) [ 2 ] .
the gf with a homogeneous field approximation is considered . in order to examine the effect of the gf ,
eas events are also simulated by switching off earth s magnetic field . on the observation level ,
the detection kinetic energy thresholds are chosen as @xmath32 mev for electrons ( @xmath5 ) and @xmath33 mev for muons .
the eas events have been generated for proton ( p ) and iron ( fe ) primaries at three different limited primary energy regions as mentioned in the sect .
i. at each energy range , we have taken five different narrow ranges of @xmath7 , @xmath34 , @xmath35 , @xmath36 , @xmath37 and @xmath38 respectively with flat option . a small sample of eas events
have also been simulated with ranges @xmath39 and @xmath40 with curved option of the atmosphere . in order to have at least some hints of fluctuations to our important observables ,
we have generated showers in those limited ranges of energy and @xmath7 .
if we simulate eas events in a wider range instead ( say , @xmath41 ) , most of the events will then have @xmath7 falling in @xmath42 range , but for which @xmath9-s will not remain a crucial component compare to other charged secondaries in the eas .
the proportion of @xmath9-s in an eas rises with increasing @xmath7 relative to the em component , and hence the gf on @xmath9-s will be effective at higher @xmath7 only .
we have used thinning option of _ corsika _ for the primary energy range @xmath43 pev by taking @xmath44 as thinning factor according to the optimum weight limitation [ 38 ] .
we have restricted all showers to two azimuthal angles : @xmath45 and @xmath46 ( only north and west directions in the coordinate system of _ corsika _ ) .
about @xmath47 showers have been generated for each @xmath7 range in the energy region @xmath13 pev and @xmath48 events for each @xmath7 range in other two energy regions .
when the gf is disabled ( setting @xmath49 times smaller than the magnetic fields at the kascade location ) in the simulation program , the lateral distribution of eas charged particles possess cylindrical symmetry for all radial distances around the eas axis in the shower front plane . in the observed plane , however , such cylindrical symmetry is distorted for inclined eas due to the geometrical and the atmospheric attenuation effects . when the gf is switched on , polar asymmetry arising from the ge
is superimposed with those caused by the geometric and attenuation effects .
therefore , for isolating or correcting out the effect contributed by the gf only from other distortions , such as those added by other two effects , the following part of this section has to be worked out during data analysis .
observables such as the muon density ( @xmath50 ) or @xmath51 obtained from the analysis of simulated data are overestimated in the early region of an eas while in the late region are underestimated .
hence , @xmath9-s in the first zone do undergo lesser attenuation than those arriving in the later zone .
this development feature of the eas accounts the attenuation contribution to the overall polar asymmetry in a particular observable under consideration , and is called the attenuation effect .
the polar asymmetries are also present , if the given observable is measured in the ground plane .
such an asymmetry appears from another effect , is known as the geometric effect . to extract the actual variation introduced by the geometric and attenuation effects in eas observables , as a first step it is necessary to take away the contribution added geometrically from data .
the correction results from the geometric effect can be briefly summarized by defining the relevant quantities involved .
a projection procedure is applied for transforming the positional information of the point of impact of each particle in the ground plane onto the shower front plane . in the figure 2 , we have shown the transformation of a point of impact by a cascade @xmath9 in the the ground ( @xmath52 mark ) with polar coordinates , ( @xmath53,@xmath54 ) onto the shower plane with the set of coordinates , ( @xmath55,@xmath56 or @xmath57 , @xmath58 ) .
the necessary transformation relations are trivial and will be as follows , @xmath59 using the figure 2 , the corresponding cartesian coordinates @xmath60,@xmath61 at the shower plane can be easily obtained as , @xmath62 @xmath63 , where @xmath64 denotes azimuthal angle of the eas . in the geometric correction ,
interaction due to attenuation process of @xmath9-s was not considered in the region between the two planes .
the polar asymmetries in the shower front plane reflects the effect of attenuation of the @xmath9 component when the magnetic field is disabled , provided interaction of muons from the intermediate space between the two planes is taken into account .
there are a number of reasons to ignore the attenuation effect on the @xmath9 component in the space between the two planes .
all these reasons have been substantiated by the following studies using simulated data .
first , we have studied the average longitudinal development of @xmath65 and @xmath66 for higher @xmath7 , to see the attenuation power between the em and muonic components of eas after the shower maximum .
the curves in the figure 3 illustrate how @xmath66 and @xmath65 contents of eas vary during the eas evolution from the depth of its size maximum .
it is clear from the figure that the @xmath9 component remains nearly insulated from the atmospheric attenuation especially in the space between the two planes .
this particular study involves all the charged muons with @xmath67 mev , but the present study considers very energetic muons with @xmath68 gev / c .
important reasons behind such momenta selection will be discussed later .
these energetic @xmath66 , generated from their parent particles at great atmospheric heights are expected to be much weaker against attenuation .
secondly , polar asymmetries of truncated @xmath65 and @xmath66 contents in the ground plane normally arise due to both the geometric and attenuation effects when the gf is switched off in the simulation .
but , a distinguishing character relating to the attenuation effect can be realized between these two polar variations for @xmath65 and @xmath66 numbers in the ground plane at @xmath69 .
it is expected that the geometric effect will enhance the polar asymmetry equally to both @xmath65 and @xmath66 variations for a given truncated region within a polar angle range .
the polar variations of @xmath65 and @xmath66 numbers in three different truncated regions ( @xmath70 ) m within a polar angle interval of @xmath71 , each centered around increasingly different mean values of @xmath72 in the ground plane , are shown in plots through the figure 4 ( @xmath73 m is not considered owing to large fluctuations near the eas core ) . at a glance , these plots as if simply showing almost similar polar variations of @xmath65 and @xmath66 arising from these two above mentioned effects .
but , some apparently unseen quantities as ratios of the highest-*h * to the lowest-*l * values of numbers of either types of particles separately from the figures 4a and 4b indicate that asymmetry in the em component is about @xmath74 range higher than the @xmath9-s for three truncated limits in @xmath75 .
we have verified that these differences will increase further with increasing @xmath7 and @xmath76 .
hence , it is found that the attenuation contribution to the asymmetry by @xmath9-s is lower than the em component from the comparison between these two figures , and may even be neglected in the space between these two planes under consideration . moreover , some other studies related to the attenuation power of the @xmath9-component from the work in [ 39 - 40 ] , backing the insignificant contribution by @xmath9-s in the concerned space between two planes according to the figure 2 .
the kascade - grande studied the @xmath51 attenuation with the slant depth ( sec@xmath7 ) , using the constant intensity cut ( cic ) method [ 41 ] in the primary energy range @xmath77 pev , and they obtained little higher value for the attenuation length ( @xmath78 ) of @xmath51 compared to mc data .
the observed value of ( @xmath78 ) was @xmath79 g @xmath80 , while the mc data yielded @xmath81 g @xmath80 and @xmath82 g @xmath80 respectively for lighter and relatively heavier compositions . at these energies , the attenuation length ( @xmath83 ) of @xmath84 obtained by the kascade experiment was less than @xmath85-th of the @xmath78 [ 39 , 41 - 42 ] .
this result also reiterates the fact that the attenuation of the @xmath9-component is much smaller than the em component , and therefore has negligible effect on the polar asymmetry in the shower plane .
a very recent study on the atmospheric attenuation of different eas particles in the concerned space , introducing a newly defined parameter , called a shift of the eas core also agrees with the insignificant trait of the attenuation effect from @xmath9-s [ 40 ] .
@xmath86 & @xmath87 & total - @xmath9 & q + & & & & & + @xmath20 & @xmath88 & @xmath89 & @xmath90 & 1816 & n + & & & & & + @xmath8 & @xmath88 & @xmath91 & @xmath92 & 1805 & n + & & & & & + @xmath20 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & * y * + & & & & & + @xmath8 & @xmath93 & @xmath97 & @xmath98 & @xmath99 & * y * + & & & & & + @xmath20 & @xmath100 & @xmath101 & @xmath102 & 1319 & n + & & & & & + @xmath8 & @xmath100 & @xmath103 & @xmath104 & 1273 & n + & & & & & + ( gev / c ) & @xmath86 & @xmath87 & @xmath105 ( gev / c ) & @xmath86 & @xmath87 + & & & & & & & + @xmath20 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xmath112 + & & & & & & & + @xmath8 & @xmath106 & @xmath107 & @xmath113 & @xmath114 & @xmath110 & @xmath115 & @xmath116 + & & & & & & & + @xmath20 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & @xmath110 & @xmath121 & @xmath122 + & & & & & & & + @xmath8 & @xmath117 & @xmath107 & @xmath123 & @xmath124 & @xmath110 & @xmath125 & @xmath126 + & & & & & & & + @xmath20 & @xmath127 & @xmath107 & @xmath128 & @xmath129 & @xmath110 & @xmath94 & @xmath130 + & & & & & & & + @xmath8 & @xmath127 & @xmath107 & @xmath131 & @xmath132 & @xmath110 & @xmath133 & @xmath98 + & & & & & & & + the analysis selects @xmath9-s which qualify two selection criteria simultaneously related to @xmath9 detection area and their energies or momenta . as already mentioned that this work considers relatively high energy @xmath9-s , those are expected to originate from first few interactions in the upper part of the atmosphere and travel much longer paths , and experiencing the gf for a longer time .
this is consistent with the expectations since only the high momenta @xmath9-s will survive against attenuation especially in highly inclined showers .
the main observables , tmbs and mtmbs ( maximum value of tmbs ) are expected to be dependent upon the energy of incoming @xmath9-s .
this work has identified some suitable conditions for detection and measurement of these observables with reasonable @xmath9 numbers or densities .
this situation essentially demands a best compromise among @xmath9 detector size , @xmath9 energy threshold and @xmath51 .
the visible effects of the gf are emphasized in the case of very inclined showers with high momentum @xmath9-s ( @xmath134 gev / c ) .
the selection of high altitude site is useful for observing high energy @xmath9-s too .
otherwise , simulations of showers from very high energy primaries even at sea level , such as the kascade here , may produce high energy @xmath9-s as well .
the practical realization of the present approach in the work requires @xmath9 detectors at least at two opposite regions as shown in the last figure 12 ( corresponding to showers coming from the north , west directions ) in an array containing several scintillation detectors .
therefore , one of the main objectives of the work is to employ @xmath9 detectors into the array with some reasonable sizes from the point of view of their construction cost .
we have used different combinations of momentum thresholds and @xmath9 detection areas as trials so that a better option may be surfaced , which will deliver an optimal asymmetry between @xmath11 and @xmath12 particles with a reasonable statistics .
it has been seen that the low momenta @xmath9-s ( below @xmath135 gev / c ) offer nearly symmetrical distribution in the x - y shower plane whereas @xmath9-s with momenta falling in the range @xmath134 gev / c manifest a better polar asymmetry in @xmath11 and @xmath12 numbers in the annular region between @xmath136 m and @xmath137 m from the eas core .
table i and table ii give a clear view on @xmath9-s selection explained above .
the highlighted figures in both the tables correspond better combinations between @xmath9-s thresholds and their detection areas , for which an optimum local contrast in the abundance of @xmath20 and @xmath8 could be achieved .
data other than highlighted ones in those tables would either provide lower percentage of @xmath20 and @xmath8 abundances at two detector locations out of a relatively higher total or higher percentage of @xmath20 and @xmath8 abundances at detector locations out of a relatively lower total .
such combinations are found unsuitable for estimating tmbs and mtmbs with some anticipated uncertainties . in the upcoming sections
, we shall use only these selected @xmath9-s having @xmath138 gev / c and @xmath139 m for our important results .
to examine the polar asymmetries of the lateral distributions of @xmath20 and @xmath8 due to the gf , we have estimated total number of each variety of particles over an arc region of truncated core distance range of @xmath140 m and central angle @xmath141 at different polar positions in both the planes .
the polar variations of @xmath20 and @xmath8 ( @xmath138 gev / c ) in the shower plane with @xmath142 and @xmath143 are shown in figures 5a and 5b for fe initiated showers corresponding to @xmath144 and @xmath145 . in the figure 5c ,
the polar asymmetries of these @xmath9-s in both the planes are depicted , while in the figure 5d , we have used showers having @xmath146 and the polar variation is limited to the shower plane only .
however , it should be mentioned that these studies were also covered by other papers [ 25 , 27 ] using @xmath9 muon densities from different perspectives . and
@xmath8 for iron primary arriving from the north and west directions.,title="fig:",scaledwidth=45.0% ] and @xmath8 for iron primary arriving from the north and west directions.,title="fig:",scaledwidth=45.0% ] and @xmath8 for iron primary arriving from the north and west directions.,title="fig:",scaledwidth=45.0% ] and @xmath8 for iron primary arriving from the north and west directions.,title="fig:",scaledwidth=45.0% ] for clear understanding of the influence of the gf , we have studied the polar variations of @xmath20 and @xmath8 contents separately in figures 5a and 5b by turning on and off ( dividing the components of the gf by a factor of @xmath147 ) the gf keeping all remaining parameters unchanged in the mc simulation program .
the nature of these polar variations are as per our expectation .
the figure 5c is a combined study on @xmath20 and @xmath8 variations including the ground plane for @xmath148 only .
inclusion of the ground plane in the figure is merely for observing asymmetries compared to the shower plane .
our subsequent studies will not include the ground plane at all because it does not provide any additional information .
the figure 5d represents the polar variations of @xmath20 and @xmath8 for showers coming from the west direction .
it is noticed that the enhancements of @xmath20 and @xmath8 occur around @xmath149 and @xmath150 when @xmath151 respectively . for @xmath152 ,
such enhancements are found around @xmath153 and @xmath154 .
these results would therefore validate our selection of polar angle ranges which have been worked out in tables i and ii . most of the figures cited above consider fe initiated showers only , because of having little higher nuon contents ( a generic feature ) compared to p and other nuclei which may receive noticeable influence from the gf . in the following ,
few relatively new eas parameters related to the polar asymmetry of charged muons due to gf will be defined , and their sensitivity to the primary mass will be analyzed .
these mass sensitive parameters are namely the tmbs , mtmbs and the eccentricity ( @xmath155 ) of the elliptic lateral distribution of charged muons .
to quantify the influence of gf as well as to identify some typical signatures of the nature of shower initiating primaries , we have estimated for each shower the coordinates of barycenters of @xmath20 and @xmath8 particles in the shower plane and thereby estimated tmbs , which actually measures the separation length between the barycenter positions of @xmath20 and @xmath8 . for this purpose , we have introduced an operation that involves a rotation either clockwise or anti - clockwise sense of a hypothetical interior quadrant sector ( iqs ) for counting the @xmath20 and @xmath8 particles : the iqs is a region in the interior of a circle enclosed by a pair of arcs on opposite sides and a pair of diagonally aligned diameters on the other two sides making a central angle of @xmath141 .
however , the implemented iqs - s ( special ) in the work is designed by two isolated sectors from the iqs as defined above within the core distance range @xmath140 m along the diameters in opposite sides from the eas core ( a sketch is given - figure 6 ) . and @xmath8 particles by rotating iqs or iqs - s in anti - clockwise sense from @xmath156 to @xmath157.,scaledwidth=45.0% ]
the variation of the tmbs with iqs - s rotation for p and fe initiated eass falling in the @xmath7-range @xmath158 and arriving from north and west directions are studied using simulated data . in figures 7a and 7b ,
a comparison of such variations are shown for showers initiated by p and fe primaries arriving from north and west directions at @xmath159 and @xmath160 pev respectively .
it is expected that @xmath20 particles experience gf greatly around polar angles @xmath161 while @xmath8 around @xmath162 ( peaks in figures 5a and 5b ) when @xmath151 .
the tmbs accordingly takes higher values corresponding to the orientation of the iqs - s through @xmath163 in the shower plane which is reflected in the figure 7a . on the other hand , the gf effects on the @xmath9-component for showers coming from the west direction as the variation of the tmbs is exhibited in the figure 7b . in this case
the parameter tmbs increases from the polar angle @xmath156 to @xmath164 and then decreases till around @xmath165 , and again rises to a value that was obtained at @xmath166 .
it is clear that the tmbs is greater for fe compared to p initiated showers in all the cases . in figures 8a - 8d , we have repeated the same study independently for p and fe primaries for @xmath151 and @xmath167 where various curves in each of the figure 8 correspond different @xmath168 of incidence of showers .
as the @xmath7 increases , @xmath9-s travel increased paths through the atmosphere receiving the gf effect for longer duration that induces more deflections to @xmath9-s and hence yield higher values for tmbs resulting from their greater deflections under the gf .
the polar distribution of @xmath9 exhibits more and more asymmetry with increasing @xmath7 and the feature is revealed from the gradual increase of tmbs in the neighborhood of polar angles @xmath161 and @xmath164 corresponding to @xmath64 with @xmath156 and @xmath167 respectively irrespective of primaries .
@xmath7 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath173 + & & & & & + proton & @xmath174 & @xmath175 & @xmath176 & @xmath177 & @xmath178 + & & & & & + iron & @xmath179 & @xmath180 & @xmath181 & @xmath182 & @xmath183 + & & & & & + @xmath7 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath173 + & & & & & + proton & @xmath184 & @xmath185 & @xmath186 & @xmath187 & @xmath188 + & & & & & + iron & @xmath189 & @xmath190 & @xmath191 & @xmath192 & @xmath193 + & & & & & + and @xmath12 with increasing @xmath7 for @xmath194 . at @xmath195
the overall ellipse gets 8-shaped pattern.,scaledwidth=45.0% ] and dependence on the high - energy hadronic interaction models
. model dependence is exhibited through comparison of figures c and d. the lines are only a guide for the eye.,title="fig:",scaledwidth=45.0% ] and dependence on the high - energy hadronic interaction models .
model dependence is exhibited through comparison of figures c and d. the lines are only a guide for the eye.,title="fig:",scaledwidth=45.0% ] and dependence on the high - energy hadronic interaction models .
model dependence is exhibited through comparison of figures c and d. the lines are only a guide for the eye.,title="fig:",scaledwidth=45.0% ] and dependence on the high - energy hadronic interaction models .
model dependence is exhibited through comparison of figures c and d. the lines are only a guide for the eye.,title="fig:",scaledwidth=45.0% ] and fig .
b with @xmath196.,title="fig:",scaledwidth=45.0% ] and fig .
b with @xmath196.,title="fig:",scaledwidth=45.0% ] to assign a single tmbs to each shower instead of several values corresponding to various positions of the iqs - s over the polar angle between @xmath156 to @xmath157 , we have introduced a mean maximum tmbs i.e. mtmbs , which is the average of four tmbs values for four different polar positions of the iqs - s over the polar angle between @xmath197 to @xmath198 ( between @xmath199 to @xmath200 is the region for iqs - s in the opposite side ) when @xmath151 . for @xmath152 ,
the mtmbs is estimated by taking average of tmbs from the polar angle range between @xmath201 to @xmath197 ( opposite side : between @xmath202 to @xmath199 ) .
the parameter mtmbs appears as a suitable mass sensitive parameter for the measurement of cr chemical composition and an experimental approach to estimate it in a ground - based eas experiment will be discussed in the sec .
the variation of mtmbs against @xmath7 is shown in figures 10a and 10b for showers induced by p and fe , and coming from the north and west directions respectively at @xmath203 ev .
mtmbs values are higher for fe compared to p primaries in all cases irrespective of @xmath7 , @xmath64 , @xmath204 and models .
according to the figure 10b , mtmbs values for p and fe showers are approaching closer to one another as the @xmath7 increases . in figures
10c and 10d , we have presented the effect of high - energy hadronic models on the parameter under consideration .
results obtained from the epos look relatively better in comparison with the qgsjet as far as the composition study of crs is concerned .
such a behavior of the epos validates the generation of little higher @xmath9-s by the model .
errors for mtmbs are evaluated by averaging standard deviations in tmbs values within the polar angle range between @xmath205 and @xmath206 when @xmath151 ( i.e. @xmath207 ) . for @xmath152
, averaging is done using values from the range @xmath208 .
the primary cr composition study has also been carried out by the eccentricity parameter ( @xmath155 ) of the polar asymmetric distribution of the @xmath9 component under the gf .
we have defined the @xmath155 parameter as @xmath209 , where @xmath210 and @xmath211 denote the core distances of a pair of points possessing same @xmath9 densities in an overall iso - density contour ( elliptic ) made by @xmath20 and @xmath8 .
one of the densities is estimated along an arbitrary axis which is being formed by the tmbs itself and other , being @xmath212 to the tmbs axis passing through the eas core in the shower plane .
actually , @xmath20 and @xmath8 in inclined showers under the gf are distributed asymmetrically in the shower plane ( @xmath20 and @xmath8 distributions also develop elliptic contours covering smaller regions but with some overlapping region ) .
combination of these smaller contours of @xmath20 and @xmath8 will appear as an overall pattern which is equivalent to an iso - density ellipse .
such a pattern of the combined elliptic contour resulting from a pair of smaller iso - density contours is presented by the sketch in the figure 9 .
the @xmath155 factors are evaluated by the above procedure at five @xmath213 ranging from @xmath214 to @xmath215 for p and fe showers using the qgsjet and epos models .
we have taken a particular @xmath151 and @xmath216 ev for the exercise and results are shown in the table iii .
it reveals from the table that the lateral @xmath9 distribution in the shower plane gets more and more stretched along the axis constituted by the tmbs as @xmath7 increases .
the eccentricity factor correspondingly increases as one moves from lighter to heavier primaries .
it has been noticed here that the parameter @xmath155 could nt be estimated by the above procedure for showers with @xmath217 .
according to the figure 9 for @xmath195 , the stretching of the @xmath9 lateral distribution along the tmbs axis is so high that iso - density concentric ellipse formation could nt be possible ( from iso - density ellipse to stretched out geometric @xmath218-shaped density pattern ) .
hence the parameter can not be useful for very inclined showers to measure the chemical composition of crs .
the parameter @xmath155 gets higher values consistently for all @xmath7 in case of epos model compared to the qgsjet . in epos
, there exists an additional particle production source which induces a larger number of muons from simulations with the model [ 43 ] .
hence , an increased number of muons come under the influence of the gf that would result into higher @xmath155 . from the experimental standpoint ,
the correlation of mtmbs with @xmath219 has been considered as a basis for extracting information on the nature of primary crs .
the problem of determining the mass composition of crs is a complicated one because many parameters in tandem are required . from the exploration of the geomagnetic spectroscopy
it is found that parameters like tmbs , mtmbs and @xmath155 extracted from the asymmetric @xmath20 and @xmath8 distributions are found sensitive to the cr mass composition .
practicability of the present approach to determine the cr chemical composition requires a correlation between the observables mtmbs with @xmath219 on an average shower with some @xmath213 corresponding to either @xmath151 or @xmath167 .
it might be noted that the present approach on the measurement of primary mass composition with the observable tmbs or mtmbs , a new eas technique , that is proposed in this work , and therefore no published data / results are available so far for comparison . in figures 11a and 11b ,
the variation for the mean mtmbs with @xmath220 for p and fe primaries are shown at @xmath221 and @xmath222 respectively for @xmath151 , corresponding to three primary energy ranges @xmath223 pev .
the figure 11a looks better for composition study compare to figure 11b so far mtmbs versus @xmath220 curve is concerned . relatively less sensitivity on primary mass at higher @xmath224 might be due to less number of muons surviving against attenuation .
it has been found that the influence of the earth s magnetic field plays a dominant role for detailed features of @xmath9-s lateral distribution on the shower plane at least for highly inclined showers .
the proposed method employs the polar asymmetry in @xmath9-s distribution to determine the primary cr chemical composition at a given observation level for three small energy intervals restricting showers from the north and west directions only .
this new eas method was validated using simulated data samples corresponding to pure proton and iron primaries .
although not done here , there is a scope to extend the present analysis in a much wider primary energy range for the measurement of chemical composition of crs with numerous features ( e.g. any indication of possible change in the chemical composition of pcrs with energy ) .
this can be carried out for other geomagnetic positions as well to correlate some geomagnetic properties with high - energy astro - particle phenomena . in order to present important results at fixed shower size and muon size under different geomagnetic intensities ,
a high - speed computation with adequate storage capacity has to be ensured .
our analysis concerning the effects of the gf on @xmath20 and @xmath8 particles of inclined eass reveals several interesting features such as polar asymmetries , sectoral @xmath9 abundances , elliptic footprints of @xmath9 distributions , and amplitude of fluctuations between proton and iron induced showers . taking a small sample of data such effects
are also found to persist with comparable magnitude ( not shown here ) if we replace the urqmd code in the simulation by the fluka [ 44 ] in the treatment of low energy hadron collisions .
the present method might help to design eas arrays with induction of @xmath9 detectors to derive more information about the primary crs employing the trio * eas direction * ( @xmath7 and @xmath64 ) , * eas core * ( @xmath225 , @xmath226 ) and * geomagnetic field*. for implementing the present technique in a possible experimental set up , resolution of the eas array for selected @xmath7 ranges corresponding to some fixed or @xmath64-range at a given geomagnetic position should be a prerequisite .
this is usually realized by performing mc simulations of eass , convoluted with the eas array structure and detector responses . we know
that at highly inclined showers ( present work ) the angular resolution of an array generally deteriorates with increasing @xmath7
. a reasonable array resolution should be provided for implementing the present method from the employment of mc showers .
a certain triggering level has to be set by some central triggering detectors in the array using mc showers . for the present purpose , the eas experimental set up firstly should have a close - packed detector array ( @xmath227 m x @xmath227 m like kascade ) for determining global parameters of an eas ( here , @xmath7 , @xmath64 and @xmath225 , @xmath226 etc . ) .
it should be mentioned that for a closely packed array , estimation for the eas core location by simple weight averaging of particle ( @xmath5 ) densities from discrete detector locations does not differ appreciably from the core obtained from the shower reconstruction . secondly , estimation of the main parameter in the work i.e. the mtmbs , one has to incorporate a pair of @xmath9 detectors with sizes at least covering the present regions around @xmath167 and @xmath228 in a possible geometrical layout shown in the figure 12 for @xmath151 .
two uncovered sectors in the figure 12 within a pair of dotted lines for each , in opposite sides within the polar regions of @xmath229 and @xmath230 ( possible locations for @xmath9 detectors for eass coming from the west ) .
the sizes for @xmath9 detectors proposed here are having the least size from the point of view of their production cost . as a result ,
the measurement of the mtmbs may involve slightly higher uncertainty . as we know that a @xmath9 detector consists of number of proportional counters having some energy threshold ( here , @xmath231 gev ) . in the figure 12
, the green patch shows @xmath9 detectors layer which is under a shielding absorber containing several layers ( concrete , soils , lead or iron sheets etc . shown by different color patches ) to set the threshold energy at @xmath135 gev for @xmath9-s .
it would be an interesting task to apply such a method using observed eas data if the kascade or grapes - iii experiments runs with the facility of concurrent @xmath9 measurements ( @xmath9 tracking detectors ( mtd ) [ 45 ] ) after restructuring the present @xmath9 detecting systems .
a new experiment consisting of underground installations of muon detectors and an eas array above the large underground would be appropriate for the study .
there are some recent proposals of studying positive and negative @xmath9-s separately in individual eas event .
in fact , a few ongoing experiments , such as the willi detector [ 26 ] or the okayama university , japan eas installation , have the capability to extract charge information of high energy @xmath9-s but these experiments do not have large @xmath9 detection area , which is needed to extract information about the nature of primaries from the study of geomagnetic influence on eas @xmath9-s . if in future , these experiments are extended in order to cover larger detection area , or new installation of large @xmath9 detection area with capability of charge identification are come up , the present proposal can be exploited to extract the nature of primary particles .
the main sources of error are the fluctuations in @xmath9 numbers from shower to shower , non - uniform behavior of the local geomagnetic field , and the uncertainties in arc , and radial distance estimations by @xmath232 values of each @xmath9 within a polar region of a selected isq - s . from the cost - productive viewpoint ,
a slim isq - s ( figure 6 or 12 ) with truncated radial bin ( @xmath140 m ) covering a pair of @xmath9 detectors in opposite sides has been proposed in a real experiment .
but , in such a situation a slightly higher uncertainty in the parameter estimation may result .
rkd thanks the serb , department of science and technology ( govt . of india ) for financial support under the grant no . emr/2015/001390 .
v. f. hess , _ phys .
z. _ , * 13:1084 - 1091 * , ( 1912 ) .
w. d. apel _
_ , _ astropart .
* 24 467 - 483 * ( 2006 ) .
the pierre auger collaboration , _ nucl .
meth a _ * 798 172 - 213 * ( 2015 ) .
g. aielli _
_ , _ nucl .
instr . meth .
a _ * 661 s50 * ( 2012 ) .
s. k. gupta _
_ , _ nucl .
instr . meth . a _ * 540 311 * ( 2005 ) .
the alice collaboration , _ j. cos .
_ * 01 032 * ( 2016 )
. t. pierog _
_ , _ phys . rev .
c _ * 92 ( 3 ) 034906 * ( 2015 ) .
s. ostapchenko , _ epj web conf .
_ _ 52 02001 _ ( 2013 ) .
d. heck _ et al .
_ , _ the corsika air shower simulation program forchungszantrum karlsruhe report _ * fzk6019 * ( 1998 ) .
p. k. f. grieder , _ extensive air showers : a tutorial , reference manual and data book _ , * vol 1 * isbn-*978 - 3 - 540 - 76940 - 8 * ( 2010 ) .
montanus , _ astropart .
phys . _ * 59 411 * ( 2014 ) .
k. greisen , _ progress in cosmic ray physics vol _ * iii * ( amsterdam : north holland ) ( 1956 ) .
g. cocconi , _ phys . rev . _
* 93 , 646 * ( 1954 ) .
h. r. allan , _ acta phys . hung .
* 29 , 699 * ( 1970 ) .
s. a. colgate , _ j. geophys .
* 72 , 4869 * ( 1967 ) .
n. a. porter , _
nuovo cimento lett .
_ * 8 , 481 * ( 1973 ) .
r. browning and k. e. turver , _
nuovo cimento a _ * 38 , 223 * ( 1977 ) .
o. c. allkofer _
_ , _ j. geophys .
* 90 3537 * ( 1985 ) .
c. c. g. bowden _ et al .
_ , _ j. phys .
g : nucl . part .
phys . _ * 18 , l55 * ( 1992 )
. m. j. lang _
_ , _ j. phys .
g : nucl . part .
phys . _ * 20 , 1841 * ( 1994 ) .
p. m. chadwick _
_ , _ j. phys .
g : nucl . part .
* 25 , 1223 * ( 1999 ) .
a. a. ivanov _
_ , _ jetp letts .
_ , * 69 , 288 * ( 1999 ) ; _ proc .
cosmic ray conf . _ * 1 , 4031 * ( 1999 ) .
m. ave m , r. a. vasquez and e. zas , _ astropart .
* 14 91 * ( 2000 ) .
capdevielle , c. le gall , kh.n .
sanosyan , _ astropart .
* 13 , 259 * ( 2000 ) .
a. cillis and s. j. sciutto , _ j. phys .
g : nucl . part .
* 26 , 309 * ( 2000 ) .
_ , _ j. phys .
g : nucl . part .
* 35 , 085203 * ( 2008 ) .
_ , _ nucl .
instru . meth .
a _ , * 638 , 147 * ( 2011 ) .
the pierre auger collaboration . , _
jcap_*11 , 022 * ( 2011 ) .
j. a. muniz _
_ , _ astropart .
phys . _ * 59 , 29 * ( 2014 ) .
j. abraham _
et al . _ ,
( _ the pierre auger collaboration _ ) , _ phys .
* 104 , 091101 * ( 2010 ) .
r. engel , d. heck , and t. pierog , _ annual rev .
nucl . part .
sc . _ * 61 : 467 - 489 * ( 2011 ) .
j. n. capdevielle , r. k. dey and a. bhadra , proc .
cosmic ray conf . , isbn-978 - 85 - 89064 - 29 - 3 ( 2013 ) ; s. dam , r. k. dey and a. bhadra _ information systems design and intelligent applications : int .
conference _ * 339 * j. k. mandal _ et al .
_ 1 ( 2015 ) .
n. n. kalmykov , s. s. ostapchenko and a. i. pavlov , _ nucl .
b _ * 52 17 * ( 1997 ) .
k. werner _
_ , _ phys .
c _ * 74 044902 * ( 2006 ) .
m. bleicher _
_ , _ j. phys .
g : nucl . part .
* 25 , 1859 * ( 1999 ) .
w. r. nelson _
_ , _ the egs4 code system report slac265 _ ( stanford linear accelerator center , stanford , ca ) ( 1985 ) . national aeronautics and space administration ( * nasa * ) us standard atmosphere technical report nasa - tm - x-74335 .
m. kobal , _ astropart .
phys . _ * 15 259 * ( 2001 ) .
t. antoni _ et al .
_ , _ astropart .
_ * 19 703 - 714 * ( 2003 ) .
j. m. c. montanus , _ expermental astronomy _ * 41(1 ) 159 - 184 * ( 2016 ) .
arteaga - vel@xmath233zquez _ et al .
_ _ nuclear physics b ( proc .
suppl . ) _ * 196 183186 * ( 2009 ) .
et al . _ , kascade - grande coll . , _ proc . of 30th icrc mexico _ * 4 203 - 206 * ( 2008 ) . t. pierog and k. werner , _ phys .
lett . _ * 101:171101 * ( 2008 ) .
a. fass@xmath234 _ et al .
monte carlo 2000 conf .
( lisbon ) ( berlin : springer ) p * 955 * ( 2001 ) .
w. d. apel _
_ , _ astropart .
phys . _ * 65 55 - 63 * ( 2015 ) . | the effect of the geomagnetic lorentz force on the muon component of extensive air shower ( eas ) has been studied in a monte carlo generated simulated data sample .
this geomagnetic field affects the paths of muons in an eas , causing a local contrast or polar asymmetry in the abundance of positive and negative muons about the shower axis .
the asymmetry can be approximately expressed as a function of transverse separation between the positive and negative muons barycentric positions in the eas through opposite quadrants across the shower core in the shower front plane . in the present study
, it is found that the transverse muon barycenter separation and its maximum value obtained from the polar variation of the parameter are higher for iron primaries than protons for highly inclined showers .
hence , in principle , these parameters can be exploited to the measurement of primary cosmic - ray chemical composition .
possibility of practical realization of the proposed method in a real experiment is briefly discussed . |
the b - v colour difference between young and evolved bars is 0.4 mag , which can be translated to an age difference of 10 gyr .
this means that bars can be robust structures , in agreement with recent @xmath0-body simulations and observations of barred galaxies at higher redshifts .
the young bars in our sample have an average length of [email protected] kpc , while the evolved bars have an average length of [email protected] kpc , consistent with recent theoretical expectations that bars grow longer while aging .
young bars are preferentially found in late - type spirals , indicating that bar recurrence may be more frequent in gas - rich , disk - dominated galaxies .
we also found that agn are preferentially hosted by galaxies with young bars , suggesting that the fueling of agn by bars happens in short timescales and that a clearer bar - agn connection would be found in a sample of galaxies with young bars .
we have also found that bar colours might be used as a proxy for bar ages .
enlarging the sample of bars with measured ages is paramount to calibrate this relation , confirm these results , compare in more detail observations and models , and better understand secular evolution .
see ( * ? ? ?
* ; * ? ? ?
* gadotti & de souza ( 2005 , 2006 ) ) for further details . | in an effort to obtain further observational evidences for secular evolution processes in galaxies , as well as observational constraints to current theoretical models of secular evolution , we have used bvri and ks images of a sample of 18 barred galaxies to measure the lengths and colours of bars , create colour maps and estimate global colour gradients .
in addition , applying a method we developed in a previous article , we could distinguish for 7 galaxies in our sample those whose bars have been recently formed from the ones with already evolved bars .
we estimated an average difference in the optical colours between young and evolved bars that may be translated to an age difference of the order of 10 gyr , meaning that bars may be long standing structures .
moreover , our results show that , on average , evolved bars are longer than young bars .
this seems to indicate that , during its evolution , a bar grows longer by capturing stars from the disk , in agreement with recent numerical and analytical results . |
theories of structure formation were strongly constrained by the statistical measurements of clustering in some of the early galaxy redshift surveys .
surveys of infrared galaxies , in particular , were able to rule out the then standard cold dark matter model @xcite .
present day redshift surveys such as the 2dfgrs @xcite , sdss @xcite and , in the far - infrared , the point source catalog redshift survey , psc - z @xcite are now able to provide definitive measurements of the galaxy clustering in the local universe . despite this success , we have always known that galaxies are biased tracers of the matter distribution and yet we have a poor observational or theoretical understanding of this bias , although it is assumed to be related to the process of galaxy formation and evolution . to understand bias and , by inference , galaxy formation
, we need to better understand the clustering of different galaxy types and the evolution of this clustering with redshift .
in this paper we attempt to provide an estimate of the clustering of infrared galaxies a factor of ten deeper ( in redshift ) than those seen in the iras surveys . to do this we provide the first estimate of clustering from any of the extragalactic iso surveys .
this is thus the first estimate of clustering from galaxies selected at 15@xmath1 m .
we have used part of the elais survey ( oliver et al .
2000 ) as this probes the largest volume of any of the iso surveys .
we measure the projected clustering by calculating the angular correlation function , we then discuss the constraints this places on the three dimensional clustering using limber s equation .
the european large - area iso survey ( elais , oliver et al .
2000 ) was the largest open time programme on iso .
this project surveyed 12 square degrees , larger than all but the serendipitous surveys , making it ideal for clustering studies .
the main survey bands were at 6.7 , 15 , 90 and 170 @xmath1 m .
of these bands the 15@xmath1 m catalogues contain the greatest density of galaxies ( see e.g. * ? ? ?
* ; * ? ? ?
* ) , and provide the best statistics for clustering . the final analysis of the 15@xmath1 m data using the lari method for one of the elais fields ( s1 ) has recently been completed @xcite and this is the sample that we use in this analysis .
the s1 field is located at @xmath7 @xmath8 , @xmath9 @xmath10 , covering an area of @xmath11 square degrees . the 15 survey is made from 9 raster observations , each one of @xmath12 .
the central raster s1_5 has been observed three times . using the lari method we have obtained a sample of 462 sources to @xmath13 in the flux range 0.45 - 150mjy @xcite .
besides the galaxy catalogue itself , the selection function is the most important ingredient in the calculation of clustering statistics .
errors in the selection function will invalidate the answer , whereas errors in the weighting scheme will usually make the answer more noisy .
a selection function is required for each source list that is being investigated .
the selection function , @xmath14 , is defined as the expected number density of sources as a function of @xmath15 ( which might be two or three dimensional ) , in absence of clustering ; i.e. , the expected number of galaxies @xmath16 in a volume @xmath17 is @xmath18 . with this definition , @xmath19 .
the selection function is used to simulate a catalogue with no clustering . to be selected from the elais s1 catalogue sources had to exceed a signal - to - noise threshold , @xmath20 . the signal to noise of a detected source @xmath21 ,
is @xmath22 where @xmath23 is the signal of the source and @xmath24 is the noise at the position of the source .
had this source been in a different part of the survey , @xmath15 it would have had a different signal - to - noise .
we can define then a mask , @xmath25 , which represents the detectability of each object as a function of position as follows @xmath26 where @xmath27 .
the un - normalised selection function can then be written as @xmath28 which can be normalised @xmath29 in the full elais s1 region there are 9 independent noise maps @xmath30 , corresponding to 9 independent sub - catalogues .
note that the central noise map is less noisy and the corresponding sub - catalogue deeper , because the iso data were already combined @xcite .
we constructed a selection function as follows : for each source in the sub - catalogue we calculate the hypothetical signal - to - noise ratio ( defined as the peak flux over the rms value ) at each point in the raster . where these exceed the extraction signal - to - noise threshold @xmath20 ( equation [ eq : mask ] ) , the value of the selection function at that position is incremented ( equation [ eq : sfunc ] ) .
the 9 individual selection functions are then combined into a single one .
figure [ fig : mask ] shows the final image . in the overlap region
only one selection function was used and the final catalogue excludes sources in that region that arose from the other sub - catalogues .
sources with stellar counterparts have also been removed ( see * ? ? ?
* ) from the catalogue and excluded from the calculation of the selection function .
we end up with a catalogue of 329 sources .
the selection function so obtained is then used to generate the random catalogues with no clustering , essential to properly calculate the two point correlation function .
correlation functions are widely used to study the distribution of sources in surveys and to derive large scale properties of galaxies .
the two - point spatial correlation function is defined so that @xmath31 dv_1 dv_2\ ] ] is the joint probability of finding a source in a volume element @xmath32 and another source in a volume element @xmath33 .
the function @xmath34 is the excess probability of finding an object compared to a random distribution of objects .
similarly , one can define the two point angular correlation function so that @xmath35 d\omega_1 d\omega_2\ ] ] is the joint probability of finding a source in a solid angle element @xmath36 and another source in a solid angle element @xmath37 .
these two statistics are related by limber s equation ( peebles 1980 ) . in order to calculate the angular correlation function of mid - ir sources we use the @xcite estimator @xcite @xmath38 } - 2 { \rm [ dr ] } + { \rm [ rr]}}{\rm [ rr]}\ ] ] where [ dd ] , [ rr ] and [ dr ] represents the normalized number of galaxy - galaxy , random - random and galaxy - random pairs with angular separation in @xmath39 .
errors in the calculation of the angular correlation function are dominated by poisson noise .
the error in each bin can be estimated using the following expresion @xcite : @xmath40 where , in this case , @xmath41 is the total number of galaxy - galaxy pairs ( not normalized ) .
errors calculated using this equation are comparable to the errors obtained from a bootstrap resampling technique @xcite .
a second source of errors comes from the to finite size of the sample . in order to correct for this effect we use the random sample to calculate the integral constraint as ( e.g. infante et al . 1994 ) @xmath42 and divide the calculated correlation function by this factor , @xmath43 .
figure [ fig : wtheta ] ( top ) shows the obtained angular correlation function , calculated using 200 realizations with 2000 random sources each , in intervals of @xmath44 degrees .
random catalogues have been built using the selection function calculated in previous section .
data points have been fitted by an exponential law of the form @xmath45 resulting in @xmath46 and @xmath5 ( where @xmath47 is measured in degrees ) .
the mean number of objects in each field is @xmath48 ( excluding the central raster s1_5 , with 71 sources ) , with a standard deviation of 6 objects .
since s1_5 field reaches deeper flux limits , it could in principle be subject to clustering variations that would affect the whole clustering estimation .
we repeat the above calculation removing those sources with fluxes fainter than the flux limit excluding s1_5 : a total of 27 sources with fluxes lower than 0.7 mjy are removed . by calculating again the selection function and the angular correlation function
we obtain then an estimate of the clustering from sources detectable over most of the s1 field . in this case
the integral constraint is slightly larger than the previously obtained , @xmath49 .
the angular correlation function is shown in figure [ fig : wtheta ] ( bottom ) and is fitted by the same exponential law previously calculated .
although the correlation function is now slightly larger at scales @xmath50deg , the errors are also larger and the overall correlation function is well fitted by the previous function . with @xmath51 and @xmath52 .
in case of small angles , both @xmath53 and @xmath54 can be approximated by power law shapes , and the spatial correlation function can be written as ( e.g. , * ? ? ?
* ) @xmath55 where @xmath56 is the comoving correlation length at @xmath57 and @xmath58 the comoving distance .
the parameter @xmath59 is the clustering evolution index and is interpreted as follows .
a value @xmath60 corresponds to stable clustering in physical coordinates , i.e. , galaxy clusters remain unchanged and clustering changes due to the expansion of the universe , while @xmath61 corresponds to clustering fixed in comoving coordinates , i.e. , clustering does not change with time and the galaxy clusters expand with the universe . if @xmath62 is parametrized as @xmath63 , then limber s equation becomes ( e.g. , * ? ? ? * ) @xmath64 where @xmath65 is the angular diameter distance , @xmath66 is the scale factor multiplied by the element of comoving distance @xmath67^{-1}\ ] ] and @xmath68}{\gamma(\gamma/2)}\ ] ] the only unknown quantity in equation [ eq : limber_small ] is the redshift distribution of the sources . a distribution given by @xmath69\ ] ] where @xmath70 is the median redshift of the survey , provides good fits to the distribution of optical galaxies ( * ? ? ? * and references therein ) and as been widely used to invert limber s equation .
however , the redshift distribution of mid - ir sources is more extended and the equation above is no longer valid .
figure [ fig : n_z ] ( top ) shows the redshift distribution of the elais sources , obtained from optical spectroscopy ( la franca et al , 2004 , in prep ; perez - fournon et al , 2004 , in prep ) and photometric redshifts @xcite . instead of the optical @xmath71
we use @xmath72 \label{eq : dndz}\ ] ] which fits reasonably well the distribution of elais sources .
we can now integrate equation [ eq : limber_small ] in order to obtain @xmath56 .
assuming @xmath73 and @xmath60 and using a median redshift of @xmath74 in equation [ eq : dndz ] , we obtain @xmath75mpc with a 95% confidence level in the range @xmath76mpc to @xmath77mpc .
an alternative redshift distribution of the elais sources has been presented by @xcite , computed from the luminosity function fit of galaxies on elais s1 and s2 ( see figure [ fig : n_z ] bottom ) . when using this redshift distribution as input to equation [ eq : limber_small ] we obtain similar results of @xmath78mpc with a 95% confidence level in the range @xmath79mpc to @xmath80mpc .
we use a sample of 329 15 galaxies detected in the elais s1 survey covering a region of 4 square degrees of sky to determine the angular correlation function of the galaxies .
we measure @xmath62 up to scales of 1 degree , corresponding to @xmath81mpc at the mean redshift of @xmath82 . the resulting correlation function
is well fitted by a power law @xmath83 with @xmath84 and @xmath52 . assuming a redshift distribution we invert the angular correlation using limber s equation in order to determine the correlation length @xmath56 .
we find a value of @xmath75mpc with a 95% confidence level in the range @xmath76mpc to @xmath77mpc .
.summary of correlation lengths values obtained from different surveys . [ cols="<,^,^,^,^",options="header " , ] references : 1 @xcite ; 2 @xcite ( assuming a einstein - de sitter cosmology ) ; 3 @xcite ; 4 @xcite ; 5 @xcite table [ tab : r0 ] lists the correlation lengths obtained from several optical and infrared surveys .
while optical surveys show a significant clustering with @xmath85mpc , iras survey shows a lower value .
the data described in this paper is consistent with this trend : that mid infrared selected sources show a smaller correlation length .
this is expected because optical surveys favour elliptical galaxies which are more strongly clustered than spiral galaxies , while infrared surveys preferentially select spirals and star - forming galaxies .
we note that the elais selected galaxies appear to have a marginally steeper two point correlation function than the optical and the iras surveys . by fixing @xmath86 to the lower value allowed by the fit @xmath87 we obtain @xmath88 , bringing the clustering amplitude closer to that seen by iras .
it is interesting to note that the optical correlation function at @xmath89 @xcite is consistent with that at @xmath90 @xcite , i.e. there is no apparent evolution in the correlation function over this albeit small redshift range . likewise
the infrared galaxy correlation function estimate from this work at @xmath91 is consistent with the infrared galaxy correlation function estimated from iras galaxies at @xmath57 @xcite .
this apparent lack of evolution may be because evolutionary effects are small or that evolution in mass clustering is compensated by evolution in galaxy bias .
it will be interesting to see if this apparent non - evolution in clustering of different galaxy population mixes is still found in deeper surveys ( e.g. swire , lonsdale et al .
2003 ) as this might imply a striking conspiracy in the evolution of the bias in different galaxy types . by performing a systematic analysis of all density peaks in the redshift distribution of field galaxies
, @xcite recently found an excess of iso selected galaxies over the whole range of density peaks .
this suggests than infrared galaxies are more strongly clustered than optical galaxies , in apparent contradiction to our results .
since the iso - cam surveys are deeper than elais , it is plausible that there is an evolution of the clustering over the redshift interval spanned by these two iso surveys .
infrared galaxies would then become more clustered towards higher redshift , while the clustering of optical galaxies changes little .
this would agree qualitatively with hierarchical pictures of structure formation .
such models predict that star formation and galaxy formation is driven by merger rate that would be a function of environment .
star formation would thus have occurred first and vigorously in the denser ( more clustered ) regions of the universe , taking more time to initiate in lower density ( less clustered ) regions ( see also * ? ? ?
* ) . the strong evolution in luminosity function of infrared galaxies ( e.g. pozzi et al . 2004
) might then be coupled with an evolution in their clustering .
optically selected galaxies sample regions where past star formation activity was high as well as those where current activity is high and are thus less sensitive to these effects . finally this apparent contradiction
may simple be that the deep iso - cam surveys do not sample a sufficiently large volume to be representative of the rest of the universe .
eduardo gonzalez - solares was supported by pparc grant ppa / g / s/2000/00508 and ec marie curie fellowship mcfi-2001 - 01809 .
this paper is based on observations with _
iso _ , an esa project , with instruments funded by esa member states ( especially the pi countries : france , germany , the netherlands and the united kingdom ) and with participation of isas and nasa .
some of this work was supported by eec training mobility research network `` probing the origin of the extra - galactic background light '' ( poe ) hprn - ct-2000 - 00138 .
the author also wants to thank d. elbaz for useful comments on the discussion of the results presented . | we present an analysis of the two - point angular correlation function of the elais s1 survey .
the survey covers 4 deg@xmath0 and contains 462 sources detected at 15@xmath1 m to a 5@xmath2 flux limit of 0.45 mjy . using the 329 extragalactic sources not repeated in different observations , we detect a significant clustering signal ;
the resulting angular correlation function can be fitted by a exponential law @xmath3 with @xmath4 and @xmath5 . assuming a redshift distribution of the objects , we invert limber s equation and deduce a spatial correlation length @xmath6mpc .
this is smaller than that obtained from optical surveys but it is in agreement with results from iras .
this extends to higher redshift the observational evidence that infrared selected surveys show smaller correlation lengths ( i.e. reduced clustering amplitudes ) than optical surveys .
[ firstpage ] galaxies : clusters : general galaxies : evolution infrared : galaxies cosmology : observations large - scale structure of universe |
in this paper we study the doubly nonlinear ( dnl ) reaction - diffusion problem posed in the whole euclidean space @xmath2 we want to describe the asymptotic behaviour of the solution @xmath3 for large times and for a specific range of the parameters @xmath4 and @xmath5 .
we recall that the @xmath6-laplacian is a nonlinear operator defined for all @xmath7 by the formula @xmath8 and we consider the more general diffusion term @xmath9 called `` doubly nonlinear''operator . here
, @xmath10 is the spatial gradient while @xmath11 is the spatial divergence .
the doubly nonlinear operator ( which can be though as the composition of the @xmath12-th power and the @xmath6-laplacian ) is much used in the elliptic and parabolic literature ( see @xcite and their references ) and allows to recover the porous medium operator choosing @xmath13 or the @xmath6-laplacian operator choosing @xmath14 .
of course , choosing @xmath15 and @xmath13 we obtain the classical laplacian . before proceeding ,
let us fix some important restrictions and notations .
we define the constants @xmath16 and we make the assumption : @xmath17 that we call `` fast diffusion assumption''(cfr . with @xcite ) .
note that the shape of the region depends on the dimension @xmath1 .
two examples are reported in figure [ fig : simulfastcaserange ] ( note that the region in the case @xmath18 is slightly different respect to the case @xmath19 and we have not displayed it ) .
we introduce the constant @xmath20 since its positivity simplifies the reading of the paper and allows us to make the computations simpler to follow .
-plane.,title="fig : " ] -plane.,title="fig : " ] the case @xmath21 , i.e. @xmath22 , has been recently studied in @xcite . in this
setting , the authors have showed that the equation in possesses a special class of travelling waves which describe the asymptotic behaviour for large times of more general solutions ( see subsection [ sectionpreviousresultsfast ] for a summary of the results of the case @xmath21 ) .
our main goal is to prove that the case @xmath23 presents significative departs in the asymptotic properties of the solutions of problem . in particular
, we will see that general solutions do not move with constant speed but with exponential spacial propagation for large times .
this fact is the most interesting deviance respect to the classical theory in which tws play an important role in the study of the asymptotic behaviour .
the function @xmath24 is a reaction term modeled on the famous references by fisher @xcite , and kolmogorov - petrovski - piscounoff @xcite in their seminal works on the existence of traveling wave propagation .
the classical example is the logistic term @xmath25 , @xmath26 .
more generally , we will assume that @xmath27 \to { \mathbb{r}}\text { and } f \in c^1([0,1 ] ) \\
f(0 ) = 0 = f(1 ) \ ; \text { and } \ ; f(u ) > 0 \text { in } ( 0,1 ) \\
f \text { is concave in } [ 0,1 ] \end{cases}\ ] ] see @xcite for a more complete description of the model .
moreover , we will suppose that the initial datum is a lebesgue - measurable function and satisfies @xmath28 note that the previous assumption is pretty much general than the more typical continuous with compact support initial data .
moreover , since @xmath29 , all data satisfying are automatically integrable , @xmath30 .
[ [ main - results - and - organization - of - the - paper . ] ] main results and organization of the paper .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the paper is divided in parts as follows : in section [ sectionpreviousresultsfast ] we present some known theorems about problem . our goal is to give to the reader a quite complete resume on the previous work and related bibliography , to connect it with the new results contained in this paper . in section [ convergencetozerofast ]
we begin the study of the asymptotic behaviour of the solutions of problem -- , with restriction . in particular
, we firstly introduce the critical exponent @xmath31 by giving a formal motivation and , later , we prove the following theorem .
[ convergencetozerofastdiffusion ] fix @xmath1 .
let @xmath4 and @xmath5 such that @xmath29 .
then for all @xmath32 , the solution @xmath3 of problem with initial datum satisfies @xmath33 for all @xmath32 , we call @xmath34 `` exponential outer set '' or , simply , `` outer set '' .
the previous theorem shows that , for large times , the solution @xmath3 converges to zero on the `` outer set '' and represents the first step of our asymptotic study . in section [ sectionexponentialexpansionsuperlevelsets ]
we proceed with the asymptotic analysis , studying the solution of problem with initial datum @xmath35 where @xmath36 and @xmath37 are positive real numbers and @xmath38 .
we show the following crucial proposition .
[ expanpansionofminimallevelsets ] fix @xmath1 .
let @xmath4 and @xmath5 such that @xmath29 and let @xmath39 .
then there exist @xmath40 , @xmath41 and @xmath42 such that the solution @xmath3 of problem with initial datum satisfies @xmath43 this result asserts that for all initial data `` small enough '' and for all @xmath44 , the solution of problem is strictly greater than a fixed positive constant on the `` exponential inner sets '' ( or `` inner sets '' ) @xmath45 for large times .
hence , it proves the non existence of travelling wave solutions ( tws ) since `` profiles '' moving with constant speed of propagation can not describe the asymptotic behaviour of more general solutions ( see section [ sectionpreviousresultsfast ] for the definition of tws ) .
moreover , this property will be really useful for the construction of sub - solutions of general solutions since , as we will see , it is always possible to place an initial datum with the form under a general solution of and applying the maximum principle ( see lemma [ lemmaplacingbarenblattundersolution ] ) . in section [ sectionasymptoticbehaviourfast ]
we analyze the asymptotic behaviour of the solution of problem , in the `` inner sets '' @xmath45 . along with theorem [ theoremboundsforlevelsetsfast ]
the next theorem is the main result of this paper .
[ convergencetoonefastdiffusion ] fix @xmath1 .
let @xmath4 and @xmath5 such that @xmath29 .
then for all @xmath44 , the solution @xmath3 of problem with initial datum satisfies @xmath46 this theorem can be summarized by saying that the function @xmath47 converges to the steady state 1 in the `` inner sets '' for large times . from the point of view of the applications
, we can say that the density of population @xmath3 invades all the available space propagating exponentially for large times .
, for all @xmath44.,title="fig : " ] , for all @xmath44.,title="fig : " ] in section [ sectionboundsforlevelsetsfast ] we consider the classical reaction term @xmath25 .
we find interesting bounds for the level sets of the solution of problem , .
in particular , we prove that the information on the level sets of the general solutions is contained , up to a multiplicative constant , in the set @xmath48 , for large times .
[ theoremboundsforlevelsetsfast ] fix @xmath1 .
let @xmath4 and @xmath5 such that @xmath29 , and take @xmath49 .
then for all @xmath50 , there exists a constant @xmath51 and a time @xmath52 large enough , such that the solution of problem with initial datum and reaction @xmath25 satisfies @xmath53 for all @xmath54 . in particular
, we have @xmath55 an important feature of this result is that for all @xmath50 , the set @xmath56 does not depend on some @xmath57 , while in theorem [ convergencetozerofastdiffusion ] and theorem [ convergencetoonefastdiffusion ] the `` outer sets '' and the `` inner sets '' depend on @xmath32 and @xmath44 , respectively .
moreover , taking a _ spatial logarithmic scale _ we can write the estimate @xmath58 for @xmath59 large enough .
actually , this result was not known for `` fast '' nonlinear diffusion neither for the porous medium case , nor for the @xmath6-laplacian case .
however , it was proved by cabr and roquejoffre for the fractional laplacian @xmath60 in @xcite , in dimension @xmath61 . in order to fully understand the importance of theorem [ theoremboundsforlevelsetsfast ]
, we need to compare it with the linear case @xmath15 and @xmath62 , see formula .
as we will explain later , in the linear case the location of the level sets is given by a main linear term in @xmath59 with a logarithmic shift for large times , see @xcite .
in other words , the propagation of the front is linear `` up to '' a logarithmic correction , for large times .
now , theorem [ theoremboundsforlevelsetsfast ] asserts that this correction does not occur in the `` fast diffusion '' range . using the logarithmic scale , we can compare the behaviour of our level sets with the ones of formula for linear diffusion , noting that there is no logarithmic deviation , but the location of the level sets is approximately linear for large times ( in spatial logarithmic scale , of course ) , and moreover there is a bounded interval of uncertainty on each level set location . in section [ sectionmaxprinccyldomains ]
we prove a maximum principle for a parabolic equation of @xmath6-laplacian type in non - cylindrical domains , see proposition [ maxprinnoncyldomains ] .
the idea of comparing sub- and super - solutions in non - cylindrical domains comes from @xcite and it will turn out to be an extremely useful technical tool in the proof of theorem [ convergencetoonefastdiffusion ] .
section [ appendixselfsimsolincinitdata ] is an appendix .
we present some knew results on the existence , uniqueness and regularity for solutions of the `` pure diffusive '' parabolic equation with @xmath6-laplacian diffusion and non - integrable initial data .
in particular , we focus on radial data @xmath63 , @xmath64 and we study some basic properties of the self - similar solutions with datum @xmath65
. the results of this section are needed for proving theorem [ convergencetoonefastdiffusion ] . finally , in section [ sectionfinalremarksfast ]
we conclude the paper with some comments and open problems related to our study . in particular , we focus on the range of parameters @xmath66 the case @xmath67 is critical in our study while the range @xmath68 is also known in literature as `` very fast '' diffusion range .
one of the problems of this range is the lack of basic tools and basic theory ( existence , uniqueness , regularity of the solutions and estimates ) known for the porous medium equation and for the @xmath6-laplacian equation , but not in the doubly nonlinear setting .
in this brief section , for the reader s convenience , we recall some known results about problem with related bibliography , and we introduce some extremely useful tools such as `` barenblatt solutions '' , we will need through the paper . as we have explained before , we present here the literature and past works linked to our paper , in order to motivate our study . basically , the goal is to give to the reader a suitable background on the fisher - kpp theory , so that our new results can be compared and fully understood .
[ [ finite - propagation - the - doubly - nonlinear - case . ] ] finite propagation : the doubly nonlinear case .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in @xcite , we studied problem assuming @xmath69 and initial datum satisfying @xmath70 before enunciating the main results we need to introduce the notion of travelling waves ( tws ) .
they are special solutions with remarkable applications , and there is a huge mathematical literature devoted to them .
let us review the main concepts and definitions .
fix @xmath4 and @xmath71 and assume that we are in space dimension 1 ( note that when @xmath72 , the doubly nonlinear operator has the simpler expression @xmath73 ) .
a tw solution of the equation @xmath74 is a solution of the form @xmath75 , where @xmath76 , @xmath77 and the _ profile _ @xmath78 is a real function . in the application to the fisher - kpp problem
, the profile is assumed to satisfy @xmath79 in that case we say that @xmath75 is an _ admissible _ tw solution .
similarly , one can consider admissible tws of the form @xmath75 with @xmath80 , @xmath81 decreasing and such that @xmath82 and @xmath83 .
but it is easy to see that these two options are equivalent , since the the shape of the profile of the second one can be obtained by reflection of the first one , @xmath84 , and it moves in the opposite direction of propagation .
+ finally , an admissible tw is said _ finite _ if @xmath85 for @xmath86 and/or @xmath87 for @xmath88 , or _ positive _ if @xmath89 , for all @xmath90 .
the line @xmath91 that separates the regions of positivity and vanishing of @xmath47 is then called the _
free boundary_. now , we can proceed .
we proved that the existence of admissible tw solutions depends on the wave s speed of propagation .
in particular , we showed the following theorem , cfr . with theorem 2.1 and theorem 2.2 of @xcite .
[ theoremexistenceoftws ] let @xmath4 and @xmath5 such that @xmath21 .
then there exists a unique @xmath92 such that equation possesses a unique admissible tw for all @xmath93 and does not have admissible tws for @xmath94 .
uniqueness is intended up to reflection or horizontal displacement . moreover ,
if @xmath95 , the tw corresponding to the value @xmath96 is finite ( i.e. , it vanishes in an infinite half - line ) , while the tws corresponding to the values @xmath97 are positive everywhere . finally , when @xmath98 , any admissible tw is positive everywhere .
the concept of admissible tws and the problem of their existence was firstly introduced in @xcite and @xcite .
then aronson and weinberger , see @xcite , proved theorem [ theoremexistenceoftws ] in the case of the linear diffusion , i.e. @xmath15 and @xmath62 ( note that the choice @xmath15 and @xmath62 is a subcase of @xmath98 ) .
later , the problem of the existence of critical speeds and admissible tws for the fisher - kpp equation has been studied for the porous medium diffusion ( @xmath99 and @xmath62 ) , see @xcite and @xcite .
recently , see @xcite , it has been proved the existence of admissible tws and admissible speeds of propagation when @xmath15 and @xmath100 , i.e. @xmath6-laplacian diffusion . in theorem [ theoremexistenceoftws ]
we generalized these results when doubly nonlinear diffusion is considered and @xmath21 .
then we focused on the pde part in which we studied the asymptotic behaviour of more general solutions , proving the following theorem ( theorem 2.6 of @xcite ) .
[ ntheoremconvergenceinneroutersets ] fix @xmath1 .
let @xmath4 and @xmath5 such that @xmath21 .
\(i ) for all @xmath94 , the solution @xmath47 of the initial - value problem with initial datum satisfies @xmath101 ( ii ) moreover , for all @xmath97 it satisfies , @xmath102 in the case in which the function @xmath3 stands for a density of population , the statement of the previous theorem means that the individuals tend to occupy all the available space and , for large times , they spread with constant speed , see @xcite . from the mathematical point of view
, we can state that the steady state @xmath103 is asymptotically stable while the null solution @xmath104 is unstable and , furthermore , the asymptotic stability / instability can be measured in terms of speed of convergence of the solution which , in this case , is asymptotically linear in distance of the front location as function of time .
again we recall that for @xmath15 and @xmath62 , the previous theorem was showed in @xcite , while for @xmath105 and @xmath62 in @xcite .
we point out that in this last paper , the authors worked with a slightly different reaction term , they called `` strong reaction '' , see also @xcite . in the linear case @xmath15 and @xmath62 , the statements of theorem [ ntheoremconvergenceinneroutersets ]
were improved .
indeed , when @xmath61 , bramson showed an interesting property of the level sets @xmath106 , @xmath107 , of the solution @xmath3 of equation ( with @xmath15 and @xmath62 ) with reaction term satisfying . in particular , in @xcite and @xcite , it was proved that for all @xmath107 there exist constants @xmath108 , @xmath109 and @xmath51 such that @xmath110\ ] ] for @xmath59 large enough , where @xmath111 .
the previous formula is interesting since it allows to estimate the `` delay '' of the solution @xmath3 from the positive tw with critical speed @xmath112 which , according to , grows in time and consists in a logarithmic deviance .
furthermore , he showed that general solutions converge uniformly to the tw with critical speed of propagation ( once it is `` shifted '' of a logarithmic factor @xmath113 ) , for large times .
more recently , similar results have been proved in @xcite with pdes techniques .
[ [ exponential - propagation - for - porous - medium - fast - diffusion . ] ] exponential propagation for porous medium fast diffusion .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let s now resume briefly the results of king and mccabe ( @xcite ) , which have inspired this paper .
they considered the porous medium case , which is obtained by taking @xmath62 in the equation in : @xmath114 in the fast diffusion range , @xmath115 ( note that we absorbed a factor @xmath12 by using a simple change of variables ) .
they considered non - increasing radial initial data @xmath30 decaying faster than @xmath116 as @xmath117 and studied radial solutions of problem .
they showed that when @xmath118 , the radial solutions @xmath119 of the previous equation converge pointwise to 1 for large times with exponential rate @xmath120 , for @xmath121 , and that the `` main variation in concentration occurs on the scale @xmath122 '' , ( see @xcite pag .
2544 ) where @xmath123 in their notation .
we will present an adaptation of their methods to our case in section [ convergencetozerofast ] .
note that `` our '' critical exponent @xmath124 generalizes the value @xmath125 to the case @xmath5 and to more general reaction terms than @xmath25 .
this is a severe departure from the tw behavior of the standard fisher - kpp model since there are no tw solutions .
instead , they found that radial solutions of have non - tws form for large times : @xmath126 where @xmath123 and @xmath127 .
the case @xmath128 is studied too and , as we have anticipated , we will discuss this range in the final section with some comments .
exponential propagation happens also with fractional diffusion , both linear and nonlinear , see for instance @xcite and the references therein .
we will not enter here into the study of the relations of our paper with nonlinear fractional diffusion , though it is an interesting topic . finally , we recall that infinite speed of propagation depends not only on the diffusion operator but also on the initial datum . in particular ,
in @xcite , hamel and roques found that the solutions of the fisher - kpp problem with linear diffusion i.e. , ( @xmath15 and @xmath62 ) propagate exponentially fast for large times if the initial datum has a power - like spatial decay at infinity .
the scene is set for us to investigate what happens in the presence of a fast doubly nonlinear diffusion .
now we present some basic results concerning the barenblatt solutions of the `` pure diffusive '' doubly nonlinear parabolic equation which are essential to develop our study in the next sections ( the reference for this issue is @xcite ) .
moreover , we recall some basic facts on existence , uniqueness , regularity and maximum principles for the solutions of problem . [ [ barenblatt - solutions . ] ] barenblatt solutions .
+ + + + + + + + + + + + + + + + + + + + + fix @xmath4 and @xmath5 such that @xmath29 and consider the `` pure diffusive '' doubly nonlinear problem : @xmath129 where @xmath130 is the dirac s function with mass @xmath131 in the origin of @xmath132 and the convergence has to be intended in the sense of measures .
it has been proved ( see @xcite ) that problem admits continuous weak solutions in self - similar form @xmath133 , called barenblatt solutions , where the _ profile _ @xmath134 is defined by the formula : @xmath135^{-\frac{p-1}{\widehat{\gamma}}},\ ] ] where @xmath136 @xmath137 is determined in terms of the mass choosing @xmath138 ( see @xcite for a complete treatise ) .
we point out that there is an equivalent formulation ( see @xcite for the case @xmath62 ) in which the barenblatt solutions are written in the form @xmath139^{-\frac{p-1}{\widehat{\gamma } } } , \qquad r(t ) = \big[(n/\alpha)t \big]^{\frac{\alpha}{n}},\ ] ] where @xmath140 is a new constant
. it will be useful to keep in mind that we have the formula @xmath141 which describes the relationship between the barenblatt solution of mass @xmath142 and mass @xmath143 and the estimates on the profile corresponding to the barenblatt solution of mass @xmath131 : @xmath144 for suitable positive constants @xmath145 and @xmath146 depending on @xmath131 . [
[ existence - uniqueness - regularity - and - maximum - principles . ] ] existence , uniqueness , regularity and maximum principles .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + before presenting the main results of this paper , we briefly discuss the basic properties of the solutions of problem .
results about existence of weak solutions of the pure diffusive problem and its generalizations , can be found in the survey @xcite and the large number of references therein .
the problem of uniqueness was studied later ( see for instance @xcite ) . the classical reference for
the regularity of nonlinear parabolic equations is @xcite , followed by a wide literature . for the porous medium case ( @xmath62 ) we refer to @xcite , while for the @xmath6-laplacian case we suggest @xcite and the references therein . finally ,
in the doubly nonlinear setting , we refer to @xcite .
the results obtained show the hlder continuity of the solution of problem .
finally , we mention @xcite for a proof of the maximum principle .
in this section we study the asymptotic behaviour of the solution of the cauchy problem with non - trivial initial datum @xmath147 satisfying : @xmath148 for some constant @xmath149 .
we recall here the definition of the critical exponent @xmath150 before proceeding , let us see how to formally derive the value of the critical exponent @xmath124 in the case @xmath29 and @xmath25 ( note that @xmath151 ) .
we follow the methods used in @xcite .
first of all , we fix @xmath29 and we consider radial solutions of the equation in , which means @xmath152 note that the authors of @xcite worked with a slightly different equation ( they absorbed the multiplicative factor @xmath153 with a simple change of variables ) . we linearize the reaction term and we assume that @xmath119 satisfies @xmath154 now , we look for a solution of of the form @xmath155 for @xmath156 which agrees with the assumption on the initial datum and with the linearization .
it is straightforward to see that for such solution , the function @xmath157 has to solve the equation @xmath158 note that since @xmath159 , we have that @xmath160 is well defined and positive , while @xmath161 .
equation belongs to the famous bernoulli class and can be explicitly integrated : @xmath162 hence , for all fixed @xmath163 , we obtain the asymptotic expansion for our solution @xmath164 now , for all fixed @xmath165 , we consider a solution @xmath166 of the _ logistic _ equation @xmath167 which describes the state in which there is not diffusion and the dynamics is governed by the reaction term .
we assume to have @xmath168 where the leading - order term @xmath166 satisfies @xmath169 for some unknown function @xmath170 , with @xmath171 , as @xmath172 .
now , matching with for @xmath59 large and @xmath156 , we easily deduce @xmath173 thus , substituting @xmath174 in and taking @xmath175 for @xmath176 , we have @xmath177 where @xmath178
. the previous formula corresponds to a `` similarity reduction '' ( see @xcite , pag .
2533 ) of the logistic equation with @xmath179 .
note that taking @xmath180 and @xmath181 , we have @xmath182 for @xmath176 while if @xmath183 and @xmath184 we have @xmath185 for @xmath176 .
this means that setting @xmath186 , @xmath187 is a `` critical '' curve , in the sense that it separates the region in which the solution @xmath119 converges to @xmath188 to the one which converges to @xmath189 .
we will show this property in theorem [ convergencetozerofastdiffusion ] and theorem [ convergencetoonefastdiffusion ] . in what follows , we prove that the solution of problem with initial datum converges uniformly to the trivial solution @xmath104 in the outer set @xmath190 as @xmath191 if @xmath32 . in the nex sections we will prove that this solution converges uniformly to the equilibrium point @xmath103 in the inner set @xmath192 as @xmath191 if @xmath44 . [ [ proof - of - theorem - convergencetozerofastdiffusion . ] ]
proof of theorem [ convergencetozerofastdiffusion ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + fix @xmath1 , @xmath29 , and @xmath32 .
first of all , we construct a super - solution for problem , using the hypothesis on the function @xmath24 . indeed , since @xmath193 for all @xmath26 , the solution of the linearized problem @xmath194 gives the super - solution we are interested in and , by the maximum principle , we deduce @xmath195 in @xmath196 .
now , consider the change of the time variable @xmath197 , \quad \text{for } t \geq 0,\ ] ] with @xmath198 .
then the function @xmath199 solves the problem @xmath200 from the properties of the profile of the barenblatt solutions and the hypothesis on the initial datum , it is evident that there exist positive numbers @xmath201 and @xmath202 such that @xmath203 in @xmath132 and so , by comparison , we obtain @xmath204 now , since the profile of barenblatt solutions satisfies @xmath205 for some constant @xmath206 and for all @xmath207 ( see ) , we can perform the chain of upper estimates @xmath208 where we set @xmath209 and we used the first relation in in the third inequality .
note that we used that @xmath210 , too .
now , supposing @xmath211 in the last inequality , we get @xmath212 since we have chosen @xmath32 , completing the proof . @xmath213
this section is devoted to prove that for all @xmath214 and initial data `` small enough '' , the solution of problem lifts up to a ( small ) positive constant on the `` inner sets '' @xmath45 for large times .
let @xmath36 and @xmath37 be positive real numbers and , for all @xmath29 , consider the initial datum @xmath215 where @xmath38 . note that @xmath216 has `` tails '' which are asymptotic to the profile of the barenblatt solutions for @xmath217 large ( see formula )
. the choice will be clear in the next sections , where we will show the convergence of the solution of problem , to the steady state @xmath103 .
the nice property of the initial datum is that it can be employed as initial `` sub - datum '' as the following lemma shows .
[ lemmaplacingbarenblattundersolution ] fix @xmath1 and let @xmath4 and @xmath5 such that @xmath29 .
then for all @xmath218 , there exist @xmath219 , @xmath41 , and @xmath42 , such that the solution @xmath3 of problem with nontrivial initial datum @xmath220 satisfies @xmath221 where @xmath216 is defined in .
_ let @xmath3 the solution of problem with nontrivial initial datum @xmath220 and consider the solution @xmath222 of the purely diffusive cauchy problem : @xmath223 it satisfies @xmath224 in @xmath225 thanks to the maximum principle .
let @xmath218 .
since @xmath226 is continuous in @xmath132 and non identically zero ( the mass of the solution is conserved in the `` good '' exponent range @xmath29 ) , we have that it is strictly positive in a small ball @xmath227 , @xmath228 and @xmath229 . without loss of generality , we may take @xmath230 .
so , by continuity , we deduce @xmath231 in @xmath232 $ ] , for some small @xmath233 and @xmath234 .
now , let us consider the function @xmath235 and the exterior cylinder @xmath236 we compare @xmath237 with a `` small '' barenblatt solution @xmath238 ( we mean that @xmath201 is small ) at time @xmath239 and on the boundary of @xmath240 .
recall that barenblatt solutions have the self - similar form @xmath241^{-\frac{p-1}{\widehat{\gamma}}},\ ] ] where @xmath242 and @xmath243 are positive constants defined in subsection [ preliminariesintro ] , and @xmath244 depends on the mass . the comparison at time @xmath245 is immediate since @xmath246 and @xmath247 for all @xmath248 .
now , let us take @xmath249 . we want to show that @xmath250 for all @xmath251 .
a simple computation shows that we can rewrite the barenblatt solution as @xmath252^{\frac{p-1}{\widehat{\gamma } } } } \leq \bigg ( \frac{\tau}{k^{p-1}\varrho^p } \bigg)^{\frac{1}{\widehat{\gamma}}},\ ] ] since @xmath251 .
thus , since @xmath253 , it is sufficient to have @xmath254 this condition is satisfied taking @xmath234 small enough .
we may now use the maximum principle to obtain the conclusion @xmath255 in the whole of @xmath240 .
in particular , we evaluate the comparison at @xmath256 , and we have @xmath257 where we set @xmath258 .
let us fix @xmath259 , @xmath260 , and @xmath261 . by taking @xmath41 smaller
, we can assume @xmath262 .
now , we verify that @xmath263 and some suitable constant @xmath244 .
writing the expression for the barenblatt solutions , the previous inequality reads : @xmath264^{\frac{1}{p-1 } } } |x|^{\frac{p}{p-1 } } : = k|x|^{\frac{p}{p-1 } } , \quad \text { for all } |x| \geq \widetilde{\varrho}_0.\ ] ] note that the coefficient @xmath265 of @xmath266 is positive thanks to our assumptions on @xmath41 .
now , since @xmath267 , we deduce that a sufficient condition so that the previous inequality is satisfied is @xmath268 . consequently , we have shown that for all @xmath218 , there exist @xmath219 , @xmath41 , and @xmath42 , such that @xmath269 which is our thesis .
@xmath213 we ask the reader to note that improved global positivity estimates were proved in @xcite and @xcite for the porous medium equation .
now , with the next crucial lemma , we prove that the expansion of the super - level sets of the solution @xmath3 of problem with initial datum is exponential for all @xmath44 and large times .
[ lemmaexpandingfastlevelsets ] fix @xmath1 .
let @xmath4 and @xmath5 such that @xmath29 , and let @xmath39 .
+ then there exist @xmath40 and @xmath270 which depend only on @xmath12 , @xmath6 , @xmath271 and @xmath272 , such that the following hold .
for all @xmath273 , there exists @xmath42 ( large enough depending on @xmath41 ) , such that the solution @xmath3 of problem with initial datum satisfies @xmath274 _ proof . _
we prove the assertion of the thesis by induction on @xmath275 .
we follow the ideas presented by cabr and roquejoffre in @xcite and , later , in @xcite , for fractional diffusion .
_ step0 . _
we set @xmath276 , @xmath39 and introduce some basic definitions and quantities we will use during the proof .
first of all , let @xmath277 be the constant corresponding to the profile @xmath278 ( see section [ preliminariesintro ] ) and let @xmath145 and @xmath146 be defined as in with @xmath143 .
in order to avoid huge expressions in the following of the proof , we introduce the constants @xmath279 we fix @xmath280 sufficiently small such that @xmath281 then , we consider @xmath282 sufficiently large such that @xmath283 ( note that such a @xmath282 exists thanks to ) and we define @xmath284 . finally , fix @xmath273 and choose @xmath37 large enough such that @xmath285 we anticipate that the choice of the subtle conditions , and will be clarified during the proof . _
_ construction of a sub - solution of problem , in @xmath286 $ ] .
first of all , we construct a barenblatt solution of the form @xmath287 such that @xmath288 since the profile of the barenblatt solution is decreasing , we impose @xmath289 in order to satisfy in the set @xmath290 .
moreover , using and noting that @xmath291 , it simple to get @xmath292 and so , it is sufficient to require @xmath293 , so that is valid in @xmath294 . thus , it is simple to obtain the relations @xmath295 now , consider the linearized problem @xmath296 and the change of variable @xmath297 , \quad \text{for } t \geq 0.\ ] ] note that @xmath298 and the function @xmath299 solves the `` pure diffusive '' problem @xmath300 since @xmath301 for all @xmath302 , from the maximum principle we get @xmath303 hence , using the concavity of @xmath272 and the second inequality in we get @xmath304\ ] ] and so , since @xmath305 implies @xmath306 , we have that @xmath307 is a sub - solution of problem , in @xmath286 $ ] . finally ,
using the first inequality in , we obtain @xmath308.\ ] ] _ step2 . _ in this step , we show that the choices made in , and allow us to find positive numbers @xmath309 and @xmath310 such that @xmath311 for all @xmath302 , where @xmath312 which implies the thesis for @xmath276 .
now , in order to find @xmath309 and @xmath310 we proceed with the chain of inequalities in for the values @xmath313 , @xmath314 and @xmath315 . imposing @xmath316^{-\alpha / n } \geq 1,\ ] ] using and observing that @xmath317 for all @xmath318 , we look for @xmath309 and @xmath310 such that @xmath319^{-\frac{\alpha p}{\;n\widehat{\gamma } } } \big\}^{-1 } \\ & \geq \frac{k_2}{2}e^{\lambda t_0 } m_1^{1+\alpha\widehat{\gamma}}(\theta_1 + \tau_0)^{-\alpha}\big\ { \widetilde{\varrho}_1 \big[m_1^{-\widehat{\gamma}}(\theta_1 + \tau_0)\big]^{-\frac{\alpha } { n } } \big\}^{-\frac{p}{\widehat{\gamma } } } \\ & = \frac{k_2}{2}e^{\lambda t_0 } ( \theta_1 + \tau_0)^{\frac{1}{\widehat{\gamma}}}\widetilde{\varrho}_1^{-\frac{p}{\widehat{\gamma } } } \\ & \geq \widetilde{\varepsilon } = a_1\widetilde{\varrho}_1^{-\frac{p}{\widehat{\gamma}}}. \end{aligned}\ ] ] thus , we get @xmath320 taking , for instance , @xmath321 such that @xmath322 note that this choice of @xmath321 performs the equality at the end of the previous chain and the value @xmath323 is determined too . _
remark 1_. note that the conditions @xmath324 and are sufficient to assure @xmath325 in @xmath132 .
indeed , it is guaranteed in the set @xmath326 since the profile @xmath278 is non - increasing . on the other hand , if @xmath327 we have @xmath328^{-\alpha / n } \geq 1 $ ] by and so , following the chain of inequalities as before , we get @xmath329 now , in order to conclude the proof of the case @xmath276 , we must check that the conditions and actually represent a possible choice and the value of @xmath282 , defined at the beginning , performs their compatibility .
the compatibility between and can be verified imposing @xmath330^{\frac{\alpha p}{\;n\widehat{\gamma}}},\ ] ] which can be rewritten using the definitions as @xmath331 now , it is simple to verify that condition implies @xmath332 and so it holds @xmath333 too .
hence , a sufficient condition so that is satisfied and does not depend on @xmath41 is @xmath334 i.e. , our initial choice of @xmath282 in which proves the compatibility between and . _
remark 2_. rewriting formula using the definition of @xmath335 , it is simple to deduce @xmath336 and , using the second hypothesis on @xmath282 in , it is straightforward to obtain @xmath337 .
in particular , we have shown @xmath338 i.e. , the thesis for @xmath276 .
[ [ iteration . ] ] iteration .
+ + + + + + + + + + set @xmath339 , @xmath340 and @xmath341 for all @xmath342 and define @xmath343 we suppose to have proved that the solution of problem , satisfies @xmath344 and we show @xmath345 in @xmath132 for the values @xmath346 and @xmath347 . from the induction hypothesis
, we have that the solution @xmath348 of the problem @xmath349 is a sub - solution of problem , in @xmath350 which implies @xmath351 in @xmath350 and so , it is sufficient to prove @xmath352 in @xmath132 .
since we need to repeat almost the same procedure of the case @xmath276 , we only give a brief sketch of the induction step . _
step1_. construction of a sub - solution of problem , , in @xmath353 $ ] . with the same techniques used in _
, we construct a barenblatt solution @xmath354 in @xmath132 with parameters @xmath355 and a sub - solution of problem , : @xmath356 in @xmath353 $ ] , where @xmath357 , \quad \text{for } t \geq t_{j-1}.\ ] ] in particular , note that @xmath358 , @xmath359 , @xmath360 and @xmath361 _ step2_. we have to study a chain of inequalities similar to the one carried out in _ step2 _ verifying that @xmath362 thus , imposing conditions similar to and and requiring their compatibility , we have to check the validity of the inequality @xmath363 since @xmath364 , we have @xmath365 and so , a sufficient condition so that the previous inequality is satisfied is @xmath366 , which is guaranteed by the initial choice of @xmath282 .
finally , following the reasonings of the case @xmath276 it is simple to obtain the relation @xmath367 which implies @xmath368 and we complete the proof . @xmath213 * proof of proposition [ expanpansionofminimallevelsets ] . *
the previous lemma proves that for the sequence of times @xmath369 and for any choice of the parameter @xmath39 , the solution of problem , reaches a positive value @xmath36 in the sequence of sets @xmath370 where @xmath42 is chosen large enough ( in particular , we can assume @xmath371 ) . actually , we obtained a more useful result .
first of all , note that , for all @xmath39 , lemma [ lemmaexpandingfastlevelsets ] implies @xmath372 for all @xmath373 .
moreover , since conditions are satisfied for all @xmath374 , we can repeat the same proof of lemma [ lemmaexpandingfastlevelsets ] , modifying the value of @xmath375 and choosing a different value @xmath376 , which is smaller but strictly positive for all @xmath374 .
hence , it turns out that for all @xmath377 , it holds @xmath378 now , iterating this procedure as in the proof of lemma [ lemmaexpandingfastlevelsets ] , it is clear that we do not have to change the value of @xmath379 when @xmath380 grows and so , for all @xmath377 , we obtain @xmath381 then , using the arbitrariness of @xmath380 , we complete the proof .
@xmath213 [ [ remark .
] ] remark .
+ + + + + + + note that , to be precise , in the proof of proposition [ expanpansionofminimallevelsets ] , we have to combine lemma [ lemmaplacingbarenblattundersolution ] with lemma [ lemmaexpandingfastlevelsets ] as follows .
let @xmath3 the solution of problem with initial datum .
we wait a time @xmath382 given by lemma [ lemmaplacingbarenblattundersolution ] , in order to have @xmath383 for all @xmath42 and some @xmath41 depending on @xmath384 .
now , thanks to the maximum principle , we deduce @xmath385 in @xmath225 , where we indicate with @xmath386 the solution of problem with initial datum @xmath387 . in this way
, we deduce the statement of lemma [ lemmaexpandingfastlevelsets ] for more general initial data satisfying and we can prove proposition [ expanpansionofminimallevelsets ] .
as mentioned in the introduction , we now address to the problem of showing the convergence of a general solution of problem , to the steady state @xmath103 . as anticipated , we find that the convergence to 1 is exponential for large times , with exponent @xmath44 .
this fact represents an interesting deviance , respect to the case @xmath21 ( i.e. @xmath22 ) in which the solutions converge with constant speed for large times and show a tw asymptotic behaviour .
[ [ proof - of - theorem - convergencetoonefastdiffusion . ] ] proof of theorem [ convergencetoonefastdiffusion ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + fix @xmath29 , @xmath39 and set @xmath388 in @xmath196 .
we will prove that for all @xmath389 , there exists @xmath390 such that @xmath391 which is equivalent to the assertion of the thesis . _
_ fix @xmath392 and consider the inner set @xmath393 , where @xmath382 is initially arbitrary .
we recall that proposition [ expanpansionofminimallevelsets ] assures the existence of @xmath41 and @xmath40 such that @xmath394 in the set @xmath395 . in particular , for all @xmath396 ,
we have that @xmath3 is bounded from below and above in the inner set : @xmath397 moreover , it is not difficult to see that , setting @xmath398 and @xmath399 , the function @xmath400 solves the problem @xmath401^m \qquad\qquad & \text{in } { \mathbb{r}}^n . \end{aligned } \end{cases}\ ] ] using , it is simple to see that @xmath402 where @xmath403 for what concerns @xmath404 , it is bounded from below in @xmath405 : @xmath406 where @xmath407 and depends on @xmath36 and @xmath12 . indeed , if @xmath115 we have that @xmath408 for all @xmath26 .
hence , we get our bound from below recalling and noting that @xmath409 if @xmath410 , we have the formula @xmath411 and so , since @xmath412 and arguing as in the case @xmath115 , we deduce @xmath413 for some @xmath407 depending on @xmath36 and @xmath12 . in particular , it follows that @xmath414 satisfies @xmath415 i.e. , @xmath414 is a sub - solution for the equation in problem in the set @xmath405 .
_ in this step , we look for a super - solution @xmath416 of problem with @xmath417 in @xmath418 .
we consider the solution of the problem @xmath419 according to the resume presented in section [ appendixselfsimsolincinitdata ] , problem is well posed if @xmath420 when @xmath421 .
further assumptions are not needed when @xmath422 . furthermore , since @xmath407 can be chosen smaller and @xmath423 larger , we make the additional assumption @xmath424 now , we define the function @xmath425 \;\;\quad & \text{if } 1 < p < 2 \\ \frac{1}{a_1}(t - t_1 ) \qquad\qquad\qquad\qquad\qquad\quad & \text{if } p = 2 \\ \frac{1}{c_0(p-2)}\big[1 - e^{-(c_0/a_1)(p-2)(t - t_1 ) } \big ] \quad & \text{if } p > 2 .
\end{aligned } \end{cases}\ ] ] note that @xmath426 is increasing with @xmath427 for all @xmath5 .
moreover , we define the limit of @xmath428 as @xmath191 with the formula @xmath429^{-1 } \quad & \text{if } p > 2 .
\end{aligned } \end{cases}\ ] ] then , the function @xmath430 ( with @xmath426 ) solves the `` pure diffusive '' problem @xmath431 as we explained in section [ appendixselfsimsolincinitdata ] , for all @xmath432 the problem @xmath433 admits self - similar solutions @xmath434 , with self - similar exponents @xmath435 and profile @xmath436 with @xmath437 for all @xmath438 , where we set @xmath439 . note that since we assumed @xmath420 when @xmath421 , the self - similar exponents are well defined with @xmath440 and @xmath441 for all @xmath5 . finally , recall that it is possible to describe the spacial `` decay '' of the self - similar solutions for large values of the variable @xmath439 , with the bounds @xmath442 for a constant @xmath443 large enough , see formula .
now , it is not difficult to see that @xmath444 , for all fixed delays @xmath432 .
moreover , we compute the time derivative : @xmath445 \big\ } \\ & = -(\tau + \tau_1)^{-\alpha_{\lambda}-1}e^{-\frac{c_0}{a_1}(t - t_1 ) } \bigg\ { \frac{c_0}{a_1}(\tau + \tau_1)^{\alpha_{\lambda}+1 } + \bigg[\frac{c_0}{a_1}(\tau + \tau_1 ) + \alpha_{\lambda } \tau ' \bigg]f(\xi ) + \beta_{\lambda } \tau ' \xi f'(\xi ) \bigg\ } , \end{aligned}\ ] ] where @xmath446 and @xmath447 stands for the derivative respect with the variable @xmath448 .
let s set @xmath449 since , @xmath450 , @xmath451 , and @xmath452 are non - negative and @xmath441 , in order to have @xmath453 , it is sufficient to show @xmath454 for all @xmath455 and a suitable choice of @xmath456 .
if @xmath13 , this follows from a direct and immediate computation , choosing @xmath456 large enough .
if @xmath457 , we may proceed similarly .
it is simple to see that condition @xmath454 for @xmath455 reads @xmath458e^{(c_0/a_1)(2-p)(t - t_1 ) } \geq 1 - \frac{\tau_1}{\tau_{\infty}}.\ ] ] consequently , since @xmath459 , it is sufficient to choose @xmath460 . finally , when @xmath421 it holds @xmath461 for all @xmath455 .
hence , it is simple to see that the choice @xmath462 is a sufficient condition so that @xmath454 for all @xmath455 .
we stress that the choice of @xmath456 is independent of @xmath382 .
now , using the fact that @xmath463 in @xmath418 and that @xmath464 in @xmath405 , it is straightforward to see that @xmath465 _ step3 .
_ now we compare the functions @xmath466 and @xmath467 , applying the maximum principle of section [ sectionmaxprinccyldomains ] .
hence , we have to check that the assumptions in proposition [ maxprinnoncyldomains ] are satisfied .
it is simple to see that it holds @xmath468 in @xmath132 .
indeed , we have @xmath469 while @xmath470^m \leq 1 $ ] .
we have to check that @xmath471 on the boundary of @xmath405 , i.e. , on the set @xmath472 .
we use the first estimate in : @xmath473 \\ & \geq e^{-(c_0/a_1)(t - t_1 ) } ( 1 + h_2|x|^{\lambda } ) = e^{-(c_0/a_1)(t - t_1 ) } ( 1 + h_2e^{\nu\lambda t } ) \\ & = e^{(c_0/a_1)t_1}(h_2 + e^{-(c_0/a_1)t } ) \geq 1 \geq \underline{w } \quad \text { in } \{|x| = e^{\nu t } , t \geq t_1\}. \end{aligned}\ ] ] first of all , we point out that the last equality in the preceding chain is satisfied thanks to the first assumption in , i.e. , @xmath474 .
secondly , we note that the first inequality holds only if @xmath475 , which means @xmath476 as the reader can easily check , when @xmath62 , is satisfied by taking @xmath396 so that @xmath477 . if @xmath421 , it is sufficient to fix @xmath396 to have @xmath478 .
the case @xmath457 is a little bit subtle .
first of all , set @xmath479 and note that , thanks to assumption , we have that is automatically satisfied if @xmath480^{-1 } e^{\nu\lambda(2-p)(t - t_1 ) } + \tau_1 \big\},\ ] ] which , since @xmath481 , is equivalent to @xmath482^{-1}h^{b_{\lambda } } \big\ } \geq h^{b_{\lambda}}\big\{[c_0(2-p)]^{-1}e^{-\nu \lambda(2-p)t_1 } + \tau_1 \big\}.\ ] ] finally , it is straightforward to see that the last inequality is satisfied for all @xmath483 so that @xmath484^{-1 } + \tau_1 \big\}.\ ] ] hence , we have that condition is satisfied for all @xmath5 when @xmath396 is taken large enough .
( a3 ) . to check this third assumption
it is sufficient to combine and , restricting they validity to the set @xmath485 .
hence , we deduce @xmath471 in @xmath485 by applying proposition [ maxprinnoncyldomains ] .
so we have @xmath486 \leq e^{-\frac{c_0}{a_1}(t - t_1)}[1 + u(e^{\nu t},\tau + \tau_1 ) ] \\ &
\leq e^{-\frac{c_0}{a_1}(t - t_1)}[1 + h_1|x|^{\lambda } ] \leq e^{-\frac{c_0}{a_1}(t - t_1)}[1 + h_1e^{\nu\lambda t } ] \end{aligned}\ ] ] in the set @xmath487 , thanks to .
now , let us fix @xmath389 and take a time @xmath488 , and a constant @xmath489 such that @xmath490 these choices combined with the previous chain of inequalities give us @xmath491 in the set @xmath492 . finally , noting that @xmath493 for all @xmath494 and for some @xmath390 large enough
, we complete the proof of the theorem using the arbitrariness of @xmath392 .
@xmath213 [ [ remarks . ] ] remarks .
+ + + + + + + + we end this section with two remarks .
first of all , we ask the reader to note that at the beginning of the previous proof , we have made the change of variable @xmath400 in order to obtain problem , which has non - constant coefficients , but the diffusion operator simplifies to a @xmath6-laplacian .
this is a considerable advantage since we can employ the well known theory of @xmath6-laplacian diffusion and non - integrable initial data ( see section [ appendixselfsimsolincinitdata ] for more details and references ) to construct the super - solution given by problem .
a different approach could be studying the existence , uniqueness and regularity of solutions for the doubly nonlinear equation and non - integrable initial data in the fast diffusion range @xmath29 . up to our knowledge , this theory has not been developed yet . secondly , we point out that in the previous proof we showed a slightly different result too stated in the following corollary
. it will be very useful in section [ sectionboundsforlevelsetsfast ] , where we will study the behaviour of the solution @xmath3 on the set @xmath48 .
[ conjectureforlevelsetssigmaastfast ] let @xmath4 and @xmath5 such that @xmath29 and let @xmath3 be the solution of the problem with initial datum .
suppose that there exist @xmath495 , @xmath496 , @xmath497 and @xmath498 such that @xmath499 then , for all @xmath50 , there exist @xmath51 large enough and @xmath500 such that @xmath501 the proof coincides with the one of theorem [ convergencetoonefastdiffusion ] .
notice indeed that we have begun by assuming that @xmath502 in @xmath503 and for all @xmath389 , we proved the existence of @xmath504 and @xmath505 such that @xmath506 we point out that in the previous statement the value of the exponent @xmath495 does not change . in the proof of theorem [ convergencetoonefastdiffusion ] , we need to take @xmath494 to obtain a `` convergence inner set '' not depending on @xmath507 .
indeed , in the second one we prove the convergence of the solution @xmath3 to the steady state 1 in the set @xmath45 for all @xmath44 , while now the exponent @xmath495 is arbitrary .
we devote this section to the proof of theorem [ theoremboundsforlevelsetsfast ] .
a similar result was showed in @xcite for the fisher - kpp equation with fractional diffusion . in particular ,
they studied the case of the fractional laplacian @xmath60 and worked in dimension @xmath61 , see theorem 1.6 of @xcite for more details . in our setting , we consider the classical reaction term @xmath25 and the problem @xmath508 where @xmath147 satisfies , i.e. , @xmath509 and @xmath510 , for some constant @xmath511 . as always
, we do not pose restrictions on the dimension @xmath1 , and we will work in the `` fast diffusion '' range @xmath29 .
we divide the proof in two main parts .
the first one is devoted to prove the `` upper bound '' i.e. , the first inclusion . in the second one
, we show the `` lower bound '' ( the second inclusion ) which is the most difficult part .
we point out that in this part , we have to give to separate proofs for the ranges @xmath512 and @xmath513 . [ [ proof - of - theorem - theoremboundsforlevelsetsfast . ] ] proof of theorem [ theoremboundsforlevelsetsfast ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the proof is divided in some steps .
we construct a super - solution and sub - solutions for problem and we use these sub- and super - solutions to prove the level sets bounds .
in particular , we employ the super - solution to prove the upper bound , whilst the sub - solutions for the lower bound .
[ [ upper - bound . ] ] upper bound .
+ + + + + + + + + + + + we begin to prove the first inclusion in , i.e. , for all @xmath50 , there exists a constant @xmath51 large enough such that it holds @xmath514 where @xmath186 ( recall that in this setting @xmath151 ) .
as we mentioned before , we need a special super - solution .
thus , we basically repeat the computations carried out at beginning of section [ convergencetozerofast ] , by looking for radial solutions of the equation @xmath515 with separate variables @xmath516 we have seen that for such solution , the function @xmath157 has to solve the equation @xmath517 note that since @xmath159 we have that @xmath160 is well defined , positive and @xmath161 .
as we have mentioned , we deal with a bernoulli equation and we can integrate it : @xmath518 hence , we found that for all @xmath519 , the function @xmath520 is a super - solution of the equation in .
now , in order to apply the maximum principle , we show @xmath521 in @xmath132 .
we consider the function @xmath522 where @xmath149 is taken as in , see also the beginning of this section .
it is simple to see that we have @xmath523 for all @xmath302 .
so , taking @xmath524 , we have @xmath525 consequently , we have @xmath521 in @xmath132 and we obtain @xmath526 in @xmath225 , by applying the maximum principle .
we are ready to prove the first inclusion in .
thus , let us fix @xmath50 and consider @xmath302 satisfying @xmath527 . using the super - solution @xmath528 constructed before , we have @xmath529 which , setting @xmath530 , implies @xmath531 and we conclude the proof of the first inclusion in .
[ [ lower - bound . ] ] lower bound .
+ + + + + + + + + + + + we want to show the second inclusion in : for all @xmath50 , there exist @xmath51 and @xmath532 large enough , such that @xmath533 the idea consists in constructing a sub - solution @xmath534 that will act as a barrier from below and then employ corollary [ conjectureforlevelsetssigmaastfast ] .
even though this idea was firstly used in @xcite , we stress that our construction is completely independent by the previous one , and the comparison is not done in the whole space , but only on a sub - region .
moreover , we will divide the range @xmath29 in two sub - ranges : @xmath512 and @xmath513 .
of course , these two new ranges are always accompanied with the `` fast diffusion assumption '' @xmath29 .
we will specify when we will need to distinguish between these different cases .
we divide the proof in some steps . _
_ in this first step , we fix some important notations and we define the candidate sub - solution . we know
the following two facts : @xmath535 first , for all @xmath389 ( small ) and for all @xmath536 ( large ) , there exists @xmath537 ( large enough ) such that @xmath538 this is a direct consequence of theorem [ convergencetoonefastdiffusion ] .
@xmath535 secondly , by applying lemma [ lemmaplacingbarenblattundersolution ] , we have that for all @xmath218 , there exist @xmath219 , @xmath41 , and @xmath42 satisfying @xmath539 where @xmath38 , see .
now , for all @xmath4 and @xmath5 satisfying @xmath29 , and @xmath1 , we consider the constants @xmath540 , \quad d_3 : = ( p/\widehat{\gamma})^{2-p}m^{1-p}.\ ] ] note that the assumption @xmath29 guarantees that @xmath541 are positive for all @xmath542 .
the importance of these constants will be clear later .
now , they are simply needed to choose @xmath543 .
we take @xmath544 large ( depending only on @xmath12 , @xmath6 , @xmath271 and @xmath545 ) , satisfying @xmath546 the first assumption will be needed in the range @xmath512 , while the second when @xmath513 .
then we take @xmath547 and @xmath548 and we fix @xmath41 and @xmath42 corresponding to the value of @xmath40 .
let us define the candidate sub - solution .
we consider the function @xmath549 where , of course , @xmath550 . we fix @xmath551 the parameter @xmath552 will be chosen in the next step , independently from @xmath553 .
we ask the reader to note that @xmath553 depends only on @xmath389 .
the fact that @xmath554 will be important later .
_ now , we consider the region @xmath555 and , we show that @xmath556 in order to assure that @xmath557 and @xmath47 are well - ordered at time @xmath313 and on the boundary of @xmath558 .
@xmath535 comparison in @xmath559 at time @xmath313 . since both the parameters @xmath560 and @xmath553 are positive
, we have @xmath561 and so , comparing with @xmath387 , we obtain @xmath562 by taking the parameter @xmath552 large enough depending on @xmath543 ( but not on @xmath553 ) : @xmath563 @xmath535 comparison on the boundary @xmath564 .
this part is simpler .
indeed , we have @xmath565 thanks to our assumptions on @xmath40 and @xmath77 , see .
_ in this step we prove that @xmath566 is a sub - solution of the equation in in the region @xmath558 . for the reader convenience
, we introduce the expression @xmath567 we proceed by carrying out some computations .
we have @xmath568a(r , t)^{-2}.\ ] ] now , we need to compute the radial @xmath6-laplacian of @xmath569 , which is given by the formula @xmath570 first of all , setting @xmath571 and using the fact that @xmath572 , it is not difficult to obtain @xmath573 where @xmath574 .
consequently , we have @xmath575 \\ & = b(t)|\psi'(r)|^{p-2}a(r , t)^{\widehat{\gamma}-p-1}\bigg\{\bigg[\frac{n-1}{r}\psi'(r ) + ( p-1)\psi''(r)\bigg ] a(r , t ) - b(p-\widehat{\gamma})(\psi'(r))^2 \bigg\}. \end{aligned}\ ] ] combining the last quantity with the ones in and multiplying by @xmath576 , we obtain @xmath577 \\ & = \frac{c-1}{(mb)^{p-1}}e^{(1+\widehat{\gamma})t}|\psi'(r)|^{2-p}a(r , t)^{p-1-\widehat{\gamma } } + \big[(p-1)\psi''(r ) + \frac{n-1}{r}\psi'(r ) \big ] a(r , t ) \ , + \\ & - b(p - \widehat{\gamma})(\psi'(r))^2 . \end{aligned}\ ] ] let us take @xmath578 with @xmath579 and @xmath580 . since @xmath581 and recalling that @xmath582 , we substitute in the previous equation deducing @xmath577 \\ & = -d_3\frac{c-1}{b^{p-1}}e^{(1+\widehat{\gamma})t}r^{(2-p)(p/\widehat{\gamma } - 1 ) } \big(br^{p/\widehat{\gamma } } + ce^t\big)^{p-1-\widehat{\gamma } } + d_2ce^tr^{p/\widehat{\gamma } - 2 } - bd_1r^{2(p/\widehat{\gamma}-1 ) } , \end{aligned}\ ] ] where @xmath583 , @xmath542 are chosen as in : @xmath584 , \quad d_3 : = ( p/\widehat{\gamma})^{2-p}m^{1-p}.\ ] ] now , multiplying by @xmath585 and setting @xmath586 , it is not difficult to obtain @xmath587 \\ & = -d_3\frac{c-1}{b^{p-1 } } \,r^p\ , \xi^{1+\widehat{\gamma } } \big(b + c\xi\big)^{p-1-\widehat{\gamma } } + d_2c\xi - bd_1 \\ &
\leq -d_3\frac{c-1}{b^{p-1 } } \,r_0^p\ , \xi^{1+\widehat{\gamma } } \big(b + c\xi\big)^{p-1-\widehat{\gamma } } + d_2c\xi - bd_1 : = -c_{r_0}(\xi ) , \end{aligned}\ ] ] for all @xmath588 . [
[ case - boldsymbolwidehatgamma - leq - p-1 . ] ] case @xmath589 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to prove that @xmath566 is a sub - solution , it is sufficient to check that @xmath590 for all @xmath591
. we will prove the previous inequality in two separate intervals @xmath592 and @xmath593 , where @xmath594 will be suitably chosen .
suppose @xmath592 . in this interval
we have @xmath595 and so , a sufficient condition so that is satisfied ( for @xmath592 ) is @xmath596 suppose @xmath593 and assume to be true .
since we are in the range @xmath512 , we have @xmath597 , and so @xmath598 now , we note that condition not only implies @xmath599 , but also @xmath600 for all @xmath592 . hence , in order to prove that @xmath600 for all @xmath593 , it is sufficient to show that the minimum point of @xmath601 is attained for some @xmath602
. it is straightforward to compute the minimum point @xmath603 of @xmath601 : @xmath604 for our purpose we may choose @xmath605 now , since @xmath606 depends on @xmath553 , we need to check that our choice of @xmath594 is compatible with , which we have assumed to be true .
thus , substituting the value of @xmath606 in , we obtain that the parameter @xmath553 has to satisfy the inequality @xmath607 the crucial fact is that the previous expressions do not depend on @xmath560 . indeed , taking @xmath608 as in
, we can rewrite the previous inequality as @xmath609 which exactly our first assumption in on @xmath543 .
this proves that for all @xmath512 and @xmath29 , the function @xmath566 is a sub - solution for the equation in in the region @xmath610 .
[ [ case - boldsymbolwidehatgamma - p-1 . ] ] case @xmath611 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this range the proof is similar , but there are some technical changes that have to be highlighted .
we rewrite @xmath612 as @xmath613 so , in order to show that @xmath566 is a sub - solution , we can verify that @xmath614 for all @xmath591 .
again we will pick a `` good '' @xmath594 and prove in the intervals @xmath592 and @xmath593 .
suppose @xmath592 . as before ,
in this interval we have @xmath595 and so , taking again @xmath553 as in , i.e. @xmath615 then is automatically satisfied ( for @xmath592 ) . now , suppose @xmath593 and assume again to be true .
since we are in the range @xmath513 , the function @xmath616 is increasing ( in @xmath617 ) , we have @xmath618 where we used in the second inequality .
exactly as in the previous case , condition implies both @xmath599 and @xmath619 for all @xmath592 .
hence , we show that the minimum point of @xmath620 is attained for @xmath621 and this gives us @xmath619 for all @xmath593 .
the minimum point @xmath603 of @xmath620 is given by the formula : @xmath622 so we ask @xmath621 , i.e. : @xmath623 again we must check the compatibility between our choice of @xmath594 and .
so , we substitute the value of @xmath606 in and we obtain the inequality @xmath624 in the parameter @xmath553 . also in this this case is is really important that the previous expressions do not depend on @xmath560 .
we take @xmath608 as in and we rewrite the previous inequality as @xmath625}.\ ] ] since @xmath626 the last inequality is satisfied thanks to the assumption on @xmath543 in .
hence , we have showed that @xmath566 is a sub - solution for the equation in in the region @xmath610 , for the range @xmath513 , too .
consequently , for all @xmath29 , we obtain @xmath627 thanks to the comparison at time @xmath313 and on the boundary of @xmath558 done in _
step2_. note that the maximum principle can be applied since @xmath628 in @xmath225 and @xmath25 can be re - defined outside @xmath629 $ ] to be lipschitz continuous . _
_ in this last step , we conclude the proof .
the following procedure holds for all @xmath29 ( see also @xcite ) .
thanks to corollary [ conjectureforlevelsetssigmaastfast ] , to deduce the second inclusion in : @xmath630 it is sufficient to prove @xmath631 in @xmath632 , for some @xmath497 .
so , in the set @xmath633 , we have @xmath634 note that the bound @xmath631 can be extended to the region @xmath635 , thanks to our assumption on @xmath282 and theorem [ convergencetoonefastdiffusion ] .
consequently , applying corollary [ conjectureforlevelsetssigmaastfast ] with @xmath636 , @xmath637 and @xmath638 , we end the proof of the theorem .
in this brief section , we give the proof of a maximum principle for a certain class of parabolic equations with @xmath6-laplacian diffusion .
as mentioned in the introduction , a similar result have been introduced in @xcite , but proved with different techniques .
this comparison principle is crucial in the study of the asymptotic behaviour of the general solutions of the fisher - kpp problem , see theorem [ convergencetoonefastdiffusion ] . before proceeding we need to introduce some definitions .
first of all , let @xmath639 be a positive and non - decreasing function , and consider the `` inner - sets '' @xmath640 now , for all @xmath5 , we consider the equation @xmath641 where @xmath642 is a continuous function in @xmath196 , with @xmath643 in @xmath405 , @xmath407 and @xmath30 . the next definition is given following @xcite .
see also @xcite , chapter 8 for the porous medium setting .
a nonnegative function @xmath644 is said to be a `` local strong '' super - solution of equation in @xmath645 if ( i ) @xmath646 , and @xmath647 ; \(ii ) @xmath644 satisfies @xmath648\eta + |\nabla \overline{u}|^{p-2}\nabla\overline{u}\,\nabla\eta \geq 0,\ ] ] for all test function @xmath649 , @xmath650 .
a nonnegative function @xmath534 is said to be a `` local strong '' sub - solution of equation in @xmath645 if ( i ) @xmath651 , and @xmath652 ; \(ii ) @xmath534 satisfies @xmath653\eta + |\nabla \underline{u}|^{p-2}\nabla\underline{u}\,\nabla\eta \leq 0,\ ] ] for all test function @xmath649 , @xmath650 .
[ maxprinnoncyldomains ] consider two functions @xmath644 and @xmath534 defined and continuous in @xmath196 .
assume that : ( a1 ) @xmath654 in @xmath132 .
( a2 ) @xmath655 in @xmath656 .
( a3 ) finally , assume that @xmath644 is a `` local strong '' super - solution and @xmath534 is a `` local strong '' sub - solution of equation in @xmath645 .
then @xmath657 in @xmath645 . _ proof .
_ let s fix @xmath658 . for all @xmath659 ,
we define the subset of @xmath132 @xmath660 we show that for all @xmath661 , it holds @xmath662_+\big\|_{l^1(\omega_{i , t } ) } \leq \big\|[\underline{u}(0 ) - \overline{u}(0)]_+\big\|_{l^1({\mathbb{r}}^n)},\ ] ] where @xmath663_+$ ] stands for the positive part .
consequently , we deduce the thesis thanks to assumption ( a1 ) .
we proceed with a standard argument , see for instance chapter 8 of @xcite for the porous medium equation .
let s consider a function @xmath664 such that @xmath665 and a sequence @xmath666 such that @xmath667 as @xmath668 in @xmath669 .
note that we can suppose @xmath670 thanks to assumption ( a2 ) .
hence , if @xmath671)$ ] with @xmath672 , we can take as test function @xmath673 thus , by the definition of sub- and super - solutions , it is simple to deduce @xmath674p(w_j)h + \big<\,|\nabla\underline{u}|^{p-2}\nabla\underline{u } - |\nabla\overline{u}|^{p-2}\nabla\overline{u},\nabla w_j\,\big > p'(w_j)h \,dxdt \leq 0.\ ] ] the second integral converges to @xmath675 thanks to the fact that @xmath676 for all @xmath677 and @xmath5 , see the last section of @xcite .
hence , taking the limit in the second integral we deduce @xmath674p(\underline{u}-\overline{u})h \,dxdt\leq 0,\ ] ] and , letting @xmath678_+(\cdot)$ ] , we obtain @xmath679 now , we have @xmath680_+ = \partial_t(\underline{u}-\overline{u})sign_+(\underline{u}-\overline{u}),\ ] ] and , since @xmath681_+ = s\cdot sign_+(s ) \geq 0 $ ] , @xmath682 , and @xmath407 we easily get @xmath683_+ dx\bigg ) h(t ) \,dt \leq 0 \quad \text { for all } \ ; h \in c_c^1([0,t ] ) , \ ; 0 \leq h \leq 1.\ ] ] thus , thanks to arbitrariness of @xmath684 , we deduce that @xmath685_+ dx \leq 0,\ ] ] for all @xmath661 . using assumption ( a2 ) again
, it is not difficult to deduce @xmath686_+ dx\bigg ) \leq 0,\ ] ] which implies @xmath687_+\big\|_{l^1(\omega_{i , t } ) } \leq \big\|[\underline{u}(0 ) - \overline{u}(0)]_+\big\|_{l^1(\omega_{i,0 } ) } \leq \big\|[\underline{u}(0 ) - \overline{u}(0)]_+\big\|_{l^1({\mathbb{r}}^n)},\ ] ] i.e. , the thesis .
@xmath213 [ [ remark.-1 ] ] remark .
+ + + + + + + we point out that the functions we use in the proof of theorem [ convergencetoonefastdiffusion ] satisfy the assumptions of regularity required in the statement of proposition [ maxprinnoncyldomains ] , as we have remarked in the introduction .
see also the bibliography reported in the next section .
in this section , we recall some basic facts about the existence of barenblatt solutions for the cauchy problem @xmath688 where @xmath5 . in particular , we focus on the specific initial datum @xmath689 a more complete analysis of the self - similarity of the @xmath6-laplacian equation can be found in @xcite .
we have decided to dedicate an entire appendix to this topic since solutions of problem play a main role in the proof of theorem [ convergencetoonefastdiffusion ] .
moreover , we think it facilitates the reading and gives us the occasion to present the related bibliography . before proceeding with our analysis , we need to recall some important properties about problem .
[ [ case - boldsymbolp-2 . ] ] case @xmath690 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the existence and uniqueness of solutions for the heat equation for continuous non - integrable initial has been largely studied , see tychonov @xcite and the references therein . in particular , he proved that if the initial datum satisfies @xmath691 and @xmath560 , then problem , admits a unique ( classical ) solution defined in @xmath692
. more work on this issue can be found in @xcite .
[ [ case - boldsymbolp-2.-1 ] ] case @xmath693 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + this range was studied in @xcite , by dibenedetto and herrero .
the authors showed that , under the assumptions @xmath694 for some @xmath149 and @xmath695 , there exists a unique weak solution of problem , defined in @xmath196 ( see theorem 1 , theorem 2 , and theorem 4 of @xcite ) .
furthermore , they proved that @xmath696 ( i.e. @xmath697 is a `` local strong solution '' ) and the function @xmath698 is locally hlder continuous in @xmath196 ( see also @xcite ) .
[ [ case - boldsymbol1-p-2 . ] ] case @xmath699 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the same authors ( see @xcite ) considered problem with @xmath457 and nonnegative initial data @xmath700 without any assumption on the decay at infinity of @xmath147 .
first of all , they show existence and the uniqueness of weak solutions of problem , by using the benilan - crandall regularizing effect , see @xcite
. then they posed their attention on the regularity of these solutions when the initial datum is a non - negative @xmath701-finite borel measure in @xmath132 , in the range @xmath702 .
in particular , they showed the existence and the uniqueness of a locally hlder continuous weak solution in @xmath196 , with @xmath703 ( i.e. they are `` local strong solutions '' ) , with @xmath698 locally hlder continuous in @xmath196 .
the sub - critical range @xmath704 was studied later by bonforte , iagar and vzquez in @xcite .
they proved new local smoothing effects when the initial datum is taken in @xmath705 and @xmath6 sub - critical , and special energy inequalities which are employed to show that bounded local weak solutions are indeed `` local strong solutions '' , more precisely @xmath706 .
then , thanks to the mentioned smoothing effect and known regularity theory ( @xcite and @xcite ) they found that the local strong solutions are locally hlder continuous .
[ [ barenblatt - solutions - for - problem - eqcauchyproblemplaplacianincreasinginitialdata - eqassumptionindatumpowertype . ] ] barenblatt solutions for problem ( [ eq : cauchyproblemplaplacianincreasinginitialdata ] ) , ( [ eq : assumptionindatumpowertype ] ) .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + from now on we take @xmath707 , @xmath64 .
we do not make any other assumptions on @xmath64 if @xmath422 , whilst when @xmath421 we assume @xmath420 , according to the theory developed in @xcite , and presented before . as mentioned before , the assumptions on the parameter @xmath708 guarantees the existence , the uniqueness and the hlder regularity of the solution of problem , , for all @xmath5 .
we look for solutions in _ self - similar _
form @xmath709 where @xmath710 and @xmath711 are real numbers and @xmath450 is called profile of the solution .
let @xmath712 and write @xmath713 .
it is not difficult to compute @xmath714 and , by taking @xmath715 we have @xmath716 , and so we obtain the equation of the profile @xmath717 furthermore , since guarantees that the equation in is invariant under the transformation @xmath718 , @xmath719 , we use the uniqueness of the solution of problem , to deduce @xmath720 hence , we get @xmath721 and , combining it with , we obtain the precise expressions for the self - similar exponents @xmath722 we point out that , thanks to the assumption @xmath420 when @xmath421 , we have @xmath723 for all @xmath724 , and so @xmath440 while @xmath441 . [ [ properties - of - the - barenblatt - solutions . ] ] properties of the barenblatt solutions .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we are going to prove that the profile @xmath450 of the barenblatt solutions is positive and monotone non - decreasing by applying the aleksandrov s symmetry principle .
later , we show some asymptotic properties of the profile @xmath450 .
let @xmath707 , with @xmath420 and , for all @xmath342 , consider the approximating sequence of initial data @xmath725 note that @xmath726 are both radial non - decreasing and bounded in @xmath132 .
now , consider the sequence of initial data @xmath727 and the sequence of solutions @xmath728 of problem with initial data @xmath729 , for all @xmath342 .
hence , by applying the aleksandrov s symmetry principle , we deduce that for all times @xmath730 , the solutions @xmath731 are radially non - increasing in space too .
finally , we define the sequence @xmath732 which are radially non - decreasing in space and solve problem with initial data @xmath733 , for all @xmath342 .
hence , passing to the limit as @xmath668 , we have @xmath734 and the limit @xmath735 , solution of problem with initial datum @xmath736 , inherits the same radial properties of the sequence @xmath737 .
now , we show the existence of two constants @xmath738 such that the following asymptotic bounds hold @xmath739 estimates follow directly from that fact that @xmath740 as @xmath741 .
indeed , for all fixed @xmath742 , we have that @xmath743 since , the left expression converges to 0 as @xmath741 , we deduce that @xmath744 , as @xmath745 and , from this limit , we get .
@xmath213 [ [ aleksandrovs - symmetry - principle . ] ] aleksandrov s symmetry principle .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the aleksandrov - serrin symmetry method was firstly introduced in @xcite and @xcite to show monotonicity of solutions of both ( eventually nonlinear ) elliptic and parabolic equations . here ,
following @xcite , we give a short proof for the case of the `` pure diffusive '' @xmath6-laplacian equation in , for all @xmath724 . before proceeding with the statement , we fix some notations .
let @xmath746 be an hyperplane in @xmath132 , @xmath747 and @xmath748 the two half - spaces `` generated '' by @xmath746 , and @xmath749 the reflection with respect to the hyperplane @xmath746 .
let @xmath750 be a solution of the initial - value problem with initial datum @xmath30 .
suppose that @xmath751 then , for all times @xmath661 it holds @xmath752 in particular , radial initial data generate radial solutions . _
proof_. first of all , thanks to the rotation invariance of the equation in , we can assume @xmath753 and @xmath754 .
moreover , it follows that @xmath755 solves problem in @xmath196 with initial datum @xmath756 .
now , we have @xmath757 in @xmath747 and @xmath758 in @xmath759 . hence , since the solution is continuous , we get the thesis by applying the maximum principle . note that , to be precise , we should consider solutions of the cauchy - dirichlet problem posed in the ball @xmath760 with zero boundary data .
these solutions approximate @xmath3 and @xmath761 .
consequently , we can apply the maximum principle to these approximate solutions and , finally , pass to the limit as @xmath762 .
see chapter 9 of @xcite for more details .
if @xmath147 is radial , we can apply the statement for all hyperplane @xmath746 passing through the origin of @xmath132 and deducing that for all times @xmath661 , the solution @xmath763 is radial respect with the spacial variable too .
we end the paper by discussing some open problems . moreover , we present some final comments and remarks to supplement our work .
as we have mentioned in the introduction , nonlinear evolution processes give birth to a wide variety of phenomena .
indeed we have seen that solutions of problem exhibit a travelling wave behaviour for large times when @xmath21 , i.e. @xmath22 , while infinite speed of propagation when @xmath29 .
it is natural to ask ourselves what happens in the range of parameters @xmath764 that we call `` very fast diffusion assumption '' .
however , respect to the porous medium and the @xmath6-laplacian case , we have to face the problem of lack of literature and previous works related to the doubly nonlinear operator ( in this range of parameters ) .
for this reason , in the next paragraphs we will briefly discuss what is known for the porous medium and the @xmath6-laplacian case , trying to guess what could happen in the presence of the doubly nonlinear operator .
we stress that our approach is quite _ formal _ , but can be interesting since it gives a more complete vision of the fast diffusion range , and allows us to explain what are ( or could be ) the main differences respect to the range @xmath29 .
this critical case was firstly studied by king @xcite and later in @xcite for the porous medium setting , i.e. @xmath62 and @xmath766 , with @xmath767 .
when @xmath768 it follows @xmath769 , choice of parameter which goes out of our range and we avoid it .
king studied the asymptotic behaviour of radial solutions of the pure diffusive equation @xmath770 with @xmath13 , @xmath771 , and @xmath767 .
actually , he considered a slightly different equation absorbing a factor @xmath153 in the time varible and he studied the cases @xmath18 and @xmath772 , too .
note that the choice @xmath67 corresponds to @xmath773 when @xmath62 .
in @xcite , the author described the asymptotic behaviour of radial solutions of equation , given by the formula @xmath774 where @xmath775 is a constant depending on @xmath271 and on the initial datum ( see formula ( 2.34 ) of @xcite ) .
in particular , it follows that the solutions of have spacial power like decay @xmath776 `` corrected '' by a logarithmic term for @xmath156 .
we are interested in seeing that an analogue decay holds when @xmath5 and @xmath67 in the doubly nonlinear setting .
we proceed as in @xcite , see section 2 .
_ asymptotic behaviour for large @xmath777 .
_ let s take for a moment @xmath29 .
seeking solutions @xmath566 of equation in separate form as in _ step1 _ of theorem [ theoremboundsforlevelsetsfast ] , it is simple to see that if the initial datum satisfies , then for all @xmath661 we have @xmath778 for some suitable constant @xmath109 .
note that it corresponds to fix @xmath661 and take the limit as @xmath172 in the formula of the barenblatt solutions , see subsection [ preliminariesintro ] .
now , motivated by the previous analysis , we fix @xmath779 ( in order to remain in the ranges @xmath780 and @xmath5 ) , @xmath67 , and we look for solutions of equation in the form @xmath781 for some correction function @xmath782 as @xmath172 and some constant @xmath109 . in what follows
we ask @xmath783 as @xmath172 , too .
it is simple to compute @xmath784 as @xmath156 , where we have used the fact that @xmath785 .
hence , it is simple to see that @xmath566 solves if and only if @xmath786 now , it is clear that a possible choice is @xmath787 , for some @xmath552 , and a straightforward computation shows that the previous equation is satisfied by taking @xmath788 so that for all @xmath661 , we obtain @xmath789 which generalizes the case @xmath13 and @xmath773 .
_ barenblatt solutions for @xmath67 .
_ as was observed in @xcite ( see pag .
346 ) , does not respect the self - similarity reduction of equation .
indeed , it admits `` pseudo - barenblatt '' solutions which , following the notation of , can be written in the form @xmath790 where @xmath791^{-\frac{(p-1)}{p}n } , \qquad \xi = xr(t)^{-1 } , \qquad r(t ) = e^t,\ ] ] and @xmath140 is a free parameter ( cfr . with @xcite for the case @xmath62 and with formula for the range @xmath29 ) .
we point out that the profile @xmath792 satisfies the inequalities in with @xmath67
. however , these self - similar solutions ( also called `` of type iii '' , see @xcite ) are quite different from the ones in the range @xmath29 .
in particular , they are eternal , i.e. defined for all @xmath793 and they do not converge to a dirac delta as @xmath741 ( see also @xcite ) . finally , for all fixed @xmath793 , these self - similar solutions are _ not integrable _ respect with to the spacial variable and show the spacial decay @xmath794 taking into account these facts , when @xmath67 it seems reasonable to study problem with nontrivial initial datum satisfying @xmath795 for some constant @xmath149 , and trying to extend the techniques used for the range @xmath29 , to the this critical case .
first of all , we can define @xmath796 by continuity .
thus , it is possible to repeat the proof of theorem [ convergencetozerofastdiffusion ] by using `` pseudo - barenblatt '' solutions instead of the usual ones . in this way , for all @xmath32
, we show the convergence of the solutions to 0 in the `` outer sets '' @xmath34 , as @xmath191 .
however , it is clear that the methods employed for showing proposition [ expanpansionofminimallevelsets ] can not be used in this case too .
indeed , in the range @xmath29 , this crucial proposition has been proved by constructing barriers from below with barenblatt solutions .
this has been possible since the initial datum @xmath387 in shares the same spacial decay of these self - similar solutions . in the critical case
@xmath67 , this property would not be preserved as suggests . in particular , `` pseudo - barenblatt '' solutionscannot be placed under an initial datum satisfying and so the validity of proposition [ expanpansionofminimallevelsets ] in this critical case remains an open problem . before discussing the doubly nonlinear diffusion , let us recall what is known in the porous medium setting in the corresponding range of parameters , @xmath802 , @xmath62 , and @xmath767 . consider the porous medium equation @xmath803 where @xmath804 .
it has been proved that the corresponding solution @xmath222 extinguishes in finite time ( see for instance @xcite and the references therein ) . in other words , there exists a critical `` extinction time '' @xmath805 such that @xmath806 , for all @xmath807 .
again , the cases @xmath769 , @xmath72 and @xmath769 , @xmath808 are critical and we refer to @xcite , chapters 5 to 8 . _ barenblatt solutions for @xmath68
_ so , even though there is not literature on the subject ( at least to our knowledge ) , it seems reasonable to conjecture that the doubly nonlinear diffusion shows a similar property in the range @xmath68 , with @xmath779 . in particular , also in this case we have `` pseudo - barenblatt '' solutions written in the form @xmath809^{-\frac{p-1}{\widehat{\gamma}}},\ ] ] where @xmath810 and , with a strong departure from , @xmath811^{-\frac{|\alpha|}{n}},\ ] ] where @xmath812 is fixed and stands for the `` extinction time''(cfr . with @xcite pag . 194 or @xcite for the case @xmath62 , and with formula for the range @xmath29 ) .
the existence of this kind of self - similar solutions ( also said in @xcite `` of type ii '' ) strengthen the idea that a larger class of solutions have an extinction time , i.e. they vanish in finite time .
_ application to the fisher - kkp equation . _ in section [ convergencetozerofast ] we have seen that the linearized problem @xmath194 gives a super - solution for the fisher - kpp problem with nontrivial initial datum @xmath30 , @xmath510 . again , with the change of variable @xmath197 , \quad \text{for } t \geq 0,\ ] ] we deduce that the function @xmath199 solves the problem @xmath813 now , set @xmath814 and note that @xmath815 .
now , let @xmath816 be the `` extinction time '' of the solution of problem .
thus , we deduce @xmath817 , for all @xmath818 and , if @xmath819 , it follows @xmath820 which implies @xmath821 for all @xmath822 $ ] , and so the solution @xmath3 of the fisher - kpp problem with initial datum @xmath823 extinguishes in finite time , too .
this conclusion holds under the assumption @xmath819 , which should be guaranteed if the the initial datum is `` small enough '' ( in terms of the mass ) , see @xcite chapter 5 , for the porous medium setting .
the analysis of the case in which the initial mass is infinite is an interesting open problem .
* acknowledgments . *
both authors have been partially funded by projects mtm2011 - 24696 and mtm2014 - 52240-p ( spain ) .
work partially supported by the erc advanced grant 2013 n. 339958 `` complex patterns for strongly interacting dynamical systems - compat '' .
we thank flix del teso for providing us with the numerical simulations that we display . | the famous fisher - kpp reaction diffusion model combines linear diffusion with the typical fisher - kpp reaction term , and appears in a number of relevant applications .
it is remarkable as a mathematical model since , in the case of linear diffusion , it possesses a family of travelling waves that describe the asymptotic behaviour of a wide class solutions @xmath0 of the problem posed in the real line .
the existence of propagation wave with finite speed has been confirmed in the cases of `` slow '' and `` pseudo - linear '' doubly nonlinear diffusion too , see @xcite .
we investigate here the corresponding theory with `` fast '' doubly nonlinear diffusion and we find that general solutions show a non - tw asymptotic behaviour , and exponential propagation in space for large times . finally , we prove precise bounds for the level sets of general solutions , even when we work in with spacial dimension @xmath1 . in particular , we show that location of the level sets is approximately linear for large times , when we take spatial logarithmic scale , finding a strong departure from the linear case , in which appears the famous bramson logarithmic correction . |
asteroseismology of solar - type stars has been one of the major successes of the nasa _ kepler _ mission ( gilliland et al .
the nominal mission provided data of exquisite quality for unprecedented numbers of low - mass main - sequence stars and cool subgiants .
asteroseismic detections were made in more than 600 field stars ( chaplin et al . 2011a ; 2014 ) , including a sample of _ kepler _ planet hosts ( huber et al .
these data have enabled a range of detailed asteroseismic studies ( see chaplin & miglio 2013 and references therein ) , many of which are ongoing . the nominal mission ended in 2013 may with the loss of a second of the spacecraft s four onboard reaction wheels .
this meant the spacecraft could no longer maintain three - axis stabilized pointing .
however , thanks to the ingenuity of the mission teams , _ kepler _ data collection has continued as a new ecliptic - plane mission , k2 ( howell et al .
targeting stars in the ecliptic minimizes the now unconstrained roll about the spacecraft boresight , thereby helping to compensate for the loss of full three - axis stability .
the degraded photometric performance presents particular challenges for the detection of oscillations in solar - type stars .
the oscillations are stochastically excited and intrinsically damped by near - surface convection .
while this mechanism gives rise to a rich spectrum of potentially observable overtones , having periods of the order of minutes , it also limits the modes to tiny amplitudes , typically several parts - per - million in brightness .
the opportunity to continue asteroseismic studies of solar - type stars with k2 would provide fresh data on stars in the solar neighborhood for application to both stellar and galactic chemical evolution studies .
the new fields have also led to the possibility of detecting oscillations of solar - type stars in open clusters and eclipsing binaries .
this would provide independent data to test the accuracy of asteroseismic estimates of fundamental stellar properties .
other specific targets of interest would potentially benefit from the provision of asteroseismic data , known exoplanet host stars being obvious examples .
in this paper we report the detection of oscillations in several subgiants using k2 short - cadence ( sc ) data collected during campaign1 ( c1 ) .
we describe the target selection and data analysis , and also discuss the implications of our results for future k2 campaigns .
our selected target list started with the hipparcos catalog ( van leeuwen 2007 ) .
use of these data allows us to make robust predictions for many bright , potential k2 targets in the ecliptic .
effective temperatures were estimated from the @xmath2 color data in the catalog , using the calibration of casagrande et al .
( 2010 ) , and luminosities , @xmath3 , were estimated from the parallaxes .
these calculations used reddening estimates from drimmel et al .
( 2003 ) ( negligible for many of our targets ) .
we adopted @xmath4 ( torres 2010 ) , and consistent bolometric corrections from the flower ( 1996 ) polynomials presented in torres ( 2010 ) , which use the estimated @xmath5 as input .
we also applied a cut on parallax , selecting only those stars having fractional parallax uncertainties of 15% or better .
stellar radii were then estimated from @xmath3 and @xmath5 , and approximate masses were estimated from a simple power law in @xmath3 ( which was sufficient for selecting targets ) .
the estimated hipparcos - based stellar properties were used as input to well - tested procedures ( chaplin et al .
2011b ) that enabled us to predict seismic parameters and relevant detectability metrics .
we narrowed down the sample to 23 well - characterized bright ( @xmath6 to 9 ) solar - type stars to be proposed for k2 observations .
all targets were predicted to show solar - like oscillations on timescales of the order of minutes , necessitating sc observations .
we also collected ground - based spectroscopic data on our selected c1 targets to help us check the hipparcos - based predictions , and to better understand the final yield of asteroseismic detections .
observations were made using the tres spectrograph ( frsz 2008 ) on the 1.5-m tillinghast telescope at the f. l. whipple observatory .
atmospheric parameters were derived using the stellar parameter classification pipeline ( spc ; see buchhave et al .
spc was used to match observed spectra taken at a resolution of 44000 to sets of synthetic model spectra to derive estimates of @xmath7 , @xmath8 , metallicity , and @xmath9 . in what follows
we assume that relative metal abundances [ m / h ] returned by spc are equivalent to relative iron abundances , [ fe / h ] .
table [ tab : tab1 ] contains the derived spectroscopic parameters .
there are four rapidly rotating stars in the sample , and some caution is advised regarding their estimated parameters .
overall , we found good agreement between the spectroscopic parameters and the hipparcos - based values .
table [ tab : tab1 ] also includes the hipparcos - based estimates of the luminosities . to understand the limits on k2 performance in c1 , we deliberately sampled the region of the hr diagram across which detections had been made in the nominal mission , as shown in the top panel of fig .
[ fig : fig1 ] .
the symbols denote stars that provided firm asteroseismic detections ( black ) , marginal detections ( gray ) , no detections ( open ) or no detections with a high measured @xmath9 ( red asterisks ) .
details are given below ( notably in section [ sec : det ] ) .
llcccccc 201162999 & 56884 & @xmath10 & @xmath11 & @xmath12 & @xmath13 & @xmath14 & @xmath15 + & & & & & & & + 201164031 & 56907 & @xmath16 & @xmath17 & @xmath18 & @xmath19 & @xmath20 & @xmath21 + & & & & & & & + 201182789 & 57275 & @xmath22 & @xmath23 & @xmath24 & @xmath25 & @xmath26 & @xmath27 + & & & & & & & + 201215315 & 57456 & @xmath28 & @xmath29 & @xmath30 & @xmath31 & @xmath32 & @xmath33 + & & & & & & & + 201343968 & 55379 & @xmath34 & @xmath35 & @xmath36 & @xmath37 & @xmath38 & @xmath39 + & & & & & & & + 201353392 & 55288 & @xmath40 & @xmath41 & @xmath42 & @xmath43 & @xmath44 & @xmath45 + & & & & & & & + 201367296 & 58093 & @xmath46 & @xmath47 & @xmath48 & @xmath49 & @xmath50 & @xmath51 + & & & & & & & + 201367904 & 58191 & @xmath52 & @xmath53 & @xmath24 & @xmath54 & @xmath55 & @xmath56 + & & & & & & & + 201421619 & 55438 & @xmath57 & @xmath58 & @xmath59 & @xmath60 & @xmath61 & @xmath62 + & & & & & & & + 201436411 & 56282 & @xmath63 & @xmath23 & @xmath64 & @xmath65 & @xmath66 & @xmath67 + & & & & & & & + 201592408 & 56755 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath72 & @xmath73 + & & & & & & & + 201601162 & 54675 & @xmath74 & @xmath75 & @xmath36 & @xmath76 & @xmath77 & @xmath78 + & & & & & & & + 201602813 & 55022 & @xmath79 & @xmath80 & @xmath81 & @xmath82 & @xmath83 & @xmath84 + & & & & & & & + 201614568 & 54857 & @xmath85 & @xmath41 & @xmath86 & @xmath87 & @xmath88 & @xmath89 + & & & & & & & + 201620616 & 58643 & @xmath90 & @xmath91 & @xmath92 & @xmath71 & @xmath93 & @xmath94 + & & & & & & & + 201626704 & 54541 & @xmath95 & @xmath96 & @xmath97 & @xmath98 & @xmath99 & @xmath100 + & & & & & & & + 201698809 & 55638 & @xmath101 & @xmath102 & @xmath97 & @xmath103 & @xmath104 & @xmath105 + & & & & & & & + 201729267 & 55574 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 & @xmath111 + & & & & & & & + 201733406 & 55467 & @xmath112 & @xmath107 & @xmath113 & @xmath114 & @xmath115 & @xmath116 + & & & & & & & + 201756263 & 57034 & @xmath117 & @xmath80 & @xmath118 & @xmath119 & @xmath120 & @xmath121 + & & & & & & & + 201820830 & 55778 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & @xmath126 & @xmath127 + & & & & & & & + 201853942 & 57136 & @xmath128 & @xmath129 & @xmath86 & @xmath130 & @xmath131 & @xmath132 + & & & & & & & + 201860743 & 57676 & @xmath133 & @xmath134 & @xmath118 & @xmath65 & @xmath135 & @xmath136 + & & & & & & & + [ tab : tab1 ] each target was observed by k2 for just over 82days in c1which lasted from 2014 may 30 to 2014 august 21with data collected in sc mode ( gilliland et al . 2010b ) .
we used the k2p@xmath137 pipeline ( lund et al .
2015 ) to prepare sc lightcurves for asteroseismic analysis . in brief , the pipeline took the sc target pixel data as input .
masks for all targets in a given frame were defined manually , and flux and position data were then extracted for our chosen targets of interest .
corrections were then applied to the lightcurves to mitigate the impact of changes of the target positions on the ccd .
finally , additional corrections were made using the filtering described by handberg & lund ( 2014 ) .
the lightcurves were distributed to several teams , who each attempted to detect signatures of solar - like oscillations in the power spectra of the data .
a complementary range of well - tested analysis codes was used , which had been applied extensively to data from the nominal _ kepler _ mission ( e.g. christensen - dalsgaard et al .
2008 ; huber et al .
2009 ; mosser & appourchaux 2009 ; roxburgh 2009 ; hekker et al .
2010 ; kallinger et al . 2010 ; mathur et al . 2010 ; gilliland et al . 2011 ; benomar et al . 2012
; campante 2012 ; see also comparison of methods in verner et al .
2011 ) . in cases where oscillations were detected ,
each team was asked to return estimates of the two most commonly used global or average asteroseismic parameters : @xmath138 , the average frequency spacing between consecutive overtones of the same angular degree ; and @xmath139 , the frequency at which the oscillations present their strongest observed amplitudes .
we checked the asteroseismic detection yield using the spectroscopic data .
the bottom panel of fig .
[ fig : fig1 ] provides a visual summary of these checks .
the horizontal axis shows the predicted @xmath139 for each target , made using the spectroscopic @xmath7 and @xmath140 as input .
estimates were calculated using the widely - used scaling relation ( brown et al .
1991 , kjeldsen & bedding 1995 ) : @xmath141 with the solar value @xmath142 ( see chaplin et al .
2014 ) providing the absolute calibration .
these predicted @xmath139 are given in table [ tab : tab1 ] .
the vertical axis on the bottom panel of fig .
[ fig : fig1 ] relates to our ability to detect solar - like oscillations . in the frequency domain , where the analysis of the k2 data is conducted ,
peaks due to the oscillations are superimposed on a slowly varying , broad - band background . our ability to make a detection depends on the prominence of the oscillation peaks above that underlying background .
for solar - like oscillators , the background in the frequency range occupied by the most prominent modes has contributions from granulation , shot noise , and other instrumental noise . the symbols with error bars show measured background levels , @xmath143 , in the k2 spectra at the predicted @xmath144 .
solar - like oscillators with lower frequencies of maximum variability show larger amplitude oscillations : the lower @xmath144 , the larger is the maximum amplitude ( i.e. , brightness variation , in ppm ) and hence the easier it is to make a detection for a given @xmath145 . using the results from over 600 stars from the nominal mission we can predict oscillation amplitudes as a function of @xmath139 . with the predicted amplitudes in hand
, we may estimate threshold background levels that would permit a significant detection of the oscillations .
we calculated these thresholds using the detection recipe in chaplin et al .
( 2011b ) .
the lines in the bottom panel of fig .
[ fig : fig1 ] show the threshold levels , @xmath146 , below which the observed backgrounds @xmath145 must lie to have a @xmath147 ( black dashed ) and @xmath148 ( black dotted ) chance of making a detection .
the gray curves show uncertainties on the plotted thresholds . the results returned by the mode - detection teams indicated that we had four good asteroseismic detections .
these stars are plotted in fig .
[ fig : fig1 ] using filled black symbols ; numbers are the final three digits of the associated k2 ecliptic plane catalog ( epic ) numbers ( huber et al . , in preparation ) .
two other targets showed marginal detections , and they are shown in gray .
it is important to stress that these stars all lie in the part of the bottom panel where we would expect to make detections , i.e. , where @xmath149 . even though there are other targets that show lower or similar background levels in their k2 spectra , they lie at higher predicted @xmath139 where the intrinsic oscillation amplitudes , and hence the chances of making a detection , are lower .
the observed @xmath139 were in good agreement with the spectroscopic predictions , at the level of precision of the data .
note that the spectroscopic predictions of @xmath139 have fractional uncertainties of @xmath150 , which are significantly larger than the typical uncertainties given by the asteroseismic measurements ( which are in contrast at the few - percent level ) .
[ fig : fig2 ] shows the k2 power spectra of the four stars with firm detections .
each spectrum has been smoothed with a @xmath151 boxcar filter .
sets of vertical gray solid and dashed lines are separated by the estimated average @xmath138 , and mark the spacing on which we would expect to see modes .
the power envelope of the oscillation spectrum of epic201820830 is somewhat flatter in frequency than the more classic gaussian - like envelopes shown by the other three stars .
this is a characteristic of hot f - type stars ( e.g. , arentoft et al .
the oscillations are strongly damped and this tends to wash out the visual appearance of the oscillation spectrum .
the insets show the power spectrum of the power spectrum ( psps ) of each star , computed from the region around @xmath139 .
the significant peak in each psps lies at @xmath152 , and is the detected signature of the near - regular spacing of oscillation peaks in the frequency spectrum .
these detection signatures persisted for each star when the respective c1 lightcurves were divided into two equal halves in the time domain and analyzed separately .
there are two stars in the the bottom panel of fig .
[ fig : fig1 ] that did not yield firm or even marginal detections , but might be expected to do so .
one of these stars ( epic201756263 ) is plotted in red .
there are also three other stars shown in red on the diagram .
these are all rapidly rotating solar - type stars with @xmath153 in the range @xmath154 to @xmath155 .
they are therefore presumably young and may be very active , which is known to lead to significant suppression of the oscillation amplitudes ( e.g. , garca et al .
2010 ; chaplin et al . 2011c ; campante et al .
2014 ) making it much harder to detect oscillations .
the detection recipe does not yet make allowance for suppression of the oscillation amplitudes by activity . the other star that might be expected to show a detection ( epic201602813 , plotted with an open symbol ) is noteworthy because it is by far the most metal - poor star in the sample .
this may lead to attenuation of the observed amplitudes , relative to the basic predictions ( e.g. , houdek et al .
1999 ; samadi et al .
other explanations are that the @xmath139 scaling underpredicts the @xmath139 value for this star ; or that the spectroscopic parameters are incorrect . we think that this latter explanation is unlikely . the target - selection ( hipparcos - based ) prediction for @xmath139 was in good agreement with the spectroscopic estimate .
the target is also the primary component of a single - lined spectroscopic binary ( carney et al .
2001 ) , which will be spatially unresolved in the k2 pixel data .
however , flux contamination from the secondary ( a suspected white dwarf ) is likely to be low .
it is of course also possible that residual artifacts and other problems relating to the lightcurve extraction , preparation and filtering may have prevented a good detection being made .
there are clearly persistent artifact peaks in some of the frequency spectra
. moreover , the high - frequency noise levels for all stars are not close to being shot - noise limited , unlike the nominal _ kepler _ data .
analysis of the k2 sc spectra therefore demands some degree of careful , manual scrutiny to be sure that a claimed detection is not the result of a chance combination of noise peaks . not including the active target epic201756263 , of the seven cases for which we would hope to detect oscillations , there are firm detections in four and marginal detections in two
this is a good return given the challenges posed by the k2 photometry .
we consolidated the @xmath138 and @xmath139 estimates returned by the mode detection teams to give final asteroseismic parameters ready for asteroseismic modeling .
the final parameters were those returned by an updated version of the octave pipeline ( chaplin et al .
, in preparation ; see also hekker et al .
its estimates had the smallest average deviation from the median estimates in a global comparison made over all pipelines and all stars .
uncertainties on each final parameter were given by adding ( in quadrature ) : the formal parameter uncertainty given by the chosen pipeline ; the standard deviation of the parameter estimates given by all other pipelines ; and a small contribution to account for uncertainties in the solar reference values ( which are employed in the grid - modeling ; see chaplin et al .
2014 for further discussion ) .
the final parameters are listed in table [ tab : tab2 ] . a second set of teams
were then asked to independently apply grid - based modeling to estimate fundamental properties of the four stars with firm detections .
these teams , like their mode - detection counterparts , used codes that have been applied extensively to nominal - mission _ kepler _ data ( e.g. , see : stello et al .
2009 ; basu et al .
2010 ; kallinger et al . 2010 ; quirion et al .
2010 ; gai et al .
2011 ; bazot et al .
2012 ; creevey et al . 2013 ; hekker et al .
2013 ; lundkvist et al .
2014 ; miglio et al . 2013
; serenelli et al .
2013 ; hekker & ball 2014 ; rodrigues et al . 2014 ; silva aguirre et al .
further details may also be found in chaplin et al .
( 2014 ) and pinsonneault et al .
( 2014 ) .
we tested the impact on the estimated stellar properties of using different sets of inputs , i.e. , a first set with \{@xmath138 , @xmath139 , @xmath5 , [ fe / h ] } , a second set with \{@xmath138 , @xmath5 , [ fe / h ] } , and further sets with the parallax - based luminosities also included .
we found very good agreement between the properties given by the different sets , and by the different pipelines .
results from the first two sets were consistent with the independent luminosity estimates .
using the luminosities as an additional input constraint did not have a significant impact on the results .
the parallax uncertainties which range from @xmath156 to 12%are too large to add anything useful to the seismic constraints . the fact that the second set of inputs provided results that were consistent with the estimated @xmath139 lends further confidence to the claimed detections and suggests that , at least for these data , potential bias in the @xmath139 estimates arising from spurious noise peaks is not a significant cause for concern .
table [ tab : tab2 ] gives final values for the estimated properties , using the first set of inputs .
the properties were calculated using the bespp pipeline ( serenelli et al .
2013 ) , which uses individual model frequencies to calculate model predictions of @xmath138 for comparison with the observations .
the uncertainties include a contribution from the scatter between pipelines ( following the procedure outlined above for the input seismic parameters ; see also the in - depth discussions in chaplin et al .
2014 ) .
these grid - modeling results demonstrate that k2 has returned solid results that allow asteroseismic modeling to be performed on the targets .
two cases here epic201367296 and epic201367904 will be amenable to more in - depth modelling studies since it will be possible to extract precise and robust individual frequencies of several overtones of each star .
cccccccc 201367296 & 58093 & @xmath157 & @xmath158&@xmath159 & @xmath160 & @xmath161 & @xmath162 + 201367904 & 58191 & @xmath163 & @xmath164&@xmath165 & @xmath166 & @xmath167 & @xmath168 + 201820830 & 55778 & @xmath169 & @xmath170&@xmath171 & @xmath172 & @xmath173 & @xmath174 + 201860743 & 57676 & @xmath175 & @xmath176&@xmath159 & @xmath177 & @xmath178 & @xmath179 [ tab : tab2 ]
we analysed k2 short - cadence ( sc ) data for 23 solar - type stars observed in c1 .
of the seven targets where we would hope to detect oscillations , there are firm asteroseismic detections in four cases , and marginal detections in a further two .
this represents a good return , in spite of the challenges posed by the k2 photometry . in sum
, we have a very good understanding of the asteroseismic yield .
the results put us in a good position to hone target selections for future campaigns .
current performance levels mean we can detect oscillations in sub - giants , but not in main - sequence stars .
changes to the operation of the fine - guidance sensors are expected to give significant improvements in the high - frequency performance of k2 from c3 onwards .
the high - frequency noise is currently a crucial limitation to making asteroseismic detections , in particular in main - sequence stars . with reference to the bottom panel of fig .
[ fig : fig1 ] , we note that a reduction of excess high frequency noise by a factor of two - and - a - half in amplitude ( just over six in power ) would lead to consistent detections of oscillations in main sequence stars with @xmath139 as high as @xmath1 , as well as converting marginal detection cases to ones for which detailed modeling could be performed .
the prospects are therefore very encouraging .
solar - type stars in the pleiades and hyades open clusters have already been observed by k2 in sc during c4 .
more stars will be observed in sc during c5 in the open clusters m44 and m67 .
there is also now clear potential to build up a statistical sample of solar - type field stars in the ecliptic that have good asteroseismic data , and to target specific stars of interest for asteroseismic study such as bright eclipsing binaries and known exoplanet host stars .
funding for this discovery mission is provided by nasa s science mission directorate .
the authors wish to thank the entire _ kepler _ team , without whom these results would not be possible .
we also thank all funding councils and agencies that have supported the activities of kasc working group1 . | we present the first detections by the nasa k2 mission of oscillations in solar - type stars , using short - cadence data collected during k2 campaign1 ( c1 ) .
we understand the asteroseismic detection thresholds for c1-like levels of photometric performance , and we can detect oscillations in subgiants having dominant oscillation frequencies around @xmath0 .
changes to the operation of the fine - guidance sensors are expected to give significant improvements in the high - frequency performance from c3 onwards .
a reduction in the excess high - frequency noise by a factor of two - and - a - half in amplitude would bring main - sequence stars with dominant oscillation frequencies as high as @xmath1 into play as potential asteroseismic targets for k2 . |
physical processes that occur during reionization are numerous and complex .
nevertheless , ionization of neutral gas ( hydrogen & helium ) and heating of the inter - galactic medium ( igm ) can be considered the two primary influences of radiating objects during reionization .
currently , the most promising `` direct '' probe of reionization is the redshifted 21-cm radiation emanating from neutral hydrogen during the epoch of reionization ( eor ) , which are to be measured using upcoming telescopes like lofar , mwa , paper and 21cma . the intensity of the observed 21-cm radiation depends on the ratio between the number density of electrons in the hyperfine states in the ground state of a neutral hydrogen atom .
this ratio is normally expressed in terms of the so - called 21-cm spin temperature , @xmath1 . at the onset of the formation of the first reionizing objects the spin temperature is equal to the cmb temperature since at these redshifts
the ratio between excited and ground hyperfine state electrons is completely determined by the cmb . however , as the number of ionizing sources increases , @xmath1 starts departing from @xmath7 ; slowly at the beginning , then rapidly approaching values larger than @xmath7 .
this evolution is typically ignored in most previous studies of reionization which assumes @xmath8 at all times @xcite .
recently , @xcite have relaxed this assumption on @xmath1 at the dawn of reionization and explored its impact on the brightness temperature .
they found a considerable considerable deviation from assuming @xmath9 at the beginning of reionization . towards the end of reionization
though , this assumption holds ground .
but , in order to track the evolution of @xmath1 accurately , like in @xcite , it is necessary to perform a detailed 3-d ly@xmath5 radiative transfer calculation .
the ly@xmath5 photons undergo a large number ( @xmath10 ) of scatterings even in a marginally neutral medium before it is sufficiently off line - centre to `` free stream '' .
the scattering angle after each encounter is completely random and therefore the radiative transfer is often done in a monte carlo sense @xcite to capture this random nature of ly@xmath5 scatterings .
unfortunately these monte carlo radiative transfer schemes are computationally very expensive , especially if we need to simulate large fields of view necessary to generate mock data sets for next generation radio telescopes . in order to circumvent the need to perform such computer - intensive calculations to obtain @xmath1
, we develop an algorithm along the lines of bears @xcite as an approximation . in this paper
we present an algorithm that follows the decoupling of @xmath1 from @xmath7 owing to ly@xmath5 photons , which couples the spin temperature to the colour / kinetic temperature via the wouthuysen - field effect @xcite .
collisional excitation and heating caused by secondary electrons resulting from hard x - ray radiation are also included .
the dominant source of ly@xmath5 flux is the background created by the redshifting of photons in the lyman band into ly@xmath5 .
these photons are blueward of ly@xmath5 and is injected into ly@xmath5 at some distance away from the source .
the amount of intrinsic ly@xmath5 , ionizing and `` heating '' photons is a function of the source spectral energy distribution ( sed ) .
thus the evolution of the spin temperature critically depends on the source of reionization .
different reionization sources manifest themselves by influencing the igm in markedly different ways .
for example , deficiency of hard photons in the seds of `` first stars '' , limit the extent to which they heat the igm @xcite , while miniquasars ( or miniqsos , characterized by central black hole masses less than a million solar ) , abundant in x - ray photons , cause considerable heating @xcite .
ionization profiles similarly have their characteristic source - dependent behavior .
although the question on which sources did the bulk of the reionization is up for debate , it is conceivable from observations of the local universe up to redshifts around 6.5 , that sources of reionization could have been a mixture of both stellar and quasar kinds ( their respective roles again are uncertain ) .
implementing radiative transfer that include both ionizing and hard x - ray photons has been difficult and as a result most 3-d radiative transfer schemes restrict themselves to ionization due to stars @xcite . in @xcite , a `` semi '' hybrid model of stars and miniqsos , like
the one hinted above , has been used albeit in sequential order instead of a simultaneous implementation .
that is , pre - ionization due to miniqsos was invoked between @xmath11 , after which , stars reionize the universe at redshift 7 .
we in this paper would like to address the issue of simulating the propagation of both the uv and hard x - ray photons , exactly in 1-d and as approximation in 3-d .
the focus of this paper is therefore to introduce the algorithm that is used to implement igm heating in bears along with the procedure to estimate the spin temperature of the igm . as an application of this technique
we explore the effects of heating due to miniqsos , stars and , for the first time , a mixed `` hybrid population '' .
subsequently , we provide quantitative and qualitative analysis of the differences in the 21-cm eor signal with and without the usual assumption of @xmath1 being always decoupled from @xmath7 . the paper is organized as follows ; [ sec : simulations ] describes briefly the n - body and 1-d radiative transfer codes used . in
[ sec : bearheat ] we describe the adaptation of bears to include @xmath6 , followed by the calculation of the @xmath1 and @xmath3 within the simulation box .
bears is then applied to three different scenarios of reionization in [ sec : applications ] , _
viz_. , ( 1 ) the primary source being stars , ( 2 ) miniqsos and ( 3 ) a hybrid population of stars and miniqsos .
subsequently , observational cubes of @xmath12 are generated for each of these scenarios and its properties discussed . in
[ sec : stats ] we provide statistics on the simulated boxes and interpret the finding mainly from the point of view of the differences in @xmath3 with and without the usual assumption that @xmath1 is decoupled from @xmath13 .
we also compare our work to that of @xcite in the same section .
conclusions and discussions of the results are presented in [ sec : conclusion ] , along with a mention of a few topics that can be addressed using the data set simulated in this paper .
in this section we briefly mention the dark - matter - only n - body simulations employed , the 1-d radiative transfer code developed by @xcite and provide an overview of the bears algorithm designed in @xcite .
the dark matter only n - body simulations are the same as in @xcite and was performed by joop schaye and andreas pawlik at the leiden observatory .
the n - body / treepm / sph code gadget @xcite was used to perform a dark matter ( dm ) cosmological simulation containing @xmath14 particles in a box of size @xmath15 with each dm particle having a mass of @xmath16 .
initial conditions for the particle s position and velocity were obtained from glass - like initial conditions using cmbfast ( version 4.1 ; @xcite ) and the zeldovich approximation was used to linearly evolve the particles down to redshift z = 127 .
a flat @xmath17cdm universe is assumed with the set of cosmological parameters @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 and @xmath23 @xcite .
snapshots were written out at @xmath24 equally spaced redshifts between @xmath25 and @xmath26 .
halos were identified using the friends - of - friends algorithm ( @xcite ) , with linking length @xmath27 .
the smallest mass haloes we can resolve are a few times @xmath28 . the density field was smoothed on the mesh with a gaussian kernel with standard deviation @xmath29 comoving @xmath30 ( 2 grid cells ) . the implementation of the 1-d radiative transfer code is modular and hence straightforward to include different spectra corresponding to different ionizing sources .
our 1-d code can handle x - ray photons , the secondary ionization and heating it causes , and as a result perform well both for high ( quasars / miniqsos ) and low energy ( stars ) photons .
further details of the code can be found in @xcite .
the parameter space simulated by the 1-d radiative transfer scheme @xcite include ; 1 ) different seds ( stars or miniqsos with different power - law indices ) , 2 ) duration of evolution ( depending on the life - time of the source ) , 3 ) redshifts at which these sources turn on and 4 ) clumping factor or over - density around a given source .
information on the ionization and temperature profiles from these simulations were cataloged to be used later as in @xcite .
density profiles around sources are assumed to be homogeneous although the code could be initialized using any given profile .
bears is a fast algorithm that simulates the underlying cosmological 21-cm signal from the eor .
it is implemented by using the nbody / sph simulation in conjunction with a 1-d radiative transfer code ; both discussed in the previous sections .
the basic steps of the algorithm are as follows : first , a catalogue of 1-d ionization profiles of all atomic hydrogen and helium species and the temperature profile that surround the source is calculated for different types of ionizing sources with varying masses , luminosities at different redshifts .
subsequently , photon rates emanating from dark matter haloes , identified in the n - body simulation , are calculated semi - analytically .
finally , given the spectrum , luminosity and the density around the source , a spherical ionization bubble is embedded around the source , whose radial profile is selected from the catalogue as generated above . in case of overlap between ionization bubbles , the excess ionized atoms in the overlap area
is redistributed at the edge of the bubbles . for more details
see thomas et al .
the outputs are data cubes ( 2-d slices along the frequency / redshift direction ) of density ( @xmath31 ) , radial velocity ( @xmath32 ) and hydrogen and helium fractions ( @xmath33 ) .
each data cube consists of a large number of slices each representing a certain redshift between 6 and 11.5 . because these slices are produced to simulate a mock dataset for radio - interferometric experiments , they are uniformly spaced in frequency ( therefore , not uniform in redshift ) .
thus , the frequency resolution of the instrument dictates the scales over which structures in the universe are averaged / smoothed along the redshift direction . the relation between frequency and redshift space z
is given by:@xmath34 , where @xmath35 mhz is the rest frequency that corresponds to the 21-cm line .
in @xcite the bears algorithm was introduced ; a special - purpose 3-d radiative transfer scheme used to simulate cosmological eor signals . in that paper
we detailed the philosophy behind the code and its implementation to incorporate ionization due to a variety of sources . in this section ,
we extend the features of bears to include heating ( on which the brightness temperature is sensitively dependent ) due to hard photons within the same framework . the algorithm to implement heating begins in a manner similar to that for ionization .
the @xmath6-profile from the 1-d radiative transfer code is used to embed a spherically symmetric `` @xmath6 bubble '' at the locations of the centre - of - mass of dark matter haloes .
the luminosity of the source is a function of the halo mass .
the problem embedding a temperature profile in the simulation box is that , the radius of the bubble is large ( @xmath36 mpc ) and results in extensive overlap , even at high redshifts ; a problem encountered for ionization bubbles only towards the end of reionization . in a quasar
dominated part of the universe this is particularly severe because of its large sphere - of - influence in heating the igm .
overlap in the spheres of ionized regions was treated by redistributing the excess ionizing photons in the overlapped region uniformly around the overlapping spheres @xcite .
treating overlapped zones directly in terms of temperatures makes it difficult to come up with any `` conserved quantity '' . to elaborate , consider a pre - ionized zone in the simulation box close to the radiating source .
although at close proximity to the source , this region is not heated efficiently because of the absence of neutral hydrogen to capture the photons . on the other hand , the same source can dump more energy into an initially neutral zone that maybe far away from it .
thus , embedding the @xmath6-profile directly from the catalog ( which assumes uniform igm density ) is not appropriate .
here we adopt a different approach where we estimate the total energy deposited in every region of the simulation box and invoke the conservation of energy deposited by the sources in the overlapped regions .
we integrate the contribution to the energy budget , at a given location , from all sources whose sphere - of - influence extents to that location .
this total energy is then redistributed equally to all the contributing sources , i.e. , the energy output of each source is modified to account for the overlapping . to illustrate , consider @xmath37 sources that overlap at a particular location @xmath38 and the total energy estimated at that location to be @xmath39 , where @xmath40 is the energy deposited by the @xmath41 source at location @xmath38 .
the fraction of `` excess '' energy attributed to each of the @xmath37 sources is @xmath42 .
this energy @xmath43 is then added to the total energy from the source to estimate a new normalization constant for the sed of the source .
for example in case of a power - law - type source , the sed includes a normalization term that is determined by the total energy of the source , @xmath44 .
however , in the overlap case the total energy of the source is replaced by @xmath45 .
now a `` new '' bubble of kinetic temperature is embedded whose profile depends on a source whose output energy has gotten a @xmath43 boost . in this manner @xmath46
is estimated at every location in the simulation box . from this point on we need to estimate the coefficients that couple @xmath6 to @xmath1 that will eventually lead to the estimation of @xmath12 .
it is worthwhile also to note that the results are robust to variation in the exact algorithm used to treat the overlapped region .
for example , if we choose to sum the energies deposited directly in a given region or choose max(@xmath47 ) , where @xmath48 are energies deposited by sources 1 ... n respectively , the final results are very similar .
this is obviously an approximation which assumes that the efficiency of the photons in heating the igm does not depend on their frequency . for the case of power - law sources where most of the energy comes from hard photons
this is probably a reasonable approximation .
however , in the case of blackbody sources this is not a very good assumption because evidently the heating is done by a very small fraction of photons .
but in these cases the overlap problem is not as severe as for the power - law sources since the extent of the heating is much more limited . in summary , despite the approximate nature of this approach , it provides an efficient and reasonable resolution for the issue of the overlapping of temperature bubbles .
we note that self consistent treatment of this problem requires taking into account the modification of each of sources sed in a different manner depending on the amplitude of its sed .
obviously this is computationally very expensive .
spin temperature can be thought of as a short hand notation to represent the level population of the hyperfine states of the ground level of a hydrogen atom .
depending on the physical processes and background radiation that dominate a medium , @xmath1 is either coupled to the background cmb temperature or to the kinetic temperature of the hydrogen gas in the medium .
formally @xcite derived @xmath49 as a weighted sum of @xmath50 and @xmath51 , as ; @xmath52 where @xmath53 and @xmath54 are parameters that reflect the coupling of @xmath1 to @xmath6 via ly@xmath5 excitation and collisions respectively .
the efficiency of ly@xmath5-coupling dominates over that of collisions , especially further away from the source @xcite . in our treatment of calculating @xmath49 , we estimate both @xmath53 and @xmath54 .
the coefficient @xmath54 is a function of @xmath50 and the ionized fraction of the medium , @xmath55 .
these informations on ionization and heating are available from the prescription outlined in the previous section . while simulating the 1-d profiles we simultaneously calculate @xmath56 as ; @xmath57 here
, @xmath58 is the ly@xmath5 flux density at distance @xmath59 from the source .
for miniqsos with high energy photons , ly@xmath5 coupling is mainly caused by collisional excitation due to secondary electrons @xcite .
this process is accounted for by the following integral , @xmath60 where @xmath61 is the number of photons with energy e at radius r , and time t per unit area @xcite .
this number is obtained directly from the 1-d radiative transfer by taking into account the the absorption due to the optical depth along the line - of - sight to the source at a distance r. the 1-d profiles generated are catalouged also as a function of time .
therefore the profiles that are embedded within the simulation are chosen to obey causality .
the ionization cross - section of neutral hydrogen in the ground state is given by @xmath62 @xmath63 is the oscillator strength of the ly@xmath5 transition , @xmath64 & @xmath65 are the electron s charge and mass , respectively , @xmath66 is the fraction of the absorbed photon energy that goes into excitation @xcite .
note here that the fitting formula is a function of both ionization fraction and energy as given in the appendix of @xcite . for the case of stars , the dominant source of ly@xmath5 flux results from the redshifting of the source spectrum blueward of ly@xmath5 into the resonant line at different distances from the source .
frequency @xmath67 at the redshift of emission ( or the source ) @xmath68 is redshifted into @xmath69 at redshift @xmath70 such that ; @xmath71 thus , at every radius @xmath59 away from the source the ly@xmath5 flux @xmath72 . where @xmath73 [ ev ] .
now , instead of embedding a sphere of the @xmath1 , as in the case of @xmath6 , we embed instead the a bubble of ly@xmath5 flux , @xmath58 , as estimated from eq .
[ eq : lyalphaflux ] .
since @xmath58 is basically the number of ly@xmath5 photons at a given location , the overlap of two `` @xmath74 bubbles '' implies that the photons and hence the @xmath74 has to be added , i.e. , at a given spatial ( pixel ) location , @xmath75 , @xmath76 , @xmath77 and time @xmath78 , the total ly@xmath5 flux is given by ; @xmath79 where @xmath80 is the ly@xmath5 flux contributed by @xmath41 source at the pixel location in the box , @xmath75 , @xmath76 , @xmath77 and time @xmath78 .
equipped with the quantities @xmath81 , and @xmath82 , we can calculate @xmath53 as in eq .
[ eq : yalpha ] , and subsequently the spin temperature through eq .
[ eq : tspin ] . now all terms required for the calculation of @xmath3 , as in eq . [ eq : dtb ] , are obtained .
@xmath83 \left(\frac{\omega_b h^2}{0.0223}\right ) \left[\left(\frac{1 + z}{10}\right)\left(\frac{0.24}{\omega_m}\right)\right]^{1/2 } & & , \label{eq : dtb}\end{aligned}\ ] ]
in this section we apply bears with its extended feature of including heating of the igm , to three different scenarios of reionization .
the models described in this section are not `` template '' reionization scenarios by any measure nor is any particular model favoured w.r.t the other .
in fact , these models may be far from reality and only serve as examples of the potential of bears to provide a reasonable estimate of the 21-cm brightness temperature for widely different scenarios of reionization .
the first scenario explored using bears is the case in which the sources of reionization are _ only _ stars . in this section
we describe the model used to describe the stellar component and the prescription adopted to embed these stars into haloes of dark matter identified in an n - body simulation .
most stars , to first order , behave as blackbodies at a given temperature , although the detailed features in the sed depends on more complex physical processes , age , metallicity and mass of the star .
this blackbody nature imprints characteristic signatures on the igm heating and ionization patterns @xcite .
@xcite showed that the temperature of the star only weakly depends on its mass .
the blackbody temperature of the stars in our simulation was thus fixed at @xmath84 to perform the radiative transfer .
we sample the parametre space of redshifts ( 12 to 6 ) , density profiles around the source and masses ( 10 to 1000 @xmath85 ) .
the total luminosity is normalized depending on the mass of the star according to table . 3 in @xcite . for blackbody with a temperature around @xmath86 , similar to our adopted value , the spectrum peaks between @xmath87 to @xmath88 ( wien s displacement law ) .
thus , from the form of the blackbody spectrum we can apriori expect , in the case of stars , that the ionization induced by these objects will be high , but the heating they cause will not be substantial given the exponential cutoff of the radiation towards higher frequencies .
now , following the prescription in @xcite we associate stellar spectra with dark matter halos using the following procedure . the global star formation rate density @xmath89 as a function of redshift
was calculated using the empirical fit ; @xmath90}{\beta - \alpha + \alpha\exp\left[\beta(z - z_m)\right]}~~[{\rm m}_\odot~{\rm yr}^{-1}~{\rm mpc}^{-3 } ] , \label{eq : starformrate}\ ] ] where @xmath91 , @xmath92 , @xmath93 marks a break redshift , and @xmath94 $ ] fixes the overall normalisation @xcite .
now , if @xmath95 is the time - interval between two outputs in years , the total mass density of stars formed is @xmath96.\ ] ] notice that this approximation is valid only if the typical lifetime of the star is much smaller than @xmath95 , which indeed is the case in our model because the stellar mass we input ( 100 @xmath85 ) has a lifetime of about a few myrs @xcite , compared to a typical time difference of few tens of myrs between simulation snapshots .
therefore , the total mass in stars within the simulation box is @xmath97 $ ] .
this mass in stars is then distributed among the halos weighted by their mass ; @xmath98 where @xmath99 is the mass in stars in the dark matter `` halo '' , @xmath100 the mass of the halo and @xmath101 the total mass in halos within the simulation box .
we then assume that all of the mass in stars is distributed in stars of 100 @xmath85 , which implies that the number of stars in the halo is @xmath102 .
the luminosity of a 100 @xmath85 star is obtained from fig . 1 of @xcite
, assuming zero metallicity .
the luminosity of 100 @xmath85 star thus derived is in the range of @xmath103 @xmath104 and this value is multiplied by @xmath105 to get the total luminosity emanating from the `` halo '' and the radiative transfer is done by normalizing the blackbody spectrum to this value .
the escape fractions of ionizing photons from early galaxies is assumed to be 10% .
results of the evolution of the @xmath6 , @xmath1 and @xmath3 of the igm , when the sources of reionization comprise only of stars is shown in figs [ fig : starheat ] , [ fig : starspin ] and [ fig : starbright ] , respectively .
corresponding to redshifts 10 , 8 , 7 and 6 are plotted ( _ panels left to right and top to bottom , respectively _ ) .
temperature are indicated in log t [ k ] on the color bar above .
the extent of heating , both in amplitude ( maximum around @xmath106 k towards the center ) and spatial extent ( @xmath107 100 kpc ) , is extremely small .
basically , only the central part ( ionized region ) is at a high temperature and falls sharply during the transition into the neutral igm.,scaledwidth=50.0% ] figure [ fig : starheat ] shows the kinetic temperature for stellar sources at redshifts 10 , 8 , 7 and 6 at which their ionized fractions are 0.12 , 0.5 , 0.83 , and 0.98 respectively .
the blackbody type stellar spectra do not have sufficient high energy photons to heat the igm substantially far away from the source .
thus we see compact regions ( bubbles ) of high temperature in the immediate vicinity of the source . in the inner parts , where the ionized fraction is high ( @xmath108 ) , the temperatures are of the order of @xmath106 k and drops sharply in the transition zone to neutral igm .
the ionized region is restricted to about 100 kpc in physical coordinates around a halo @xcite . .
collisional coupling is efficient to couple the @xmath6 to @xmath1 at the centre but the figures also indicate that there is sufficient @xmath74 to couple @xmath6 to @xmath1 well beyond the ionized region , where @xmath109 .
the color bar indicates temperature on a log t [ k ] scale.,scaledwidth=50.0% ] figure .
[ fig : starspin ] shows @xmath1 at four different redshifts .
collisional coupling ( @xmath54 ) is important close to the radiating source ( @xmath110 ) .
the primary source of coupling between @xmath49 and @xmath50 is the redshifting of photons blueward of the ly@xmath5 line into the ly@xmath5 frequency .
it has to be emphasized here that the ly@xmath5 photons are produced _ only _ due to the source and secondary processes .
thus we see that stellar sources are efficient in building up a significant background of @xmath74 far away from the source which is sufficient to couple @xmath49 to @xmath50 during reionization @xcite . as an example , in fig .
[ fig : histcompinf ] we show the reionization history ( @xmath3 ) assuming a high background @xmath74 , in other words , assuming perfect @xmath1@xmath6 coupling . .
the temperature is now contoured on a linear scale with values indicated on the colorbar . in these slices ,
@xmath3 , starting from zero within ionized regions attains large negative value outside because stars are not efficient at heating the igm beyond their ionized bubble and hence can not render the igm visible in emission .
early on , ( top - left ) at a redshift of 10 , there are only a few sources and the heating induced colliosional coupling or the @xmath74 is not high enough to render the universe visible , but the photons blueward of ly@xmath5 can reach far outside the ionized region where they are injected into ly@xmath5 frequency causing substantial coupling to the kinetic temperature which is much below the cmb . and , at later times , the ionized bubbles overlap significantly driving the brightness temperature to zero.,scaledwidth=50.0% ] in fig .
[ fig : starbright ] @xmath3 corresponding to the same redshifts as in the previous figures is shown . in this particular model @xmath3 values
are not very high ( @xmath111mk ) . at early epochs , ( top - left ) at @xmath112 , there are only a few sources and the heating of the igm or the secondary @xmath74 is not high enough to cause substantial differential brightness temperatures . at later times ,
the ionized bubbles overlap significantly driving the brightness temperature to zero .
thus there is only a small portion of the universe ( in volume ) that has sufficient ly@xmath5 coupling and neutral hydrogen density to cause large differential brightness temperatures .
this behaviour is specific to the model assumed and will be significantly different if the parameters such as the star - formation rate , sed and escape fractions are altered . in this section
we will consider heating due to high energy x - ray photons emanating from power - law type sources ( eg . , miniqsos ) .
because of their large mean free path ( @xmath113 ) it has been difficult to incorporate the effect of heating by power - law sources self - consistently in a 3-d radiative transfer simulation .
observations reveal the energy spectrum of quasars - type sources typically follow a power - law of the form @xmath114 @xcite . here
we assume @xmath115 and thus the sed of the miniqsos is given by : @xmath116 with the normalization constant , @xmath117 where @xmath44 is the total energy output of the miniqso within the energy range , @xmath118 .
any complex , multi - slope spectral templates as in @xcite can be adopted .
the number of plausible seds that can be considered are numerous and serves as a reminder of the extent of `` unexplored parameter space '' even while considering the case of miniqsos alone , and argues for an extremely quick rt code like bears .
the miniqsos are assumed to accrete at a constant fraction @xmath119 ( normally 10% ) of the eddington rate .
therefore , the luminosity is given by : @xmath120 the luminosity derived from the equation above is used to normalize the relation in eq .
[ eq : spectrum ] according to eq .
[ eq : normalization ] .
the energy range considered is between 10.4 @xmath121 and 10 @xmath122 .
simulations were carried out for a range of masses between @xmath123 and @xmath124 . to justify comparing the case of stars and miniqsos
we argue that , although the number of photons at different energies is a function of the total luminosity and spectral index in the case of miniqsos , if we assume that all photons from the miniqso are at the hydrogen ionization threshold , then the number of ionizing photons obtained , for the mass range fixed above , is about @xmath125 ; the same order of magnitude as the number of ionizing photons in the case of stars and it also matches the numbers being employed for simulations by various other authors like @xcite and @xcite .
the miniqsos are embedded into an n - body output following the prescription in @xcite .
dark - matter haloes are identified using the friends - of - friends ( fof ) algorithm and masses of black holes assigned according to , @xmath126 where the factor @xmath127 reflects the magorrian relation @xcite between the halo mass ( @xmath128 ) and black hole mass ( @xmath129 ) multiplied with the radiative efficiency assumed to be 10% , and the fraction @xmath130 gives the baryon ratio @xcite .
contour plots of @xmath6 , @xmath1 and and @xmath12 for the case of _ only _ miniqsos being the sources of reionization , at four different redshifts , are plotted in figs .
[ fig : qsoheat ] , [ fig : qsospin ] and [ fig : qsobright ] , respectively .
the high energies of the x - ray photons substantially affects the photo - ionization and thermal energy balance of the igm . with temperatures indicated on a log t [ k ] scale .
here we see the striking difference between miniqsos and stars in the extent of heating .
the amplitude is about @xmath131 k towards the center and spatial extent of the heating is about a few mpcs .
already at redshift 10 ( top - left ) the average temperature is of the order @xmath132 k.,scaledwidth=50.0% ] the marked difference between power - law type and stellar sources , in the extent to which they heat the igm , is evident in fig .
[ fig : qsoheat ] .
the temperatures are shown at redshifts 10 , 8 , 7 and 6 with ionized fractions 0.01 , 0.54 , 0.96 , 0.999 respectively . in the inner parts of the `` @xmath6 bubble '' the amplitude of @xmath6 is about @xmath131
k. the secondary electrons created by high energy x - ray radiation thermalizes to these high temperature values . also contributing to extremely high temperatures in the very inner parts of the bubble
is compton heating @xcite .
the mean free path of x - rays are extremely large resulting in a spatial extent of heating that reaches many mpcs .
thus , we see that already at redshift 10 ( top - left ) the average temperature of the igm is of the order of @xmath132 k and overlap of the temperature bubbles occur much before that of the ionization bubble
. this example might be unrealistic , because the soft x - ray background constraint @xcite is not strictly observed , but again this serves just as an illustrative example . with the temperatures indicated on a log t [ k ] scale .
spin temperatures show an interesting ring - like behaviour ( easily visible in the top - left panel ) .
the reason for this is discussed in the text . , scaledwidth=50.0% ] figure .
[ fig : qsospin ] shows the evolution of @xmath1 as a function of redshift . owing to a high secondary @xmath74 produced due to x - ray photons ,
@xmath1 is coupled to @xmath6 for the large part .
the snapshots of @xmath1 show a peculiar ring - like behaviour ( refer . fig .
[ fig : qsospin ] ) .
this is because towards the central part of the bubble around the source , the igm is highly ionized and @xmath133 , obtains an extremely low value . on
the other hand @xmath74 progressively gets lower away from the source , regions ( @xmath1345mpc ) .
therefore only a relatively narrow zone ( few hundred kpcs ) in between the ionizing front and regions further out has @xmath74 high enough to decouple @xmath1 from @xmath7 . .
note here that the brightness temperature ( mk ) in plotted on a linear scale with values as indicated on the color bar .
we see again from the top - left panel that although the number of sources in the field are few , they were efficient in both increasing @xmath6 dramatically and providing sufficient @xmath74 to `` light - up '' the universe in 21-cm around them .
the spheres in the top - left panel have @xmath3 of zero because they are highly ionized . , scaledwidth=50.0% ] figure .
[ fig : qsobright ] shows @xmath3 for redshifts 10 , 9 , 8 and 7 in panels top - left to bottom - right , respectively .
the @xmath3 plotted in this figure are high enough to be detectable by upcoming telescope s like lofar and mwa , whose sensitivities should reach @xmath135 @xcite .
we see in the top - left panel of this figure that even though the number of sources in the field is small , they are efficient in both increasing @xmath6 dramatically and providing sufficient @xmath74 to `` light - up '' the universe in the 21-cm differential brightness temperature .
the @xmath3 inside the spheres are zero because they are highly ionized ( @xmath136 ) . in the observable universe
we know that stars and quasars do co - exist .
although , there are indications of the quasar number - density peaking around redshift two @xcite and declining thereof , there are also measurements by @xcite of high - mass ( @xmath137 ) quasars at @xmath138 .
this allows us to envisage a scenario of reionization in which stars and quasars ( or miniqsos ) contributed to the ionization and heating of the igm in a combined fashion . as a final example - application of bears , we present our model of a `` hybrid '' star - miniqso sed and examine its results on the ionization and heating of the igm .
the uncertainty regarding the properties and distributions of objects during the dark ages allows us to come up with strikingly different ways of incorporating the co - existence of stars and miniqsos .
for example , consider that there are @xmath37 different haloes within our simulation box .
one approach would be to suppose that each of these haloes host both a miniqso and stars , whose masses / luminosity is derived according to the previous two examples . the other approach would be to place one or two massive miniqsos ( almost quasars ) , still conforming to the black hole mass density predictions of @xcite , and the rest of the haloes in the simulation box with populations of stars .
for this example , given the minimum mass halo our simulation can resolve , we have chosen to adopt the former approach , i.e. , a miniqso and stars in every halo . in our approach , given the halo mass we embed them with stellar sources as described in
[ sec : stellar ] .
then the black hole mass is set as @xmath127 times the stellar mass .
radiative transfer is performed using this hybrid sed , i.e. , a superposition of power - law type miniqso sed and black body spectrum normalized according to the criterion in the previous sections .
the results of ionization and heating are presented below .
the hybrid model proposed does have large amounts of high - energy x - ray photons from miniqsos that influence the temperature and ionization of the igm prior to full reionization .
the x - rays photons have larger mean free paths than their uv counterparts , and their secondary electrons have significant influence on the thermal and ionization balance .
the uv photons on the other hand are efficient at keeping the igm in the vicinity of the halo highly ionized . for the redshifts 10 , 8 , 7 and 6 we plot the ionization due to the hybrid model in fig .
[ fig : combiion ] and the ionized fraction at these redshifts are 0.01 , 0.38 , 0.86 and 0.995 respectively . .
shown in these slices are the contour levels of neutral hydrogen in a logarithmic scale .
we see here that for the hybrid model we have assumed the ionization proceeds quickly and the universe is substantially ionized by redshift 6 ( refer fig .
[ fig : histcomp ] ) for the reionization history .
the extent of ionizations interpolates between the cases of black holes and stars.,scaledwidth=50.0% ] for this hybrid model the number of uv photons was not high enough to ionize the universe completely by redshift 6 ( the ionized fraction reaches 0.95 ) .
full reionization thus could be achieved either by increasing the mass of the black hole and/or by changing the ratio between the black hole and stellar masses . the hybrid model considered here yielded @xmath6 , @xmath1 and @xmath3 as in figs .
[ fig : combiheat ] , fig .
[ fig : combispin ] and fig .
[ fig : combibright ] , repectively .
, with temperatures indicated on a log t [ k ] scale .
the heating pattern is peculiar in that the temperature shows a much `` cuspier '' behaviour towards the center .
the reason is that the stellar component of the source ionized the very central part and the x - ray radiation can thus heat the central part where to even higher temperature . on the other hand the large extent of the heating
is attributed to the power - law component of the sed.,scaledwidth=50.0% ] the hybrid model produces heating patterns that interpolates between the results for the previous two cases and thus is qualitatively different ( fig .
[ fig : combiheat ] ) .
the power - law components take over the role of heating an extended region and the stellar component maintains a very high temperature in the central parts . .
spin temperatures for the case of the hybrid model again shows the ring like behaviour as in the case of power - law sources .
temperatures are indicated on a log t [ k ] scale.,scaledwidth=50.0% ] the spin temperature ( fig .
[ fig : combispin ] ) again shows the characteristic ring - like structure around the source . the amount of secondary ly@xmath5 radiation produced due to the miniqsos results in an efficient coupling of the spin temperature and the kinetic temperature . . the temperatures ( mk )
are indicated on a linear scale in the color bar above .
the large extent of heating and sufficient @xmath74 ensures that the igm well beyond the location of the source is rendered visible in 21-cm emission .
, scaledwidth=50.0% ] the brightness temperature for the hybrid model is plotted in fig .
[ fig : combibright ] .
the secondary @xmath74 produced due to the power - law component efficiently couples the high ambient @xmath6 of the igm to @xmath1 and this enables a large fraction of the universe to be visible in the brightness temperature albeit with a few sources ( top - left panel of the figure ) .
the brightness temperature does not go to zero at redshift six because , in the model we have assumed , the universe is not entirely ionized by then ( for recent papers on this issue see @xcite ) . in this section
we compare the reionization histories for the three scenarios explored above . to create a contigious observational cube or `` frequency cube '' ( right ascension ( ra ) @xmath139 declination ( dec ) @xmath139 redshift ) , we adopt the procedure described in @xcite .
the procedure involves the stacking of ra and dec slices , taken from individual snapshots at different redshifts ( or frequency ) , interpolated smoothly to create a reionization history .
this data cube is then convolved with the point spread function of the lofar telescope to simulated the mock data cube of the redshifted 21-cm signal as seen by lofar . for further details on creating
this cube refer to @xcite . as expected the signatures ( both visually and in the r.m.s ) of the three scenarios ( fig .
[ fig : histcomp ] ) are markedly different . in the miniqso - only scenario ,
reionization proceeds extremely quickly and the universe is almost completely ( @xmath140 ) reionized by around redshift 7 .
the case in which stars are the only source see reionization end at a redshift of 6 .
also in this case , compared to the previous one , reionization proceeds in a rather gradual manner . the hybrid model , as explained previously , interpolates between the previous two scenarios . the @xmath3 in fig .
[ fig : histcomp ] is calculated based on the effectivness of @xmath74 , produced by the source , to decouple the cmb temperature ( @xmath7 ) from the spin temperature ( @xmath1 ) .
this flux , both in spatial extent and amplitude , is obviously much larger in the case of miniqsos compared to that of stars resulting in a markedly higher brightness temperatures in both the miniqso - only and hybrid models when compared to that of the stars .
however , we know that stars themselves produce ly@xmath5 radiation in their spectrum . apart from providing sufficient ly@xmath5 flux to their immediate surrounding ,
this radiation builds up , as the universe evolves , into a strong background @xmath74 @xcite , potentially filling the universe with sufficient ly@xmath5 photons to couple the spin temperature to the kinetic temperature everywhere .
thus , we plot in fig .
[ fig : histcompinf ] the same set of reionization histories as in fig .
[ fig : histcomp ] but now assuming that @xmath1 coupled to @xmath6 .
clearly the `` cold '' regions in the universe show up as regions with large negative brightness temperature . this assumption though is not strictly applicable towards the beginning of reionization . in [ sec : stats ] we quantify the differences in the brightness temperature evolution when this assumption is made .
it has to be noted that the results we are discussing here are extremely model dependent and any changes to the parametres can influence the results significantly .
this on the other is the demonstration of the capability and the need for a `` bears like '' algorithm to span the enormous parametre space of the astrophysical unknowns in reionization studies . to test the feasibiliy of 21-cm telescopes ( specifically lofar ) in distinguishing observational signatures of different sources of reionization , we need to propagate the cosmological 21-cm maps generated in [ subsec : reionhist ] using lofar s telescope response .
the point spread function ( psf ) of the lofar array was constructed according to the latest configuration of the antenna layouts for the lofar - eor experiment .
the lofar telescope will consist of up to 48 stations in the netherlands of which approximately 22 will be located in the core region @xcite .
the core marks an area of 1.7 @xmath139 2.3 kilometres and is essentially the most important part of the telescope for an eor experiment .
the core station consists of two sets of antennae , the low band ( lba ; 30 - 90 mhz ) and the high band antenna ( hba ; 110 - 240 mhz ) .
the hba antennae in the core stations is further split into two `` half - stations '' of half the collecting area ( 35-metre diameter ) , separated by @xmath141 130 metres .
this split further improves the uv - coverage . for details on the antenna layout and the synthesis of the antenna beam pattern
refer to upcoming paper by @xcite . in its current configuration
, the resolution of the lofar - core is expected to be around 3 arcmins .
thus , as an example at redshift 10 , all scales below @xmath141 800 kpc will be filtered out .
[ fig : histcompafterconv ] shows this effect for the reionization histories corresponding to fig .
[ fig : histcomp ] .
the corresponding changes in the r.m.s of the brightness temperature is also plotted .
we see that , although smoothed to a large extent , there still exists qualitative differences between the different scenarios .
this of course is a reflection of the vastly contrasting models of reionization compared in this paper and difference will disappear if the models investigated are more similar to each another .
the previous section provided a qualitative description of the differences in the 21-cm @xmath3 fluctuations for two cases i.e. , with and without the assumption that @xmath142 . in this section ,
a number of statistical tests are performed to quantify these differences . for this purpose
we focus on the scenario with the miniqsos .
this exercise can be repeated for the other two scenarios as well . at four different redshifts ( as indicated on the panels )
are plotted .
the solid curves indicate the power spectra resulting from the self - consistent inclusion of @xmath1 and the dashed curve , under the assumption @xmath143 . at the onset of reionization ,
@xmath144 in our model , @xmath3 is larger at all scales under this assumption because we have enforced , in effect , an _ artificial _ decoupling of @xmath1 from @xmath7 in regions of the simulated universe where @xmath145 .
but by redshift 10 the power spectra match exactly as the ly@xmath5 becomes efficient everywhere in the box to couple @xmath1 to @xmath146 .
notice that the power keeps dropping as @xmath147 and universe gets reionized at lower redshifts.,scaledwidth=50.0% ] firstly , fig .
[ fig : powspec ] shows the 3-d power spectra of @xmath12 at redshift 12 , 10 , 8 and 6 . in our model
, reionization begins at around redshift 12.7 and thus z=12 can be considered the onset of reionization .
the dashed curve reflects the assumption that @xmath143 while the solid curve takes into account the evolution of @xmath1 self - consistently . at the initial stages of reionization ,
only regions at close proximity to the source have been ionized and heated , and the volume filling by `` spheres of ionization '' has just begun , leaving large portions of the universe neutral and cooler than the cmb . here
we assume that after the epoch of recombination the universe cools adiabatically , i.e. , @xmath148 and @xmath149 , which results in @xmath145 at the onset of reionization .
thus , we have a situation in which we have assumed @xmath143 , with @xmath150 , resulting in an _ artificial _ decoupling of @xmath1 and @xmath7 , and is manifested as increased 21-cm power .
the discrepancy quickly ( by @xmath151 ) disappears as the spheres of ly@xmath5 overlap to volume - fill the simulation box and hence increase the @xmath1 @xmath6 coupling efficiency .
therefore , in our model , it is safe to make the assumption that @xmath1 follows @xmath6 after about @xmath151 . as a function of redshift .
line styles correspond to fig.[fig : powspec ] .
because at earlier redshifts ( around 12 ) @xmath145 within a large fraction in the simulation box , the minimum value attained by assuming @xmath143 is much lower than when only regions with ly@xmath5 excitation ( which is typically near the source ) is allowed to couple @xmath1 with @xmath6 . and for the same reason the maximum value in both cases are similar because the assumption holds close to the sources as the presence of ly@xmath5 flux ensures efficient coupling . as a result
the global average @xmath3 is also lower for the approximated model at the onset of reionization and catches up quickly to the `` proper '' model as the universe fills up with sufficient ly@xmath5 flux.,scaledwidth=50.0% ] from eq .
[ eq : dtb ] it follows that @xmath3 attains large negative values when two conditions are satisfied , @xmath152 and a high neutral hydrogen overdensity .
we therefore plot the mean , minimum and maximum values of @xmath3 as a function of redshift in fig .
[ fig : dtbmmm ] . as expected the mean and minimum values of @xmath3 are much lower for the case in which the spin temperature has been set to the kinetic temperature , because the universe has cooled substantially below the cmb and coupling @xmath6 everywhere in the universe to @xmath1 enforces the conditions required for large negative @xmath3 as stated above .
the maximum positive values though do not differ because a positive @xmath3 indicates that @xmath153 and hence has to be heated by the radiating source .
and in regions where the source has heated the igm , the presence of ly@xmath5 flux and colliosional coupling almost always guarantees the coupling of @xmath1 to @xmath6 anyway .
( top ) and the ratio , @xmath154 ( bottom ) , are plotted as a function of redshift . if @xmath155 was assumed to track @xmath156 , then the top panel would always be saturated at unity .
the kinetic temperature likewise merge with the spin temperature only by a redshift of @xmath157 , as seen in the bottom panel.,scaledwidth=50.0% ] as another simple diagnostic of the efficiency of @xmath1 @xmath6 coupling we plot , ( i ) the differential temperature , i.e. , @xmath158 , ( ii ) their ratio , @xmath159 , as a function of redshift ( fig . [ fig : meantstk ] ) .
note here that if the spin temperature was artificially set to the kinetic temperature , the differential temperature would be saturated at unity .
but the coupling ensures that the spin temperature remains below that of the cmb till about a redshift of 10 .
although significantly different at the early phases of reionization , in the model we have assumed , @xmath1 catches up with @xmath6 at a redshift of about eight , when the coupling due to ly@xmath5 and collisions become significant enough to tie @xmath1 to the local @xmath6 .
is considerably lower than the @xmath6 initially and then merges with @xmath6 as reionization proceeds .
refer text for more details.,scaledwidth=50.0% ] as an alternative probe to investigate the influence of ly@xmath5-coupling coupling , in fig.[fig : powspectemp ] we plot the power spectra , now of the kinetic ( solid line ) and the spin ( dashed line ) temperatures at four different redshifts .
there are a few interesting attributes to this figure .
firstly we see that at all scales @xmath1 is considerably lower than @xmath6 at the onset of reionization ( @xmath160 ) .
this scenario changes rapidly as @xmath1 approaches @xmath6 .
note that it is only towards the end of reionization ( z=6 ) that @xmath1 is identical to @xmath6 at all scales .
the pecularity observed is that all scales in @xmath1 do not approach @xmath6 simultaneously . as seen from the top two panels of the figure at @xmath161 and @xmath162 , @xmath1 approaches @xmath6 at large scales ( small @xmath163 values ) .
and then subsequently at small scales towards the end of reionization .
this is because the central parts of the ionized bubble around the source was highly ionized and because @xmath164 , the spin temperature obtains extremely low values .
now , as you go further away from the source , the igm is becomes neutral , but @xmath74 from the source and the secondary ly@xmath5 photons is high enough to couple the spin temperature to the kinetic temperature .
thus , the large scales ( large volumes outside the ionized bubble ) get `` matched '' first followed by the small scales , when the igm is hot and collisions become important . here
we briefly compare the output of our simulation with that of @xcite , wherein a semi - analytical scheme is developed to look at ionization and heating during the epoch of reionization .
fig.[fig : compare ] shows the comparison between our simulations which has now been performed for the case of a power - law source starting at 100ev to facilitate a balanced comparison with the 100ev case of @xcite .
plotted in this figure are the average gas temperature ( solid line and crosses ) and volume filling factor of x - ray radiation ( dashed line and filled hexagons ) .
the lines correspond to our simulation and points to @xcite .
their case with the normalization factor @xmath165 , ergs / s/@xmath166 . ] and photon energies of 100.0 ev shows similar trends . because we start reionization a bit later ( @xmath167 ) the temperature is a bit lower initially .
( solid line ) and the volume filling factor ( dashed line ) of the x - ray radiation as a function of redshift for the case of the miniqsos .
for comparison , points from @xcite are shown for @xmath6 ( crosses ) and filling factor ( filled hexagons ) with their case of @xmath168 and photon energies 100.0 ev.,scaledwidth=50.0% ] their model is qualitatively different because the x - ray emissivity is tied to their star - formation rate .
otherwise , the manner in which heating is implemented to similar to the one described in our case albeit without terms involving compton heating / cooling .
these effects though are important towards the central regions of the source @xcite .
the focus of this paper was three - fold .
1 ) to introduce an extension to bears in order to incorporate heating ( including x - ray heating ) and discuss its application to two cases of reionization and heating , i.e. , for blackbody ( stellar ) and power - law ( miniqso ) type sources , 2 ) to monitor the evolution of spin temperature ( @xmath1 ) self - consistently and contrast its influence against the usual assumption made in most reionization simulation that @xmath169 , and 3 ) use the extended bears code to study reionization due to a hybrid population of stars and miniqsos as reionizing sources . to incorporate heating into bears , we embeded spheres of `` kinetic temperature '' around the sources of radiation , much like in the algorithm used to obtain the ionized fraction @xcite .
the bears algorithm implemented is extremely fast , in that it takes @xmath170 hours ( 2ghz , 16-core , 32 gb shared ram ) to perform the radiative transfer ( including heating ) on about 35 different boxes ( @xmath14 and 100 @xmath171 comoving mpc ) between redshifts 12 and 6 .
these snapshots are interpolated between redshifts to produce a contiguous data cube spanning redshifts @xmath172 , i.e. , the observational range of the lofar - eor experiment .
the efficiency of the code facilitates the simulation of many different scenarios of reionization and test their observational signatures .
apart from predicting the nature of the underlying cosmological signal , these simulations can be used in conjunction with others probes of reionization to enhance the detectability and/or constrain parameters concerning reionization . in @xcite
these simulations were cross - correlated with the simulations of the cmb anisotropy ( mainly the kinetic and thermal sz - effect ) to probe reionization . in the same paper jelic et al . , repeated the same exercise by using reionization simulations from @xcite and found similar results , further emphasizing the validity of the approximate manner in which we solve the evolution of the kinetic temperature in our simulations .
the improved bears code was used to simulate three distinct reionization scenarios , i.e. , miniqsos , stellar and a mixture of both .
these simulations further emphasized that miniqsos were not only very efficient in increasing @xmath6 of the igm but the secondary ly@xmath5 flux produced by these sources was enough to drive @xmath1 away from @xmath7 in an extended region around the source thereby rendering the igm around the source `` bright '' in @xmath3 .
this implies , if a miniqso hosting a black hole in the mass range @xmath173 is within the observing window of lofar , the value of @xmath3 and its spatial extent would be conducive to a possible detection .
the code was also applied to simulate reionization with a hybrid population of stars and miniqsos .
every dark matter halo was embedded with both a quasar and a stellar component .
the relation between the two was set according to the `` magorrian relation '' we observe today .
the total mass in quasars were constrained following black hole mass density estimates by @xcite .
the effect of the hybrid population on the igm was an interpolation between the scenarios of stars and miniqsos i.e. , although the heating ( and ionization ) was not as extended as in the case of miniqsos , it was definitely larger than that for stars .
our ultimate goal is to simulate mock data sets for the lofar - eor experiment . towards this end
, reionization simulations performed on 35 different snapshots between redshifts 6 and 12 were combined to form `` frequency cubes '' ( reionization histories , @xcite ) for all three scenarios i.e. , stars , miniqsos and the hybrid case . to fold - in the effect of the instrument we smoothed these frequency cubes according to the lofar beam - scale .
notwithstanding the convolution by the lofar beam , stark differences between the scenarios clearly persists .
the r.m.s @xmath3 cleary shows this difference between scenarios .
all three scenarios of reionization discussed above was performed for two different cases , i ) assuming that @xmath1 is always coupled to @xmath6 and ii ) following the evolution of @xmath1 self - consistently in the simulation .
figures [ fig : histcomp ] and [ fig : histcompinf ] reflect the differences between the two cases . to further quantify the influence of the self - consistent evolution of @xmath1 , especially at the beginning of reionization , a suite of statistical analyses was performed .
flux and heating induced by miniqsos . ]
firstly , the 3-d power spectra of @xmath12 ( fig .
[ fig : powspec ] ) for the two cases was calculated at four redshifts ( 12 , 10 , 8 and 6 ) .
we see that although there is a considerable difference at the onset ( z=12 ) of reionization the power spectra of the brightness temperature match - up by a redshift of 8 .
this is attributed to the volume filling nature of the spin temperature coupling induced by miniqsos .
the minimum , maximum and mean value of @xmath174 as a function of redshift reveal similar differences between the two cases .
the maximum value of @xmath174 is the same in both cases because regions where the igm is considerably heated also corresponds to regions with a large supply of ly@xmath5 photons and they are also regions where colliosional coupling is effective .
the minimum and average values reveal a difference since under the assumption that @xmath175 , regions much below @xmath7 are artificially coupled to @xmath1 . as a final statistic we compute the 3-d power spectra of @xmath6 and @xmath1 and compare them ( fig .
[ fig : powspectemp ] . vast differences do exist between the two almost till the end of reionization , when they become identical .
an interesting result of this comparison is that the large scales of the two spectra get matched before the small scales in the box .
the interpretation of this behaviour is given in section [ sec : stats ] .
we also compared our results to that of @xcite and found good match for the evolution of gas temperature and the volume filling factor of x - ray radiation . in the near future calibration errors expected from a lofar - eor observational run
will be added to these simulations along with the effects of the ionosphere and expected radio frequency interference ( rfi ) .
subsequently , galactic and extra - galactic foregrounds modeled as in @xcite will be merged with these simulations to create the final `` realistic '' data cube .
these data cubes will be processed using the lofar signal extraction and calibration schemes being developed to retrieve the underlying cosmological signal in preparation for the actual lofar - eor experiment which has already begun .
the author rmt would also like to thank the institute for the physics and mathematics of the universe , tokyo , for the support provided during an extended stay .
lofar is being partially funded by the european union , european regional development fund , and by `` samenwerkingsverband noord - nederland '' , ez / kompas .
the authors would like to thank the anonymous referee was his / her extensive comments on the manuscript and for improving its contents significantly .
rmt and sz would also like to thank joop schaye and andreas h. pawlik for providing us with the n - body simulations and for useful discussions .
99 abel t. , bryan g. l. , norman m. l. , 2002 , science , 295 , 93 abel t. , bryan g. l. , norman m. l. , 2000 , apj , 540 , 39 baek s. , di matteo p. , semelin b. , combes f. , revaz y. , 2009 , a&a , 495 , 389 bromm , v. , coppi , p.s . ,
larson , r.b .
, 2002 , apj , 564 , 23 chuzhoy l. , alvarez m. a. , shapiro p. r. , 2006 , apj , 648 , l1 ciardi b. , madau p. , 2003 ,
apj , 596 , 1 ciardi b. , ferrara a. , marri s. , raimondo g. , 2001 , mnras , 324 , 381 davis m. , efstathiou g. , frenk c. s. , white s. d. m. , 1985 , apj , 292 , 371 dijkstra m. , haiman z. , loeb a. , 2004 , apj , 613 , 646 elvis m. , et al . , 1994 , apjs , 95 , 1 gnedin n. y. , abel t. , 2001 , newa , 6 , 437 furlanetto s. r. , zaldarriaga m. , hernquist l. , 2004 , apj , 613 , 1 fan x. , et al .
, 2006 , aj , 131 , 1203 ferrarese l. , 2002 , apj , 578 , 90 field g. b. , 1958 , proc i.r.e . , 46 , 240 fukugita m. , kawasaki m. , 1994 , mnras , 269 , 563 furlanetto s. r. , loeb a. , 2002 , apj , 579 , 1 jeli v. , et al . , 2009 , arxiv , arxiv:0907.5179 jeli v. , et al . , 2008 , mnras , 891 kuhlen m. , madau p. , 2005
, mnras , 363 , 1069 laor a. , fiore f. , elvis m. , wilkes b. j. , mcdowell j. c. , 1997 , apj , 477 , 93 labropoulos p. , et al .
, 2009 , arxiv , arxiv:0901.3359 laursen p. , razoumov a. o. , sommer - larsen j. , 2009 , apj , 696 , 853 madau p. , meiksin a. , rees m. j. , 1997 , apj , 475 , 429 magorrian j. , et al .
, 1998 , aj , 115 , 2285 mellema g. , iliev i. t. , alvarez m. a. , shapiro p. r. , 2006 , newa , 11 , 374 mesinger a. , 2009 , arxiv , arxiv:0910.4161 mesinger a. , furlanetto s. , 2007 , apj , 669 , 663 nakamoto t. , umemura m. , susa h. , 2001 , mnras , 321 , 593 nusser a. , 2005 , mnras , 359 , 183 nusser a. , silk j. , 1993 , apj , 411 , l1 page l. , et al . , 2007 , apjs , 170 , 335 pawlik a. h. , schaye j. , 2008 , arxiv , 802 , arxiv:0802.1715 pierleoni m. , maselli a. , ciardi b. , 2009 , mnras , 393 , 872 razoumov a. o. , cardall c. y. , 2005 , mnras , 362 , 1413 ricotti m. , ostriker j. p. , 2004 ,
mnras , 352 , 547 rijkhorst e .- j .
, plewa t. , dubey a. , mellema g. , 2006 , a&a , 452 , 907 ritzerveld j. , icke v. , rijkhorst e .- j . , 2003 ,
astro , arxiv : astro - ph/0312301 santos m. g. , amblard a. , pritchard j. , trac h. , cen r. , cooray a. , 2008 , apj , 689 , 1 sazonov s. y. , ostriker j. p. , sunyaev r. a. , 2004 , mnras , 347 , 144 schaerer d. , 2002 , a&a , 382 , 28 seljak u. , zaldarriaga m. , 1996 , apj , 469 , 437 shull j. m. , van steenberg m. e. , 1985 , apj , 298 , 268 spergel d. n. , et al .
, 2007 , apjs , 170 , 377 springel v. , 2005 , mnras , 364 , 1105 springel v. , hernquist l. , 2003 , mnras , 339 , 312 susa h. , 2006 , pasj , 58 , 445 thomas r. m. , et al . , 2009 , mnras , 393 , 32 thomas r. m. , zaroubi s. , 2008 , mnras , 384 , 1080 tasitsiomi a. , 2006 , apj , 645 , 792 vanden berk d. e. , et al . , 2001 ,
aj , 122 , 549 vignali c. , brandt w. n. , schneider d. p. , 2003
, aj , 125 , 433 volonteri m. , lodato g. , natarajan p. , 2008 ,
mnras , 383 , 1079 valds m. , ferrara a. , 2008 , mnras , 387 , l8 whalen d. , norman m. l. , 2006 , apjs , 162 , 281 wouthuysen s. a. , 1952 , aj , 57 , 31 wyithe j. s. b. , loeb a. , 2004 , apj , 610 , 117 yoshida n. , sokasian a. , hernquist l. , springel v. , 2003 , apj , 598 , 73 zahn o. , lidz a. , mcquinn m. , dutta s. , hernquist l. , zaldarriaga m. , furlanetto s. r. , 2007 , apj , 654 , 12 zaroubi s. , thomas r. m. , sugiyama n. , silk j. , 2007 , mnras , 375 , 1269 zaroubi s. , silk j. , 2005 , mnras , 360 , l64 | simulations estimating the brightness temperature ( @xmath0 ) of the redshifted 21-cm from the epoch of reionization ( eor ) often assume that the spin temperature ( @xmath1 ) is decoupled from the background cmb temperature and is much larger than it , i.e. , @xmath2 .
although a valid assumption towards the latter stages of the reionization process , it does not necessarily hold at the earlier epochs .
violation of this assumption will lead to fluctuations in @xmath3 that are neither driven by density fluctuations nor by @xmath4 regions .
therefore , it is vital to calculate the spin temperature self - consistently by treating the ly@xmath5 and collisional coupling of @xmath1 to the kinetic temperature , @xmath6 . in this paper
we develop an extension to the bears algorithm , originally developed to model reionization history , to include these coupling effects . here
we simulate the effect in ionization and heating for three models in which the reionization is driven by stars , miniqsos or a mixture of both .
we also perform a number of statistical tests to quantify the imprint of the self - consistent inclusion of the spin temperature decoupling from the cmb .
we find that the evolution of the spin temperature has an impact on the measured signal specially at redshifts higher than 10 and such evolution should be taken into account when one attempts to interpret the observational data .
[ firstpage ] physical data processes : radiative transfer cosmology : theory observation diffuse radiation radio lines : general . |
gravitational wave observations will be sensitive to the spins of black holes @xcite .
simulated search pipelines indicate that accounting for black hole spin can substantially improve the sensitivity of gravitational wave searches for binaries containing one @xcite or two @xcite black holes
. however , this improvement comes at a cost of increased computational resources and is only realised if the spins of black holes are indeed significant .
it is therefore necessary to balance the increased cost with the expectation that black holes will indeed have large spins .
here we discuss that evidence .
truly isolated black holes are expected to satisfy the kerr bound , @xmath0 , where @xmath1 and @xmath2 denote the angular momentum and mass of the black hole and @xmath3 and @xmath4 the speed of light and newton s gravitational constant respectively .
this bound is manifested in the formula for the spatial coordinate of the event horizon in the kerr solution , @xmath5 .
the value of @xmath6 can formally be computed for any object that has a mass and angular momentum and it is worth recalling that the kerr bound is only a bound on black holes .
many non - compact objects such as the earth , the sun and extremely rapidly rotating massive stars like vfts 102@xcite do not satisfy this bound , whilst compact objects like the rapidly rotating millisecond pulsar psr j1748 - 2446ad @xcite and the near - extremal black hole candidate cygnus x-1@xcite do .
this is described in table [ table : spins ] where the spin values for the solid objects assume they are constant density spheres . while the exterior spacetimes of these objects are all approximately vacuum , axisymmetric and stationary as required by the kerr solution , except for cygnus x-1 they do not describe black holes ( do not contain horizons ) and
so can have a different multipole structure to the kerr solution and are not constrained by the kerr bound .
things are even more extreme for elementary particles .
the quantum spin of an elementary particle can be related to an asymptotic classical angular momentum in the sense of the einstein - de haas effect . in this way we can simply calculate values for the mass , specific angular momentum and charge of elementary particles .
an electron s mass is @xmath7secs and its specific angular momentum is @xmath8 so its @xmath6 value is a whopping @xmath9 .
( in fact things are even more extreme in the context of the kerr - newman solution as the electron has a square root charge of @xmath10 seconds ) .
the reason for the violation of the kerr bound in these cases is that elementary particles typically have planckian values of spin , but not of mass . of the standard model elementary particles , only the higg s boson satisfies the kerr bound because its charge and spin are thought to be zero .
[ cols= " < , < , < , < , < " , ] [ table : spins ] * table 1 . * approximate values of mass and specific angular momentum for the earth , sun , a rapidly spinning massive star vfts 102 , a rapidly spinning neutron star psr j1748 - 2446ad and a rapidly spinning black hole cygnus x-1 . for ease of comparison , both
the mass and specific angular momentum values are given in seconds .
x - ray observations of accretion disks have been able to measure the spins of around 10 stellar mass black holes @xcite .
these are displayed in fig [ fig : xray_spins_names.png ] .
, assuming ns natal spins of 0 , bh natal spins of 0.5 and merging in 15 billion years
. adapted from @xcite . ] , assuming ns natal spins of 0 , bh natal spins of 0.5 and merging in 15 billion years . adapted from @xcite . ]
of these most are low mass x - ray binary ( lmxb ) sources and are unlikely to form the double compact object systems necessary to be seen by the current generation of ground - based gravitational wave detectors .
three of the systems ( lmc x-3 , m33 x-7 and cygnus x-1 ) are high mass x - ray binaries ( hmxb ) where the companion to the black hole is a massive star , with mass greater than @xmath11 and these systems do have a chance to form either neutron star - black hole binaries or binary black holes systems .
these systems in themselves are not targets for current ground - based gw observatories as they are many millions of years away from merging but it is interesting to note that all three of these hmxbs have black holes with large spins @xmath12 and this suggests that the population of black holes in compact binary systems might be dominated by black holes with large spins .
the probability of obtaining three values all above @xmath13 from a flat distribution is only @xmath14% . on the other hand , observations of neutron star
pulsars suggest that the spins of neutron stars in compact binaries are very low @xmath15 .
if this is indeed the case it implies that the formation channels of these objects are very different in terms of their angular momentum evolution .
neutron stars may be born with low spins or they may spin down rapidly after birth .
the observation of young slowly spinning pulsars , such as the crab pulsar rotating at only @xmath16hz with a spin down rate of @xmath17hz per year , 1000 years after its birth , constrains models of rapid spin - down for neutron stars .
to form a compact binary system requires two large heavy stars in a binary .
single massive stars are typically slowly rotating at the end of their lives dues to stellar wind losses and not differentially rotating due to angular momentum redistribution by magnetic torques @xcite .
but stars in binaries can spin up , increasing their rotational angular momentum by mass transfer @xcite .
the spin rate of both stars can be affected by tides and mass transfer @xcite .
tidal interactions will tend to circularize the orbits of massive stars@xcite .
massive stars in binaries can have rotational velocities @xmath18km / s @xcite .
close binary systems are expected to be tidally locked @xcite .
the tidal locking timescale is between 100 and 10,000 years for helium core stars before they turn into black holes @xcite .
the tidal locking is more efficient on the larger object the closer to equal mass the bodies , perhaps giving an explanation why bhs in lmxbs have a wide range of spins , but predominantly high spins in hmxbs .
it has been shown that there is a correlation between the black hole mass and orbital period in x - ray transients that may be partly explained by tidal locking @xcite . for simplicity here
we will assume that the moment of inertia for each star is given by @xmath19 where @xmath20 is a constant ( equal to @xmath21 for solid balls of uniform density ) .
the relationship between radius and mass is often assumed to be @xmath22 where typical values of @xmath23 for massive stars are @xmath24 to @xmath25 .
then we find that the dimensionless spin parameter @xmath6 for an object with mass @xmath26 is given by a _ * = k ( ) ^3/2()^ 2- 1 . where @xmath27 denotes the semi - major axis of the orbit .
note that the three hmxb systems would have @xmath28 if they were tidally locked stars at their current mass and separation values . observed rotation rates of massive stars in binaries tend to be significantly higher than the tidal locking rate @xcite , since a @xmath29 binary would need a separation of @xmath30 to be tidally locked if the rotational velocity is @xmath31km / s , corresponding to @xmath32 .
stars that are further apart will have a lower rotational velocity .
a separation of 1000 solar radii would have a tidally locked velocity of @xmath31 m/s , since @xmath33 .
different scenarios have been proposed for how collapse proceeds for massive stars .
these distinguish between failed core collapse ( collapsar ) with no explosion and successful core collapse ( supernova ) where the outer envelope is partially ejected @xcite .
direct collapse may be favoured for heavy systems as the energy flux from the core is not enough to expel the outer shell .
if a star collapses directly to a black hole , without shedding any matter and conserving its mass and its angular momentum , then its @xmath6 value would be conserved .
for failed core collapse models , oconnor and ott @xcite find a linear relationship between initial specific angular momentum of the star and final black hole spin parameter ( their fig 11 ) .
the relation is approximately @xmath34 .
the specific angular momentum before collapse , @xmath35 , can be calculated using the tidal lock model above , to obtain a final @xmath6 after collapse of , a _ * = 155k()^3/2()^ 2 + 0.5 .
it should be noted that for rapidly spinning black holes , centrifugal support will tend to prevent the black hole from accreting large amounts of matter during a collapse scenario .
if an initial , near extremal black hole forms , subsequent shells with @xmath28 will not be accreted but expelled @xcite . in this case , as in the successful core collapse , detailed modeling is required to derive final mass and spin values for the remnant .
the distribution of upper limits on the final bh spins , assuming natal spins of zero for neutron stars and @xmath24 for black holes , for merging nsbh systems in @xcite is shown in fig .
[ fig : hist_finalspinns0bh0pt5.png ] .
this shows that if accretion in the common envelope phase does behave as a thin disk , then the final spins of bhs in nsbh binaries may be quite large , with most values falling in the range @xmath25 to @xmath36 . the distribution in fig .
[ fig : hist_finalspinns0bh0pt5.png ] actually contains two distinct populations with different formation channels .
most of the high spin systems are actually light black holes that formed initially after collapse as neutron stars and subsequently accreted sufficient material during common envelope to form black holes , where the threshold for black hole formation is set at @xmath37 . for these light systems ,
the accretion of a small amount of matter is enough to spin them up considerably .
the other distribution consists of objects that collapsed to black holes before the common envelope phase .
these systems are typically more massive and hence do not spin up as much during common envelope .
the limited x - ray evidence suggests that black holes in compact binaries may have high spins .
we have discussed several mechanisms that could lead to these high spins , including tidal locking of massive progenitor stars , rapid collapse to black holes and common envelope accretion .
there are still large uncertainties of all these processes , although they tend to reinforce one another .
measuring the spins of a large number of black holes with gravitational wave detectors will help to constrain some of these model uncertainties .
this work was supported by the max planck society .
the author would like to thank richard oshaughnessy for helpful conversations and sharing data for fig .
[ fig : hist_finalspinns0bh0pt5.png ] . | black hole spin will have a large impact on searches for gravitational waves with advanced detectors . while only a few stellar mass black hole spins have been measured using x - ray techniques , gravitational wave detectors
have the capacity to greatly increase the statistics of black hole spin measurements .
we show what we might learn from these measurements and how the black hole spin values are influenced by their formation channels . |
@xmath55&=x^mp^n - p^nx^m\\ & = x^mp^n - p^{n-1}(x^mp-\frac{im}{2}x^{m-1})\\ & = x^mp^n+\frac{im}{2}p^{n-1}x^{m-1}-p^{n-2}px^mp\\ & = x^mp^n+\frac{im}{2}p^{n-1}x^{m-1}-p^{n-2}x^mp^2+\frac{im}{2}p^{n-2}x^{m-1}p\\ & = x^mp^n+\frac{im}{2}p^{n-1}x^{m-1}+\frac{im}{2}p^{n-2}x^{m-1}p - p^{n-2}x^mp^2\\ \vdots\\ & = \frac{im}{2}\sum_{k=0}^{n-1 } p^kx^{m-1}p^{n - k-1}\end{aligned}\ ] ] @xmath55&=\frac{im}{4}\sum_{k=0}^{n-1 } p^kx^{m-1}p^{n - k-1}+p^{n - k-1}x^{m-1}p^k\\ & = \frac{im}{4}\sum_{k=0}^{n-1 } \left ( ( x^{m-1}p^k-[x^{m-1},p^k])p^{n - k-1 } + p^{n - k-1}([x^{m-1},p^k]+p^kx^{m-1 } ) \right)\\ & = \frac{im}{4}\sum_{k=0}^{n-1 } \left ( x^{m-1}p^{n-1}-[x^{m-1},p^k]p^{n - k-1 } + p^{n - k-1}[x^{m-1},p^k]+p^{n-1}x^{m-1 } \right)\\ & = \frac{im}{4}\sum_{k=0}^{n-1 } \left ( x^{m-1}p^{n-1}+p^{n-1}x^{m-1}+ [ p^{n - k-1},[x^{m-1},p^k ] ] \right)\\ & = \frac{imn}{4 } \left ( x^{m-1}p^{n-1}+p^{n-1}x^{m-1}\right)+ \frac{im}{4}\left ( \sum_{k=1}^{n-2 } [ p^{n - k-1},[x^{m-1},p^k ] ] \right)\\\end{aligned}\ ] ] as a result : @xmath56 -\frac{1}{n+1}\sum_{k=1}^{n-1}[p^{n - k},[x^m , p^k]]\ ] ] note that for @xmath57 , the summation term in the identity above is zero .
also , due to the jacobi identity , we have @xmath58=[p^{k},[x^m , p^{n - k}]]$ ] , and this may also lead to some simplification depending on the value of @xmath59 .
here , we shall illustrate the necessity for having better ( than any commonly used ) approximations . the approximations that we employ
are explicitly introduced below in this supplemental material .
for instance , for a nested commutation approximation of an interaction strength @xmath60 and a dominant error term of @xmath61 , we would need @xmath62 operations corresponding to an eighth - order approximation . for comparison , note that it is also possible to use lloyd s method @xcite .
lloyd s idea was originally intended as a proof of principle , but it has also been used in the literature as an approximation tool .
the required number of operators using lloyd s method is evaluated below .
the notation @xmath63 $ ] is used to indicate that a number of @xmath64 operators is required . for the commutation operator
, we have @xmath65}&=\left(e^{ib\frac{t}{n}}e^{ia\frac{t}{n}}e^{-ib\frac{t}{n}}e^{-ia\frac{t}{n}}\right)^{n^2}+f\left(\frac{t^3}{n},a , b\right)\label{basic_comm}\\ e^{t[a , b]}&=n[4\times n^2]+f\left(\frac{t^{3/2}}{n}\right)\nonumber\end{aligned}\ ] ] while for the nested commutation , we obtain @xmath66}&=e^{itb}e^{t^2[b , a]}e^{-itb}e^{-t^2[b , a]}+f'(t^4,a , b)\label{basic_nested}\\ & = \left(e^{i\frac{t}{m}b}e^{\frac{t^2}{m^2}[b , a]}e^{-i\frac{t}{m}b}e^{-\frac{t^2}{m^2}[b , a]}\right)^{m^3}+f'\left(\frac{t^4}{m}\right)\nonumber\\ & = \left(e^{i\frac{t}{m}b}\left(e^{ib\frac{t}{ml}}e^{ia\frac{t}{ml}}e^{-ib\frac{t}{ml}}e^{-ia\frac{t}{ml}}\right)^{l^2}e^{-i\frac{t}{m}b } \left(e^{-ib\frac{t}{ml}}e^{-ia\frac{t}{ml}}e^{ib\frac{t}{ml}}e^{ia\frac{t}{ml}}\right)^{l^2}\right)^{m^3}+ f'\left(\frac{t^4}{m}\right)+f\left(\frac{t^3}{l}\right)\nonumber\\ e^{it[b,[b , a]]}&=n[8\times l^2\times m^3]+f'\left(\frac{t^{4/3}}{m}\right)+f\left(\frac{t}{l}\right)\nonumber\end{aligned}\ ] ] thus , using these approximations , the number of ( elementary ) operations will be of the order of @xmath52 for an interaction strength of @xmath60 and a dominant error term of @xmath61 .
for the decomposition of a kerr gate with coefficient 0.1 and dominant error term @xmath61 , we need to approximate the following operators : @xmath67 $ ] , @xmath68 $ ] .
the kerr gate is then : @xmath69 . here
@xmath70 is the fourier transform of @xmath71 , i.e. , @xmath72 .
first , using second - order three - party splitting ( see refs .
@xcite ) , we split into separate elements : @xmath73 and then we insert the approximations for the commutation and the nested commutation operators ( see below )
. approximations for these operators can be calculated considering the necessary approximation order . for the nested commutation operator with coefficient @xmath74 and dominant error term smaller than @xmath61 ,
we need the fourth - order approximation which corresponds to 9 operators , @xmath75}&=&e^{-i0.11157p^2}e^{i0.02231x^3}e^{i0.02231p^2}e^{-i0.02231x^3}e^{-i0.02231p^2}\\&&e^{-i0.02231x^3}e^{i0.02231p^2}e^{i0.02231x^3}e^{-i0.11157p^2 } + 0.55326\times 10^{-3}\times f(x^3,p^2 ) + ... \end{aligned}\ ] ] similarly , the commutation operator with coefficient @xmath76 and dominant error term smaller than @xmath61 also requires the fourth - order approximation , this time corresponding to 10 operators , @xmath77}&=&e^{-i0.25298x^3}e^{i0.210819p^3}e^{i0.01918x^3}e^{-i0.28476p^3}e^{i0.36163x^3}\\&&e^{i0.36053p^3}e^{i0.12861x^3}e^{-i0.05805p^3}e^{-i0.25644x^3}e^{-i0.22853p^3 } + 0.41643\times 10^{-3}\times f(x^3,p^3 ) + ... \end{aligned}\ ] ] as we have to apply @xmath71 four times and @xmath78 only once , we require @xmath79 elementary operations .
every single @xmath71 operation as well as the @xmath78 operation each consume @xmath80 additional fourier transforms in order to switch from the elementary @xmath22 gates to the necessary @xmath23 gates [ here , @xmath80 indicates that some fourier gates cancel in the sequence when switching between @xmath71 and @xmath70 ] . as a result , summing up
, we will need @xmath82 extra fourier gates , in addition to the @xmath83 elementary quadratic and cubic @xmath22 gates . in total , this leads to @xmath84 elementary operations from the universal gate set needed to simulate a kerr interaction of size @xmath60 with errors scaling smaller than @xmath61 . without using these more powerful approximation techniques ( as explained in detail below ) and using instead standard techniques such as the well - known trotter formula for splitting and the standard approximations for the commutation operator and the nested commutation operator , we would need @xmath85 operations to simulate this kerr interaction gate with the same precision .
as a second example , we discuss the decomposition of the hamiltonian evolution of parametric down conversion with a quantized pump . this example illustrates that one can sometimes find exact operator relations and thus reduce the use of operator approximations to some extent .
the operator to be decomposed is : @xmath86 , where the subscripts denote the two optical modes .
as described , we first divide the hamiltonian into its simplest hermitian parts @xmath87 through splitting approximations .
now the operators @xmath88 and @xmath89 can be realized through exact operator relations instead of operator approximations : @xmath90 here , @xmath91 and @xmath92 are some free parameters and @xmath93 are equivalent to @xmath94 up to fourier transforms .
the only remaining operator , @xmath95 , can be realized through a commutation approximation of @xmath96 $ ] .
in this section , we shall explain a very powerful method that we propose to employ for deriving the necessary approximations in our scheme .
we also present explicitly some efficient approximations which are used throughout the paper .
let us first define the problem .
suppose we have access to a finite set of non - commuting exponential operators , @xmath97 , where @xmath98 , @xmath99 , @xmath100 , ... are arbitrary operators . through different and finite concatenations of these operators ,
our aim is to obtain another operator , @xmath101 , approximately , i.e. , up to a certain error , @xmath102 where @xmath103 are scalars .
also , we further assume that @xmath104 are not fixed , i.e. , they are adjustable parameters being functions of @xmath5 as a result of the approximation problem .
if the taylor expansion of both sides of eq . matches for the orders of @xmath5 up to @xmath34 , then the right hand side of the equation is called @xmath35th order approximation for the operator @xmath105 , and the problem is to find a method for this approximation , i.e. , to find the appropriate values for @xmath104 . in the present context , we are interested in the case where @xmath106 is an arbitrary nested commutation operator comprised of @xmath107 and @xmath108 : @xmath109]]$ ] , which is to be approximated through the set @xmath110 . throughout this section
we call an exponential operator with exponent @xmath111 $ ] _ commutation operator _ , and similarly for the nested commutators . in general
, we omit hats to indicate operators except when confusions are possible .
the method we employ combines various elements of existing splitting and approximation techniques ( see , for instance , page 393 of the splitting review ref .
our execution of these methods relies upon the generalized baker - campbell - hausdorff series ( gbch ) which is basically the logarithm of an arbitrary exponential operator concatenation involving more than two operators ; an extension of the standard bch , as the name indicates .
it is possible to calculate the gbch in various ways , for instance , by repeatedly utilizing formulas for the standard bch . with the aid of computer software , this is straightforward . in the present work , however , we shall employ a very convenient calculation technique proposed by reinsch @xcite . for any necessary orders of the operator approximation , we use two calculation steps . in the first step , we employ a brute force and hence efficient method for approximations . in the second step ,
we obtain higher - order approximations using the results of the first step .
this second step will be necessary in order to be able to achieve orders of approximations otherwise unattainable through the first step alone . in order to derive a certain order of approximation , we start with a specific operator concatenation like the one in eq . .
we calculate the logarithm of this concatenation through the gbch and this results in a linear combination of a set of operators .
the coefficients of these operators correspond to polynomials of the input parameters .
we then solve all the polynomial equations in order to eliminate the undesired operators in the linear combination .
the solutions of the polynomial equation set give us the proper values of the input parameters for a particular approximation .
hence , the whole problem of obtaining a specific order of approximation is reduced to a computationally well defined problem , namely that of solving polynomial equations .
for example , consider a fourth - order approximation of the commutation operator @xmath112}$ ] through the operators @xmath113 .
here , @xmath114 is the square root of -1 .
we start from a concatenation of alternating @xmath115 and @xmath116 operators .
the concatenation consists of ten operators , for which the reason will become apparent below .
thus , we have @xmath117+f_4t^3[a[a , b]]+f_5t^3[b,[a , b]]\\ & + f_6t^4[a,[a,[a , b]]]+f_7t^4[b,[a,[a , b]]]+f_8t^4[b,[b,[a , b]]]+ ...
\end{split}\end{aligned}\ ] ] where the linear dependent operators such as @xmath49 $ ] are omitted .
the eight coefficients , @xmath118 , are polynomials of the input parameters , @xmath119 .
to these parameters , we assign the values @xmath120 and @xmath121 to ensure a real solution set .
then we solve these polynomials simultaneously with the conditions @xmath122 and @xmath123 , @xmath124 . as a result
, we obtain sets of appropriate values for the input parameters . consequently , the left hand side of eq .
can be considered a fourth - order approximation for the commutation operator .
more specifically , for the first step of the calculation , we need to take into account the following points .
_ constructing real solutions _ the number of operators in the concatenation should always match the number of polynomials to be solved .
however , the solutions might be all complex , which is useless for physical scenarios . in this case
, one has to add more operators to the concatenation to be able to adjust the input parameters and obtain an all real - valued solution set .
this is why in the above example , we added two more operators to the concatenation and chose the specific values 1.2 and -1 .
we choose the number of extra operators and the assigned values by trial and error . _ removing linear dependency _
one should reduce the polynomials for simplification .
the polynomial orders depend on the order of the operators that belong to the polynomials ( for instance , in the above example , @xmath125 is a second - order polynomial , but @xmath126 and @xmath127 are third - order polynomials ) , and for every order there exist many linear dependent polynomials . especially for higher orders , these will consist of thousands of elements .
thus , in order to reduce the complexity , one should eliminate the linear dependent polynomials .
for example , a concatenation series comprised of 14 operators , for a fifth - order approximation , in general , gives rise to 51 polynomials , but , after the elimination of the linear dependent ones , this reduces to 14 .
_ solving polynomial equations _ as an algorithm for polynomial equation solving , we do not recommend to employ a conventional analytical method such as those standard methods based upon grbner bases or other standard commercial computer software .
these approaches hardly help for obtaining solutions of orders higher than four . instead ,
for polynomial equation solving , we use the so - called homotopy continuation method @xcite for the first step of the calculation scheme ( as well as for the second step , as described later ) .
the homotopy continuation method is an iterative numerical method .
even though it has its own complications and subtleties , we shall only use the basic principle of that method .
we employ it for iterating newton s polynomial solving algorithm .
newton s algorithm works only provided one can guess the roots of polynomials very close to their actual values .
homotopy continuation is basically a general idea to circumvent this kind of problems by just starting from a simple toy problem and slowly converting it into the real problem while tracing the solutions . in our case
, we start with a set of polynomial equations the roots of which are obvious , and make small changes on these initial polynomials .
we then solve the new equations using the results from the previous polynomial solving step as the starting point for newton s method .
further , we iterate this process such that at some point we get the original polynomials that we wanted to solve , and in addition some of their solutions . besides the overall time efficiency of this approach , the method allows us to deal with each solution separately , thus , eliminating the need of finding all the solutions .
this property indirectly improves the efficiency dramatically .
we can obtain solutions for up to fifth - order approximations and with a little more expertise on polynomial equation solving and patience , it should be possible to obtain even sixth - order approximations .
however , in order to obtain higher - order approximations , we believe that the first step alone is not sufficient .
therefore , we propose to employ yet another step in order to be able to go to higher orders in order to circumvent the problem of having an intractable calculation , we have to employ another step . in this second step , to derive higher order approximations , we now use a concatenation of those approximations that we obtain through the first step .
this second step is similar to the suzuki - yoshida method @xcite .
the difference is , instead of improving step by step and obtaining every higher order recursively , here , by utilizing the gbch , we obtain a by many orders higher approximation in a single step .
let us turn again to our example of commutation operators .
suppose we want to obtain a ninth - order approximation for a given commutation operator .
in this case , we can obtain the fifth order with the first step , corresponding to a approximation for an operator of the form @xmath112+t^6\hat{f}+t^{7}\hat{g}+ ... }$ ] .
let us denote this as @xmath128 .
we then take a concatenation series of two such operators @xmath128 and @xmath129 . obviously , the approximation of @xmath129 is the reverse ordered approximation of @xmath130 , and hence is an operator of the form @xmath131-t^6\hat{f}-t^{7}\hat{g}- ... }$ ] .
the concatenation we use is @xmath132 where @xmath133 and @xmath134 are parameters that we have to find ( this kind of concatenation is also used in ref .
@xcite for a recursive approximation in the splitting problem ) .
we again employ the gbch to obtain the resultant operator combination and eliminate the operators from order six to nine . for this particular example
, we need to solve seven polynomials , but due to the constraint of a real solution set , we use nine operators .
when we rewrite each fifth - order operator in terms of their approximated form , in total , we need 144 operators .
obviously , this method also heavily depends on the efficiency of polynomial solving . in this section
, we present some commutation and nested commutation approximations .
although for each case there exist many solution sets , we always present only one of these and we only give the explicit numbers up to six digits out of fifteen digits .
also , note that here some parameters are indicated as zero in order to avoid confusion in the general concatenation ansatz that we used .
given an initial approximation , it is also possible to use the symmetry argument of suzuki @xcite and construct recursively symmetrical approximations for the nested commutation operator , since @xmath136}e^{i(-t)^3[a,[a , b]]}=i$ ] .
hence , a sixth - order approximation for nested commutation can be constructed through fourth orders using @xmath137 operations . | we present a general , systematic , and efficient method for decomposing any given exponential operator of bosonic mode operators , describing an arbitrary multi - mode hamiltonian evolution , into a set of universal unitary gates .
although our approach is mainly oriented towards continuous - variable quantum computation , it may be used more generally whenever quantum states are to be transformed deterministically , e.g. in quantum control , discrete - variable quantum computation , or hamiltonian simulation .
we illustrate our scheme by presenting decompositions for various nonlinear hamiltonians including quartic kerr interactions .
finally , we conclude with two potential experiments utilizing offline - prepared optical cubic states and homodyne detections , in which quantum information is processed optically or in an atomic memory using quadratic light - atom interactions . _
introduction_ since the proposal of quantum computation as a generalization of computer science , an important theoretical challenge has been how to decompose an arbitrary gate into a universal set .
the corresponding theory of discrete - variable decompositions is very extensive and mostly employs matrix representations of logic gates utilizing matrix decomposition techniques @xcite .
in contrast to discrete - variable theory , there is not an established method to decompose an arbitrary operator in the continuous - variable ( cv ) regime except the proof - of - principle results on universal gate sets in refs . @xcite . in particular , ref .
@xcite makes use of an exponential operator approximation and proves that by employing certain elementary gate sets ( discussed below ) one can derive any operator up to a certain error .
however , none of these works intend to present a constructive and efficient decomposition recipe .
the problem of decomposition is intrinsically related to the concept of universality .
universality means to have a set of operators that allows you to simulate any operator on a certain hilbert space through concatenations of the elements of the universal set . for our purpose ,
achieving universality is then equivalent to decomposing , at least approximately , an arbitrary unitary exponential operator to a set of elementary unitary exponential operators : @xmath0 here , @xmath1 and @xmath2 are annihilation and creation operators , respectively , and @xmath3 are fixed hermitian functions of mode operators . the coefficients @xmath4 are interaction times of the hamiltonians and are functions of @xmath5 .
thus , different concatenations of elements of this set for varying interaction times should enable one to simulate an arbitrary operator .
we assume that we have access to arbitrary interaction times for the initial set . in our setting , there are two important criteria for cv gate decompositions : how systematic and how efficient the decompositions are . here
we shall derive methods according to these criteria and present a systematic and efficient framework for decomposing any given unitary operator that acts on bosonic modes into a universal set of elementary cv gates .
our general method consists of first expressing operators in terms of linear combinations of commutation operators and then realizing each commutation operator and their combinations through approximations . we will discuss the efficiency of the decompositions and present guidelines to obtain an arbitrary order of error . for this purpose
, we employ a powerful technique for obtaining efficient approximations ( see supplemental material for details ) . throughout
, we use the convention @xmath6 , i.e. , the fundamental commutation relation is @xmath7=i/2 $ ] with @xmath8 and @xmath9 . _
general gaussian decompositions_ for gaussian operators , i.e. , second - order operators , exact and finite decompositions to elementary sets are known ( here , order is defined as the polynomial order of the mode operators in the hamiltonian of a given operator ) .
for example , the bloch - messiah decomposition allows to decompose any second - order operator , i.e. , any unitary gaussian operation , to passive linear multi - mode optics , single - mode squeezing , and displacement operations @xcite . in ref .
@xcite the following set is presented as a single - mode gaussian universal set : @xmath10 , and in ref .
@xcite , similar to the bloch - messiah decomposition , a recipe is given to decompose any single - mode transformation of second order to this set with no more than four steps .
these exact decompositions emerge from the fact that mode transformations through second - order unitary operations are linear , and thus , one can utilize matrix representations and matrix decomposition techniques . in fact , for gaussian operator decompositions , one can find infinitely many elementary sets and decompositions , and instead of those sets above , one may choose the one that suits best the given situation and purpose . _
universal decompositions_ in order to decompose an arbitrary single - mode operator in cv systems , it has been shown that , in principle , adding a nonlinear element ( of order three or more ) to the toolbox is sufficient @xcite . in the present work we use the following set : @xmath11 this set is not unique , and one may use different gaussian elements as explained above and a different nonlinear element . however , this particular set turns out to be useful for describing cv quantum computation in the one - way model using cv cluster states @xcite .
note that one can simplify this elementary set further by omitting the second - order hamiltonian , since @xmath12 , and using the fourier transformation whose action is given in eq . .
even though this simplification has value from an academical point of view , as it reduces the minimal number of elementary gates , we are basically motivated by decomposing an arbitrary gate to a set of experimentally accessible gates .
all second - order gates are relatively easy to implement and replacing them by third - order hamiltonians will increase the complexity of the gate sequence .
therefore we shall use the overcomplete set in our decompositions without loss of generality .
one may also prefer a further extended set depending on a certain experimental situation in order to reduce the complexity of the decompositions .
in addition , one can obtain _ some _ nonlinear operations through unitary conjugation : @xmath13 .
an important unitary conjugation is the fourier transform : @xmath14 employing unitary conjugation , with the set , one can now generate certain nonlinear gates exactly .
for example , @xmath15 , which is a fourth - order operator .
however , there is only a limited number of such decomposable nonlinear operators , and therefore , for generality , we will make use of the idea of operator approximations .
besides the abstract notion of universality @xcite how can a given unitary exponential operator be decomposed to the elementary set in a systematic and efficient way ?
let us first introduce the available toolbox employed to achieve a decomposition as efficient as possible ( while efficiency will be defined and discussed later ) .
the tools we use for cv gate decompositions include gaussian operator decompositions , unitary conjugation , and exponential operator relations as well as approximations . before proceeding to the general case ,
let us demonstrate how to realize a particular nonlinear exponential operator using the above tools and the set .
a very important example is the kerr interaction operator .
it allows to convert a coherent state into a cat state @xcite and to realize a controlled quantum gate for qubits @xcite .
the one - mode self - kerr interaction , up to a gaussian element , corresponds to @xmath16 . in order to decompose the kerr operation to the set , we would first write the full kerr hamiltonian as a linear combination of commutators and then realize each through operator approximations . the following relations , together with the fourier transform , are enough to realize this gate up to a phase : @xmath17 $ ] , @xmath18 $ ] . thus , it is sufficient to realize the above commutators and their linear combinations ( see supplemental material for more details ) .
let us now present the general scheme for an arbitrary hamiltonian .
obviously , any single - mode hamiltonian , as a polynomial of bosonic mode operators , consists of operators of the form @xmath19 .
we can show that any such operator can be written as a linear combination of commutation operators .
first note that @xmath20 $ ] , and then one can derive the following two identities ( see supplemental material for a derivation ) , @xmath21,\label{eq : x_order_increment}\\ x^mp^n+p^nx^m=&-\frac{4i}{(n+1)(m+1)}[x^{m+1},p^{n+1}]\label{eq : fundm_hamilton}\\ & -\frac{1}{n+1}\sum_{k=1}^{n-1}[p^{n - k},[x^m , p^k]].\nonumber\end{aligned}\ ] ] equation is necessary to obtain arbitrary powers of @xmath22 and @xmath23 operators with the fourier conjugation , and eq
. basically prescribes how to systematically decompose an elementary hamiltonian to commutation operations of orders of @xmath22 and @xmath23 and their combinations where we can use the tools we have . for multi - mode operators one needs an extended elementary set including an entangling operation @xcite , for example in the optical context it can be the beam - splitter operation . using gaussian decomposition methods , for simplicity
, we may assume that we have access to the following gate without loss of generality , @xmath24 .
for multi - mode hamiltonians we can again use the simplifications for a single mode and eq .
, because of the fact that the operators on one mode commute with the operators on the other mode .
however , for this purpose , we initially need to realize the two - mode operations with arbitrary powers of @xmath22 and @xmath23 in both modes , similar to the single - mode relation .
the following relation together with fourier conjugation and eq . , is sufficient to realize the two - mode operations with arbitrary powers of @xmath22 and @xmath23 , @xmath25.\ ] ] then
, we can use eq . again with the single - mode operations to realize an arbitrary two - mode expression . as an example , consider the cross - kerr hamiltonian ( up to a gaussian transformation ) : @xmath26 . in this case ,
the nested commutator @xmath27 $ ] will suffice .
another , even simpler example for decomposing a nonlinear two - mode evolution , namely cubic parametric down conversion with a quantized pump , is discussed in the supplemental material . in form of a dispersive interaction ,
it may also be used to mimic a kerr gate @xcite . from an academic perspective ,
eqs . , , and
are universal not only for any hamiltonian , but also for any initial universal set with a nonlinear gate different from @xmath28 because of the well - known equations : @xmath29 $ ] , @xmath30 $ ] , where @xmath31 is a function of operators @xmath22 and @xmath23 .
thus , any initial nonlinear hamiltonian can be reduced to a form @xmath32 . from a practical perspective , however , it is unwise to use eqs . and for arbitrary initial sets because of a typical increase of complexity .
instead , one should derive an optimized expression ( in terms of the decomposition efficiency , see below ) utilizing the available tools for every other hamiltonian and every other universal set . _
efficiency_ besides having a systematic method , we also require the decompositions to be relatively efficient .
we define efficiency as the number of operators needed to realize , in an approximate fashion , a given operator with a certain negligible error ( note that this definition slightly differs from previous ones @xcite where efficiency is the scaling of the number of operators with respect to the error ) .
let us give a few more definitions .
if in @xmath33 the taylor expansion of both sides matches for the orders of @xmath5 up to @xmath34 , it is called @xmath35th - order decomposition . for example , an important case is when @xmath36 for which we will use the term splitting .
another important case is when @xmath37 $ ] which , from now on , we call commutation operator . for example , the identity below is a well - known second - order approximation for a commutation operator .
it has been used already in quantum control @xcite , discrete - variable quantum computation @xcite , cv quantum computation @xcite , or , in general , hamiltonian simulation theory @xcite , @xmath38}=e^{itb}e^{ita}e^{-itb}e^{-ita}+f(t^3,a , b)+\ldots\label{eq : bch3}\\\ ] ] it basically says that , for @xmath39 , the corresponding operator concatenation is the same as applying the commutation operator of @xmath40 and @xmath41 , up to some error where the dominant term is of the order @xmath42 . now in order to obtain more reliable gates , a straightforward and common way to improve accuracy is using smaller interaction times , @xmath43 , and applying the decomposition @xmath44 times to obtain the same interaction time as before , @xmath45}=\left(e^{ib\frac{t}{n}}e^{ia\frac{t}{n}}e^{-ib\frac{t}{n}}e^{-ia\frac{t}{n}}\right)^{n^2}+f\left(\frac{t^3}{n},a , b\right).\nonumber\ ] ] let us call this approach rescaling ( @xcite and refs . therein ) . besides improving accuracy
, rescaling is absolutely necessary to realize nested commutations .
for example , one may replace @xmath46 by @xmath47 $ ] in eq . to simulate the nested commutation operator @xmath48}=e^{itb}e^{t^2[b , a]}e^{-itb}e^{-t^2[b , a]}+f'(t^4,a , b)$ ] , and similarly for further nested commutations .
however , in this identity , an approximation of @xmath49 $ ] is still needed and eventually , using again eq . , we obtain an operator whose interaction time is of the same order as the dominant error term . in order to obtain a reasonable decomposition
, the order of the dominant error should be smaller than @xmath42 .
thus , again rescaling is needed , requiring relatively many operators to enhance accuracy .
for instance , using the approximations described so far , the number of operators to approximate a single commutation operator with coefficient 0.1 and dominant error term @xmath50 requires 4000 operators , while for the nested commutation operator , with the same values , the number of operators will be @xmath51 ( see supplemental material ) .
hence , better approximations are crucial to achieve reasonable decompositions .
what we propose to use is a general and powerful method for obtaining higher - order approximations in a direct fashion ( details on this rather technical part of our proposal can be found in the supplemental material ; see also , for instance , page 393 of ref .
@xcite ) . in this approach
, concatenations converge much faster to an arbitrary set of commutations and linear combinations of these , reducing the number of operators from the order of @xmath52 to the order of tens . _
experiment_ an immediate consequence of our work is that it bridges the originally huge gap between abstract notions of cv decomposition theory and possible experimental implementations , for instance , within quantum optics .
highly nonlinear quantum gates ( such as quartic kerr gates ) may then be realized in a deterministic fashion by concatenating tens of quadratic and cubic gates .
figure 1 illustrates two experimental schemes .
they both combine ideas for conditional optical state preparation , in order to probabilistically prepare and distill high - fidelity cubic phase states @xmath53 @xcite , with those for temporally encoded @xcite and fully homodyne - detection - based @xcite cv cluster computation .
the complication of realizing nonlinear gates is shifted offline into the preparation of the cubic ancillae . in fig .
1a ) , the conditional state preparer ( csp ) is linked with a quantum memory ( qm ) through a classical channel in order to signal whenever an offline state is available such that an optical pulse carrying the latest quantum information is released from the quantum memory .
after the first application of the optical @xmath54 gate between a first input pulse coming from the left and a first ancilla pulse coming from the top ( emerging from the csp ) , the @xmath54-transformed input pulse is measured through homodyne detection ( hd ) , while the @xmath54-transformed ancilla pulse is sent to the quantum memory with its quantum state transferred onto the memory .
after preparation of the next ancilla pulse , a new input pulse emerges from the memory for a second application of the @xmath54 gate and so on .
this is similar to the scheme of ref .
@xcite for a single quantum wire ( corresponding to a linear cv cluster chain ) , but this time including cubic states and quantum memories , and excluding non - linear measurements such as photon counting ( except for the csp ) .
note that storage of non - classical states has been already demonstrated experimentally @xcite .
the scheme in fig .
1b ) uses an atomic ensemble and a quadratic light - matter interaction @xcite .
the light - matter interaction is always delayed until an appropriate offline state is available .
quantum information can be stored and processed at the same time within the ensemble @xcite by swapping the optical and atomic states after every interaction , simply using additional quadratic interactions @xcite . as opposed to refs .
@xcite , the optical state is either a squeezed vacuum or a conditionally prepared cubic state , inserted into the temporal cluster chain whenever needed @xcite .
thus , we make explicit use of the atomic memory to compensate for the probabilistic optical ancilla - state preparations . only with these cubic ancillae
is it possible to perform universal gates solely through homodyne detection @xcite . in principle
, our decomposition method would then determine what homodyne local - oscillator angles to choose in order to realize a given hamiltonian in a deterministic fashion .
currently available squeezing levels of almost 13db are , in principle , sufficient to perform up to 73 elementary teleportations in a nonclassical fashion ; four such elementary gaussian gates have been implemented already in a fully optical , spatial encoding using 5.5db - squeezed light sources @xcite . in summary , we proposed a systematic framework to decompose arbitrary cv unitaries ( from arbitrary multi - mode hamiltonians ) into an experimentally realizable elementary gate set .
different from previous proof - of - principle demonstrations , our treatment brings the abstract notions of decomposition theory for cv quantum computation close to experimental implementations .
we acknowledge support from the emmy noether program of the dfg in germany and thank nick menicucci and akira furusawa for useful discussions .
20 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 m. a. nielsen and i. l. chuang , _ quantum computation and quantum information _ , cambridge ( 2000 ) . , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) .
r. raussendorf and h. j. briegel , phys .
rev .
lett . * 86 * , 5188 ( 2001 ) .
n. c. menicucci _ et al .
_ , phys .
rev .
lett . * 97 * , 110501 ( 2006 ) . , * * , ( ) . , * * , ( )
. a. b. klimov , l. l. snches - soto , and j. delgado , opt .
communic .
* 191 * , 419 ( 2001 ) .
, , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) .
d. gottesman , a. kitaev , and j. preskill , phys .
rev .
a * 64 * , 012310 ( 2001 ) .
s. ghose and b. c. sanders , j. of mod
. opt . * 54 * , 855 ( 2007 ) .
p. marek , r. filip , and a. furusawa , arxiv:1105.4950 .
n. c. menicucci , x. ma , and t. c. ralph , phys .
rev .
lett .
* 104 * , 250503 ( 2010 ) .
a. j. roncaglia _ et al .
_ , phys .
rev .
a * 83 * , 062332 ( 2011 ) .
m. gu _ et al .
_ , phys .
rev .
a * 79 * , 062318 ( 2009 ) .
m. d. eisaman _ et al .
_ , nature * 438 * , 837 ( 2005 ) .
t. chaneliere _ et al .
_ , nature * 438 * , 833 ( 2005 ) .
b. julsgaard _ et al .
_ , nature * 432 * , 482 ( 2004 ) .
j. fiursek , phys .
rev .
a * 68 * , 022304 ( 2003 ) .
z. kurucz and m. fleischhauer , phys .
rev .
a * 78 * , 023805 ( 2008 ) .
r. ukai _ et al .
_ , phys .
rev .
lett . * 106 * , 240504 ( 2011 ) . , * * , ( ) . , * * , ( ) . , _ _ ( , ) . , * * , ( ) . |
grb030329/sn2003dh @xcite and grb980425/sn1998bw @xcite show that type ib / c supernovae are the parent population of long grbs .
type ib / c sne are believed to represent core - collapse events of massive stars in compact binaries @xcite .
they are probably part of a continuous sequence adjacent to type ii sne , ordered by increasing compactness of the binary in which the hydrogen ( ib / c ) and the helium ( ic ) envelope are removed in a common envelope phase @xcite .
the remaining naked star rotates rapidly at the orbital period by tidal spin - up . as the inactive iron - core succumbs to its own weight and that of the surrounding he - envelope , a rotating black hole nucleates during core - collapse @xcite .
some of the binding energy liberated during gravitational collapse will be channeled to eject matter , producing an accompanying hydrogen ( and helium ) deficient type ib ( type ic ) supernova @xcite .
the branching ratio of type ib / c sne to grb - sne can be calculated from the ratio @xmath2 of observed grbs - to - type ii supernovae @xcite , a beaming factor of 450 @xcite to 500 @xcite and a rate of about 0.2 of type ib / c - to - type ii supernovae @xcite , giving @xmath3=\frac{n(\mbox{grb - sne})}{n(\mbox{type~ib / c } ) } \simeq ( 2 - 4)\times 10^{-3}. \label{branch}\end{aligned}\ ] ] this ratio is remarkably small , suggesting a higher - order down - selection process .
the small branching ratio ( [ branch ] ) can be attributed to various factors in the process of creating grbs in type ib / c supernovae @xcite , e.g , not all baryon poor jets successfully punch through the remnant stellar envelope @xcite , and not all massive progenitors making type ib / c supernovae nucleate rapidly rotating black holes .
it is unlikely that either one of these down - selection processes by itself accounts for the smallness of @xmath1 .
rather , a combination of these might effectively contribute to a small branching ratio . by tidal interaction with the companion star ,
the naked star is not spherical prior to collapse .
black holes nucleated in nonspherical collapse possess recoil by bekenstein s gravitational radiation recoil mechanism @xcite .
tidal deformation produces a systematic recoil velocity , which may combine with random multipole mass - moments to produce a distribution in recoil velocities . some of the black holes will leave the central high - density core prematurely , before completion of the stellar collapse process .
these events are _ decentered_. other holes will remain centered and surge into a high - mass object surrounded by a high - density accretion disk or torus .
these events are _
centered black holes becomes luminous in a state of suspended accretion , if they rotate rapidly .
they spin down against emissions in gravitational radiation and other radiation channels @xcite .
the latter includes a burst in high - energy radiation from torus winds which radiatively drives a supernova @xcite , closely related to @xcite . here
, we quantify the various stages in the nucleation of black holes in stellar collapse .
we favor an association with binaries @xcite based on the type ii / ib event sn1993j @xcite and the proposed association of grb - supernovae remnants with soft x - ray transients @xcite .
we shall identify a branching ratio of core - collapse events producing centered nucleation of black holes with the probability of low kick velocities based on the bekenstein recoil mechanism . a related but different mechanism for explaining the small branching ratio based on kick velocities in core - collapse
poses fragmentation into two or more objects @xcite . in this scenario , grbs are associated with the formation of a fireball in the merger of binaries possessing small kick velocities .
it is motivated , in part , in the search for delay mechanisms in creating a grb , after the onset of the supernova on the basis of x - ray line - emissions in grb011211 . however
, x - ray line - emissions produced in radiatively powered supernovae allow the same time - of - onset of the grb and the supernova , obviating the need for any delay mechanism @xcite .
rotating black holes are described by kerr @xcite . in core - collapse of massive stars , rotating black holes nucleate by accumulation of mass and angular momentum from infalling matter .
the kerr solution describes the constraint @xmath4 for a black hole of mass @xmath5 and angular momentum @xmath6 , where @xmath7 is newton s constant and @xmath8 is the velocity of light .
table i summarizes the key quantities of kerr black holes .
quite generally , initial collapse of a rotating core produces a torus @xcite , which initially satisfies @xmath9 thus , the nucleation of black holes takes place through a _ first - order _ phase - transition : a torus forms of increasing mass by accumulation of matter , diluting its angular momentum until it satisfies ( [ eqn_jm ] ) and collapses into an extremal black hole .
the alternative of a second - order phase transition which initially forms a sub - solar mass black hole , requires rapid shedding of excess angular momentum by gravitational radiation .
however , limited mass - densities in core - collapse probably render this mechanism ineffective in competition with mixing on the free - fall timescale of the core . nevertheless , gravitational radiation emitted from a non - axisymmetric torus prior to the nucleation of the black hole is potentially interesting @xcite .
gravitational radiation in the formation of black holes through a first - order phase transition is important in non - spherical collapse , even when its energy emissions are small relative to the initial mass of the black hole .
the bekenstein gravitational radiation - recoil mechanism operates already in the presence of initial asphericities of about @xmath10 , producing a recoil of 300km / s or less .
the radius of the accretion disk or torus around a newly formed stellar mass black hole is @xmath11 cm .
a torus of a few tenths of a solar mass forms by accumulation of matter spiralling in , compressed by a factor @xmath12 as it stalls against the angular momentum barrier outside the inner most stable circular orbit ( isco ) .
the time - of - collapse of stellar matter from a radius @xmath13 is about the free - fall timescale , @xmath14 where @xmath15 denotes the mass of the progenitor he - star .
the newly formed low - mass black hole is typically kicked out of the central high - density region into surrounding lower - density regions _ before _ core - collapse is completed .
it continues to grow off - center by accretion of relatively low - density matter a high - density accretion disk never forms . with low but non - zero probability , the black hole has small
recoil , remains centered and surges into a high mass black hole surrounded by a high - density torus .
after nucleation of the black hole , an accretion disk may form provided that the specific angular momentum @xmath16 of infalling matter exceeds that of the inner most stable circular orbit ( isco ) , @xmath17 on the angular momentum @xmath6 of a black hole of mass @xmath5 . here
, @xmath18 denotes the dimensionless specific angular momentum of matter in circular orbits on the isco , where @xmath19 denotes the specific angular momentum of the black hole .
explicitly , we have @xcite @xmath20 in terms of @xmath21^{1/2}$ ] with @xmath22,~~ z_2=\left(3q^2+z_1 ^ 2\right)^{1/2}$ ] and @xmath23 .
notice that @xmath24 , between an extremal black hole @xmath25 ) and a non - rotating black hole @xmath26 ) .
the evolution of the newly nucleated black hole continues to be governed by angular momentum loss of the surrounding matter , until the inequality in ( [ eqn_jl ] ) is reversed .
the black hole rapidly grows uninhibited , while the inequality ( [ eqn_jl ] ) is reversed ( @xmath27 . this _ surge _ continues until once again ( [ eqn_jl ] ) holds . in dimensionless form , ( [ eqn_jl ] ) becomes @xmath28 where @xmath29 in terms of the dimensionless integrals @xmath30 and @xmath31 of the normalized lane - emden density distribution with @xmath32 at the origin and the zero @xmath33 at @xmath34 . here , @xmath35 in cylindrical geometry
for which @xmath36 , and @xmath37 in spherical geometry for which @xmath38 ; @xmath39 denotes the binary period in days , @xmath40 denotes the radius in units of the solar radius @xmath41 cm @xcite , and @xmath15 the mass of the progenitor he - star .
1 shows the solutions as a function of dimensionless period @xmath42 .
the upper branch shows that rapidly spinning black holes plus accretion disk form in small - period binaries @xcite , following a surge for periods beyond the bifurcation points @xmath43 @xmath44 the resulting mass and energy fractions as a function of @xmath42 are shown in fig .
these two geometries serve to bound the range of values in more detailed calculations , e.g. , through multi - dimensional numerical simulations . _ the bardeen trajectory _ corresponds to continuing accretion beyond surge , wherein matter remaining in the remnant envelope forms an accretion disk outside the isco . at this point ,
magnetohydrodynamical stresses within the disk as well as disk winds may drive continuing accretion .
accretion from the isco onto the black hole further increases the black hole mass and spin according to @xmath45 @xcite , generally causing spin - up towards an extremal state of the black hole . in fig . 1
this is indicated by accretion _
upwards _ beyond the upper isco - branch .
_ radiative spin - down _ corresponds to a long - duration burst of gravitational radiation emitted by a non - axisymmetric torus @xcite , described by a frequency and energy @xmath46 where @xmath47 , and @xmath48 and @xmath49 denote the relative mass and angular velocity of the torus .
this takes place if the torus is uniformly magnetized with the remnant magnetic field of the progenitor star . in fig .
1 , this radiative spin - down is indicated by a transition _ downwards _ from the upper isco branch to the branch on which the angular velocities of the black hole and of matter at the isco match ( @xmath50 and @xmath51 ) .
this radiative transition lasts for the lifetime of rapid spin of the black hole a dissipative timescale of tens of seconds @xcite .
additional matter accreted is either blown off the torus in its winds , or accumulates and accretes onto the black hole after spin - down .
in what follows , we shall consider a two - dimensional gaussian distribution of kick velocities in the equatorial plane associated with the the tidal deformation of the progenitor star by its companion , assuming a velocity dispersion @xmath52km / s in bekenstein s recoil mechanism .
the probability of centered nucleation during @xmath53 s is that of a kick velocity @xmath5410km / s , i.e. : @xmath55 while the numerical value has some uncertainties , the selection mechanism by gravitational radiation - recoil effectively creates a small probability of centered nucleation .
we identify the branching ratio of type ib / c sne into grbs with the probability of centered nucleation , @xmath3=p_{c}\simeq 0.5\%,\end{aligned}\ ] ] effectively creating a small , higher - order branching ratio .
the proposed centered and decentered core - collapse events predicts a differentiation in gravitational wave - signatures .
these signatures are of interest to the newly commissioned gravitational wave - detectors ligo , virgo and tama , both as burst sources and through their collective contributions to the stochastic background in gravitational radiation @xcite .
the black hole nucleation process is accompanied by a short burst in gravitational radiation , specifically in response to non - axisymmetric toroidal structures and fragmentation @xcite .
its gravitational radiation signature depends on details of the hydrodynamical collapse .
centered nucleation is followed by a long burst in gravitational radiation .
a short burst in gravitational radiation is hereby common to all type ib / c supernovae .
this may apply to type ii events as well .
type ii events are possibly associated with low spin - rates and could represent delayed core - collapse via an intermediate nucleon " star ( sn1987a , e.g. @xcite ) .
their gravitational wave - emissions are thereby essentially limited to that produced by kick ( if any ) and collapse of this nucleon star . for a recent review of the short - duration ( @xmath56s ) bursts of gravitational waves in core - bounce , more closely related to type ii supernovae ,
see @xcite and references therein . on account of ( [ eqn_aa1]-[eqn_aa2 ] ) ,
core - collapse compact can produce high - mass , rapidly spinning black holes in centered nucleation , whose rotational energy can reach about one - half the maximal spin - energy of a kerr black hole . in a suspended accretion state , these black holes spin - down in the process of emitting a long - duration burst of tens of seconds in gravitational waves @xcite .
this long duration burst comes as a second burst , after the short burst of gravitational radiation in centered nucleation .
to conclude , type ib / c supernovae produce a short , single burst of gravitational radiation at birth of a low - mass black hole .
the sub - population of grb - supernovae produce a subsequent long burst of gravitational radiation representing spin - down of the black hole .
the second burst takes place after a quiescent or subluminous @xcite surge of the black hole into a high - mass object .
* acknowledgement . *
the author thanks p. saulson , j. bekenstein and the referee for constructive comments .
this research was completed at a joint apctp - tpi meeting at the university of alberta , and is supported by the ligo observatories , constructed by caltech and mit with funding from nsf under cooperative agreement phy 9210038 .
the ligo laboratory operates under cooperative agreement phy-0107417 .
this paper has been assigned ligo document number ligo - p040014 - 00-r .
bardeen , j.m . , 1970 ,
nature , 226 , 64 bardeen j.m . , press w.h .
, teukolsky s.a . , 1972 , 178 , 347 bekenstein , j. , 1973 , apj , 183 , 657 bethe , h.a . ,
brown , g.e . , & lee , c .- h . , 2003 , selected papers : formation and evolution of black holes in the galaxy ( world scientific ) brown , g.e .
, lee , c .- h . , wijers r.a.m.j . ,
lee , h.k . , israelian g. & bethe h.a . , 2000 , newa , 5 , 191 cappellaro , e. , barbon , r. , & turatto , m. , 2003 , astro - ph/0310859 davies , m.b . , king , a. , rosswog , s. , & wynn , g. , 2002 , apj , 579 , l63 duez , m.d . ,
shapiro , s.l .
, & yo , h .- j . , 2004 , gr - qc/0401076 frail , d.a . , et al . , 2001 , apj , 562 , l55 fryer , c.l . ,
holz , d.e .
, & hughes , s.a . , 2004 , astro - ph/0403188 galama , t.j . , et al . 1998 , nature , 395 , 670 ghisellini , g. , lazatti , d. , rossi , e. , & rees , m.j . , 2002 , a&a , 389 , l33 hjorth , j. , et al . , 2003 ,
nature , 423 , 847 hflich , p.j . ,
wheeler , j.c .
, & wang , l. , 1999 , apj , 521 , 179 kerr , r.p . , 1963 ,
lett . , 11 , 237 kippenhahn , r. & weigert , a. 1990 , stellar structure and evolution ( springer - verlag : new york ) , p178 kouveliotou , c. , et al . , 1993 , apj , 413 , l101 lee , c .- h . ,
brown , g.e . , &
wijers , r.a.m.j . , 2002 , apj , 575 , 996 macfadyen , a.i . and woosley , s.e . , 1999 , apj , 524 , 262 macfadyen , a.i . , 2003 , astro - ph/0301425 maund , j.r . ,
smartt , s.j . ,
kudritzki , r.p . ,
podsiadlowski , p. , & gilmore , g.f . , 2004 , nature , 427 , 129 mineshige , s. , hosokawa , t. , machida , m. , & matsumoto , r. , 2002 , pasj , 54 , 655 nomoto , k. , iwamoto , k. , & suzuki , t , 1995 , phys .
256 , 173 paczyski , b. , 1998 , apj , 494 , l45 podsiadlowski , ph . ,
mazzali , p.a . ,
nomoto , k. , lazzati , d. , & cappellaro , e. , 2004 , apj , 607 , l17 popov , s.b .
, 2004 , astro - ph/0403710 porciani , c. , & madau , p. , 2001 ,
apj , 548 , 522 reeves , j.n . , et al . , 2002 , nature , 416 , 512 turatto , m. , 2003 , in supernovae and gamma - ray bursters , ed .
weiler ( springer - verlag , heidelberg ) , p21 van putten m.h.p.m . , & levinson , a. , 2003 , apj , 584 , 937 van putten m.h.p.m .
, levinson , a. , regimbau , t. , punturo , m. , & herry , g.m .
, 2004 , phys .
d. 69 , 044007 van putten , m.h.p.m . , 1999 , science , 284 , 115 van putten , m.h.p.m . ,
& regimbau , t. , 2003 , apj , 593 , l15 rees , m.j . ,
ruffini , r. , wheeler , j.a . , 1974 , _ black holes , gravitational waves and cosmology : an introduction to current research _ ( gordon & breach , new york ) , section 7 stanek , k.z . , et al . 2003 , apj , 591 , l17 woosley , s.e . , 1993 ,
apj , 405 , 273 | we study centered and decentered nucleation of black holes in core - collapse of massive stars in binaries . by bekenstein
s gravitational - radiation recoil mechanism , a newly nucleated black hole typically leaves the central core prematurely . with low probability ,
the black hole remains centered and matures to a high - mass black hole which spins rapidly if the binary is compact .
grb030329/sn2003dh demonstrates that type ib / c supernovae are the parent population of long grbs , whose branching ratio is @xmath0 .
we identify @xmath1 with the low probability of centered nucleation in compact binaries .
decentered events are predicted to produce a single short burst in gravitational radiation .
centered events are predicted to produce a second , long - burst in gravitational radiation powered by a luminous black hole .
these signatures are of interest to ligo , virgo and tama . |
let @xmath0 be an algebraically closed field of characteristic @xmath1 . in sharp contrast with the situation in characteristic @xmath12 ,
there exist galois covers @xmath7 ramified only over infinity . by abhyankar s conjecture @xcite , proved by raynaud and harbater @xcite , @xcite
, a finite group @xmath2 occurs as the galois group of such a cover @xmath13 if and only if @xmath2 is quasi-@xmath6 , i.e. , @xmath2 is generated by @xmath6-groups .
this result classifies all the finite quotients of the fundamental group @xmath14 .
it does not , however , determine the profinite group structure of @xmath14 because this fundamental group is an infinitely generated profinite group .
there are many open questions about galois covers @xmath7 ramified only over infinity . for example , given a finite quasi-@xmath6 group @xmath2 , what is the smallest integer @xmath15 for which there exists a cover @xmath7 ramified only over infinity with @xmath9 of genus @xmath15 ?
as another example , suppose @xmath2 and @xmath16 are two finite quasi-@xmath6 groups such that @xmath16 is a quotient of @xmath2 .
given an unramified galois cover @xmath17 of @xmath18 with group @xmath16 , under what situations can one dominate @xmath17 with an unramified galois cover @xmath13 of @xmath18 with galois group @xmath2 ?
answering these questions will give progress towards understanding how the finite quotients of @xmath14 fit together in an inverse system .
these questions are more tractible for quasi-@xmath6 groups that are @xmath6-groups since the maximal pro-@xmath6 quotient @xmath19 is free ( of infinite rank ) @xcite . in this paper , we study galois covers @xmath7 ramified only over @xmath8 whose galois group is a semi - direct product of the form @xmath3 , where @xmath5 is a prime distinct from @xmath6 .
such a cover @xmath13 must be a composition @xmath20 where @xmath21 is unramified and @xmath22 is an artin - schreier cover ramified only over @xmath8 .
the cover @xmath17 has an affine equation @xmath23 for some @xmath24 $ ] with degree @xmath25 prime - to-@xmath6 .
the @xmath5-torsion @xmath26 $ ] of the jacobian of @xmath27 is isomorphic to @xmath28 . when @xmath29 , we determine how an automorphism @xmath30 of @xmath27 of order @xmath6 acts on @xmath26 $ ] .
this allows us to construct a galois cover @xmath31 ramified only over @xmath8 which dominates @xmath17 , such that the galois group of @xmath32 is @xmath33 where @xmath10 is the order of @xmath5 modulo @xmath6 ( section [ exist ] ) .
we prove that the genus of @xmath34 is minimal among all natural numbers that occur as the genus of a curve @xmath9 which admits a covering map @xmath7 ramified only over @xmath8 with galois group of the form @xmath3 .
we also prove that the number of curves @xmath9 of this minimal genus which admit such a covering map is at most @xmath11 when @xmath6 is odd ( section [ unique ] ) .
we recall which groups occur as galois groups of covers of @xmath35 ramified only over @xmath8 .
[ quasi - p ] a finite group is a quasi @xmath6-group if it is generated by all of its sylow @xmath6-subgroups .
it is well - known that there are other equivalent formulations of the quasi-@xmath6 property , such as the next result .
[ lem : quasi - p ] a finite group is a quasi @xmath6-group if and only if it has no nontrivial quotient group whose order is relatively prime to @xmath6 .
the importance of the quasi-@xmath6 property is that it characterizes which finite groups occur as galois groups of unramified covers of the affine line . [ tabconj ] a finite group occurs as the galois group of a galois cover of the projective line @xmath35 ramified only over infinity if and only if it is a quasi-@xmath6 group .
this is a special case of abhyankar s conjecture @xcite which was jointly proved by harbater @xcite and raynaud @xcite .
we now restrict our attention to groups @xmath2 that are semi - direct products of the form @xmath3 .
the semi - direct product action is determined by a homomorphism @xmath36 .
[ lgroup ] suppose a quasi-@xmath6 group @xmath2 is a semi - direct product of the form @xmath3 for a positive integer @xmath4 . 1 .
then @xmath2 is not a direct product .
2 . moreover , @xmath37 where @xmath38 is the order of @xmath5 modulo @xmath6 .
part ( 1 ) is true since @xmath39 can not be a quotient of the quasi-@xmath6 group @xmath2 . for part ( 2 ) , the structure of a semi - direct product @xmath3 depends on a homomorphism @xmath40 . by part ( 1 )
, @xmath41 is an inclusion .
thus @xmath42 has an element of order @xmath6 .
now @xmath43 thus @xmath44 for some positive integer @xmath45 which implies @xmath37 .
[ lexistgb ] if @xmath46 , then there exists a semi - direct product of the form @xmath33 which is quasi-@xmath6 .
it is unique up to isomorphism .
if @xmath46 , then there is an element of order @xmath6 in @xmath47 and so there is an injective homomorphism @xmath48 .
thus there exists a non - abelian semi - direct product @xmath2 of the form @xmath33 . to show that @xmath2 is quasi-@xmath6 , suppose @xmath49 is a normal subgroup of @xmath2 whose index is relatively prime to @xmath6
. then @xmath49 contains an element @xmath30 of order @xmath6 . by (
* , thm .
9 ) , since @xmath2 is not a direct product and @xmath50 is normal in @xmath2 , the subgroup @xmath51 is not normal in @xmath2 .
thus @xmath51 is a proper subgroup of @xmath49 .
it follows that @xmath5 divides @xmath52 and so @xmath49 contains an element @xmath53 of order @xmath5 by cauchy s theorem .
recall that @xmath54 contains no element of order @xmath6 for any positive integer @xmath55 .
thus the group generated by the conjugates of @xmath53 under @xmath30 has order divisible by @xmath56 .
thus @xmath57 and @xmath2 has no non - trivial quotient group whose order is relatively prime to @xmath6 . by lemma [ lem : quasi - p ] , @xmath2 is quasi-@xmath6 .
the uniqueness follows from ( * ? ? ?
* lemma 6.6 ) .
in this section , we give concrete examples of galois covers @xmath7 ramified only over @xmath8 with galois group of the form @xmath33 . to compute the genus of the covering curve @xmath9
, we will need to determine the higher ramification groups of @xmath13 .
let @xmath58 be a galois extension of function fields of curves with galois group @xmath2 and let @xmath59 be primes of @xmath60 and @xmath61 such that @xmath62 .
let @xmath63 and @xmath64 be the corresponding valuation function and valuation ring for @xmath65 . for any integer @xmath66 ,
the @xmath67th ramification group of @xmath62 is @xmath68 [ lem : genus ] suppose @xmath24 $ ] is a polynomial of degree @xmath25 for a positive integer @xmath25 prime to @xmath6 .
let @xmath69 be the cover of curves corresponding to the field extension @xmath70/ ( y^p - y - f(x)).\ ] ] 1 .
then @xmath69 is a galois cover with galois group @xmath71 ramified only at the point @xmath72 over @xmath8 .
2 . the @xmath67th ramification group at @xmath72 satisfies @xmath73 3 . the genus @xmath74 of @xmath27 is equal to @xmath75 for part ( 1 ) , note that the extension @xmath76/ ( y^p - y - f(x))$ ] is cyclic of degree @xmath6 , with galois group generated by the automorphism @xmath77 of order @xmath6 .
let @xmath78 be a finite prime of @xmath79 and let @xmath80 be the corresponding valuation .
then @xmath81 , hence @xmath78 is unramified by ( * ? ? ?
iii.7.8(b ) ) . for the infinite prime @xmath8 with corresponding valuation @xmath82
, we have @xmath83 for all @xmath84 $ ] thus @xmath85 is totally ramified by ( * ? ? ?
iii.7.8(c ) ) . to prove part
( 2 ) , we note that furthermore @xmath86 which implies that @xmath87 now let @xmath88 be the completion of the valuation ring of @xmath89/(y^p - y - f(x))$ ] at @xmath85 , and let @xmath90 be a generator of the unique prime in @xmath88 .
then write @xmath91 , where @xmath92 is a unit in @xmath93 $ ]
. since @xmath0 is algebraically closed , @xmath94{u } \in \widehat{\theta}$ ] , and so @xmath94{y } \in \widehat{\theta}$ ] .
after possibly changing @xmath90 , we can assume without loss of generality that @xmath94{y } = \pi_\infty^{-1}$ ] .
recalling that @xmath30 acts on @xmath95 by @xmath96 , we have @xmath97 thus @xmath98 , which completes the proof of part ( 2 ) . to find the genus @xmath74 of @xmath27 for part ( 3 ) , we make use of the riemann - hurwitz formula @xmath99 where @xmath100 denotes the @xmath67th ramification group at @xmath72 , ( * ? ?
7.27 & 11.72 ) ) . from part
( 2 ) , we then obtain that @xmath101 .
recall the following facts about the @xmath6th cyclotomic polynomial @xmath102 , which is the minimal polynomial over @xmath103 of a primitive @xmath6th root of unity @xmath104 .
now @xmath105 is a galois extension of @xmath106 , unramified over @xmath5 since @xmath107 , and all primes over @xmath5 have the same residue field degree .
the irreducible factors of @xmath108 modulo @xmath5 are in one - to - one correspondence with the primes of @xmath109 $ ] over @xmath5 , and each of their degrees is equal to the residue field degree of the corresponding prime over @xmath5 .
the latter equals the order @xmath110 of @xmath5 modulo @xmath6 ( * ? ? ?
12.2 , exercise # 20 ) .
we shall soon explicitly construct a cover of @xmath35 ramified only over @xmath8 with galois group @xmath33 .
but before we do so , we start with a specific example .
let @xmath6 be an odd prime .
consider the artin - schreier cover @xmath111 corresponding to the field extension @xmath76/ ( y^p - y - x^2)$ ] .
by lemma [ lem : genus](3 ) , the genus of @xmath112 is @xmath113 .
let @xmath114 be the jacobian of @xmath27 .
the automorphism @xmath30 of @xmath27 given by @xmath96 defines an automorphism of @xmath114 of order @xmath6 .
now we describe the action of @xmath30 on the subgroup @xmath115 $ ] of 2-torsion points of @xmath114 explicitly .
note that since @xmath116 , then @xmath115 $ ] is isomorphic to @xmath117 by ( * ? ? ?
thus we can represent @xmath30 as an element of @xmath118 . for @xmath119 ,
let @xmath120 denote the closed point of @xmath27 at which the function @xmath121 vanishes . for each @xmath67 ,
the divisors @xmath120 and @xmath122 on @xmath27 can be identified with elements of @xmath114 .
let @xmath123 be the identity element of @xmath114 , i.e. , the linear equivalence class of principal divisors .
then the divisor @xmath124 is equivalent to @xmath123 since @xmath125 .
moreover since @xmath126 is equivalent to @xmath12 , we have @xmath127 $ ] with the only relation @xmath128 . in particular , @xmath129 form a basis of @xmath115 $ ] .
with respect to this basis , the action of @xmath30 can be represented by the @xmath130-matrix @xmath131 the characteristic polynomial of @xmath30 is @xmath132 $ ] , which factors into irreducible polynomials each of degree equaling the order of @xmath133 modulo @xmath6 . in particular ,
@xmath30 acts irreducibly on @xmath115 $ ] if and only if @xmath133 is a primitive root modulo @xmath6 , i.e. , if and only if @xmath6 is an artin prime .
for example , if @xmath134 , then @xmath30 acts irreducibly on @xmath115 $ ] with minimal polynomial @xmath135 . if @xmath136 , then @xmath133 has order @xmath137 modulo @xmath138 and the factorization of @xmath139 into irreducible polynomials is @xmath140 modulo @xmath133 .
thus the action of @xmath30 on @xmath115 $ ] can be represented by the @xmath141-matrix @xmath142 where @xmath143 and @xmath144 are the irreducible @xmath137-dimensional companion matrices of @xmath145 and @xmath146 respectively .
for the rest of the paper , let @xmath147 be the artin - schreier cover corresponding to the field extension @xmath70/ ( y^p - y - x^s).\ ] ] we show that @xmath148 can be dominated by a galois cover of @xmath35 with galois group of the form @xmath33 for @xmath10 equal to the order of @xmath5 modulo @xmath6 .
[ pexist ] let @xmath25 and @xmath5 be primes distinct from @xmath6 .
let @xmath149 be the artin - schreier cover with affine equation @xmath150 .
let @xmath110 be the order of @xmath5 modulo @xmath6 .
then there exists an unramified galois cover @xmath151 with galois group @xmath152 such that @xmath153 is a galois cover of @xmath35 ramified only over @xmath8 whose galois group is a semi - direct product of the form @xmath154 . by lemma [ lem : genus](1 )
, @xmath155 is a galois cover with galois group @xmath71 ramified only at the point @xmath72 over @xmath8 .
the genus @xmath156 of @xmath157 is @xmath158 .
consider two commuting automorphisms of @xmath157 defined by @xmath159 let @xmath160 be the jacobian of @xmath157 .
then @xmath30 and @xmath161 define commuting automorphisms of @xmath160 of orders @xmath6 and @xmath25 respectively .
therefore , @xmath162 contains a ring isomorphic to @xmath163 \cong { \mathbb{z}}[\zeta_{ps}]$ ] , which is a @xmath164-module of rank @xmath165 . then @xmath166 is contained in @xmath167 . in other words
, @xmath160 has complex multiplication by @xmath166 . for a prime @xmath5 distinct from @xmath6 ,
the automorphism @xmath30 induces an action on the subgroup @xmath168 $ ] of @xmath5-torsion points of @xmath160 .
recall that there is a bijection between @xmath5-torsion points @xmath169 of @xmath160 and unramified @xmath170-galois covers @xmath171 ( * ? ? ?
also @xmath169 has order @xmath5 if and only if @xmath172 is connected .
this induces a bijection between orbits of @xmath30 on the set of unramified @xmath170-galois covers @xmath171 and on the set of @xmath5-torsion points of @xmath160 . for a point @xmath169 of order @xmath5 of @xmath160 ,
consider the compositum @xmath173 of all of the conjugates @xmath174 for @xmath175 : @xmath176 \ar[dl ] \ar[d ] \ar[drr ] & & \\
z_d \ar[drr]_{({\mathbb z}/\ell { \mathbb z } ) } & z_{\tau(d ) } \ar[dr]^{({\mathbb z}/\ell { \mathbb z } ) } & z_{\tau^2(d ) } \ar[d]^{({\mathbb z}/\ell { \mathbb z } ) } & \ldots & z_{\tau^{p-1}(d ) } \ar[dll]^{({\mathbb z}/\ell { \mathbb z})}\\ & & y_s & & } \ ] ] then @xmath9 is invariant under @xmath30 and so @xmath177 is galois .
moreover , @xmath178 is the galois closure of @xmath179 .
suppose there is a non - trivial one - dimensional @xmath30-invariant subspace of @xmath168 $ ] with eigenvalue @xmath180 ; i.e. @xmath30 acts trivially on this subgroup of order @xmath5 .
this yields a cover @xmath181 .
since the action of @xmath30 is trivial , @xmath182 is galois , ramified only over @xmath8 , with abelian galois group @xmath183 .
this contradicts lemma [ lgroup ] .
since @xmath30 has order @xmath6 , the minimal polynomial @xmath184 of @xmath30 divides @xmath185 in @xmath186 $ ] . from the preceding paragraph
, there is no non - trivial one - dimensional @xmath30-invariant subspace of @xmath168 $ ] with eigenvalue @xmath180 .
this implies that @xmath184 divides the @xmath6th cyclotomic polynomial @xmath187 in @xmath186 $ ] .
the irreducible factors of @xmath108 in @xmath186 $ ] all have degree @xmath10 .
thus the degree of @xmath184 equals @xmath10 . since @xmath188
, we have @xmath168 \cong ( { \mathbb{z}}/\ell{\mathbb{z}})^{(p-1)(s-1)}$ ] , so we can represent @xmath30 as an element of @xmath189 .
we can choose a basis of @xmath168 $ ] such that @xmath30 is represented as an element of @xmath189 in block form .
the first irreducible subrepresentation of @xmath30 has dimension @xmath10 .
moreover , since @xmath166 is a galois extension of @xmath103 , the block form of @xmath30 consists entirely of irreducible blocks of the same size .
in particular , the number of irreducible blocks is @xmath190 . in other words ,
@xmath30 can be represented by an element of @xmath191 of the form @xmath192 where @xmath193 is an @xmath194 matrix representing an @xmath10-dimensional irreducible subrepresentation of @xmath30 on @xmath168 $ ] . using the bijection between orbits of @xmath168 $ ] and orbits of @xmath195-covers of @xmath157 under @xmath30 and the above observation for the action of @xmath30 on @xmath168 $ ]
, there exists an unramified @xmath196-galois cover @xmath151 such that @xmath153 is a galois cover of @xmath35 with galois group of the form @xmath154 .
also @xmath32 is ramified only over infinity since @xmath148 is ramified only over @xmath8 and since @xmath197 is unramified .
in this section , we find the minimal genus of a curve @xmath9 that admits a covering map @xmath198 ramified only over @xmath8 , with galois group of the form @xmath3 .
the minimal genus depends only on @xmath5 and @xmath6 .
we consider the cases @xmath6 odd and @xmath199 separately .
we also prove that the number of curves @xmath9 of this minimal genus which admit such a covering map is at most @xmath11 when @xmath6 is odd and at most @xmath200 when @xmath199 . the following lemma will be useful .
[ lunram ] let @xmath2 be a semi - direct product of the form @xmath3 where @xmath5 is a prime distinct from @xmath6 .
if @xmath7 is a galois cover ramified only over @xmath8 with galois group @xmath2 , then the subcover @xmath21 with galois group @xmath201 is unramified . the quotient of @xmath2 by the normal subgroup @xmath202 is @xmath203 .
thus the cover @xmath13 is a composition @xmath20 where @xmath204 has galois group @xmath203 and is totally ramified at the unique point @xmath72 over @xmath8 and where @xmath21 has galois group @xmath49 and is branched only over @xmath72 . then @xmath197 is a prime - to-@xmath6 abelian cover of @xmath27 .
let @xmath15 be the genus of @xmath27 .
then by ( * ? ? ?
* xiii , cor . 2.12 ) , the prime - to-@xmath6 fundamental group of @xmath205 is isomorphic to the prime - to-@xmath6 quotient @xmath206 of the free group on generators @xmath207 subject to the relation @xmath208 = c^{-1}$ ] .
the cover @xmath197 corresponds to a surjection of @xmath206 onto @xmath49 where @xmath209 maps to the canonical generator of inertia @xmath210 of a point of @xmath9 over @xmath72 .
thus @xmath49 is generated by elements @xmath211 subject to the relation @xmath212 = \gamma^{-1}$ ] .
since @xmath49 is abelian and so @xmath197 is unramified . 1 .
there exists a galois cover @xmath31 ramified only over @xmath8 whose galois group is a semi - direct product of the form @xmath33 such that @xmath214 .
the integer @xmath215 is the minimal genus of a curve @xmath9 which admits a covering map @xmath7 ramified only over @xmath8 with galois group of the form @xmath3 for any positive integer @xmath4 .
3 . there are at most @xmath11 isomorphism classes of curves @xmath9 which admit a galois covering map as in part ( 1 ) with minimal genus @xmath215 . by the construction in proposition [ pexist ] ,
there exists a galois cover @xmath31 ramified only over @xmath8 whose galois group is a semi - direct product of the form @xmath33 .
we compute the genus of the curve @xmath34 . recall that @xmath32 is a composition @xmath216 where @xmath217 is an unramified @xmath152-galois cover and @xmath218 has artin - schreier equation @xmath219 .
then @xmath112 has genus @xmath220 by lemma [ lem : genus](3 ) . by the riemann - hurwitz formula , @xmath221 , i.e. , @xmath222
this completes part ( 1 ) . for part ( 2 ) , suppose @xmath7 is a galois cover ramified only over @xmath8 with galois group of the form @xmath3 . if @xmath15 is the genus of @xmath9 , we will show that @xmath223 . as described in the proof of lemma [ lunram ] ,
the cover @xmath13 is a composition @xmath20 where @xmath22 has galois group @xmath224 and is ramified only over @xmath8 and where @xmath197 is unramified with group @xmath201 . by the riemann - hurwitz formula , @xmath225 . by artin - schreier theory , @xmath17 is given by an equation @xmath23 where @xmath226 $ ] has degree @xmath25 for some integer @xmath25 relatively prime to @xmath6 .
since the genus @xmath74 of @xmath27 is @xmath158 by lemma [ lem : genus ] ( 3 ) , we should make @xmath25 as small as possible .
the value @xmath227 is impossible since then @xmath27 is a projective line and there do not exist galois covers of the projective line ramified only over one point with galois group @xmath228 .
thus @xmath229 yields the smallest possible value for @xmath74 , namely @xmath230 . recall that @xmath231 by lemma [ lgroup ] .
thus @xmath232 . for part ( 3 ) , suppose @xmath7 is a galois cover ramified only over @xmath8 with galois group of the form @xmath33 and the genus of @xmath9 satisfies @xmath233 . as in part
( 2 ) , @xmath13 factors as @xmath234 where @xmath21 is an unramified @xmath152-galois cover , where @xmath204 is an artin - schreier cover ramified only over @xmath8 , and where @xmath27 has genus @xmath230 . by lemma [ lem : genus](3 )
, @xmath27 has an affine equation @xmath235 for some @xmath236 and @xmath237 . since @xmath6 is odd and @xmath0
is algebraically closed , it is possible to complete the square and write @xmath238 where @xmath239 . after modifying by an automorphism of the projective line , specifically by the affine linear transformation @xmath240 ,
the equation for @xmath27 can be rewritten as @xmath241 .
since @xmath0 is algebraically closed , there exists @xmath242 such that @xmath243 .
let @xmath244 .
after the change of variables @xmath245 , the curve @xmath27 is isomorphic to the curve @xmath112 with affine equation @xmath246 .
thus there is a unique possibility for the isomorphism class of the curve @xmath27 . from the proof of proposition [ pexist ] , there is a bijection between @xmath30-invariant connected unramified @xmath152-galois covers of @xmath112 and orbits of @xmath30 on points @xmath169 of order @xmath5 on @xmath247 .
the action of @xmath30 on @xmath248 $ ] decomposes into @xmath11 irreducible subrepresentations .
each of these is distinct , because the irreducible factors of @xmath249 $ ] are distinct .
thus there are @xmath11 choices for a @xmath30-invariant unramified @xmath152-galois cover of @xmath112 .
thus there are at most @xmath11 isomorphism classes of curves @xmath9 which admit a galois covering map as in part ( 1 ) with minimal genus @xmath215 . 1 .
there exists a galois cover @xmath7 ramified only over @xmath8 with galois group of the form @xmath250 .
the minimal genus of a curve @xmath9 which admits a covering map as in part ( 1 ) is @xmath251 .
3 . there are at most @xmath252 isomorphism classes of curves @xmath9 which admit a galois covering map as in part ( 1 ) with minimal genus @xmath251 .
note that the order of @xmath5 modulo @xmath133 is @xmath253 .
for part ( 1 ) , lemma [ lexistgb ] shows that there exists a semi - direct product of the form @xmath250 which is quasi-@xmath133 .
the result is then immediate from theorem [ tabconj ] .
suppose @xmath7 is a galois cover ramified only over @xmath8 with galois group as in part ( 1 ) .
as before , @xmath13 factors as a composition @xmath234 . where @xmath21 has galois group @xmath228 and @xmath204 is an artin - schreier extension with affine equation @xmath254 for some @xmath24 $ ] of odd degree @xmath25 .
by lemma [ lunram ] , @xmath197 is unramified .
the minimal genus for @xmath9 will thus occur when @xmath25 is as small as possible .
as before , @xmath227 is impossible , and so @xmath255 is the smallest choice . in this case , by lemma [ lem : genus](3 ) , @xmath256 , i.e. , @xmath27 is an elliptic curve . by the riemann - hurwitz formula ,
the minimal genus for @xmath9 is @xmath257 , which completes part ( 2 ) . for part ( 3 )
, since @xmath0 is algebraically closed , we can complete the cube of @xmath258 and make the corresponding change of variables , which is a scaling and translation of @xmath259 .
so we can assume that @xmath27 has affine equation @xmath260 for some @xmath261 .
then it follows from ( * ? ? ?
* appendix a , prop .
1.1c ) that the @xmath262-invariant of @xmath27 is @xmath263 and that the discriminant is @xmath264 .
since @xmath0 is algebraically closed , by ( * ? ? ?
* appendix a , prop .
1.2b ) , all elliptic curves @xmath27 with @xmath263 are isomorphic over @xmath0 .
thus there is a unique choice for @xmath27 up to isomorphism . without loss of generality
, we may assume that @xmath265 has affine equation @xmath266 . from the proof of proposition [ pexist ]
, the action of @xmath30 on @xmath267 $ ] decomposes into the direct sum of two @xmath180-dimensional subrepresentations .
in other words , the action of @xmath30 is diagonal with both eigenvalues equal to @xmath268 .
the number of non - trivial @xmath30-invariant subgroups of @xmath267 $ ] is the number of subgroups of order @xmath5 in @xmath269 , which is @xmath252 . as in theorem
[ exist - minimal ] , this implies that there are at most @xmath270 isomorphism classes of curves @xmath9 which admit a galois covering map as in part ( 1 ) with minimal genus @xmath251 .
springer - verlag , berlin , 1971 .
sminaire de gomtrie algbrique du bois marie 19601961 ( sga 1 ) , dirig par alexandre grothendieck . augment de deux exposs de m. raynaud , lecture notes in mathematics , vol . 224 | let @xmath0 be an algebraically closed field of characteristic @xmath1 .
let @xmath2 be a semi - direct product of the form @xmath3 where @xmath4 is a positive integer and @xmath5 is a prime distinct from @xmath6 . in this paper , we study galois covers @xmath7 ramified only over @xmath8 with galois group @xmath2 .
we find the minimal genus of a curve @xmath9 which admits a covering map of this form and we give an explicit formula for this genus in terms of @xmath5 and @xmath6 .
the minimal genus occurs when @xmath4 equals the order @xmath10 of @xmath5 modulo @xmath4 and we also prove that the number of curves @xmath9 of this minimal genus which admit such a covering map is at most @xmath11 when @xmath6 is odd . |
the recovery of sparse signals of high dimensions on the basis of noisy linear measurements is an important problem in the field of signal acquisition and processing .
when the number of linear observations is significantly lower than the dimension of the signal to be recovered , the signal recovery may exploit the property of sparsity to deliver correct results .
the field of research that studies such problems is often referred to as _ compressed sensing _ or _ compressive sensing _ ( cs ) @xcite .
+ several computationally tractable methods to address cs problems have been developed in the last two decades @xcite . among them , greedy methods prove to be valuable choices as their complexity is significantly lower than that of algorithms based on @xmath0-minimization @xcite . + while many cs problems involve only one sparse signal and the corresponding _ measurement vector _ , _
i.e. _ , the vector gathering all the linear observations of this signal , some applications either require or at least benefit from the presence of several sparse signals and measurement vectors .
examples of such applications are available in section [ subsec : applications ] .
models involving one measurement vector are referred to as single measurement vector ( smv ) models while multiple measurement vector ( mmv ) models involve at least two measurement vectors @xcite .
+ when the supports of the sparse signals are similar , it is possible to improve the reliability of the recovery by making joint decisions to determine the estimated support @xcite .
thereby , all the measurement vectors intervene in the estimation of the support and the final support is common to all the sparse vectors .
algorithms performing joint recovery are also capable to weaken the influence of additive measurement noise on the performance provided that the noise signals are statistically independent and exhibit some degree of isotropy . + orthogonal matching pursuit ( omp ) is one of the most extensively used greedy algorithm designed to solve smv problems @xcite . among several greedy algorithms
conceived to deal with multiple measurement vectors , the extension of omp to the mmv paradigm , referred to as simultaneous orthogonal matching pursuit ( somp ) , is of great interest as it remains simple , both conceptually and algorithmically @xcite .
the classical somp algorithm does not account for the possibly different measurement vector noise levels . in some sense
, all the measurement vectors are considered equally worthy .
however , it is clear that an optimal joint support recovery method should necessarily take into account the noise levels by accordingly weighting the impact of each measurement vector on the decisions that are taken .
the first aim of this paper is to extend somp by gifting it with weighting capabilities .
the new algorithm will be referred to as somp with noise stabilization ( somp - ns ) and basically extends the decision metric of somp to weight the impact of each measurement vector onto the decisions that are taken .
+ the second objective is to provide theoretical and numerical evidence that the proposed algorithm indeed enables one to achieve higher performance than the other greedy alternatives when the noise levels , or more generally the signal - to - noise ratios ( snr ) , vary from one measurement vector to another .
we study partial and full support recovery guarantees of somp - ns for a mmv signal model incorporating arbitrary sparse signals to be recovered and statistically independent additive gaussian noise vectors exhibiting diagonal covariance matrices , _
i.e. _ , the entries within each vector are statistically independent . it is assumed that the variances of the entries within each noise vector are identical although they may be different for each measurement vector . the signal model is thoroughly detailed in section [ subsec : signalmodel ] .
+ our first contribution is the proposal of somp - ns which generalizes somp by weighting the measurement vectors .
the second contribution is a novel theoretical analysis of somp and somp - ns in the presence of additive gaussian noise on the measurements . to the best of the authors knowledge
, the theoretical analysis in this paper has never been proposed , neither for somp nor for somp - ns .
+ finally , numerical simulations will show that the weighting capabilities of somp - ns enable one to improve the performance with regards to somp when the noise vectors exhibit different powers .
the numerical results will also provide evidence that the theoretical analysis accurately depicts key characteristics of somp - ns .
in particular , closed - form formulas for the optimal weights will be derived from the theoretical analysis and will be compared to the simulation results .
several authors have worked on similar problems .
the study of full support recovery guarantees for omp with @xmath1 or @xmath2-bounded noises as well as with gaussian noises has been performed in @xcite .
the authors of @xcite also provided conditions on the stopping criterion to ensure that omp stops after having picked all the correct atoms .
+ our analysis is similar to that performed by tropp in @xcite for convex programming methods in a smv setting .
together with gilbert @xcite , they analyzed the probability of full support recovery by means of omp for gaussian measurement matrices in the noiseless case .
their result has subsequently been refined by fletcher and rangan in @xcite to account for additive measurement noise by means of a high - snr analysis , _
i.e. _ , it is assumed that the signal - to - noise ratio scales to infinity .
all of the papers discussed so far only focus on the smv framework .
+ the theoretical analysis of our paper is partially inspired from @xcite and has been generalized to the mmv framework .
it is worth pointing out that our analysis does not require the high snr assumption of @xcite , properly captures the benefits provided by multiple measurement vectors but nevertheless presents some inaccuracies that are to be discussed at the end of this paper .
+ gribonval _ et al . _ have performed an analysis of somp for a problem similar to ours in @xcite .
they were interested in providing a lower bound on the probability of correct support recovery when the signal to be estimated is sparse and its non - zero entries are statistically independent mean - zero gaussian random variables exhibiting possibly different variances . +
while our statistical analysis considers the additive measurement noise as a random variable and the sparse signals to be recovered as deterministic quantities , the results obtained in @xcite use the opposite approach , _
i.e. _ , the sparse signals are random and the noise is deterministic .
thus , the problem addressed in our paper differs from that presented in @xcite but both papers use similar mathematical tools and the criteria to ensure full support recovery with high probability share analogous properties .
this last remark will be further discussed in section [ subsec : gribonvalrelatedthm ] .
first of all , section [ sec : sigmodel ] progressively introduces the context , provides a detailed description of the signal model and depicts an associated application .
afterwards , section [ sec : sompandsompns ] provides descriptions of somp and somp - ns . + before deriving the theoretical analysis , section [ sec : background ] introduces the mathematical tools necessary for its execution .
section [ sec : section3 ] then provides general theoretical results on the proper recovery of sparse vectors by means of somp - ns . on the basis of the results from section [ sec : section3 ] ,
we show in section [ sec : recovguarantees ] that , for gaussian noises , the probability of failure of somp - ns decreases exponentially with regards to the number of measurement vectors . + in section [ sec :
numresults ] , extensive numerical simulations show that adequate weighting strategies enable somp - ns to outperform somp whenever the noise variances for each measurement vector are different . also , a closed - form weighting strategy is derived from the theoretical analysis of the previous sections and these weights are compared to the optimal ones obtained by simulation .
finally , the simulation results show which aspects of the behavior of somp - ns are properly captured by the proposed theoretical analysis .
the reasons why our analysis fails to capture some characteristics of somp are discussed and potential workarounds are proposed for investigation .
we find preferable to introduce here the common notations used in this paper .
first of all , @xmath3 : = \left\lbrace 1 , 2 , \dots , n \right\rbrace$ ] . for any set @xmath4
, @xmath5 refers to its cardinality .
matrices are denoted by upper case bold letters while vectors are written in lower case bold letters .
the tranpose operation is denoted by the superscript @xmath6 . for @xmath7
, @xmath8 denotes the @xmath9-th component of @xmath10 .
similarly , the @xmath11-th column vector of any matrix is denoted by the corresponding lower case bold letter with subscript @xmath11 , _
e.g. _ , the @xmath11-th column of @xmath12 is @xmath13 .
the @xmath14-th entry of a matrix is denoted by the upper case letter with subscript @xmath15 , _
e.g. _ , the @xmath14-th entry of @xmath12 is written @xmath16 . the @xmath17 norm ( @xmath18 ) of @xmath19
is defined as @xmath20 .
moreover , the @xmath2 norm of vector @xmath19 is defined as @xmath21 } |\phi_i|$ ] . for any matrix @xmath22
, @xmath23 denotes the space spanned by its column vectors .
also , the trace of @xmath12 is @xmath24 and the frobenius norm of @xmath12 is denoted by @xmath25 . for @xmath26 , @xmath27 denotes the submatrix of @xmath12 that comprises its columns indexed by @xmath28 .
likewise , @xmath29 is the subvector of @xmath10 comprising only the components indexed within @xmath28 .
the moore - penrose pseudoinverse of any matrix @xmath12 is denoted by @xmath30 .
the orthogonal complement of a vector subspace @xmath4 is given by @xmath31 .
for any random variable @xmath32 , its cumulative density function ( cdf ) is denoted by @xmath33 while its probability density function ( pdf ) is written @xmath34 ( when it exists ) .
similarly , the joint cdf and pdf of the random variables @xmath35 are written as @xmath36 and @xmath37 respectively .
the probability measure is given by @xmath38 while the mathematical expectation is denoted by @xmath39 .
the statistical independence symbol is written @xmath40 .
we define the support of a vector @xmath10 as @xmath41 .
the @xmath42 `` norm '' of a vector @xmath10 is defined as @xmath43 . loosely speaking
, we say that @xmath7 is sparse whenever @xmath44 .
moreover , @xmath10 is said to be @xmath45-sparse whenever @xmath46 .
+ let us consider a collection of @xmath47 signals @xmath48 ( @xmath49 ) that are sparse in an orthonormal basis , _
i.e. _ , @xmath50 where @xmath51 represents the orthonormal basis and @xmath52 is the sparse coefficient vector of @xmath53 expressed in @xmath54 .
+ we now consider a unique linear measurement process to recover each one of the @xmath47 sparse coefficient vector @xmath55 .
moreover , the measurement process is assumed to deliver a number of observations @xmath56 significantly lower than @xmath57 .
additive measurement noise vectors @xmath58 are also accounted for .
formally , the latter statements rewrite @xmath59 where the _ measurement matrix _
@xmath22 denotes the linear measurement process and the _ measurement vectors _ @xmath60 gather the observations .
+ since the number of observations @xmath56 is lower than @xmath57 , arbitrary vectors @xmath55 can not be recovered from @xmath61 , even in the noiseless case . however , in the noiseless case , it has been shown @xcite that @xmath55 can be recovered provided that it is sparse enough , _
i.e. _ , the cardinality of its support is below a certain threshold , and that the measurement matrix @xmath12 exhibits specific properties such as the restricted isometry property that is described afterwards .
+ the orthonormal basis @xmath54 is often assumed to be the canonical basis .
the reason that explains this simplification relies on the fact the the measurement matrix @xmath12 is usually generated as a realization of a subgaussian random matrix .
it can be shown that such random matrices are well - condtionned for sparse support recovery , even when multiplied by orthonormal matrices , _
i.e. _ , @xmath62 remains a subgaussian random matrix .
this phenomenon is more thoroughly discussed in ( * ? ? ?
* section 9.1 ) .
thereby , the signal model will be simplified to @xmath63 if several sparse signals @xmath55 share similar supports , then it is interesting to simultaneously recover their joint support @xmath64 instead of performing @xmath47 independent and possibly different estimations of the support of every vector @xmath55 .
the reason that explains why such a strategy yields performance improvements is twofold : 1 .
it has been shown @xcite that when the sparse vectors @xmath55 share a similar support whose associated entries are highly variable from one vector to another , then the probability of correct support recovery increases with the number of available measurement vectors @xmath47 .
2 . in the noisy setting , the vectors @xmath65 are often statistically independent and isotropic which suggests that a joint support recovery procedure could be capable of filtering them
. this property is part of what the theoretical results and simulations of this paper establish .
once the joint support has been recovered , @xmath55 is easily recovered by solving a least squares problem involving only the non - zero entries of @xmath55 and the associated column vectors from @xmath12 .
we wish to provide here a formal and precise statement of the signal model to be used in the rest of this paper . +
as mentioned previously , we consider @xmath47 measurement vectors @xmath61 that are generated on the basis of @xmath47 sparse signals @xmath55 whose supports are similar .
we consider the following mmv model : @xmath66 where @xmath67 is composed of the @xmath47 measurement vectors @xmath68 .
@xmath69 comprises the @xmath47 sparse signals @xmath70 .
finally , the columns of @xmath71 correspond to measurement errors .
each error vector @xmath65 is distributed as @xmath72 and all of them are statistically independent .
the purpose of ( [ eq : mmvsignalmodel ] ) is to aggregate the @xmath47 equations defined in ( [ eq : introsigmodel2 ] ) into a single relation .
this representation will be preferred throughout the rest of this paper . + before going further , we will point out that the mathematical problem of the joint support recovery is equivalent to finding the columns of @xmath12 that enable one to fully express @xmath73 .
thereby , @xmath12 may be seen as a _
dictionary matrix _ whose columns @xmath74 are the _ atoms _ of the associated dictionary .
the problem of joint support recovery then boils down to determining which atoms to choose to simultaneously express the @xmath47 measurement vectors as their linear combinations .
although viewing @xmath12 as a measurement process is well suited to the description of typical applications , the dictionary approach is more appropriate for the sake of presenting the mathematical results and will thus be adopted for the rest of this paper .
however , the term _ measurement vector _ will not be replaced to stay consistent with the standard mmv terminology .
+ the dictionary matrix @xmath12 is assumed to satisfy the restricted isometry property of an order equal to or higher than the cardinality of the joint support @xmath75 .
moreover , it will be assumed that each column of this matrix exhibits a unit @xmath1 norm .
we briefly review two procedures to obtain a dictionary matrix that satisfies both properties above with high probability : 1 .
generate a random gaussian matrix and then normalize the @xmath1 norm of its columns .
in such a way , the atoms are uniformly distributed on the unit hypersphere @xmath76 of dimension @xmath57 .
it is possible to show that this class of random matrices satisfies the restricted isometry property with high probability ( see @xcite for example ) .
2 . generate a matrix whose entries are statistically independent rademacher random variables .
each column of the resulting matrix is then normalized by multiplying the matrix by @xmath77 to obtain atoms exhibiting unit @xmath1 norms .
the typical scenario associated with signal models ( [ eq : introsigmodel2 ] ) and ( [ eq : mmvsignalmodel ] ) is depicted in figure [ fig : firstscenario ] .
the idea is to observe a physical quantity , _
e.g. _ , a chemical composition , a wireless signal , etc . at different locations and/or time instants by means of @xmath47
_ sensing nodes _ @xmath78 whose only purpose is to acquire observations , _
i.e. _ , measurement vectors , and repatriate them to a central node ( cn ) that will simultaneously process all the data .
in such a configuration , the sensing nodes are generally cheap and exhibit very limited computational capabilities and power consumption while the central node is more costly because it has to deliver higher performance
. + nodes with different noise levels generate @xmath47 measurement vectors on which joint estimation of the sparse support @xmath79 can be performed ] although several applications of the signal model ( [ eq : introsigmodel2 ] ) are presented in ( * ? ? ?
* section 3.3 ) , we propose to take as an example the spectrum sensing problem .
spectrum sensing aims at scanning multi - gigahertz electromagnetic ( em ) spectrums at a rate that is below that of nyquist .
the reason that motivates this objective is that , although most of the available frequency bands are licensed to specific users and thus costly to acquire , it has been observed that the spectrum occupancy is limited , _
i.e. _ , the spectrum is sparse in the frequency domain at a given time instant and location .
therefore , it would be interesting to observe this spectrum through an appropriate linear measurement process , as described in equation ( [ eq : introsigmodel2 ] ) , and then use algorithms tailored for cs problems to determine , in real - time , which frequency bands are free and can be used to transmit information .
+ in this application , each entry of the sparse signals @xmath55 represents the power of a given frequency band .
since most of the spectrum is assumed to be unused at a given time instant and location , the vectors @xmath55 are expected to be sparse . although the nodes should ideally be exposed to the same spectrum , this is not the case in practice because of the rayleigh fading that strongly attenuates some frequency bands , thus being invisible to some nodes . in multiple input multiple output ( mimo )
wireless communications , this issue is circumvented by placing the receivers sufficiently far away from one another so that the fading becomes statistically independent for each node and thus highly unlikely to strongly attenuate the same frequency band for every node . in a likewise fashion
, the same solution will work for our framework since occasional `` holes '' will reveal to be a nonissue when performing joint decisions .
+ finally , the different nodes may exhibit different noise levels because of discrepancies in the fabrication process or because the hardware components ( _ e.g. _ , amplifiers , multipliers , filters ) of each node are different .
this last remark justifies the multiple noise variances hypothesis of this paper .
the two next subsections present two methods for addressing the problem of joint support recovery envisioned in section [ subsec : signalmodel ] .
the first method , somp , is standard in the literature but does not include noise stabilization .
the second method , our contribution , generalizes the first one by multiplying each measurement vector @xmath61 ( @xmath49 ) by a weight @xmath80 .
the original omp algorithm @xcite has been generalized in several ways to deal with matrix signals @xmath81 , _ i.e. _ , mmv problems
. simultaneous orthogonal matching pursuit ( somp ) is a possible generalization of omp @xcite .
+ somp is a greedy algorithm that provides approximate solutions to the joint support recovery problem by successively picking atoms from @xmath12 to simultaneously approximate the @xmath47 measurement vectors @xmath82 .
somp is described in algorithm [ alg : somp ] .
+ algorithm [ alg : somp ] : + simulatenous orthogonal matching pursuit ( somp ) + @xmath81 , @xmath22 , @xmath83 initialization : @xmath84 and @xmath85 @xmath86 determine the atom of @xmath12 to be included in the support : + @xmath87 update the support : @xmath88 projection of each measurement vector onto @xmath89 : + @xmath90 projection of each measurement vector onto @xmath91 : + @xmath92 @xmath93 @xmath94 we now explain how somp proceeds .
the residual at iteration @xmath95 , denoted by @xmath96 , consists of the projection of each one of the original signals @xmath61 onto the orthogonal complement of @xmath97 .
in such a way , the residual is orthogonal to every atom that has been chosen so far .
initially , the residual is chosen equal to the original signal . the decision on which atom to choose ( step 4 ) is based on the sum of the inner products of every atom @xmath98 with each residual measurement vector @xmath99 ( where @xmath99 refers to the @xmath100-th column of @xmath96 ) since @xmath101 the index of the atom maximizing the @xmath0 norm is included in the support ( step 5 ) .
then , the original signal @xmath102 is projected onto the orthogonal complement of @xmath89 ( steps 6 and 7 ) .
+ in this setting , somp stops after exactly @xmath45 iterations . however , it is worth mentioning that the stopping criterion usually comprises a criterion based on the number of iterations as well as another one relying on the norm of the residual , _
i.e. _ , if the norm of the residual is below a certain threshold , then omp stops .
different norms can be used for the second criterion but these considerations will not be further discussed in this paper .
the interested reader can consult @xcite for related matters .
+ furthermore , maximizing the @xmath0 norm in step @xmath103 is not the unique choice .
other authors have investigated different norms , _
e.g. _ , the @xmath1 and @xmath2 norms .
nevertheless , some numerical simulations reveal that the choice of the norm has very little effect on the performance ( see ( * ? ? ?
* figure 3 ) ) .
we now present the development of a noise stabilization strategy to be used in conjunction with somp that has low computational requirements .
the equivalent new algorithm is referred to as somp - ns where ns stands for _ noise stabilization_. algorithm [ alg : sompns ] describes the first form of somp - ns . +
this novel algorithm is a generalization of somp that weights the impact of each measurement vector within matrix @xmath81 on the decisions performed at each iteration .
+ algorithm [ alg : sompns ] : + somp with noise stabilization ( somp - ns ) first form + @xmath81 , @xmath22 , @xmath104 , @xmath83 @xmath84 and @xmath85 @xmath86 determine the atom of @xmath12 to be included in the support : + @xmath105 @xmath88 projection of each measurement vector onto @xmath89 : + @xmath90 projection of each measurement vector onto @xmath91 : + @xmath92 @xmath93 @xmath94 somp - ns is actually very close to somp .
both algorithms decide on which atom to pick on the basis of a sum of absolute values of inner products , each term in the sum depending only on one measurement vector .
somp gives the same importance to each measurement vector whereas its weighted counterpart introduces weights @xmath80 ( @xmath49 ) so as to give more or less importance to each measurement vector .
+ a second form of somp - ns that is more computationally efficient is available in algorithm [ alg : sompnsv2 ] . in the second form , * somp*(@xmath106 )
refers to the regular somp algorithm described in algorithm [ alg : somp ] .
algorithm [ alg : sompnsv2 ] : + somp with noise stabilization ( somp - ns ) second form + @xmath81 , @xmath22 , @xmath104 , @xmath83 @xmath107 the columns of @xmath102 are weighted beforehand : @xmath108 apply the regular somp algorithm : @xmath109 * somp*(@xmath106 ) @xmath28 both forms are equivalent since @xmath110 and @xmath111 .
the last equality holds true because the orthogonal projector @xmath112 is applied to each column of @xmath113 separately .
although the residual matrix @xmath96 is different for the two forms of somp - ns , the difference only consists in a multiplicative term @xmath114 for each column @xmath100 of @xmath96 and it does not modify the atoms added to the estimated support .
this section briefly explains the mathematical tools needed to conduct the theoretical analysis .
let @xmath22 be a matrix composed of @xmath57 column vectors @xmath115 .
moreover , @xmath116 $ ] , @xmath117 . the coherence @xmath118 of @xmath12 is defined as @xmath119 similarly , one can define the cumulative coherence function ( also referred to as the babel function ) of @xmath12 as @xmath120 trivially , @xmath121 . for the sake of brevity , @xmath118 and @xmath122
will be respectively denoted as @xmath123 and @xmath124 from now on . + it is also frequent to use the bound @xmath125 if @xmath126 where @xmath127 ( @xmath128 ) , then @xmath129 } { \text{supp}}({\boldsymbol{u}}_j)$ ] .
the definition above extends the notion of support to matrices , _
i.e. _ , the support of a matrix is the union of the supports of its columns .
similarly to the vector case , if @xmath130 , then @xmath131 } { \text{supp}}({\boldsymbol{u}}_j ) \right|$ ] .
+ we define @xmath132 } |\phi_i|$ ] which is not a norm .
furthermore , for @xmath22 , @xmath133 is defined as ( * ? ? ?
* equation a.8 ) @xmath134 where @xmath135 . for the sake of simplifying the notations
, we will adopt the convention @xmath136 .
it can be shown that , for @xmath22 , ( * ? ? ?
* lemma a.5 ) @xmath137 } \sum_{j=1}^{n } \left| \phi_{i , j } \right| = \|{\boldsymbol{\phi}}^{\mathrm{t } } \|_{1}. \ ] ] the sparse rank @xcite of @xmath22 , denoted by @xmath138 , is given by @xmath139 @xmath138 is thus the smallest number of linearly dependent columns of @xmath12 .
equivalently , it means that , if @xmath140 , then , for every support @xmath28 such that @xmath141 , the columns of @xmath27 are linearly independent , _
i.e. _ , @xmath27 has full column rank .
+ note that computing @xmath138 for a given matrix @xmath12 is not computationally tractable as this problem is even harder to solve than a @xmath42 norm minimization problem which is known to be np - hard @xcite .
the matrix @xmath22 satisfies the so - called restricted isometry property ( rip ) @xcite of order @xmath45 if there exists a constant @xmath142 such that @xmath143 for all @xmath45-sparse vectors @xmath144 .
the smallest @xmath145 that satisfies equation ( [ eq : ripdef ] ) is called the rectricted isometry constant ( ric ) of order @xmath45 .
the ric of order @xmath45 can theoretically be computed by considering @xmath146 \\
|s| = s } } \lambda_{\mathrm{max}}({\boldsymbol{\phi}}_{s}^{\mathrm{t } } { \boldsymbol{\phi}}_{s } ) - 1 \\
l_s = 1 - \min_{\substack{s \subseteq \left[n \right ] \\
|s| = s } } \lambda_{\mathrm{min}}({\boldsymbol{\phi}}_{s}^{\mathrm{t } } { \boldsymbol{\phi}}_{s})\end{aligned}\ ] ] where @xmath147 and @xmath148 denote the smallest and largest eigenvalues respectively . then
, @xmath149 evaluating @xmath150 and @xmath151 is not computationally tractable as it requires to determine the smallest and largest eigenvalues of @xmath152 matrices of size @xmath153 .
in particular , this problem has been shown to be np - hard in the general case @xcite .
it is therefore interesting to find an upper bound on @xmath145 that can be easily computed .
[ lem : riclambda ] if @xmath154 , then @xmath155 \\ |s| = s } } \lambda_{\mathrm{max}}({\boldsymbol{\phi}}_{s}^{\mathrm{t } } { \boldsymbol{\phi}}_{s } ) \leq 1 + \mu_1(s-1 ) \leq 1 + ( s-1 ) \mu\\
\min_{\substack{s \subseteq \left[m \right ] \\
|s| = s } } \lambda_{\mathrm{min}}({\boldsymbol{\phi}}_{s}^{\mathrm{t } } { \boldsymbol{\phi}}_{s } ) \geq 1 - \mu_1(s-1 ) \geq 1 - ( s-1 ) \mu.\end{aligned}\ ] ] the first inequality of each line holds if @xmath156 .
this result is obtained in @xcite .
a consequence of lemma [ lem : riclambda ] is that , if @xmath154 , then @xmath157 it is worth noticing that if @xmath142 , then @xmath140 .
the reason is that @xmath142 implies that @xmath158 , |s| = s } \lambda_{\mathrm{min}}({\boldsymbol{\phi}}_{s}^{\mathrm{t } } { \boldsymbol{\phi}}_{s } ) > 0 $ ] which in turn implies that , for every support @xmath28 of cardinality equal to or less than @xmath45 , @xmath159 has full column rank .
a lipschitz function @xmath160 with regards to metric @xmath1 is a function that satisfies @xmath161 the constant @xmath162 is called the lipschitz constant .
we now wish to derive basic theoretical results needed to analyze the performance of somp - ns in both the noiseless and noisy cases .
the theoretical framework developed hereafter will be used in section [ sec : recovguarantees ] to give lower bounds on the probability of partial or full recovery using somp - ns when the additive noise is gaussian .
we now wish to build an exact recovery criterion ( erc ) that guarantees , for dictionary matrix @xmath12 , the recovery of every sparse signal of support @xmath28 by means of somp - ns .
the result that will be obtained is similar to an earlier result of j. a. tropp , sometimes referred to as tropp s erc @xcite .
+ first of all , it is worth noticing that the maximum correlation in step 4 of algorithm [ alg : sompns ] can be written as @xmath163 } \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{r}}^{(t)}_k , { \boldsymbol{\phi}}_{j } \right\rangle \right| q_k \right ) = \left\|{\boldsymbol{\phi}}^{\mathrm{t } } { \boldsymbol{r}}^{(t ) } { \boldsymbol{q } } \right\|_{\infty}\ ] ] where @xmath164 is the diagonal matrix that contains the weights to be applied to each column vector of @xmath96 .
it is assumed that @xmath80 ( @xmath49 ) .
+ the next result allows to state the erc for somp - ns .
it will also be used to develop the exact recovery criterion for the noisy case later .
[ lem : erc ] let @xmath22 where @xmath159 has full column rank for some support @xmath165 .
moreover , @xmath166 denotes the orthogonal projector onto @xmath97 where @xmath167 .
let @xmath168 , @xmath81 and @xmath169 where @xmath170 .
it is assumed that @xmath171 . under these conditions ,
the following inequality holds true @xmath172 where @xmath173 denotes the relative complement of @xmath28 with respect to @xmath174 $ ] . following the steps of ( * ? ? ?
* theorem 4.5 ) , the lemma is easily obtained .
each column of the matrix @xmath96 can be expressed as a linear combination of the columns of @xmath159 .
the reason that explains this last statement is that @xmath175 and @xmath176 .
+ moreover , @xmath159 is guaranteed to have full column rank which implies that the moore - penrose pseudoinverse @xmath177 is equal to @xmath178 and consequently that @xmath179 . furthermore , it is easily established ( * ? ? ?
* lemma 4.4 ) that for two matrices @xmath180 and @xmath181 , @xmath182 . combining the results above yields
@xmath183 finally , using equation ( [ eq : inftytoonenorm ] ) shows that @xmath184 and concludes the proof .
we are now ready to provide an erc for somp - ns .
[ thm : erc ] let @xmath22 and @xmath169 .
if @xmath185 and @xmath159 has full column rank , then a sufficient condition for somp - ns to properly retrieve the support @xmath28 of @xmath186 after exactly @xmath187 iterations is @xmath188 where @xmath173 is the relative complement of @xmath28 with respect to @xmath174 $ ] .
let us assume that somp - ns has made correct decisions before iteration @xmath95 .
the greedy selection ratio ( at iteration @xmath95 ) is defined as : @xmath189 clearly , @xmath190 ensures that somp - ns performs a correct decision at iteration @xmath95 since it implies that the largest sum of inner products is obtained for one of the atoms belonging to the support @xmath28 , _
i.e. _ , @xmath191 .
since it is assumed that somp - ns has only made correct decisions so far , lemma [ lem : erc ] shows that @xmath190 holds true whenever @xmath192 .
+ furthermore , @xmath192 ensures that a correct atom is picked during iteration @xmath193 . by induction ,
the hypothesis that correct decisions have been made so far will be satisfied for every iteration and full support recovery at iteration @xmath194 is thus guaranteed .
it can be shown that theorem [ thm : erc ] is sharp in the sense that if the inequality @xmath192 is not satisfied for some support @xmath28 , it is always possible to find a @xmath195 whose support is @xmath28 for which somp - ns identifies an incorrect atom at the first iteration .
the sharpness property for somp - ns directly derives from ( * ? ? ?
* theorem 3.10 ) which provides an equivalent property in the smv case for omp .
indeed , if omp fails to choose a correct atom with vector @xmath196 , then somp - ns also fails with @xmath197 as , in this particular case , omp with signal @xmath198 and somp - ns with signal @xmath199 make the same decisions .
+ moreover , the theorem above is generally of theoretical use only since computing @xmath200 requires to know the support beforehand .
also , computing @xmath200 for all the possible supports of a given size is not computationally tractable .
+ however , it is shown in @xcite that if @xmath201 , then @xmath202 for all the supports of size @xmath187 .
it is worth noticing that @xmath27 is full rank for all supports @xmath28 of size @xmath45 if and only if @xmath140 . + although theorem [ thm : erc ] proves to be interesting for @xmath203 in signal model ( [ eq : mmvsignalmodel ] ) , it should be extended to the noisy case which is the purpose of the next section .
we develop here an erc for the noisy case which generalizes theorem [ thm : erc ] to this context . to reach this result ,
let us assume that somp - ns has made correct decisions before iteration @xmath95 .
first of all , we separate the contribution of the noise @xmath204 and that of the useful signal @xmath186 : @xmath205 then , the somp - ns correlation is lower bounded by @xmath206 moreover , the triangle inequality yields @xmath207 since somp - ns makes a correct decision at step @xmath95 if @xmath208 , the inequalities above show that this condition is always satisfied whenever @xmath209 it is worth noticing that , due to lemma [ lem : erc ] , the following relationship holds true : @xmath210 note that the assumption that correct decisions have been so far is crucial for this result to be true .
furthermore , as @xmath159 and @xmath211 are column submatrices of @xmath12 , one easily obtains @xmath212 a sufficient condition for ( [ eq : sompanalysis1 ] ) to hold is thereby @xmath213 theorem [ thm : theoframework1 ] summarizes the previous discussion by explicitly stating the erc in the noisy case .
[ thm : theoframework1 ] let @xmath186 be the sparse matrix to be retrieved and let @xmath171 denote its support .
let @xmath159 exhibit a full column rank .
let us assume that only correct atoms have been picked before iteration @xmath214 and that the reduced dictionary matrix @xmath159 has full column rank .
somp - ns with dictionary matrix @xmath12 and signal @xmath215 is guaranteed to make a correct decision at iteration @xmath95 whenever @xmath216 where @xmath217 , @xmath218 and @xmath219 .
noticeably a necessary condition for satisfying ( [ eq : sompanalysis2 ] ) reads @xmath220 which is precisely the erc obtained before for the noiseless case .
moreover , low values of @xmath200 imply a better robustness against the noise as the condition hereabove is then more easily satisfied .
this observation is not surprising since the robustness in the noiseless case determines the amplitude of the noise we can apply without ruining the support recovery .
+ theorem [ thm : theoframework1 ] is the cornerstone of the theoretical analysis conducted in this paper .
however , as done hereafter , it remains desirable to find a lower bound for @xmath221 that expresses in a simpler manner the impact of the signal to be estimated , the weights and the dictionary matrix .
also , the term @xmath222 will be dealt with by means of a statistical analysis presented in section [ sec : recovguarantees ] when the noise is gaussian .
we now focus on deriving a lower bound for @xmath223 that can be easily evaluated .
verifying equation ( [ eq : sompanalysis2 ] ) with this computable lower bound will provide a sufficient condition for the erc to hold .
in particular , we desire to obtain a lower bound that does not rely on the knowledge of the particular support @xmath28 that is chosen . + to reach this goal , we chose here to extend the method proposed in @xcite to mmv problems .
the following theorem mainly has a theoretical interest and is afterwards particularized in corollary [ corr : ripbasedbtbound ] that will used in the rest of the paper .
corollary [ corr : mubasedbtbound ] will provide a variant of corollary [ corr : ripbasedbtbound ] relying on both the coherence of the dictionary ( instead of the rip ) and on the quantity @xmath200 .
+ [ thm : firstboundgeneral ] let @xmath224 $ ] and let @xmath225 denote the indexes of the atoms chosen by somp - ns at iteration @xmath95 .
it is assumed that @xmath226 , _
i.e. _ , only correct decisions have been made before iteration @xmath95 .
@xmath227 contains the indexes of the correct atoms yet to be selected at iteration @xmath95 .
let @xmath228 where @xmath166 denotes the orthogonal projector onto @xmath97 .
then , for any @xmath229 ( @xmath49 ) , @xmath230 where @xmath231 .
moreover , if @xmath12 satisfies the rip with @xmath232-th restricted isometry constant @xmath233 , then @xmath234 denoting the @xmath100-th column of @xmath235 by @xmath236 , we first observe that @xmath237 the maximum being taken over @xmath238 since @xmath236 is orthogonal to @xmath97 because of the orthogonal projector @xmath239 . since @xmath240 , @xmath241 which implies @xmath242 the triangle inequality yields @xmath243 thus , @xmath244 the first inequality results from the observation that , for any vector @xmath245 , we have @xmath246 .
the inequality @xmath247 is available in ( * ? ? ?
* lemma 5 ) .
the first part of the theorem is now proved . + if @xmath12 satisfies the rip with ric @xmath233 , then equation ( [ eq : ricasmax ] ) yields @xmath248 , which proves the second part of the theorem .
the result above shows that the decision metric in corresponding to the correct atoms , _
i.e. _ , @xmath221 , is closely related to the @xmath1 norm of the signal to be recovered and to the singular values of @xmath249 .
it is clear that the ability of @xmath12 to conserve the norm of sparse vectors is necessary to ensure that the measurement noise @xmath250 does not absorb @xmath235 .
+ through the following corollary , we wish to obtain a simple term that replaces @xmath251 .
[ corr : ripbasedbtbound ] let @xmath224 $ ] and let @xmath225 denote the indexes of the atoms chosen by somp - ns at iteration @xmath95 .
it is assumed that @xmath226 , _
i.e. _ , only correct decisions have been made before iteration @xmath95 .
@xmath252 contains the indexes of the correct atoms yet to be selected at iteration @xmath95 .
let @xmath228 where @xmath166 denotes the orthogonal projector onto @xmath97 .
if @xmath12 satisfies the rip with @xmath232-th restricted isometry constant @xmath233 , then @xmath253 it is easy to show that for any @xmath254 @xmath255 it implies that @xmath256 in particular , the choice of the @xmath257 is arbitrary so that the best lower bound is given by @xmath258 one easily notices that @xmath259 . if @xmath260 , then choosing @xmath261 is optimal , _ i.e. _ , @xmath262 . together with theorem
[ thm : firstboundgeneral ] , this result shows that the first inequality holds true .
the last inequality is then trivially obtained . both inequalities in the result above explicitly emphasize the impact of the weights , the ric and the amplitude of the coefficients to be recovered . however , regarding the weights , it is clear that they will also impact the value of @xmath222 .
moreover , equation ( [ eq : firstboundcorr1_2 ] ) suggests that , in order to reliably retrieve an atom , the sum of the absolute values of the associated coefficients in @xmath186 should be high enough .
+ as already mentioned in the introduction , theorem [ thm : firstboundgeneral ] appears to be a new result while corollary [ corr : ripbasedbtbound ] has already been obtained in the literature ( * ? ? ?
* theorem 5 ) .
+ it is now possible to use the coherence - based inequalities provided by lemma [ lem : riclambda ] so as to derive a new bound on the basis of the previous one .
we obtain corollary [ corr : mubasedbtbound ] which should be understood as the coherence counterpart of corollary [ corr : ripbasedbtbound ] .
[ corr : mubasedbtbound ] let @xmath224 $ ] and let @xmath225 denote the indexes of the atoms chosen by somp - ns at iteration @xmath95 .
it is assumed that @xmath226 , _
i.e. _ , only correct decisions have been made before iteration @xmath95 .
@xmath252 contains the indexes of the correct atoms yet to be selected at iteration @xmath95 .
let @xmath228 where @xmath166 denotes the orthogonal projector onto @xmath97 .
if @xmath263 , then @xmath264 moreover , if @xmath265 , then both equation ( [ eq : firstboundvmu1 ] ) and ( [ eq : firstboundvmu ] ) hold true . @xmath266 using lemma [ lem : riclambda ] and the equality @xmath267 , one obtains @xmath268 the first inequality makes sense only if @xmath263 while the last inequality requires @xmath265 .
although corollary [ corr : mubasedbtbound ] is less powerful and general than corollary [ corr : ripbasedbtbound ] , it provides an interesting insight into how the coherence of the dictionary influences @xmath221 .
the previous section provided a non - probabilistic analysis of the quantity @xmath221 by deriving lower bounds that are more simple to evaluate than the original quantity . regarding the noise - related quantity @xmath222 , we will in this section
perform a stochastic analysis to derive a lower bound on the probability that it does not exceed a threshold @xmath269 for gaussian noises . + as shown by theorem [ thm : theoframework1 ] , it is possible to examine whether somp - ns succeeds in choosing a correct atom at step @xmath95 by evaluating separately quantities linked to the sparse signal to be estimated and the noise vectors , @xmath221 and @xmath222 respectively . since several simple lower bounds for @xmath221 have been found , it becomes possible to evaluate a lower bound on @xmath270\ ] ] _ i.e. _ , a lower bound on the probability that somp - ns makes correct decisions for signal model ( [ eq : mmvsignalmodel ] ) according to equation ( [ eq : sompanalysis2 ] ) of theorem [ thm : theoframework1 ] .
corollaries [ corr : ripbasedbtbound ] and [ corr : mubasedbtbound ] yield @xmath271 \\
\geq \mathbb{p}\left [ \|{\boldsymbol{\phi}}^{\mathrm{t } } { \boldsymbol{e}}^{(t ) } { \boldsymbol{q } } \|_{\infty } < 0.5 \left(1 - \|{\boldsymbol{\phi}}_s^+ { \boldsymbol{\phi}}_{\overline{s } } \|_{1 } \right ) \|{\boldsymbol{\phi}}_s^{\mathrm{t } } { \boldsymbol{z}}^{(t ) } { \boldsymbol{q } } \|_{\infty}^{\mathrm{(rip ) } } \right ] \\
\geq \mathbb{p}\left [ \|{\boldsymbol{\phi}}^{\mathrm{t } } { \boldsymbol{e}}^{(t ) } { \boldsymbol{q } } \|_{\infty } < 0.5 \left(1 - \|{\boldsymbol{\phi}}_s^+ { \boldsymbol{\phi}}_{\overline{s } } \|_{1 } \right ) \|{\boldsymbol{\phi}}_s^{\mathrm{t } } { \boldsymbol{z}}^{(t ) } { \boldsymbol{q } } \|_{\infty}^{(\mu ) } \right ] \end{aligned}\ ] ] where @xmath272 a statistical analysis of @xmath222 is proposed when @xmath273 and @xmath274 for @xmath275 .
the advantage of our approach is to take into account the isotropic nature of statistically independent gaussian random vectors .
our main result is theorem [ thm : finalgaussianthm ] , which shows that the probability of making incorrect decisions from iteration @xmath193 to iteration @xmath276 included decreases exponentially with regards to a certain number of parameters .
this theorem is then particularized so as to make use of the coherence @xmath123 of @xmath12 instead of the rip and the erc .
+ first of all , the statistical properties of @xmath277 for a single and arbitrary atom @xmath13 are investigated in section [ subsec : oneatom ] .
these properties are then extended to @xmath278 } \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right)$ ] in section [ subsec : natoms ] . on the basis of this last result ,
theorem [ thm : finalgaussianthm ] is obtained in section [ subsec : fullrecovery ] . a particular case of the signal model
is then examined in order to ease , in section [ sec : numresults ] , the comparison of the results provided by means of the theoretical bound and those obtained by simulation . in this section
, we restate some assumptions and define quantities used later on .
+ it is assumed that the entries of each @xmath65 ( @xmath49 ) are i.i.d .
mean - zero gaussian random variables of variance @xmath279 , _
i.e. _ , @xmath273 .
furthermore , the noise vectors @xmath65 are statistically independent .
finally , the columns of @xmath159 are assumed to be linearly independent . as already mentioned in section [ subsec : ripdef ]
, the latter condition is true for all supports @xmath28 of cardinality @xmath45 whenever @xmath280 .
+ we define @xmath281 before going on further , we also define @xmath282 and @xmath283 corollary [ corr : ripbasedbtbound ] shows that a sufficient condition for to choose a correct atom at iteration @xmath193 is given by @xmath284 } \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right ) < 0.5 \left(1 - \|{\boldsymbol{\phi}}_s^+ { \boldsymbol{\phi}}_{\overline{s } } \|_{1 } \right ) ( 1-\delta_{|s| } ) \min_{j
\in s } \sum_{k=1}^{k } | x_{j , k } | q_k . \end{aligned}\ ] ] using the upper bound @xmath285 is a recurrent solution in the literature and provides satisfactory performance in a smv setting ( see @xcite ) .
however , using such an approach in a mmv setting does not properly capture the performance gains obtained whenever @xmath47 increases as this upper bound assumes that all the error vectors @xmath286 are aligned with a single atom at each iteration .
while this approximation is acceptable whenever only one measurement vector is available , it proves to be highly pessimistic as soon as one considers many independent ( and usually isotropic ) error vectors .
+ this observation motivates an in - depth analysis of the statistical properties of @xmath287 } \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right)$ ] .
the analysis conducted hereafter mainly relies on the notion of lipschitz functions and the related concentration inequalities . in this section ,
we are interested in providing a lower bound for @xmath288 for an arbitrary atom @xmath13 .
the main result of this section is lemma [ lem : finalboundindivprob ] .
[ thm : concentrationineqlipschitz ] ( * ? ? ? * theorem 8.40 . )
let @xmath289 be a lipschitz function ( with regards to the metric @xmath1 ) with lipschitz constant @xmath162 .
let @xmath290 be a vector of independent standard gaussian random variables .
then , for all @xmath291 @xmath292 and consequently @xmath293 the theorem above shows that lipschitz functions @xmath294 tend to concentrate around their expectations when @xmath295 is distributed as a standard gaussian random vector .
moreover , the concentration gets better as the lipschitz constant @xmath162 decreases .
+ this theorem is intended to be used in conjunction with the function @xmath296 where @xmath297 ( @xmath49 ) .
this function will be shown to be equivalent to @xmath277 when @xmath65 is gaussian .
+ we now wish to establish that @xmath298 is a lipschitz function , compute the associated lipschitz constant and determine its expectation .
let @xmath299 , then , using the reverse triangle inequality and the cauchy - schwarz inequality , @xmath300 therefore , a valid lipschitz constant @xmath162 of @xmath298 is equal to @xmath301 .
this is the best lipschitz constant since , for @xmath302 and @xmath303 , @xmath304 .
+ using the concentration inequalities for lipschitz functions requires to know the value of @xmath305 $ ] . using the linearity of the expectation and the fact that for @xmath306 , @xmath307 , one easily obtains @xmath308 = \sqrt{\dfrac{2}{\pi } } \| { \boldsymbol{q}}^{({\boldsymbol{\sigma}})}\|_1 = b({\boldsymbol{q } } , { \boldsymbol{\sigma}}).\ ] ] by using theorem [ thm : concentrationineqlipschitz ]
, it is now possible to conclude that , for @xmath291 , @xmath309 the final step of the development is to prove that the function @xmath298 defined above is distributed as @xmath294 for @xmath310 .
indeed , @xmath311 and @xmath312 .
the following lemma summarizes the discussion above and will be used to establish the theorem of the next section .
[ lem : finalboundindivprob ] let @xmath313 where @xmath314 ( @xmath128 ) .
let @xmath65 ( @xmath49 ) be independent random variables respectively distributed as @xmath72 where @xmath297 .
it is also assumed that @xmath80 .
then , for @xmath291 , @xmath315 it is important to highlight that lemma [ lem : finalboundindivprob ] provides an upper bound of the probability that @xmath277 is higher than @xmath316 only if @xmath317 is higher than @xmath318 .
the next problem to be tackled is that we would like to estimate an upper bound of the probability that @xmath287 } ( \sum_{k=1}^{k } | \langle { \boldsymbol{\phi}}_j , { \boldsymbol{e}}_{k } \rangle | q_k ) $ ] is higher than a threshold @xmath317 instead of the same probability obtained for @xmath319 .
the issue we are facing is that the @xmath57 random variables @xmath319 ( for @xmath128 ) are not statistically independent which implies that the probability of upper bounding all these @xmath57 random variables simultaneously is not equal to the product of the probability to upper bound them separately .
however , it remains possible to find a workaround using the union bound , as demonstrated by the following theorem .
[ thm : finalboundmanyatomsprob ] let @xmath313 where @xmath314 ( @xmath128 ) .
let @xmath65 ( @xmath49 ) be independent random variables respectively distributed as @xmath72 where @xmath297 .
it is also assumed that @xmath80 .
then , for @xmath291 , @xmath320 \leq n \exp \left ( - \kappa({\boldsymbol{q } } , { \boldsymbol{\sigma } } ) \varepsilon^2 \right).\ ] ] equivalently , @xmath321 \geq 1 - n \exp \left ( - \kappa({\boldsymbol{q } } , { \boldsymbol{\sigma } } ) \varepsilon^2 \right).\ ] ] first of all , we observe that @xmath322 } \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right)$ ] .
then , by union bound , @xmath323 } \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right ) \geq b({\boldsymbol{q } } , { \boldsymbol{\sigma } } ) + \varepsilon \right ] \\ = & \mathbb{p } \left [ \bigcup_{j=1}^{n } \left [ \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right ) \geq b({\boldsymbol{q } } , { \boldsymbol{\sigma } } ) + \varepsilon \right ] \right ] \\
\leq & \sum_{j=1}^{n } \mathbb{p } \left [ \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right ) \geq b({\boldsymbol{q } } , { \boldsymbol{\sigma } } ) + \varepsilon \right ] \\
\leq & n \exp \left ( - \kappa({\boldsymbol{q } } , { \boldsymbol{\sigma } } ) \varepsilon^2 \right).\end{aligned}\ ] ] the first inequality results from the union bound while the second inequality holds because of lemma [ lem : finalboundindivprob ] .
theorem [ thm : finalboundmanyatomsprob ] implicitly provides a lower bound on the probability of making correct decisions during iteration @xmath193 .
it is indeed possible to use either equation ( [ eq : eqripbound ] ) or ( [ eq : eqmubound ] ) in conjunction with equation ( [ eq : probinequalities ] ) and theorem [ thm : finalboundmanyatomsprob ] to obtain the desired lower bound .
nevertheless , a lower bound on the probability that somp - ns makes correct decisions from iteration @xmath193 to iteration @xmath276 remains to be found .
this is the purpose of this section .
+ before establishing the main result , one needs lemmas [ lem : fullsupportlem1 ] and [ lem : fullsupportlem2 ] . + [
lem : fullsupportlem1 ] if @xmath286 is distributed as @xmath72 and @xmath324 is a fixed orthogonal projector matrix , then @xmath325 is distributed as @xmath326 where @xmath327 .
we have @xmath328 an auxiliary atom @xmath329 can be defined for index @xmath11 , @xmath330 .
this atom satisfies the inequality @xmath331 .
let us now observe that if the entries of @xmath65 are i.i.d .
random variables distributed as @xmath332 , then @xmath333
. the result immediately follows .
lemma @xmath334 basically establishes that , for a fixed projection matrix @xmath324 , @xmath325 is distributed as a mean - zero gaussian random variable whose variance is always lower than that of the entries of @xmath335 .
this result will enable us to apply a statistical analysis similar to that of the previous section by keeping the gaussian hypothesis .
[ lem : fullsupportlem2 ] we consider the random variables @xmath32 , @xmath336 and @xmath337 where @xmath338 .
it is assumed that @xmath339 and @xmath340 .
then , for all @xmath341 , @xmath342 \leq \mathbb{p } \left [ x + |y_2| \leq \varepsilon \right].\ ] ] we know that @xmath343 and @xmath344 where @xmath345 since @xmath346 is a monotonically increasing function , one obtains @xmath347 .
thus , @xmath348 & = \int_{-\infty}^\varepsilon \mathbb{p } \big [ |y_1| \leq \varepsilon - x \big ] f_x(x ) \mathrm{d}x \\ & \leq \int_{-\infty}^\varepsilon \mathbb{p } \big [ |y_2| \leq \varepsilon - x \big ] f_x(x ) \mathrm{d}x\\ & = \mathbb{p } \left[x + |y_2| \leq \varepsilon \right ] .
\qedhere\end{aligned}\ ] ] in the rest of this paper , the random variable @xmath32 of lemma [ lem : fullsupportlem2 ] will be replaced with a sum of @xmath349 independent random variables exhibiting half - normal distributions , _
i.e. _ , @xmath350 where the @xmath351 ( @xmath352 ) are independent and @xmath353 .
thereby , an immediate corollary of the lemma above is that the probability of upper bounding by @xmath291 the sum of statistically independent random variables exhibiting half - normal distributions is always decreased whenever the variance of at least one of the random variables is increased .
+ let us now state the key theoretical result of this paper , which provides a lower bound on the probability that somp - ns selects correct atoms during the first @xmath354 iterations .
[ thm : finalgaussianthm ] let @xmath65 ( @xmath49 ) be statistically independent gaussian random vectors respectively distributed as @xmath72 .
let @xmath355 .
let @xmath356 be the signal matrix whose support is @xmath28 .
let also @xmath357 be unit - norm vectors in @xmath358 and @xmath359 be the corresponding matrix which is assumed to satisfy the rip with @xmath187-th ric @xmath360 .
let @xmath361 . then
, somp - ns with dictionary matrix @xmath12 , weights @xmath114 ( @xmath49 ) and signal @xmath215 is ensured to make correct decisions during the first @xmath354 iterations , _
i.e. _ , from iteration @xmath193 to iteration @xmath45 included , with probability higher than @xmath362 whenever @xmath363 where @xmath364 @xmath365 and @xmath318 being defined in equation ( [ eq : defkappa ] ) and ( [ eq : defb ] ) , respectively .
first , we observe that @xmath366 where it is convenient to define the @xmath11-th auxiliary atom at iteration @xmath95 as @xmath367 . according to ( [ eq : probinequalities ] ) , at iteration @xmath95 ( for @xmath368 ) , a sufficient condition for to make a correct decision at iteration @xmath95 is given by @xmath284 } \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j^{(t ) } , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right ) \leq 0.5 ( 1 - \|{\boldsymbol{\phi}}_s^+ { \boldsymbol{\phi}}_{\overline{s } } \|_{1 } ) ( 1-\delta_{|\mathcal{j}_t| } ) \min_{j \in s } \sum_{k=1}^{k } | x_{j , k } | q_k . \end{aligned}\ ] ] since @xmath369 implies @xmath370 , then @xmath371 which shows that a less tight sufficient condition for somp - ns to make a correct decision at iteration @xmath95 is @xmath372 } \left ( \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{\phi}}_j^{(t ) } , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \right ) \leq \varepsilon ( { \boldsymbol{q } } ) = 0.5 ( 1 - \|{\boldsymbol{\phi}}_s^+ { \boldsymbol{\phi}}_{\overline{s } } \|_{1 } ) ( 1-\delta_{|s| } ) \min_{j \in s } \sum_{k=1}^{k } | x_{j , k } | q_k .
\end{aligned}\ ] ] therefore , if the condition above holds for @xmath373 , somp - ns is guaranteed to pick a correct atom at iteration @xmath193 . at iteration @xmath374
, the orthogonal projector @xmath375 can take @xmath187 different values ( each value of @xmath375 corresponds to a specific support @xmath376 ) .
therefore , if ( [ eq : suffcondwsompstept ] ) holds for every possible projector matrix at iteration @xmath374 , somp - ns is guaranteed to pick a correct atom at iteration @xmath374 . thus , we need to satisfy @xmath377 equations of the form @xmath378 ( where @xmath379 ) to ensure that somp - ns picks correct atoms during the first two iterations .
+ extending the previous train of thought from iteration @xmath193 to iteration @xmath45 , one easily comes to the conclusion that @xmath380 equations of the form @xmath378 ( where @xmath379 ) should be satisfied since , at iteration @xmath95 , there exist @xmath381 possible realizations of the orthogonal projector @xmath382 . using
the union bound as in the proof of theorem [ thm : finalboundmanyatomsprob ] shows that the probability of satisfying the @xmath383 equations is lower bounded by @xmath384\ ] ] where @xmath385 is the @xmath9-th possible realization of the orthogonal projector at iteration @xmath95 assuming that only correct atoms have been picked before iteration @xmath95 .
+ lemmas [ lem : fullsupportlem1 ] and [ lem : fullsupportlem2 ] imply that @xmath386 \leq \mathbb{p } \left [ \sum_{k=1}^{k } \left| \left\langle { \boldsymbol{{\boldsymbol{\phi}}_j } } , { \boldsymbol{e}}_{k } \right\rangle \right| q_k \geq \varepsilon({\boldsymbol{q } } ) \right ] .
\end{aligned}\ ] ] furthermore , since @xmath387 , lemma [ lem : finalboundindivprob ] shows that @xmath388 \leq \exp \left ( - \kappa({\boldsymbol{q } } , { \boldsymbol{\sigma } } ) \overline{\varepsilon}({\boldsymbol{q } } , { \boldsymbol{\sigma}})^2 \right).\ ] ] whenever @xmath389 .
finally , the probability that somp - ns chooses correct atoms from iteration @xmath193 to iteration @xmath45 ( let us denote this event by @xmath390 ) satisfies @xmath391 \geq 1 - n \mathcal{c}_s \exp \left ( - \kappa({\boldsymbol{q } } , { \boldsymbol{\sigma } } ) \overline{\varepsilon}({\boldsymbol{q } } , { \boldsymbol{\sigma}})^2 \right ) .
\qedhere\ ] ] the theorem above translates several intuitive realities .
first , by examining the expression of @xmath269 , one concludes that several quantities influence the probability of correct recovery : * the expression @xmath392 quantifies the robustness of the recovery in the noiseless case .
it is evident that the margin of error when deciding which atom to choose in the noiseless case will contribute to determine the admissible noise level in the noisy case .
* the term @xmath393 translates the ability of the dictionary matrix @xmath12 to maintain the @xmath1 norm of @xmath187-sparse signals , which ensures that the norm of the columns of @xmath235 remain sufficiently high to avoid being absorbed by the noise matrix @xmath250 . *
the minimized sum @xmath394 depicts the idea that the weighted sum of the coefficients associated with every atom should be high enough to allow somp - ns to identify them when noise is added to the measurements .
this term simultaneously captures the influence of the signal to be recovered as well as that of the weights .
moreover , @xmath365 indicates that increasing the noise variances decreases the probability of support recovery .
also , increasing the weights naturally augments the power of the noise signal @xmath250 .
nevertheless , it should be expected that this effect is counterbalanced by @xmath395 as the latter variable is also a function of the weights .
finally , @xmath57 translates the fact that the noise affects every atom while @xmath396 takes into account the existence of @xmath45 iterations .
+ it is worth noticing that @xmath397 so that it indicates that the probability of full support recovery is lower bounded by @xmath398 on the basis of past results in the literature , it will be suggested in section [ subsec : gribonvalrelatedthm ] that @xmath396 is an artifact of our proof and should ideally be replaced with a function that increases more slowly as @xmath187 is augmented .
+ also , as a last remark , theorem [ thm : finalgaussianthm ] can be rephrased by stating that the probability of failure of somp - ns from iteration @xmath193 to iteration @xmath45 , _
i.e. _ , at least one wrong atom is chosen during the first @xmath354 iterations , admits the upper bound @xmath399
theorem [ thm : finalgaussianthm ] is now to be particularized using the coherence of @xmath12 instead of the rip and the erc .
[ thm : finalgaussianthmcoherence ] let @xmath65 ( @xmath49 ) be statistically independent gaussian random vectors respectively distributed as @xmath72 .
let @xmath400 .
let also @xmath356 be the signal matrix whose support is @xmath28 .
let also @xmath357 be unit - norm vectors in @xmath358 and @xmath359 be the corresponding matrix whose coherence @xmath123 is assumed to satisfy @xmath201 .
let @xmath361 .
then , somp - ns with dictionary matrix @xmath12 , weights @xmath114 ( @xmath49 ) and signal @xmath215 is ensured to make correct decisions during the first @xmath354 iterations , _
i.e. _ , from iteration @xmath193 to iteration @xmath45 included , with probability higher than @xmath401 whenever @xmath402 , where @xmath403 @xmath365 and @xmath318 being defined in equation ( [ eq : defkappa ] ) and ( [ eq : defb ] ) , respectively . as it has already been pointed out in section [ subsec : ercnoiseless ] , if @xmath201 , then @xmath202 for all the supports of size @xmath187 . moreover , equation ( [ eq : rictocoherence ] ) shows that @xmath404 whenever @xmath405 thus , using the two inequalities above in conjunction with theorem [ thm : finalgaussianthm ] yields @xmath406 elementary algebraic manipulations show that the expression above simplifies to @xmath407 a similar although stronger result can be obtained by means of the cumulative coherence function @xmath408 . however , the resulting expression of @xmath409 is more complicated .
moreover , coherence - based bounds only have a theoretical interest as they prove to be pessimistic for practical cases , even when using the cumulative coherence function .
a result similar to theorem [ thm : finalgaussianthm ] has already been obtained in @xcite .
the striking similarities between our result and that obtained in @xcite motivate this section .
they prove the following theorem .
* theorem 7 ) [ thm : gribonval ] let @xmath410 with @xmath411 a @xmath412 matrix of standard gaussian random variables , @xmath413 and @xmath204 an error term orthogonal to the atoms in @xmath187 .
assume that the dictionary matrix @xmath12 satisfies the restricted isometry property ( rip ) with restricted isometry constant ( ric ) ( of order @xmath414 ) @xmath415 and @xmath416 then , the probability that @xmath187 iterations of somp fail to exactly recover the support @xmath28 on the basis of @xmath102 does not exceed @xmath417 with @xmath47 the number of measurement vectors , @xmath57 the number of atoms and @xmath418 in @xcite , the statistical analysis has been focused on the sparse signal to be estimated , _
i.e. _ , @xmath186 , while no particular statistical distribution is assumed for @xmath204 .
since the noise vectors are assumed to be orthogonal to the columns of @xmath186 , their purpose is to model an approximation noise , _
i.e. _ , the part of the signal @xmath102 that can not be mapped by @xmath73 .
conversely , the noise signals envisioned in our paper represent additive measurement noises that can not be assumed to be orthogonal to any vector subspace . + in the noiseless case , a result similar to theorem [ thm : gribonval ] has been obtained in ( * ? ? ?
* theorem 6.2 ) for a variant of somp entitled 2-somp , _
i.e. _ , a somp algorithm where the decision on which atom to pick is performed on the basis of the maximization of a @xmath1 norm instead of a @xmath0 norm . +
although the problem addressed in this paper is different from that of theorem [ thm : gribonval ] , the authors of @xcite have used approaches similar to ours and they also noticed that the parasitic term @xmath396 seems difficult to remove .
the rationale of this discussion is thus that it is difficult to avoid the suboptimal term @xmath396 when conducting developments similar to that presented in this paper .
theorem [ thm : finalgaussianthm ] provides interesting insights into how successful somp - ns is whenever some parameters are modified .
however , the theoretical developments presented in this paper do not properly capture all the characteristics of somp - ns .
+ first of all , regarding @xmath419 , the value of @xmath57 should be lowered in practice .
the reason why @xmath57 should be replaced by @xmath420 is because our analysis assumes that all the atoms not to be picked ( of index @xmath11 ) are such that @xmath421 while , in practice , only a few atoms are likely to exhibit a significant value of @xmath422 . + we conjecture that @xmath423 may be replaced by a linear function of @xmath187 .
the reason why we failed at obtaining such a result is probably linked to the proof of theorem [ thm : finalgaussianthm ] . all the possible supports at each iteration are considered and it is ensured that the sufficient condition for making a correct decision is satisfied for each support .
this is very pessimistic as only one support out of all the numerous possibilities actually matters .
as indicated in section [ subsec : gribonvalrelatedthm ] , other researchers working on similar problems have stumbled upon this issue and no solution has been found so far to the best of the authors knowledge .
the simulation results presented in the end of this paper will however not address this problem .
+ furthermore , as it will be explained in section [ subsec : summarynumres ] , the bias @xmath318 should be removed in order to deliver results compatible with what has been observed in our numerical simulations .
this term is most likely an artifact linked to equation ( [ eq : firstbigapprox ] ) and equation ( [ eq : secondbigapprox ] ) .
equation ( [ eq : firstbigapprox ] ) basically assumes that @xmath424 and @xmath425 always have opposite signs for atoms @xmath13 whose indexes @xmath11 belong to the support @xmath28 .
conversely , equation ( [ eq : secondbigapprox ] ) assumes that @xmath424 and @xmath425 always have identical signs for atoms whose indexes do not belong to the support @xmath28 .
thus , a statistical analysis directly performed on @xmath426 may prevent the bias term from appearing .
+ therefore , considering all the conjectures above , a better bound may be given by @xmath427 where @xmath57 has been replaced with @xmath420 , @xmath428 has been substituted to @xmath396 and @xmath318 has been canceled out . + finally , it is worth pointing out that equation ( [ eq : b1 ] ) will likely not deliver perfect results . indeed ,
equation ( [ eq : trueequationmodel ] ) suggests that the probability of sparse support recovery success is actually a sum of exponential functions , each function corresponding to a single atom , iteration and support .
moreover , possibly different values of @xmath429 should be chosen for each exponential function .
it is also probably suboptimal to make use of the union bound .
this section aims at demonstrating that : 1 .
somp - ns provides significant performance improvements when compared to somp provided that the noise vectors exhibit different variances and that the weights are properly chosen .
2 . depending on the signal model that is chosen , it is possible to accurately estimate the optimal weights by using simple closed - formed formulas derived from equation ( [ eq : b1 ] ) .
3 . equation ( [ eq : b1 ] ) properly predicts the performance improvements obtained when the number of measurement vectors increases .
the purpose of the first point is to demonstrate numerically that the gains provided by somp - ns are significant .
the last two points rather focus on the numerical validation of the theoretical analysis presented in this paper . with these goals in mind
, a particular signal model will be chosen .
in particular , this signal model will be sufficiently simple to allow the computation of the optimal weights when the model ( [ eq : b1 ] ) is assumed to be correct .
the objective of this section is to particularize theorem [ thm : finalgaussianthm ] to models for which @xmath430 ( @xmath49 , @xmath431 ) and @xmath432 ( @xmath49 , @xmath433 ) where the terms @xmath434 denote rademacher random variables , _
i.e. _ , random variables that return either @xmath374 or @xmath435 with probability @xmath436 for both outcomes .
+ two different sign patterns will be distinguished .
sign pattern 1 refers to the case where the sign pattern is identical for all the sparse vectors @xmath55 to be recovered , _
i.e. _ , @xmath437 for all @xmath438 , @xmath431 and @xmath439 whenever @xmath440 .
sign pattern 2 corresponds to the situation where the sign pattern is independent for each and within each @xmath55 , _
i.e. _ , @xmath441 whenever @xmath440 and/or @xmath275 .
+ in both cases , it is worth mentioning that the absolute values of the entries of each @xmath55 are equal to @xmath442 . thus , @xmath443 and , according to theorem [ thm : finalgaussianthm ] , the probability of full support recovery is always higher than @xmath444 where @xmath445 .
also , the bound above only holds if @xmath446 . + by using the conjecture about the elimination of the bias term @xmath318 described in section [ subsec : conjectures ] ,
one obtains equation ( [ eq : modelconjecture ] ) , which is reminiscent of equation ( [ eq : b1 ] ) , except that ( [ eq : modelconjecture ] ) does not include the conjectures linked to the term @xmath419 . if our only objective is to derive the optimal weights according the theoretical model , only the argument of the exponential matters so that adjusting the term @xmath419 has no effect .
@xmath447 let @xmath448 denote the arithmetic mean of the entries of vector @xmath10 .
then , @xmath449 and @xmath450 .
thus , the probability of full support recovery is always higher than @xmath451 as a particular case , if @xmath452 ( @xmath453 ) , the latter probability also rewrites @xmath454 } \rangle } \varepsilon'^2 \right).\ ] ] let us now focus our attention on the weights that maximize equation ( [ eq : b2 ] ) or , equivalently , @xmath455 .
we can restrict our attention to the maximization of @xmath456 we define @xmath457 so that the expression above also reads @xmath458 the quantity @xmath459 represents the direction of @xmath460 so that we know that a global maximizer is obtained whenever @xmath460 and @xmath461 have the same direction .
it means that a global maximizer is obtained if and only if @xmath462 where @xmath463 .
this is equivalent to requiring @xmath464 since @xmath457 . by choosing @xmath465 ,
one concludes that @xmath466 provides an optimal weighting strategy according to ( [ eq : b2 ] ) .
now that the optimal weighting strategy for the signal models envisioned in section [ subsec : specialsigmodel ] has been derived , it becomes possible to determine whether the bound ( [ eq : b2 ] ) properly predicts the impact of the weights on the performance that is achieved .
+ our matlab simulation software is available in @xcite .
all the scripts needed to generate the figures presented in this section are also available . the reader should know that every simulation result exposed hereafter has been performed by using single precision floating point representations .
the reason for such a choice is that single precision arithmetic is faster and thus preferred for algorithms intended to run on real - time platform such as somp - ns .
for the same reason , the simulation results are obtained faster when using the single precision format . + it is assumed that @xmath467 and @xmath468 .
the simulation framework consists of a fixed dictionary matrix @xmath469 whose entries were generated on the basis of independent and identically distributed gaussian random variables and then normalized in such a way that each column of the matrix exhibits a unit @xmath1 norm .
this matrix is fixed for all the simulations and is available in @xcite .
+ two simulation frameworks have been envisioned to demonstrate the three points introduced at the very beginning of section [ sec : numresults ] .
the first framework consists of simulations for the case @xmath470 and addresses the first two objectives while the last framework examines how the probability of successful support recovery evolves as @xmath47 increases .
first of all , it is worth pointing out that the performance achieved by is invariant if the weight vector @xmath471 is multiplied by a positive constant .
therefore , only the angle @xmath472 is investigated .
the weights are thus generated on the basis of the polar coordinate system @xmath473 , @xmath474 where @xmath475 . in practice ,
the grid of weighting angles @xmath476 will consist of @xmath477 uniformly spaced angles from @xmath478 to @xmath479 .
+ the noise standard deviation vector @xmath480 is generated on the basis of the polar coordinate system @xmath481 where @xmath482 describes the orientation of the noise vector .
the grid of values for @xmath482 is composed of @xmath477 uniformly spaced angles ranging from @xmath483 to @xmath484 .
extremely high or low angles have been avoided because they correspond to situations for which the noise concentrates essentially on one measurement vector .
therefore , appropriate weighting strategies would be able to cancel most of the noise and would lead to probabilities of correct support recovery that are too high to be reliably estimated on the basis of a limited number of monte carlo cases . + a total of six simulation configurations have been run .
each configuration corresponds to one of the two sign patterns described in section [ subsec : specialsigmodel ] , to a support size @xmath187 and to a value of @xmath442 .
simulation cases have been generated for each value of @xmath482 belonging to the grid defined beforehand .
once the support size is fixed , the actual support is randomly and uniformly chosen among all the possibilities .
the support that is simulated is independent for each case .
+ although @xmath485 , the input signal - to - noise ratio ( snr ) , referred to as @xmath486 and defined in equation ( [ eq : defsnrinput ] ) , is to be modified by means of the quantity @xmath442 .
+ @xmath487 table [ tab : configssimk2 ] describes all the configurations that have been investigated numerically .
the values of @xmath442 have been chosen in such a way that the probability of full support recovery when @xmath488 is approximately equal to @xmath489 and @xmath490 for the sign pattern 1 and sign pattern 2 respectively .
these probabilities have been chosen in such a way that the probability of successful full support recovery over the grid defined for @xmath491 never reaches values so high that it can not be reliably estimated on the basis of a limited number of simulation experiments .
the lower value of the probability of successful recovery for sign pattern @xmath374 is linked to the fact that having a sign pattern independent for each measurement vector , as for sign pattern 2 , provides performance improvements .
this observation is actually reminiscent to what has been established in theorem [ thm : gribonval ] .
+ in table [ tab : configssimk2 ] , the input snr , _
i.e. _ , @xmath486 , is estimated by generating @xmath492 cases for the configuration of interest , then the input snr ( in db ) is computed for each case and the results are finally averaged .
the matlab script implementing this estimation is available in @xcite .
.simulation configurations | @xmath470 [ cols= " < , < , < , < , < , < " , ] as an example , figure [ fig : simanglesplotk2conf2 ] depicts the probability of full support recovery for configuration @xmath493 as a function of @xmath491 .
+ ) for configuration @xmath493 ( sign pattern @xmath374 , @xmath494 , @xmath495 and @xmath496 db ) probability of full support recovery as a function of @xmath476 and @xmath482 the black curve refers to the optimal weights derived from equation ( [ eq : b2 ] ) each pixel of the figure has been computed on the basis of @xmath497 simulation cases , width=491 ] the first objective we would like to fulfill is demonstrating that somp - ns is capable to outperform somp whenever the noise standard deviations are not identical for each measurement vector . to do so , we will examinate , for each configuration and for each value of @xmath482 , what is the ratio of the probability of failure obtained for @xmath498 , _
i.e. _ , the weights corresponding to somp , to the lowest probability of failure , _
i.e. _ , that obtained for the truly optimal weights . figure [ fig : simanglessompvsoptimal ] plots the aforementioned quantity for the all the configurations of table [ tab : configssimk2 ] .
note that the optimal weights are determined on the basis of the numerical results , the formula @xmath499 obtained on the basis of equation ( [ eq : b2 ] ) is not used .
+ it is observed that the gains provided by the proper application of somp - ns are significant , especially for low values of the support size .
the only case for which the gain is almost nonexistent is for sign pattern @xmath374 , @xmath500 and @xmath501 , _ i.e. _ , configuration @xmath502 .
this may actually be a consequence of the normalization procedure of @xmath442 described previously which ensures that the probability of successful support recovery for @xmath488 is equal to @xmath489 .
the value of @xmath442 had to be chosen significantly higher than that of the other cases to attain the @xmath489 goal , which limits the impact of the noise and thus hinders the improvement of the performance by modifying the weights .
+ let us now attack the second objective of section [ sec : numresults ] .
we wish to show that the formula @xmath499 obtained in section [ subsec : specialsigmodel ] delivers reliable estimates of the optimal weights .
figure [ fig : simanglesoptimalvstheory ] summarizes the results .
first of all , it is observed that , for the first sign pattern , the solution @xmath499 always corresponds to the truly optimal weights while this is not true for the second sign pattern . in particular , the discrepancy between the numerical results and the theoretical formula ( [ eq : b2 ] ) increases as the size of the support augments .
+ the first observation is explained by the different sign patterns for each measurement vector .
although the weights have been introduced to better filter the influence of the noise , they also have an impact on the relative importance of each @xmath55 in the decisions that are taken . given that the sparse vectors @xmath55 to be recovered have identical distributions , it is to be expected that , without noise , the optimal weights are obtained by choosing @xmath498 for symmetry reasons .
figure [ fig : simanglesplotk2conf6nonoise ] displays the simulation results obtained for a configuration identical to configuration @xmath503 except that @xmath504 db , _
i.e. _ , the influence of the noise is negligible .
it is observed that the interaction of the weights and the sparse vectors to be recovered exists and that the optimal weighting angle is equal to @xmath505 .
a possible interpretation of the observations above is that the optimal weighting strategy is a mixture of the strategy that optimizes the support recovery in the noiseless case and of that which minimizes the impact of the noise on the decisions that are taken , _
i.e. _ , @xmath499 .
nevertheless , further theoretical developments should be conducted to assess whether the proposed interpretation is correct .
finally , it is worth pointing out that the phenomenon described above is not observed for sign pattern @xmath374 because @xmath506 and @xmath507 are identical in this case .
+ the reasons that explain why the optimal weights get closer to @xmath498 when the support size increases , as shown in figure [ subfig : sp2optangles ] , is not clear and would require additional theoretical investigations that fall outside of the framework of this paper . + ) for configuration @xmath503 with probability of full support recovery as a function of @xmath476 and @xmath482 sign pattern @xmath493 @xmath501 @xmath508 # cases @xmath509,width=491 ] the final question of whether the proposed theoretical analysis properly conveys the properties of somp - ns whenever @xmath47 increases is to be discussed in this section .
our main objective is to show that , as predicted by equation ( [ eq : b2 ] ) , the probability of failure of somp - ns decreases linearly with @xmath47 when it is plotted is semi - logarithmic axes . indeed ,
equation ( [ eq : b2 ] ) yields @xmath510 where @xmath511 denotes the probability of failure of correct support recovery . + the simulation framework consists of a fixed weighting strategy for which all the weights are equal , _
i.e. _ , the weighting strategy corresponds to somp . for each value of @xmath47 ,
the noise vector is given by @xmath512 so that it is reminiscent of the noise vector defined in section [ subsec : simresk2 ] for @xmath470 .
the results are plotted in figure [ fig : simplotk ] .
the configurations that have been chosen are directly inspired of those presented in table [ tab : configssimk2 ] .
the number of cases simulated for each curve and each value of @xmath47 is equal to @xmath513 .
some configurations have been discarded because they do not exhibit a probability of failure equal to @xmath193 without noise , which is incompatible with the implicit assumption of our theoretical model that no errors are committed in the noiseless case . indeed , if @xmath514 , _
i.e. _ , the erc is the noiseless case is not satisfied , then equation ( [ eq : sompanalysis2 ] ) can not hold .
for example , figure [ fig : simanglesplotk2conf6nonoise ] shows that configuration @xmath503 from table [ tab : configssimk2 ] exhibits a non - zero probability of failure without noise .
+ sp refers to sign pattern , width=491 ] the principal observation for figure [ fig : simplotk ] is that the slope of @xmath511 in semi - logarithmic axes is linear with regards to @xmath47 .
this observation provides evidence that the theoretical model conveys the behavior of somp - ns when @xmath47 increases .
the numerical results have revealed the following interesting facts : 1 .
somp - ns provides significant performance improvements when compared to somp provided that the weights are properly chosen and that the noise variances are different for each measurement vector .
the formula @xmath499 derived from equation ( [ eq : b2 ] ) corresponds to the truly optimal weights whenever * the sparse vectors @xmath55 to be recovered are identical . *
the support size @xmath187 is low enough .
+ the exact reason why the formula @xmath499 gets less accurate as the size of the support increases remains an open question .
the theoretical analysis properly predicts the characteristics of the decrease of the probability of failure of somp - ns whenever the number of measurement vectors increases . besides the three points above , which answer the three questions enumerated at the beginning of section [ sec : numresults ] , the close fitting of @xmath499 and the numerical results for sign pattern @xmath374 suggests that the bias term @xmath318 is indeed an artifact of our developments as adding it would change the theoretically optimal weights and we would then observe a mismatch between them and those obtained by simulation .
as suggested in section [ subsec : conjectures ] , the bias @xmath318 could be removed by avoiding to make use of the inequalities ( [ eq : firstbigapprox ] ) and ( [ eq : secondbigapprox ] ) . using a more subtle approach than the use of the union bound could also yield performance improvements .
+ although it appears to be difficult , replacing the term @xmath423 by a function that depends linearly on @xmath187 would be of great interest , especially since it would close a hole in the literature regarding the performance of the well - known somp algorithm that is a particular case of ours .
+ finally , extending the presented analysis by performing a joint statistical analysis of both the noise and the sparse signals to be recovered would be of great interest . to begin with
, it would provide a theoretical model that predicts the truly optimal weights by simultaneously taking into account how they impact the sparse signals to be recovered and the noise vectors .
next , it would also enable one to comprehend why the discrepancy between the formula @xmath499 and the truly optimal weights increases as the support size augments ( see figure [ fig : simanglesoptimalvstheory ] ) .
finally , the statistical analysis of the sparse signals could replace @xmath419 by @xmath515 ( where @xmath516 is significantly lower than @xmath57 ) as conjectured in section [ subsec : conjectures ] .
a novel algorithm entitled somp - ns that generalizes somp by associating weights with each measurement vector has been proposed . a theoretical framework to analyze
this algorithm has been built .
lower bounds on the probability of full support recovery by means of somp - ns have been developed in the case where the noise corrupting the measurements is gaussian .
numerical simulations have revealed that the developed theoretical results accurately depict key components of the behavior of somp - ns while they also fail to capture some of its properties . in particular , it has been shown that , under the right circumstances , the weights of somp - ns can be efficiently optimized on the basis of the proposed theoretical bounds . finally , the reasons that explain why some characteristics of somp - ns are not properly conveyed by the theoretical analysis have been discussed and potential workarounds to be investigated have been suggested .
r. adamczak , a. e. litvak , a. pajor , and n. tomczak - jaegermann , `` restricted isometry property of matrices with independent columns and neighborly polytopes by random sampling , '' _ constructive approximation _ , vol .
1 , pp . 6188 , 2011 , springer .
e. j. cands , j. romberg , and t. tao , `` robust uncertainty principles : exact signal reconstruction from highly incomplete frequency information , '' _ information theory , ieee transactions on _ , vol .
2 , pp . 489509 , 2006 .
j. f. determe , j. louveaux , l. jacques , and f. horlin , `` somp - ns software package , '' version 0.3.2 , available online at https://opera-wireless.ulb.ac.be/owncloud/index.php/s/lhkmncgou66mtes or http://bit.ly/1qg0y1g .
d. l. donoho , and m. elad `` optimally sparse representation in general ( nonorthogonal ) dictionaries via @xmath5171 minimization , '' _ proceedings of the national academy of sciences _ , vol . 100 , no .
5 , pp . 21972202 , 2003 .
d. l. donoho , m. elad , and v. n. temlyakov `` stable recovery of sparse overcomplete representations in the presence of noise , '' _ information theory , ieee transactions on _ , vol .
1 , pp . 618 , 2006 .
r. gribonval , h. rauhut , k. schnass and p. vandergheynst , `` atoms of all channels , unite ! average case analysis of multi - channel sparse recovery using greedy algorithms , '' _ journal of fourier analysis and applications _ , vol .
14 , no . 5 - 6 , pp .
655687 , 2008 .
d. needell , and r. vershynin , `` signal recovery from incomplete and inaccurate measurements via regularized orthogonal matching pursuit , '' _ selected topics in signal processing , ieee journal of _ , vol .
2 , pp . 310316 , 2010 .
j. a. tropp , and a. c. gilbert , `` simultaneous sparse approximation via greedy pursuit , '' _ acoustics , speech , and signal processing , 2005 .
proceedings.(icassp05 ) .
ieee international conference on _ , vol . 5 ,
pp . 721724 , 2005 .
a. m. tillmann , and m. e. pfetsch , `` the computational complexity of the restricted isometry property , the nullspace property , and related concepts in compressed sensing , '' _ information theory , ieee transactions on _ , vol .
2 , pp . 12481259 , 2013 . | this paper studies the joint support recovery of similar sparse vectors on the basis of a limited number of noisy linear measurements , _
i.e. _ , in a multiple measurement vector ( mmv ) model . the additive noise signals on each measurement vector
are assumed to be gaussian and to exhibit different variances . the simultaneous orthogonal matching pursuit ( somp ) algorithm is generalized to weight the impact of each measurement vector on the choice of the atoms to be picked according to their noise levels .
the new algorithm is referred to as somp - ns where ns stands for noise stabilization . to begin with , a theoretical framework to analyze the performance of the proposed algorithm is developed .
this framework is then used to build conservative lower bounds on the probability of partial or full joint support recovery .
numerical simulations show that the proposed algorithm outperforms somp and that the theoretical lower bound provides a great insight into how somp - ns behaves when the weighting strategy is modified . |
quantum coherent control of photo - induced processes @xcite directs a quantum system from its initial state to desired final state(s ) by manipulating interferences among coherent excitation pathways . when using strong femtosecond laser pulses , control can be realized based on the broad spectrum , high intensity and ultrashort duration of the pulses by tailoring their temporal shape @xcite .
coherent control using tailored short pulses is nowadays successfully employed in many different areas of quantum physics , including photoionization and fragmentation , nuclear magnetic resonance and quantum information processing .
however , the dream of coherently controlling photo - induced bimolecular chemical reactions @xcite which was at the origin of coherent control , still stands open .
for example , it is not yet understood under which conditions a bimolecular reaction can be controlled and which reaction mechanisms are particularly susceptible to control .
this dream has attracted significant interest over the years including that of moshe shapiro @xcite since it holds the promise of creating a new type of photochemistry .
two possible scenarios for the coherent control of a bimolecular reaction have been discussed in the literature .
the first one relies on preparing coherent superposition states , in order to determine the outcome of the reaction @xcite . realizing this scenario requires control over reactive collisions which currently seems to be extremely challenging .
given , however , recent advances with cold @xcite and ultracold molecules @xcite , we may witness the advent of such a coherent chemistry in the not too distant future .
the second scenario utilizes interference in time rather than frequency domain , i.e. , it is built on short , shaped pulses to induce bond making ( photoassociation ) and breaking ( photodissociation ) @xcite .
the coherence of the reactants required for controlling bond making must then be prepared during the photo - induced process .
this is the reason why coherent control of a binary reaction has proven so difficult .
the quantumness of an ensemble of reaction partners can be measured in terms of the purity , @xmath0 $ ] , where @xmath1 describes the state of the reaction partners , initially a thermal ensemble , @xmath2 $ ] .
note that the purity is closely related to the entropy of the sample ; and thus the coherent interaction of the reaction partners with the laser light does not change the purity .
preparation of coherence can still be achieved by addressing a subensemble of reactants .
one possibility is given by filtering out collision pairs in a small franck - condon window close to the repulsive barrier of the potential , as demonstrated by us in multi - photon femtosecond photoassociation of hot magnesium atoms @xcite .
the selected collision pairs are correlated due to the strong interaction close to the barrier .
correspondingly , the subensemble of photoassociated , electronically excited molecules shows coherent transient vibrational and rotational dynamics that were identified using pump - probe femtosecond spectroscopy @xcite .
the generation of molecular coherence was shown to be possible , even though the thermal ensemble of atoms was completely incoherent , with an initial purity of essentially zero .
the generated coherences are amenable to coherent control , as demonstrated by us very recently @xcite : the yield of detected magnesium dimers is enhanced by positively chirped pulses and suppressed by negatively chirped pulses . our _ ab initio _ model has revealed that control is achieved by purification combined with chirp - dependent raman transitions .
then , in closed - loop optimization experiments , an improved pulse was obtained that displays an effective temporal linear chirp and is composed of two sub - pulses .
the latter allows for utilizing coherent vibrational dynamics in addition to the chirp - dependent raman transitions @xcite . here
, we utilize the insight gained earlier @xcite and demonstrate an even higher degree of coherent control over bond making in femtosecond photoassociation of hot magnesium atoms , achieving up to an order - of - magnitude enhancement of the yield compared to the unshaped transform - limited pulse .
the highly performing phase - shaped pulses all display an overall temporally increasing instantaneous frequency and are composed of several overlapping sub - pulses , where the time separation between the first two sub - pulses almost perfectly fits the vibrational frequency of the generated intermediate wavepacket .
the corresponding spectral phases have the general form of a cosine function .
these findings can again be rationalized by chirp - dependent raman transitions and exploitation of vibrational dynamics as underlying control mechanisms .
-state scattering states as well as the band of rovibrational levels after two - photon excitation and that from which uv emission is observed . ]
pairs of magnesium atoms are photoexcited to form a magnesium dimer , @xmath3 .
figure [ extplot ] presents the bond - making excitation scheme , starting from the van der waals ground electronic state , @xmath4 , and using a non - resonant two - photon transition to the @xmath5 as a first step @xcite .
initially , the magnesium atoms , held at a temperature of 1000@xmath6k , thermally populate a manifold of @xmath7-state scattering states , indicated by the blue - shaded area in fig .
[ extplot ] .
the excitation is driven by a strong , linearly polarized , shaped femtosecond pulse , characterized by a transform - limited ( tl ) duration of 70@xmath6fs and a central wavelength of 840@xmath6 nm .
the pulse energy corresponds to a tl peak intensity of 7.2@xmath810@xmath9w/@xmath10 . following the irradiation , atom pairs with short internuclear separation
undergo bond formation via a free - to - bound transition into bound levels of a manifold of electronically excited states of mg@xmath11 .
the gateway for this excitation is the excited state @xmath12 , and the corresponding range of rovibrational levels that are populated in the @xmath13 state is indicated in fig .
[ extplot ] ( in - between red dashed lines ) .
the photoassociated molecules are detected by measuring the intensity of their uv fluorescence back to the ground electronic state , with a lifetime of a few ns , in the wavelength range of 285 - 292@xmath6 nm .
this corresponds to uv emission from below the atomic asymptote @xmath14 of excited mg atoms ( see the dashed blue arrow schematically shown in fig .
[ extplot ] ) .
the corresponding range of uv - emitting levels which belong to the @xmath15 , @xmath16 , and @xmath17 electronically excited states , is also indicated in fig .
[ extplot ] ( in - between dashed black lines ) .
the total uv signal originates from an incoherent summation over the signals obtained for all thermally populated initial scattering states which are characterized by their angular momentum @xmath18 and collision energy @xmath19 .
we have previously shown @xcite that the main contribution to the signal is from high partial waves , with @xmath18 around 70 , where due to the centrifugal barrier the shallow van - der - waals potential well of the x - state supports only very few bound levels or no bound levels at all .
this also implies that , as we have calculated @xcite , the bound - bound contribution to our signal corresponding to excitation from bound levels in the @xmath7-state amounts to less than 20@xmath20 of our total signal .
the bond - making excitation is controlled using phase - shaped femtosecond pulses .
our earlier study @xcite employed pulses with a linear chirp having a spectral phase of the form @xmath21 , where @xmath22 is the central frequency and @xmath23 is the linear chirp parameter .
the best linearly chirped pulse , with @xmath24ps@xmath25 , enhances the uv signal , compared to the tl pulse , by a factor of 5.4 .
closed - loop optimization experiments have led to an improved pulse , yielding an enhancement factor of 7.4 , which features an effective temporal linear chirp and is composed of two sub - pulses . here ,
we investigate two additional sets of shaped pulses , all having a cosine spectral phase of the form @xmath26\,,\ ] ] characterized by three parameters , @xmath27 , @xmath28 and @xmath29 .
such a spectral phase temporally yields a sequence of sub - pulses with a constant time difference @xmath28 between subsequent sub - pulses . once the value of @xmath28 is smaller than or comparable to the tl pulse duration , 70@xmath6fs in our case , the overall pulse results from several overlapping sub - pulses that interfere with each other . hence , with @xmath29=0 , and @xmath27 and @xmath28 having appropriate values , the spectral phase of eq . results in a pseudo - chirped pulse .
such a pulse has a symmetric temporal intensity profile with no splitting into separated sub - pulses .
it therefore can be assigned a well - defined pulse duration calculated as its full width at half maximum . at the same time , during the pulse s main intensity peak , the instantaneous temporal frequency increases , respectively decreases , monotonically .
hence , a pseudo - chirped pulse can be characterized by pulse duration and slope of the instantaneous frequency .
examples of pseudo - chirped pulses are shown below in fig .
[ ataumap](a)-(f ) . for a given bandwidth ,
a set of such pulses includes more combinations of pulse duration and instantaneous frequency slope than a set of linearly chirped pulses .
this allows for a systematic investigation of the interplay of chirp - dependent raman transitions and exploitation of vibrational coherence by temporally spaced pulses , and is the first set of pulses we employ here .
the second set of shaped pulses we utilize has cosine spectral phases with @xmath30 . with an additional proper choice for the values of @xmath27 and @xmath28 ,
the resulting shaped pulses are asymmetric in time and effectively composed of several , partially overlapping sub - pulses .
each sub - pulse has its own intensity and width , and the time difference between subsequent sub - pulses can change across the complete sequence .
these shaped pulses also exhibit an instantaneous frequency that overall increases , respectively decreases , in time .
examples of these asymmetric pulses are found below in fig .
[ ataumapm57](a)-(f ) . for a given bandwidth
, this set of pulses allows for a systematic investigation of the role of the time delay between the sub - pulses as well as their relative intensity .
the magnesium photoassociation experiments are conducted in a heat pipe with ar buffer gas heated to 1000 k , in which the pressure of the mg vapor is around 5 torr .
although a buffer gas is needed for the stable and continuous operation of the heated cell over long time , it does not play a part in the reaction .
this is confirmed by observation of a uv signal even without the buffer gas .
furthermore , the observed uv signal remains the same when replacing argon with helium as a buffer gas .
as we have previously discussed @xcite , the uv signal can be ruled out to result from mg ionization @xcite .
past experimental work shows that the existence of mg clusters larger than a dimer is highly negligible in our heat pipe @xcite . we therefore conclude that our measured signal results only from excitation of magnesium dimers .
the shaped femtosecond laser pulses irradiate the sample at a repetition rate of 1@xmath6khz , after undergoing shaping using a liquid - crystal spatial light phase modulator @xcite .
the measured uv fluorescence signal is collected at the laser beam entrance to the heat pipe , at a small angle from the beam axis , using a dichroic mirror , filter and lens , focusing the uv light onto an optical fiber that is coupled to a spectrometer and a time - gated camera system with a 5@xmath6ns gate .
we employ the theoretical framework developed in refs .
@xcite to describe photoassociation of hot atoms by an intense laser field .
our first principles model combines _ ab initio _ electronic structure theory for the mg@xmath31 molecule with quantum molecular dynamics to describe the light - matter interaction in a non - perturbative way .
random phase wavefunctions are employed to account for thermal averaging @xcite . in brief
, we explicitly include all electronic states shown in fig .
[ extplot ] .
the @xmath32 ground electronic state is coupled to the @xmath12 state by a two - photon coupling .
then , the @xmath12 state is coupled by resonant one - photon couplings with a several states of _ ungerade _ symmetry , the most important of which are @xmath15 , @xmath16 and @xmath17 .
in addition , there are near - resonant one - photon couplings with the @xmath33 states . moreover , since the system is exposed to a strong field , we also include the stark shifts due to one - photon couplings with off - resonant states . the strong nonadiabatic interaction between the first two @xmath34 states is taken into account by transformation into the diabatic basis .
the general structure of the hamiltonian is given in eq .
( 12 ) of ref .
@xcite ; it is augmented here by the additional states shown in fig .
[ extplot ] and their one - photon couplings to the @xmath12 state .
the time - dependent schrdinger equation is solved using the chebyshev propagator @xcite separately for each thermal random - phase basis wavefunction and rotational quantum number @xmath18 @xcite .
any coupling between rotational and vibrational degrees of freedom is neglected to ease the computational effort .
time - dependent , thermally averaged expectation values are calculated by proper weighting for each partial wave , cf .
( 25 ) and ( 26 ) of ref .
when the propagation is over , the final populations of the bound levels in the @xmath15 , @xmath16 and @xmath17 excited states are used to generate the integrated emission spectrum , which is the experimental observable . for each electronic state ,
the emission spectrum is obtained by means of the einstein coefficients with appropriate transition dipole moments and including the line broadening .
) . left : experimental uv signal enhancement relative to the tl pulse ( ef ) as a function of @xmath28 and @xmath27 , cf .
right : examples of pseudo - chirped pulses yielding large enhancement .
the intensity profile in time domain and the corresponding instantaneous frequency are plotted for : ( a ) @xmath35 , @xmath36 , ( b ) @xmath37 , @xmath38 , ( c ) @xmath39 , @xmath40 , ( d ) @xmath41 , @xmath36 , ( e ) solid line : @xmath42 , @xmath43 , dashed line : @xmath44 , @xmath43 , ( f ) solid line : @xmath45 , @xmath46 , dashed line : @xmath47 , @xmath46 .
the corresponding values of the measured enhancement ( ef ) are also indicated . ]
we first discuss the case @xmath48 in eq . .
figure [ ataumap ] presents the measured enhancement factor ( ef ) , i.e. , the ratio of the uv signal for the shaped pulse over the signal for the tl pulse , as a function of the phase shape parameters @xmath28 and @xmath27 .
the maximal enhancement for cosine phase shapes with @xmath48 is found to be @xmath49 , slightly larger than the value for linearly chirped pulses . for positive values of @xmath27 , we find suppression of the signal ( ef@xmath501 ) , similarly to the results for linearly chirped pulses with negative chirp @xcite .
the right - hand side of fig .
[ ataumap ] displays examples of pseudo - chirped pulses which produce a large enhancement .
as explained above in section [ sec : scheme ] , pseudo - chirped pulses can be characterized by their duration and slope of the instantaneous frequency . translating the spectral phase parameters @xmath27 and @xmath28 into these characteristics , fig .
[ slopesdursmap ] shows the enhancement factor as a function of pulse duration and temporal frequency slope . for comparison
the inset shows the same map for linearly chirped pulses .
clearly , the range of pulse durations between approximately @xmath51fs and 400@xmath6fs and instantaneous frequency slopes between approximately @xmath52thz / fs and @xmath53thz / fs is optimal for achieving large enhancement factors , between 4.5 to 6.6 .
this finding is similar for both cosine phase - shaped pulses and linearly chirped pulses . for the highest enhancement factors , above 6.0
, the optimal range of pulse durations and frequency slopes becomes narrower , with the corresponding values of 200@xmath54280@xmath6fs and [email protected]@xmath6thz / fs , respectively . the dependence on the sign of the chirp , respectively slope of the instantaneous frequency , is reflected in the asymmetry of the map with respect to the axis of zero slope : for negative slope , suppression rather than enhancement ( ef@xmath55 ) is found .
the corresponding pulses are shown by dashed lines in fig .
[ ataumap](e),(f ) .
the results show that there is an extended set of chirped and pseudo - chirped pulses , with an extended range of pulse durations and frequency slopes , that lead to a large enhancement of the detected photoassoication yield .
we associate the significant tolerance to the exact shape of the optimal pulse with the fact that , due to the initial thermal conditions , the photoassociation takes place from many different initial scattering states of the colliding atoms .
since each such initial state generally has a different energy and a slightly different wavefunction , the corresponding optimal pulse has a slightly different shape . ) . left : uv signal enhancement ( ef ) as a function of @xmath28 and @xmath27 , cf .
right : examples of asymmetric pulses with a sub - pulse structure yielding large enhancement .
the intensity profile in time domain and the corresponding instantaneous frequency are plotted for : ( a ) @xmath56 , @xmath57 , ( b ) @xmath58 , @xmath59 , ( c ) @xmath60 , @xmath61 , ( d ) @xmath45 , @xmath62 , ( e ) @xmath63 , @xmath61 , ( f ) @xmath39 , @xmath61 .
the corresponding values of the measured enhancement ( ef ) are also indicated . ]
the second set of cosine phase - shaped pulses uses @xmath30 in eq . , leading to pulses with a sub - pulse structure .
the experimental enhancement as a function of the parameters @xmath28 and @xmath27 is presented in the left - hand side of fig .
[ ataumapm57 ] , together with six pulses producing a large enhancement over tl pulses in the right - hand side .
the best asymmetric pulses result in higher enhancement factors than the best pseudo - chirped pulses . here
, the enhancement reaches an order of magnitude compared to tl pulses and is larger by about a factor of @xmath64 compared to the pseudo - chirped pulses . , i.e. , delay between the first and second sub - pulse ( red ) , @xmath65 ( blue ) , and @xmath66 ( green ) .
( b ) time delay ratio @xmath67 for the data shown in panel ( a ) .
the average value corresponds to 0.52 .
( c ) spectrum of the quantum beat frequencies obtained in the pump - probe experiment @xcite and frequencies corresponding to the delays shown in panel ( a ) , which fit the vibrational peak of the quantum beat spectrum . ]
the duration of these pulses is more difficult to determine since they are divided into several sub - pulses .
a better understanding of the advantage of these pulses over pseudo - chirped and linearly chirped pulses is obtained by inspecting the time delay between the sub - pulses .
the optimal time delays between sub - pulses are identified by considering only those asymmetric pulses that produce enhancement factors larger than @xmath68 , see fig .
[ ef_above8onqb](a ) . for the delay between the first two sub - pulses ( @xmath69 )
, the optimum is found to be around @xmath70fs ( red squares in fig .
[ ef_above8onqb](a ) ) .
the optimal delay between the second and third sub - pulse ( @xmath71 ) is around @xmath72fs ( blue squares in fig .
[ ef_above8onqb](a ) ) , consistent with half of the first delay , see also fig . [ ef_above8onqb](b ) .
the delay between the first two sub - pulses corresponds to a frequency of @xmath73@xmath74 . using femtosecond pump - probe spectroscopy ,
we have earlier determined the quantum beat frequencies relevant in the photoassociation process @xcite .
the corresponding fourier spectrum of the experimental transient pump - probe signal is compared in figure [ ef_above8onqb](c ) to the frequencies corresponding to the time delays of the asymmetric pulses .
the delay between the first two sub - pulses ( red squares ) is found to fit the vibrational part of the pump - probe signal .
the delay @xmath75 between the first and third sub - pulse provides a rough estimate of the overall pulse pulse duration . for the well performing asymmetric pulses of fig .
[ ef_above8onqb](a ) , @xmath75 is found to be in the range of 220@xmath6fs to 275@xmath6fs .
this is in good agreement with the optimal pulse durations of the well performing pseudo - chirped pulses reported in sec .
[ subsec : pseudochirped ] . , @xmath28 , @xmath76 ) , and their experimental ef values .
panels ( a ) and ( b ) , respectively ( c ) and ( d ) compare pulses with the same temporal intensity profile but overall increasing vs. overall decreasing instantaneous frequency .
additionally , the relative intensity between the sub - pulses is compared between ( a ) , ( b ) on one hand and ( c ) , ( d ) on the other . ( a )
@xmath56 , @xmath57 , @xmath30 , ( b ) @xmath77 , @xmath78 , @xmath30 , ( c ) @xmath56 , @xmath57 , @xmath79 , ( d ) @xmath77 , @xmath57 , @xmath30 . ]
all well performing asymmetric pulses , cf .
[ ef_above8onqb](a ) and examples in fig .
[ ataumapm57 ] , have a first sub - pulse that is of higher intensity and longer duration than the subsequent sub - pulses
. these pulses also display an overall increasing instantaneous frequency where the onset of the increase is located in the second part of the first sub - pulse . the role of delay and instantaneous frequency
is further analyzed in fig .
[ pulsesfreqsubscomp ] which displays four pulses together with the enhancement that they yield .
comparison of panels ( a ) and ( b ) , respectively ( c ) and ( d ) , which show pulses with the same temporal intensity profile but overall increasing vs. overall decreasing instantaneous frequency , confirm the earlier observation of a clear advantage of increasing compared to decreasing instantaneous frequency : an overall increasing ( decreasing ) instantaneous frequency results in enhancement ( suppression ) of the uv signal .
figure [ pulsesfreqsubscomp ] also allows to examine the role of relative intensity between the sub - pulses . comparing ,
for example , panels ( a ) and ( c ) which both feature an overall increasing instantaneous frequency but differ in the increase vs decrease of the sub - pulse peak amplitudes , reveal the advantage of a strong first sub - pulse : the enhancement in panel ( a ) is larger by a factor of 5 compared to ( c ) .
let us summarize the features of asymmetric pulses that are particularly advantageous for enhancing the observed uv signal : ( i ) a stronger and longer first sub - pulse , ( ii ) an overall increasing instantaneous frequency which starts to increase in the second part of the first sub - pulse , ( iii ) an optimal sub - pulse delay @xmath69 fitting exactly the @xmath12 vibrational period , and ( iv ) an optimal sub - pulse delay @xmath71 fitting half the @xmath12 vibrational period .
these findings suggest the following interpretation : the first sub - pulse creates a @xmath12 vibrational wavepacket , most likely at the right turning point , via free - to - bound two - photon photoassociation transitions .
the non - resonant nature of these transitions requires a strong and temporally wide sub - pulse . in the second part of the first sub - pulse ,
the frequency starts to increase such that , after half a vibrational period , when the wavepacket has reached the left turning point , part of the amplitude is efficiently transferred into lower vibrational levels of the @xmath12 state by resonant raman transitions via higher electronically excited states .
these raman transitions are favored by an instantaneous frequency increase @xcite .
then , after another half vibrational period , i.e. , one vibrational period after the photoassociation transition , the wavepacket has reached again the right turning point .
this coincides with the peak amplitude of the second sub - pulse , i.e. , with @xmath69 , such that another part of the wavepacket , larger than before , is efficiently and coherently excited into lower @xmath12 vibrational levels via up - chirped raman transitions .
last , after one more half vibrational period ( @xmath71 ) , the wavepacket is again at the left turning point , where the third sub - pulse drives another coherent raman transition into lower @xmath12 vibrational levels . as we have previously shown and discussed @xcite , the coherent enhancement of population in the lower vibrational levels of the @xmath12 state
is directly reflected in the final population of the observed uv emitting band , which is accessed from the former by a weak one - photon transition .
we now compare these experimental findings to simulation results .
for the two most efficient pulses , shown in fig .
[ ataumapm57](a ) and ( b ) , with experimental ef of 10 and 9.9 , the theoretically obtained enhancement amounts to 6.8 and 18.3 , respectively .
qualitatively , this confirms the strong enhancement of the uv signal for phase - shaped pulses compared to the tl pulse .
in particular , the largest theoretically found enhancement for cosine phase - shaped pulses is also larger than the maximum enhancement obtained with linearly chirped pulses .
however , due to the sensitivity of the dynamics on the details of the electronic structure data , as discussed in ref .
@xcite , it would be very challenging to reproduce the exact values for the experimental enhancement factors .
in particular , the maximum enhancement was found for larger chirp rates in the calculations than in experiment @xcite .
moreover , the agreement between experimental and theoretical results for the vibrational spectra of the pump - probe transients was satisfactory but not perfect @xcite .
a pulse which is experimentally found to maximally exploit the vibrational dynamics will therefore not necessarily be the best performing pulse in the calculations where the vibrational dynamics is slightly different .
state for two optimal pulses compared with a tl pulse .
the shaded area shows the band which can be transfered by a one - photon transition to the uv emission window below the @xmath80 atomic threshold . ]
figure [ pigvib ] presents the final vibrational distribution in the intermediate @xmath12 state . assuming that the final probe transition from the @xmath12 state into the uv emitting states can be treated perturbatively , it is the @xmath12 vibrational distribution that reveals the control mechanism @xcite .
for example , linearly chirped pulses were found to up- , respectively down - shift the @xmath12 vibrational distribution by raman transitions between the @xmath12 and higher lying electronic states .
this modifies in particular the population that is transferred by the probe photon , marked by the shaded area in fig .
[ pigvib ] , into the band of rovibrational levels in the _ ungerade_-states that produce the measured uv emission @xcite .
inspection of fig .
[ pigvib ] clearly shows that cosine phase - shaped pulses tend to transfer more population to the most deeply bound levels in the @xmath12 band than the tl pulse .
thus , the raman cycling process discussed in ref .
@xcite is efficient also in case of the new family of asymmetric sub - pulse structured pulses .
control of binary reactions is one of the most difficult tasks of coherent control .
the first step of two - photon franck condon filtering purifies the initial thermal state and thus sets the stage for control .
the target task in the present study was to control the generation of bound mg@xmath11 molecules in the band of vibrational levels of _ ungerade _ states just below the @xmath81p+@xmath81s dissociation threshold .
population in this band is detected experimentally by its transition back to the ground state which results in emission of uv light .
once a target is set we can decipher the interference control mechanism optimizing this task .
the present work demonstrates the power of a constructive approach based on cosine spectral phase patterns .
the constructive approach limits the number of control parameters and therefore enables a rational approach to optimization .
we are able to identify two control mechanisms electronic and vibrational .
the electronic component employs the full manifold of electronic surfaces generating interference pathways which are raman - like .
the electronic control mechanism enhances population in intermediate levels which are transfered to the target band by a weak final probe step .
in addition , we identify by the time delay between sub - pulses a vibrational mechanism which employs the interference between a direct excitation and a delayed excitation . we find a relatively high tolerance to the exact shape of the optimal pulse for enhancing the detected uv signal .
we associate it with the highly thermal character of the initial atomic ensemble which is held at a temperature of 1000@xmath6k .
the roadmap explored in the control of strong - field photoassociation of magnesium dimers opens the door to more complex binary reactions .
the main lesson learnt with mg@xmath31 can be summarized as follows : bond making requires a purification step which implies that only a relatively small number of reactants can be addressed .
it is then crucial to employ shaped femtosecond pulses of high intensity for maximizing this number as much as possible and , at the same time , coherently controlling the excitation dynamics to a very high degree .
on the other hand , it is also crucial to ensure an almost background - free detection of the photoassociated molecules .
it was achieved here by the choice of closed - shell reactants and a non - resonant multiphoton photoassociation transition .
this insight provides a blueprint for studying e.g. triatomic reactions which would allow for stable reaction products and two possible reaction outcomes .
the basic scenario of utilizing interference in time domain for the coherent control of binary reactions has thus been established .
thirty years after conception of the idea , perspectives for the coherent control of binary reactions finally look bright . | the first step in the coherent control of a photoinduced binary reaction is bond making or photoassociation .
we have recently demonstrated coherent control of bond making in multi - photon femtosecond photoassociation of hot magnesium atoms , using linearly chirped pulses [ levin et al .
, arxiv:1411.1542 ] .
the detected yield of photoassociated magnesium dimers was enhanced by positively chirped pulses which is explained theoretically by a combination of purification and chirp - dependent raman transitions .
the yield could be further enhanced by pulse optimization resulting in pulses with an effective linear chirp and a sub - pulse structure , where the latter allows for exploiting vibrational coherences . here
, we systematically explore the efficiency of phase - shaped pulses for the coherent control of bond making , employing a parametrization of the spectral phases in the form of cosine functions .
we find up to an order of magnitude enhancement of the yield compared to the unshaped transform - limited pulse .
the highly performing pulses all display an overall temporally increasing instantaneous frequency and are composed of several overlapping sub - pulses .
the time delay between the first two sub - pulses almost perfectly fits the vibrational frequency of the generated intermediate wavepacket .
these findings are in agreement with chirp - dependent raman transitions and exploitation of vibrational dynamics as underlying control mechanisms . |
one of the mysteries of elementary particle physics is the origin of electroweak symmetry breaking ( ewsb ) . in the standard model ( sm )
, ewsb occurs via the higgs mechanism , a consequence of which is the existence of a fundamental scalar particle , the higgs boson @xmath4 @xcite .
comparison of precision measurements of electroweak parameters with sm predictions indicates a preference for a light higgs boson @xmath5 gev @xcite .
the higgs boson has been searched for at collider experiments .
the current best limit on its mass is @xmath6 gev from searches for @xmath7 at lep2@xcite .
the cern lep2 collider is expected to ultimately reach a center - of - mass energy @xmath8 gev and integrated luminosity of @xmath9 per experiment , allowing an exploration of @xmath10 up to approximately @xmath11 gev .
experiments at the cern lhc ought to be able to observe @xmath4 for @xmath12 gev , although if @xmath10 is in the intermediate mass regime ( @xmath13 gev ) , several years of running may be required to extract the @xmath14 signal from qcd two photon backgrounds .
it has been pointed out that a high luminosity fermilab tevatron @xmath0 collider has significant reach for a higgs boson@xcite .
the most promising channel at the tevatron is @xmath15 , where @xmath16 and @xmath17 .
simulations of signal and sm backgrounds@xcite ( mainly @xmath18 , @xmath19 , @xmath20 , @xmath21 and @xmath22 production ) have shown that a @xmath23 signal ought to be detectable above background if @xmath24 gev , provided that an integrated luminosity of 25 fb@xmath3 can be accumulated at @xmath25 tev . in many particle physics models with weak scale supersymmetry ( susy ) and the low energy particle content of the minimal supersymmetric standard model ( mssm ) ,
the lightest higgs scalar @xmath26 has a mass that is typically @xmath27 gev@xcite .
furthermore , frequently the lightest susy higgs boson @xmath26 behaves much like the sm higgs boson .
thus , the higgs boson mass reach of the tevatron collider is particularly fascinating in that it may nearly saturate the parameter space of many interesting supersymmetric models .
the implication is that , if susy exists , then high luminosity upgrades of the fermilab tevatron @xmath28 collider will either discover the lightest susy higgs boson @xmath26 , or will exclude much of the parameter space of many susy models !
our goal in this paper is to translate the already calculated tevatron sm higgs boson mass reach into a reach in parameter space of three specific models involving weak scale supersymmetry .
these models are used for most phenomenological analyses of supersymmetry .
the first model assumes the generic structure of the minimal supersymmetric standard model ( mssm)@xcite with no assumptions about physics at scales beyond @xmath29 tev . in this case , we set all dimensional susy parameters ( such as soft susy breaking sfermion and gaugino masses , and @xmath30 ) to @xmath31 tev , except @xmath32-parameters which are set to zero , so that the relevant parameter space consists of @xmath33 where @xmath34 is the mass of the pseudoscalar higgs boson and @xmath35 is the ratio of higgs field vacuum expectation values .
several papers have presented susy higgs search projections for lep2 and lhc in this parameter space@xcite .
the second model we examine is the minimal supergravity ( msugra ) model@xcite with radiative electroweak symmetry breaking . in this model
, it is assumed that susy breaking takes place in a hidden sector , and susy breaking effects are communicated to the observable sector via gravitational interactions . in the minimal rendition of this model ,
all scalars have a common mass @xmath36 at the gut scale , while all gauginos have a common gut scale mass @xmath37 , and all trilinear scalar couplings unify to @xmath38 , where the universality of the various parameters occurs at some ultra - high scale @xmath39 . weak scale sparticle and higgs masses are obtained via renormalization group running of soft susy breaking masses and couplings from @xmath40 down to @xmath41 , where radiative electroweak symmetry breaking occurs .
ultimately , all sparticle masses and mixings are calculated in terms of the parameter set @xmath42 where @xmath30 is the superpotential higgs mass parameter , whose magnitude is fixed by the condition of radiative electroweak symmetry breaking .
the last model we consider is the simplest gauge mediated susy breaking model @xcite . in this model ,
susy breaking again takes place in a hidden sector , but susy breaking is communicated to the visible sector via messenger fields which also interact via usual gauge interactions .
sparticle masses are proportional to their gauge couplings , and their overall scale is set by the parameter @xmath43 , where @xmath44 is the susy breaking scale and @xmath45 the mass scale for the messenger particles .
the model is parameterized in terms of @xcite @xmath46 where @xmath47 is the number of complete @xmath48 messenger multiplets ( @xmath49 if @xmath50 is @xmath51 tev ) , and @xmath52 is the ratio of hidden sector to messenger sector vacuum expectation values of auxiliary fields .
these susy models are incorporated in the event generator isajet 7.37@xcite .
therein the susy higgs boson masses are calculated by minimizing the renormalization - group - improved one - loop effective potential . the minimization is performed at an optimized scale choice @xmath53 , which effectively includes the dominant two - loop contributions@xcite to @xmath54 .
we input susy parameter space values into isajet to calculate the various higgs boson and susy particle masses and mixing angles , as well as higgs boson branching fractions to sm and susy particles@xcite .
the associated and pair production cross sections of susy higgs bosons at @xmath55 colliders can be expressed as @xcite @xmath56 where @xmath57 } .\ ] ] they are written in terms of the sm result for associated higgs production given by @xmath58 with @xmath59 , @xmath60 , and @xmath61 .
the initial state qed corrections to the above processes are sizeable @xcite and are therefore included in our analysis .
this is accomplished by a resummation ( exponentiation ) of large logarithms due to soft photon emission@xcite .
the final form is a convolution of the leading order cross section with a resummed piece : @xmath62 where the resummed piece @xmath63 the soft / virtual and hard photon contributions represented by @xmath64 and @xmath65 may be be found in ref .
@xcite to @xmath66 .
they are polynomials in @xmath67 .
the term in the exponent @xmath68 where @xmath69 .
we obtain contours of lep2 excluded regions for @xmath70 or @xmath71 searches by calculating the sm cross section for @xmath72 production for @xmath73 gev and @xmath74 gev@xcite , and then requiring that the corresponding @xmath75 and @xmath71 cross sections be less than this value .
the lep2 ultimate reach contours are calculated by assuming lep2 to run at @xmath76 gev , with each experiment accumulating @xmath77 pb@xmath3 of integrated luminosity .
this has been projected @xcite to yield a discovery limit of about 107 gev for @xmath4 via the bjorken process ; we normalize our ultimate susy higgs @xmath75 and @xmath71 cross sections to the corresponding sm cross section . to obtain contours of the lep2 excluded region from @xmath78 searches , we normalize the @xmath79 production cross section to the exclusion contours presented in ref .
@xcite , which , for instance , exclude @xmath80 gev for @xmath81 at @xmath82 gev .
we also project the ultimate lep2 reach via @xmath83 production assuming that discovery will be possible if the corresponding cross section at @xmath76 gev is larger than 0.1 pb , while the charged higgs boson will be detectable if it is lighter than 95 gev .
detailed simulations of the @xmath84 signal and background have been performed in ref .
@xcite and @xcite , using the vecbos , pythia and herwig event generators . the event selection cuts were as follows@xcite : * @xmath85 gev , * @xmath86 gev , * number of jets ( @xmath87 gev , @xmath88)@xmath89 , * number of jets ( @xmath90 gev , @xmath88)@xmath91 , * @xmath92 , where @xmath93 and @xmath94 are variable end - points for the dijet mass bin ( for @xmath95 gev , @xmath96 gev and @xmath97 gev , for example ) .
in addition , * the two jets are tagged via the cdf secondary vertex ( svx ) @xmath98-tagging algorithm . in ref .
@xcite , the signal was increased by adopting the cdf standard soft lepton tag ( slt ) and the cdf loose svx @xmath98-tag to the second @xmath98-jet .
finally , some additional optimization of signal to background was obtained by imposing the cut * @xmath99 , in the @xmath100 center - of - mass system .
tables for signal and background were presented for @xmath101 , 80 , 100 and 120 gev .
we performed a simple fit to the signal presented in table iv of ref .
@xcite , and found that the number of signal events @xmath102 in 10 fb@xmath3 could be represented by @xmath103 while the number of background events could be represented by @xmath104 with @xmath105 the numerical results of table iv of ref . @xcite and
the above fitted curves are shown in fig .
[ fig0 ] .
the supersymmetric @xmath106 and @xmath107 production cross sections are given by @xmath108 where @xmath109 .
thus , @xmath110,\ ] ] with a similar equation for @xmath107 production .
the susy and sm branching fractions are extracted from isajet and are given in @xcite . as a criterion for observability at the tevatron , we use @xmath111 and @xmath112 .
we note that @xmath113 throughout most of parameter space .
our first results are presented in fig . [ fig1 ] for the mssm @xmath114 plane , where all susy mass parameters have been set to 1 tev except the @xmath32-parameters which have been set to zero , and @xmath115 gev .
the region labelled lep1 is excluded because the rate for @xmath116 and @xmath117 would violate limits placed on higgs masses from lep1 runs at the @xmath118-pole .
the region below the contour labelled `` @xmath119 current '' yields a @xmath75 or @xmath71 signal in excess of the @xmath72 rate for @xmath73 gev . in the region below the `` @xmath119 ultimate '' contour ,
the signal ought to be detectable by lep2 experiments sensitive to @xmath120 gev at @xmath76 gev .
the region to the left of the contour labelled `` @xmath121 ''
is excluded by current lep2 experiments searching for @xmath122 .
the ultimate lep2 @xmath79 reach is only a slight extension of the current @xmath121 contour and is not shown .
the reach via @xmath117 production lies well within the region probed by @xmath79 searches .
finally , there may exist a region of this parameter plane at high @xmath35 that is excluded by @xmath123 , @xmath124 and @xmath125 searches at the tevatron , where the higgs bosons decay to @xmath126 ; refined calculations of this are in progress@xcite .
the reach of the fermilab tevatron experiments for @xmath127 events with 10 fb@xmath3 of integrated luminosity corresponds to the region below the contour marked `` @xmath106 at 10 fb@xmath3 '' .
if the integrated luminosity is increased to 25 fb@xmath3 , then the region to the right of the contour labelled `` @xmath106 at 25 fb@xmath3 '' can be seen .
in addition , at low values of @xmath34 , the region interior to the contour labelled `` @xmath107 at 25 fb@xmath3 '' can be seen .
we find a window of non - observability ( labelled no ) at @xmath128 gev where _ none _ of the susy higgs bosons can be seen at lep2 or the fermilab tevatron collider . in this region ,
both @xmath26 and @xmath129 may contain a significant portion of the sm higgs scalar , so that @xmath130 is significantly smaller than unity and @xmath131 is suppressed , while @xmath129 is somewhat heavier than 120 gev @xcite .
this window is similar to but smaller than the one noted in refs . @xcite
that also occurs at the cern lhc .
most or all of the window may be filled in at the cern lhc by high luminosity running , and by searches for @xmath126 , @xmath132 or @xmath133 modes from higgs production . if the tevatron integrated luminosity is increased to 30 fb@xmath3 , the @xmath107 region increases towards lower @xmath35 , but most of the unobservable domain persists even at this higher integrated luminosity , and is not filled in even by our projection of the ultimate reach of lep2 .
the no window persists even for an integrated luminosity of 50 fb@xmath3 .
a similar plot for @xmath134 gev was constructed , but was indistinguishable from fig .
[ fig1 ] . finally , to test the robustness of our conclusions , we re - mapped our contours assuming an increase of 3 gev in all the higgs masses throughout parameter space .
the resulting plot was qualitatively similar to the one presented in fig .
[ fig1 ] , with a modest increase in the size of the window of non - obervability .
a similar analysis of susy higgs observability can be made within the context of the msugra model , except that in this case there exists the possibility of light superpartners which can affect both the higgs boson masses and branching fractions .
the results are presented in fig .
[ fig2 ] for @xmath135 and for _ a _ ) @xmath136 and _ b _ ) @xmath137 .
the th region is excluded either due to a lack of appropriate electroweak symmetry breaking , or because the lightest neutralino is not the lightest susy particle ( lsp ) .
the ex region is excluded by the negative results from lep2 searches for chargino pair production , which require @xmath138 gev@xcite .
the higgs boson search at lep2 presently excludes the region below the contours labelled `` @xmath119 current '' .
we see that for @xmath135 and @xmath136 , the current lep2 higgs bound rules out a significant portion of msugra parameter space up to @xmath139 gev , corresponding to @xmath140 gev ! however , for @xmath137 it gives only a slight additional excluded area beyond the chargino bound .
for both parameter planes for @xmath135 , experiments at lep2 will ultimately probe the entire parameter space shown , so that these will either discover a higgs boson or rule out msugra if @xmath35 happens to be small@xcite .
the fermilab tevatron collider has no reach for higgs bosons with 2 fb@xmath3 of integrated luminosity , but ought to be able to explore the entire plane of fig .
[ fig2]_a _ for @xmath136 with 10 fb@xmath3 . for @xmath137 , the tevatron reach is to @xmath141 gev for 10 fb@xmath3 , while the entire plane can be explored with 25 fb@xmath3 of integrated luminosity that might be accumulated at the tev33 upgrade of the tevatron .
the entire @xmath135 parameter space can be explored by lep2 and a 25 fb@xmath3 tevatron even if we allow @xmath142 gev everywhere .
an intermediate value of @xmath143 is shown for the msugra model in fig .
[ fig3]_a _ and [ fig3]_b _ for @xmath136 and @xmath137 , respectively . as @xmath35 increases ,
@xmath54 generally increases@xcite , and in fact the current lep2 sm higgs search imposes no additional constraints beyond those from other susy particle searches .
in fact , the lep2 ultimate reach lies only slightly beyond the current chargino mass exclusion contour , making higgs searches at lep2 difficult for larger values of @xmath35 . at @xmath143 ,
the fermilab tevatron experiments have _ no reach _ for higgs bosons with 10 fb@xmath3 of integrated luminosity . however ,
if an integrated luminosity of 25 fb@xmath3 can be accumulated , then the tevatron higgs search may probe @xmath37 values as high as @xmath144 gev ( 650 gev ) for @xmath136 ( @xmath137 ) via the @xmath106 channel .
( these values change to @xmath145 gev if @xmath142 gev ) .
an increase in integrated luminosity to 30 fb@xmath3 allows complete coverage of the parameter space shown . finally , we show in fig .
[ fig4]_a _ and _ b _ the corresponding msugra reach plots for @xmath146 . in this case
, there are some additional th excluded regions at large @xmath36 and @xmath147 due to lack of radiative ewsb , though the exact computation of this region is numerically somewhat sensitive . in this large @xmath35 case
, there will be _ no reach _ by lep2 higgs searches beyond the regions already excluded by current chargino searches .
however , fermilab tevatron experiments can explore via @xmath106 searches up to @xmath148 gev with 25 fb@xmath3 of integrated luminosity ( @xmath141 gev if @xmath142 gev ) , and they can probe once again the entire parameter space shown with 30 fb@xmath3 of integrated luminosity . from these results , we conclude that the fermilab tevatron collider with 30 fb@xmath3 of integrated luminosity ought to either discover a light higgs boson , or can exclude the msugra model .
we do not expect our results to change significantly with changes in the @xmath38 parameter .
results for the gmsb model are shown in fig .
[ fig5 ] , in the @xmath149 plane for _ a _ ) @xmath136 and _ b _ ) @xmath150 .
we take the messenger scale @xmath151 tev and first assume the messenger sector is comprised of just a single set @xmath152 of `` quark '' and `` lepton '' superfields in a @xmath153 representation of @xmath48 .
the th regions are excluded because they do not lead to proper radiative ewsb : in these regions , either @xmath154 or @xmath155 .
the ex regions are excluded by the aleph bound @xmath156 gev on the unstable neutralino which decays via @xmath157 from searches for @xmath158 @xcite , or by @xmath159 gev , from searches for staus in gmsb models@xcite .
the d0 and cdf experiments at the tevatron @xcite , from a non - observation of di - photon events with large @xmath160 have inferred a lower bound of about 150 gev on the chargino mass : this excludes essentially the same region as the aleph neutralino bound .
the current bound from lep2 due to higgs boson searches is shown as the region below the contour labelled `` lep2 current '' in the very low @xmath35 region of each plot .
the lep2 ultimate reach is for the area below the contour labelled `` @xmath119 ultimate '' , and can only probe parameter space with @xmath161 .
experiments at the fermilab tevatron will explore the region below the `` @xmath106 at 10 fb@xmath3 '' contour with @xmath162 fb@xmath3 of integrated luminosity . by increasing the integrated luminosity to @xmath163 fb@xmath3 ,
almost the entire parameter space should be explorable via higgs boson searches .
the exception is the region at @xmath164 tev and @xmath165 .
this additional region , however , can be covered with an integrated luminosity of @xmath166 fb@xmath3 .
finally , there is a narrow sliver of region at @xmath167 which is not seeable at the tevatron via @xmath106 searches ; however , most of this sliver is accessible via @xmath107 production at @xmath163 fb@xmath3 .
we have also again examined the effect of setting @xmath142 gev for the gmsb model .
the most obvious impact is on the size of the region that can not be explored with 25 fb@xmath3 of data : the boundary of this region shifts from @xmath168 tev to @xmath169 tev for both signs of @xmath30 , and furthermore , this region then extends down to @xmath170 .
finally , we have also explored the higgs boson reach for the case @xmath171 , the largest value of @xmath47 for which couplings remain perturbative up to a large scale when @xmath50 is 1000 tev .
the ex region may be somewhat underestimated @xcite since we obtain it using the same bound @xmath172 gev used in the @xmath152 case , in conjunction with the bound on @xmath173 . in this case
, @xmath54 is typically larger for a given @xmath174 and @xmath35 value than for the @xmath175 case .
these results are shown in fig .
the overall structure of the plot is similar to the results from fig.[fig5 ] , except that the @xmath106 reach for 25 fb@xmath3 is considerably smaller .
however , the whole parameter plane shown can once again be covered by 30 fb@xmath3 of integrated luminosity .
the lightest neutral scalar higgs boson is lighter than @xmath176 gev in a wide variety of interesting models commonly used for phenomenological studies of weak scale supersymmetry .
this observation , together with the fact that the most careful study to date yields a sm higgs boson reach of about 120 gev at proposed luminosity upgrades of the tevatron , motivated us to translate this reach to the corresponding reach in three classes of susy models : the mssm with @xmath31 tev , the msugra model , and finally , the gmsb model . while there are viable models where the @xmath177 gev limit on the lightest of the neutral higgs scalars can be significantly surpassed , these models generally contain additional fields ( most commonly higgs singlets ) which serve no specific purpose , and hence are less attractive theoretically .
assuming that the conclusion about the reach for the sm higgs boson holds up to closer scrutiny @xcite , we find that it should be possible to discover at least one of the higgs bosons over the entire parameter space of the msugra and gmsb models , and over most of the parameter space of the mssm , provided tevatron experiments can accumulate an integrated luminosity of 25 - 30 fb@xmath3 . the large integrated luminosity is crucial for an exploration of the parameter space : if the integrated luminosity is `` just '' 10 fb@xmath3 the signal is detectable over a much smaller range of parameters .
specifically , for the case of the mssm , experiments at tev33 should be able to probe the entire model parameter space given 25 fb@xmath3 of integrated luminosity , except for a window of unobservability around @xmath178 gev .
for the case of the msugra model , both lep2 and tev33 should be able to probe the entire parameter space for low @xmath35 ; for higher @xmath35 , the reach of lep2 will be limited , and tevatron experiments will need 30 fb@xmath3 of integrated luminosity to probe the entire model parameter space . for gmsb models , the ultimate reach of lep2 is limited to low values of @xmath35 , while the fermilab tevatron will need @xmath179 fb@xmath3 of integrated luminosity to probe the entire parameter space .
this conclusion is insensitive to the specific value of @xmath47 as long as the messenger scale @xmath50 is not too large .
in summary , if future luminosity upgrades of the tevatron are able to accumulate an integrated luminosity of 25 - 30 fb@xmath3 , experiments at these facilities should be able to probe most of the parameter space of these models via a search for @xmath180 events . while a discovery of a signal would be very exciting ,
it would not , by itself , allow us to conclude that we had discovered susy .
if instead tevatron experiments do accumulate the required data sample and conclusively find no signal for the higgs boson , it would exclude most of the parameter space of currently popular susy models , and cause us to rethink our ideas of how weak scale supersymmetry might be realized in nature .
we thank m. h. reno for collaboration in the early stages of this work .
this work was supported in part by the u. s. department of energy under contract numbers de - fg - fg03 - 94er40833 and de - fg02 - 97er41022 .
after submission of this paper for publication , another paper@xcite appeared in which the detectability of susy higgs bosons was examined within the mssm . in ref .
@xcite , it was pointed out that the region of the @xmath181 plane ( our fig .
2 ) where there is an observable higgs boson signal shows some dependence on the sign of @xmath182 .
this does not show up in our figure 2 since we have set @xmath183 .
we have checked and confirm ( for @xmath184 tev ) that the signal shows this sign dependence : the reach region is somewhat smaller for @xmath185 than for @xmath137 . for @xmath186 gev and large @xmath35 ( @xmath187 ) , the signal goes from being just above observability ( for @xmath137 ) to just below observability ( for @xmath185 ) with 25 @xmath188 . with 30 @xmath188 , the signal dependence on the sign of @xmath182 still makes some difference , but only for a limited range of @xmath34 around 120 - 150 gev . for a review ,
e.g. _ _ the higgs hunter s guide _
, j. f. gunion , h. e. haber , g. kane and s. dawson ( addison wesley , reading , ma , 1990 ) .
p. renton , talk given at 4th international conference on physics potential and development of @xmath189 colliders ( mumu 97 ) , san francisco , ca , 10 - 12 dec 1997 . y. pan , talk given at 4th international conference on physics potential and development of @xmath189 colliders ( mumu 97 ) , san francisco , ca , 10 - 12 dec 1997 .
a. stange , w. marciano and s. willenbrock , phys .
* d*49 , 1354 ( 1994 ) and phys . rev . *
d*50 , 4491 ( 1994 ) ; s. mrenna and g. l. kane , calt-68 - 1938 ( 1994 , unpublished ) ; a. belyaev , e. boos , and l. dudko , mod .
a10 * , 25 ( 1995 ) .
see contribution by u. heintz , s. kuhlmann and s. willenbrock to _ the tev-2000 group report _
, ed . by d. amedei and r. brock , ( fermilab - pub-96/082 , batavia , il , 1996 ) .
s. kim , s. kuhlmann and w. m. yao , in _ new directions for high energy physics , snowmass 96 _ , ed . by d. g. cassel , l. trindle gennari and r. h. siemann ( stanford linear accelerator center , stanford , ca , 1997 ) . g. kane , c. kolda and j. wells , phys . rev
. lett . * 70 * , 2686 ( 1993 ) ; j. r. espinosa and m. quiros , phys . lett . *
b279 * , 92 ( 1992 ) ; _ ibid .
_ , * b302 * , 51 ( 1993 ) ; j. r. espinosa and m. quiros , phys .
lett . * 81 * , 516 ( 1998 ) . for recent reviews ,
see x. tata , _ lectures presented at the ix jorge swieca summer school _ , campos do jordo , brazil , feb .
1997 , uh-511 - 872 - 97 , hep - ph/9706307 ; s. dawson , lectures at tasi 97 , hep - ph/9712464 ( 1997 ) . see also , h. e. haber , hep - ph/9709450 ( 1997 ) .
h. baer , m. bisset , c. kao and x. tata , phys .
* d*46 , 1067 ( 1992 ) and h. baer , m. bisset , d. dicus , c. kao and x. tata , phys . rev . * d*47 , 1062 ( 1993 ) . j. f. gunion and l. orr , phys . rev .
* d*46 , 2052 ( 1992 ) .
z. kunszt and f. zwirner , nucl . phys . * b*385 , 3 ( 1992 ) .
v. barger , m. berger , a. stange and r.j.n .
phillips , phys . rev . *
d45 * , 4128 ( 1992 ) ; v. barger , k. cheung , r. phillips and a. stange , phys . rev .
* d46 * , 4914 ( 1992 ) .
e. richter - was , d. froidevaux , f. gianotti , l. poggioli , d. cavalli and s. resconi , int . j. mod
a13 * , 1371 ( 1998 ) .
d. denegri ( cms collaboration ) , in _ beyond the desert 1997 _ , ed . by h. v. klapdor - kleingrothaus and h. pas , ( institute of physics publishing , 1998 ) .
a. chamseddine , r. arnowitt and p. nath , phys .
lett . * 49 * , 970 ( 1982 ) ; r. barbieri , s. ferrara and c. savoy , phys . lett .
* b*119 , 343 ( 1982 ) ; l.j .
hall , j. lykken and s. weinberg , phys .
rev . d*27 * , 2359 ( 1983 ) ; for a review , see h. p. nilles , phys . rep . *
110 * , 1 ( 1984 ) .
m. dine and a. nelson , phys .
* d48 * , 1277 ( 1993 ) ; m. dine , a. nelson , y. shirman , phys . rev .
* d51 * , 1362 ( 1995 ) ; m. dine , a. nelson , y. nir and y. shirman , phys . rev . *
d53 * , 2658 ( 1996 ) . see _
e.g. _ h. baer , p. mercadante , f. paige , x. tata and y. wang , hep - ph/9806290 ( 1998 ) , phys .
* b * ( in press ) .
f. paige , s. protopopescu , h. baer and x. tata , hep - ph/9804321 ( 1998 ) .
m. carena , j. espinosa , m. quiros and c. wagner , phys .
* b355 * , 209 ( 1995 ) ; m. carena , m. quiros , c. wagner , nucl . phys . * b461 * , 407 ( 1996 ) ; h. haber , r. hempfling and a. hoang , z. phys . * c75 * , 539 ( 1997 ) . m. bisset , ph .
d. thesis , hawaii preprint , uh-511 - 813 - 94 ( 1994 ) .
j. f. gunion _
_ , phys . rev .
* d*38 , 3444 ( 1988 ) .
a. djouadi , j. kalinowski , and p. zerwas , z. phys .
* c*57 , 569 ( 1993 ) ; a. djouadi , j. kalinowski , p. ohmann , and p. zerwas z. phys .
* c*74 , 93 ( 1997 ) .
e. gross , b. a. kniehl , and g. wolf , z. phys .
* c*63 , 417 ( 1994 ) .
f. berends , w. l. van neerven , and g. burgers , nucl .
phys . * b*297 , 429 ( 1988 ) and erratum * b*304 , 921 ( 1988 ) .
talk by n. konstantinidis at the first general meeting of the physics at run ii supersymmetry / higgs workshop , fnal , may , 1998 .
a. sopczak , hep - ph/9712283 ( 1997 ) .
m. drees , m. guchait and p. roy , phys .
* 80 * , 247 ( 1998 ) , erratum - ibid * 81 * 2394 ( 1998 ) ; c. balazs , j. l. daiz - cruz , h. j. he , t. tait , and c. p. yuan , anl - hep - pr-98 - 55 , hep - ph/9807349 ( 1998 ) . in the window of non - observability , @xmath26 and @xmath129 tend to be close in mass .
if they are close enough so that @xmath190 falls into a common bin , the signal from the two could be combined , and may then satisfy our observability criterion .
we have not attempted to do so here , but an analysis which uses variable bins may optimize the signal .
r. barate _
_ , ( aleph collaboration ) , eur . phys
. j. * c2 * , 417 ( 1998 ) .
h. baer , m. brhlik , r. munroe and x. tata , phys . rev .
* d*52 , 5031 ( 1995 ) .
h. baer , c. h. chen , m. drees , f. paige and x. tata , phys .
lett . * 79 * , 986 ( 1997 ) ; _ ibid .
_ ; * 80 * , 642 ( 1998)(e ) .
r. barate _
et al . _ , ( aleph collaboration ) , phys .
b420 * , 127 ( 1998 ) .
r. barate _
et al . _ , ( aleph collaboration ) , cern - ep-98 - 077 ( 1998 ) . b. abbott _ et al . _ , phys . rev . lett . * 80 * , 442 ( 1998 ) ; see also s. abachi _ et al .
lett . * 78 * , 2070 ( 1997 ) and f. abe _ et al .
_ , hep - ex/9801019 ( 1998 ) .
these issues are being addressed at the yearlong physics at run ii
supersymmetry / higgs workshop organized by fermi national laboratory , 1998 .
m. carena , s. mrenna , and c. wagner , anl - hep - pr-98 - 54 , hep - ph/9808312 ( 1998 ) .
fig . 1 : : number of events expected after cuts at the fermilab tevatron for 10 fb@xmath3 of integrated luminosity for the @xmath100 signal ( solid ) and sm background ( dashes ) , as a function of @xmath10 .
the crosses denote points from the calculations of kim _ et al .
the curves denote our fit to these points .
fig . 2 : : regions of the mssm @xmath191 plane accessible to current and future higgs boson searches at lep2 and the fermilab tevatron .
we take @xmath115 gev and set @xmath31 tev .
fig . 3 : : regions of msugra model parameter space accessible to current and future higgs boson searches at lep2 and the fermilab tevatron .
we take @xmath115 gev , @xmath192 and @xmath135 .
fig . 4 : : regions of msugra model parameter space accessible to current and future higgs boson searches at lep2 and the fermilab tevatron .
we take @xmath115 gev , @xmath192 and @xmath143 .
fig . 5 : : regions of msugra model parameter space accessible to current and future higgs boson searches at lep2 and the fermilab tevatron .
we take @xmath115 gev , @xmath192 and @xmath146 .
fig . 6 : : regions of gmsb model parameter space accessible to current and future higgs boson searches at lep2 and the fermilab tevatron
. we take @xmath115 gev , @xmath151 tev and @xmath152 .
fig . 7 : : regions of gmsb model parameter space accessible to current and future higgs boson searches at lep2 and the fermilab tevatron
. we take @xmath115 gev , @xmath151 tev and @xmath171 . | luminosity upgrades of the fermilab tevatron @xmath0 collider have been shown to allow experimental detection of a standard model ( sm ) higgs boson up to @xmath1 gev via @xmath2 events .
this limit nearly saturates the parameter space for many models of weak scale supersymmetry ( susy ) with a minimal particle content .
it is therefore interesting to examine the susy higgs reach of future tevatron experiments .
contours are presented of higgs boson reach for cern lep2 and tevatron luminosity upgrades for three models of weak scale susy : the minimal supersymmetric standard model ( mssm ) , the minimal supergravity model ( msugra ) and a simple gauge mediated susy breaking model ( gmsb ) . in each case
we find a substantial gain in reach at the tevatron with integrated luminosity increasing from 10 fb@xmath3 to 25 - 30 fb@xmath3 . with the larger integrated luminosity , a higgs search at the tevatron should be able to probe essentially the entire parameter space of these models .
while a discovery would be very exciting , a negative result would severely constrain our ideas about how weak scale supersymmetry is realized . * to appear in physical review d * |
in the last decade a wide variety of modified theories of gravity has been studied with the aim of solving or smoothing some puzzling features of conventional gravity and cosmology .
the reasons for considering such a modified gravitational schemes rely mainly on two damages that are intrinsic features of general relativity ( gr ) . on one hand ,
it is widely accepted by high energy physicists that the description of the gravitational field provided by general relativity must be doomed at scales of the order of the planck length , where the spacetime structure itself must be represented in terms of a quantum regime . on the other hand , and in the opposite extreme of the physical phenomena , gr also faces an intriguing dilemma in connection with the late cosmic speed up stage of the universe .
one of the newest extended theories of gravity is the so called @xmath0 gravity , which is a theory formulated in a spacetime possessing absolute parallelism @xcite .
the first investigations within this framework can be traced back to refs .
@xcite@xcite , where an ultraviolet deformation of einstein gravity was considered with the aim of smoothing singularities , namely the initial singularity of friedmann - robertson - walker ( frw ) cosmological models .
after that , the attention was focused in low energy deformations of gr which serve to tackle those aspects of the cosmological evolution connected with the late speed up stage of the universe in spatially flat models @xcite@xcite .
quite more recently , some fundamental aspects of @xmath0 theories , like the presence of extra degrees of freedom and the lack of lorentz invariance , had been addressed in refs .
@xcite@xcite . in spite of their success in smoothing the big - bang singularity @xcite ,
the ultraviolet deformations in the context of the @xmath0 theories fail to smooth singularities of gr vacuum solutions like black holes @xcite . for this reason
, we introduced in ref .
@xcite a more general theory in which the singularities of einstein theory receive a more appropriate treatment , even those that emerge from vacuum solutions of einstein equations . in this framework
the gravitational action is chosen , in close analogy with the born - infeld ( bi ) procedure used in electrodynamics , as the determinant of a particular combination of the torsion tensor that assures the correct low energy limit given by gr .
both the bi theory and the @xmath0 s mentioned above are based on actions built with just first derivatives of the dynamical field -the vierbein @xmath1- , so assuring second order motion equations .
most of the gravitational theories with absolute parallelism lack the invariance under local lorentz transformations of the vierbein .
this means that they should be regarded as theories describing the dynamics of the vierbein field rather than the metric field ( which is invariant under local lorentz transformations of the vierbein ) . as a consequence ,
the symmetry of the metric tensor is not enough to suggest the form of the vierbein , which could be elusive in many physically meaningful solutions . in this article
we will examine this issue by working out the proper vierbein field for closed and open frw cosmologies .
for this purpose , in section 2 we introduce the teleparallel equivalent of general relativity , that constitutes the geometrical pillar on which the @xmath0 and bi theories of gravity , also summarized there , are built . in section 3
we explain the lack of invariance of these theories under local lorentz transformation of the vierbein field in a cosmological context , discuss about the meaning of this feature , and work out the proper vierbein field for closed and open frw cosmologies . finally , in section 4 we display the conclusions .
in this section we will focus the attention on two gravitational schemes based on absolute parallelism . on one hand
, we will discuss a family of modified teleparallel theories constructed as a deformation of the einstein - hilbert lagrangian in weitzenbck spacetime the so called @xmath0 gravities. on the other hand , we will consider a quite different ultraviolet modification of gr with determinantal structure : born - infeld gravity .
we will start by introducing the geometrical concepts that serve as the building blocks of both schemes .
the basic idea is that gravity can be described by providing the spacetime with a torsion @xmath2 , @xmath3 , where @xmath4 is a vierbein ( a basis of the cotangent space ) in a @xmath5-dimensional manifold .
the vierbein @xmath4 is the co - frame of an associated basis @xmath6 in the tangent space .
if @xmath7 and @xmath8 are respectively the components of the 1-forms @xmath9 and the vectors @xmath10 in a given coordinate basis , then the relation between frame and co - frame is expressed as @xmath11 contracting with @xmath12 one also gets @xmath13 the components @xmath14 of the torsion tensor in the coordinate basis is related to the 2-forms @xmath15 through the equation @xmath16 this means that the spacetime is endowed with a connection @xmath17 ( since @xmath18 ) . the first term in eq .
( [ wei ] ) is the weitzenbck connection .
the metric is introduced as a subsidiary field given by @xmath19 where @xmath20 .
( [ metric ] ) can be inverted with the help of eq .
( [ coframe ] ) to obtain @xmath21 which means that the vierbein is orthonormal .
teleparallelism uses the weitzenbck spacetime , where the connection is chosen as @xmath22 as a consequence of the choice of the weitzenbck connection ( [ wei1 ] ) , the riemann tensor is identically null .
so the spacetime is flat : the gravitational degrees of freedom are completely encoded in the torsion @xmath2 . in terms of parallelism
, the choice of the weitzenbck connection has a simple meaning .
in fact , the covariant derivative of a vector yields @xmath23 in particular , eq .
( [ orto ] ) implies that @xmath24 ; so , the weitzenbck connection is metric compatible .
in general , eq . ( [ transport ] ) means that a given vector is parallel transported along a curve if its projections on the co - frame remain constant .
so , the vierbein parallelizes the spacetime .
of course , this nice criterion of parallelism would be destroyed if local lorentz transformations of the co - frame were allowed in the theory .
teleparallelism is a dynamical theory for the vierbein , which is built from the torsion . according to eq .
( [ metric ] ) , a set of dynamical equations for the vierbein also implies a dynamics for the metric .
this dynamics coincides with einstein s dynamics for the metric when the teleparallel lagrangian density is chosen as @xcite @xmath25=\frac{1}{16\pi g}\;e\;(t-2\lambda)\ ; , \label{lagrangiant}\ ] ] where @xmath26 , and @xmath27 stands for the absolute value of the determinant . in eq .
( [ lagrangiant ] ) we have defined the _
weitzenbck invariant _ as @xmath28 the tensor @xmath29 appearing in the last equation is defined according to @xmath30 in fact , the lagrangian ( [ lagrangiant ] ) just differs from the einstein - hilbert lagrangian with cosmological constant @xmath31 in a total derivative @xmath32=e\ t-2\;\partial _ { \nu } ( e\;t_{\sigma } ^{\ \
\sigma \nu } \,)\ ; , \label{divergence}\ ] ] where @xmath33 is the scalar curvature for the levi - civita connection . when gr dresses this costume , the gravitational degrees of freedom are gathered in the torsion instead of the levi - civita curvature .
it is a very curious and fortunate fact that both pictures enable to construct a gravitational action with the same physical content .
however , it is remarkable that the lagrangian ( [ lagrangiant ] ) involves just first derivatives of its dynamical field , the vierbein . in some sense
, the teleparallel lagrangian picks up the essential dynamical content of einstein theory without the annoying second order derivatives appearing in the last term of eq .
( [ divergence ] ) .
such lagrangian is a better starting point for considering modified gravity theories , since any deformation of its dependence on @xmath34 will always lead to second order dynamical equations . on the contrary ,
the so called @xmath35 theories lead to fourth order equations .
analogously to the @xmath35 scheme , a @xmath0 theory replaces the weitzenbck invariant @xmath34 in eq .
( [ lagrangiant ] ) with a general function @xmath0 .
so , the dynamics is described by the action @xmath36where @xmath37 is the matter lagrangian density .
undoubtedly , the whole family of actions gathered in ( [ action ] ) constitutes a vast territory worth to be explored , specially when one is aware that the dynamical equations arising by varying the action ( [ action ] ) with respect to the vierbein components @xmath38 are of second order .
this distinctive feature makes weitzenbck spacetime a privileged geometric structure to formulate modified theories of gravitation .
in fact , the dynamical equations for the vierbein tell how the matter distribution organizes the orientation of the basis @xmath9 at each point , in such a way that the field lines of @xmath39 realize the parallelization of the manifold .
after this vierbein field is obtained , one uses the assumption of orthonormality to get the metric ( [ metric ] ) . let us deal now with born - infeld gravity
this high energy modification of einstein gravity is based on the born - infeld procedure used first in the context of electrodynamics , and is governed by the @xmath40-dimensional action in weitzenbck spacetime given by @xcite @xmath41 , \label{acciondet}\]]where @xmath42 is quadratic in the weitzenbck torsion , and reads @xmath43 being @xmath44 and @xmath45 non - dimensional constants .
such a combination ensures the correct gr limit since both terms in @xmath46 have trace proportional to the weitzenbck invariant @xmath47 . in order to show this fact , we can factor out @xmath48 from expression ( [ acciondet ] ) and use the expansion of the determinant , @xmath49where @xmath50 and @xmath51 is the identity .
this last equation expresses the fact that in any number of dimensions we have @xmath52hence , at the lowest order we retrieve the low energy regime described by the einstein theory with cosmological constant @xmath53 $ ] .
due to the fact that general relativity is a theory for the metric , it is invariant under local lorentz transformations of the vierbein .
however , the equivalence between teleparallelism ( [ lagrangiant ] ) and gr dynamics , expressed in the eq .
( [ divergence ] ) , implies that @xmath34 changes by a boundary term under local lorentz transformations .
because of this reason , the teleparallel equivalent of gr does not provide the manifold with a parallelization but only with a metric . on the contrary , in a @xmath0 theory the boundary term " in @xmath34
will remain encapsulated inside the function @xmath54 .
this means that a @xmath0 theory is not invariant under local lorentz transformations of the vierbein .
so , a @xmath0 theory will determinate the vierbein field almost completely ( up to global lorentz transformations ) . in other words ,
a @xmath0 theory will describe more degrees of freedom than the teleparallel equivalent of gr .
this is an important issue in the search for solutions to the @xmath0 dynamical equations , since pairs of vierbein fields connected by local lorentz transformations ( i.e. , leading to the same metric tensor ) are _ inequivalent _ from the point of view of the theory .
clearly , the same considerations are also valid for born - infeld gravity because the determinantal action ( [ acciondet ] ) is constructed with quadratic combinations of the torsion tensor , which fails to be invariant under a local lorentz change of the tetrad . due to the lack of local lorentz invariance ,
the vierbein fields solving the equations of theories with absolute parallelism are harder to obtain than the metric field .
in fact , the considerations of symmetry about the form of the metric can be of little help to determine which among a set of locally lorentz related frames is the right one to solve the dynamical equations .
as an example , we will find here the adequate frames in order to deal with frw universes with non flat spatial sections .
for these geometries , the line element can be described in hyper - spherical coordinates as @xmath55 , \label{metesferica}\ ] ] where @xmath56 for the closed universe and @xmath57 for the open universe . here
, one is tempted to think that the vierbein that solves the dynamical equations could have the form @xmath58 however , this choice turn out to be incompatible with the field equations coming from the actions ( [ action ] ) and ( [ acciondet ] ) , because the vierbein ( [ naive ] ) does not correctly parallelize the spacetime .
the symptom that the choice ( [ naive ] ) will not work is the form acquired by the weitzenbck invariant in such case , which turn out to be @xmath59.\ ] ] this form of @xmath34 would be unable of giving a proper reduced lagrangian for the dynamics of the scale factor @xmath60 , as a consequence of its dependence on @xmath61 .
this @xmath62-dependent weitzenbck invariant is not consistent with the isotropy and homogeneity of the frw cosmological models .
we shall here sumarise the technique in order to obtain the parallelized frames . for details , we refer the reader to ref .
@xcite .
the procedure starts taking the _ naive _ frame of eq .
( [ naive ] ) and performing on it a suitable local lorentz transformation .
this transformation will locally twist the diagonal frame ( [ naive ] ) to convert it in a global well defined parallel field . for the closed universe ( @xmath63 ) ,
the local lorentz transformation consist in a local rotation of the spatial tetrad @xmath64 , with euler angles @xmath65 ; thus both frames are related via the euler matrix @xmath66 where @xmath67 as usual , the angular coordinates range in the intervals @xmath68 , @xmath69 and @xmath70 . the aspect of the so locally rotated frame is now @xmath71\notag \\
\mathring{e}^{3 } & = & - s ( \theta)s ( \phi)d\psi - s ( \psi ) [ ( s ( \psi ) c ( \phi ) + c ( \psi ) c ( \theta ) s ( \phi ) ) \ d\theta + ( c ( \psi ) c ( \phi ) - s ( \psi ) c ( \theta ) s ( \phi ) ) \ s ( \theta )
\ d\phi ] , \notag \\ \label{autop1}\end{aligned}\ ] ] where , of course , @xmath72 and @xmath73 .
a similar procedure is applicable to the open frw universe ; take the same rotation ( [ rotacion ] ) starting from the spatial part of the naive vierbein ( [ naive ] ) with @xmath57 , but now replace the euler angle @xmath61 by @xmath74 .
the transformed fields read ( here we put @xmath75 and @xmath76 ) @xmath77 \notag \\ \breve{e}^{3 } & = & s(\theta)s(\phi)d\psi + sh(\psi ) \ [ ( i\ sh(\psi)c ( \phi ) + ch(\psi)c(\theta)s(\phi))\ d\theta + ( ch(\psi)c(\phi ) -i\ sh ( \psi)c(\theta)s(\phi ) ) s(\theta ) \ d\phi ] .\notag
\\ \label{autop3}\end{aligned}\ ] ] then , as the frame field of 1-forms we take @xmath78 where the @xmath79 s allude to the dreibein @xmath80 and @xmath81 , @xmath82 , of eqs .
( [ autop1 ] ) and ( [ autop3 ] ) respectively .
the fact that the naive vierbein ( [ naive ] ) is connected with ( [ autop4 ] ) by a local rotation guarantees that the latter still describes the spatially curved frw metric given in eq .
( [ metesferica ] ) .
the weitzenbck invariant associated to the vierbein ( [ autop4 ] ) is@xmath83where the positive sign corresponds to the closed case and the negative to the open one , and @xmath84 is the hubble rate . ) can be summarized in the expression @xmath85 .
see ref .
@xcite for the flat case . ]
the absence of angular vestiges in ( [ invarcerrado ] ) foretell that the vierbein ( [ autop4 ] ) will be adequate to solve the dynamical equations .
let us prove this statement first for the @xmath0 schemes represented by the action ( [ action ] ) . for this purpose
, we must replace the vierbein ( [ autop4 ] ) in the field equations coming from the variation of the action ( [ action ] ) with respect to the components of the vierbein .
when varying with respect to the temporal component we obtain the modified version of friedmann equation , which turns out to be @xmath86 the non trivial equations for the spatial sector are equal to @xmath87 in the right hand side of the equations appear the energy density @xmath88 and the pressure @xmath89 of the fluid that fills the spacetime in a homogeneous and isotropic way , i.e. , the energy - momentum tensor reads @xmath90 in the comoving frame .
note that eq .
( [ friedcerrada ] ) is of first order in time derivatives of the scale factor , irrespective of the function @xmath54
. equations ( [ friedcerrada ] ) and ( [ espcerradas ] ) are two differential equations for just one unknown function @xmath60 ; so , they can not be independent .
the way to see that this is indeed the case , is to take the time derivative of ( [ friedcerrada ] ) and combine it with the conservation equation , @xmath91 to obtain eq .
( [ espcerradas ] ) .
conversely , if the system ( [ friedcerrada ] ) and ( [ espcerradas ] ) is consistent , then the conservation of energy in the matter sector is given automatically and eq .
( [ densidad de energia ] ) holds .
a further characterization of the scale factor dynamics depends on the particular choice of the function @xmath0 , but the general considerations about the correctness of the frames ( [ autop4 ] ) are valid for all @xmath54 s .
the frames ( [ autop4 ] ) are also valid for other theories with absolute parallelism , because they reflect the procedure of parallelization performed in the frw manifold .
we shall use them now in order to study the dynamics of the curved frw models in four dimensional born - infeld gravity . for simplicity
, we will work in the subspace of the parameter space @xmath92 defined by the condition @xmath93 ( see eq .
( [ tensorf ] ) ) .
the dynamics of the scale factor is encoded in the modified friedmann equation , which emerges from the replacement of the ansatz ( [ autop4 ] ) in the motion equation coming from varying the action ( [ acciondet ] ) with @xmath94 respect the temporal component of the tetrad @xmath95 , and it looks @xmath96 in the last equation , the signs @xmath97 refer again to the close and open cases respectively , and the bars mean absolute value .
again , the equation coming from the spatial sector is just the time derivative of ( [ 00k=1 ] ) combined with the conservation equation ( [ densidad de energia ] ) , so we can ignore it in the analysis .
an exact solution of the motion equation ( [ 00k=1 ] ) is beyond our present scope .
however , the main physical features can be derived if one thinks about eq .
( [ 00k=1 ] ) as an energy conservation equation of the type @xmath98 the potential @xmath99 can be found easily , and it can be written as @xmath100 , \label{potbik=1}\ ] ] where we have defined the two constants @xmath101 and @xmath102 .
additionally , the two constants @xmath103 and @xmath104 indicate present date values of the involved quantities , and @xmath105 is the barotropic index of the perfect fluid with state equation @xmath106 . in fig .
[ potenciales ] we can see the potentials ( [ potbik=1 ] ) for a radiation - dominated universe ( i.e. , @xmath107 ) , with @xmath108 , so @xmath109 .
the curves in black refers to the closed case and the ones in grey to the open case .
the behavior of gr is depicted in dashed lines for both cases .
while the gr potentials become divergent as @xmath110 , we see that the bi dynamics behave drastically different in that ( high energy ) limit .
as the scale factor tends to zero , the bi potentials go to zero with null derivative . in fig .
[ potenciales ] , we display the curves for three nominal values of the bi parameter , @xmath111 , @xmath112 and @xmath113 .
note that the energy level in eq .
( [ efpotk=1 ] ) is null , so the closed universe possesses a turning point in the low energy regime , where the potential is null . the non divergence of the bi potentials in the regime when @xmath110 denotes a very nice feature of the theory
actually , in that limit we have @xmath114 so the eq .
( [ efpotk=1 ] ) has , in the same limit , the solution @xmath115 then , in born - infeld cosmology we obtain that the early universe is characterized by an inflationary stage of geometrical nature .
the exponential behavior of the scale factor is driven by the bi parameter through the maximum hubble rate defined according to @xmath116 .
the existence of this early inflationary stage does not require the presence of the inflaton , and it seems to be an intrinsic feature of the bi regularization procedure .
in fact the spacetime so obtained is regular and there is not big bang singularity in it , because the scale factor tends to zero when the proper time @xmath117 .
in this article we have discussed certain conceptual features related with the lack of local lorentz invariance in gravitational theories with absolute parallelism and illustrated its consequences by means of two examples . in every gravitational theory possessing absolute parallelism , the spacetime structure is materialized in the coframe field @xmath118 which defines an orthonormal basis in the cotangent space @xmath119 of the manifold @xmath120 at each point @xmath121 . as an example we analyzed the dynamics of frw models in the so called @xmath0 theories , which are enjoying a growing popularity in our days .
when @xmath122 , i.e. , when one considers general relativity in weitzenbck spacetime , the basis @xmath118 at two different points of the manifold are completely uncorrelated , and it is not possible to define a global smooth field of bases unambiguously .
this is so because the theory is invariant under the local lorentz group acting on the coframes @xmath118 . in turn , when @xmath123 , local lorentz rotations and boosts are not symmetries of the theory anymore .
because of this lack of local lorentz symmetry , the theory picks up a preferential global reference frame constituted by the coframe field @xmath118 that solves the field equations . in such case
, the bases at two different points become strongly correlated in order to realize the parallelization of the manifold . nevertheless , the appearance of a preferred reference frame is a property coming from the symmetries of the spacetime , and it is not ruled by the specific form of the function @xmath0 .
for instance , when one is dealing with frw cosmological spacetimes , the field of frames that will lead to consistent field equations will be given by ( [ autop4 ] ) , whatever the function @xmath0 be .
additionally , the fields ( [ autop4 ] ) are also valid in more general theories with absolute parallelism which are not related with the @xmath0 schemes , for instance , born - infeld gravity ( [ acciondet ] ) . in this framework
, we have seen that the behavior of the early universe is characterized by an inflationary stage of infinite duration and geometrical character .
the exponential form of the scale factor is ruled by the bi parameter @xmath124 and it does not rely on the existence of the inflaton . finally , and due to the very nature of the born - infeld action , the spacetime results free of the big bang singularity .
f. fiorini wants to thank the organizers of the 8th alexander friedmann international seminar for their hospitality at rio de janeiro .
this research was supported by conicet and universidad de buenos aires . | the vierbein ( tetrad ) fields for closed and open friedmann - robertson - walker cosmologies are hard to work out in most of the theories featuring absolute parallelism .
the difficulty is traced in the fact that these theories are not invariant under local lorentz transformations of the vierbein .
we illustrate this issue in the framework of @xmath0 theories and born - infeld determinantal gravity .
in particular , we show that the early universe as described by the born - infeld scheme is singularity free and naturally inflationary as a consequence of the very nature of born - infeld gravitational action . |
a0535 + 26 is a be / x - ray binary system , discovered by _
ariel v _ during a giant outburst in 1975 @xcite .
the binary system consists of the neutron star pulsar a0535 + 26 and o9.7-b0 iiie optical companion hde245770 @xcite .
the neutron star moves in an eccentric orbit with _
= 0.47 , an orbital period @xmath4 = [email protected] days , and exhibiting a pulse period of @xmath0103.25 s ( * ? ? ?
* and references therein ) . the estimated distance to the system _ d _ = 2 kpc @xcite . extensive reviews of the system are given in @xcite and @xcite .
a0535@xmath626 belongs to a class of high mass systems , called be / x - ray binaries , known for having outbursts where matter accretes onto the magnetic poles via an accretion disk that is filled at periastron passage from a circumstellar decretion disk of the primary .
when this stellar disk retreats , the accretion disk shrinks and the source enters quiescence .
this allows the observer to study other source(s ) of accretion that may be present .
@xcite have studied high mass systems in quiescence where the neutron star spin period is relatively short ( @xmath7 5 s ) , and they conclude that x - ray emission is due to material falling onto the magnetospheric boundary or burning on the neutron star surface .
longer period high mass systems are less well studied .
+ the x - ray intensity of a0535 + 26 varies by almost three orders of magnitude with three basic intensity states : 1 ) quiescence , 2 ) normal , or type i , outbursts , generally associated with periastron passages , and 3 ) giant , or type ii , outbursts that may occur at any orbital phase .
the companion , hde245770 , has an equatorial circumstellar disk whose size has varied over time .
it is this material that drives the normal outbursts at periastron .
the giant outbursts may arise from large asynchronous mass outflows from the companion .
an indicator of the disk size is the h@xmath8 equivalent width , as well as that of he i and the overall visual magnitude @xcite .
the results of monitoring of these quantities are given in figure 1 of @xcite for the past 37 years .
a large reduction in the h@xmath8 strength occurred in 1998 ( mjd @xmath051000 ) in conjunction with the ceasing of outbursts for 7 years .
new outbursts began anew with the giant outburst of 2005 ( fig .
[ fig : bat ] ( left ) ) . beginning in 2009 ,
the h@xmath8 strength again declined with an apparent leveling - off at the time of the 2011 february outburst .
since then the h@xmath8 strength has continued to decline ( camero - arranz pvt .
comm . ) and no outbursts have been detected . since its discovery , nine giant outbursts have been detected : in 1977 december @xcite , 1980 october @xcite , 1983 june @xcite , 1989 march / april @xcite , 1994 february @xcite , 2005 may / june @xcite , 2009 december @xcite , 2010 october @xcite , and 2011 february @xcite .
normal outbursts are seen at periastron passages ( see figure 4 in * ? ? ?
* ) , but may not occur for years at a time .
after the 1998 normal outburst , the source went into the quiescent state until resuming in 2005 ( fig .
[ fig : bat ] ( left ) ) .
the last outburst was the giant one in 2011 february .
since then , a0525@xmath626 has exhibited no outbursts at all ( fig .
[ fig : bat ] ( right ) ) .
+ + + in quiescence , the system is expected to be in the centrifugally inhibited regime @xcite , preventing the continual accretion of matter onto the neutron star .
nevertheless , all historic observation campaigns during quiescence have found pulsations in at least part of their data , indicating that matter is still being accreted along the magnetic field lines .
observations of the quiescent state have been performed by _ exosat , rxte , bepposax _ , and _ xmm - newton_. the source was observed four times by _ exosat _ between 1985 and 1986 and pulsations were detected during two of the observations @xcite .
_ rxte _ performed two observations in quiescence between 1998 august and 1998 november @xcite , and pulsations were detected during the 1998 november observation , while in august only a weak indication ( @xmath9 ) was found for periodicity at the nominal pulse period .
_ bepposax _ observed the source during quiescence in 2000 september - october and 2001 march @xcite .
@xcite divided the _
data into 5 ever - decreasing count - rate bins , and significantly detected pulsations in all but the faintest .
we report here on three _ rxte _ observations made in 2011 after the giant outburst of 2011 february , and pulsations were only detected in the second of the new observations . combining these three observations with the previous observations and a new _ xmm - newton _ observation made a year after the last outburst ( doroshenko pvt . comm . )
, we present a detailed study of a0535@xmath626 in quiescence , where accretion onto the magnetic poles continues , possibly from matter accumulated at the magnetospheric radius or a stellar wind .
section 2 describes the data reduction and analysis , while section 3 discusses the results and gives the conclusions of this investigation .
rrrrr 2011 may 15 & 96421 - 01 - 02 - 00 & [email protected] & 2,432 & 0.749 + 2011 july 28 & 96421 - 01 - 03 - 00 & [email protected] & 2,944 & 0.415 + 2011 august 22 & 96421 - 01 - 04 - 00 & [email protected] & 2,672 & 0.640 _ rxte _ observed a0535@xmath626 in quiescence on three occasions in 2011 in order to study its spectral and temporal behavior ( table [ tab : obs ] ) .
the data reduction was carried out using heasoft version 6.11 software , and xspec version 12.7.0 @xcite .
the data were extracted observing the standard ucsd / iaat / ecap filtering criteria of pointing within 0.01 degrees of the target , elevation above the earth s horizon greater than 10@xmath10 , time since the center of the last south atlantic anomaly passage of 30 minutes , and an electron rate of less than 0.1 .
rccccc 2011 may 15 & _ rxte _ & 55696 & 77 & 0.749 & no + 2011 july 28 & _ rxte _ & 55770 & 155 & 0.415 & yes + 2011 aug 22 & _ rxte _ & 55795 & 180 & 0.640 & no + 2012 feb 28 & _ xmm _ & 55985 & 370 & 0.350 & yes + 1985 sept 23 & _ exosat _ & 46331 & 600 & 0.545 & yes + 1986 feb 10 & _ exosat _ & 46471 & 740 & 0.284 & no + 1986 feb 18 & _ exosat _ & 46479 & 748 & 0.212 & yes + 1986 mar 9 & _ exosat _ & 46498 & 767 & 0.041 & no + 1998 aug 21 & _ rxte _ & 51046 & 1455 & 0.105 & yes + 1998 nov 12 & _ rxte _ & 51129 & 1538 & 0.358 & yes + 2000 sept 4 - 5 & _ bepposax _ & 51791 & 2200 & 0.400 & no + 2000 oct 5 - 6 & _ bepposax _ & 51822 & 2231 & 0.121 & no + 2001 mar 6 - 7 & _ bepposax _ & 51974 & 2383 & 0.753 & yes we have also re - analyzed the 1998 _ rxte _ observations of a0535@xmath626 reported by @xcite and the 19851986 _ exosat _ observations of @xcite using updated response and background files10@xmath11 ergs @xmath2 s@xmath3 . ] .
the re - analysis of the _ exosat _ spectra for a0535@xmath626 was performed by downloading the spectral and response files for the _ exosat _ me detector from the heasarc for the 4 observations .
the _ xmm - newton _ results will be the subject of a separate publication , and only the 210 kev flux is presented here .
the _ bepposax _
results are from the publications of @xcite and @xcite . in addition
, we have used the abundances of @xcite and cross sections of @xcite .
table [ tab : quiescence ] summarizes the 13 observations made in quiescence since 1984 in order of their time since the last outburst preceding them . to perform a period search , proportional counter unit 2 ( pcu2 ; *
* ) lightcurves from the observations in 2011 july ( obsid 96421 - 01 - 03 - 00 ) and 2011 august ( obsid 96421 - 01 - 04 - 00 ) were extracted in data mode b_250ms_128 m and the lightcurve from the observation in 2011 may ( obsid 96421 - 01 - 02 - 00 ) in goodxenon mode . due to the low luminosity of the source ,
no significant flux was detected above @xmath020 kev , and therefore only data from pulse height channels 5 to 49 , corresponding to energies between @xmath03 and @xmath020kev , were used in the timing analysis ( see table [ tab : obs ] for the comparative counting rates ) .
for each observation the data were collected into consecutive 2 s time bins . with uniformly sampled and uninterrupted data , the lomb - scargle algorithm was applied over the frequency range of 0 to 0.020 hz ( periods greater than 50 s ) in 160 frequency steps ( approximately 1 s steps in period ) .
the resulting lomb - scargle significance is plotted versus frequency for the three data sets in fig [ fig : epochfold ] ( left column ) .
the probability of a frequency component occurring in pure gaussian noise is pr(occ ) @xmath12 ( 0.01)@xmath13 , where @xmath14 is the significance .
thus a significance of 66.7 , for the middle observation at 0.0097 hz , implies a very significant detection .
the interpretation of the actual significance is limited by systematic effects , such as aliasing and non - gaussian variations .
the peak of the lomb - scargle plot for the middle observation was at 102.36 s. note that the lomb - scargle analysis assumes sinusoidal variations and gaussian noise , while the pulse profile of a0535@xmath626 is somewhat sinusoidal ( fig .
[ fig : epochfold ] ( lower - right ) ) and the statistics are poisson .
epoch folding for periods near the lomb - scargle peak value was used to further define the pulse period .
the first and last observations produce very little lomb - scargle significance near 0.0097 hz ( 103 s ) , and we conclude that pulsing was not detected for those observations .
epoch folding for those data was performed to validate the lomb - scargle analysis .
an iterative epoch folding method was utilized whereby pulse - to - pulse changes in luminosity were taken into account by averaging the count rate over the pulse period and subtracting the resulting pulse averaged lightcurve from the original one .
this increases the strength of a signal in the epoch folding @xmath15-distribution @xcite .
periodic signals were searched between 94 and 112 s , with 200 test periods in between and 8 phase bins .
the resulting @xmath15 distribution for the 2011 july observation is shown in figure [ fig : epochfold ] ( top - right ) , and the corresponding 16 bin folded light curve for a peak period of 103.48(28)s is given on the middle - right . if one compares the second observation profile to those seen in outburst ( e.g. * ? ? ?
* ) , this 320 kev profile resembles the higher energy ( @xmath1615 kev ) ones in outburst with the broader minimum , as compared to lower energy ones with a narrow feature defining flux minimum .
it also resembles many of the previous quiescent 210 kev profiles . the uncertainty in the epoch fold period for the 2011 july data is at the @xmath17-confidence limit and was determined by monte carlo simulations .
the 2011 july pulse period is consistent with the detected pulse period of @xmath0103.25s @xcite during the last outburst in february 2011 , and considering that the pulse period changes relatively little during non - outburst times .
it is surprising that the pulsations were only detected in the 2011 july observation , which has an intermediate flux of the three observations .
even if the lack of a detection of pulsations in the last of the three observations could be attributed to insufficient statistics , the non - detection in the brightest observation remains a puzzle . with exposure roughly equivalent to the 2011 july observation and higher counting rate
, one would expect it to have been more sensitive to pulsations .
the energy spectra of all 3 observations are quite similar ( fig .
[ fig : spectral_power ] ) , so the presence or lack of pulsations does not appear attributable to some kind of state change .
the spectra for each _ rxte _ observation were generated from the pcu2 standard2 data files , which represent the full ( 290 kev ) energy range in 129 variable width spectral bins .
these histograms are independent of the individualized data modes for a given observation .
the pcu2 faint background estimate , provided by the pca team and based upon detector rates acquired while observing blank fields during the mission , was used in fitting of background subtracted data .
the normalization of the background histogram depends upon the dead time estimate , which is also based upon average detector counting rates .
the xspec recorn function was initially used to correct for the small difference between the estimated and actual pcu2 background normalization , which can not be measured directly due to the proportional counter array detectors being non - imaging . consequently the background must be modeled .
comparison of results with and without the recorn function in the model showed that these data were insensitive to this background optimization , and thus results are given for fits without the recorn function included in the modeling .
the two spectral models fitted to the three observations in the 320 kev range were an absorbed power law ( cflux*(tbabs*pegpwrlw ) ) and absorbed thermal bremsstrahlung ( cflux*(tbabs*bremss ) ) , both modified by a maximum in galactic line of sight absorption ( fixed at 4.5@xmath18 @xmath2 , * ? ? ?
cflux was the basis for the measurement of 210 kev flux plus 90% errors .
the simple power law was used since the statistics did not warrant the use of a more complex model with a high energy cutoff , as utilized for fitting outburst spectra .
both models gave acceptable and similar fits ( fig .
[ fig : spectral_power ] , tables [ tab : power ] and [ tab : bremss ] ) .
the errors reported in the two tables are 90% uncertainties .
the power law and thermal bremsstrahlung best - fit values for absorption at the source differ by about a factor of two , with the power law one averaging about 9@xmath1910@xmath20 @xmath2 and bremsstrahlung one averaging a bit less than 4@xmath1910@xmath20 @xmath2 .
this is due to the curvature of the bremsstrahlung spectrum with respect to that of a power law at low energies . above @xmath010 kev ,
the bremsstrahlung spectrum rolls over with respect to the power law , but poor statistics at these energies preclude the determination of which model provides the better representation of the incident spectrum .
the 1998 _ rxte _ spectral data for a0535@xmath626 were extracted with the same acceptance criteria as the 2011 _ rxte _ data , and re - analyzed with updated response and background files . for the present _
rxte _ re - analysis , only data from pcu2 were used for consistency with the later _ rxte _ results .
only 2@xmath21 upper limits to the column density were found , and the best - fit values of the power law indices were less than but consistent with those found by @xcite . when applying the bremsstrahlung model ,
the column density upper limits were again lower than those for the power law fits , and the electron temperatures were somewhat higher , but again consistent with the values found by @xcite .
tables [ tab : power ] and [ tab : bremss ] give the best - fit values from the re - analysis . while @xcite reported average _ exosat
_ values for the best - fit parameters for both power law and thermal bremsstrahlung models , we report the individual observation values here ( tables [ tab : power ] and [ tab : bremss ] ) .
we found an average power law index of 2.27 and an average electron temperature of 7.5 kev .
no simultaneous fitting between different instruments was performed due to the variations in column density , flux , and power law index .
consequently the relative normalizations between the instruments are assumed to be unity .
+ + + rcccc 2011 may 15 & 9.1@xmath22 & 2.40@xmath23 & 4.01@xmath24 & 0.67 + 2011 july 28 & 9.1@xmath25 & 2.59@xmath26 & 2.37@xmath27 & 0.75 + 2012 aug 22 & 5.0@xmath28 & 2.44@xmath29 & 1.55@xmath30 & 0.42 + 1998 aug 21 & @xmath3111.5 & 2.30@xmath32 & 0.79@xmath33 & 0.50 + 1998 nov 12 & @xmath3113.7 & 2.40@xmath34 & 0.96@xmath35 & 0.46 + 1985 sept 23 & 2.4@xmath36 & 1.83@xmath37 & 2.46@xmath38 & 1.33 + 1986 feb 10 & 4.0@xmath39 & 2.42@xmath40 & 1.18@xmath41 & 0.96 + 1986 feb 18 & 2.5@xmath42 & 1.98@xmath43 & 1.44@xmath41 & 0.81 + 1986 mar 9 & 6.7@xmath44 & 2.85@xmath45 & 1.56@xmath46 & 0.89 rcrrr 2011 may 15 & 3.7@xmath47 & 8.22@xmath48 & 4.22@xmath49 & 0.60 + 2011 july 28 & 3.7@xmath50 & 6.59@xmath51 & 2.48@xmath52 & 0.65 + 2012 aug 22 & @xmath315.4 & 7.44@xmath53 & 1.62@xmath54 & 0.41 + 1998 aug 21 & @xmath317.0 & 9.06@xmath55 & 0.80@xmath56 & 0.48 + 1998 nov 12 & @xmath318.1 & 7.17@xmath57 & 1.00@xmath58 & 0.42 + 1985 sept 23 & 1.6@xmath59 & 11.14@xmath60 & 2.42@xmath61 & 1.25 + 1986 feb 10 & 2.4@xmath62 & 4.66@xmath63 & 1.11@xmath64 & 0.93 + 1986 feb 18 & 1.6@xmath65 & 8.42@xmath66 & 1.40@xmath67 & 0.81 + 1986 mar 9 & 4.1@xmath68 & 3.84@xmath69 & 1.51@xmath70 & 0.90 the present 2011 _ rxte _ observations that occurred 3 , 5 , and 6 months after the last outburst ( 2011 feb 23 ) show a clear steeply dropping flux ( fig .
[ fig : time_since ] ) . however , when the _ xmm - newton _ value from 12 months after the outburst is added , the interpretation is less clear , and a flux level of @xmath02@xmath71 ergs @xmath2 s@xmath3 612 months after the outburst is indicated .
the addition of the earlier _ rxte , bepposax _ and _ exosat _ observations performed 26 years into quiescence sample the declining flux from 0.25 to 6 years after an outburst ( fig .
[ fig : time_since ] ( left ) ) , albeit not the same outbursts and not all either normal or giant outbursts .
the dashed line in fig . [ fig : time_since ] ( left ) is a least squares fit to all but the first _
rxte _ point , and reflects a pearson correlation coefficient of @xmath720.785 ( a null hypothesis probability of 0.012 ) . if taken at face value , the quiescent flux values indicate a slowly declining quiescent level with a slope of 3.8@xmath73 ergs @xmath2 s@xmath3 per day after these 3 outbursts .
the 20112012 _ rxte _ data show a slightly raised flux in the first observation , as do the _ exosat _ data
. this may be indicative of small levels of flaring during quiescence , which would be yet another indication of continuing , variable accretion .
it may also be an effect of orbital phase in an isotropic stellar wind .
as the observed intensity decreased linearly as a function of time since outburst , the column density decreased exponentially ( fig .
[ fig : time_since ] ( right ) ) .
the pearson correlation coefficient for the log of @xmath74 versus time since outburst is @xmath720.897 ( a null hypothesis probability of 0.001 ) .
this naturally results in column density increasing with the 210 kev power law flux ( fig .
[ fig : index_flux ] ( right ) ) , and in this case the pearson correlation coefficient is 0.772 ( a null hypothesis probability of 0.015 ) .
finally , the power law index does not appear to correlate with the power law flux ( fig .
[ fig : index_flux ] ( left ) ) .
a0535@xmath626 continues to show pulsations , as seen in most but not all of the observations .
the fact that the first and brightest of the 3 new _ rxte _ observations in quiescence did not exhibit detectable pulsations , while the second , @xmath050% weaker observation did with a similar live time , is strong evidence that the pulsations are not always present in quiescence .
this would indicate in general that accretion onto the polar caps was still occurring , but that the accretion may not always be channeled to the magnetic poles . in quiescence the neutron star in a0535@xmath626
is expected to be centrifugally inhibited @xcite , and plasma entry into the magnetosphere may be mediated by various instabilities affecting the extent of the emission region @xcite .
@xcite have modeled the outbursts from be / x - ray binaries using a viscous decretion model , where the presence of outbursts depends upon the neutron star orbit with respect to various resonances of the gas in the system . if the neutron star s orbit is outside a given resonance , matter from that resonance will not flow onto the neutron star , whereas if it is within a given resonance , the matter will flow .
@xcite state for a0535@xmath626 that the disk is truncated at the 4:1 resonance when outbursts are detected , and that a slight decrease in viscosity or disk temperature would lower the disk radius to the 5:1 resonance and thus have the system go into dormancy .
the observation that the quiescent luminosity is declining in the composite light curve ( fig .
[ fig : time_since ] ) could imply that an accumulated mass at the corotation radius @xcite is being depleted over time .
then quiescence in a0535@xmath626 would be characterized by reduced accretion from the magnetospheric boundary , while the bright outbursts would be characterized by high levels of accretion from the companion s circumstellar disk .
another possibility is that the companion may have a stellar wind in addition to that from the circumstellar disk .
this would mean that the system could transition from disk accretion to stellar wind accretion . if the disk were to recede , as the h@xmath8 observations imply in 1998 and 20112012 ( figure 1 of @xcite and v. mcbride , pvt .
, one might expect the system to transition to spherical wind accretion dominating over that of the disk . if there is a sufficient stellar wind to provide wind accretion as in gx 301 - 2 ( e.g. * ? ? ?
* ) , then the generation of x - rays and pulsing seen in quiescence might be explained . the very small number of quiescent data points after each outburst as a function of orbital phase precludes any test of variations with orbital phase , which might have been evidence of stellar wind accretion .
@xcite adopt a wind mass loss rate for hde245770 of 10@xmath75 m@xmath76 yr@xmath3 from measurements of uv resonance lines and infrared fluxes .
spherical bondi accretion for a 1.4 m@xmath76 neutron star at a distance of 267 light - seconds ( @xmath77 for the a0535@xmath626 system ) and a 2000 km s@xmath3 wind velocity at infinity ( e.g. , table 1 in * ? ? ?
* ) would predict the mass accretion rate onto the neutron star to be @xmath78 of the wind mass loss rate . if the bolometric quiescent x - ray luminosity is about twice the 2 - 10 kev luminosity , or 2.5@xmath79 ergs s@xmath3 for a distance of 2 kpc
, then one would predict a wind mass loss rate of 10@xmath80 m@xmath76 yr@xmath3 , if the conversion efficiency of gravitational to x - ray energy is 0.1 .
this estimate is 10% of the @xcite value , which may be reasonable if 90% of the wind mass loss is due to the circumstellar disk . @xcite
predict a factor of 10@xmath81 for the difference in x - ray luminosities between polar and equatorial mass flows for hde 245770 ( 10@xmath82 ergs s@xmath3 vs 10@xmath83 ergs s@xmath3 ) .
if the orbit of the neutron star is inclined to the equator of the companion , and if the mass flow changes with angle from the pole , the neutron star may sample a mass flow that produces fluxes in the 10@xmath84 ergs s@xmath3 range .
should we see orbital variations of the flux due to the eccentric orbit in the _ rxte _ data that was taken within one orbital period ?
a 2000 km s@xmath3 wind would predict at most a factor of @xmath03 variation in flux between the 3 _ rxte _ phases .
the _ rxte _
fluxes are in agreement with this limited variation .
the spectra of a0535@xmath626 indicate a decreasing amount of absorption at the source as time since an outburst increases ( fig .
[ fig : time_since ] ( right ) ) . when comparing the various values of @xmath74 versus orbital phase ( given in table [ tab : quiescence ] )
it is clear that this material is present in the line of sight independent of orbital phase .
thus , it must be either all pervasive in the system ( the wind ? ) or very close to the neutron star ( a declining accumulation of matter at the corotation radius ? ) .
for the former , variations in @xmath74 might imply a variable stellar wind density , which in turn would result in a correlated variable x - ray flux ( fig .
[ fig : index_flux ] ( right ) ) . for the latter ,
the disk might thin as the material is drained from it .
this could be further evidence for a continual draining of the material trapped near the corotation radius @xcite .
the power law index is independent of flux over the flux range of 14 @xmath85 ergs @xmath2 s@xmath3 ( fig .
[ fig : index_flux ] ) .
in addition , the power law indices measured for all of the quiescent observations are significantly steeper ( @xmath0 2.5 versus @xmath31 1 ) than those seen during outbursts ( e.g. * ? ? ?
a steeper power law would imply less compton up - scattering of lower energy photons due to either a thinning of the hot electon scattering plasma , or a cooling of the electron distribution .
this may be further indication that the accretion mode is different between outburst and quiescence . the pulse profile exhibited by the 2011 july 28 _ rxte _
data is single peaked and somewhat sinusoidal , as has been seen in the past for observations made at quiescent luminosities , as noted above .
_ cgro_/batse observations hint at a change in the pulse profile to the single peak variety that occurs below 10@xmath86 ergs @xmath2 s@xmath3 @xcite .
this change from the double peaked structure of the outburst pulse profiles possibly indicates a change in the accretion regime . for low luminosities
the beam pattern is expected to be more pencil beam - like , and more like a fan beam at higher luminosities ( * ? ? ? * and references therein ) .
_ rxte _ observations of a0535@xmath626 months after a giant outburst were combined with three other sets of multiple observations of a0535@xmath626 from other x - ray observatories after other outbursts to reveal that they all had a similar ( within a factor of 2 ) , or slowly decreasing flux level .
at least half of the observations detected pulsations , indicating that accretion onto the magnetic poles was still occurring , but perhaps intermittently .
optical observations of h@xmath8 emission at the times of the measurements indicated that the 1998 and 2011 data were taken during times of a highly diminished be star disk , allowing for the possibility of sources of accreted material other than that responsible for the outbursts .
when combining all 13 observations as a function of time since the last outburst ( albeit not the same outbursts ) , the quiescent flux appears to drop by a factor of 2 over 6.5 years , or @xmath87 @xmath88 ergs @xmath2 s@xmath3 with small ( @xmath89 ) flares or possible orbital variations present .
based upon very limited statistics , the line - of - sight column density decreases by a factor of a few over the combined 6.5 years of quiescence , while the power law remained relatively constant .
the quiescent flux and subsequent , correlated line of sight column depth could be due to a draining of material built - up at the corotation radius or due to spherical accretion from an isotropic , but diminishing , component of the stellar wind .
further characterization of the quiescent mode of a0535@xmath626 will require more observations over a range of times since the last outburst in order to see if the slow downward trend in the flux continues , and to see if one could detect the orbital variations in flux expected for wind accretion in the eccentric orbit .
we acknowledge the assistance and mentoring of eric michelsen concerning lomb - scargle analysis and its interpretation , along with use of his program.thanks go out to ron remillard for supplying the weekly a0535@xmath626 asm data for the duration of the _ rxte _ mission .
these results were provided by the asm / rxte teams at mit and at the rxte sof and gof at nasa s gsfc .
we thank evan smith for his excellent efforts in scheduling of the rxte observations .
we thank issi for hosting and financing international group meetings on modeling and observations of cyclotron lines in high mass x - ray binaries .
we acknowledge the gsfc heasarc for providing extracted spectra and response files for the _ exosat _ observations that allowed for the re - analysis of the a0535@xmath626 observations .
this work was supported by the bundesministerium fr wirtschaft und technologie through dlr grant 50 or 1113 .
we would like to thank j. e. davis for the slxfig module which was used to create some of the plots through out this paper .
aca thanks the support from grants aya2012 - 39303 , sgr2009- 811 , and ilink2011 - 0303 . | we have analyzed 3 observations of the high mass x - ray binary a0535 + 26 performed by the _ rossi x - ray timing explorer _ ( _ rxte _ ) 3 , 5 , and 6 months after the last outburst in 2011 february .
we detect pulsations only in the second observation .
the 320 kev spectra can be fit equally well with either an absorbed power law or absorbed thermal bremsstrahlung model .
re - analysis of 2 earlier _ rxte _ observations made 4 years after the 1994 outburst , original _
bepposax _
observations 2 years later , re - analysis of 4 _ exosat _ observations made 2 years after the last 1984 outburst , and a recent _ xmm - newton _ observation in 2012 reveal a stacked , quiescent flux level decreasing from @xmath02 to @xmath1 ergs @xmath2 s@xmath3 over 6.5 years after outburst .
detection of pulsations during half of the quiescent observations would imply that accretion onto the magnetic poles of the neutron star continues despite the fact that the circumstellar disk may no longer be present .
the accretion could come from material built - up at the corotation radius or from an isotropic stellar wind . |
experimental realization of trapped one - dimensional ( 1d ) cold atom systems @xcite are triggering more and more theoretical efforts to study the 1d many - body physics beyond the mean - field theory .
for the ultracold quantum gases tightly confined in waveguides , the dynamics are effectively described by a 1d model due to the radial degrees of freedom are frozen @xcite .
further , the ability of tuning the effective 1d interactions by feshbach resonance allows experimental access to the very strongly interacting regime where correlation effects are greatly enhanced @xcite . in the limit of the tonks - girardeau ( tg ) @xcite gas with
effective coupling constant @xmath2 , the many - body problem of a tg gas can be mapped to that of a free fermi gas by the bose - fermi mapping , which has been verified by two experimental groups @xcite .
this bose - fermi duality was generalized to show the equivalence between a 1d fermionic system and a bosonic one with the reversed role of strong and weak couplings @xcite .
recently , the exact ground state of the fermionic tg ( ftg ) gas , defined as a 1d spin - polarized fermionic gas with infinitely strong attractive @xmath0-wave interactions , has been determined by inversely fermi - bose mapping to the ideal bose gas @xcite . a key experimental challenge is to obtain superfluidity with pairs in nonzero orbital angular momentum states by using @xmath0-wave , or maybe even @xmath3-wave feshbach resonances . in general , the @xmath0-wave interaction is very weak comparing with the @xmath4 -wave interaction .
however , for a spin - polarized fermionic gas , the @xmath4-wave scattering is forbidden due to the pauli exclusion principle and thus the @xmath0 -wave interaction is dominant .
furthermore , the @xmath0-wave interaction can be greatly enhanced by the feshbach resonances @xcite and using a @xmath0-wave feshbach resonance between @xmath5 atoms jin s group at jila have successfully produced and detected molecules with lifetimes on the order of milliseconds on both the bec and the bcs side of the resonance @xcite . for a 1d gas , the additional confinement induced resonance permits one to tune the 1d effective interaction via a 3d feshbach resonance @xcite . in this paper , we report on a detailed study of the 1d fermi gases in the infinitely deep square potential well .
we will show that the model of fermionic gases with attractive @xmath0-wave interactions in such a one - dimensional hard - wall trap is exactly solvable by the bethe - ansatz method .
the experimental efforts in trapping ultracold gases near micro - fabricated surfaces , the so - called `` atom chips '' @xcite , and various innovative features in designing the optical box trap @xcite , are specifically aimed at studying the surprisingly rich variety of physical regimes predicted for the 1d bose gas and have stimulated many theoretical studies on the physics in a box trap @xcite .
different from the harmonic trap , the interacting model in a hard - wall trap is integrable and thus could provide us some exact pictures for understanding the trapped many - body systems .
so far , there has been a growing interest in the exactly solved models in the hard - wall trap @xcite , but most of them focus on the bose gas and the fermi gas with odd - wave interactions is not addressed . while the theoretical understanding of the correlation effect of bosonic system has been investigated extensively @xcite , the fermionic system is not well understood except in the so - called ftg limit @xcite .
we consider an @xmath6-particle system with finite , attractive @xmath0-wave interaction in a one - dimensional box of length @xmath7 , which obviously fills the gap between the ftg limit and free fermions .
the schrdinger equation can be formulated as @xmath8 \psi = e\psi , \ ] ] where @xmath9 is the pseudo - potential describing the @xmath0 -wave scattering .
it has been shown that the @xmath0-wave scattering of two spin - polarized fermions in a tightly confined waveguide can be well described by the contact condition @xcite @xmath10 where @xmath11 ^{-1}\ ] ] is the effective 1d scattering length with @xmath12 the @xmath0-wave scattering length and @xmath13 the transverse oscillator length @xcite
. the contact condition can be reproduced by using the following pseudo - potential @xcite @xmath14 where @xmath15 and @xmath16 the scattering length can be tuned readily from @xmath17 to @xmath18 by sweeping an external magnetic field - the feshbach resonance , or by changing the geometry of the trapping potential - the confinement induced resonance and in this paper the full physical regimes @xmath19 will be studied .
similar to the case of bose gas , the important parameter characterizing the different physical regimes of the 1d fermi gas is @xmath20 , where @xmath21 and @xmath22 .
a standard rescaling procedure brings the schrdinger equation into a dimensionless one ( for simplicity we keep the original notations ) @xmath23 with @xmath24 where in the dimensionless interaction constant @xmath25 we intentionally keep the factor 2 in accordance with the bosonic case @xcite .
the wave function takes the general form @xmath26 where we have used @xmath27 to label the region @xmath28 .
the wave function of fermions should follow the antisymmetry of exchange , so our model is simplified into the solution of @xmath29 in the region @xmath27 with the open boundary condition @xmath30 using the bethe ansatz method we obtain the wavefunction parameterized by the set of quantum number @xmath31 known as quasi - momenta or rapidities @xcite @xmath32 \sin \left ( k_{p_1}x_{q_1}\right ) \\ & & \times \prod_{1<j < n}\sin \left ( k_{p_j}x_{q_j}-\sum_{l < j}\omega _ { p_lp_j}\right ) \\ & & \times \exp \left ( ik_{p_n}l\right ) \sin \left ( k_{p_n}\left ( l - x_{q_n}\right ) \right)\end{aligned}\ ] ] with @xmath33 and @xmath34 . here @xmath35 and @xmath36 denote sign factors associated with even / odd permutations of @xmath37 and @xmath38 , respectively .
the quasi - momenta @xmath39 can be easily determined from the bethe ansatz equations @xmath40 with @xmath41 .
these quasi - momenta lead us immediately to important physical quantities for our system .
for example , the energy eigenvalue is given by @xmath42 and the total momentum by @xmath43 .
it is clear that these bethe ansatz equations are the same as those in the case of bose gas if we simply make a substitution @xmath44 @xcite . in the regime @xmath27 ,
there is one - to - one correspondence between the quasi - momentum solution of attractive @xmath0-wave fermi gas and that of repulsive bose gas , but the total wave functions @xmath45 take different forms because of their distinct exchange symmetries .
this difference can be easily seen from the one - body density matrix and the momentum distribution .
taking the logarithm of bethe ansatz equations , we have @xmath46 for the ground state the set of integer @xmath47 ( @xmath48 ) . for simplicity , in the following evaluation
the length @xmath7 will be taken as one .
it is obvious that the solutions of @xmath49 is only relevant to @xmath50 for different @xmath7 .
the subsequent procedure is standard . by numerically solving the transcendental equations eqs.([bae ] ) , we obtain the quasi - momenta @xmath51 and thus the ground state wave function . in principle , all necessary information about the system can be inferred , including the one - body density matrix , the momentum distribution , and the excitation spectrum .
furthermore , one can apply the thermodynamic formalism for dealing with the one dimensional interacting systems developed by yang and yang @xcite .
-wave attractive interactions as a function of the dimensionless coordinates @xmath52 and @xmath53 for @xmath54 .
( a ) @xmath55 ; ( b ) @xmath56 ; ( c ) @xmath57 ; ( d ) @xmath58.,width=307 ] -wave attractive interactions for @xmath54.,width=336 ] before we proceed to the general case with intermediate interaction strength , we d like to take a look at the situation in the two limiting cases . in the limit of free fermion case , we have @xmath59 ( @xmath60 ) . in this exceptional case of @xmath55 ,
the quasi - momentum acquires the genuine physical meaning of momentum and the atoms occupy the single particle momentum states according to pauli exclusion principle .
we thus have the total wave function @xmath61 with @xmath62 the normalization constant , which is nothing but the slater determinant of the lowest @xmath6 eigenstates @xmath63 of the system of single particle in the hard wall potential . in the other limit of strongly attractive interaction ,
i.e. , the ftg limit , all quasi - momenta take the same value @xmath64 ( @xmath60 ) and @xmath65 .
thus we have the total wave function @xmath66 on the other hand , the fermi - bose mapping method has been used to give exactly the ground state of the ftg gas in @xcite @xmath67 with the antisymmetric function @xmath68sgn@xmath69 .
it is obvious that the above two wave functions match each other .
we now turn to the system with finite interaction strength @xmath70 . in this case
the quasi - momenta are decided by numerically solving the bethe ansatz equations and the total wave function is obtained by eq.([wf ] ) through @xmath71 under the restriction of exchange antisymmetry .
for the one dimensional interacting system , a quantity of fundamental importance in many - body physics is the one - body density matrix , which , in terms of the ground state wave function @xmath72 , is given by @xmath73 this quantity furnishes the expectation values of single particle observables such as the position density distribution @xmath74 , and the momentum distribution which is simply the fourier transformation of @xmath75 , @xmath76 we display the one - body density matrix and the position density distribution in fig . 1 and in fig . 2 for different @xmath0-wave attractive interactions .
the one - body density matrix expresses the self correlation and it means the probability that two successive measurements , one immediately following the other , will find the particle at the point @xmath52 and @xmath77 , respectively .
we notice that for all interacting strengthes there exists a strong enhancement of the diagonal contribution @xmath78 along the line @xmath79 . to see this more clearly the position density distribution @xmath80 for @xmath54 particles
is shown in fig .
2 with more variable interacting strengthes . for noninteracting system the atoms
behave as ideal fermions and the density profiles show obvious spatial oscillation structure . increasing the strength of the attractive @xmath0-wave interaction , as seen in fig .
2 , first leads to the depression of the amplitude of the oscillations of the density profile , followed by the emergence of the typical gaussian - like bosonic behavior and contraction of the half - width of the density . in the limit of infinitely strong attractive interaction between the fermions , the system enters into the ftg regime , and the density shows the same smooth profile as that of noninteracting bosons . particularly , the density distribution in fig
. 2 for ftg is identical to that of the free bose gas and there exists one - to - one correspondence for the density distribution between the attractive fermionic gas and the repulsive bosonic gas with the interacting strength related by @xmath81 @xcite .
-wave attractive interactions for @xmath54.,width=336 ] .,width=336 ] .
( a ) @xmath82 ; ( b ) @xmath83 ; ( c ) @xmath84 ; ( d ) @xmath85 ; ( e ) @xmath86 ; ( f ) @xmath87 .
, width=336 ] it is worth to emphasize that the density distribution is identical for tg bosons and noninteracting fermions , and also for ftg fermions and noninteracting bosons .
nevertheless , their momentum distributions show remarkable differences as shown in fig . 3 for fermions with different attractive interaction strengths and in fig . 4 for bosons with the corresponding repulsive interaction strengths .
the momentum distribution for fermions oscillate in the full regime and the number of oscillation peaks remains equal to the number of atoms in the system .
it becomes more and more nonuniform in the momentum space and and two sharp spikes indicating that there is high probability of finding the atom in momentum states around @xmath88 .
two other peaks at higher momentum diminish with increasing interaction but remain there even in the ftg limit due to the fermionic statistics .
the half - width of the profiles become larger as the attractive interaction increases . as a comparison ,
the momentum distributions of bosons with repulsive interactions are given in fig .
4 which are obtained by the fourier transformation of the one - body density matrix shown in fig .
obviously , the momentum distribution for the bosons exhibits quite different behaviors . as shown in fig .
4 , there is an obvious peak around the zero momentum point and the height of the peak shrinks with the increase of the repulsive interaction .
furthermore we find no oscillation at all in the momentum distribution for bosons . even in the tonks limit
, the momentum distribution of bosons does not show shell structure like free fermions .
the largest probability of distribution appears around the zero point of momentum and decreases rapidly for finite momentum values .
stronger repulsive interaction between the bosonic atoms tends to spread out the distribution into higher momentum space .
although the momentum distribution for the interacting bosonnic gases has been studied by different numerical methods @xcite , an exact result has never been given except in the tg limit @xcite . from the eqs .
( [ energy ] ) and ( [ bae ] ) , we see that the energy level structure of our fermionic model is exactly the same as the corresponding bosonic model with interaction strength related by @xmath44 .
therefore the thermodynamic properties of the fermionic atoms with attractive @xmath0-wave interactions are the same as the well known thermodynamic properties of the 1d boson gas @xcite with inverse coupling @xmath44 .
this implies that there exists a bose - fermi duality between the @xmath0-wave fermionic model and bosonic one with arbitrary interactions which can not be distinguished by the thermodynamic properties .
however , due to the different exchange symmetry of the wave functions , the observables associated with the wave functions rather than the square of wave functions ( density distribution ) should display different behaviors . as we have shown ,
the off - diagonal density matrix and the momentum distributions are different greatly , which result from the different statistics properties followed by bosons and fermions .
in summary , we have investigated the ground - state properties of fermionic gases with attractive @xmath0-wave interactions in a one - dimensional hard - wall trap . with the bethe ansatz method ,
the explicit wave function of the ground state and therefore the one - body density and momentum distributions are obtained .
it turns out that the density distributions show one - to - one correspondence between fermions with attractive @xmath0-wave interactions and bosons with repulsive interactions . for weak attractive interaction the density distributions of fermions display shell structures and the boson - like distributions
appear as the interactions become stronger . in the ftg limit
, the fermi gas exhibits the same distribution as that of the ideal bose gas .
this again confirms the bose - fermi duality : strongly interacting bosons behave like fermions , and vice versa .
nevertheless , from the viewpoint of momentum distribution , the conclusion is rather different . in the full interacting regime
the momentum distributions of fermions show typical oscillations , which is in sharp contrast with the case of bosonic atoms . in the tonks limit of infinite interaction , although the density profiles of one dimensional bosons display the fermion - like distribution , the momentum distribution is still boson - like .
is supported by nsf of china under grant no .
10574150 and programs of chinese academy of sciences .
y.z . is supported by 973 program under grant no .
2006cb921102 and shanxi province youth science foundation under grant no
. 20051001 .
the authors would like to acknowledge the hospitality of kitpc in beijing . | we investigate the ground state of the one - dimensional fermionic system enclosed in a hard - wall trap with attractive contact @xmath0-wave interactions .
based on the bethe ansatz method , the explicit wave function is derived by numerically solving the bethe ansatz equations for the full physical regimes ( @xmath1 ) . with the exact wave function some quantities which are important in many - body physics
are obtained , including the one - body density matrix and the momentum distribution of the ground state for finite system .
it is shown that the shell structure of the density profiles disappears with the increase of the interaction and in the fermionic tonks - girardeau ( ftg ) limit the density distribution shows the same behavior as that of an ideal bose gas .
however the one - body density matrix and the momentum distribution exhibit completely different structures compared with their bosonic counterparts . |
physical parameters of glaxy clusters , such as total mass and gas mass , are commonly studied through scaling relations .
these relations assume that both growing and mature clusters are relaxed , self - similar systems such that relations between e.g. @xmath9 , @xmath10 , @xmath11 , @xmath2 , @xmath7 , etc .
are simple power laws ( see e.g. @xcite and @xcite for a recent review ) . deviations from hydrostatic equilibrium ( hse ) ( or from virialisation ) and self - similarity during cluster mergers
will cause scatter around the scaling relations .
studies in the literature aim to use these relations to make accurate determinations of e.g. total cluster mass , and therefore often focus on minimising the scatter either by careful sample selection of low - redshift , relaxed clusters ( e.g. , , @xcite , @xcite ) , or by finding a particularly low - scatter mass proxy ( e.g. @xcite , , , @xcite ) .
these approaches often produce low - scatter relations that agree with the self - similar predictions .
however , @xcite , using simulations of two - body cluster mergers to track the evolution of a merger ( from a relaxed system before the start of the merger through to relaxation of the merged system ) in the plane of a scaling relation , find large changes in cluster observables _ along _ the relation with little perpendicular displacement .
assessment of these cluster parameter values through calculation from sunyaev
zeldovich ( sz , @xcite ) and x - ray observation provides a critical probe of the dynamical state of the cluster gas due to the difference in dependencies of the sz and x - ray flux densities on the electron number density , @xmath0 .
the sz effect is the inverse compton scattering of cmb photons by hot cluster gas , and is @xmath12 , where @xmath7 is the plasma temperature and @xmath13 the line element along the line of sight through the cluster .
the x - ray bremsstrahlung signal is @xmath14 , where @xmath15 is the cooling function ( @xmath15@xmath16 @xmath17 for the clusters in this paper ) .
parameter values estimated from measurement of sz and x - ray signals will , therefore , also depend differently on @xmath0 and @xmath7 .
as cluster mergers are known to produce regions of higher density gas , through processes such as shocking , x - ray parameter estimation is likely more sensitive to dynamical state , and will produce larger displacements along scaling relations during a merger than sz parameter values .
this implies that merger activity can be identified by looking at discrepancies between sz and x - ray measurements . to test this observationally ,
we use the clash sample of well - studied clusters selected by @xcite to form a sample of massive clusters , most of which are classified in the literature as relaxed , plus a small number of clusters with pronounced strong gravitational lensing ( see section [ sec : sample ] ) . here
we discuss measurements of a sub - sample of clash clusters via the sz effect using the arcminute microkelvin imager ( ami , @xcite ) .
the sz signal measures the comptonization parameter , @xmath18 , the line - of - sight integral of the number of collisions multiplied by the mean fractional energy change of the cmb photons per collision : @xmath19 where @xmath20 is the thomson scattering cross - section , @xmath21 the electron mass , @xmath22 the speed of light .
equation [ eq : ypar ] shows that the sz surface brightness is proportional to the electron pressure , @xmath23 , assuming an ideal gas law , integrated along the line of sight .
integrating @xmath18 over the solid angle @xmath24 subtended by the cluster gives @xmath25 , which quantifies the internal energy of the cluster gas , providing a proxy for total mass , given redshift information . in x - ray studies
@xmath1 , found from @xmath26 , is used as an analogue of @xmath25 which is proportional to the product of the gas mass and the mean temperature measured from sz within a sphere ( or a cylinder ) .
@xcite find , using simulated data , that @xmath1 provides an equally good proxy for total mass as @xmath25 .
the mean cluster temperature has also been widely used as a proxy for total cluster mass .
cluster @xmath7 has traditionally been measured through x - ray spectroscopy ; with good enough sensitivity and angular resolution , annular averaging gives temperature profiles out to , for some clusters , radii of @xmath271mpc ( see e.g. accept database , @xcite , @xcite , ) . @xcite and @xcite
show that a gas temperature profile can also be obtained via sz observation , given assumed geometry and dynamical state , and given a prior on the gas fraction @xmath28 at @xmath29 . in this study ,
cluster parameters are derived from our ami sz measurements in a fully bayesian way using the model described in @xcite and ( 2013 ) .
this model uses a navarro , frenk and white ( nfw ) profile to describe the dark matter density , which is believed , from cosmological n - body simulations , to accurately model all dark matter halos @xcite .
a generalised navarro , frenk and white ( gnfw ) profile is used to describe the gas pressure , shown to follow self - similarity more closely than the density or temperature at high radius @xcite .
further conditions of spherical symmetry , hse , and a small @xmath30 compared to unity , produces cluster properties as functions of radius . throughout , we assume @xmath31 = 70 km @xmath32 and a concordance @xmath15cdm cosmology with @xmath33 = 0.3 , @xmath34 = 0.7 , @xmath35 = 0 , @xmath36 = 0.041 , @xmath37 = @xmath381 , @xmath39 = 0 and @xmath40 = 0.8 .
all cluster parameter values are at the redshift of the cluster .
we emphasise that we denote @xmath41 as @xmath1 for either sz or x - ray .
the clash ( cluster lensing and supernova survey with hubble ) sample consists of 25 massive clusters , covering a large redshift range ( @xmath42 from 0.213 to 0.888 ) , selected for strong and weak lensing observation with hubble and _ subaru _ @xcite . 20 of these clusters were selected from @xmath4 x - ray observations to be dynamically relaxed .
the remaining five clusters were selected solely based on their high lensing - strength and consequently include some of the most disturbed clusters known @xcite .
eleven of these clusters are visible to ami which is currently restricted to a declination range of 20@xmath43 to 85@xmath43 . since the clash sample
is composed mainly of abell and macs clusters , all eleven clusters in the sub - sample had in fact been observed ( with varying sensitivities ) by ami in 2009 to 2012 .
these observations showed that the field of view of maj1720 + 3536 contains too bright a source to allow sz mapping with ami .
we discard this cluster leaving an ami - clash sub - sample of ten .
we have searched the literature to find additional assessments of the dynamical states and x - ray morphologies of the ten clusters ; table [ tab : litrev ] summarises these findings .
lcc & sa & la + antenna diameter & 3.7 m & 12.8 m + number of antennas & 10 & 8 + baseline lengths ( current ) & 520 m & 18110 m + primary beam ( 15.7 ghz ) & 20.1 arcmin & 5.5 arcmin + typical synthesized beam & 3 arcmin & 30 arcsec + flux sensitivity & 30 mjy s@xmath44 & 3 mjy s@xmath44 + observing frequency & + bandwidth & + channel bandwidth & + lcccc cluster & & + & 2009 - 12 & new & 2009 - 12 & new + a611 & 18 & 16 & 24 & 14 + a1423 & 57 & & 18 & 8 + a2261 & 60 & 8 & 7 & 124 + clj1226 + 3332 & 77 & & 48 & + maj0647 + 7015 & 25 & 8 & 21 & 8 + maj0717 + 3745 & 49 & 9 & 30 & 38 + maj0744 + 3927 & 16 & 9 & 10 & 8 + maj1149 + 2223 & 38 & 7 & 18 & 16 + maj1423 + 2404 & 42 & 7 & 27 & 8 + rxj1532 + 3021 & 63 & 77 & 25 & 33 + ll parameter & prior + cluster position @xmath45 & gaussian at sa pointing centre , @xmath46 set at 60 arcsec regardless of precision of cluster position + mass @xmath47/@xmath48 & uniform in log space between 1@xmath49 and 6@xmath50 + gas fraction @xmath51 & gaussian with @xmath52=0.13 , @xmath46= 0.02 + shape parameters @xmath53 & delta - function with
universal " values , see @xcite + source position @xmath54 & delta - function at la position + source flux density @xmath55 @xmath564@xmath57 & gaussian at la value , with @xmath46=40 per cent of la flux density + source spectral index when @xmath58 @xmath564@xmath57 & gaussian centred at la - fitted spectral index with @xmath46=la error , or prior based on 10c , see section [ sec : priors ] + source flux density @xmath55 @xmath594@xmath57 & delta - function at la value , unless close to cluster + source spectral index when @xmath58 @xmath594@xmath57 & delta - function centred at la - fitted spectral index , or based on 10c , see section [ sec : priors ] +
sz observations were carried out with ami , a dual array of interferometers observing at centre - frequency 15.7ghz , located near cambridge .
the arrangement of ten 3.7-m antennas with baselines of 520 m in the small array ( sa ) and eight 12.8-m antennas with baselines of 18110 m in the large array ( la ) allows for study of the large - scale sz effect from clusters along with subtraction of otherwise confusing radio sources . for a full description of the instrument see @xcite .
our pointing strategy for each cluster is as follows .
a single sa pointing centre is observed , centred on the cluster x - ray centre . for the la , the smaller primary beam size ( see table [ table_ami ] )
requires that the same area is covered with a mosaiced 61-point la observation ; the central 19 pointings of this are observed 3@xmath60 longer than the outer pointings to reach lower noise levels near the cluster centre .
our sensitivity aim for each cluster has been to achieve thermal noise levels below 100@xmath52jy in the centre of the sa maps and averaging below 80@xmath52jy over the central 19-pointings of the la maps .
the 2009 - 2012 observations of clj1226 + 3332 on both arrays and the sa 2009 - 2012 observations of a1423 and maj1423 + 2404 have the sensitivities that we require . for the rest we have made new observations so that these , plus the 2009 - 2012 observations , achieve the sensitivity aim . a summary of the observations carried out is presented in table [ table : obs ] .
note that , for rxj1532 + 3021 , 2009 - 2012 observations were made with a somewhat erroneous pointing centre , and we use only new , correctly centred , observations .
data processing is carried out with our in - house software package reduce , in which the raw data are calibrated and flagged for interference and telescope errors .
note that all ami data , whenever taken , have been reduced with our latest reduction pipeline with optimised interference flagging and calibration .
flux calibration is carried out with observations of 3c48 , 3c147 and 3c286 , with 3c286 flux densities calibrated against very large array ( vla ) measurements @xcite .
data are fourier transformed into frequency channels and then written out as @xmath61-fits files .
data @xmath61-fits files from 2009 - 2012 and from new observations are concatenated in @xmath61-space to produce a single data set in each array .
for source - finding and for initial visual inspection of a cluster field , the @xmath61-data are imaged in aips using automated clean procedures with noise levels determined using imean .
the la channel maps are passed through the source - find algorithm @xcite which estimates la source positions , flux densities and spectral indices to use as priors in mcadam @xcite , our bayesian analysis system ( section [ sec : analysis ] ) .
if sources in the fields exhibit variability , concatenating @xmath61-data taken a few years apart could introduce inaccuracies in source flux estimates between the arrays .
if the proportion of old data and new data is different between the arrays , the average sa flux may be significantly different from the average la flux that is being subtracted , leaving residuals .
only a few sources in the maps show variability between old ( 2009 - 2012 ) and new ( 2013 - 2014 ) observations at a level that would effect parameter estimation if the la flux was directly subtracted from the sa data .
this effect is accounted for in mcadam ( section [ sec : analysis ] ) .
analysis of ami data is carried out in @xmath61-space using our bayesian analysis package mcadam @xcite which uses the fast sampler multinest ( @xcite , @xcite and @xcite ) to estimate cluster parameters given a cluster model ; mcadam simultaneously fits the radio source environment and takes full account of the power spectrum of the primordial cmb structure as a function of angular scale as seen by ami , thermal noise in the @xmath61-data , and confusion noise , in the form of a generalised noise covariance matrix .
concatenated sa visibilities must first be binned to reduce the size of the data , easing memory demands ; for minimal loss of information , bin sizes are less than aperture illumination function fwhm .
the model we employ for this analysis is described in @xcite and @xcite , and is based on a number of key assumptions .
the first two are the functional forms of the dark matter density profile and the gas pressure profile within a spherical geometry .
the non - baryonic matter distribution of a cluster is described using a nfw model , @xmath62 where @xmath63 is an overall normalisation coefficient and the scale radius , @xmath64 , is the value of @xmath65 at which @xmath66 .
a gnfw model is used to describe the gas pressure profile @xmath67 where @xmath68 is an overall normalisation coefficient , and @xmath69 is the scale radius .
the parameters @xmath70 , @xmath71 and @xmath22 describe the slopes of the pressure profile at @xmath72 , @xmath73 and @xmath74 respectively .
the third assumption is of hydrostatic equilibrium throughout the cluster .
the final assumption is that the gas mass fraction @xmath30 is small compared to unity , so that @xmath75 .
these assumptions allow the total mass within a radius @xmath65 to be derived as @xmath76 using the assumption of hse , the electron number density profile @xmath77 can be found and , taking the gas to be ideal , the electron temperature profile is calculated by @xmath78 , @xmath79 \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}\ ] ] parameter values are estimated at or within an overdensity radius @xmath80 , the cluster radius internal to which the mean density is @xmath81 times the critical density at the cluster redshift . throughout this paper
we use parameters estimated from this model at or within @xmath29 and @xmath8 . [ cols="<,^,^,^,^,^,^ " , ] internal cluster dynamics during merger activity
have been investigated through simulations covering a wide variety of merger scenarios and individual systems , see e.g. @xcite , @xcite , @xcite and @xcite .
we focus on the comprehensive work of @xcite , which follows the cluster merger process for three merger mass ratios for each of three merger impact parameters ( see also @xcite , @xcite and @xcite ) .
from example merger stages in the paper itself and the suite of simulation videos online , at http://visav.phys.uvic.ca/~babul/merger_paperi/ , this study provides simulations of both the x - ray and the sz signals , ideal for the focus of this paper , as well as the x - ray temperature , gas surface density and entropy .
while these simulations include the effects of radiative cooling , star formation and minimal feedback from supernovae , @xcite make clear that they do not include factors such as feedback from agn , magnetic fields , pressure due to cosmic rays , and conduction . in the following ,
we compare the simulated x - ray morphology and temperature maps with real x - ray maps and temperature profiles on the @xmath4 accept database . for clusters that appear relaxed , or very close to relaxed , we class their dynamical state as @xmath825gyr since first pericentre ( the first closest approach of the secondary cluster to the primary ) , as depicted by poole et al . :
we class a611 , a1423 , a2261 and maj1423 + 2404 in this way .
the x - ray morphologies of a611 and a1423 do not look similar to any of the simulated merger stages , with circular shapes and no visible concentrations of x - ray emission .
neither cluster is present in the @xcite sample , so are not included in figure [ fig : ymratio ] , but do not show significantly higher @xmath6 values relative to @xmath5 in figure [ fig : tempszx ] .
a2261 appears in both figures [ fig : ymratio ] and [ fig : tempszx ] , also showing no significant difference in x - ray and sz estimated parameter values .
the x - ray morphology most closely resembles the poole et al .
simulations after 5gyr apart from an area of isolated substructure lying approximately 0.7mpc from the cluster centre ( see e.g. maughan et al . ) .
given the low x - ray temperature and agreement of sz and x - ray parameter estimates , it could be the start of a very high mass - ratio merger , well before first pericentre .
maj1423 + 2404 also has a relaxed x - ray morphology and displays little discrepancy between x - ray - estimated and sz - estimated @xmath1 vs @xmath2 in figure [ fig : ymratio ] and @xmath5 vs @xmath6 in figure [ fig : tempszx ] , also supporting a relaxed classification .
rxj1532 + 3021 is reportedly a very strong cool - core cluster , supporting a relaxed classification .
@xcite investigate the x - ray morphology using @xmath83-@xmath84 and deep @xmath4 observations and reveal central substructure from substantial agn activity and a cold front .
the authors conclude that the origin of the cold front is either cool gas dragged out by the agn outburst or turbulence from sloshing of the cool core induced by a minor merger .
figure [ fig : tempszx ] shows good agreement between @xmath5 and @xmath6 values but sz- and x - ray - derived @xmath1 and @xmath2 values in figure [ fig : ymratio ] are more discrepant .
this supports the theory of a low - level merger proposed by hlavacek - larrondo et al . which may be old enough to no longer exhibit the higher temperatures associated with mergers .
although clj1226 + 3332 is classed by @xcite and @xcite as relaxed given its x - ray morphology , differences in sz and x - ray parameter values suggest otherwise .
figures [ fig : ymratio ] and [ fig : tempszx ] show large discrepancies between x - ray and sz in @xmath1 and @xmath2 and in average temperature estimates .
the average x - ray temperature and temperature profiles from @xmath4 on the accept database , and @xmath83-@xmath84 in @xcite , show very high temperatures for the low mass of clj1226 + 3332 , suggesting an early - stage merger . in conjunction with its circular shape ,
slightly displaced x - ray peak brightness from the cluster centre and some substructure towards the south - west , clj1226 + 3332 is likely a recent , close to head - on , minor merger .
maughan et al . and
@xcite also discuss the non - relaxed state of clj1226 + 3332 .
jee & tyson compare their weak - lensing mass reconstruction with the x - ray temperature map of maughan et al . : they note the correlation of a high - temperature asymmetry with an area of low - luminosity substructure to the south - west of the cluster centre , concluding that the substructure has just passed through the primary cluster . maj0647 + 7015 ( strong - lensing selected ) has high values of @xmath85 and @xmath86 , both of which are very discrepant from the sz values in figure [ fig : tempszx ] .
figure [ fig : ymratio ] shows large differences also between sz and x - ray @xmath1 and @xmath2 values .
the x - ray morphology is non - circular with no central peak in x - ray surface brightness which , along with the high @xmath6 values , indicate maj0647 + 7015 has recently undergone a close to head - on major merger .
maj0717 + 3745 has a complicated x - ray morphology that suggests multiple mergers : a triple merger , according to @xcite .
figure [ fig : ymratio ] shows large differences between x - ray and sz measurements of @xmath1 and @xmath2 .
no significant discrepancy is evident between the sz and x - ray temperatures at 700kpc , in figure [ fig : tempszx ] .
this is likely due to effects of annular averaging over a complex x - ray temperature distribution that is poorly approximated by the model profile in both x - ray and sz analyses .
this is supported by a larger discrepancy in average temperature , for which annular averaging was not used .
parameter estimation for maj0744 + 3927 suggests a relaxed state .
both @xmath85 and @xmath86 are consistent with the values estimated from sz , shown by figure [ fig : tempszx ] , and there is no significant discrepancy in @xmath1 and @xmath2 between sz and x - ray in figure [ fig : ymratio ] .
the accept x - ray temperature profile shows a low peak temperature for the mass of the cluster and some evidence for a cool - core .
however , the x - ray morphology does not appear relaxed : there is a clear displacement in x - ray peak brightness from the centre and some extended clump features .
these indicate an old merger returning to a relaxed state , supported by the small cool core . in a strong - lensing investigation of maj1149
+ 2223 , @xcite found the mass distribution to be best described as a main dark matter halo plus three , group - sized halos .
@xcite suggest that the high velocity dispersion , measured by @xcite , arises from merging along the line of sight .
there is , however , no significant discrepancy between x - ray estimated and sz estimated parameter values in figures [ fig : ymratio ] and [ fig : tempszx ] ; we argue that there are two reasons that may explain this .
the first reason concerns the mass of the main halo compared with any of the masses of the group halos .
given the findings of smith et al . and the low x - ray temperature , we expect the individual merging groups to be of low mass relative to the primary cluster : @xcite show the x - ray temperatures reached in binary mergers of 1:1 and 3:1 are much higher than in 10:1 mergers , where the temperature is lower with larger impact parameter .
the lack of a high x - ray surface - brightness region in the accept @xmath4 maps also indicates that mergers are of high impact - parameter .
the second reason concerns the complex shape of the system , evident in the accept images , for which annular averaging will be problematic .
we expect the x - ray temperature profile to be biased low due to cooler , higher - radius gas being included in each annulus .
we have observed the eleven clusters in the clash sample that are accessible to ami , and discard one due to a very bright source on the edge of the field of view .
the remaining ten clusters have been analysed in a fully bayesian way to give estimated parameter values from sz measurement . 1 .
although 20 out of the 25 clash sample members were selected to be relaxed , we find disagreement in the literature on the dynamical states of many of our ami - clash sub - sample , illustrating the difficulty in determining cluster dynamical state and identifying mergers .
we investigate the correlations in our model and use it to calculate @xmath87@xmath88 curves , varying @xmath30 and redshift .
we discuss the effect these parameters have on the power - laws fitted to the curves : @xmath87@xmath88 is dependent on the cluster redshift and on the value of @xmath30 .
x - ray studies have found sensitivity of the scatter about scaling relations to the dynamical states of the clusters included . here
, we investigate the sensitivity of our model correlations to dynamical state by introducing variations in cluster pressure profile shape parameters : a consequence of merger activity that can be induced in the analysis . in our analysis
these are fixed to the universal " values .
varying @xmath70 , the shape parameter best constrained by ami sz data , over a large range induces only small changes in scaling relation power - laws .
this is consistent with @xcite who track cluster observables of merging systems in the plane of scaling relations , finding very little perpendicular scatter but large variations along the scaling relation , over the course of a merger .
4 . due to the difference in dependence of sz and x - ray measurement on @xmath0 ,
x - ray observables are much more sensitive to changes in mass , pressure , temperature and density during a merger .
discrepancies between parameter estimates derived from sz and x - ray measurement can help identify clusters undergoing mergers . denoting @xmath41 by @xmath1 ( for x - ray and sz ) , we compare two @xmath1@xmath2 scaling relations of the sub - sample members : one plotted using ami sz parameter values and the second using @xmath4 x - ray parameters from @xcite .
in addition to a difference in scatter we see an apparent movement " of sample members along the line of the relation , between sz and x - ray .
these discrepancies are visualised by plotting the ratios of ami and @xmath4 @xmath1 and @xmath2 values showing a population of our sub - sample for which the sz and x - ray parameters agree well .
other clusters are discrepant by up to @xmath602 , all towards higher x - ray @xmath1 and @xmath2 values _ along _ the line of the relation .
we also plot temperature estimates made from sz and x - ray observation and find a similar split of clusters into two populations that are in agreement with those found from the ratios of ami and @xmath4 @xmath1 and @xmath2 values .
this result comes from the comparison of ami sz parameter estimates with those from three independent x - ray analyses , addressing possible inconsistencies between methods .
dynamical state classifications in the literature of the ten clusters report : a611 , maj1423 + 2404 and rxj1532 + 3021 are relaxed ; maj0647 + 7015 , maj0717 + 3745 and maj1149 + 2223 are mergers ; and a1423 , a2261 , clj1226 + 3332 , and maj0744 + 3927 have mixed reports .
using discrepancies in sz and x - ray parameter estimates and comparisons of @xcite merger simulations with x - ray morphology we determine the dynamical states of the ten clusters in the sub - sample .
we class a611 , a1423 , a2261 and maj1423 + 2404 as relaxed .
maj0647 + 7015 and maj0717 + 3745 , selected for their lensing strength , we find to be mergers .
as expected from the mixed classifications in the literature , maj0744 + 3927 and clj1226 + 3332 , although selected as relaxed , are also mergers .
maj1149 + 2223 , although reported in the literature as highly disturbed , shows no significant discrepancies between sz and x - ray parameters .
we conclude that low mass infalling cluster groups with high impact - parameters will cause less gas shocking and/or fractionation than lower mass - ratio mergers .
we find evidence supporting the presence of an old , low level merger in rxj1532 + 3021 , postulated by @xcite .
we thank the anonymous reviewer for considered thoughts and useful comments that have improved the paper .
we also thank the staff of the mullard radio astronomy observatory for their invaluable assistance in the commissioning and operation of ami , which is supported by cambridge university .
wjh and cr are grateful for the support of stfc studentships .
cr also acknowledges the support of cambridge university .
mo and ycp acknowledge support from research fellowships from sidney sussex college and trinity college , cambridge , respectively .
we thank arif babul for his assistance in accessing the poole et al .
online materials .
much of this work was undertaken on the cosmos shared memory system at damtp , cambridge university , operated on behalf of the stfc dirac hpc facility .
this equipment is funded by bis national e - infrastructure capital grant st / j005673/1 and stfc grants st / h008586/1 , st / k00333x/1 . | using arcminute microkelvin imager ( ami ) sz observations towards ten clash clusters we investigate the influence of cluster mergers on observational galaxy cluster studies .
although selected to be largely relaxed , there is disagreement in the literature on the dynamical states of clash sample members .
we analyse our ami data in a fully bayesian way to produce estimated cluster parameters and consider the intrinsic correlations in our nfw / gnfw - based model .
varying pressure profile shape parameters , illustrating an influence of mergers on scaling relations , induces small deviations from the canonical self - similar predictions in agreement with simulations of @xcite who found that merger activity causes only small scatter perpendicular to the relations .
we demonstrate this effect observationally using the different dependencies of sz and x - ray signals to @xmath0 that cause different sensitivities to the shocking and/or fractionation produced by mergers .
plotting @xmath1@xmath2 relations ( where @xmath3 ) derived from ami sz and from @xmath4 x - ray gives ratios of ami and @xmath4 @xmath1 and @xmath2 estimates that indicate movement of clusters _ along _ the scaling relation , as predicted by @xcite .
clusters that have moved most along the relation have the most discrepant @xmath5 and @xmath6 estimates : all the other clusters ( apart from one ) have sz and x - ray estimates of @xmath2 , @xmath7 and @xmath1 that agree within @xmath8 .
we use sz vs x - ray discrepancies in conjunction with @xmath4 maps and @xmath6 profiles , making comparisons with simulated cluster merger maps in @xcite , to identify disturbed members of our sample and estimate merger stages .
= 1 [ firstpage ] galaxies : clusters : general methods : observational techniques : interferometric large - scale structure of the universe .
lcclcluster & relaxed & unrelaxed & notes + + a611 & & & widely agreed relaxed , see e.g. @xcite , @xcite + & & & relaxed : e.g. classified ` non - distorted ' by @xcite and ` regular ' by @xcite + & & & unrelaxed : due to centroid shift , cuspiness and cooling time
@xcite + & & & widely considered relaxed , see e.g. @xcite , @xcite + & & & unrelaxed due to extra structure to the south - west , @xcite + & & & classified relaxed e.g. @xcite , @xcite + & & & merger given x - ray morphology , @xcite , and temperature map , @xcite + maj0647 + 7015 & & & strong - lens selected , highly unrelaxed : @xcite + maj0717 + 3745 & & & strong - lens selected , highly unrelaxed and complex merger : @xcite + & & & in samples of large relaxed clusters @xcite , @xcite , and @xcite + & & & substructure identified , e.g. @xcite , @xcite .
disturbed , @xcite + maj1149 + 2223 & & & strong - lens selected , highly complex merger , see @xcite + maj1423 + 2404 & & & highly relaxed state , pronounced cool - core , e.g. @xcite and @xcite + & & & presence of a cool - core e.g. @xcite , @xcite classifies it as relaxed + & & & cold front possibly from low - level merger turbulence @xcite + |
qcd at low energies is hard to solve , perhaps too hard for mere mortals to solve , even when assisted with the latest supercomputers .
qcd is the theory that describes the interactions of quarks and gluons .
qcd has been well tested in high energy scattering experiments where perturbation theory is valid .
however , qcd should also describe nuclear physics and the mass spectrum of hadrons .
hadron masses depend on the coupling ( @xmath0 ) like @xmath1 hence perturbation theory ca nt be used to compute the masses of hadrons such as the proton . the only technique that offers any prospect of computing masses and matrix elements non - perturbatively , from first principles , is lattice qcd . in lattice
qcd , qcd is transcribed to a lattice and the resulting equations are solved numerically on a computer .
the computation of the hadron spectrum using lattice qcd started in the early 80 s @xcite .
the modern era in lattice qcd calculations of the hadron spectrum started with the results of the gf11 group @xcite .
the gf11 group were the first to try to quantify the systematic errors in taking the continuum and infinite volume limits .
the goal of a `` numerical solution '' to qcd is not some kind of weird and misguided reductionist quest .
our inability to solve qcd has many profound consequences .
a major goal of particle physics is to look for evidence for physics beyond the standard model of particle physics .
one way of doing this is to extract the basic parameters of the standard model and look for relations between them that suggest deeper structure . to test the quark sector of the standard model
requires that matrix elements are computed from qcd @xcite .
the problem of solving qcd is symbolically summarised by the errors on the quark masses . for example , the allowed range on the strange quark mass in the particle data table @xcite is 80 to 155 mev ; a range of almost 100% .
the value of top quark mass , quoted in the particle data table , is @xmath2 gev as the mass of the quark increases its relative error decreases .
the dynamics of qcd becomes simpler as the mass of the quarks gets heavier .
wittig has reviewed the latest results for the light quark masses from lattice qcd @xcite irrespective of applications of solutions to qcd to searches for physics beyond the standard model , qcd is a fascinating theory in its own right .
qcd does allow us to test our meagre tools for extracting non - perturbative physics from a field theory . in this review i will focus on the results from lattice gauge theory for the masses of the light mesons and baryons .
i will not discuss flavour singlet mesons as these have been reviewed by michael @xcite .
there has been much work on the spectroscopy of hadrons that include heavy quarks @xcite , however i will not discuss this work .
the treatment of heavy quarks ( charm and bottom ) on the lattice has a different set of problems and opportunities over those for light quarks .
although the spectroscopy of hadrons with heavy quarks in them can naturally be reviewed separately from light quark spectroscopy , the physics of heavy hadrons does depend on the light quarks in the sea .
in particular the hyperfine splittings are known to have an important dependence on the sea quarks @xcite . until recently
, the computation of the light hadron spectrum used to be just a test of the calculational tools of lattice qcd .
the light hadron spectrum was only really good for providing the quark masses and estimates of the systematic errors .
however , the experimental program at places such as the jefferson lab @xcite has asked for a new set of quantities from lattice qcd .
in particular the computation of the spectrum of the @xmath3 s is now a goal of lattice qcd calculations .
as the aim of the review is to focus more on the results of lattice calculations , i shall mostly treat lattice calculations as a black box that produces physical numbers .
however , `` errors are the kings '' of lattice qcd calculations because the quality and usefulness of a result usually depends on the size of its error bar , hence i will discuss the systematic errors in lattice calculations . most of systematic errors in lattice qcd calculations can be understood using standard field theory techniques .
i have also included an appendix [ cmn : se : technicaldetails ] . on some of the `` technical tricks '' that are important for lattice qcd insiders , but of limited interest to consumers of lattice results .
however , it is useful to know some of the jargon and issues , as they do effect the quality of the final results .
there are a number of text books on lattice qcd .
for example the books by montvay and munster @xcite , rothe @xcite , smit @xcite and creutz @xcite provide important background information .
the large review articles by gupta @xcite , davies @xcite and kronfeld @xcite also contain pertinent information .
the annual lattice conference is a snap - shot of what is happening in the lattice field every year .
the contents of the proceedings of the lattice conference have been put on the hep - lat archive for the past couple of years @xcite .
the reviews of the baryon spectroscopy from lattice qcd by bali @xcite and edwards @xcite describe a different perspective on the field to mine .
there used to be a plenary reviews specifically on hadron spectroscopy at the lattice conference @xcite .
the subject of hadron spectroscopy has now been split into a number of smaller topics , such as quark masses .
if the reader wants to play a bit with some lattice qcd code , then the papers by di pierro @xcite , contain some exercises and pointers to source code .
the milc collaboration also make their code publicly available ( try putting `` milc collaboration '' into a search engine ) .
in this section , i briefly describe the main elements of numerical lattice qcd calculations .
quantum chromodynamics ( qcd ) is the quantum field theory that describes the interactions of elementary particles called quarks and gluons .
the key aspect of a quantum field theory is the creation and destruction of particles .
this type of dynamics is crucial to qcd and one of the reasons that it is a hard theory to solve . in principle , because we know the lagrangian for qcd , the quantum field theory formalism should allow us to compute any quantity .
the best starting point for solving qcd on the computer is the path integral formalism .
the problem of computing bound state properties from qcd is reduced to evaluating equation [ eq : reallyqcd ] .
@xmath4 where @xmath5 and @xmath6 are the actions for the fermion and gauge fields respectively .
the path integral is defined in euclidean space for the convergence of the measure .
the fields in equation [ eq : reallyqcd ] fluctuate on all distance scales .
the short distance fluctuations need to be regulated . for computations of non - perturbative quantities
a lattice is introduced with a lattice spacing that regulates short distance fluctuations .
the lattice regulator is useful both for numerical calculations , as well for formal work @xcite ( theorem proving ) , because it provides a specific representation of the path integral [ eq : reallyqcd ] .
a four dimensional grid of space - time points is introduced .
a typical size in lattice qcd calculations is @xmath7 .
the introduction of a hyper - cubic lattice breaks lorentz invariance , however this is restored as the continuum limit is taken .
the lattice actions do have a well defined hyper - cubic symmetry group @xcite .
the continuum qcd lagrangian is transcribed to the lattice using `` clever '' finite difference techniques . in the standard lattice qcd formulation
, the decision has been made to keep gauge invariance explicit .
the quark fields are put on the sites of the lattice .
the gauge fields connect adjacent lattice points .
the connection between the gauge fields in lattice qcd and the fields used in perturbative calculations is made via : @xmath8 the gauge invariant objects are either products of gauge links between quark and anti - quark fields , or products of gauge links that form closed paths .
all gauge invariant operators in numerical lattice qcd calculations are built out of such objects .
for example the lattice version of the gauge action is constructed from simple products of links called plaquettes . @xmath9
the wilson gauge action @xmath10 is written in terms of plaquettes .
@xmath11 is the number of colours .
the action in [ eq : wgaugeaction ] can be expanded in the lattice spacing using equation [ eq : connect ] , to get the continuum gauge action . @xmath12
the coupling is related to @xmath13 via @xmath14 the coupling in equation [ eq : barecoupling ] is known as the bare coupling .
more physical definitions of the coupling @xcite are typically used in perturbative calculations . the fermion action is generically written as @xmath15 where @xmath16 is called the fermion operator , a lattice approximation to the dirac operator .
one approximation to the fermion operator on the lattice is the wilson operator .
there are many new lattice fermion actions , however the basic ideas can still be seen from the wilson fermion operator .
@xmath17 the wilson action can be expanded in the lattice spacing to obtain the continuum dirac action with lattice spacing corrections . to the dirac lagrangian .
the @xmath18 parameter is called the hopping parameter .
it is a simple rescaling factor that is related to the quark mass via @xmath19 at tree level .
an expansion in @xmath18 is not useful for light quarks , because of problems with convergence , however for a few specialised applications , it is convenient to expand in terms of @xmath18 @xcite . the fermion operator in equation [ eq : wilsonfermion ] contains a term called the wilson term that is required to remove fermion doubling @xcite .
the wilson term explicitly breaks chiral symmetry so equation [ eq : kappatomass ] gets renormalised . currently , there is a lot of research effort in designing lattice qcd actions for fermions with better theoretical properties .
i briefly describe some of these developments in the appendix [ cmn : se : technicaldetails ] .
most lattice qcd calculations obtain hadron masses from the time sliced correlator @xmath20 . @xmath21 where the average is defined in equation [ eq : reallyqcd ] . any gauge invariant combination of quark fields and gauge links
can be used as interpolating operators ( @xmath22 ) in equation [ eq : timeslicedcorr ] .
an example for an interpolating operator in equation [ eq : timeslicedcorr ] for the @xmath23 meson would be @xmath24 the interpolating operator in equation [ eq : localoper ] has the same @xmath25 quantum numbers as the @xmath23 .
the operator in equation [ eq : localoper ] is local , as the quark and anti - quarks are at the same location .
it has been found to better to use operators that build in some kind of `` wave function '' between the quark and anti - quarks . in lattice - qcd - speak
we talk about `` smearing '' the operator .
an extended operator such as @xmath26 might have a better overlap to the @xmath23 meson than the local operator in equation [ eq : nonlocaloper ] .
the @xmath27 function is a wave - function like function .
the function @xmath28 is designed to give a better signal , but the final results should be independent of the choice of @xmath28
. typical choices for @xmath28 might be hydrogenic wave functions .
unfortunately equation [ eq : nonlocaloper ] is not gauge invariant , hence vanishes by elitzur s theorem @xcite .
one way to use an operator such as equation [ eq : nonlocaloper ] is to fix the gauge .
a popular choice for spectroscopy calculations is coulomb s gauge . on the lattice coulomb s gauge
is implemented by maximising @xmath29 the gauge fixing conditions that are typically used in lattice qcd spectroscopy calculations are either : coulomb or landau gauge .
different gauges are used in other types of lattice calculations .
non - local operators can be measured in lattice qcd calculations , @xmath30 however these do not have any use in phenomenology @xcite , because experiments are usually not sensitive to single @xmath31 fock states .
there are also gauge invariant non - local operators that are used in calculations .
@xmath32 where @xmath33 is a product of gauge links between the quark and anti - quark .
there are many different paths between the quark and anti - quark .
it has been found useful to fuzz @xcite .
the gauge links by adding bended paths : @xmath34 where @xmath35 projects onto su(3 ) .
there are other techniques for building up gauge fields between the quark and anti - quark fields , based on computing a scalar propagator @xcite .
the path integral in equation [ eq : reallyqcd ] is evaluated using algorithms that are generalisations of the monte carlo methods used to compute low dimensional integrals @xcite .
the physical picture for equation [ eq : timeslicedcorr ] is that a hadron is created at time 0 , from where it propagates to the time t , where it is destroyed .
this is shown in figure [ eq : mesonprop ] .
equation [ eq : timeslicedcorr ] can be thought of as a meson propagator .
duncan et al .
@xcite have used the meson propagator representation to extract couplings in a chiral lagrangian .
the algorithms , usually based on importance sampling , produce @xmath36 samples of the gauge fields on the lattice .
each gauge field is a snapshot of the vacuum .
the qcd vacuum is a complicated structure .
there is a community of people who are trying to describe the qcd vacuum in terms of objects such as a liquid of instantons ( for example @xcite ) .
the lattice qcd community are starting to create publicly available gauge configurations @xcite .
this is particularly important due to the high computational cost of unquenched calculations .
an archive of gauge configurations has been running at nersc for many years .
the correlator @xmath20 is a function of the light quark propagators ( @xmath37 ) averaged over the samples of the gauge fields .
@xmath38 the quark propagator @xmath39 is the inverse of the fermion operator .
the quark propagator depends on the gauge configurations from the gauge field dependence of the fermion action ( see [ eq : wilsonfermion ] for example ) .
the correlator in equation [ eq : algorexplain ] is essentially computed in two stages
. first the samples of the gauge configurations are generated .
these are the building blocks of most lattice qcd calculations .
the correlators for specific processes are computed by calculating the hadron correlator for each gauge configuration and then averaging over each gauge configuration .
to better explain the idea behind computing quark propagators in lattice qcd calculations , it is helpful to consider the computation of the propagator in perturbation theory .
the starting point of perturbative calculations is when the quarks do not interact with gluons .
this corresponds to quarks moving in gauge potentials that are gauge transforms from the unit configuration @xmath40 under these conditions the quark propagator can be computed analytically from the fermion operator using fourier transforms ( see @xcite for derivation ) .
@xmath41 @xmath42 although in principle the propagator in equation [ eq : freeprop ] could be used as a basis of a perturbative expansion of equation [ eq : algorexplain ] , the physical masses depend on the coupling like @xmath43 ( see section [ se : latticespacing ] ) , so hadron masses can not be computed using this approach .
the quark propagator on its own is not gauge invariant .
quark propagators are typically built into gauge invariant hadron operators . however , in a fixed gauge quark propagators can be computed and studied . this is useful for the attempts to calculate hadron spectroscopy using dyson - schwinger equations @xcite . in lattice
qcd calculations the gauge fields have complicated space - time dependence so the quark propagator is inverted numerically using variants of the conjugate gradient algorithms @xcite .
weak coupling perturbation theory on the lattice is important for determining weak matrix elements , quark masses and the strong coupling .
the sum over @xmath44 in the time slice correlator ( equation [ eq : timeslicedcorr ] ) projects onto a specific momentum at the sink .
traditionally for computational reasons , the spatial origin had to be fixed either at a point , or with a specific wave function distribution between the quarks .
physically , it would be clearly better to project out onto a specific momentum at the origin .
the number of spatial positions at the origin is related to the cost of the calculation .
there are new lattice techniques called `` all - to - all '' that can be used to compute a quark propagator from any point to another without spending prohibitive amounts of computer time @xcite .
there have been some studies of the point to point correlator @xcite @xmath45 it was pointed out by shuryak @xcite that the correlator in equation [ eq : pointtopoint ] may be more easily compared to experiment .
also , the correlator [ eq : pointtopoint ] could give some information on the density of states . in practise the point to point correlators were quite noisy .
the physics from the time sliced correlator is extracted using a fit model @xcite : @xmath46 where @xmath47 ( @xmath48 ) is the ground ( first excited ) state mass and the dots represent higher excitations .
there are simple corrections to equation [ eq : fitmodel ] for the finite size of the lattice in the time direction . in practise as recently emphasised by lepage et al .
@xcite , fitting masses from equation [ eq : fitmodel ] is nontrivial .
the fit region in time has to be selected as well as the number of exponentials .
the situation is roughly analogous to the choice of `` cuts '' in experimental particle physics .
the correlators are correlated in time , hence a correlated @xmath49 should be minimised . with limited statistics it can be hard to estimate the underlying covariance matrix making the @xmath50 test nontrivial @xcite . as ensemble sizes have increased the problems resulting from poorly estimated covariance matrices
have decreased .
it is usually better to measure a matrix of correlators so that variational techniques can be used to extract the masses and amplitudes @xcite . @xmath51
the correlator in equation [ cmn : eq : timeslicedvary ] is analysed using the fit model in equation [ cmn : eq : timeslicedvaryanal ] . @xmath52
the matrix @xmath53 is independent of time and in general has no special structure .
the masses can be extracted from equation [ cmn : eq : timeslicedvaryanal ] either by fitting or by reorganising the problem as a generalised eigenvalue problem @xcite .
obtaining the matrix structure on the right hand of equation [ cmn : eq : timeslicedvaryanal ] is a non - trivial test of the multi - exponential fit .
one piece of `` folk wisdom '' is that the largest mass extracted from the fit is some average over the truncated states , and hence is unphysical .
this may explain why the excited state masses obtained from @xcite in calculation with only two basis states were higher than any physical state the main ( minor ) disadvantage of variational techniques is that they require more computer time , because additional quark propagators have to be computed , if the basis functions are `` smearing functions '' .
if different local interpolating operators are used as basis states , as is done for studies of the roper resonance ( see section [ cmn : se : baryonexcite ] ) , then there is no additional cost .
the amount of computer time depends linearly on the number of basis states .
also the basis states have to be `` significantly '' different to gain any benefit .
apart from a few specialised applications @xcite the efficacy of a basis state is not obvious until the calculations is done .
although in principle excited state masses can be extracted from a multiple exponential fit , in practise this is a numerically non - trivial task , because of the noise in the data from the calculation .
there is a physical argument @xcite that explains the signal to noise ratio .
the variance of the correlator in equation [ eq : timeslicedcorr ] is : @xmath54 the square of the operator will couple to two ( three ) pion for a meson ( baryon ) .
hence the noise to signal ratio for mesons is @xmath55 and noise to signal ratio for baryons is @xmath56 the signal to noise ratio will get worse as the mass of the hadron increases . in figure [ eq : protoncorr ]
i show some data for a proton correlator .
the traditional way to plot a correlator is to use the effective mass plot .
@xmath57 when only the first term in equation [ eq : fitmodel ] dominates , then equation [ eq : effectivemass ] is a constant .
there is a simple generalisation of equation [ eq : effectivemass ] for periodic boundary conditions in time . in the past
the strategy used to be fit far enough out in time so that only one exponential contributed to equation [ eq : fitmodel ] .
one disadvantage of doing this is that the noise increases at larger times , hence the errors on the final physical parameters become larger @xcite .
many of the most interesting questions in hadronic physics involve the hadrons that are not the ground state with a given set of quantum numbers , hence novel techniques that can extract masses of excited states are very important .
= 0.1350 , @xmath13 = 5.2 . ]
recently , there has been a new set of tools developed to extract physical parameters from the correlator @xmath20 in equation [ eq : fitmodel ] .
the new techniques are all based on the spectral representation of the correlator in equation [ eq : timeslicedcorr ] .
@xmath58 the fit model in equation [ eq : fitmodel ] corresponds to a spectral density of @xmath59 spectral densities are rarely a sum of poles , except when the particles ca nt decay such as in the @xmath60 limit .
shifman reviews some of the physics in the spectral density @xcite . as the energy increases it
is more realistic to represent the spectral function as a continuous function .
leinweber investigated a sum rule inspired approach to studying hadron correlators @xcite on the lattice .
allton and capitani @xcite generalised the sum rule analysis to mesons .
the sum rule inspired spectral density is @xmath61 the operator product expansion is used to obtain the spectral density .
@xmath62 where @xmath63 is the quark mass and @xmath64 is the quark condensate .
the @xmath65 factors in equation [ cmn : eq : opespect ] are obtained by putting the ope expression for the quark propagator in [ eq : timeslicedcorr ] .
the final functions that are fit to the lattice data involve exponentials multiplied by polynomials in @xmath66 ( where @xmath67 is the time ) .
these kind of methods can also help to understand some of the systematic errors in sum rule calculations @xcite .
this hybrid lattice sum rule analysis is not in wide use in the lattice community .
see @xcite for recent developments .
there has been an interesting calculation of the mass of the charm quark using sum rule ideas and lattice qcd data @xcite .
there is currently a lot of research into using the maximum entropy method to extract masses from lattice qcd calculations . here
i follow the description of the method used by cp - pacs @xcite .
there are other variants of the maximum entropy method in use .
the maximum entropy approach @xcite is based on bayes theorem .
@xmath68 in equation [ eq : cmn : bayestheorem ] @xmath69 is the spectral function , @xmath70 is the data , and @xmath71 is the prior information such as @xmath72 , and @xmath73 is the conditional probability of getting @xmath69 given the data @xmath70 and priors @xmath71 .
the replacement of the fit model in equation [ eq : fitmodel ] is @xmath74 where the kernel @xmath75 is defined by @xmath76 with @xmath77 the number of time slices in the time direction . @xmath78 where @xmath79 is a known normalisation constant and @xmath80 is the number of time slices to use .
the correlation matrix is defined by @xmath81 @xmath82 where n is the number of configurations and @xmath83 is the correlator for the n - th configuration .
equation [ cmn : fig : chisq ] is the standard expression for the likelihood used in the least square analysis .
the prior probability depends on a function @xmath84 and a parameter @xmath85 .
@xmath86 with @xmath87 a known constant .
the entropy @xmath88 is defined by : @xmath89\ ] ] the @xmath85 parameter , that determines the weight between the entropy and @xmath90 , is obtained by a statistical procedure . the physics is built into the @xmath84 function .
cp - pacs build some constraints from perturbation theory into the function @xmath84 . the required function @xmath28 is obtained from the condition @xmath91 in principle the maximum entropy method offers the prospect of computing the masses of excited states in a more stable way than fitting multi - exponentials to data .
it will take some time to gain more experience with these types of methods .
i review some of the results from these techniques in sections [ se : mesonexcite ] and [ cmn : se : baryonexcite ] .
a more pragmatic approach to constraining the fit parameters was proposed by lepage et al .
the basic idea is to constrain the parameters of the fit model into physically motivated ranges .
the standard @xmath49 is modified to the @xmath92 .
@xmath93 with the prior @xmath94 defined by @xmath95 the priors constrain the fit parameters to @xmath96 = @xmath97 and @xmath98 = @xmath99 , the derivation of equation [ smn : eq : lepagechisq ] comes from bayes theorem plus the assumption that the distribution of the prior distribution is gaussian the idea is to `` teach '' the fitting algorithm what the reasonable ranges for the parameters are .
lepage et al .
@xcite do test the sensitivity of the final results against the priors .
it is difficult to understand why certain things are done in lattice calculations without an appreciation of the computational costs of lattice calculations .
the sesam collaboration @xcite estimated that the number of floating point operations ( @xmath100 ) needed for @xmath101 = 2 full qcd calculations as : @xmath102 for a box size of @xmath90 , lattice spacing @xmath103 , and @xmath104 is the number of sample of the gauge fields in equation [ eq : timeslicedcorr ] .
a flop is a floating point operation such as multiplication . with appropriate normalisation factors equation [ eq : manyflops ]
shows how `` big '' a computer is required and how long the programs should run for .
equation [ eq : manyflops ] shows that it is easy to reduce the statistical errors on the calculation as they go like @xmath105 , but more expensive to change the lattice spacing or quark mass . in some sense equation [ eq : manyflops ] ( or some variant of it )
is the most important equation in numerical lattice qcd .
to half the value of the pion mass used in the calculations requires essentially a computer that is seven times faster .
equation [ eq : manyflops ] is not a hard physical limit .
improved algorithms or techniques may be cheaper .
there are disagreements between the various collaborations that do dynamical calculations as to the exact cost of the simulations .
in particular , a formulation of quarks on the lattice called improved staggered fermions @xcite has a much less pessimistic computational cost .
the reason for this is not currently understood .
the big selling point of lattice qcd calculations is that the systematic errors can in principle be controlled . to appreciate lattice qcd calculations and to correctly use the results ,
the inherent systematic errors must be understood . for computational reasons ,
an individual lattice calculation is done at a finite lattice spacing in a box of finite size with quarks that are heavy relative to the physical quarks .
the final results from several calculations are then extrapolated to the continuum and infinite volume limit .
it is clearly important that the functional forms of the extrapolations are understood .
most of the systematic errors in lattice qcd calculations can be understood in terms of effective field theories @xcite .
a particularly good discussion of the systematic errors from lattice qcd calculations in light hadron mass calculations is the review by gottlieb @xcite .
the milc collaboration make an amusing observation pertinent to error analysis @xcite . in the heavy quark limit ,
the ratio of the nucleon mass to rho mass is @xmath106 , requiring one flop to calculate .
this is within 30 % of the physical number ( 1.22 ) .
this suggests for a meaningful comparison at the @xmath107 level , the error bars should be at the 10% level .
the high computational cost of the fermion determinant led to development of quenched qcd , where the dynamics of the determinant is not included in equation [ eq : reallyqcd ] , hence the dynamics of the sea quarks is omitted . until recently the majority of lattice qcd calculations were done in quenched qcd .
i will call a lattice calculation unquenched , when the dynamics of the sea quarks are included . the integration over the quark fields in equation [ eq : reallyqcd ]
can be done exactly using grassmann integration .
the determinant is nonlocal and the cause of most of the computational expense in equation [ eq : manyflops ] .
the determinant describes the dynamics of the sea quarks .
quenched qcd can be thought of as corresponding to using infinitely heavy sea quarks .
chen @xcite has made some interesting observations on the connection between quenched qcd and the large @xmath11 limit of qcd .
figure [ eq : quenchedlargen ] shows some graphs of the pion two point function .
in unquenched qcd all three diagrams in figure [ eq : quenchedlargen ] contribute to the two point function . in quenched qcd , only diagrams ( a ) and ( c ) contribute . in the large @xmath11 ( number of colours ) limit only graphs of type ( a ) in figure [ eq : quenchedlargen ] contribute .
this argument suggests that in the large @xmath11 limit quenched qcd and unquenched qcd should agree , hence for the real world @xmath108 case , quenched qcd and unquenched qcd should differ by @xmath109 @xmath110 30% .
chen @xcite firms up this heuteristic argument by power counting factors of @xmath11 and discusses the effect of chiral logs .
this analysis suggests that the quenching error should be roughly 30% , unless there is some cancellation such that the leading @xmath109 corrections cancel .
quenched qcd is not a consistent theory , because omitting the fermion loops causes problems with unitarity .
bardeen @xcite et al .
have shown that there is a problem with the non - singlet @xmath111 correlator in quenched qcd .
the problem can be understood using quenched chiral perturbation theory .
the non - singlet @xmath111 propagator contains an intermediate state of @xmath112 .
the removal of fermion loops in quenched qcd has a big effect on the @xmath113 propagator .
the result is that a ghost state contributes to the scalar correlator , that makes the expression in equation [ eq : fitmodel ] inappropriate to extract masses from the calculation .
bardeen et al .
@xcite predict that the ghost state will make the @xmath114 mass increase as the quark mass is reduced below a certain point .
this behaviour was observed by weingarten and lee @xcite for small box sizes ( l @xmath115 1.6 fm ) .
the negative scalar correlator was also seen by degrand @xcite in a study of point to point correlators .
damgaard et al .
@xcite also discuss the scalar correlator in quenched qcd .
one major problem with quenched qcd is that it does not suppress zero eigenvalues of the fermion operator @xcite .
a quark propagator is the inverse of the fermion operator , so eigenvalues of the fermion operator that are zero , or close to zero , cause problems with the calculation of the quark propagator . in unquenched
qcd , gauge configurations that produce zero eigenvalues in the fermion operator are suppressed by the determinant in the measure .
gauge configurations in quenched qcd that produce an eigenvalue spectrum that cause problems for the computation of the propagator are known as `` exceptional configurations '' .
zero modes of the fermion operator can be caused by topology structures in the gauge configuration .
the problem with exceptional configurations get worse as the quark mass is reduced .
the new class of actions , described in the appendix [ cmn : se : technicaldetails ] that have better chiral symmetry properties , do not have problems with exceptional configurations please note that this section should not be taken as an apology for quenched qcd .
as computers and algorithms get faster the parameters of unquenched lattice qcd are getting `` closer '' to their physical values .
hopefully quenched qcd calculations will fade away .
lattice qcd calculations produce results in units of the lattice spacing .
one experimental number must be used to calculate the lattice spacing from : @xmath116 as the lattice spacing goes to zero any choice of @xmath117 should produce the same lattice spacing this is known scaling .
unfortunately , no calculations are in this regime yet .
the recent unquenched calculations by the milc collaboration @xcite may be close . popular choices to set the scale are the mass of the rho , mass splitting between the s and p wave mesons in charmonium , and a quantity defined from the lattice potential called @xmath118 . the quantity @xmath118
@xcite is defined in terms of the static potential @xmath119 measured on the lattice .
@xmath120 many potential @xcite models predict @xmath121 @xmath122 0.5 fm .
the value of @xmath121 can not be measured experimentally , but is `` easy '' to measure on the lattice .
the value of @xmath121 is a modern generalisation of the string tension . although it may seem a little strange to use @xmath118 to calculate the lattice spacing , when it is not directly known from experiment .
there are problems with all methods to set the lattice spacing .
for example , to set the scale from the mass of the rho meson requires a long extrapolation in light quark mass .
also it is not clear how to deal with the decay width of the rho meson in euclidean space
. the physics results from lattice calculations should be independent of the lattice spacing .
a new lattice spacing is obtained by running at a different value of the coupling in equation [ eq : barecoupling ] . in principle the dependence of quantities on the coupling can be determined from renormalisation group equations : @xmath123 the renormalisation group equations can be solved to give the dependence of the lattice spacing on the coupling @xcite .
@xmath124 the @xmath125 term in equation [ eq : asymp ] prevents any weak coupling expansion converging for masses .
equation [ eq : asymp ] is not often used in lattice qcd calculations .
the bare coupling @xmath126 does not produce very convergent series .
if quantities , such as wilson loops are computed in perturbation theory and from numerical lattice calculations the agreement between the two methods is very poor .
typically couplings defined in terms of more `` physical '' quantities , such as the plaquette are used in lattice perturbative calculations @xcite .
allton @xcite tried to use variants of [ eq : asymp ] with some of the improved couplings .
he also tried to model the effect of the irrelevant operators .
an example of allton s result for the lattice spacing determined from the @xmath23 mass as a function of the inverse coupling is in figure [ eq : rhomass ] . )
@xcite ] there are corrections to equation [ eq : asymp ] from lattice spacing errors .
for example the wilson fermion action in equation [ eq : wilsonfermion ] differs from the continuum dirac action by errors that are @xmath127 .
the lattice spacing corrections can be written in terms of operators in a lagrangian .
these operators are known as irrelevant
. ratios of dimensional quantities are extrapolated to the continuum limit using a simple polynomial .
@xmath128 the improvement program discussed in appendix [ cmn : se : technicaldetails ] designs fermion actions to reduce the lattice spacing dependence of ratios of dimensional quantities . when computationally feasible , calculations are done at ( least ) three lattice spacings and the results
are extrapolated to the continuum .
this was the strategy of the large scale gf11 @xcite and cp - pacs calculations @xcite .
it is very important to know the exact functional form of the lattice spacing dependence of the lattice results for a reliable extrapolation to the continuum .
the quantum field theory nature of the field renormalises the correction polynomial .
there are potentially @xmath129 errors .
this is particularly a problem for states with a mass that is comparable to the inverse lattice spacing .
there is physical argument that the compton wavelength of a hadron should be greater than the lattice spacing .
this is a problem for heavy quarks such as charm and bottom , that is usually solved by the use of effective field theories @xcite .
the masses of the excited light hadrons are large relative to the inverse lattice spacing , so there may be problems with the continuum extrapolations . there have been a number of cases where problems with continuum extrapolations have been found .
for example morningstar and peardon @xcite found that the mass of the @xmath111 glueball had a very strange dependence on the lattice spacing .
morningstar and peardon @xcite had to modify the gauge action to obtain results that allowed a controlled continuum extrapolation .
the alpha collaboration @xcite discuss the problems of extrapolating the renormalisation constant associated with the operator corresponding to the average momentum of non - singlet parton densities to the continuum limit , when the exact lattice spacing dependence is not known . in principle
the formalism of lattice gauge theory does not put a restriction on the size of the lattice spacing used .
the computational costs of lattice calculations are much lower at larger lattice spacings ( see equation [ eq : manyflops ] )
. however , there may be a minimum lattice spacing , set by the length scale of the important physics , under which the lattice calculations become unreliable @xcite .
there has been a lot of work to validate the instanton liquid model on the lattice @xcite .
instantons are semi - classical objects in the gauge fields .
the instanton liquid model models the gauge dynamics with a collection of instantons of different sizes .
some lattice studies claim to have determined that there is a peak in the distribution of the size of the instantons between 0.2 to 0.3 fm in quenched qcd @xcite .
if the above estimates are correct , then lattice spacings of at least 0.2 fm would be required to correctly include the dynamics of the instanton liquid on the lattice . however , determining the instanton content of a gauge configuration is non - trivial , so estimates of size distributions are controversial .
at least one group claims to see evidence in gauge configurations against the instanton liquid model @xcite .
there have been a few calculations of the hadron spectrum on lattices with lattice spacings as coarse as 0.4 fm @xcite .
there were no problems reported in these coarse lattice calculations , however only a few quantities were calculated .
another complication for unquenched calculations is that the lattice spacing depends on the value of the sea quark mass @xcite . as well as the coupling .
this is shown in figure [ cmn : eq : r0withkappa ] .
the dependence of the lattice spacing on the sea quark mass is a complication , because for example the physical box size now depends on the quark mass .
some groups prefer to tune the input parameters in their calculations so that the lattice spacing @xcite is fixed with varying sea quark mass .
other groups prefer to work with a fixed bare coupling @xcite . from @xcite .
] the cost of lattice qcd calculations ( see equation [ eq : manyflops ] ) forces the calculations to be done at unphysical large quark masses .
the results of lattice calculations are extrapolated to physical quark masses using some functional form for the quark mass dependence .
the dependence on the light quark masses is motivated from effective field theories such as chiral perturbation theory .
the modern view is that the lattice qcd calculations do not need to be done at exactly the masses of the physical up and down quarks , but the results can be matched onto an effective theory @xcite . in this section
i briefly discuss the chiral extrapolation fit models used to extrapolate hadron masses from lattice qcd data to the physical points .
the relationship between effective theories and lattice qcd is symbiotic , because the results from lattice qcd calculations can also test effective field theory methods .
the extrapolation of lattice qcd data with light quark masses is currently a `` hot '' and controversial topic in the lattice qcd community .
the controversy is over the different ways to improve the convergence of the effective theory .
there was a discussion panel on the different perspectives on using effective field theories to analyse lattice qcd data , at the lattice 2002 conference @xcite . in section [
se : params ] , i show that the masses of the quarks used in lattice calculations are getting lighter , that will help alleviate many problems . in the past chiral perturbation theory provided functional forms that could used to analyse the quark mass dependence of lattice data . in practise , the extrapolations of lattice data were mostly done assuming a linear dependence on the quark mass @xcite , the mass dependence from effective field theory was essentially treated like a black box . however , a number of high profile studies @xcite have shown that a deeper understanding of the physics behind the expressions from chiral perturbation theory is required .
the subject of effective lagrangians for hadronic physics is very large .
the reviews by georgi @xcite , lepage @xcite and kaplan @xcite contain introductions to the physical ideas behind effective field theory calculations .
scherer @xcite reviews chiral perturbation theory for mesons and baryons .
the chiral perturbation theory for the lowest pseudoscalar particles is the perhaps the most `` well defined '' theory .
the effective field theories for the vectors and baryon particles are perhaps more subtle with concerns about convergence of the expansion . in this section
, i will briefly describe effective field theories , starting with the pseudoscalars and working up to nucleons via the vector particles .
i will try to explain some of the physical pictures behind the functional dependence from effective field theories , rather than provide an exhaustive list of expressions to fit to . in georgi
s review @xcite of the basic ideas behind effective theory , he quotes the essential principles behind effective field theory as * a local lagrangian is used .
* there are a finite numbers of parameters that describe the interaction each of dimension @xmath130 . * the coefficients of the interaction terms of dimension @xmath130 is less than or of the order of @xmath131 where @xmath132 for some mass independent @xmath133 .
the above principles allow that physical processes to be calculated to an accuracy of @xmath134 for a process of energy @xmath135 .
the energy scale @xmath16 helps to organise the power counting .
the accuracy of the results can be improved in a systematic way if more parameters are included .
this is one of the main appeals of the formalism .
as the energy of the process approaches @xmath16 , then in the strict effective field theory paradigm , a new effective field theory should be used with new degrees of freedom @xcite .
for example , fermi s theory of weak interactions breaks down as the propagation of the @xmath136 becomes important . for these process
the electroweak theory can be used .
the effective field theory idea can be applied to hadronic physics . indeed historically , the origin of the effective field theory idea was in hadronic physics .
a lagrangian is written down in terms of hadron fields , as hadrons are the more appropriate degrees of freedom at low energies .
the lagrangian is chosen to have the same symmetries as qcd , hence in its domain of validity it should give the same physics results as qcd .
this is known as weinberg s theorem @xcite . for `` large '' momentum scales the quark and gluon degrees of freedom will become evident ,
so the theory based on the chiral lagrangian with hadron fields must break down .
unfortunately , it is not clear what the next effective field theory is beyond the hadronic effective lagrangian , because it is hard to compute anything in low energy qcd .
the lowest order chiral lagrangian for pseudoscalars is @xmath137 @xmath138 @xmath139 the expansion parameter @xcite for the chiral lagrangian of pseudoscalar mesons is @xmath140 the power counting theorem of weinberg @xcite guarantees that the tree level lagrangian in equation [ ew : lowestchiral ] will generate the @xmath141 lagrangian and non - analytic functions .
the next order terms in the lagrangian for the pseudoscalars are @xmath142 the coefficients @xmath143 are independent of the pion mass .
they represent the high momentum behaviour .
the normalisation used in lattice calculations is connected to the one in gasser and leutwyler @xcite by @xmath144 .
for example , one of the terms in the @xmath145 lagrangian is @xmath146 ^ 2\ ] ] it can be convenient to have quarks with different masses in the sea , than those used in the valence correlators . in lattice qcd jargon , this is known as `` partial quenching '' .
these partially quenched theories can provide information about the real unquenched world @xcite .
the chiral lagrangian predicts @xcite the mass of the pion to depend on the sea and valence quark masses @xmath147 \label{eq : pionmodel}\end{aligned}\ ] ] the function @xmath148 is related to the quark mass via @xmath149 where @xmath150 is the pion decay constant . the leading order term in equation [ eq : pionmodel ] represents the pcac relation .
the terms like @xmath151 are known as `` chiral logs '' are caused by one loop diagrams . as
the chiral logs are independent of the gasser - leutwyler coefficients ( @xmath152 , @xmath153 , @xmath154 , @xmath155 ) , they are generic predictions of the chiral perturbation theory .
a major goal of lattice qcd calculations is to detect the presence of the chiral logs .
this will give confidence that the lattice qcd calculations are at masses where chiral perturbation theory is applicable ( see @xcite for a recent review of how close lattice calculations are to this goal ) .
there have been some attempts to use equation [ eq : pionmodel ] to determine the gasser - leutwyler coefficients from lattice qcd @xcite . . where the mass of the sea quark is @xmath156 and the masses of the valence quarks are @xmath48 and @xmath157 .
comparison of the results for the gasser - leutwyler coefficients from lattice qcd ( with two flavors ) @xcite , to non - lattice estimates @xcite .. [ cols="^,^,^",options="header " , ] there has some work on computing electromagnetic polarizabilities @xcite from lattice qcd .
the electric ( @xmath135 ) and magnetic ( @xmath158 ) polarizabilities measure the interaction of a hadron with constant electromagnetic fields . under the electromagnetic field
the mass of the hadron is shifted by @xmath159 .
@xmath160 the @xmath85 and @xmath13 quantities should be computable from qcd .
holstein @xcite reviews the theory and experiments behind the nucleon polarizabilities .
the experimental values for @xmath85 and @xmath13 are extracted from compton scattering experiments ( see @xcite for example3 ) .
a comparison of the results from lattice qcd to models and experiment can be found in two recent papers @xcite .
in contrast to work of duncan et al . the formalism used for electromagnetic polarizations
@xcite uses static electromagnetic fields .
the @xmath108 gauge fields are modified by multiplying @xmath161 where @xmath162 is the euclidean time variable and @xmath135 is the constant electric field . the phase factor in equation
[ cmn : eq : emgauge ] can be linearised .
smit and vink have described how to put a constant magnetic field on the lattice @xcite .
there are speculations that in very strong magnetic fields ( @xmath163 t ) , the proton will become unstable to the decay to neutrons @xcite .
magnetic fields of this intensity may be realised in the universe @xcite .
the original estimate @xcite of the instability of the proton was done in the quark model . in an attempt to remove some of the uncertainty in the hadronic calculation , rubinstein et al .
@xcite used lattice qcd to study the dependence of the masses of the proton and neutron on the magnetic field .
some early lattice calculations included the magnetic fields to look at the magnetic moments of hadrons @xcite .
however , it is best to calculate magnetic moments from form factors @xcite , so electromagnetic fields are no longer used .
electric fields were used in the first ( unsuccessfully ) attempts to compute the electric dipole moment of the neutron @xcite .
the calculation of the neutron electric dipole moment has recently been reformulated @xcite in a way that does not require the use of electromagnetic fields .
the start of the book numerical methods for scientists and engineers by hamming has the immortal phrase : `` the purpose of computing is insight not numbers . '' in the previous sections i have described how lattice qcd is used to compute the masses of hadrons
. this may give the impression that lattice qcd is essentially just a black box that produces the masses of hadrons without any insight into the physical mechanisms or relevant degrees of freedom .
in this section i hope to show that lattice qcd can also help to explain the physical mechanisms behind the hadron mass spectrum .
a very good overview of the type of insight wanted from hadronic physics , that contrasts the hadron spectroscopy approach to the study of confinement with the results from dis type studies , is the paper by capstick et al .
note however for the b physics experimental program , high precision numbers with reliable error bars are required to look for evidence for physics beyond the standard model of particle physics @xcite .
isgur s motivation @xcite for studying the @xmath164 particles is based on trying to understand the important degrees of freedom that describe low energy qcd .
increasingly , lattice qcd calculations are being used to provide insight into the dynamics of qcd . some of isgur s last papers @xcite were devoted to using lattice qcd to validate the quark model picture of hadronic physics . at first sight the model building approach to studying qcd appears to give more insight into the dynamics of qcd .
it is usually quite easy to study the effect of adding new interactions to the model .
there are many models for qcd interactions : quark model @xcite , instanton liquid model @xcite , or bag model @xcite .
the main problem with the model based approach to hadron spectroscopy is that is very difficult to judge whether the assumptions in the models are valid .
the models are only believed when they provide a reasonable description of `` most '' experimental data with a good @xmath49 , but this does not prove that they are correct .
different models of the qcd dynamics can be based on very different physical pictures , but may give equally valid descriptions of experimental data .
for example the physical assumptions behind the bag model seem to be very different to the assumptions behind the instanton liquid model @xcite .
the major advantage of lattice qcd calculations is that the theory can be mutilated in a controllable way , so the physical mechanism underlying a process can be studied .
for example , the question about what mechanism in qcd causes a linearly rising potential for heavy quarks at intermediate distances is not something that can be answered by the quark model .
greensite @xcite reviews the work on studying `` confinement '' using lattice qcd ) . in this section
i will focus on various attempts to explain the value of the mass splitting between the nucleon and delta . according to the quark model @xcite ,
the masses of the nucleon and delta are split by a spin - spin hyperfine term : @xmath165 where the sum runs over the constituent quarks and @xmath166 is the set of su(3 ) unitary spin matrices ( a runs from 1 to 8) . in perturbative
qcd a term of the form [ eq : hfterm ] is naturally generated , but it is not clear whether the equation [ eq : hfterm ] is relevant for light relativistic quarks .
lattice qcd calculations can produce more masses than experiment , so the hyperfine interaction can be tested @xcite .
a pictorial measure of the quark mass dependence of the masses of the octet and decuplet is given by the `` edinburgh '' plot from cp - pacs @xcite is in figure [ cmn : cppacsedplot ] .
the continuous curve is from a quark model by ono @xcite that uses : @xmath167 the agreement between the lattice data and ono s model is reasonable in figure [ cmn : eq : onomodel ] .
the lattice qcd data from cp - pacs @xcite in figure [ cmn : cppacsedplot ] is almost precise enough to show deviations from the model by ono @xcite ( equation [ cmn : eq : onomodel ] ) . in the instanton liquid model @xcite
, the vacuum is made up of a liquid of interacting instantons .
there has been a lot of work on comparing the instanton liquid model against lattice qcd .
the basic idea is to cool the gauge configuration @xcite .
this removes the perturbative part of the gauge field and leaves the classical configurations , that can be compared against the predictions of the instanton liquid model .
the cooling procedure is essentially a way of smoothing the perturbative noise from the configurations .
this perturbative noise presumably has something to do with the one gluon exchange term . in the first study of this on the lattice @xcite ,
the masses of the nucleon , @xmath23 , @xmath168 and @xmath169 were measured in the usual way .
the gauge configurations were then cooled and the simple hadron spectrum was measured again .
the masses for the @xmath23 @xmath169 and nucleon particles were qualitatively the same before and after cooling .
as the cooling does not effect the instantons this suggested that the mass splittings are largely due to instantons .
the mass splitting between the nucleon and @xmath168 was reduced by smoothing .
chu et al .
@xcite claimed that this was due to the problems with extracting the mass of the @xmath168 from the lattice data .
later work on this issue @xcite has not returned to the effect of instantons on the mass splitting between the nucleon and @xmath168 . rosner and shuryak @xcite have shown that simple instanton interactions can give a reasonable representation of some baryon mass splittings . in a series of papers , glozman @xcite and
collaborators have argued that a interaction based on goldstone boson exchange between constituent quarks gives a better description of the mass spectroscopy of baryons , than interactions of the form in equation [ eq : hfterm ] .
this is not the place for a detailed review of the case for and against a interaction based on the exchange of goldstone bosons ( gbe ) . for a critique of gbe
you can look at the paper by isgur @xcite and the review article by capstick and roberts @xcite . here
i will just discuss the evidence for gbe from lattice qcd .
the kentucky group @xcite have introduced valence qcd , a mutilated version of lattice qcd , that omits `` z '' graphs from the formalism .
the aim was to study the foundation of the quark model . in
quenched qcd the sea quark loops are omitted , however there are still higher fock states from intermediate `` z '' states .
figure [ cmn : fig : zgraph ] shows that a quark going backwards in time can be interpreted as meson state .
the lattice version of valence qcd is the wilson fermion action ( equation [ eq : wilsonfermion ] ) with the backwards hopping terms removed .
@xmath170 the kentucky group @xcite studied valence qcd in a lattice calculation at @xmath171 with a volume of @xmath172 , and sample size of 100 .
the kentucky group @xcite also used valence qcd to study form factors and matrix elements , but i just will focus on their results for the hadron spectrum . in figure [ cmn
: fig : valencespectrum ] , i show a comparison of some hadron masses ( from the comment by isgur @xcite ) the main conclusion from figure [ cmn : fig : valencespectrum ] is that the hyperfine splittings seem to have been reduced for light hadrons .
valence qcd still has the physics of gluon exchange , so the near degeneracy of the nucleon and @xmath168 , suggests that the `` z '' graphs are the important mechanism behind the nucleon - delta mass splitting . for heavy hadrons the results from vqcd are not suppressed relative to the those from quenched qcd .
for example , the vqcd prediction for the mass splitting between @xmath173 and @xmath158 is @xmath174 mev , compared to the experimental value of 46 mev .
so the one gluon exchange is important for heavy quarks ( beyond charm ) , but less important for light quarks .
the vqcd calculations have much reduced hadron masses .
for example .
the nucleon mass in vqcd is reduced by 700 mev from quenched qcd .
the reduction of the vector meson mass was 537 mev .
the kentucky group @xcite attributed the reduction in hadron masses to reduction of the dynamical or constituent quark mass due to the omission of z graphs .
isgur @xcite criticised the conclusions from the study of lattice vqcd . in particular , isgur notes that the one boson exchange interactions , based on z graphs , operates between two quarks , so should not effect mesons made from a quark and anti - quark .
so the near degeneracy between of the @xmath23 and @xmath169 masses in figure [ cmn : fig : valencespectrum ] would not be expected from the suppression of z graphs .
also isgur @xcite noted that the hadron spectrum from valence qcd was radically different from experiment ( and quenched qcd ) .
this might cause additional problems , if for example the wave - functions were very different for valence qcd compared to the real world , then this would effect the hyperfine splittings .
the kentucky group did study matrix elements that are related to the wave function of the states , but did not see any problems @xcite .
the current lattice calculations are currently not able to determine the correct mechanism for the mass splitting between the @xmath168 and nucleon .
much of the lattice work has concentrated on validating one particular model , rather than also falsifying competing models .
this type of physics is necessarily qualitative .
for example , in the spirit of theory mutilation , ideally only one piece of physics must be removed at one time .
it is not clear what effect that valence qcd has on the instanton structure of the vacuum .
the modification of the wilson fermion operator in equation [ eq : wilsonfermionvalence ] will also effect the zero mode structure that is of crucial importance to instanton inspired models .
the operator in equation [ eq : wilsonfermionvalence ] does not obey @xmath175 for the wilson operator @xmath16 .
there is a generalisation of equation [ cmn : eq : hermlatt ] in valence qcd .
there has been some work on trying to `` disprove '' the instanton liquid model by studying lattice qcd gauge configurations @xcite .
edwards reviews the work by many groups on this @xcite .
lattice qcd simulations have been used to test other assumptions made in models of the qcd dynamics .
for example , there are some models of hadronic structure that are based on diquarks @xcite .
a critical assumption in diquark models is that two quarks actually do cluster to form a diquark .
this assumption has been tested in a lattice gauge theory calculations by the bielefeld group @xcite , where they found no deeply bound diquark state in landau gauge .
leinweber @xcite has claimed that lattice qcd data on the charge radii of hadrons provides evidence against scalar diquark clustering . as a test of the mit bag model @xcite and skyrme model @xcite ,
density operators from the models were compared against results from quenched qcd .
the large @xmath11 limit of qcd provides much insight into qcd ( see @xcite for a review ) .
teper @xcite and collaborators have studied the glueball masses in the large @xmath11 limit .
lattice qcd calculations can be done with any gauge group .
teper @xcite et al . studied the glueball masses for @xmath11 = 2,3,4,5 .
this allowed them to estimate the size of the corrections to the @xmath176 limit .
as the @xmath177 corrections are small , it was important to control both the finite volume and lattice spacing errors .
in the previous sections i have implicitly assumed that all the hadrons are stable . in the real world , most hadrons are unstable to strong decays . for example the @xmath23 meson has a mass of 770 mev and a decay width of 150 mev .
most lattice practitioners never `` worry '' about the @xmath23 s decay width ( in public at least ) .
the determination of a hadron s mass from experiment is inextricably linked to the determination of the decay width .
this is perhaps most dramatically demonstrated by the problem of missing baryon resonances @xcite .
the quark model predicted more excited baryons @xcite than were actually seen in @xmath178 reactions .
it was claimed that the additional states were not seen because they coupled very weakly to the @xmath178 channel .
the quark model did predict that the missing baryons might be seen in @xmath179 reactions @xcite .
there are experiments at the jefferson lab that are trying to detect these `` missing resonances '' @xcite .
lattice qcd calculations have to be done in euclidean space for convergence of the path integral @xcite .
this implies that the amplitudes and masses from lattice calculations are real .
this makes the study of resonances nontrivial , because decay widths are inherently complex quantities .
this is also a problem for calculations with a finite chemical potential , although there has been some progress in this area @xcite .
i do not discuss the very elegant formalism of lscher @xcite for studying unstable particles in a finite volume .
the formalism and results from the scattering formalism are reviewed by fiebig and markum in this volume @xcite . in this section
, i would like to describe the possible implications for mass determinations from standard correlation functions .
the momentum is quantised on a lattice of length @xmath90 and periodic boundary conditions .
the momentum of mesons can only take values : @xmath180 where @xmath181 is an integer between 0 and @xmath182 . for a typical lattice , @xmath183 and @xmath184 = 2.0 gev ,
so the quantum of momentum is 0.79 gev .
the quantisation of momentum makes the coupling of the state to the scattering states different to that in the continuum .
this `` feature '' has been used to advantage by lscher @xcite in his formalism .
the quantisation of momentum has important consequences for mesons that decay via @xmath185 wave decays such as the @xmath23 meson .
a @xmath23 meson at rest can only decay to two pions with momentum @xmath186 and @xmath187 .
the decay threshold is @xmath188 .
the quantisation of momentum on the lattice does not effect the threshold for @xmath189wave decays , an example of a s - wave decay is the decay of the @xmath111 meson to pairs of mesons .
the quarks in current lattice calculations are almost light enough for the strong decay of the flavour singlet @xmath111 @xcite .
hadron masses are extracted from lattice qcd calculations using two point correlators ( equation [ eq : fitmodel ] ) .
however the use of equation [ eq : fitmodel ] may not be appropriate for hadrons that can decay .
the most naive modification of the lattice qcd formalism caused by the introduction of decay widths is the replacement @xmath190 where @xmath191 is the decay width .
this modifies equation [ eq : fitmodel ] @xmath192 that is used to extract the masses from correlators .
there are a number of problems with equation [ eq : fitmodelmink ] , so a more thorough derivation is required .
i review the work by michael @xcite ( see also the text book by brown @xcite ) on the effect of decays on two point functions .
i will consider two scalar fields ( @xmath193 ) interacting with the interaction @xmath194 the mass of the @xmath169 particle is @xmath195 . to study the implications of particle decay on the two point correlator , consider the renormalisation of the propagator of the @xmath196 particle in euclidean space @xmath197 the effect of the interaction of the @xmath169 particle with the @xmath196 particle renormalises the @xmath196 propagator .
@xmath198 where @xmath53 is the self energy .
masses are extracted from lattice qcd calculations using the time sliced propagator : @xmath199 where @xmath200 if the pole around @xmath201 is neglected then the expression for the spectral density is @xmath202 in the limit @xmath203 @xmath204 in figure [ cmn : plot : rescorr ] , i plot separately the log of the first and second terms in equation [ cmn : eq : corrresmodel ] using the parameters : ( @xmath205 ) .
the breakdown of a pure exponential decay can be seen for the second term .
there is no evidence for the breakdown of the simple exponential fit model in current lattice calculations ( apart from the effects of chiral artifacts in quenched qcd @xcite ) .
this effect may become apparent as the sea quark masses are reduced to where particle decay is energetically allowed .
the maximum entropy approach @xcite to extracting masses from correlators produces an estimate of the spectral density , so in principle could be used to extract the decay widths of particles .
it is not clear to me whether a decay width obtained from an analysis based on maximum entropy method would be physical .
yamazaki and ishizuka @xcite have recently compared the maximum entropy approach to studying unstable particles to the method advocated by lscher @xcite in a model theory .
yamazaki and ishizuka claimed good agreement between the two methods .
another way of looking for `` evidence '' of resonant behaviour is to look for peculiarities in the quark mass dependence of the hadron masses .
one of the first applications of this idea was by degrand who studied the effect of @xmath23 decay on the quark mass dependence of the rho mass @xcite .
the mixing of the @xmath23 correlator with @xmath206 states at the nonzero momentum makes the mass dependence of the @xmath23 correlator more complex than described by the theory in section [ eq : massdepend ] .
the milc collaboration @xcite have recently claimed to see some evidence for the decay of the @xmath114 ( non - singlet @xmath111 ) meson by comparing the mass dependence of the @xmath114 particle in unquenched and quenched qcd .
the milc collaboration @xcite tried to look for the evidence of @xmath23 meson decay by studying the dispersion relation of the rho particle .
the momentum can be injected into particle correlators , so that the hadron masses can be computed at nonzero momentum .
the signal to noise gets worse , as the momentum increases , so typically the dispersion relation is only known for a few values of momentum . on the lattice
the dispersion relation is of the form @xcite @xmath207 the `` improvement '' program described in the appendix , aims to make the lattice dispersion relation closer to the continuum one .
some clever lattice theorists call this computing `` the speed of light '' @xcite . the relationship in equation [ eq : cmn :
dispersioncont ] predicts that the dependence of the hadron mass as a function of momentum .
consider an interpolating operator for the @xmath23 meson : @xmath208 .
if a unit of momentum is injected in the @xmath209 direction , then the @xmath210 operator will couple to the @xmath23 with one unit of moment , as well as two pions with unit momentum .
the two states will mix and the masses will be modified . however , the operators : @xmath211 , @xmath212 do not couple to the two pion states , essentially because @xmath213 interaction is zero for these kinematics . a signal for
rho decay would be a different mass from the @xmath23 s polarised perpendicular and parallel to the momentum of the state .
the milc collaboration did not see this effect , perhaps because of artifacts with the type of fermions used .
the ukqcd collaboration have recently claimed to see evidence for @xmath23 decay via this mechanism @xcite that was consistent with the results from other methods . in this review
i have focused on just computing the masses of hadrons .
the computation of decay widths from first principles is the holy grail of hadron spectroscopy .
there have been very few attempts at the calculation of the decay widths from lattice qcd .
the hadronic coupling constants are computed from matrix elements , rather than trying to fit expressions like equation [ cmn : eq : corrresmodel ] to data .
additional correlators over the standard two point correlators are required to be calculated .
there have been some attempts to compute the rho to two pion coupling constant @xcite .
perhaps the most famous calculation decay widths is the calculation of partial widths for the decay of @xmath111 glueball were calculated by the gf11 group @xcite .
a clarification of the formalism for studying decay widths and mixing in lattice qcd calculations is described in @xcite .
the computation of the light hadron mass spectrum is essential to checking the validity of lattice qcd techniques .
for example , to reliably extract quark masses requires a precise and consistent calculation of the masses of the light mesons .
a precise determination of the light hadron mass spectrum would unequivocally demonstrate that non - perturbative quantities can be extracted from a physically relevant quantum field theory .
the recent large scale quenched lattice qcd calculations of the spectrum of the lightest hadrons from cp - pacs @xcite are the benchmark for future calculations . after nearly twenty years of work the deviations of the predictions of quenched qcd from experiment have been quantified .
the next generation of large scale quenched qcd calculations ( if they are worth doing at all ) will probably use the new fermion operators , such as overlap - dirac operator , to push to lighter quark masses below @xmath214 of 0.4 .
the light quark mass region of quenched qcd is full of pathologies , such as quenched chiral logs , that will be fun to study theoretically , but of limited or no relevance to experiment .
quenched qcd calculations will still be of some value for many interesting quantities where the uncertainty is larger than the inherent error of quenched qcd .
the main challenges in unquenched hadron spectrum calculations is determining the lattice spacing dependence and reducing the size of the quark masses used in the calculations .
the unquenched calculations of the cp - pacs collaboration @xcite have shown that at least three lattice spacings will be needed to obtain high `` quality '' results .
lattice qcd calculations with 2 + 1 flavours of quarks are starting to produce important results @xcite .
the effective field theory community is increasingly doing calculations specifically to analyse data from lattice qcd calculations .
motivated by the `` new nuclear physics '' experimental programs at facilities , such as jefferson lab @xcite , lattice calculations are starting to be used to study interesting particles such as the @xmath3s .
the use of more sophisticated interpolating operators and more advanced statistical techniques ( such as maximum entropy techniques ) may allow some information to be obtained on some of the lowest excited states of hadrons .
the determination of the masses of excited states from lattice qcd would be a big step forward for hadron spectroscopy , if this was indeed possible .
the theory of hadronic physics is the ultimate postmodern playground , as it sometimes seems that the use of a particular model for hadronic physics , from a mutual incompatible set of possibilities , is almost a matter of personal preference @xcite .
one advantage of lattice qcd calculations is that they provide qualitative information about physical mechanisms , that is not directly accessible from experiment .
if as shuryak @xcite claims , the physical picture behind the bag model and the instanton liquid model are different , then only one picture is correct , so one of them must be discarded .
i hope that qualitative lattice qcd calculations can help simplify the theories behind hadronic spectroscopy by ruling out the underlying pictures behind certain classes of models .
the aim of simplifying the theory of hadronic physics is a stated aim of the current experimental program @xcite .
there are number of interesting `` challenges '' for hadron spectrum calculations beyond the critical task of reducing the errors in lattice qcd calculations : * can the resonant nature of the @xmath23 meson be determined from lattice qcd ?
* can improved lattice calculations determine the structure of the roper resonance ( first excited state of the nucleon ) . for over thirty five years , there have been many speculations on the nature of the roper resonance .
can lattice qcd close this issue ? * can the physical mechanism behind the hyperfine splittings in mesons and baryons be determined from lattice qcd ?
* can lattice qcd calculations be used to simplify hadronic physics by ruling out ( or perhaps even validating ) the bag model ?
this work is supported by pparc .
i thank members of the ukqcd and milc collaborations for discussions .
i thank chris michael for reading the manuscript .
to outsiders ( physicists who live in continuum ) , the lattice gauge theory community most seem like a very inward looking bunch .
a large fraction of research in lattice gauge is on improving the methods used in lattice calculations .
improving the methods used in lattice calculations leads to smaller and more believable error bars and hence is a good thing ! currently , the biggest improvements in the methodology of lattice qcd calculations are coming from better lattice representations of the continuum dirac operator .
the `` new '' lattice representations of the dirac operator have either reduced lattice spacing dependence or a better chiral symmetry . the importance of the dependence of the results on the lattice spacing has been stressed through out this review . as the results of lattice calculations are extrapolated to the continuum
, the calculations would be more precise if the lattice spacing dependence of quantities was weak .
the computational cost of reducing the lattice spacing used in lattice qcd calculations from equation [ eq : manyflops ] is very large , hence it is advantageous to use coarser lattice spacings @xcite .
a standard technique from numerical analysis is to use derivatives that are closer approximations to the continuum derivatives .
for example the lattice derivative in equation [ eq : better ] should be more accurate with a larger lattice spacing than derivative in equation [ eq : ok ] .
@xmath215 @xmath216 however in a quantum field theory there are additional complications , such as the operators in equation [ eq : better ] mixing under renormalisation .
there is a formalism due to symanzik @xcite called improvement where new terms are added to the lattice action that cancel @xmath127 terms ( irrelevant operators ) in a way that is consistent with quantum field theory .
the required terms in the improved lagrangian can be simplified by the use of field redefinitions in the path integral @xcite . a very elegant numerical procedure to improve the wilson fermion action has been developed by the alpha collaboration ( see the review @xcite by lscher ) .
the wilson fermion operator in equation [ eq : wilsonfermion ] differs from the continuum lagrangian by @xmath127 terms .
the improvement scheme used in most lattice qcd calculations with wilson fermions is called clover improvement .
the clover term @xcite is added to the wilson fermion operator in equation [ eq : wilsonfermion ] .
@xmath217 where @xmath218 is the lattice field strength tensor .
if the @xmath219 coefficient computed in perturbation theory is used then the errors in the results from the lattice calculation are @xmath220 .
the alpha collaboration @xcite have computed @xmath219 to all orders in @xmath221 using a numerical technique .
the result for @xmath219 from alpha is @xmath222 for @xmath223 , where @xmath0 is the coupling .
the estimate of @xmath219 , by alpha collaboration , agrees with the one loop perturbation theory for @xmath224 .
the clover improvement program for wilson fermions has had many practical successes .
unfortunately , it is computationally very costly to reach light quark masses in quenched or unquenched lattice calculations that use the clover fermion operator @xcite .
hence , attention has focused on also improving the eigenvalue spectrum of the lattice representation of the dirac operator .
the design of fermion operators on the lattice has a deep connection with chiral symmetry and the global chiral anomaly .
the theoretical complications with transcribing the dirac operator to the lattice are reviewed in many places @xcite .
our understanding of chiral symmetry on the lattice has recently increased by the rediscovery of the ginsparg - wilson relation @xcite @xmath225 where @xmath16 is the fermion operator in equation [ eq : faction ] at zero mass .
equation [ eq : gweqn ] smoothly matches onto the chiral symmetry equation in the continuum as the lattice spacing is taken to zero .
lattice fermion operators that obey the ginsparg - wilson relation ( equation [ eq : gweqn ] ) have a form of lattice chiral symmetry @xcite .
explicit solutions , such as overlap - dirac @xcite or perfect actions @xcite , to equation [ eq : gweqn ] are known .
actions that obey the ginsparg - wilson relation are increasingly being used for quenched qcd calculations @xcite .
domain wall actions , that can loosely be thought of as being approximate solutions to the ginsparg - wilson relation , have been used in calculations @xcite of the matrix elements for the @xmath226 .
the main downside of fermion operators that obey the ginsparg - wilson relation is that they are computationally expensive . in a review of the literature ,
jansen argues @xcite that overlap - dirac type operators are roughly 100 times more expensive computationally than calculations with standard wilson fermions .
the development of new algorithms should reduce this difference in computational cost @xcite .
the various versions of the overlap - dirac operator are cheap enough to use for quenched calculations .
i speculate that there will be an increasing trend to use overlap - dirac type operators for quenched calculations .
it will be some time before unquenched calculations , that are useful for phenomenology , are performed with overlap - dirac operators .
unquenched calculations of qcd with domain wall fermions have just started @xcite .
a more pragmatic development in the design of light fermion actions is the development of improved staggered fermion actions @xcite .
this class of action is being used for unquenched lattice qcd calculations with very light quarks ( see table [ tb : dynamicalparams ] ) by the milc collaboration .
the problem with standard kogut - susskind quarks was that the formalism broke flavour symmetry .
so numerical calculations usually had fifteen pions split by a considerable amount from the goldstone boson pion .
the new variants of fermion operators in the staggered formulations have much reduced flavour symmetry breaking .
the improved staggered quark formalism is quite ugly compared to actions that are solutions of the ginsparg - wilson relation , but lattice qcd is a pragmatic subject and utility wins out over beauty .
it is not understood why calculations using improved staggered quarks are much faster @xcite than calculations using wilson fermions @xcite .
d. weingarten , phys .
b109 * , 57 ( 1982 ) , h. hamber and g. parisi , phys .
lett . * 47 * , 1792 ( 1981 ) , f. butler , h. chen , j. sexton , a. vaccarino , and d. weingarten , phys .
lett . * 70 * , 2849 ( 1993 ) , hep - lat/9212031 , f. butler , h. chen , j. sexton , a. vaccarino , and d. weingarten , nucl .
b430 * , 179 ( 1994 ) , hep - lat/9405003 , m. beneke , ( 2002 ) , hep - lat/0201011 , particle data group , k. hagiwara _ et al .
_ , phys . rev . *
d66 * , 010001 ( 2002 ) , h. wittig , ( 2002 ) , hep - lat/0210025 , c. michael , ( 2003 ) , hep - lat/0302001 , c. michael , phys .
scripta * t99 * , 7 ( 2002 ) , hep - lat/0111056 , c. davies , the heavy hadron spectrum , 1997 , hep - ph/9710394 . c. mcneile , ( 2002 ) , hep - lat/0210026 , clas , p. rossi , ( 2003 ) , hep - ex/0302032 , v. d. burkert , ( 2002 ) , hep - ph/0210321 , v. d. burkert , ( 2001 ) , hep - ph/0106143 , i. montvay and g. munster , , cambridge , uk : univ .
( 1994 ) ( cambridge monographs on mathematical physics ) .
a. s. kronfeld , ( 2002 ) , hep - lat/0205021 , c. t. h. davies _
et al . _ , ( 1998 ) , hep - lat/9801024 , t. degrand , c. detar , r. sugar , and d. toussaint , ( 1998 ) , hep - lat/9811023 , m. mueller - preussker _ et al . _ , ( 2002 ) , hep - lat/0203004 , g. s. bali , ( 2003 ) , nucl - th/0302039 , r. g. edwards , ( 2002 ) , hep - lat/0210027 , s. gottlieb , nucl .
phys . proc
* 53 * , 155 ( 1997 ) , hep - lat/9608107 , r. d. mawhinney , nucl .
. suppl . *
83 * , 57 ( 2000 ) , hep - lat/0001032 , s. aoki , nucl .
proc . suppl . *
94 * , 3 ( 2001 ) , hep - lat/0011074 , d. toussaint , nucl .
106 * , 111 ( 2002 ) , hep - lat/0110010 , t. kaneko , ( 2001 ) , hep - lat/0111005 , m. di pierro , ( 2000 ) , hep - lat/0009001 , m. di pierro , ( 1998 ) , hep - lat/9811036 , m. di pierro , ( 2000 ) , hep - lat/0004007 , j. glimm and a. jaffe , new york , usa : springer ( 1987 ) 535p .
j. e. mandula , g. zweig , and j. govaerts , nucl . phys .
* b228 * , 91 ( 1983 ) , j. e. mandula and e. shpiz , nucl . phys . * b232 * , 180 ( 1984 ) , g. p. lepage and p. b. mackenzie , phys . rev .
* d48 * , 2250 ( 1993 ) , hep - lat/9209022 , ukqcd , d. s. henty and r. d. kenway , phys .
b289 * , 109 ( 1992 ) , hep - lat/9206009 , s. elitzur , phys . rev .
* d12 * , 3978 ( 1975 ) , b. velikson and d. weingarten , nucl . phys .
* b249 * , 433 ( 1985 ) , s. gottlieb , presented at conf .
advances in lattice gauge theory , tallahassee , fl , apr 10 - 13 , 1985 . a. duncan , e. eichten , and h. thacker , phys . lett . *
b303 * , 109 ( 1993 ) , m. w. hecht _
_ , phys . rev .
* d47 * , 285 ( 1993 ) , hep - lat/9208005 , m. w. hecht and t. a. degrand , phys . rev .
* d46 * , 2155 ( 1992 ) , ukqcd , p. lacock , a. mckerrell , c. michael , i. m. stopher , and p. w. stephenson , phys . rev . *
d51 * , 6403 ( 1995 ) , hep - lat/9412079 , a. billoire , e. marinari , and g. parisi , phys .
b162 * , 160 ( 1985 ) , ukqcd , c. r. allton _
_ , phys . rev . *
d47 * , 5128 ( 1993 ) , hep - lat/9303009 , j. w. negele and h. orland , redwood city , usa : addison - wesley ( 1988 ) 459 p. ( frontiers in physics , 68 ) .
a. duncan , s. pernice , and j. yoo , phys . rev . *
d65 * , 094509 ( 2002 ) , hep - lat/0112036 , j. w. negele , nucl .
suppl . * 73 * , 92 ( 1999 ) , hep - lat/9810053 , ukqcd , c. mcneile , data storage issues in lattice qcd calculations , 2000 , hep - lat/0003009 .
ukqcd , c. t. h. davies , a. c. irving , r. d. kenway , and c. m. maynard , ( 2002 ) , hep - lat/0209121 , p. maris and c. d. roberts , ( 2003 ) , nucl - th/0301049 , c. d. roberts and a. g. williams , prog .
* 33 * , 477 ( 1994 ) , hep - ph/9403224 , a. frommer , nucl .
* 53 * , 120 ( 1997 ) , hep - lat/9608074 , g. m. de divitiis , r. frezzotti , m. masetti , and r. petronzio , phys
* b382 * , 393 ( 1996 ) , hep - lat/9603020 , ukqcd , c. michael and j. peisa , phys . rev . * d58
* , 034506 ( 1998 ) , hep - lat/9802015 , a. duncan and e. eichten , phys . rev .
* d65 * , 114502 ( 2002 ) , hep - lat/0112028 , m. c. chu , j. m. grandy , s. huang , and j. w. negele , phys . rev . * d48 * , 3340 ( 1993 ) , hep - lat/9306002 , ukqcd , s. j. hands , p. w. stephenson , and a. mckerrell , phys . rev . *
d51 * , 6394 ( 1995 ) , hep - lat/9412065 , t. degrand , phys . rev . *
d64 * , 094508 ( 2001 ) , hep - lat/0106001 , e. v. shuryak , rev .
65 * , 1 ( 1993 ) , g. p. lepage _
et al . _ , nucl .
* 106 * , 12 ( 2002 ) , hep - lat/0110175 , c. michael , phys .
rev . * d49 * ,
2616 ( 1994 ) , hep - lat/9310026 , c. michael and a. mckerrell , phys . rev . * d51 * , 3745 ( 1995 ) , hep - lat/9412087 , c. michael and m. teper , nucl .
b314 * , 347 ( 1989 ) , m. luscher and u. wolff , nucl . phys . *
b339 * , 222 ( 1990 ) , ukqcd , c. mcneile and c. michael , phys . rev . *
d63 * , 114503 ( 2001 ) , hep - lat/0010019 , t. draper and c. mcneile , nucl .
. suppl . * 34 * , 453 ( 1994 ) , hep - lat/9401013 , g. p. lepage , nucl .
* 26 * , 45 ( 1992 ) , t. a. degrand and m. w. hecht , phys . rev . *
d46 * , 3937 ( 1992 ) , hep - lat/9206011 , ukqcd , c. r. allton _
_ , phys . rev . *
d65 * , 054502 ( 2002 ) , hep - lat/0107021 , m. a. shifman , world sci .
notes phys .
* 62 * , 1 ( 1999 ) , d. b. leinweber , phys
. rev . * d51
* , 6369 ( 1995 ) , nucl - th/9405002 , c. allton and s. capitani , nucl . phys . *
b526 * , 463 ( 1998 ) , hep - lat/9712006 , c. allton , d. blythe , and j. clowser , nucl .
. suppl . * 109 * , 192 ( 2002 ) , hep - lat/0202024 , a. bochkarev and p. de forcrand , nucl . phys . *
b477 * , 489 ( 1996 ) , hep - lat/9505025 , cp - pacs , t. yamazaki _ et al .
_ , phys . rev . *
d65 * , 014501 ( 2002 ) , hep - lat/0105030 .
y. nakahara , m. asakawa , and t. hatsuda , phys . rev . *
d60 * , 091503 ( 1999 ) , hep - lat/9905034 , txl , t. lippert , nucl .
. suppl . * 106 * , 193 ( 2002 ) , hep - lat/0203009 , s. gottlieb , nucl .
* 106 * , 189 ( 2002 ) , hep - lat/0112039 , c. w. bernard _
et al . _ ,
phys . rev .
* d48 * , 4419 ( 1993 ) , hep - lat/9305023 , j .- w .
chen , phys .
* b543 * , 183 ( 2002 ) , hep - lat/0205014 , w. bardeen , a. duncan , e. eichten , n. isgur , and h. thacker , phys . rev . * d65 * , 014509 ( 2002 ) , hep - lat/0106008 , w .- j .
lee and d. weingarten , phys . rev . * d61 * , 014015 ( 2000 ) , hep - lat/9910008 , p.
h. damgaard , m. c. diamantini , p. hernandez , and k. jansen , nucl . phys . *
b629 * , 445 ( 2002 ) , hep - lat/0112016 , w. bardeen , a. duncan , e. eichten , g. hockney , and h. thacker , phys . rev . * d57 * , 1633 ( 1998 ) , hep - lat/9705008 , c. w. bernard _
et al . _ ,
phys . rev . * d64 * , 054506 ( 2001 ) , hep - lat/0104002 , milc , c. w. bernard _
et al . _ ,
phys . rev . * d61 * , 111502 ( 2000 ) , hep - lat/9912018 , hpqcd , a. gray _
et al . _ , ( 2002 ) .
r. sommer , nucl .
* b411 * , 839 ( 1994 ) , hep - lat/9310022 , g. s. bali , phys . rept . * 343 * , 1 ( 2001 ) , hep - ph/0001312 , m. creutz , l. jacobs , and c. rebbi , phys . rept . *
95 * , 201 ( 1983 ) , c. r. allton , ( 1996 ) , hep - lat/9610016 , cp - pacs , s. aoki _
et al . _ ,
phys . rev . * d67
* , 034503 ( 2003 ) , hep - lat/0206009 , c. j. morningstar and m. j. peardon , phys . rev .
* d60 * , 034509 ( 1999 ) , hep - lat/9901004 , c. morningstar and m. j. peardon , nucl .
. suppl . * 83 * , 887 ( 2000 ) , hep - lat/9911003 , m. guagnelli , k. jansen , and r. petronzio , phys .
* b457 * , 153 ( 1999 ) , hep - lat/9901016 , t. degrand and a. hasenfratz , phys . rev . * d64 * , 034512 ( 2001 ) , hep - lat/0012021 , t. schafer and e. v. shuryak , rev .
phys . * 70 * , 323 ( 1998 ) , hep - ph/9610451 , m. teper , nucl .
suppl . * 83 * , 146 ( 2000 ) , hep - lat/9909124 , a. hasenfratz and c. nieter , phys . lett . *
b439 * , 366 ( 1998 ) , hep - lat/9806026 . a. hasenfratz , phys . lett . *
b476 * , 188 ( 2000 ) , hep - lat/9912053 , i. horvath , n. isgur , j. mccune , and h. b. thacker , phys . rev .
* d65 * , 014502 ( 2002 ) , hep - lat/0102003 , i. horvath _ et al .
_ , phys . rev . *
d66 * , 034501 ( 2002 ) , hep - lat/0201008 , m. g. alford , w. dimm , g. p. lepage , g. hockney , and p. b. mackenzie , phys .
* b361 * , 87 ( 1995 ) , hep - lat/9507010 , m. g. alford , t. r. klassen , and g. p. lepage , phys . rev . * d58
* , 034503 ( 1998 ) , hep - lat/9712005 , m. g. alford and r. l. jaffe , nucl . phys . *
b578 * , 367 ( 2000 ) , hep - lat/0001023 , milc , t. degrand , phys . rev . * d58 * , 094503 ( 1998 ) , hep - lat/9802012 , ukqcd , c. r. allton _
et al . _ ,
phys . rev . * d60 * , 034507 ( 1999 ) , hep - lat/9808016 , ukqcd , a. c. irving _
et al . _ ,
phys . rev . *
d58 * , 114504 ( 1998 ) , hep - lat/9807015 , s. r. sharpe , ( 1998 ) , hep - lat/9811006 , c. bernard _
et al . _ , ( 2002 ) , hep - lat/0209086 , w. detmold , w. melnitchouk , j. w. negele , d. b. renner , and a. w. thomas , phys .
lett . * 87 * , 172001 ( 2001 ) , hep - lat/0103006 , h. georgi , ann .
nucl . part .
sci . * 43 * , 209 ( 1993 ) , g. p. lepage , ( 1997 ) , nucl - th/9706029 , d. b. kaplan , ( 1995 ) , nucl - th/9506035 , s. scherer , ( 2002 ) , hep - ph/0210398 , s. weinberg , physica * a96 * , 327 ( 1979 ) , a. manohar and h. georgi , nucl . phys . *
b234 * , 189 ( 1984 ) , j. gasser and h. leutwyler , nucl . phys . *
b250 * , 465 ( 1985 ) , c. w. bernard and m. f. l. golterman , phys . rev .
* d49 * , 486 ( 1994 ) , hep - lat/9306005 , s. r. sharpe and n. shoresh , phys . rev .
* d62 * , 094503 ( 2000 ) , hep - lat/0006017 , s. r. sharpe , phys . rev . *
d56 * , 7052 ( 1997 ) , hep - lat/9707018 , alpha , j. heitger , r. sommer , and h. wittig , nucl . phys . *
b588 * , 377 ( 2000 ) , hep - lat/0006026 , w. bardeen , a. duncan , e. eichten , and h. thacker , phys . rev . * d62
* , 114505 ( 2000 ) , hep - lat/0007010 , ukqcd , a. c. irving , c. mcneile , c. michael , k. j. sharkey , and h. wittig , phys . lett .
* b518 * , 243 ( 2001 ) , hep - lat/0107023 , d. r. nelson , g. t. fleming , and g. w. kilcup , phys .
lett . * 90 * , 021601 ( 2003 ) , hep - lat/0112029 , ukqcd , a. c. irving , ( 2002 ) , hep - lat/0208065 .
j. bijnens , g. ecker , and j. gasser , ( 1994 ) , hep - ph/9411232 , g. ecker , j. gasser , a. pich , and e. de rafael , nucl . phys . *
b321 * , 311 ( 1989 ) , d. b. kaplan and a. v. manohar , phys .
rev . lett . * 56 * , 2004 ( 1986 ) , cp - pacs , a. ali khan _ et al .
_ , phys . rev . *
d65 * , 054505 ( 2002 ) , hep - lat/0105015 , n. h. fuchs , h. sazdjian , and j. stern , phys . lett . *
b269 * , 183 ( 1991 ) , m. knecht and j. stern , ( 1994 ) , hep - ph/9411253 , l. giusti , f. rapuano , m. talevi , and a. vladikas , nucl . phys . *
b538 * , 249 ( 1999 ) , hep - lat/9807014 , p. hernandez , k. jansen , and l. lellouch , phys . lett . *
b469 * , 198 ( 1999 ) , hep - lat/9907022 , g. ecker , ( 1998 ) , hep - ph/9805500 , t. morozumi , a. i. sanda , and a. soni , phys . rev .
* d46 * , 2240 ( 1992 ) , a. morel , j. phys .
( france ) * 48 * , 1111 ( 1987 ) , s. r. sharpe , phys . rev . *
d41 * , 3233 ( 1990 ) , c. w. bernard and m. f. l. golterman , phys . rev .
* d46 * , 853 ( 1992 ) , hep - lat/9204007 , m. golterman , ( 1997 ) , hep - ph/9710468 , g. rupak and n. shoresh , ( 2002 ) , hep - lat/0201019 , s. r. sharpe and j. singleton , robert , phys . rev . * d58
* , 074501 ( 1998 ) , hep - lat/9804028 , qq+q , f. farchioni , c. gebert , i. montvay , e. scholz , and l. scorzato , phys . lett . *
b561 * , 102 ( 2003 ) , hep - lat/0302011 , u. g. meissner , phys . rept . * 161 * , 213 ( 1988 ) , g. ecker , j. gasser , h. leutwyler , a. pich , and e. de rafael , phys .
b223 * , 425 ( 1989 ) , j. bijnens , p. gosdzinsky , and p. talavera , nucl . phys . *
b501 * , 495 ( 1997 ) , hep - ph/9704212 , e. jenkins , a. v. manohar , and m. b. wise , phys .
lett . * 75 * , 2272 ( 1995 ) , hep - ph/9506356 , m. booth , g. chiladze , and a. f. falk , phys . rev . *
d55 * , 3092 ( 1997 ) , hep - ph/9610532 , c .- k . chow and
rey , nucl .
b528 * , 303 ( 1998 ) , hep - ph/9708432 , t. a. degrand , phys . rev . *
d43 * , 2296 ( 1991 ) , d. b. leinweber and t. d. cohen , phys . rev . *
d49 * , 3512 ( 1994 ) , hep - ph/9307261 , p. geiger and n. isgur , phys . rev . *
d41 * , 1595 ( 1990 ) , d. b. leinweber , a. w. thomas , k. tsushima , and s. v. wright , phys . rev . *
d64 * , 094502 ( 2001 ) , hep - lat/0104013 , a. w. thomas , ( 2002 ) , hep - lat/0208023 , d. b. leinweber , a. w. thomas , k. tsushima , and s. v. wright , phys . rev . *
d61 * , 074502 ( 2000 ) , hep - lat/9906027 , e. jenkins and a. v. manohar , phys . lett . *
b259 * , 353 ( 1991 ) , v. bernard , n. kaiser , j. kambor , and u. g. meissner , nucl . phys . *
b388 * , 315 ( 1992 ) , b. borasoy and u .- g .
meissner , annals phys . * 254 * , 192 ( 1997 ) , hep - ph/9607432 , j. f. donoghue , b. r. holstein , and b. borasoy , phys . rev . * d59 * , 036002 ( 1999 ) , hep - ph/9804281 , d. b. kaplan , m. j. savage , and m. b. wise , nucl .
b478 * , 629 ( 1996 ) , nucl - th/9605002 , e. jenkins , nucl . phys . *
b368 * , 190 ( 1992 ) , a. w. thomas , d. b. leinweber , k. tsushima , and s. v. wright , nucl . phys . * a663 * , 973 ( 2000 ) , nucl - th/9909041 , u .- g .
meissner , ( 1998 ) , hep - ph/9810276 , r. d. young , d. b. leinweber , and a. w. thomas , ( 2002 ) , hep - lat/0212031 , r. lewis and p .-
a. ouimet , phys . rev . *
d64 * , 034005 ( 2001 ) , hep - ph/0010043 , b. borasoy , r. lewis , and p .-
a. ouimet , phys . rev . *
d65 * , 114023 ( 2002 ) , hep - ph/0203199 , t. becher and h. leutwyler , eur .
j. * c9 * , 643 ( 1999 ) , hep - ph/9901384 , a. r. hoch and r. r. horgan , nucl .
b380 * , 337 ( 1992 ) , c. r. allton , v. gimenez , l. giusti , and f. rapuano , nucl
b489 * , 427 ( 1997 ) , hep - lat/9611021 , ukqcd , p. lacock and c. michael , phys . rev . * d52
* , 5213 ( 1995 ) , hep - lat/9506009 , l. maiani and g. martinelli , phys . lett . *
b178 * , 265 ( 1986 ) , k. c. bowler _
et al . _ ,
b162 * , 354 ( 1985 ) , ape , p. bacilieri _ et al .
_ , nucl . phys . *
b317 * , 509 ( 1989 ) , m. fukugita , h. mino , m. okawa , and a. ukawa , phys .
lett . * 68 * , 761 ( 1992 ) , b. sakita , world sci .
notes phys .
* 1 * , 1 ( 1985 ) , m. luscher , phys
b118 * , 391 ( 1982 ) , m. luscher , nucl .
b219 * , 233 ( 1983 ) , alpha , j. garden , j. heitger , r. sommer , and h. wittig , nucl . phys . *
b571 * , 237 ( 2000 ) , hep - lat/9906013 , j. gasser and h. leutwyler , phys . lett . *
b184 * , 83 ( 1987 ) , g. colangelo , s. durr , and r. sommer , ( 2002 ) , hep - lat/0209110 , qcdsf , a. ali khan _ et al .
_ , ( 2002 ) , hep - lat/0209111 , s. antonelli _ et al .
_ , phys . lett . *
b345 * , 49 ( 1995 ) , hep - lat/9405012 , i. montvay , rev .
phys . * 59 * , 263 ( 1987 ) , b. orth _
et al . _ ,
. suppl . * 106 * , 269 ( 2002 ) , hep - lat/0110158 , m. luscher , ( 1998 ) , hep - lat/9802029 , j. d. bjorken , presented at the slac summer institute on particle physics , stanford , calif . ,
jul 9 - 20 , 1979 .
p. van baal , ( 2000 ) , hep - ph/0008206 , ukqcd , k. c. bowler _
et al . _ ,
phys . rev .
* d62 * , 054506 ( 2000 ) , hep - lat/9910022 , f. james and m. roos , comput .
. commun . * 10 * , 343 ( 1975 ) , ukqcd , c. r. allton _
et al . _ ,
b284 * , 377 ( 1992 ) , hep - lat/9205016 , alpha , m. guagnelli , r. sommer , and h. wittig , nucl . phys . *
b535 * , 389 ( 1998 ) , hep - lat/9806005 , cp - pacs , s. aoki _
et al . _ ,
lett . * 84 * , 238 ( 2000 ) , hep - lat/9904012 , bgr , c. gattringer _
et al . _ , ( 2003 ) , hep - lat/0307013 , m. della morte , r. frezzotti , and j. heitger , nucl .
. suppl . * 106 * , 260 ( 2002 ) , hep - lat/0110166 , m. gockeler _ et al .
_ , phys . rev . *
d57 * , 5562 ( 1998 ) , hep - lat/9707021 , ukqcd , c. mcneile and c. michael , nucl .
. suppl . * 106 * , 251 ( 2002 ) , hep - lat/0110108 , p. hasenfratz , s. hauswirth , t. jorg , f. niedermayer , and k. holland , ( 2002 ) , hep - lat/0205010 , cp - pacs , a. ali khan _ et al .
lett . * 85 * , 4674 ( 2000 ) , hep - lat/0004010 , txl , u. glassner _
et al . _ ,
b383 * , 98 ( 1996 ) , hep - lat/9604014 , txl , g. s. bali _
et al . _ ,
phys . rev . *
d62 * , 054503 ( 2000 ) , hep - lat/0003012 , alpha , k. jansen and r. sommer , nucl . phys . *
b530 * , 185 ( 1998 ) , hep - lat/9803017 , milc , c. w. bernard _
et al . _ ,
phys . rev . * d56 * , 7039 ( 1997 ) , hep - lat/9707008 , v. lubicz , nucl .
. suppl . * 94 * , 116 ( 2001 ) , hep - lat/0012003 , g. martinelli and y .- c .
zhang , phys .
b125 * , 77 ( 1983 ) , a. hasenfratz , p. hasenfratz , z. kunszt , and c. b. lang , phys . lett . *
b117 * , 81 ( 1982 ) , a. patel and r. gupta , phys . lett . *
b131 * , 425 ( 1983 ) , ukqcd , p. lacock , c. michael , p. boyle , and p. rowland , phys . rev . *
d54 * , 6997 ( 1996 ) , hep - lat/9605025 , t. degrand and m. hecht , phys . lett . *
b275 * , 435 ( 1992 ) , h. b. meyer and m. j. teper , nucl
b658 * , 113 ( 2003 ) , hep - lat/0212026 , j. d. weinstein and n. isgur , phys . rev .
* d27 * , 588 ( 1983 ) , ukqcd , a. hart , c. mcneile , and c. michael , ( 2002 ) , hep - lat/0209063 , rbc , s. prelovsek and k. orginos , ( 2002 ) , hep - lat/0209132 , m. m. brisudova , l. burakovsky , and t. goldman , phys . lett . *
b460 * , 1 ( 1999 ) , hep - ph/9810296 , a. j. g. hey and r. l. kelly , phys . rept . * 96 * , 71 ( 1983 ) , j. r. forshaw and d. a. ross , cambridge , uk : univ .
( 1997 ) 248 p. ( cambridge lecture notes in physics .
a. tang and j. w. norbury , phys
d62 * , 016006 ( 2000 ) , hep - ph/0004078 , m. m. brisudova , l. burakovsky , and t. goldman , phys . rev . *
d61 * , 054013 ( 2000 ) , hep - ph/9906293 , c. w. bernard _
et al . _ ,
phys . rev .
* d64 * , 074509 ( 2001 ) , hep - lat/0103012 , g. f. chew and s. c. frautschi , phys . rev . lett .
* 8 * , 41 ( 1962 ) , a. donnachie , econf * c010430 * , t05 ( 2001 ) , hep - ph/0106197 , j. hoek and j. smit , nucl
b263 * , 129 ( 1986 ) , d. b. leinweber , phys . rev . *
d51 * , 6383 ( 1995 ) , nucl - th/9406001 , milc , c. w. bernard _
et al . _ ,
lett . * 81 * , 3087 ( 1998 ) , hep - lat/9805004 , milc , c. bernard , phys . rev . *
d65 * , 054031 ( 2002 ) , hep - lat/0111051 , s. kim and s. ohta , phys . rev . *
d61 * , 074506 ( 2000 ) , hep - lat/9912001 , r. d. mawhinney , nucl .
. suppl . * 47 * , 557 ( 1996 ) , hep - lat/9603019 , qcdsf , m. gockeler _ et al .
. suppl . * 83 * , 203 ( 2000 ) , hep - lat/9909160 , jlqcd , s. aoki _
et al . _ , ( 2002 ) , hep - lat/0212039 , milc , c. w. bernard _
et al . _ ,
. suppl . * 73 * , 198 ( 1999 ) , hep - lat/9810035 , n. isgur and g. karl , phys . rev . * d20
* , 1191 ( 1979 ) , n. isgur , phys . rev . * d60
* , 054013 ( 1999 ) , nucl - th/9901032 , s. capstick and w. roberts , ( 2000 ) , nucl - th/0008028 , n. isgur , ( 2000 ) , nucl - th/0007008 , j. n. labrenz and s. r. sharpe , phys . rev . * d54 * , 4595 ( 1996 ) , hep - lat/9605034 , qcdsf , m. gockeler _ et al .
_ , phys . lett . *
b532 * , 63 ( 2002 ) , hep - lat/0106022 , s. j. dong _
et al . _ , ( 2003 ) , hep - ph/0306199 , ukqcd , c. m. maynard and d. g. richards , ( 2002 ) , hep - lat/0209165 , s. sasaki , t. blum , and s. ohta , phys . rev . * d65 * , 074503 ( 2002 ) , hep - lat/0102010 , d. broemmel _ et al .
_ , ( 2003 ) , hep - ph/0307073 , y. nemoto , n. nakajima , h. matsufuru , and h. suganuma , ( 2003 ) , hep - lat/0302013 , n. isgur and g. karl , phys
. rev . * d19 * , 2653 ( 1979 ) , p. r. page , ( 2002 ) , nucl - th/0204031 , t. barnes , ( 2000 ) , nucl - th/0009011 , w. melnitchouk _ et al .
_ , phys . rev . *
d67 * , 114506 ( 2003 ) , hep - lat/0202022 , s. sasaki , ( 2003 ) , nucl - th/0305014 , m. gockeler _ et al .
_ , nucl . phys . *
b487 * , 313 ( 1997 ) , hep - lat/9605035 , s. kim , j. b. kogut , and m .-
lombardo , phys . rev . *
d65 * , 054015 ( 2002 ) , hep - lat/0112009 , m. burkardt , d. b. leinweber , and x .- m .
jin , phys .
b385 * , 52 ( 1996 ) , hep - ph/9604450 , a. duncan , e. eichten , and h. thacker , phys .
lett . * 76 * , 3894 ( 1996 ) , hep - lat/9602005 , a. duncan , e. eichten , and h. thacker , phys . lett . *
b409 * , 387 ( 1997 ) , hep - lat/9607032 , h. r. fiebig , w. wilcox , and r. m. woloshyn , nucl . phys . *
b324 * , 47 ( 1989 ) , l. zhou , f. x. lee , w. wilcox , and j. christensen , ( 2002 ) , hep - lat/0209128 , j. christensen , f. x. lee , w. wilcox , and l .- m .
zhou , ( 2002 ) , hep - lat/0209043 , b. r. holstein , ( 2000 ) , hep - ph/0010129 , b. e. macgibbon _
_ , phys . rev . *
c52 * , 2097 ( 1995 ) , nucl - ex/9507001 , j. smit and j. c. vink , nucl . phys . *
b286 * , 485 ( 1987 ) , m. bander and h. r. rubinstein , phys . lett . *
b311 * , 187 ( 1993 ) , hep - ph/9204224 , d. grasso and h. r. rubinstein , phys . rept . * 348 * , 163 ( 2001 ) , astro - ph/0009061 , h. r. rubinstein , s. solomon , and t. wittlich , nucl . phys . *
b457 * , 577 ( 1995 ) , hep - lat/9501001 , c. w. bernard , t. draper , k. olynyk , and m. rushton , phys .
lett . * 49 * , 1076 ( 1982 ) , g. martinelli , g. parisi , r. petronzio , and f. rapuano , phys . lett . *
b116 * , 434 ( 1982 ) , t. draper , r. m. woloshyn , w. wilcox , and k .- f .
liu , nucl .
b318 * , 319 ( 1989 ) , t. draper , r. m. woloshyn , and k .- f .
liu , phys .
b234 * , 121 ( 1990 ) , v. gadiyak , x .- d .
ji , and c .- w .
jung , phys . rev . *
d65 * , 094510 ( 2002 ) , hep - lat/0112040 , s. aoki and a. gocksch , phys .
lett . * 63 * , 1125 ( 1989 ) , s. aoki , a. gocksch , a. v. manohar , and s. r. sharpe , phys .
lett . * 65 * , 1092 ( 1990 ) , d. guadagnoli , v. lubicz , g. martinelli , and s. simula , jhep * 04 * , 019 ( 2003 ) , hep - lat/0210044 , s. capstick _ et al .
_ , key issues in hadronic physics , 2000 , hep - ph/0012238
. n. yamada , ( 2002 ) , hep - lat/0210035 , n. isgur , phys .
rev . * d61
* , 118501 ( 2000 ) , hep - lat/9908009 , n. isgur and h. b. thacker , phys . rev .
* d64 * , 094507 ( 2001 ) , hep - lat/0005006 , e. v. shuryak , lect . notes phys . *
583 * , 251 ( 2002 ) , hep - ph/0104249 , a. w. thomas and w. weise , berlin , germany : wiley - vch ( 2001 ) 389 p. j. greensite , ( 2003 ) , hep - lat/0301023 , r. d. loft and t. a. degrand , phys . rev . *
d39 * , 2678 ( 1989 ) , y. iwasaki and t. yoshie , phys . lett . *
b216 * , 387 ( 1989 ) , s. ono , phys . rev . *
d17 * , 888 ( 1978 ) , m. c. chu , j. m. grandy , s. huang , and j. w. negele , phys . rev . *
d49 * , 6039 ( 1994 ) , hep - lat/9312071 , t. g. kovacs , phys . rev . *
d62 * , 034502 ( 2000 ) , hep - lat/9912021 , e. v. shuryak and j. l. rosner , phys . lett . *
b218 * , 72 ( 1989 ) , l. y. glozman , nucl . phys . *
a663 * , 103 ( 2000 ) , hep - ph/9908423 , n. isgur , phys . rev . *
d62 * , 054026 ( 2000 ) , nucl - th/9908028 , k. f. liu _
et al . _ ,
phys . rev . * d59 * , 112001 ( 1999 ) , hep - ph/9806491 , k. f. liu _
et al . _ ,
phys . rev .
* d61 * , 118502 ( 2000 ) , hep - lat/9912049 , r. g. edwards , nucl .
. suppl . * 106 * , 38 ( 2002 ) , hep - lat/0111009 , m. anselmino , e. predazzi , s. ekelin , s. fredriksson , and d. b. lichtenberg , rev .
phys . * 65 * , 1199 ( 1993 ) , m. hess , f. karsch , e. laermann , and i. wetzorke , phys . rev . *
d58 * , 111502 ( 1998 ) , hep - lat/9804023 , d. b. leinweber , phys . rev . *
d47 * , 5096 ( 1993 ) , hep - ph/9302266 , m. lissia , m. c. chu , j. w. negele , and j. m. grandy , nucl . phys . *
a555 * , 272 ( 1993 ) , m. c. chu , m. lissia , and j. w. negele , nucl . phys . *
a570 * , 521 ( 1994 ) , hep - lat/9308012 , j. f. donoghue , e. golowich , and b. r. holstein , cambridge monogr . part .
* 2 * , 1 ( 1992 ) , b. lucini and m. teper , jhep * 06 * , 050 ( 2001 ) , hep - lat/0103027 , r. koniuk and n. isgur , phys .
lett . * 44 * , 845 ( 1980 ) , j. b. kogut , ( 2002 ) , hep - lat/0208077 , m. luscher , nucl . phys . *
b364 * , 237 ( 1991 ) , h. r. fiebig and h. markum , ( 2002 ) , hep - lat/0212037 , ukqcd , a. hart and m. teper , phys .
rev . * d65
* , 034502 ( 2002 ) , hep - lat/0108022 , c. michael , nucl . phys . *
b327 * , 515 ( 1989 ) , l. s. brown , , cambridge , uk : univ .
( 1992 ) .
m. asakawa , t. hatsuda , and y. nakahara , prog .
. phys . * 46 * , 459 ( 2001 ) , hep - lat/0011040 , t. yamazaki and n. ishizuka , phys . rev . * d67
* , 077503 ( 2003 ) , hep - lat/0210022 , t. bhattacharya , r. gupta , g. kilcup , and s. r. sharpe , phys . rev . * d53
* , 6486 ( 1996 ) , hep - lat/9512021 , ukqcd , c. mcneile and c. michael , phys . lett . *
b556 * , 177 ( 2003 ) , hep - lat/0212020 , s. gottlieb , p. b. mackenzie , h. b. thacker , and d. weingarten , phys . lett . *
b134 * , 346 ( 1984 ) , r. l. altmeyer _
et al . _ , z. phys . * c68 * , 443 ( 1995 ) , hep - lat/9504003 , r. d. loft and t. a. degrand , phys . rev . * d39 * , 2692 ( 1989 ) , j. sexton , a. vaccarino , and d. weingarten , phys
lett . * 75 * , 4563 ( 1995 ) , hep - lat/9510022 , a. pickering , , edinburgh , uk : univ . pr .
( 1984 ) .
p. lepage , nucl .
. suppl . *
60a * , 267 ( 1998 ) , hep - lat/9707026 , k. symanzik , nucl . phys . *
b226 * , 187 ( 1983 ) , k. symanzik , nucl . phys . *
b226 * , 205 ( 1983 ) , b. sheikholeslami and r. wohlert , nucl . phys . *
b259 * , 572 ( 1985 ) , m. luscher , s. sint , r. sommer , p. weisz , and u. wolff , nucl . phys . *
b491 * , 323 ( 1997 ) , hep - lat/9609035 , p. h. ginsparg and k. g. wilson , phys . rev . * d25
* , 2649 ( 1982 ) , m. luscher , phys . lett . *
b428 * , 342 ( 1998 ) , hep - lat/9802011 , h. neuberger , ann .
nucl . part .
sci . * 51 * , 23 ( 2001 ) , hep - lat/0101006 , p. hasenfratz , nucl .
b525 * , 401 ( 1998 ) , hep - lat/9802007 , p. hernandez , nucl .
. suppl . * 106 * , 80 ( 2002 ) , hep - lat/0110218 , cp - pacs , j. i. noaki _
et al . _ , ( 2001 ) , hep - lat/0108013 , rbc , t. blum _
_ , ( 2001 ) , hep - lat/0110075 , k. jansen , nucl .
. suppl . * 106 * , 191 ( 2002 ) , hep - lat/0111062 , rbc , t. izubuchi , ( 2002 ) , hep - lat/0210011 , milc , k. orginos and d. toussaint , phys . rev . * d59 * , 014501 ( 1999 ) , hep - lat/9805009 , | i review the results of hadron spectroscopy calculations from lattice qcd for an intended audience of low energy hadronic physicists .
i briefly introduce the ideas of numerical lattice qcd .
the various systematic errors , such as the lattice spacing and volume dependence , in lattice qcd calculations are discussed .
in addition to the discussion of the properties of ground state hadrons , i also review the small amount of work done on the spectroscopy of excited hadrons and the effect of electromagnetic fields on hadron masses .
i also discuss the attempts to understand the physical mechanisms behind hadron mass splittings . |
[ [ section ] ] living cells possess multiple trafficking pathways , which have in common , a regulated flux of cargo molecules , such as proteins and lipids , moving through and processed within , organized compartments or organelles .
for instance , the _ secretory pathway _ consists of molecules that move from the endoplasmic reticulum ( er ) to the cell surface via an organelle system called the golgi apparatus @xcite .
the golgi apparatus itself consists of distinct sub - compartments known as cisternae .
proteins and lipids arrive from the er , enter through one face of the golgi and undergo several chemical reactions ( processing ) ; the modified products then leave through the other face to mainly target the cell surface .
[ [ section-1 ] ] there has been much discussion about what drives the flux of molecules through the golgi @xcite .
the _ vesicular transport model _ envisages that the cisternae are stable structures with fixed enzymatic composition .
molecules shuttle from one cisterna to the next in small sacs called vesicles , and get chemically modified by the resident enzymes .
cisternal maturation model _
, on the other hand , considers the cisternae to be transient structures that are formed by fusion of incoming vesicles . in this model , it is the cisternae that progress through the golgi apparatus , carrying the biomolecules with them .
specific enzymes get attached to a cisterna in different stages of its progression , and modify its contents .
the final cisterna eventually breaks up , releasing processed biomolecules .
independent studies on a variety of cells provide evidence for both these possibilities @xcite .
[ [ section-2 ] ] not only the cargo molecules , but indeed the molecules that form the golgi organelle themselves , must also be trafficked along the same route and by the same driving forces .
this invites the following question : how does the golgi organelle form in the first place , i.e. , how does one obtain stable structures ( cisternae ) given the rules of molecular trafficking , which broadly , may be described as : ( i ) localised injection of ` particles ' , i.e. of the vesicles containing unprocessed biomolecules ( ii ) transformation of particles from one species to the other , i.e. chemical processing of the biomolecules by enzymes ( iii ) transport of particles either by chipping ( breaking off ) of a single particle ( corresponding to vesicle movement ) or through movement of bigger aggregates ( corresponding to cisternal progression ) .
the aim of this paper is to construct a statistical model incorporating these elementary processes , and use this to quantitatively address questions of structure formation and the nature of the states at long times .
[ [ section-3 ] ] with this motivation , we define the following multi - species model . particles of species a are injected into a one - dimensional ( 1d ) lattice at one boundary .
particles of _ different _ species b , c ... ( or more generally , of all types a , b , c etc . )
leave from the other boundary .
this happens by allowing a particles to convert to b particles ( and vice versa ) , b particles to c and so on .
there is no restriction on the number of particles of any species a site can hold .
the hopping of particles from one site to another can either occur collectively via movement of the whole stack or one at a time by chipping of a single particle .
chipping refers to a _
particle breaking off from a stack and hopping to a neighbouring site
. the hopping probability may be the same to the left and right ( diffusive ) or different ( driven diffusive , due to an existing chemical or electrical gradient ) .
when a particle ( or a collection of particles ) hops on to a stack , it merges with the particles already resident on that site .
thus , stacks constantly gain and lose particles .
[ [ section-4 ] ] this is a generalization of a well studied model of aggregation and chipping @xcite .
earlier studies dealt with a closed system , with a single species of particle .
the present generalization deals with an open system , with injection of particles at one end , and interconversion from one species of particle to another .
interestingly , we find that new sorts of phases can arise in some limits .
[ [ section-5 ] ] the parameter space is large .
thus , it is useful to begin with the study of the ` chipping only ' model , where there is no movement of stacks as a whole . in the remainder of this paper
, we will study the model in this limit , with chipping and interconversion rates taken to be constants , independent of the number of a or b or c ... particles on the site . with this assumption of constant rates
, we find that for some rates , the system fails to achieve steady state in the sense that unbounded growth of mass occurs .
interestingly , even in these growing states , the particle currents at each site are stationary ( time independent ) after sufficiently long times .
the indefinitely growing average mass at a given site arises simply because the particle currents , though stationary , are not balanced at that site @xcite .
thus , we call such a state quasi - stationary .
[ [ section-6 ] ] although we have defined the model for an arbitrary number of species , from now on we will focus primarily on the two - species model .
the multi - species model is a simple generalization of the two - species case and shows qualitatively similar behaviour , as discussed in sec .
[ sec:3species ] .
[ [ section-7 ] ] the rest of the paper is organised as follows .
section [ sec : modelandresults ] defines the model precisely , highlights some connections with other models , and briefly discusses the main results of the paper . in sec .
[ sec : first_site ] , we analyse the behaviour of the first site in detail . in sec .
[ sec : asym_lattice ] , we study the case of fully asymmetric hopping for a 1d lattice and solve the current equations at each site . in sec .
[ sec : continuum ] , a continuum limit is taken and results obtained for general asymmetry . in sec . [ sec:3species ] ,
we briefly discuss the generalization of the model to three species .
section [ sec : conclusion ] contains a discussion of some issues and extensions .
the model is defined on a 1d lattice . at any time @xmath0 ,
a lattice site @xmath1 has @xmath2 particles of type @xmath3 and @xmath4 particles of type @xmath5 .
we start with an empty lattice with @xmath6 sites at @xmath7 . at each instant
, a site @xmath1 is chosen at random and one of the following moves ( illustrated in fig .
[ model_schematic ] ) occurs in an infinitesimal time interval @xmath8 : 1 .
injection : if the site picked is site 1 , then an a particle is injected into it from the left boundary with probability @xmath9 : @xmath10 2 .
interconversion : with probability @xmath11 , one of the a particles residing on the selected site converts to type b : if @xmath12 , then @xmath13 and @xmath14 . with probability @xmath15 one of the b particles
residing on the site converts to type a : if @xmath16 then @xmath17 and @xmath18 .
3 . chipping : with probability @xmath19 ( @xmath20 ) , one of the a particles chips off the site and hops to the right ( left ) neighbouring site , thereby decreasing the mass of a on the site by 1 and increasing the mass of a on the neighbouring site by 1 ; if @xmath12 then @xmath13 and @xmath21 .
with probability @xmath22 ( @xmath23 ) , one of the b particles chips off the site and hops to the right ( left ) : if @xmath16 then @xmath17 , @xmath24 4 .
stack movement : with probability @xmath25 ( or @xmath26 ) , the entire mass ( i.e. all a and b particles together ) present at the site hops to the right ( left ) neighbouring site : @xmath27 , @xmath28 , @xmath29 , @xmath30 . at the last site
an a ( b ) particle can hop out of the system at a rate @xmath31 , i.e. with the same hopping rate as in the bulk .
once particles hop rightwards from the last site @xmath6 , they can not return to the system .
thus , site @xmath32 acts as a sink .
[ [ section-8 ] ] in general , the drive i.e. the asymmetry in the rates of the three hopping processes hopping of stacks , hopping of a particles after chipping and hopping of b particles after chipping , can be different .
however , for simplicity , we will take the asymmetry to be the same for all three processes , i.e. @xmath33
. then we can parametrize the rates as follows : @xmath34 , @xmath35 ; @xmath36 , @xmath37 ; @xmath38 , @xmath39 . here , @xmath40 is the measure of the asymmetry in hopping and takes on values between 0 and 1 .
[ [ section-9 ] ] the moves ( i ) to ( iv ) above define a general model in which particles can move via both stack movement and chipping .
however , as indicated in section [ sec : intro ] , in this paper we will analyse the model in the absence of stack movement i.e. in the limit @xmath41 .
[ [ section-10 ] ] several features of our model , including boundary effects , mass movement and coalescence , and two - species coupling have been studied separately in a number of different contexts .
macroscopic aggregates are known to form in models such as the the zero range process ( zrp ) @xcite and the aggregation - chipping model @xcite . in these models ,
the closed system in steady state can enter a condensate phase in which a finite fraction of the total mass resides on a single site .
a quasi - stationary state characterised by growth of total mass in the system occurs in aggregation models such as the takayasu model @xcite in which particles are injected at every site .
the in - out model @xcite , which extends the takayasu model to allow for evaporation of unit masses , exhibits a phase transition from a growing state to a steady state .
such a phase transition is also seen in our model
. however , the phases in our model are not spatially uniform as injection occurs only at the boundary .
[ [ section-11 ] ] open boundaries with injection can induce interesting features such as boundary - driven phase transitions in the asymmetric simple exclusion process ( asep ) @xcite . in the zrp with open boundaries @xcite , steady state , in the sense of constant average mass ,
is not reached for strong boundary injection ; indefinitely growing aggregates may form on one or both boundaries , while bulk sites attain steady state or show slower growth , depending on the injection , hopping rates , and asymmetry .
our model , in which injection takes place at the boundary , also shows growing phases ; somewhat counter intuitively , a steady state with constant mass is reached close to the boundary , while there is unbounded growth of the mass in the bulk .
[ [ section-12 ] ] coupling between different species of particle may or may not involve interconversion .
for example , in the two - species zrp @xcite , condensates of one or both species may form above critical densities but the number of particles of each species in the system does not change .
new effects are seen when interconversion ( or more generally , non - conservation of particle number ) occurs .
multi - species models that allow for inter - conversion have been studied earlier in the context of transport in single @xcite and multiple @xcite channels@xcite .
these models differ from the present work in that they restrict the occupancy to at most one particle per site in each channel . generically in an open boundary channel ,
if the number of particles is not conserved in the bulk ( e.g. if switching between channels or deposition and evaporation occurs ) , then for low switching rates , the system may develop a localised shock separating low density and high density regions in one of the channels @xcite .
our model too shows co - existence of two phases of one species ( a steady phase and a growing phase ) in different spatial regions separated by a domain wall with a tunable location .
our main results may be summarised as follows : + the model is parametrized by the rates @xmath42 and the asymmetry parameter @xmath40 , where @xmath43 corresponds to symmetric and @xmath44 to fully asymmetric rightward hopping . in different regions of parameter space , different kinds of behaviour are seen : 1 .
for some values of parameters , the system exists in a steady phase .
occupation probabilities of a and b at a site @xmath1 , i.e. the probability that the site has at least one a ( b ) particle are defined as @xmath45 where @xmath46 is the probability of finding @xmath47 a particles and @xmath48 b particles on site @xmath1 between time @xmath0 and @xmath49 . in the steady phase , @xmath50 and @xmath51 at each site ,
reach a time independent value _
less than 1_. the total mass in the system ( and on each site ) also reaches a finite time - independent value .
2 . for other values of parameters ,
the system exists in a quasi - stationary state characterised by stationary ( time - independent ) currents but _ non - stationary i.e. indefinitely growing total mass_. this happens even in a finite system .
we will refer to unbounded growth of mass as _ formation of pile - ups_. more precisely , if the average a ( b ) mass at a site grows indefinitely , we refer to this as an a ( b ) pile - up at the site . at the site with an a ( b ) pile - up
, the mean occupancy @xmath52 approaches @xmath53 at long times ( see sec . [
sec : first_site ] ) .
this may be treated as a functional definition of a pile - up .
a pile - up is to be distinguished from a condensate of the sort found in the zero range process ( zrp ) @xcite : a condensate at a site contains a finite fraction of the total number of particles in the system but the condensate mass does not grow in time .
3 . in the growing phase ,
a system may have pile - ups of a , of b or of both species .
for the situation with injection of only a particles , as studied in this paper , an a pile - up can be found only on site 1 ( except in the special case @xmath54 , @xmath55 ) , i.e. it is a boundary phenomenon .
b pile - ups , on the other hand , are found in the bulk . in a phase with b pile - ups ,
an interesting situation can occur where the lattice has a _ region in steady state ( with finite mean mass ) co - existing with a region of b pile - ups ( where the mean mass grows indefinitely)_. for fully asymmetric hopping , the region close to the left boundary is in steady state while pile - ups occur in the entire region to the right ( shown schematically in fig [ schematic_asym ] ) .
for the symmetric or partially asymmetric case , there are two regions in steady state close to either boundary , with a region of b pile - ups between these ( fig .
[ schematic_sym ] ) . in the rest of the paper
, we will adopt the following terminology . if the average mass of both a and b reaches a finite time - independent value , we will refer to the system as being in a steady phase .
if one region of the system attains constant average mass and another region shows unbounded growth , we will refer to the two regions as a region in steady state and a region with pile - ups respectively .
the system as a whole will be referred to as being in a growing phase .
[ [ section-13 ] ] steady and growing phases can also be distinguished by the spatial profile of the currents .
let @xmath56 , @xmath57 and @xmath58 denote respectively the net a particle current , the b particle current and the total particle current in the @xmath1th bond of the lattice i.e. from site @xmath1 to @xmath59 . figs . [ j(x)steady ] and [ j(x)growing ] show schematically how , for @xmath44 , these three currents vary across the lattice in a steady phase and a growing phase ( with b pile - ups ) respectively . in the steady phase , although @xmath60 and @xmath61 vary with the distance from the origin @xmath62 , they do so in such a way that the total particle current @xmath63 remains constant across the lattice . in contrast , in the growing phase , the _ total particle current @xmath63 also varies with @xmath62 _ in the pile - up region .
in fact , @xmath64 must be greater than @xmath65 for a pile - up to exist at site @xmath1 , implying that @xmath63 decreases with @xmath1 ( or @xmath62 ) in a pile - up region . ) , the occupation probability profiles @xmath66 and @xmath67 ( dashed lines ) are flat , being @xmath68 and @xmath69 respectively . when @xmath70 , @xmath71 , there is a net interconversion from a to b at each site , pushing @xmath72 down from the value @xmath66 and pushing up @xmath73 from the value @xmath67 . here
@xmath72 and @xmath73 ( solid lines ) are occupation probabilities with interconversion.,scaledwidth=40.0% ] [ [ section-14 ] ] in order to explain the spatial structure of the various phases , it is useful to understand what causes the spatial variation in @xmath60 and @xmath61 or equivalently , @xmath72 and @xmath73 .
for simplicity , let us consider purely rightward hopping ( @xmath44 ) in a system of size @xmath6 . in the absence of interconversion , the particle current @xmath60 is the same in each bond , and so is @xmath61 .
thus , @xmath72 and @xmath73 are constant across the lattice , being @xmath68 and @xmath69 respectively ( dotted lines in fig . [ s(x ) ] ) .
interconversion changes @xmath60 and @xmath61 and consequently @xmath72 and @xmath73 ( solid lines in fig .
[ s(x ) ] ) in the following way .
the particle current into site 1 ( @xmath74 ) is entirely of type a. at site 1 , some of the a particles convert to type b , resulting in a small probability that site 1 also has b particles .
hence , @xmath75 is different in composition , having a small b component as well . at site 2
, there is still a much larger concentration of a particles than b particles , and hence a net conversion of a to b particles .
thus , the composition of @xmath76 shifts further towards b. this process continues until we reach a site at which there is _ no net conversion_. beyond this site @xmath60 and @xmath61 do not change with @xmath62 . if interconversions are faster than hopping ( i.e. @xmath77 ) , then at a given site , the relative amounts of a and b can alter significantly through interconversion before hopping occurs . in this limit ,
@xmath60 and @xmath61 change sharply over a few sites and reach their asymptotic values very close to the left boundary . in the other limit ( @xmath78 ) , @xmath60 and @xmath61 vary slowly over a large part of the lattice . in particular , if @xmath79 and @xmath80 are not @xmath81 but @xmath82 , then the spatial variation in @xmath60 and @xmath61 becomes a bulk rather than boundary effect .
[ [ section-15 ] ] the right boundary is an absorbing boundary .
the degree of asymmetry in the hopping determines how far into the lattice the effect of the absorbing boundary condition propagates .
for fully asymmetric dynamics ( @xmath44 ) , the right boundary has no effect on any other site while for @xmath40 slightly less than 1 , the effect extends only into a very small region of the lattice close to the boundary .
for @xmath43 , the information about the right boundary propagates into the entire lattice .
thus , the asymmetry in hopping rates and the rate of interconversion relative to the rate of hopping together determine how @xmath60 and @xmath61 or @xmath72 and @xmath73 vary across the lattice .
[ [ section-16 ] ] for fully asymmetric ( rightward ) hopping ( @xmath44 ) , a site is not affected by any sites to the right of it .
thus , the first site can be analysed independently of the rest of the lattice ; this analysis yields considerable insight into the full problem .
we find four possible behaviours : steady state ; pile - up of a but not b ; pile - up of b but not a ; pile - ups of both a and b. the master equation for the probability distribution @xmath83 ( as defined in section [ sec : results ] ) at the first site can be written as follows : + [ eqn : mastereqsite1 ] + for @xmath84 and @xmath85 @xmath86 @xmath87 @xmath88 @xmath89 + to understand the behaviour of the system , we do not solve the master equation but work with the time - evolution equations for @xmath90 and @xmath91 : [ eqn : currenteqsite1 ] @xmath92 @xmath93 here @xmath94 and @xmath91 denote the average masses of a and b respectively on site 1 averaged over time histories and @xmath72 and @xmath73 the occupation probabilities of a and b ( defined in eq . ) ; all of these are functions of time .
evidently , eq . is just a continuity equation for the average masses .
[ [ section-17 ] ] the time - evolution of the first site can also be mapped to a random walk in two dimensions .
the two - dimensional ( 2d ) space here is the @xmath95 space .
the random walk starts at the origin at @xmath7 , moves in both positive and negative directions along the @xmath47 axis ( with rates @xmath96 and @xmath97 respectively ) but moves only in the negative direction along the @xmath48 axis ( with a rate @xmath98 ) as there is no injection of b particles .
diagonal moves , corresponding to interconversion , are also allowed _ but only in the direction of increasing @xmath47 and decreasing @xmath48 or vice versa_. the @xmath47 and @xmath48 axes are reflecting boundaries . a possible random walk , with all allowed moves in @xmath95 space ,
is illustrated in fig .
[ randomwalk ] .
clearly , the random walk can not be decomposed into independent walks along the @xmath47 and @xmath48 axes because of the diagonal moves .
[ [ section-18 ] ] the random walk picture is useful in the following way . to begin with ,
let us consider a simple 1d random walk in the positive half of the @xmath62-axis with a reflecting barrier at the origin .
this walk has a drift velocity @xmath99 and a diffusion constant @xmath100 .
after a length @xmath101 , drift becomes more important than diffusion . if @xmath102 , then beyond this length , the mean displacement will grow linearly with time , while if @xmath103 , the mean displacement reaches a constant value @xmath101 . if @xmath104 , then the motion is always diffusive , and the mean displacement grows as @xmath105 .
the 2-d walk is more complicated because motion along @xmath47 and @xmath48 axes is coupled due to the diagonal moves ( corresponding to interconversion ) .
the diagonal moves effectively make @xmath99 and @xmath100 in the @xmath47 direction dependent on the average displacement along the @xmath48 direction .
if the effective @xmath99 in the @xmath47 direction is greater than or equal to zero , then we get a pile - up of a on the first site ( with @xmath104 implying @xmath106 growth of the pile - up and @xmath102 , growth that is linear in time ) . also , as expected from the theory of random walks
, the fluctuation about the mean mass , i.e. @xmath107 grows as @xmath105 ( for @xmath108 ) . if @xmath103 , then the mean mass of a and the mean deviation in mass of a reach time independent values . in terms of the occupation probability ,
if @xmath109 , then @xmath47 must be finite and if @xmath110 , @xmath47 is infinite at long times ( i.e. pile - ups occur ) . to see this , note that the return from @xmath111 to @xmath112 requires @xmath113 steps , implying that @xmath114 falls as @xmath115 , which is non - vanishing only for finite @xmath113 .
so , if @xmath116 is less than 1 at long times , @xmath117 must be finite ; if @xmath72 is equal to 1 at long times , @xmath117 is infinite .
[ [ section-19 ] ] from the above discussion it is clear that depending on the rates , the effective @xmath99 in either direction ( i.e. along @xmath47 and @xmath48 axes ) can be negative or non - negative , resulting in four possibilities altogether .
figure [ trails ] shows schematically , possible random walks corresponding to the four scenarios steady state , pile - up of a , pile - up of b and pile - up of both a and b. [ [ section-20 ] ] in terms of the rates , the conditions for each of the four scenarios can be derived from eq . .
for example , if there is a pile - up of b but not of a on site 1 , then at long times , @xmath118 with @xmath119 and @xmath120 with @xmath109 . substituting these into eq . gives the two inequalities in parameter space that must be satisfied in this case .
appendix [ appendixa ] lists the necessary conditions in parameter space for each of the four cases .
[ [ section-21 ] ] these conditions can be visualised better by considering a reduced parameter space .
we look at a case where both species hop with the same rate , i.e. @xmath121 .
then the total mass evolves according to : @xmath122 clearly , if @xmath123 , then @xmath72 , @xmath124 and the system will always attain steady state .
if @xmath125 , the system can never attain steady state and must show pile - ups ( as @xmath126 even when @xmath72 and @xmath73 take on the maximum possible value i.e. @xmath53 ) . in this case , if @xmath127 , i.e. the interconversion rates are similar , both a and b pile up . however , if one of the conversion rates is much higher than the other , then the species which converts fast is able to stabilize , so that only one of the species piles up .
figure [ same_hoppingrates2 ] shows the @xmath128 phase diagram for the case @xmath125 . if @xmath129 , then @xmath72 and @xmath73 can adjust themselves such that steady state is attained
however , as before , if one of the species converts much faster than the other , it drives the latter out of steady state , causing it to pile up .
figure [ same_hoppingrates1 ] shows the @xmath128 phase diagram for the case @xmath129 .
[ [ section-22 ] ] in this section , we analyse the fully asymmetric case ( @xmath44 ) . in this limit
, the current equations involving @xmath72 and @xmath73 have a simple solution .
also , this case illustrates in a clear way , how interconversions drive spatial gradients in a and b particle currents .
[ [ section-23 ] ] current equations similar to eq . can be written for each site in the system : [ eqn : currenteqsitei ] @xmath130 @xmath131 and for site @xmath53 , as before : [ eqn : currenteqsite1a ] @xmath132 @xmath133
the above equations are exact , being just the continuity equations for the average masses . they can also be obtained by writing a master equation for the total probability @xmath134 , multiplying it by @xmath2 ( or @xmath4 ) and averaging over all configurations . for the steady phase ( @xmath135 at all sites ) , equations and give recursion relations expressing @xmath50 and @xmath51 in terms of @xmath136 and @xmath137 . iterating these recursion relations
gives the following steady phase occupation probabilities : @xmath138 \\ & s_{i}^{b}=\frac{au}{uq+vp}\left[1-\lambda^i\right ] \end{split}\ ] ] where @xmath139 . and
@xmath51 as a function of site number @xmath1 in a steady phase for @xmath44 , @xmath140 , @xmath141 , @xmath142 , @xmath143 , @xmath144 .
data points are obtained from monte carlo simulations , and solid lines are plots of the analytical results in eq . .
@xmath72 and @xmath73 approach their asymptotic values over a length @xmath145 defined in eq .
.,scaledwidth=45.0% ] .
[ sstate ] figure [ sstate ] shows the spatial profile of @xmath50 and @xmath51 in a steady phase , as obtained from monte carlo simulations , along with the analytical results , for a certain choice of parameters .
[ [ section-24 ] ] from eq .
, the following can be deduced : 1 . in a steady phase ,
occupation probabilities @xmath146 , @xmath51 must be less than @xmath53 for all @xmath1 .
this is ensured if the maxima of the two expressions in are less than 1 , i.e. @xmath147 ( for @xmath72 ) and @xmath148 ( for @xmath73 ) .
thus , having rates which satisfy these two inequalities simultaneously is a necessary and sufficient condition for the system to be in steady phase .
2 . away from the left boundary , @xmath146 and @xmath149 approach asymptotic values that satisfy the relation @xmath150 which corresponds to zero interconversion current at site @xmath1 .
the characteristic length ( @xmath145 ) associated with this ( exponential ) approach is given by @xmath151^{-1 } \label{eqn : decay_length}\ ] ] which is @xmath152 to first order in interconversion rates .
this first order term has the following interpretation .
@xmath153 is the time over which a single hopping event of an a particle takes place at each site in the lattice .
@xmath79 is the rate of conversion for an a particle .
thus , @xmath154 is the probability that a conversion event ( of an a particle ) takes place in this time .
the term @xmath155 has a similar interpretation .
thus , @xmath156 is an estimate of the average number of interconversions taking place in the system in a time interval over which or before hopping occurs . as reflected in the expression for @xmath145 ( and discussed in section [ sec : results ] ) , when this number is large , @xmath72 and @xmath73 approach their asymptotic values over just a few sites from the left boundary .
conversely , for interconversion rates that are proportional to @xmath157 , the length @xmath145 is of order @xmath6 and the gradients in @xmath158 extend over the bulk of the lattice .
( @xmath82 interconversion rates are also studied in @xcite . )
[ [ section-25 ] ] to study the structure of the growing phases in the system , we use fig .
[ network ] which is a diagrammatic version of eqs . and .
each site can be thought of as having a and b compartments .
the figure shows particle currents between the various compartments , with vertical arrows representing intra - site interconversion currents and horizontal arrows the inter - site hopping currents .
the currents into and out of any compartment can either balance each other ( resulting in a steady state for the compartment ) or the in current can be more than the out current ( resulting in a pile - up ) .
evidently , the out current can not exceed the in current at any site as this would eventually lead to negative mass .
this simple observation can be used to establish the following result : + _ in both steady and growing phases , if @xmath159 , then @xmath160 and @xmath161 @xmath162 . _ consequently , @xmath163 , i.e. interconversion current keeps decreasing as we move right , asymptotically approaching zero .
[ [ section-26 ] ] to prove this result , we note that if @xmath164 , then at site 1 , the interconversion current , being the only ` in ' current for compartment b , is always positive , i.e. from a to b. this implies that @xmath165 and @xmath166 .
we can prove this by ruling out the other three possibilities which are : + ( i ) @xmath167 , @xmath166 : this would imply that either the net current out of site 2 is more than the net current into it which is not possible ; or if @xmath168 and @xmath169 , then @xmath170 , i.e. there is a net interconversion current out of a. since the in and out hopping currents for a are equal ( @xmath168 ) , this means there is a net out current from compartment a which is not possible .
+ ( ii ) @xmath165 , @xmath171 : since @xmath172 and @xmath173 can at most be equal to @xmath53 , this implies that @xmath174 , i.e. site 2 is in steady state and currents for both compartments balance . to satisfy current balance for a with @xmath165
, interconversion current must be from a to b ( see fig .
[ network ] ) but for current balance for b with @xmath171 , it must be in the opposite direction .
this leads to a contradiction .
+ ( iii ) @xmath167 , @xmath171 : this implies that @xmath175 .
however ( from fig . [ network ] )
, to have @xmath167 , the interconversion current has to be from b to a or at least has to be zero .
this leads to an inconsistency .
+ thus , the only possibility is that @xmath165 and @xmath166 . in the same way
, the proof can be extended to all subsequent sites .
it may be noted that we have not assumed the existence of steady state * i.e. @xmath176 , * anywhere in this argument .
thus , the inequalities hold for the growing phase also .
the @xmath177 case has many sub cases and can be analysed in a similar way . in the rest of the discussion
, it is assumed that @xmath159 .
[ [ section-27 ] ] the above inequalities for the occupation probabilities have the following implications : 1 . even if a pile - up of a exists on site 1 ( @xmath178 ) , no subsequent site can have a pile - up of a ( for @xmath164 ) as @xmath179 @xmath180 .
thus , an a pile - up would be found only at site 1 , if at all .
the occupation probability @xmath73 increases from left to right .
if @xmath73 touches 1 at some site @xmath181 , then @xmath182 @xmath183 as @xmath184 @xmath183
. thus , there are _ b pile - ups at all sites right of @xmath181_. [ [ section-28 ] ] it also follows from eq . that when b pile - ups exist , then at long times , the mass of b grows linearly with time at all sites with pile - ups ( except at site @xmath181 , the earliest site with a b pile - up , where it may grow as @xmath105 ) .
moreover , if @xmath185 , then @xmath186 , @xmath187 etc .
thus , after sufficiently long times , a snapshot of the lattice at any instant would show an exponential decay in space in the mass profile of b , site @xmath188 onwards , with a decay length given by @xmath189^{-1}$ ] .
the location of the site @xmath181 depends on the parameters and can even be tuned to be at site 1 .
left of site @xmath181 , the system exists in steady state ( except for a possible a pile - up on site 1 ) while right of it , the system exists in a growing phase .
, @xmath190 , @xmath191 , @xmath192 , @xmath144 .
snapshots of the lattice showing @xmath48 profile at @xmath193 ( squares ) and @xmath194 ( circles ) ( @xmath0 in units of 500 mc steps ) .
inset : semi - log plot of @xmath195 vs @xmath1 .
solid line shows the analytical prediction.,scaledwidth=48.0% ] [ [ section-29 ] ] the system was also studied numerically by doing monte carlo simulations for small lattices , typically @xmath196 .
since later sites can not affect earlier sites , this is the same as doing a simulation with large @xmath6 but keeping track of only the first few sites .
numerical results are presented for a choice of parameters for which the theory predicts the existence of steady state at the first four sites , and pile - ups of b site 5 onwards .
figure [ profile ] shows snapshots of the b profile of the lattice at two times @xmath197 and @xmath198 ( where t is in units of 500 monte carlo ( mc ) steps ) . in agreement with the analytical prediction , upto site 4
, @xmath48 reaches a finite time - independent value , while growth of @xmath48 starts occurring from site 5 , is fastest at site 6 and becomes slower as we go further from the boundary .
the inset of fig .
[ profile ] shows @xmath199 as a function of site number in a semi - log plot .
the solid line is the analytically expected exponential decay .
there is good agreement between the solid line and the data points until site 10 but not after that . for sites after site @xmath200
, the long time behaviour , i.e. linear growth does not set in during the run time of the simulation ( @xmath201 mc steps ) .
such large relaxation times were the primary reason for choosing small system sizes in simulations .
[ [ section-30 ] ] thus , to sum up , for only rightward hopping ( @xmath44 ) , with injection of a particles at site @xmath53 , @xmath72 decreases monotonically while @xmath73 increases monotonically with site number .
the system may attain stationarity or it may exist in a growing phase characterised by ` pile - ups ' .
an a pile - up can occur only at the first ( boundary ) site ( for @xmath202 ) .
b pile - ups occur in the bulk ; in fact , it is possible to have steady state existing near the right boundary and pile - ups of b further on ( as in fig [ profile ] ) . by making interconversion much slower than hopping ( @xmath203 )
, both the steady state and the pile - up region can be made macroscopically large ( @xmath204 ) .
if a site has a pile - up of b , then all sites right of it also have b pile - ups .
these pile - ups grow as @xmath205 . in the pile - up region , at any instant , the spatial profile of @xmath48 shows an exponential decay .
[ [ section-31 ] ] when @xmath55 , i.e. b particles can not move , there are two important differences .
steady state on site 1 implies a steady state behaviour for the whole system , and it is not possible to have a region of steady state followed by a region of pile - ups .
secondly , @xmath55 is the only surface in parameter space on which _
pile - ups of a need not be localised at site 1_. if there exists a pile - up of a on site 1 , then as long as @xmath206 , there are pile - ups of a on all subsequent sites .
these pile - ups grow as @xmath105 with time .
[ [ section-32 ] ] for fully asymmetric hopping ( @xmath44 ) , steady phase occupation probabilities were obtained by solving eqs . , . for any other @xmath40
however , a given site is affected by both its left and right neighbours and the resulting time evolution equations for @xmath207 and @xmath208 , ( similar to eqs . , )
yield complicated recursion relations for the occupation probabilities which are difficult to solve .
thus , to analyse the system for any general asymmetry , we assume the total number of sites @xmath6 in the system to be large , and take a continuum limit for the lattice , thereby going from the integer valued site number @xmath1 to the real valued position co - ordinate @xmath62 , defined as @xmath209 . by this definition , @xmath62 can take on real values between @xmath69 and @xmath53 . in the continuum limit , @xmath210 [ [ section-33 ] ]
first we would like to check whether the steady phase profile given by eq . is recovered in the continuum limit . substituting from eq . into eqs . and ,
setting time derivatives equal to zero and retaining only leading order terms in @xmath157 , we get : [ eqn : asym_cont ] @xmath211 @xmath212 along with the boundary conditions @xmath213 and @xmath214 . here
, @xmath215 and @xmath216 refer to the rescaled rates @xmath217 and @xmath218 . in eq .
the full derivative with respect to @xmath62 has been used instead of the partial derivative as the steady state occupation probabilities are functions of @xmath62 only .
the solution of eq . is given by : [ eqn : asym_cont_profile ] @xmath219\ ] ] @xmath220\ ] ] with @xmath221 .
a comparison with the decay length computed for the discrete lattice ( in eq . ) shows that apart from the length rescaling factor @xmath157 , @xmath222 is just the first order term in the expansion of the exact decay length @xmath145 in terms of @xmath156 .
this first order term arises because in retaining only the leading order term in eq .
, we are making the assumption that @xmath223 are slowly varying functions of @xmath62 which in turn requires that interconversions are much slower than hopping , i.e. @xmath224 . [ [ section-34 ] ] the other limiting case is @xmath43 or perfectly symmetric hopping .
as before , current equations similar to eqs . and can be written for the discrete lattice , which on taking the continuum limit , yield the following two coupled differential equations : [ eqn : sym_cont ] @xmath226 @xmath227 here the rescaled interconversion rates are @xmath228 and @xmath229 .
the boundary conditions are given by @xmath230 , @xmath231 and @xmath232 .
the boundary condition at @xmath233 just reflects the fact that the right boundary acts as a sink , as particles can not return to the system once they leave from site @xmath6 .
equation can be solved for steady phase by setting time derivatives equal to zero , taking second derivative of eq .
with respect to @xmath62 and eliminating the @xmath73 terms by using eq . and the original eq .
this gives an equation which is the second derivative with respect to @xmath62 of a second order differential equation in @xmath72 .
the solution of this equation gives the following steady phase spatial profiles of @xmath72 and @xmath73 : [ eqn : sym_profile ] @xmath234\ ] ] @xmath235\ ] ] where @xmath236 [ [ section-35 ] ] figure [ profile_sym ] shows the spatial profile of @xmath72 and @xmath73 in steady phase for a certain choice of parameters .
clearly , steady phase profiles ( eq . ) obtained from the continuum approximation tally well with the monte carlo results .
[ [ section-36 ] ] the spatial profiles of @xmath72 and @xmath73 in fig .
[ profile_sym ] give an indication of the nature of the growing phases of the system . as in the asymmetric case with @xmath164 ,
since @xmath72 is always a decreasing function of @xmath62 , it follows that an a pile - up can exist only on the first site .
if an a pile - up exists at site 1 , then we use the boundary condition @xmath237 instead of the boundary condition for @xmath238 in eq . to solve for the steady state of the rest of the system . from fig .
[ profile_sym ] we can also deduce that if the maximum in the @xmath239 spatial profile crosses 1 , the system can not exist in a steady phase but must show pile - ups of b. moreover , these b pile - ups occur in some central region of the lattice flanked by regions of steady state on either side . to test this prediction ,
simulations were done with parameters for which @xmath239 , as obtained from eq . , crosses 1 in some range of @xmath62 values .
figure [ pileups_profile ] shows the occupation probability profiles for such a choice of parameters .
as expected , the region of pile - ups ( with @xmath240 ) exists in the middle of two steady state regions .
appendix [ appendixb ] presents the details of how we analytically calculate the spatial boundaries of the pile - up region as well as the occupation probability profiles which are shown as solid lines in fig .
[ pileups_profile ] .
the main assumption involved in this calculation is that @xmath241 vanishes at the boundaries of the pile - up region .
[ [ section-37 ] ] away from the limits of pure drift ( @xmath243 ) or pure diffusion ( @xmath244 ) , current equations similar to eq . can be written with partially asymmetric hopping rates ( @xmath245 ) .
these equations also have terms with first derivatives of @xmath72 and @xmath73 with respect to @xmath62 , and can be solved in the same manner as eq . ,
giving the following expressions for the steady phase occupation probabilities @xmath246 and @xmath239 : [ eqn : p_asym_profile ] @xmath247 \end{split}\ ] ] @xmath248 \end{split}\ ] ] where @xmath249 , @xmath250 , @xmath236 and @xmath251 [ [ section-38 ] ] equation is also valid for @xmath252
. however , if particles have a net drift back towards the source , only a small region close to @xmath253 is occupied .
thus , we confine our analysis to the more interesting case @xmath254 . strictly at the two limits ,
@xmath244 and @xmath243 , eq . is not valid .
however , taking the limit @xmath255 , or equivalently @xmath256 for eq .
gives back the profiles of eq . .
at @xmath243 , a second order description can not be used as there is only one boundary condition for @xmath72 ( or @xmath73 ) .
[ [ section-39 ] ] the spatial variation of @xmath246 and @xmath239 is governed by the rescaled variables @xmath257 and @xmath258 .
if @xmath259 and @xmath260 , this variation is confined to the boundary regions , eventually becoming a discontinuity in the limit @xmath261 .
if @xmath262 and @xmath263 , the gradients due to the two boundaries extend into the whole lattice and together , determine the behaviour of the system in a complex way .
for large but not infinite values of @xmath257 and @xmath258 ( @xmath264 , @xmath265 ) , gradients near the two boundaries can be treated independently and yet are not mere boundary discontinuities .
this situation is shown in fig .
[ partial_asym_ss ] .
for such values of @xmath257 and @xmath258 and sufficiently large @xmath6 , two distinct length scales emerge , one characterizing the variation near @xmath266 and the other near @xmath267 .
these length scales are , respectively : @xmath268^{-1 } , \qquad \tilde{l}_{2 } = \left[2\tilde{\xi}\right]^{-1}\ ] ] [ [ section-40 ] ] the length @xmath269 is the distance from @xmath253 beyond which a and b are effectively decoupled ( no net interconversion ) .
@xmath269 is small for fast interconversions ( high values of @xmath257 ) , which has been rationalized earlier in section [ sec : results ] .
@xmath269 also decreases as @xmath258 decreases , i.e. with decrease in hopping asymmetry . and
@xmath149 for @xmath270 , @xmath271 , @xmath272 , @xmath273 , @xmath274 and @xmath275 . for these parameters , @xmath276 and @xmath277 .
solid lines are plots of eq . with @xmath62 replaced by @xmath278.,scaledwidth=48.0% ] [ [ section-41 ] ] the effect of the absorbing right boundary ( at @xmath233 ) extends over a length which is given by @xmath279 .
this can be seen in the following way . since
a and b are effectively decoupled beyond @xmath269 , the spatial variation of @xmath246 ( and similarly @xmath239 ) beyond this length is described by a drift diffusion equation ( @xmath280 ) along with the boundary condition @xmath281 .
the diffusion length associated with this equation is @xmath282^{-1}$ ] which is just @xmath279 .
[ [ section-42 ] ] as @xmath269 and @xmath279 increase , the variations near the two boundaries can no longer be treated independently and the middle region of nearly constant @xmath246 and @xmath239 seen in fig .
[ partial_asym_ss ] disappears .
perfectly symmetric hopping , for which @xmath283 diverges , is an extreme case of this .
[ [ section-43 ] ] the two - species model can be extended quite easily to more species .
for example , consider the three - species generalization . particles of type a are injected at the left boundary .
an a particle can convert to b and vice versa ; in addition , b particles can convert to c ( at rate @xmath181 ) and c to b ( at rate @xmath284 ) .
note that direct interconversion between a and c is not allowed . also , as before , a single particle of any species can chip off a site ( with rates @xmath285 for species a , b , c respectively ) and move , in general , in a driven diffusive way .
[ [ section-44 ] ] the model can be analysed in exactly the same way as the two - species model by writing equations involving particle currents .
the analysis of the three - species model yields the following results : 1 . while @xmath72 decreases monotonically as a function of @xmath62 , @xmath73 and @xmath286 both show qualitatively the same behaviour of the spatial profile , i.e. both increase monotonically with @xmath62 in the driven case and have a peak in the diffusive case .
this indicates that in the three - species model , depending on the rates , pile - ups of b or c or of both can occur in the bulk of the lattice and a pile - up of a only on the first site .
2 . interestingly , although interconversion takes place in the order @xmath287 , pile - ups need not appear in the same order in space .
for example , c pile - ups may be found to the left of b pile - ups ( i.e. closer to the left boundary ) as shown in fig [ 3sp ] . ) .
parameters : @xmath288 , @xmath289 , @xmath290 , @xmath291 , @xmath292 , @xmath293 , @xmath294 , @xmath295 . data
points show @xmath4 and @xmath296 at two time intervals @xmath297 and @xmath298 .
pile - ups of c occur earlier in the lattice ( site 13 onwards ) than pile - ups of b ( site 15 onwards ) .
inset : @xmath50 , @xmath51 , @xmath299 at long times.,scaledwidth=48.0% ] [ [ section-45 ] ] thus , in the multi - species model with sequential interconversion , in different regions of parameter space , the system may have pile - ups of one , a few or all ( except a ) species of particle .
pile - ups of different species may occur in different regions of the system .
[ [ section-46 ] ] in this paper , motivated by the phenomenology of traffic in the golgi apparatus of the cell , we have studied a stochastic two - species model with boundary injection of type a , interconversion between types a and b , and transport of both species through the lattice ( in general , in a driven diffusive way ) by chipping of one particle at a time . we found that depending on the rates of various processes , the system may either eventually attain steady state or show unbounded growth of mass at all times .
pile - ups ( as defined in the text ) may be composed of a particles or b particles or both . unlike in translationally
invariant systems , the phases in our model have interesting spatial structure . generically , in growing phases , a part of the lattice attains steady state while other regions show unbounded growth of mass ( pile - ups ) .
[ [ section-47 ] ] below , we comment on some possible extensions of the model : 1 . if stack movement also occurs ( as in the more general model defined in section [ sec : model ] )
, there are _ no growing phases _ and the average mass in the system always attains a constant finite value .
this may be argued as follows .
if @xmath300 and @xmath301 are the in and out currents at a site , then the average mass @xmath302 at the site grows as @xmath303 . in the ` chipping only ' case , @xmath304 ( assuming only one type of particle ) .
if @xmath305 , then in and out currents can not balance , and @xmath302 grows indefinitely . on the other hand , with stack movement ,
@xmath301 which is given by @xmath306 , can always balance @xmath300 , resulting in zero growth rate or constant mass .
however , if the rate of stack movement is mass dependent i.e. @xmath307 , then @xmath301 which is now @xmath308 will be bounded for @xmath309 . in this case
, pile - ups can occur and the behaviour of the system would be expected to be similar to that of the ` chipping only ' model .
one extension of this model to higher dimensions is trivial . if injection takes place at a surface perpendicular to one of the spatial directions , and we assume periodic boundary conditions in other directions , the model is still effectively one dimensional as there are no @xmath310 currents between sites in the transverse direction .
generalizations to higher dimensions with other boundary conditions can be more complex .
this model can also be extended quite easily to include injection of b particles at the left boundary , at a rate @xmath311 .
now , depending on the value of the ratio @xmath312 , the system either has net interconversion from a to b ( and behaviour similar to what has been discussed in the paper ) or net interconversion from b to a ( resulting in a pile - ups in the bulk etc . ) or no net interconversion , i.e. no effective coupling between the two species . at the right boundary , introducing exit rates that are different from the chipping rates can also change the behaviour of the system .
however , this change is only a boundary effect unless the exit rate is @xmath82 times the chipping rate
. if the exit rate is small in the sense defined above , then @xmath72 and @xmath73 vary with @xmath62 differently from eqs . .
however , growing phases with b pile - ups in the bulk are still found .
[ [ section-48 ] ] finally , we comment on the possible relevance to the biological system . the model system we have studied is a simple one , but it shares the following qualitative features with the golgi apparatus .
first , there is a gradation of sizes of stacks from the source to the sink . in particular , small stacks are found close to the source while large , though growing
, aggregates occur in regions farther away .
secondly , there is a gradation of ` chemical ' species across the system , with different spatial regions having aggregates that may be predominantly of type b and/or type c etc .
these aggregates , however , show unbounded growth , which is not realistic in the context of the golgi , which consists of large but bounded and discrete structures .
as discussed above , unbounded growth is eliminated once we allow for movement of whole stacks , as in the more general model defined in sec [ sec : model ] , and discussed in this section .
[ [ section-49 ] ] in the present model , we have assumed that mass transfer occurs either by chipping of a single particle or by movement of the whole stack .
other transport processes could involve the fragmentation and movement of chunks of intermediate sizes .
the study of these different mass transport processes in systems with boundary injection is an interesting direction for future study .
[ [ section-50 ] ] we thank a.b .
kolomeisky and d. mukamel for useful discussions and m.r .
evans for very useful comments on the manuscript .
as indicated in section [ sec : first_site ] , conditions for each of the four scenarios for the first site can be derived from eq . .
they are as follows : 1 .
steady state : if @xmath313 and @xmath314 , then steady state is reached with steady state probabilities : + + @xmath315 and @xmath316 2 .
pile - up of b : if @xmath317 and @xmath318 , then @xmath91 keeps growing while @xmath94 reaches a constant value .
+ + at long times : @xmath319 , @xmath119 and @xmath320 3 .
pile - up of a : if @xmath321 and @xmath322 , then @xmath94 keeps growing while @xmath91 reaches a constant value .
+ + at long times : @xmath323 , @xmath110 and @xmath324 4 .
pile - ups of both a and b : if @xmath325 and @xmath326 , then both @xmath94 and @xmath91 keep growing . at long times : + + @xmath327 , @xmath328 and @xmath329
let @xmath330 and @xmath331 divide the lattice into three regions such that in : + region i ( @xmath332 ) : steady state ( @xmath333 , @xmath334 ) ; + region ii ( @xmath335 ) : steady state of a , pile - ups of b ( @xmath333 , @xmath336 ) ; + region iii ( @xmath337 ) : steady state ( @xmath333 , @xmath334 ) . in regions i and iii , since steady state exists , @xmath246 and @xmath239 are obtained by setting the l.h.s .
equal to @xmath69 in both of eqs . and and solving them with the following boundary conditions for : + region i : @xmath338 , @xmath339 , @xmath340 , @xmath341 + region iii : @xmath342 , @xmath343 , @xmath344 in region ii , substituting @xmath345 into eq . , and setting the left - hand - side equal to zero , we get @xmath246 upto two unknowns
. we can find @xmath346 and @xmath347 from the solutions in regions i and iii respectively . since @xmath246 must be a continuous function , these provide the boundary conditions for the solution of @xmath246 in region ii .
thus , we have @xmath246 and @xmath239 for all three regions upto the two unknowns @xmath330 and @xmath331 .
to determine @xmath330 and @xmath331 , we make the assumption that @xmath348 is continuous at @xmath330 and @xmath331 , i.e. @xmath349 and @xmath350 which yields two transcendental equations in @xmath330 and @xmath331 . for a given set of parameters ,
these can be solved numerically to get @xmath330 and @xmath331 , since we have expressions for @xmath246 and @xmath239 in terms of @xmath330 and @xmath331 , this also gives the occupation probabilities in all three regions .
this calculation involves two _ ad hoc _ assumptions , namely , that @xmath241 vanishes and @xmath348 is continuous at the two boundaries of the pile - up region .
these assumptions give results which match well with numerical data ( see fig .
[ pileups_profile ] ) , but they need to be investigated and adequately justified .
99 b. alberts et al , _ molecular biology of the cell _ , 4/e ( new york : garland , 2003 ) b.s .
glick , v. malhotra , cell * 95 * , 883 ( 1998 ) .
e. losev , c.a .
reinke , j. jellen , d.e .
strongin , b.j .
bevis , b.s .
nature * 441 * , 1002 ( 2006 ) .
k. matsuura , m. takeuchi , a. ichihara , k. mikuriya and a. nakano , nature * 441 * , 1007 ( 2006 ) .
patterson , k. hirschberg , r.s .
polishchuk , d. gerlich , r.d .
phair and j. lippincott - schwartz , cell * 133 * , 1055 ( 2008 ) .
s. n. majumdar , s. krishnamurthy , and m. barma , phys .
lett . * 81 * , 3691 ( 1998 ) .
s. n. majumdar , s. krishnamurthy , and m. barma , j. stat .
99 , 1 ( 2000 ) . as discussed in section [ sec : conclusion ] , inclusion of the stack hopping move restores the balance of currents , and eliminates unbounded growth of mass .
m. r. evans and t. hanney , j. phys .
a * 38 * , r195 ( 2005 ) .
m.takayasu and h.takayasu , _ nonequilibrium statistical mechanics in one dimension _ , edited by v. privman ( cambridge university press , cambridge , 1997 )
. s. n. majumdar , s. krishnamurthy , and m. barma , phys .
e * 61 * , 6337 ( 2000 ) .
g. m. schtz , _ phase transitions and critical phenomena _ , edited by c. domb and j. lebowitz ( academic press , san diego , 2001 ) , vol .
e. levine , d. mukamel and g. m. schtz , j. stat .
* 120 * 759 ( 2005 )
evans and t. hanney , j. phys .
a * 36 * , l441 ( 2003 ) .
t. hanney and m. r. evans , phys .
e * 69 * , 016107 ( 2004 ) .
e. levine and r. d. willmann , j. phys .
a * 37 * , 3333 ( 2004 ) .
t. chou , j. phys .
a * 35 * , 4515 ( 2002 ) .
s. muhuri and i. pagonabarraga , epl * 84 * , 58009 ( 2008 ) .
s. muhuri , l. shagolsem and m.rao , e - print arxiv:1101.5531v1 .
u. basu and p.k .
mohanty , phys .
e * 82 * , 041117 ( 2010 ) .
e. pronina and a. b. kolomeisky , j. phys .
a * 37 * , 9907 ( 2004 ) .
e. pronina and a. b. kolomeisky , physica a * 372 * , 12 ( 2006 ) .
t. reichenbach , t. franosch , and e. frey , phys .
97 * , 050603 ( 2006 ) ; t. reichenbach , e. frey , and t. franosch , new j. phys .
* 9 * 159 ( 2007 ) .
r. juhsz , phys .
e * 76 * , 021117 ( 2007 ) . in multi - channel transport
, interconversion corresponds to particles switching channels . | motivated by the phenomenology of transport through the golgi apparatus of cells , we study a multi - species model with boundary injection of one species of particle , interconversion between the different species of particle , and driven diffusive movement of particles through the system by chipping of a single particle from a site .
the model is analysed in one dimension using equations for particle currents .
it is found that , depending on the rates of various processes and the asymmetry in the hopping , the system may exist either in a steady phase , in which the average mass at each site attains a time - independent value , or in a `` growing '' phase , in which the total mass grows indefinitely in time , even in a finite system .
the growing phases have interesting spatial structure .
in particular , we find phases in which some spatial regions of the system have a constant average mass , while other regions show unbounded growth . |
rapid progress has been made in the past decade in discovering other planetary systems and in building theories for their formation and evolution .
more than 300 extrasolar planets have been found through measuring stellar radial velocities and about 60 planets have been observed to transit their stars ( extrasolar planets encyclopaedia , http://exoplanet.eu/ )
. however , these techniques are optimal in finding systems dramatically different from the solar system , with giant planets in orbits very close to their stars .
there are two primary avenues for probing systems with planets that might have structures more analogous to that of the solar system .
first , giant planets have recently been imaged around fomalhaut @xcite and hr 8799 @xcite in orbits with radii of tens of au .
second , the spitzer space telescope ( _ spitzer _ , @xcite ) has observed more than 200 debris disks , which represent dust production from colliding planetesimals , typically in zones analogous to the asteroid and kuiper belts in the solar system .
it is expected that these debris disks are sculpted by giant planets through gravitational interactions ( wyatt 2008 and references therein ) .
observations and models of well - studied examples show a wide variety of structures and behavior , possibly related to the individual quirks of their ( unseen ) planetary systems and their evolutionary states ( e.g. , @xcite ) .
hr 8799 has been known to have a prominent debris system presumably sculpted by unseen planets @xcite .
the recent detection of three giant planets orbiting this star , along with estimates of their masses and orbits , provides a new opportunity to understand in more detail the forces that shape debris disks .
this paper reports new measurements with _
spitzer _ that fully characterize the debris disk around hr 8799 .
we show that the debris system includes at least three components : warm dust interior to the three planets ; cold dust exterior to them ; and a halo of small grains surrounding the cold dust zone .
the placement of the first two disk components appears to be as expected from the orbits of the planets .
the dynamics of the three massive planets @xcite suggests substantial dynamical excitation in the overall planetary system .
the elevated level of dynamical activity in the cold dust zone , which is the source of the large halo , is possibly a result of the unstable state of the system .
the 20160 myr age of the system @xcite corresponds to a critical phase in the history of our own solar system when planet formation processes were near completion and the solar system s dynamical configuration was being established .
the hr 8799 debris system offers an exciting view of this phase of planetary system formation and evolution .
we present the new observations in 2 , describe the general observational results in 3 , build detailed models in 4 , and discuss the implications of this work in 5 .
the observations presented here are from the _ spitzer _ director s discretionary time ( ddt ) program 530 and guaranteed time observations ( gto ) program 50175 that utilize all of the major observation modes of both the infrared spectrograph ( irs , @xcite ) and the multiband imaging photometer for _ spitzer _ ( mips , @xcite ) .
the observational setting and depth for each of the modes are summarized in table [ obs_tab ] . a short ( 6 s @xmath1 1 cycle ) integration irs spectrum of hr 8799
was previously published by @xcite . here
we present an irs spectrum that is 460 times deeper along with newly obtained mips images .
observations at 24 @xmath2 m were obtained in standard small - field photometry mode , for a total integration of 210 s. the observation at 70 @xmath2 m provides a total integration of @xmath0320 s on source .
the 160 @xmath3 observation was obtained using the enhanced - mode at 4 sub - pixel - offset positions , for a total integration of @xmath0150 s. the mips sed - mode observation provides a low - resolution ( r=1525 ) spectrum from 55 to 95 @xmath3 with 600 s of integration on source . [ fig1 ] rlll 28889856 & irs staring & sl2 & 6 s @xmath1 4 cycles + & & sl1 & 14 s @xmath1 4 cycles + & & ll2 & 120 s @xmath1 3 cycles + & & ll1 & 120 s @xmath1 3 cycles + 28889088 & mips photometry & 24 @xmath3 , 5 cluster pos . &
3 s @xmath1 1 cycle + 28889344 & mips photometry & 70 @xmath3 , default scale & 10 s @xmath1 4 cycles + 28889600 & mips photometry & 160 @xmath3 , 4 cluster pos . & 10s @xmath1 1 cycle + & & ( enhanced mode ) & + 25711872 & mips sed - mode & 1 chop & 10s @xmath110 cycles + all of the mips data were processed using the data analysis tool @xcite for basic reduction ( e.g. , dark subtraction , flat fielding / illumination corrections ) , with additional processing to minimize instrumental artifacts @xcite . after correcting these artifacts in individual exposures , the final mosaics were combined with pixels half the size of the physical pixel scale ; the resulting images are shown in figure [ fig1 ] .
the calibration factors used to transfer the instrumental units to the physical unit ( mjy ) are adopted from the mips calibration papers @xcite . to ensure the best flat field result at 24 @xmath3 , a second flat field made from a median stack of the data with all bright sources masked out
was also applied in addition to the regular scan - mirror - dependent flat fielding ( for details see @xcite ) .
the flux density of the source at 24 @xmath3 was estimated using aperture photometry with both small and large apertures ( the same photometry parameters as used in @xcite ) .
the large aperture gives a flux density of [email protected] mjy , while the small aperture gives a flux density of [email protected] mjy ( @xmath02.7% lower ) , implying the source profile is more extended than a true point source .
therefore , to capture all the flux from the extended component , we conducted photometry with a very large ( radius of 35 ) aperture on an image with the dc offset and all background sources subtracted without a sky annulus ( i.e. , the aperture photometry was relative to the median pixel value for the entire frame ; therefore , an aperture correction of 1.06 was applied , see @xcite ) .
the final measured flux we adopted is 86.4 mjy ( before color correction ) .
the center of the source at 24 @xmath3 was determined using a 2-d gaussian fitting routine , and it coincides with the expected stellar position within the pointing error ( @xmath51 ) .
this is consistent with the fact that the flux contribution in the 24 @xmath3 band is mostly from the stellar photosphere ( 58 mjy ; see [ photosphere ] ) .
the stellar photosphere was then subtracted by scaling an observed blue point spread function ( psf ) .
after photospheric subtraction ( see fig .
[ fig1]b ) , the source has fwhms of 578@xmath1567 at a position angle ( p.a . )
of 17 , slightly broader than an observed red psf ( 561@xmath1555 for the @xmath6 lep disk , @xcite ) .
there is a faint ( @xmath00.8 mjy ) source located @xmath021 from hr 8799 at p.a . of 36 , presumably a background object .
the 70 @xmath3 data reduction follows the steps recommended by @xcite and uses time filtering with the source region masked out .
several region sizes were tried .
a masked radius of 55 yields the minimum value for the background variation ( 1-@xmath7 = 1.14@xmath110@xmath8 mjy arcsec@xmath8 per subpixel ) .
the source at 70 @xmath3 ( see fig . [
fig1]c ) is clearly extended with fwhms of 246@xmath1244 at a p.a .
of 77 , compared to the nominal resolution of 18 .
the source has an azimuthally symmetric morphology in the image .
the outer boundary of the source ( though masked by the instrumental psf ) can be traced out to 42 or 34 in radius at 1- or 3-@xmath7 levels , respectively .
aperture photometry was used to estimate the flux density for the source since it is extended .
an aperture of 42 in radius ( 1-@xmath9 boundary ) was used with a sky annulus of 4454 . after correcting for lost light based on the same aperture settings applied to a theoretical psf
, the final integrated 70 @xmath3 flux density is 545 mjy ( before color correction ) .
the 160 @xmath3 data were taken in the `` enhanced aot '' to allow them to be time filtered , as was also done for the 70 @xmath3 data ( for details see @xcite ) .
no additional 160 @xmath3 psf was obtained for the purpose of leak subtraction as it has been shown that the ghost image produced by the 160 @xmath3 filter leakage is less than @xmath015 times of the photospheric flux density at 160 @xmath3 .
the disk is @xmath0400 times brighter than the photosphere in this channel ( see tab .
[ tbl : fluxes ] ) .
this is also confirmed by careful inspection of the data where the expected ghost image lies .
the final 160 @xmath3 mosaic shows some large - scale extended cirrus structures surrounding the source ( see discussion in [ bkg_cirrus ] ) .
the source is clearly detected at 160 @xmath3 ( see fig .
[ fig1]d ) , and is consistent with being azimuthally symmetric above the 5-@xmath10 level at the expected star position .
the source at 160 @xmath3 is surrounded by a low level of asymmetric background cirrus .
the fwhms are 47@xmath141 ( based on an image with a field of view of 88 ) , slightly broader than a nominal observed psf ( @xmath038 , @xcite ) , possibly influenced by the background cirrus .
aperture photometry is used to determine the integrated flux . to minimize the influence of the background cirrus , we used small aperture sizes ( 16 and 24 ) , sky annuli ( 64128 ) and aperture corrections ( 4.683 and 2.613 for a red point source of 50 k ; @xcite ) .
the resultant integrated 160 @xmath3 flux density is 539 mjy ( before color - correction ) .
the centroid of the disk seen in the mips 70 and 160 @xmath3 images coincides with the expected position of the star .
rrrrrrr 23.68 & 86.6 & 58 & 29 & 21 & @xmath55 & 7.5 + 71.42 & 610 & 6.5 & 605 & & 285387 & 319217 + 155.89 & 555 & 1.4 & 554 & & 554 & + the mips sed - mode data were reduced and calibrated as described by @xcite with an extraction aperture of 5 native pixels ( @xmath050 ) in the spatial direction . since the source size ( see below ) is smaller than the extraction aperture , the slit loss was corrected based on a point source .
this procedure would under - correct for the slightly extended source and may explain why there is small offset ( @xmath017% at the fiducial wavelength of the 70 @xmath3 band ) between the 70 @xmath2 m photometry and the mips - sed spectrum .
the final mips sed - mode spectrum is smoothed to match the resolution at the long - wavelength portion of the spectrum ( r=15 ) . the basic reduction and extraction of the irs spectral data were provided by the ssc irs pipeline s18.5 .
we first trimmed a few end points from each module and manually removed large outliers by visually inspecting each of the co - added , background - subtracted spectra ( differencing the two nod positions ) .
the spectra were then averaged with sigma clipping ( s / n=3 ) .
all of the final photometry measurements ( color - corrected , see details in [ analysis : disk_components ] ) , mips sed - mode data , irs combined spectrum , along with previously published data are shown in the spectral energy distribution ( sed ) ( fig
. [ sed ] ) .
hr 8799 ( hd 218396 , hip114189 ) , located at 39.4 pc @xcite , is classified as an a5 v @xcite , @xmath11 doradus variable .
it has @xmath12 bootis type characteristics @xcite with low abundances of the heavier elements ( e.g. , [ fe / h ] = 0.55 ) , but near - solar abundances of c and o ( sadakane 2006 )
. a high - resolution optical spectrum of the star indicates @xmath13 = 7250k , log @xmath14 = 4.30 ( sadakane 2006 ) .
the galactic space motion ( uvw ) of the star resembles those of young clusters and associations in the solar neighborhood with ages of 20160 myr @xcite .
the stellar rotation velocity ( @xmath15 ) is 49 km s@xmath16 @xcite , consistent with the star being viewed close to pole - on . to determine the stellar spectral energy distribution
, we fit all available optical to near - infrared photometry ( johnson @xmath17 , strmgen @xmath18 photometry , hipparcos tycho @xmath19 photometry , 2mass @xmath20 photometry ) with the synthetic kurucz model @xcite based on a @xmath21 goodness of fit test .
a value of @xmath13=7500 k with @xmath22=1.4 @xmath23 and log g = 4.5 and sub - solar abundances gives the best match , in good agreement with the results from spectroscopy ( sadakane 2006 ) .
the resulting stellar luminosity is 5.7 @xmath24 , placing the star near the zero - age main sequence consistent with a young age .
based on the best - fit kurucz model , we then estimate the stellar photospheric flux densities at 23.6 , 71.42 , and 155.9 @xmath3 for the mips 24 , 70 and 160 @xmath3 bands , respectively .
these values are listed in table [ tbl : fluxes ] . with a stellar mass of 1.5 @xmath25 ,
the blowout size ( @xmath26 ) is @xmath02 @xmath3 assuming a grain density of 2.5 g @xmath27 around hr 8799 .
[ excesssed ] figure [ excesssed ] shows the excess sed after stellar photospheric subtraction .
it is evident that there are at least two disk components in the system : one with a characteristic temperature of @xmath0150 k ( warm ) and the other with a characteristic temperature of @xmath045 k ( cold ) .
similar disk components have been suggested by @xcite as well .
for blackbody radiators that are in thermal equilibrium with the stellar radiation field , the warm ( 150 k ) and cold ( 45 k ) components correspond to radii of @xmath09 au ( 02 ) and @xmath095 au ( 24 ) from the star .
interestingly , hr 8799 d , c and b are at projected separations of 24 , 38 and 68 au from the star , respectively @xcite .
they all lie between the two dust reservoirs . in general
, the location of the dust that is at a specific thermal equilibrium temperature depends on the grain properties ( through the absorption efficiency @xmath28 , which is size dependent ) .
the equilibrium dust temperature is computed by balancing the absorption and emission energy and ultimately it depends on cross sections ( @xmath29 , where @xmath30 is the grain radius ) ; therefore , the temperature of large grains is generally lower than that of small grains at the same distance from the heating source .
we adopt the optical constants for astronomical silicates @xcite and compute the grain properties ( @xmath31 values for the absorption and scattering efficiency for grain radii of 0.11000 @xmath3 in size ) using mie theory with additional modifications for large sizes ( @xmath3210 @xmath3 ) .
the resultant thermal equilibrium dust temperatures as a function of radius from the star are shown in figure [ radius_td ] . due to the nature of an optically thin debris system
, any dust in a continuous spatial distribution between the two distinct temperature components will have intermediate temperatures , resulting in a smoother excess sed between 2055 @xmath3 ( see fig . 2 in @xcite ) .
the sed shape and the distinct characteristic dust temperatures strongly suggest that very little dust resides between the two components .
based on figure [ radius_td ] , the emitting zone for the warm component is between @xmath05 and @xmath015 au , with an angular diameter of @xmath5 1 .
this component , which dominates the emission detected in the irs spectrum and the mips 24 @xmath3 band , should be unresolved at 24 @xmath3 ( inside planet d ) .
a similar calculation indicates that the cold component , if it has a size of @xmath0100 au in radius , should be at best only barely resolved at 24 @xmath3 ; however , the excess sed ( fig
. [ excesssed ] ) shows that the cold component would contribute only a small ( @xmath33 9% ) portion of the total signal in the mips 24 @xmath3 band .
thus , the flux well outside the psf at 24 @xmath3 suggests the detection of an extended component at this wavelength .
we estimate the flux of the unresolved point source in the 24 @xmath3 image ( star + disk ) by scaling a psf to match the peak flux of the source .
the scaling suggests that the central unresolved component is @xmath078 mjy at maximum ( assuming the extended component contributes no flux at the position of the star ) .
the stellar photosphere is expected to be 58 mjy in the 24 @xmath3 band , so the central unresolved disk component is @xmath020 mjy before color correction . due to the large temperature differences between the warm and cold components ,
different color - corrections must be applied to each of the components at 24 @xmath3 . applying a color - correction factor of 1.055 for a blackbody of 150 k
( see the mips data handbook , http://ssc.spitzer.caltech.edu/mips/dh/ , for the detailed color corrections for the mips bands ) , the total flux density of the central unresolved ( warm ) disk component is @xmath021 mjy , which agrees well with the photosphere - subtracted irs spectrum ( [email protected] mjy ) .
therefore , the extended disk component is @xmath07.5 mjy ( after applying a color correction of 0.894 for a blackbody of 50 k ) . adding these measurements , the final total flux density for the entire system ( star + two disk components after color corrections for the two different temperatures ) is 86.6 @xmath41.7 mjy ( assumed 2% error based on the nominal calibration uncertainty , @xcite ) .
the expected stellar photosphere is 6.5 mjy in the 70 @xmath3 band , suggesting the disk is @xmath0603 mjy after a color correction factor of 1.12 for a 50 k blackbody .
therefore , the total ( star + disk ) flux density at 70 @xmath3 is 610@xmath431 mjy ( assumed 5% error based on the nominal calibration uncertainty , @xcite ) . at 160 @xmath3 ,
the stellar photosphere is 1.4 mjy ; therefore , the total disk flux density is @xmath0554 mjy after a color correction of 1.03 for a blackbody of 50 k , resulting in a total ( star+disk ) flux of 555@xmath466 mjy ( assumed 12% error based on the nominal calibration uncertainty , @xcite ) . [ radius_td ]
the 70 @xmath3 image is essentially only from the disk since the star is @xmath331 % of the total flux in the far - infrared .
an observed radial surface brightness profile of the disk at 70 @xmath3 is shown in figure [ radprof70 ] . as was indicated by the fwhm measurements ,
the disk is clearly resolved .
the structure of the disk at 70 @xmath3 is similar to the one at 24 @xmath3 , a bright core and an extended disk ; however , the bright core at 70 @xmath3 can not originate from the warm component seen at 24 @xmath3 ( because the temperature is too cold ) .
this bright core is most likely to originate from a dust belt ( hereafter the planetesimal disk ) outside planet b , where emission by dust is also detected at submillimeter wavelengths @xcite .
consistent with this possibility , the emission at 850 @xmath3 is mostly confined within 20 ( diameter ) based on the jcmt / scuba measurements @xcite . however ,
if the dust emitting at 70 @xmath3 all originated from the same location as in the submillimeter , the disk could not be resolved by mips at 70 @xmath3 .
in addition , @xcite also estimate the outer edge of the cold dust component is between 125 and 170 au ( a narrow ring ) based on simple sed fitting and assumed grain properties . given the nominal resolution of 18 in the mips 70 @xmath3 band , any disk structure that is smaller than 360 au in radius is not resolvable at a distance of 40 pc .
the large measured fwhms at 70 @xmath3 suggest that the extended emission at 70 @xmath3 must originate from a component outside the planetesimal disk . limited by the large beam size at 70 @xmath3 , the current data can not differentiate whether there is any gap between the planetesimal disk and the extended component .
we can only estimate the bright core ( presumably the planetesimal disk ) flux by normalizing a psf to match the peak of the observed flux ( this procedure provides an upper limit ) or to match the flux within a 10 radius ( to provide a lower limit ) .
after color correction for 50 k , the unresolved core flux density is 285387 mjy , suggesting that the extended component is roughly equal in integrated brightness to the unresolved core . based on 10 years of astrometric data on planet b , @xcite suggest an orbital inclination of 1323 off the plane of the sky .
in addition , both @xcite and @xcite suggest that the system is unlikely to be exactly face - on from stability analyses of the planetary configuration .
the resolved image at 24 @xmath3 provides very little constraint because the emission at this wavelength is mostly dominated by the central unresolved disk . the resolved image at 70 @xmath3 appears to be relatively azimuthally symmetric . from @xmath0100
mips observations of a routine calibrator , hd 180711 ( g9 iii , 447 mjy at 70 @xmath3 ) , the measured fwhm ratio is [email protected] for a point source at 70 @xmath3 . the measured fwhm ratio ( 1.008 ) of the hr 8799 70 @xmath3 disk suggests a deviation of @xmath01.5-@xmath9 from a face - on symmetric disk . from model images ( see discussion in [ analysis : thermal_model ] ) constructed for various inclination angles , we can rule out any angles that are larger than @xmath025 because the resultant images give a ratio much larger than the measured value .
[ radprof70 ] in summary , the resolved disks at 24 and 70 @xmath3 indicate that the disk around hr 8799 has at least three components : a warm belt inside planet d , a planetesimal disk outside of planet b with extension of @xmath3310 , and a surrounding large extended disk .
the separation into these three components and their placement relative to the planets in the system appear to be largely independent of assumed grain sizes and optical properties .
all of the observed flux densities , including those for the three disk components , are listed in table [ tbl : fluxes ] .
[ radprof24 ]
we now build on the observational results of the preceding section through more detailed models of the hr 8799 debris disk .
although these specific models are under - constrained and hence not unique , they are guided by models of other debris systems and fit the observations of hr 8799 .
therefore , they give a plausible picture of the disk behavior . in [ analysis : surface_brightness_distribution ] , we constrain the disk extent by fitting the observed surface brightness profiles with assumed powerlaw profiles that are independent of grain properties . the main purpose of this exercise is to derive spatial information that can be used in a realistic thermal model . in
[ analysis : thermal_model ] , we then utilize the disk components that are inferred from the observations with the adopted grain properties to estimate the physical extent and masses required for the dust in each component . because of our low angular resolution , it is convenient to characterize the disk using radially averaged surface brightness distributions , represented as a mean value at a given radius and the standard deviation of the mean . for simplicity , we assume that the disk is face - on , so an azimuthally averaged radial profile can be used for modeling .
this approximation greatly reduces the noise by averaging the intensity at a given radius ( a similar approach has been used on the disks around vega @xcite and @xmath34 eri @xcite ) .
the surface brightness radial profile of the disk ( after psf subtraction for the stellar photosphere and masking out all nearby sources ) at 24 @xmath3 is shown in figure [ radprof24]a .
for comparison , we also show the azimuthally averaged radial profiles of the smoothed theoretical psf ( generated by the stinytim program , @xcite ) and the unresolved disk around @xmath6 lep ( su et al .
2009 , in prep . ) scaled to match the peak flux of the hr 8799 disk .
the radial profile of the unresolved @xmath6 lep disk agrees well with the theoretical psf , while it is clear that the second dark and the third bright airy rings of the hr 8799 disk ( in the range of @xmath020 to @xmath030 from the center ) are brighter than for a point source . by subtracting the scaled radial profile of the @xmath6 lep disk from the observed radial profile of hr 8799 , a rough disk surface brightness distribution for the extended disk component
is revealed ( fig .
[ radprof24]b ) .
this component is very faint , @xmath01@xmath110@xmath35 mjy arcsec@xmath8 at a distance of @xmath01000 au from the star , similar to the surface brightness level seen at 24 @xmath3 in the vega disk .
as shown in figures [ radprof70 ] and [ radprof24 ] , the true disk structure is masked by the instrumental psfs .
one way to understand the structure of the disk is to construct a trial surface brightness distribution , convolve it with the beam , and then compare with the observed data .
in particular , this approach is a practical way to constrain the true disk outer radius in the low - resolution images .
since our goal is to set constraints on the main planetesimal disk , the 24 @xmath3 profile in the following fits is for the extended disk only .
the simplest model is to assume that the surface brightness profile follows a power - law , @xmath36 , between an inner break radius ( @xmath37 ) and outer cut - off radius ( @xmath38 ) . in an optically thin debris disk where the only heating source is the stellar radiation and
the disk density is only a function of radius in a power - law form of @xmath39 , the radial surface brightness is then proportional to @xmath40 where @xmath41 is the planck function , and @xmath42 is the radial - dependent dust temperature ( generally , @xmath43 for small grains and @xmath44 for large grains ) .
this simple power - law formula has been used to fit the well - resolved vega disk at both 24 and 70 @xmath3 , and it was found that the power indices change from @xmath45 for the inner part of the disk to @xmath46 for the outer part of the disk @xcite . an inner break point is needed in this simple power - law for steep indices ( @xmath473 or 4 ) ; otherwise , the profile mimics a point source after convolution with the psf .
because the outermost planet is at a radius of @xmath02 , we also set the disk surface brightness to zero inside 2 , i.e. , @xmath480 for @xmath492 ( @xmath50 stands for @xmath51 ) .
[ radprof_fits ] we have tried four different cases for the model surface brightness distribution : ( 1 ) a larger inner hole with @xmath52=4 , ( 2 ) a flat distribution between @xmath53 and @xmath37 but with @xmath52=4 outward , ( 3 ) a single powerlaw with @xmath52=2 , and ( 4 ) two powerlaws with indices of @xmath52=2 and @xmath52=3 . in all cases , we varied @xmath37 and @xmath38 of the surface brightness distribution ( two free parameters ) and computed the reduced @xmath54 value to determine the best fit . the reduced @xmath54 value was computed for data points between 540 , and between 575 at 24 and 70 @xmath3 , respectively .
note that the bright hump seen in the 70 @xmath3 profile between 75 and 90 is where the second bright airy ring is located ( see fig .
[ radprof70 ] ) .
the match between an observed psf derived from the calibration stars and theoretical psf is poor in this range ; therefore , we disregard the 70 @xmath3 data points outside 75 in the modeling . the results are shown in figure [ radprof_fits ] . due to _ spitzer _ s large beam sizes in the far - infrared , the assumed surface brightness behavior and its number of free parameters needed to describe the model
, there is no unique fit for the observed profiles .
the best - fit outer radii range from 23 to 46 , consistent with the previous finding of emission from outside of the submillimeter planetesimal ring ( @xmath510 ) . from the model surface brightness profiles
, we can also derive the surface brightness ratio ( @xmath55 ) in the extended disk .
the ratios range from @xmath0170 to @xmath0250 , which can be used to set constraints on the grain properties since the density - dependent term cancels out in the ratio .
figure [ radius_fr ] shows the expected ratio for a size distribution of @xmath56 with a maximum size cutoff of @xmath010 @xmath3 and various minimum grain size cutoffs ( @xmath57 ) using astronomical silicate grains .
based on the range of derived @xmath55 ratios ( represented by the horizontal dashed line in figure [ radius_fr ] ) , the acceptable range of @xmath57 is @xmath332 @xmath3 ( = @xmath26 ) . outside the main planetesimal disk ( @xmath5810 ) , only grains with minimum sizes @xmath33 the radiation blowout size have flux ratios in the observed range .
this supports the idea that the disk outside of the main planetesimal disk consists of small grains being pushed outward by radiation pressure , i.e. , the hr 8799 disk has an extended halo similar to that seen around the vega disk @xcite .
[ radius_fr ] [ models ] from the preceding section , we can expand on the description of the three disk components summarized at the end of 3.2 : ( 1 ) the warm belt
inside the innermost planet d has a temperature of @xmath0150 k with parameters dubbed as `` @xmath59 '' ; ( 2 ) the main planetesimal disk outside the outermost planet b has a temperature of @xmath045 k with parameters dubbed as `` @xmath60 '' ( parent bodies ) ; and ( 3 ) the outermost component is small grains strongly affected by photon pressure with parameters dubbed as `` @xmath61 '' . the basic equations for computing the sed and surface brightness of an optically - thin debris disk can be found in @xcite , @xcite , and @xcite . to model these structures , we use grain properties for astronomical silicates with a density of 2.5 g @xmath27 and a @xmath62 size distribution appropriate for a steady - state collisional cascade @xcite . using the sed of the excess and the resolved images as constraints
, we fit the minimum and maximum grain sizes ( @xmath63 and @xmath64 ) in the size distribution and the inner and outer radii of each of the disk components ( @xmath65 and @xmath66 ) by assuming a constant surface density ( @xmath67 ) disk for the strongly bound disk and @xmath68 for the weakly - bound or unbound disk ( i.e. , the grains affected by photon pressure ) . from its characteristic temperature ( @xmath0150 k ) and
figure [ radius_td ] , the inner radius of the warm component can be as close as @xmath06 au if the grains have a size of @xmath010 @xmath3 . however , the spectral shape of the 10 to 20 @xmath3 excess emission is better fit with grains of @xmath0@xmath3 sizes .
in addition , the spectral shape suggests that sub - micron grains are absent , since their emission spectrum would result in higher contrast features at 10 and 20 @xmath3 .
we fit the inner warm component as a bound disk with parameters used in table [ model_param ] .
the dust mass only accounts for the small grains that dominate the mid - ir emission .
we lack the constraints , such as the flux of this warm component at longer wavelengths , that would be required to estimate the mass of the population of large grains and parent bodies associated with the warm emission .
it is less straightforward to fit the other two disk components , due to the lack of strong constraints on the size of the planetesimal disk .
we can , however , use the following information : ( 1 ) we know roughly how much flux comes from the unresolved component ( presumably the planetesimal disk ) at 70 @xmath3 ; ( 2 ) the minimum inner radius of the planetesimal disk is @xmath090 au , from the characteristic temperature of the cold component in the sed ; and ( 3 ) large grains ( at least a few hundred microns ) are required due to the collisional cascade nature of debris disks .
the model parameters not only have to provide satisfactory fits to the global sed , but the model images after convolution with the beam sizes have to match the observed images as well . our modeling strategy is similar to the one applied to the @xmath11 oph disk @xcite ; we first tried many combinations of parameters that can provide good fits to the sed , then refined the parameters by matching the observed radial profiles .
we started with constraining the parameters for the inner component using data shortward of 30 @xmath3 since they are most sensitive to this component , then used the constraints listed above for the parameters in the other two disk components . for the planetesimal disk
, we fixed @xmath64 and @xmath69 at 1000 @xmath3 and 90 au , respectively .
additionally , we constrained @xmath63 to be greater than @xmath26 ( @xmath02 @xmath3 ) . for the extended halo disk , @xmath70 was fixed at 1000 au ; @xmath71 and @xmath72 were required to be equal ( there is no gap between the planetesimal and halo disks ) .
initially , we fit the planetesimal disk with a narrow ring ( @xmath73 ) so the actual density distribution has less effect in the output sed .
a difficulty in this narrow ring model is that grains smaller than @xmath26 have relatively high thermal equilibrium temperatures at the location where the extended disk starts , and the resultant emission from the inner edge of the halo component is too strong in the 2535 @xmath3 range . the only way to make
the narrow ring model work is to have far less contribution from the extended disk as shown in the sed of figure [ models]a ( the top panel ) .
however , the resulting model yields insufficient flux from the extended component , and therefore , also shows a disk at 70 @xmath3 that is too small , as shown in the model radial profile ( fig .
[ models]a ) . to satisfy all the constraints listed above simultaneously
, the outer radius of the planetesimal disk has to extend to @xmath0300 au .
figure [ models]b shows one of the best - fit models in terms of sed and radial profiles . in this model , the planetesimal disk extends from @xmath69=90 au to @xmath72=300 au and consists of grains of @xmath63=10 @xmath3 to @xmath64=1000 @xmath3 with a total dust mass ( @xmath74 ) of 1.2@xmath110@xmath75 .
the halo disk extends from @xmath71=300 au to @xmath70=1000 au and consists of grains of @xmath63=1 @xmath3 to @xmath64=10 @xmath3 with a total @xmath74=1.9@xmath110@xmath76 .
note that the reduced @xmath54 values in the sed fitting are similar between the narrow - ring and broad - disk models ; however , the model flux ratio at 70 @xmath3 between the extended component ( halo ) and the unresolved core ( planetesimal disk ) is too low compared to the observed ratio ( @xmath01 ) .
this mis - match is best shown in the model surface brightness profile at 70 @xmath3 ( bottom panel of fig .
[ models]a ) , where the narrow - ring model does not fit the 70 @xmath3 profile at all .
the parameters used and derived in our thermal model for the disk components are listed in table [ model_param ] .
because of the ambiguous nature of sed modeling with multiple components and low - resolution images , these model parameters are not unique but were built from the least number of components with simple assumptions that produced a good match to all the available data .
one should not take the model grain sizes too literally since they depend greatly on the grain properties used .
however , the total dust mass in each of the components remains similar among all of the acceptable models .
cccc @xmath77 & @xmath78 & @xmath78 & @xmath79 + @xmath65 ( au ) & 6 & 90 & 300 + @xmath66 ( au ) & 15 & 300 & 1000 + @xmath63 ( @xmath3 ) & 1.5 & 10 & 1 + @xmath64 ( @xmath3 ) & 4.5 & 1000 & 10 + @xmath74 ( @xmath80 ) & 1.1@xmath110@xmath81 & 1.2@xmath110@xmath16 & 1.9@xmath110@xmath8 + @xmath82 & 2.2@xmath110@xmath83&7.5@xmath110@xmath83&1.0@xmath110@xmath84
the architecture of the hr 8799 planetary system is complex : a warm inner asteroid - belt analog located at a radius of 615 au ; a cold kuiper - belt - analog planetesimal disk located from @xmath090 au up to @xmath0300 au ; three massive planets orbiting between the two disk components ; and a prominent halo of small grains extending up to @xmath01000 au . similar components
( except for imaged planets ) have been found in a number of other debris disk systems , indicating some underlying order within the diversity in debris disk structures .
assuming the grains in the inner belt are astronomical silicates , we can rule out grains of sub - micron size , due to the absence of strong silicate features at 10 and 20 @xmath3 .
the shape of the irs spectrum strongly favors micron sizes for the grains that dominate the emission .
if we further assume that all the grains in this component are bound ( a@xmath85 = a@xmath86 ) , then the outer radius of this warm component can be as small as @xmath010 au based on temperature arguments ( see the 2 @xmath3 curve in fig .
[ radius_td ] ) .
we also find that the maximum inner radius of this component is around @xmath06 au , again from temperature arguments .
there are few , if any , grains inside this radius .
an upper limit on the dust mass inside this inner warm component can be estimated based on the observed irs excess spectrum .
a flux density of 1 mjy at 10 @xmath3 corresponds to a dust mass of 2.3@xmath110@xmath87 assuming astronomical silicates of @xmath30=10 @xmath3 at 3 au ( i.e. , t@xmath88200 k ) and a grain density of 2.5 g @xmath27 .
this inner hole can be maintained by ice sublimation if the grains are icy @xcite ; a process that will occur at a dust temperature of @xmath0150 k. alternatively , the lack of grains inside @xmath06 au suggests either that there is another ( unseen ) interior planet maintaining a dust - free zone , or that the growth of the protoplanets in the terrestrial zone ( 320 au for an a - type star ) at an early stage initiated a collisional cascade that moved outward and created this inner hole @xcite . since we lack constraints such as a measurement of the long wavelength emission from this inner component at higher spatial resolution , we can not set limits on @xmath64 . because the dust in this component is warm ,
very little mass is required , only about 1.1@xmath110@xmath89 for dust with an infrared fractional luminosity ( @xmath90 ) of 2.2@xmath110@xmath83 .
these values are rough lower limits because it is difficult to place tighter constraints without knowing additional parameters such as @xmath64 .
we also fit the inner warm component with amorphous carbon grains ( density of 1.85 g/@xmath91 , @xcite ) . as expected from the discussion in
[ sec_models ] , the model results for the placement of the disk are not very sensitive to the assumed grain composition ( @xmath92 au and @xmath9315 au ) , and we can also rule out a significant amount of sub - micron carbonaceous grains in this inner component because the equilibrium temperature of 150 k places them right at the location of planet c. a similar result of low sensitivity to grain properties is also found in @xcite , who analyzed the behavior of the vega system for grains of silicate , carbonaceous , and mixed composition . a similar inner warm component ( t@xmath94150 k ) that dominates the disk emission at the irs and mips 24 @xmath3 wavelengths
is seen relatively commonly among resolved debris systems .
for example , hr 4796a , fomalhault , and @xmath34 eri disks all have such a component @xcite .
in fact , all of their excess seds look very similar : a steep inflection between the irs spectrum and mips - sed mode data , indicating a relatively sharp inward - facing edge to the structure dominating the far - infrared emission . this feature may arise from the action of massive planets , as appears to be the case for fomalhaut @xcite .
hd 32297 and @xmath95 tel are other systems that show similar characteristics revealed by recent ground - based imaging @xcite .
in addition , a recent irs spectroscopic study of unresolved debris disks shows that warm components are very common ( @xmath050% ) among a - type excess stars @xcite . in the case of hr 8799
, there are strong indications that this warm component is physically separated from the cold , outer component and that there is little dust in the intervening region .
our best - fit model appears to indicate too much flux in the submillimeter ( fig .
[ models ] ) for the cold planetesimal disk .
however , the model 850 @xmath3 disk has a fwhm of 178 after being convolved with the nominal 850 @xmath3 beam size of 15 .
this slight extension would not be resolved by the 850 @xmath3 observation because the data were taken in photometry mode and no mapping was done .
the flux density reported in @xcite was calibrated assuming a point source . therefore , the correct way to compare the model and observed fluxes is to integrate the total flux in a 15 diameter aperture on the model image .
the 850 @xmath3 model flux within a diameter of 15 is 9.33 mjy , consistent with the scuba measurement of [email protected] mjy .
similarly , the 1.2 mm flux is also under - estimated for the source extension since it was obtained using the iram 30 m single dish telescope ( fwhm=95 ) .
the dust mass in the planetesimal disk that is derived from our three - component model is 0.12 @xmath80 , consistent with the mass derived from the 850 @xmath3 observation ( 0.1 @xmath80 , @xcite ) .
this is only twice as large as the mass required by the narrow ring model , even though the disk is much larger .
the dust mass in this cold component is similar to values derived for disks in the 12 myr - old @xmath96 pic moving group ( @xmath96 pic and hd 15115 , @xcite ) .
the amount of the excess emission in the hr 8799 disk ( @xmath97=1.5 and @xmath98=94 , compared to the stellar photosphere ) is comparable with other debris disks around a - stars of similar age ( 20160 myr ) @xcite , at least within statistical expectations .
the infrared fractional luminosity for this component is 7.5@xmath110@xmath83 , suggesting a collision - dominated disk .
since the cold component is the birth place for the second - generation debris from collisional cascades , the presence of large grains is required . the inner edge of the disk ,
although not directly resolved by imaging , has a minimum radius of @xmath090 au from temperature arguments .
we lack high - spatial - resolution submillimeter data to put strong constraints on the outer edge of the planetesimal disk .
the outer radius of @xmath0300 au in our model is based on the fewest and most straightforward assumptions .
finally , we consider the halo component of the system .
the general appearance and extent of this component are similar to the halo around the vega system . in that case , the _ spitzer _
observations provided sufficient angular resolution to fit the surface brightness profile and show that it is consistent with expectations for grains being blown out by radiation pressure . while the details of this process are complex , it is generally parametrized by @xmath96 , the ratio of radiation pressure force to gravitational force .
when collisions produce grains with @xmath96 @xmath99 1 , they will be blown out of the system quickly .
when the collisional cascade produces grains over a narrow range of @xmath96 below unity , they will tend to go into elliptical orbits .
the behavior of these barely - bound grains has been suggested to explain the broken power - laws seen in the scattered - light disk surface brightness profile around au mic @xcite and @xmath96 pic @xcite .
however , these highly - elliptical orbits will continue to cross the parent body zone and eventually it is likely that these grains will collide again , be broken into smaller ones with @xmath100 and be blown out of the system .
thus , the time scale for mass loss from the system depends on the details of the collisional fragmentation of the grains
. however , qualitatively , the halo around hr 8799 resembles the behavior of vega closely and differs substantially from the observations of most other a - stars , indicating that both systems are currently undergoing an enhanced level of dynamical stirring in their outer planetesimal disks that leads to an elevated production rate for small grains with @xmath96 near unity .
these grains will eventually be ground down to sizes that participate in a general outflow from the system .
we therefore model the halo by making assumptions similar to those used for the vega outflow @xcite . to fit the vega outflow ,
grains extending up to about five times the nominal blowout size were required .
this apparent discrepancy probably arises because , for simplicity , the models are based on solid , spherical grains , whereas real grains are likely to be non - spherical and complex in structure with a substantial volume of voids ( e.g. , @xcite and references therein ) .
the influence of photon pressure will be much larger for such grains than for solid spherical ones .
we use @xmath101 @xmath3 in the extended halo component , about five times larger than the blowout size for hr 8799 , and other parameters of the model are similar to those used for vega @xcite .
the estimated dust mass in the halo is relatively large , 1.9@xmath102 that is @xmath06 times greater than the mass in the vega outflow .
the terminal radial velocity is 24 km s@xmath16 for grains having @xmath103 and that are being produced in the planetesimal disk , 100300 au from a 1.5 m@xmath104 star @xcite . for a disk radius of 1000 au ,
the residence time for such particles is @xmath0 9002400 yr .
for a static collisionally dominated disk @xcite , the expected ratio of mass in blown out grains to that of grains in the bound disk is @xmath01.2% in the hr 8799 system ( i.e. , a 1.5 m@xmath105 star with a fractional disk luminosity of 8@xmath106 and @xmath107 100 au ) .
the observed ratio for hr 8799 is 15 times higher , qualitatively similar to the vega system , where the mass in blown out grains is about 30 times the predicted static level .
both systems have much more prominent halos than systems such as fomalhaut @xcite and @xmath96 leo ( stock et al . in prep . ) .
this comparison indicates that there is an elevated level of dynamical activity in the bound disks of vega and hr 8799 that is enhancing the production of small grains .
although we have proceeded by analogy with vega , future work should address the role of weakly bound grains in more detail .
the conclusion that the halo indicates more dynamical activity than in a stars without such halos should be robust , but more complete modeling should draw out other aspects of the hr 8799 system . based on the `` resonance overlap '' condition in the planar circular restricted three body problem , the width of the chaotic region , @xmath108 , in the vicinity of a planet with mass of @xmath109 and orbital semi - major of @xmath110 orbiting a star with mass of @xmath111 is given by @xcite , @xmath112 adopting the nominal parameters from @xcite , stellar mass 1.5 @xmath25 , planet masses 7 @xmath113 and 10 @xmath113 , and semi - major axes 24 or 68 au for planets d and b , respectively , the width of the unstable zone near each planet is is @xmath08 and @xmath023 au , respectively .
therefore , the locations of the outer edge of the inner dust belt ( @xmath015 au ) and the inner edge of the outer belt ( @xmath090 au ) are both consistent with gravitational sculpting by the innermost known planet d and the outermost known planet b , respectively .
the nearly dust - free zone interior to @xmath06 au may indicate the presence of additional unseen inner planets .
if so , then this leads to an estimate of the timescale of inner planet formation of @xmath520160 myr given the age of hr 8799 .
this is in good agreement with theoretical calculations for terrestrial planet accretion @xcite as well as isotopic constraints on the age of the earth @xcite .
the existence of the halo indicates that the outer planetesimal disk is heavily stirred due either to : ( 1 ) the planets not being in a stable configuration @xcite ; ( 2 ) planet migration and orbital resonance crossings that are causing extreme excitation in the planetesimal disk , as in models for the kuiper belt @xcite and for the late heavy bombardment @xcite ; or ( 3 ) processes in the planetesimal disk itself , such as ongoing formation of ( unseen ) ice giants or planetary embryos that stir the disk @xcite .
the first two possibilities relate the dynamical activity in the planetesimal disk to issues associated with the stability of the configuration of the three known planets , while the third hypothesis would be independent of the planets .
although no specific conclusions can be drawn , the outcomes of any of these scenarios are a heavily stirred planetesimal disk that produces an era of enhanced collisional activity .
the resulting collisional cascades and avalanches will feed an outflow of small grains @xcite .
[ a160_co32 ] the cirrus seen at 160 @xmath3 resembles some of the large - scale structure in the co ( 3 - 2 ) cloud detected by @xcite .
figure [ a160_co32 ] shows a new , 7@xmath17 co ( 3 - 2 ) map at 15 resolution around hr 8799 , taken with the harp heterodyne array on the jcmt , overlaid on the _ spitzer _ 160 @xmath3 map .
the co emission is integrated in the local standard of rest ( lsr ) velocity from -4 to -6 km s@xmath16 .
the radial velocity of the star , [email protected] km s@xmath16 , from @xcite is in the heliocentric velocity frame , however , and converts to -4.6 km s@xmath16 lsr .
thus , contrary to the conclusion in @xcite , we can not say that the cloud is not associated with hr 8799 and can not place any limits on the gas content of the disk from these observations .
this is also further confirmed by the new iram 30-m observation that found strong co ( 2 - 1 ) line emission due to the cloud at the heliocentric radial velocity of hr 8799 ( hily - blant , kastner , forveille & zuckerman , in prep . )
this new jcmt co ( 3 - 2 ) map is larger and of higher quality than the map presented in @xcite and shows the extended , low level molecular emission in more detail .
it appears to be associated with the extended arc of high latitude clouds , mbm 53 - 55 , detected in co ( 1 - 0 ) by @xcite .
the distance of these objects and their relation to one another is not well known .
if the association with the star is physical then , at 40 pc , this would be the closest known molecular cloud .
whether physically associated or not , it is likely that the low flux levels ( 1 - 3@xmath9 ) of the 160 @xmath3 emission come from the molecular cloud .
the asymmetry in the outer part of the observed source at this wavelength may also be due to cloud contamination .
nevertheless , the detected source ( with the peak flux @xmath99 8-@xmath9 ) is coincident with the expected stellar position , which is offset ( @xmath020 ) from the central emission of the nearby clump in the cloud .
the cloud therefore contributes only a very small amount of flux that can not be completely removed from our aperture photometry . at 70 @xmath3
, there is no resemblance to the large - scale background structure in the data nor an asymmetry seen in the source morphology ; therefore , we can rule out any significant contribution from this extended background cloud to the 70 @xmath3 disk emission .
the arrow in figure [ a160_co32 ] shows the direction of proper motion of hr 8799 ( @xmath114ra = 107.93 mas yr@xmath16 and @xmath114dec = 49.63 mas yr@xmath16 ; @xcite ) .
given the coincidence in space and velocity , it is plausible that accretion from the cloud might explain the abundance anomalies leading to the @xmath12 bootis designation @xcite .
using equation ( 16 ) from @xcite for bondi - hoyle accretion implies a mass accretion rate of @xmath115 with an assumed density of 100 @xmath27 , just above the threshold to produce @xmath12 bootis behavior @xcite .
however , this calculation assumes a fluid model for the ism , which is not strictly applicable at the expected density @xcite .
therefore , the accretion rates may be significantly lower than indicated @xcite .
the accretion hypothesis rests on the untested assumption that the rate can be increased through such means as small scale high - density structures in the ism and/or increases in density in the wake of the star .
we have obtained deep infrared images and spectra of the debris system around hr 8799 . to our surprise at the distance of @xmath040 pc , the hr 8799 disk is clearly resolved at both 24 and 70 @xmath3 ; the outer boundary of the disk can be traced further than 1000 au . from general ,
model - independent considerations , the star appears to have one zone of warm dust within the orbit of its innermost known planet , another broad zone of cold dust outside the orbit of its outermost known planet , and an extended halo of small grains from this outer zone .
we have constructed detailed models of the system , guided by models of other debris systems as well as the observations of hr 8799 .
although the models are not unique , because we draw on experience in modeling many debris systems , they should give a reasonably accurate picture of conditions in the hr 8799 system .
the models indicate that the inner zone lies between 6 and 15 au with a total dust mass of 1.1@xmath110@xmath89 and the outer zone is between 90 and 300 au with a dust mass of 1.2@xmath110@xmath75 .
there is little dust inside the inner zone ( @xmath1166 au ) and in between the inner and outer cold zones where the three planets reside .
the halo extends to at least 1000 au and has a dust mass of 1.9@xmath110@xmath76 , about 15 times the level expected for a static , quiescent debris system .
the implied high level of dynamical activity in the hr 8799 debris system may be related to perturbations imposed by its system of three massive planets or may be signaling ongoing planet formation in the outer parts of this system .
hr 8799 is at an age similar to that of the solar system when the terrestrial planets formed , along the way to settling into the final configuration .
the planetary system of hr 8799 ( both planets and debris disk ) provide an intriguing snapshot of processes occurring at this stage and may help us understand the formation and evolution of our own planetary system .
this work is based on observations made with the _
spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology .
support for this work was provided by nasa through contract 1255094 and 1256424 issued by jpl / caltech to the university of arizona .
we thank chian - chou chen for assistance with the jcmt co data , david wilner and eric mamajek for discussion of the background cloud , and ben zuckerman for pointing out the error in the velocity comparison of the star and the cloud .
ks also thanks glenn schneider and mike meyer for the useful discussion .
this research has made use of the simbad database , operated at cds , strasbourg , france . | we have obtained a full suite of _ spitzer _ observations to characterize the debris disk around hr 8799 and to explore how its properties are related to the recently discovered set of three massive planets orbiting the star .
we distinguish three components to the debris system : ( 1 ) warm dust ( t @xmath0 150 k ) orbiting within the innermost planet ; ( 2 ) a broad zone of cold dust ( t @xmath0 45 k ) with a sharp inner edge , orbiting just outside the outermost planet and presumably sculpted by it ; and ( 3 ) a dramatic halo of small grains originating in the cold dust component . the high level of dynamical activity implied by this halo may arise due to enhanced gravitational stirring by the massive planets .
the relatively young age of hr 8799 places it in an important early stage of development and may provide some help in understanding the interaction of planets and planetary debris , an important process in the evolution of our own solar system . |
multicomponent systems are of great theoretical and practical importance .
an example of an important ternary system , that inspired the current paper , is the formation of polymer membranes through immersion precipitation @xcite . in this process
a polymer - solvent mixture is brought in contact with a non - solvent . as the non - solvent diffuses into the mixture , the mixture phase - separates , leaving behind a complex polymer morphology which depends strongly on the processing conditions .
the dependence of the morphology on the parameters of the system is as yet poorly understood .
preliminary lattice boltzmann simulations of this system exist @xcite
. however , this work did not recover the correct drift diffusion equation .
a general fully consistent lattice boltzmann algorithm with an underlying free energy to simulate multicomponent systems is still lacking .
this paper strives to bring us a step nearer to achieving this goal .
there are several previous lattice boltzmann methods for the simulation of multi - component systems .
there are three main roots for these approaches
. there are those derived from the rothmann - keller approach @xcite that attempt to maximally phase - separate the different components .
a second approach by shan and chen is based on mimicking the microscopic interactions @xcite and a third approach after swift , orlandini and yeomans @xcite is based on an underlying free energy .
all of these have different challenges .
since we are interested in the thermodynamics of phase - separation we find it convenient to work with a method based on a free energy .
this allows us to easily identify the chemical potentials of the components .
this is convenient since the gradients of the chemical potentials drive the phase separation as well as the subsequent phase - ordering .
the challenge for the lb simulation of a multicomponent system lies in the fact that momentum conservation is only valid for the overall system but not for each component separately , and diffusion occurs in the components . for a binary system of components
@xmath0 and @xmath1 with densities @xmath2 and @xmath3 , the simulation usually traces the evolution of the total density @xmath4 and the density difference @xmath5 @xcite .
although this scheme is successful in the simulation of a binary system @xcite , its generalization for the lb simulations of systems with an arbitrary number of components is asymmetric .
for instance , to simulate a ternary system of components @xmath0 , @xmath1 , and @xmath6 with densities @xmath2 , @xmath3 and @xmath7 , the total density of the system , @xmath8 , should be traced , and the other two densities to be traced may be chosen as , e.g. , @xmath3 and @xmath9 @xcite . this approach is likely to be asymmetric because the three components are treated differently as is the case of lamura s model @xcite .
if an lb method is not symmetric , it will lose generality an will only be adequate for special applications . in this paper , we established a multicomponent lattice boltzmann method based on an underlying free energy that is manifestly symmetric .
the equation of motion for a multicomponent system are given by the continuity and navier - stokes equations for the overall system and a drift diffusion equation for each component separately .
the continuity equation is given by @xmath10 where @xmath11 is the mass density of the fluid , @xmath12 is the mass flux which is given by @xmath13 , and @xmath14 is the macroscopic velocity of the fluid .
the navier - stokes equation describes the conservation of momentum : @xmath15 where @xmath16 and @xmath17 are the pressure and viscous stress tensors respectively , @xmath18 is the component @xmath19 of an external force on a unit mass in a unit volume , and the einstein summation convention is used . for newtonian fluids , the viscous stress tensor is given by @xmath20 where @xmath21 is the shear viscosity , and @xmath22 is bulk viscosity , and @xmath23 is the spacial dimension of the system .
free energy , chemical potential , and pressure are key thermodynamic concepts to understand the phase behavior of a system .
the chemical potential of each component can be obtained by a functional derivative as @xmath24 where @xmath25 is the chemical potential of component @xmath26 ; @xmath27 is the number density of component @xmath26 ; and @xmath28 is the total free energy of the system .
the pressure in a bulk phase in equilibrium is given by @xmath29 the pressure tensor is determined by two constraints : @xmath30 in the bulk and @xmath31 everywhere . in multicomponent systems , there are two mechanisms for mass transport : convection and diffusion .
convection is the flow of the overall fluid , while diffusion occurs where the average velocities of components are different .
the velocity of the overall fluid is a macroscopic quantity because it is conserved , but the average velocities of the components are not .
the macroscopic velocity of the fluid @xmath32 can be expressed in terms of the density @xmath33 and velocity @xmath34 of each component in the form of @xmath35 with the notation @xmath36 the flux of each component can be divided into a convection part @xmath37 and a diffusion part @xmath38 : @xmath39 because mass conservation still holds for each component , the continuity equation for each component is valid : @xmath40 substituting eq .
( [ jjj ] ) into eq .
( [ onecon ] ) , the convection diffusion equation for a component can be obtained .
@xmath41 from eqs .
( [ uaverage ] ) and ( [ dsl ] ) , we see that @xmath42 which ensures the recovery of the continuity equation for the overall system .
the diffusion process between two components is related to the difference of the chemical potential of the two components , which is also called the exchange chemical potential @xcite . recognizing that the gradient of the exchange chemical potential determines the diffusion processes , we obtain a first order approximation for the diffusion flux of one component into all other components as @xmath43 where @xmath26 and @xmath44 enumerate the components ; @xmath25 and @xmath45 are the chemical potentials of components @xmath26 and
@xmath44 ; and @xmath46 is a symmetric positive definite mobility tensor .
a simple model for the diffusion process assumes that a diffusion flux between two components is proportional to the overall density and the concentration of each component .
then mobility tensor can be expressed as @xmath47 where @xmath48 is the constant diffusion coefficient between components @xmath26 and @xmath44 .
it depends on components but is independent of the total densities and concentration of each component . substituting eq .
( [ mss ] ) into eq .
( [ akak ] ) , we have @xmath49 substituting eq .
( [ bfjj ] ) into eq .
( [ condieq ] ) , the general form of a convection diffusion equation is obtained as @xmath50
to simulate a multicomponent fluid using lb we set up a lb equation for each component .
the lbe for a component @xmath26 of a multicomponent system is given by @xmath51 , \label{sslb}\end{aligned}\ ] ] where @xmath52 is the particle distribution function with velocity @xmath53 for component @xmath26 , @xmath54 is its equilibrium distribution and @xmath55 is the forcing term of component @xmath26 due to the mean potential field generated by the interaction of the component @xmath26 with the other components . the main task in setting up this lattice boltzmann method is to determine the correct form of the forcing term @xmath55 which will recover the convection diffusion equation ( [ gfcd ] ) .
the density of each component and the total density are given by @xmath56 the average velocity of one component @xmath26 and the overall fluid can be defined as @xmath57 where @xmath58 is the average velocity of the component @xmath59 , and @xmath60 is the average velocity of the overall fluid .
the moments of equilibrium distributions for one component are chosen to be @xmath61 the moments for the forcing terms of one component are @xmath62 to utilize the analysis of the one component system we can establish a lb equation for the total density by defining @xmath63 similar to the counterparts of the one - component system , the moments for the overall equilibrium distribution function are given by @xmath64 the moments for the overall force terms are then given by @xmath65 using eq .
( [ ff3 ] ) , we obtain @xmath66 where the second term of eq .
( [ olso ] ) is of a higher order smallness than the first terms , and therefore does not enter the hydrodynamic equations to second order . for the third moment
we have @xmath67 by summing eq .
( [ sslb ] ) over @xmath26 , an effective lb equation for the total density is @xmath68 , \label{sglb}\end{aligned}\ ] ] this is identical to the lb equation for a system of one component .
therefore , the continuity equation and the navier stokes equation of the overall fluid of a multicomponent system are recovered as @xmath69 where @xmath70 is the macroscopic velocity of the fluid . the navier stokes equation for the overall
fluid is : @xmath71 to recover the convection diffusion equation of each component , we performed a taylor expansion on the left of eq .
( [ sslb ] ) to second order : @xmath72 because of the recursive nature of eq .
( [ s2 lb ] ) , @xmath73 can be expressed by @xmath74 and derivatives of @xmath75 as @xmath76 substituting eq .
( [ bfe ] ) into the left side of eq .
( [ s2 lb ] ) we obtain @xmath77 summing eq .
( [ s3 lb ] ) over @xmath78 gives , @xmath79 the first moment of @xmath80 and @xmath81 are not identical , and the continuity equation can not be obtained .
( [ sgao ] ) shows that @xmath82 is of order @xmath83 , and @xmath84 is of order @xmath85 .
therefore @xmath86 is of order @xmath87 , and we get @xmath88 + o(\epsilon^3 ) .\end{aligned}\ ] ] so eq .
( [ sgao ] ) can be simplified to @xmath89 = 0 .
\label{ssim}\end{aligned}\ ] ] eqs .
( [ snse ] ) yields @xmath90 from eq .
( [ sgao ] ) it follows that @xmath91 inserting eqs .
( [ stu ] ) and ( [ sxx ] ) into eq .
( [ ssim ] ) we get @xmath92 substituting eq .
( [ jeep ] ) into eq .
( [ ssim ] ) results in @xmath93 , \label{geq}\end{aligned}\ ] ] from this we deduce that the correct form of the forcing term is @xmath94 where the coefficient is @xmath95 .
this coefficient approaches 1 as @xmath96 approaches 0 , as one would expect from the continuum limit .
this constitutes the main result of this paper .
plugging eq .
( [ definef ] ) into eq .
( [ geq ] ) , we then obtain the convection diffusion equation @xmath97 . \label{woche}\end{aligned}\ ] ] the diffusion flux of component @xmath26 is @xmath98 so that the @xmath99 in eq .
( [ gogogo ] ) is equivalent to @xmath100 in eq .
( [ bfjj ] ) .
we examined the equilibrium behavior of phase separated binary and ternary systems .
we used the flory - huggins free which is a very popular model to study polymer solutions .
it is given by @xmath101 where @xmath102 is the polymerization of the component @xmath26 , @xmath27 is its number density , and @xmath103 is its volume fraction .
it is defined as @xmath104 where @xmath105 is the mer density of component @xmath26 and @xmath106 is the mer density of the system , which is a constant in the flory - huggins model . to validate our algorithm
we compared the binodal lines obtained by our algorithm to the theoretical ones obtained by minimizing the free energy .
we used the interfacial tension parameter @xmath107 in all our lb simulations of binary and ternary systems , because there is an intrinsic surface tension in the lb simulation due to higher order terms @xcite , which did not appear explicitly in the second order taylor expansion presented in this paper .
since we are only evaluating the phase - behavior here we use a one dimensional model known as d1q3 .
this model has the velocity set @xmath108 .
this is an important test since all other frequently used higher dimensional model have this d1q3 model as a projection .
we consider two binary systems : a monomer system with @xmath109 and @xmath110 , and a polymer system with @xmath111 and @xmath112 .
for both systems , the total density was @xmath113 . throughout this paper
we choose the self interaction parameters to vanish : @xmath114 .
the critical volume fractions for the monomer system are @xmath115 and @xmath116 and for the polymer systems are @xmath117 and @xmath118 . to induce phase separation a small sinusoidal perturbation @xmath119
was added in the initial conditions .
the amplitude of the perturbation is 0.1 and its wavelength is the lattice size . the initial volume fraction of component a is given by @xmath120 .
the initial volume fraction of component b is given by @xmath121 .
the monomer system was simulated with different inverse relaxation times . in figure [ binary_binodal ]
we show results for @xmath122 , and @xmath123 .
we see that the equilibrium densities have only a very slight dependence on the relaxation time , although the range of stability depends noticeably on the relaxation time .
the polymer system was simulated with only one inverse relaxation time of @xmath124 .
starting from the critical point and increasing the @xmath125 value for each initial condition until the simulations were numerically unstable , we obtained a pair of binodal points for each initial condition .
the system reached a stable state after about 5000 time steps .
the measurement were taken after 50000 time steps to be sure that an equilibrium state had been reached .
for the polymer system , figure [ lbchem ] shows the comparison of the total density , and the volume fractions and chemical potentials of each component to the corresponding theoretical values .
the total density of a system in equilibrium by lb is essentially constant with a variation of @xmath126 .
the volume fractions of each component in the lb simulation agree well with the theoretical values .
the chemical potential of each component by the lb simulation was very close to the theoretical value .
the chemical potential @xmath127 , corresponding to the polymer component , varied slightly with a difference for the bulk values of about @xmath128 and a variation in the interface of about @xmath129 .
this is the underlying reason for the small deviation from the theoretically predicted concentration .
the potential @xmath22 was nearly constant with a variation of less than @xmath130 in the bulk and a variation of about @xmath131 at the interface . for large values of @xmath125
this discrepancy increases leading to the noticeable variation of the equilibrium densities of the polymer system as shown in figure [ binary_binodal ] .
we also performed lb simulations with two ternary systems : a monomer system with @xmath109 , @xmath110 , and @xmath132 and a polymer system with @xmath133 , @xmath112 , and @xmath134 .
the @xmath135 parameters for both systems were @xmath136 , @xmath137 , and @xmath138 .
the other @xmath135 parameters were zero .
the inverse relaxation time constant for both simulations was @xmath139 .
the critical point for the monomer system was @xmath140 , @xmath141 , and @xmath142 .
the critical point for the polymer system was @xmath143 , @xmath144 , and @xmath145 .
the initial state of each simulation were set from the critical points towards the end point ( @xmath146 , @xmath116 , @xmath147 ) .
initially a small sinusoidal wave perturbation @xmath148 of an amplitude of 0.1 and wavelength of the lattice size was superimposed on the initial volume fraction of the a component .
this perturbation was subtracted from the b component and the c component was constant .
i performed a lb simulation for each set of initial volume fractions and obtained the volume fractions of the two phases in the equilibrium state , resulting in two binodal points .
the simulation reached a stable state after about 20,000 time steps .
the measurements were taken after 200,000 time steps to make sure the equilibrium state was reached .
figure [ trilbm10phase ] shows the comparison of the binodal points by lb simulation to the theoretical binodal lines of both systems .
the binodal points obtained by the lb simulation agree fairly well with the theoretical binodal lines for the monomer and polymer systems .
the simulation becomes unstable when @xmath149 is close to zero , i.e. when one component is nearly depleted . in this region
the simulation results also deviate noticeably from the theoretical binodal lines . immediately near the critical point ,
the evolution of the system becomes extremely slow so the slight deviation between the binodal points obtained through the lb simulation and the theoretical ones probably indicates that the lb simulation was not yet fully equilibrated .
for the polymer system figure [ trilbm10ch ] shows a comparison of the volume fractions and chemical potentials of each component .
the total density of the system is again nearly constant with variation of less than @xmath150 in the bulk . at the interface
there is a small variation of @xmath151 .
the volume fractions of each phase in the simulation were very close to their theoretical values .
the chemical potential of component a was slightly different in two phases with a variation of about @xmath128 , while the chemical potentials of components b and c were much closer in the two phases with a variation of less than @xmath130 .
we have presented a general lattice boltzmann algorithm for systems with an arbitrary number of components which is based on an underlying free energy . in this algorithm
the key thermodynamic quantities such as the chemical potentials of the components are immediately accessible .
it is also manifestly symmetric for all components .
we tested the equilibrium behavior of the new algorithm for two and three component systems in each case examining both the case of monomer and polymer mixtures with an underlying flory - huggins free energy .
we obtained to expected phase - diagrams to good accuracy and the chemical potentials were constant to good accuracy for the monomer systems .
polymer systems were more challenging to simulate but still obtained acceptable results for @xmath152 .
higher polymerizations , however , become increasingly difficult to realize with the current algorithm .
there are three directions in which we hope to extend this algorithm in the future .
the current algorithm does not allow for component dependent mobility .
we are working on developing an algorithm that can recover an arbitrary mobility tensor @xmath153 .
the chemical potential is only approximately constant .
recent progress for liquid - gas systems @xcite makes us hopeful that we will be able to ensure that the chemical potential is constant up to machine accuracy . and
lastly we hope to extend to model so that it can simulate polymer systems with significantly larger polymerization . | we present a lattice boltzmann algorithm based on an underlying free energy that allows the simulation of the dynamics of a multicomponent system with an arbitrary number of components .
the thermodynamic properties , such as the chemical potential of each component and the pressure of the overall system , are incorporated in the model .
we derived a symmetrical convection diffusion equation for each component as well as the navier stokes equation and continuity equation for the overall system .
the algorithm was verified through simulations of binary and ternary systems .
the equilibrium concentrations of components of binary and ternary systems simulated with our algorithm agree well with theoretical expectations . |
subluminous stars are considered to be core helium - burning objects with a thin hydrogen shell . in a hertzsprung - russell diagram these stars are situated on the extreme horizontal branch ( ehb ) .
the evolution of these stars is far from understood since the formation of an sdb requires extreme mass loss at the tip of the rgb .
the mechanism leading to this mass loss is still under debate @xcite .
formation scenarios can be roughly divided into single star and binary channels . about 50% of the known sdbs
are found to be in close binaries with periods of few hours to several days consistent with the predictions of the common envelope ( ce ) ejection scenario @xcite .
two main sequence stars evolve in a binary system .
the more massive one becomes a red giant first and fills its roche lobe .
unstable mass transfer leads to the formation of a ce .
the core of the red giant and the immersed secondary experience a loss of orbital energy which is deposited in the ce causing a spiral - in towards the center of mass .
as soon as enough energy has been deposited in the envelope , it is ejected .
the inwards migration stops and the stars end up in a close binary .
this scenario is well suited to explain the known population of close sdb binaries .
but how are the apparently single sdbs formed ?
@xcite suggested that substellar companions like planets or brown dwarfs may be able to trigger ce ejection as well . in order to survive the common envelope and avoid evaporation inside the envelope
the secondary should have a minimum mass of more than @xmath0 .
a similar scenario has been proposed for the formation undermassive single white dwarfs by @xcite .
several substellar objects have recently been detected around hot subdwarfs in wide orbits , showing that planets can survive the red giant phase of their host star @xcite .
recently , @xcite discovered a candidate substellar companion in close orbit around the bright sdb hd 149382 , which perfectly fits into the pattern predicted by soker .
systems like that may be quite common and could have remained unnoticed up to now .
the goal of our project is to put constraints on the population of the lowest mass companions to sdb stars by measuring accurate radial velocities from multi - epoch high resolution spectra .
up to now our survey consists of 23 bright sdb stars with visual magnitudes ranging from @xmath1 to @xmath2 .
we used high resolution spectra obtained with the feros instrument mounted at the eso / mpg-2.2 m telescope .
the spectra have a resolution of @xmath3 .
spectra were reduced , calibrated and corrected for earth s orbital motion with the midas package using the feros reduction pipeline .
all programme stars are single - lined objects without visible spectral features of a companion . furthermore , those binaries with high radial velocity variability have been excluded because their unseen companions must be quite massive .
each star has been observed several times ( from 2 to 15 ) on timescales ranging from one day to some years . as an example we display a section of the feros spectrum of hd 205805 in fig [ fig : uebers ] .
subdwarf b stars are hot ( @xmath4 ) and their atmospheres radiative .
the dominant features in their spectra are the hydrogen balmer and several helium lines .
but the spectra also show up to @xmath5 weak metal absorption lines which are very sharp .
these features are best suited to measure radial velocities .
the atmospheres of sdb stars are considered to be stable .
the only known effect other than a close companion that may lead to rv variability are pulsations .
pulsating sdb stars do indeed exist with periods between @xmath6 and @xmath7 and rv amplitudes of few km s@xmath8 at most @xcite .
for our analysis we chose a set of sharp , unblended metal lines situated between @xmath9 and @xmath10(see fig .
[ fig : fitlines ] , left hand panel ) .
accurate rest wavelengths were taken from the nist database .
depending on the atmospheric parameters of the star and the quality of the data a subset was used .
the rv measurement consisted of two steps .
first , we fitted gaussian and lorentzian profiles to each line separately using the spas routine ( hirsch priv .
the resulting mean rvs had standard deviations ranging from @xmath11 to @xmath12 .
second , we used the fitsb2 routine @xcite to perform a simultaneous fit of all suitable lines .
uncertainties were calculated using a bootstrapping algorithm and found to range from @xmath13 to @xmath14 , while both methods result in consistent mean rvs .
the scatter of individual rv values measured with spas around the mean value was reviewed for possible systematic effects like correlations with certain elements .
no such effects could be found . to check the wavelength calibration for systematic errors we used telluric features as well as night - sky emission lines ( see fig .
[ fig : fitlines ] , right hand panel ) .
having their origin on earth these lines should have zero rv .
the feros instrument turned out to be very stable .
usually corrections of less than @xmath14 had to be applied . in order to derive the degree of precision we can achieve with our method , we generated synthetic model spectra with realistic atmospheric parameters including poisson noise with @xmath15 and determined the rv in the way described above .
the uncertainties turned out to range from @xmath13 to @xmath16 .
the accuracy is limited by the resolution of the spectrograph , the number of features and the intrinsic thermal broadening of the lines .
four stars were found to be significantly ( another two marginally ) rv variable .
the shift in rv ranges from @xmath17 to @xmath18 ( see tab . [ tab : a ] ) . as an example we display two rv measurements for ec 20106@xmath195248 in fig .
[ fig : bigshift ] .
follow - up photometry and high resolution spectroscopy are necessary to exclude pulsational variability and derive the orbital parameters of these binaries .
constraints can then be put on the companion masses . even if there are no low mass companions we would expect to find a certain fraction of systems with small rv shifts caused by more massive secondaries seen at low inclination .
a larger sample is needed to exclude this case .
.summary of rv results for our sample obtained using two different algorithms .
rvs for each line were measured independently and averaged with spas ( left hand column ) whereas fitsb2 uses a simultaneous fit ( right hand column ) . [ cols="<,>,>,>,>",options="header " , ] no rv variations were found in the rest of the sample .
nevertheless , the errors allow to derive upper limits for @xmath20 of undetected substellar companions in close orbits from the binary mass function ( eq . [ eq:1 ] ) .
@xcite suggested that these objects should be more massive than @xmath0 and should have orbital periods shorter than @xmath22 .
adopting this maximum period and assuming a canonical mass of @xmath23 for the sdb , the errors of the rv measurements can be used as upper limits for the rv semi - amplitudes ( @xmath24 ) . as can be seen in fig .
[ fig : const ] , tight constraints can be put on possible substellar companions , if the measurement error is smaller than @xmath25 .
given data of perfect quality , close - in planets with masses as low as @xmath26 can be firmly excluded if our assumption about the orbital period is correct . in conclusion @xmath27 of the apparently single stars in our sample show small rv variations .
the true fraction is expected to be higher , because the accuracy is limited by the s / n in most cases .
companions with masses higher than @xmath0 in close orbits ( @xmath28 ) can be excluded for @xmath29 of our sample .
one possible explanation for the fractions found is that the assumed ce scenario is not the only channel for sdb formation .
however , even with ce ejection being the only channel , the companion may be destroyed during the ce phase either by evaporation or a merger with the stellar core leaving behind a single sdb .
better statistics and reliable predictions from theoretical calculations are required to solve this problem .
s. geier , h. edelmann , u. heber , and l. morales - rueda , _ apj _ * 702 * , l96l99 ( 2009 ) .
z. han , p. podsiadlowski , p. f. l. maxted , t. r. marsh , and n. ivanova , _ mnras _ * 336 * , 449466 ( 2002 ) .
u. heber , _
a&a _ * 155 * , 3345 ( 1986 ) .
j. w. lee , s. kim , c. kim , r. h. koch , c. lee , h. kim , and j. park , _ aj _ * 137 * , 31813190 ( 2009 ) .
r. napiwotzki , l. yungelson , g. nelemans , t. r. marsh , b. leibundgut , r. renzini , d. homeier , d. koester , s. moehler , n. christlieb , d. reimers , h. drechsel , u. heber , c. karl , and e. pauli , `` double degenerates and progenitors of supernovae type ia , '' in _ spectroscopically and spatially resolving the components of the close binary stars _ , edited by r. w. hilditch , h. hensberge , & k. pavlovski , 2004 , vol .
318 of _ astronomical society of the pacific conference series _ , pp . 402410 .
s. qian , l. zhu , s. zola , w. liao , l. liu , l. li , m. winiarski , e. kuligowska , and j. m. kreiner , _ apj _ * 695 * , l163l165 ( 2009 ) .
r. silvotti , s. schuh , r. janulis , j. solheim , s. bernabei , r. stensen , t. d. oswalt , i. bruni , r. gualandi , a. bonanno , g. vauclair , m. reed , c. chen , e. leibowitz , m. paparo , a. baran , s. charpinet , n. dolez , s. kawaler , d. kurtz , p. moskalik , r. riddle , and s. zola , _ nature _ * 449 * , 189191 ( 2007 ) . | the origin of subdwarf b ( sdb ) stars is not fully understood yet since it requires high mass loss at the red giant stage .
sdbs in close binary systems are formed via common envelope ejection , but the origin of apparently single sdb stars remains unclear .
substellar companions may be able to trigger common envelope ejection and help forming sdbs that appear to be single .
+ using a sample of high resolution spectra we aim at detecting small radial velocity ( rv ) shifts caused by such low mass ( sub-)stellar companions .
the rvs are measured with high accuracy using sharp metal lines .
our goal is to test the theoretical predictions and put constraints on the population of the lowest mass companions to sdb stars .
address = dr .
karl remeis observatory & ecap , astronomical institute , friedrich - alexander university erlangen - nuremberg , sternwartstr .
7 , d-96049 bamberg , germany address = dr .
karl remeis observatory & ecap , astronomical institute , friedrich - alexander university erlangen - nuremberg , sternwartstr
. 7 , d-96049 bamberg , germany address = dr .
karl remeis observatory & ecap , astronomical institute , friedrich - alexander university erlangen - nuremberg , sternwartstr
. 7 , d-96049 bamberg , germany address = australian astronomical observatory , po box 296 , epping , nsw , 1710 , australia |
the beginning of twentieth century launched a series of major paradigm shifts which heralded the era of modern physics .
it will perhaps be surprising to the modern reader that in the advent of the revolutionary einsteinian theory of relativity , maxwell and boltzmann s kinetic theory and planck s hypothesis of quanta , the scientific world was not convinced of the fact that matter is grainy and can not be continuously divided ad infinitum @xcite .
the seed of doubt was planted by the renowned scottish botanist , robert brown , who noticed in 1827 that pollen in water suspension which he examined under his microscope displayed a very rapid , irregular , zigzag motion .
the mystery of the `` vital force '' driving the brownian motions remained unsolved for nearly 80 years , evading the pincer of conventional physics .
the answer came from einstein and smoluchowski , who showed how the behaviour of mechanical objects is driven by the statistical properties of thermal noise , postulating the existence of molecules in the fluid and linking the diffusion strength of their motion to the friction acting on a body moving in the fluid @xcite .
the explanation of brown s experiments , being at the same time a major diversion from the `` continuous '' newtonian dynamics forming the core of the contemporary physics , opened a whole new avenue of research into the behaviour of systems influenced with random noise , resulting in such fundamental discoveries as the fluctuation - dissipation theorem @xcite . since that time
, dissipation has been shown to affect such key dynamical processes as electron transfer and transport , surface dynamics , quantum tunneling , control and nonadiabatic effects .
more generally , scientists in many disciplines , from physics through biology to social sciences , have developed increasingly powerful methods of modelling _ open systems _
, which interact with their environment . in many nano - scale systems
the noise influencing the dynamics arises from quantum fluctuations . already in 1928 ,
when nyquist proposed the fluctuation - dissipation theorem @xcite , the quantum fluctuations were treated differently than the classical ones : the energy @xmath0 from the classical equipartition law was replaced by the thermally averaged energy of a quantum harmonic oscillator , a distinction becoming negligible at high temperatures .
this result has been followed by the development of the new branch of physics , the theory of _ open quantum systems _
it has found applications in almost all areas of natural sciences @xcite , from quantum optics @xcite , through condensed matter physics @xcite , nanotechnology @xcite and spintronics @xcite , through quantum information @xcite , through chemistry @xcite , to biology @xcite or even stochastic gravity and inflationary cosmology @xcite .
furthermore , it has implications for such fundamental problems as the quantum measurement theory @xcite and the emergence of classicality due to decoherence @xcite
. there exists a rich variety of methods of modelling open quantum systems , applicable to different physical regimes and based on different approximation techniques @xcite . in general , these methods are difficult to apply to large systems or arbitrary spectral densities . in this paper
we propose a new method , which describes finite - dimensional quantum systems up to a moderately large size ( ca 50 basis states on a standard pc ) , interacting with non - markovian quantum harmonic oscillator baths : from single modes to continuous spectra , as well as a wide range of interaction strengths , while having moderate computational requirements .
the non - markovianity is necessary to quantitatively analyse the properties of many physical systems encountered in the fields mentioned in the previous paragraph @xcite .
the proposed method handles large or infinite baths and a wide range of interaction strengths , while having moderate computational requirements .
it uses the heisenberg picture , which makes it particularly easy to focus the attention on the system degrees of freedom while preserving the decoherence effects due to the coupling to the bath , and conserving the total energy . in the following section we will remind shortly the theoretical background of our work and lay out the employed formalism ( secs.[sec : open - qs ] and [ sec : dynamics - heisenberg ] ) . next we will present the derivation of the reduced operator approximation approach ( sec.[sec : roa ] ) and propose its two variants : low and high - order in the systems and bath operators .
they will be optimised for typical cases of continuous and lorentzian baths in sec.[sec : baths ] . in sec.[sec : examples ] we will present the results of our method and compare it to other known techniques of modelling open quantum systems , like the pseudomode method or the quantum state diffusion .
section[sec : summary ] contains a short summary of our work .
most generally , an open quantum system is a subsystem of a larger , interacting quantum system , e.g. one of the photons in an epr pair , an atom in a resonant cavity , a quantum dot interacting with phonons in the crystal or any real object `` becoming classical '' through scattering of a vast number of air molecules and photons on it .
we consider the case of a finite - dimensional quantum system coupled to an infinite - dimensional quantum bath , composed of a possibly infinite number of modes .
in such an asymmetrical setup it is natural to ignore the details of the bath dynamics and focus on the dynamics of the reduced density matrix of the system . in this chapter
we derive it using the proposed reduced operator approximation approach .
we consider a quantum system represented in an @xmath1-dimensional hilbert space @xmath2 spanned by basis states @xmath3 , and a space of linear operators on @xmath2 , denoted by @xmath4 .
its internal dynamics is described by the hamiltonian @xmath5 , @xmath6 where @xmath7 are transition operators between the states @xmath8 and @xmath9 and @xmath10 are the interstate couplings , @xmath11 . in a more concise notation
, @xmath12 is a trace of a matrix product in @xmath13 , the space of @xmath14 matrices whose elements are operators from @xmath4 : @xmath15 in the equation above , @xmath16 is an @xmath14 matrix of system operators , @xmath17 , and @xmath18 .
since @xmath19 can be embedded in @xmath4 as @xmath20 ( where @xmath21 is the identity operator on @xmath2 ) , the matrix product between @xmath22 and @xmath13 can be defined , with values in @xmath13 .
@xmath23 denotes the trace in @xmath13 , @xmath24 by a natural extension of the scalar product from @xmath2 , @xmath25 @xmath26 is also a hilbert space and @xmath27 , the space of linear operators on @xmath28 .
hermitian conjugation in @xmath13 is defined as @xmath29 . in the case of @xmath30
, this leads to @xmath31 , since @xmath32 .
the system is coupled to a quantum bath composed of a collection of @xmath33 independent harmonic oscillators living in an infinite - dimensional hilbert space @xmath34 and described by the hamiltonian @xmath35 where @xmath36 is the annihilation operator of the @xmath37-th mode ( @xmath38 ) .
the coupling between the system and the bath is described by the operator @xmath39 where @xmath40 and @xmath41 are @xmath42 diagonal matrices describing the coupling of the @xmath37-th bath mode with the system , @xmath43 .
the fact that each @xmath44 matrix is diagonal means that the bath does not induce transitions between system basis states .
however , the matrix notation allows for an easy generalisation of the model to include such bath - induced transitions .
the total hamiltonian , generating the evolution of the system and bath in the schrdinger picture , is given by @xmath45 in the traditional `` index '' notation , @xmath46 has the form @xmath47 the reduced matrix density of the system , @xmath48 , is defined as @xmath49 where @xmath50 is the density matrix of the system and the bath as a whole , and the trace @xmath51 goes over the bath degrees of freedom only , @xmath52 . in the @xmath8 basis , @xmath53 \,,\ ] ] where the trace is taken over both system and bath degrees of freedom .
the main task of the presented method is obtaining @xmath48 without calculating @xmath50 . in the heisenberg picture
the wavefunction is time - independent , @xmath54 ( hence , the density matrix is time - independent as well ) , while an observable @xmath55 ( time - independent in the schrdinger picture ) satisfies @xmath56 \ , , \label{eq : heisenberg}\ ] ] where @xmath57 . from the last definition it follows that @xmath58 = [ o_1 , o_2](t)$ ] .
we assume that at time @xmath59 the system and the bath
represented by their initial reduced density matrices @xmath60 and @xmath61 , respectively are uncorrelated .
hence , the total density matrix @xmath62 equals @xmath63 and eq
. acquires the form @xmath64 \\ & = \operatorname{tr}[t_{mn}(t ) \rho_s\otimes\rho_b ] \\ & = \operatorname{tr}_s [ \rho_s \operatorname{tr}_b \rho_b t_{mn}(t ) ] \ , .
\end{split}\ ] ] let @xmath65 denote the heisenberg - picture counterparts of @xmath66 and @xmath36 , respectively , with @xmath67 ( where @xmath68 is the identity operator on @xmath34 ) and @xmath69 . the fact that time - dependent operators act on the complete hilbert space @xmath70 reflects the fact that the interaction couples the system and the bath .
analogously to the schrdinger picture , we define @xmath71 as @xmath72 .
we also define the trace @xmath73 as @xmath74 in order to derive their equations of motions using eq .
we calculate , for any @xmath75 ( which can be embedded in @xmath76 by embedding each element @xmath77 in @xmath78 ) , @xmath79 & = \sum_{m',n'=1}^n a_{m'n ' } [ t_{n'm'}(t ) , t_{mn}(t ) ] \\ & = [ { \hat{a}},{\hat{t}}(t)]_{mn } \ , , \end{split}\ ] ] where we have used the identity ( valid also in the schrdinger picture ) @xmath80 in the more compact notation , @xmath81 = [ { \hat{a } } , { \hat{t } } ] \ .\ ] ] the above identities can be applied to eq . to obtain the evolution of @xmath30 and @xmath82 operators generated by the hamiltonian . for the system we obtain @xmath83 + i \sum_{k=1}^k \left ( [ { \hat{g}}_k , { \hat{s}}_k^\dagger(t ) ] + [ { \hat{g}}_k^\dagger , { \hat{s}}_k ] \right ) , \ ] ] where @xmath84 are system - bath interaction operators and , since @xmath85 and bath operators commute with system operators , @xmath86 . for the bath , using the canonical commutation relations for bosonic creation / annihilation operators , we obtain @xmath87 in the index notation , the above equations have the forms ( assuming @xmath43 ) @xmath88 and @xmath89 the aim of the presented method is to model the evolution of the system , including its decoherence caused by the interaction with the bath .
the information about this process is contained in the reduced density matrix of the system @xmath48 .
as demonstrated in previous sections , eqs and , it can be obtained from the mean values of system operators @xmath90 .
thus , to calculate @xmath48 in the heisenberg picture , one has to evolve @xmath90 in time . since the evolution equation for @xmath90 involves the bath operators @xmath91 , due to the system - bath interaction , it is necessary to evolve @xmath91 as well .
however , a numerical description of both types of operators in the total system and bath basis is impossible , as @xmath92 is infinite - dimensional . according to eq . ,
given the initial system state @xmath60 we only need to know the partial expectation values of system operators , @xmath93 , to obtain @xmath48 .
the corresponding partial expectations of the bath operators , @xmath94 , contain part of the information on how interaction correlated the bath and the system if there was no such correlation , @xmath94 would be proportional to an identity operator in @xmath95 ( recall that @xmath96 ) .
thus , even after averaging over the bath degrees of freedom , we can at least approximately capture the system - bath correlations arising from the interaction terms in eqs . and .
this observation forms the basis of the proposed reduced operator approximation ( roa ) .
we represent both system and bath operators by @xmath14 complex matrices in the system state basis ( hence , @xmath30 is represented by an @xmath14 matrix of @xmath14 complex matrices ) .
let @xmath97 \in { { \mathcal{l}({\mathcal{h}_s})}}$ ] denote this _
reduced representation _ of an operator @xmath98 , defined as @xmath99 : = \operatorname{tr}_b \rho_b o(t ) \ , .
\label{eq : roa}\ ] ] the mapping @xmath97 $ ] is defined so as to preserve , at least partially , the information about how the operator @xmath100 acts on the system degrees of freedom ( hilbert space @xmath2 ) given the the interaction with the bath .
since the system - bath interaction depends on the initial state of the bath @xmath61 , so does @xmath97 $ ] .
this effect is particularly strong in non - markovian systems . as can be seen from eq . , the mapping conserves the expectation values : @xmath101 = \\ = \operatorname{tr}_s [ \rho_s m[o(t ) ] ] = \langle m[o(t ) ] \rangle \ .\end{gathered}\ ] ] from the definition , @xmath102 = m[o(t)]^\dagger$ ] .
the mapping @xmath103 $ ] is easily extended elementwise to @xmath104 \in { { { { \mathcal{l}({\mathcal{h}_s})}}^{n \times n}}}$ ] for @xmath105 .
the evolution equations for the reduced representations of system and bath operators are @xmath106 = -i \omega_k m[a_k(t ) ] - i \operatorname{tr}{\hat{g}}_{k } m[{\hat{t}}(t)]\ ] ] and @xmath107 = i [ { \hat{v}}^t , m[{\hat{t}}(t ) ] ] + \\&\qquad i \sum_{k=1}^k \left ( [ { \hat{g}}_k , m[{\hat{s}}_k^\dagger(t ) ] ] + [ { \hat{g}}_k^\dagger , m[{\hat{s}}_k(t ) ] ] \right ) \ , .
\end{split}\ ] ] since the system and the bath are correlated , @xmath108 \neq m [ { \hat{t}}(t ) ] m [ a_k(t ) ] $ ] , which means that the above evolution equations are not complete .
the simplest way to complete them is to approximate @xmath109 $ ] by the product of @xmath110 $ ] and @xmath111 $ ] , which means neglecting higher - order correlations between the system and the bath introduced by their interaction .
however , again due to the system - bath coupling , @xmath110 m[a_k(t ) ] \neq m[a_k(t ) ] m[{\hat{t}}(t)]$ ] , even though @xmath112 = 0 $ ] .
thus , we need to specify a concrete ordering of the multiplied reduced operators .
we use the approximations of the form @xmath113 & \approx \theta_l m[{\hat{t}}(t ) ] m[a_k(t ) ] + ( 1-\theta_l ) m[a_k(t ) ] m[{\hat{t}}(t ) ] \,,\\
m[{\hat{s}}^\dagger_k(t ) ] & \approx \theta_l m[a_k(t)]^\dagger m[{\hat{t}}(t ) ] + ( 1-\theta_l ) m[{\hat{t}}(t ) ] m[a_k(t)]^\dagger , \end{aligned}\ ] ] for @xmath114 $ ] . based on our numerical experiments , which have shown that simulations diverge for @xmath115
, we put @xmath116 .
appendix [ sec : symmetrisation ] contains an examination of this problem for a simplified case , supporting this choice .
in fact , the symmetrization of the product of @xmath111 $ ] and @xmath110 $ ] is consistent with the fact that @xmath91 and @xmath90 always commute . using it
, we arrive at the final form of the evolution equation of the system operators in the reduced representation , @xmath117 & = i [ { \hat{v}}^t , m[{\hat{t}}(t ) ] ] \\ & \quad + \frac{i}{2 } \sum_{k=1}^k [ { \hat{g}}_k , \ { m [ { \hat{t}}(t ) ] , m [ a_k^\dagger(t ) ] \ } ] \\ & \quad + \frac{i}{2 } \sum_{k=1}^k [ { \hat{g}}_k^\dagger , \ { m [ { \hat{t}}(t ) ] , m [ a_k(t ) ] \ } ] \ , , \end{split}\ ] ] where @xmath118 denotes the anti - commutator .
eqns and employ reduced representations which are linear in the system or bath operators .
hence , we will refer to them as the _ lower - order _ roa .
it is important to note that this form does not neglect the correlations between the system and the bath , due to the fact that the bath operators are represented by their matrices in the system basis , @xmath111 $ ] .
additional information about the system - bath correlations is provided by the @xmath109 $ ] matrix .
hence , it may be beneficial to evolve it separately in addition to @xmath111 $ ] and @xmath110 $ ] .
for this purpose , we first derive the evolution equation for @xmath119 , using eqs . and , @xmath120 + i \sum_{k'=1}^k [ { \hat{g}}^\dagger_{k ' } , { \hat{s}}_{k}(t ) ] a_{k'}(t ) \\ & + i \sum_{k'=1}^k a_{k'}^\dagger(t ) [ { \hat{g}}_{k ' } , { \hat{s}}_{k}(t ) ] - i \omega_k { \hat{s}}_{k}(t ) - i { \hat{t}}(t ) { \hat{g}}_k \ , , \end{split}\ ] ] where we have used the fact that , due to the associativity of the operator product , @xmath121 and @xmath122 . to derive the evolution equation for @xmath109 $ ] , we have to solve a similar ordering problem as in the lower - order method : @xmath123 & \approx \theta_h m[{\hat{s}}_{k}(t ) ] m[a_{k'}(t ) ] \\ & \ + ( 1 - \theta_h ) m[a_{k'}(t ) ] m[{\hat{s}}_{k}(t ) ] \\
m[a_{k'}^\dagger(t ) { \hat{s}}_{k}(t ) ] & \approx \theta_h m[a_{k'}^\dagger(t ) ] m[{\hat{s}}_{k}(t ) ] \\ & \ + ( 1 - \theta_h ) m[{\hat{s}}_{k}(t ) ] m[a_{k'}^\dagger(t ) ] \\ & \ - ( 1 - \theta_h ) \delta_{k , k ' } m[{\hat{t}}(t ) ] \ , \label{eq : } \end{split}\ ] ] where the last term arises from the fact that @xmath124 and @xmath125 do not commute . for the bath in the ground state at @xmath59 we must have @xmath126 = 0 $ ] .
it is easy to see that the only way this can be satisfied is by choosing @xmath127 , as otherwise @xmath128 = - ( 1 - \theta_h ) \delta_{k , k ' } m[{\hat{t}}(t ) ] \neq 0 \ .\ ] ] this choice is supported by numerical tests , which yield divergences when @xmath129 is used .
differently than in the low - order case , we have not chosen a symmetrised product approximation ( @xmath130 ) , due to the fact that @xmath131 \neq 0 $ ] .
for consistency , we apply the same choice of @xmath132 to the approximation of @xmath133 $ ] .
@xmath133 $ ] can be approximated by either @xmath109 m[a_{k'}(t)]$ ] , as above , or @xmath134 m[a_{k}(t)]$ ] ( note the transposition of indices @xmath37 and @xmath135 ) . similarly , we can choose between @xmath136 m[{\hat{s}}_{k}(t)]$ ] and @xmath137 m[a_{k}(t)]$ ] for @xmath138 $ ] .
to exploit fully the information about the system - bath correlations contained in @xmath134 $ ] matrices , we use an equally weighted average of the two approximations . in this way
we obtain the evolution equation for the reduced representation of the interaction operator @xmath139 = i [ { \hat{v}}^t , m[{\hat{s}}_k(t ) ] ] + \\ & \tfrac{i}{2 } \sum_{k'=1}^k \left ( [ { \hat{g}}_{k ' } , m[{\hat{s}}_{k'}^\dagger(t ) ] ] + [ { \hat{g}}_{k'}^\dagger , m[{\hat{s}}_{k'}(t ) ] ] \right ) m[a_k(t ) ] \\ & + \tfrac{i}{2 } \sum_{k'=1}^k m[a_{k'}^\dagger(t ) ] [ { \hat{g}}_{k ' } , m[{\hat{s}}_{k}(t ) ] ] \\ & + \tfrac{i}{2 } \sum_{k'=1}^k [ { \hat{g}}_{k'}^\dagger , m[{\hat{s}}_k(t ) ] ] m[a_{k'}(t ) ] \\ & \
-i \omega_k { \hat{s}}_{k}(t ) - i { \hat{t}}(t ) { \hat{g}}_k \,.\end{split}\ ] ] together with eqs . and
it defines the _ higher - order _ roa . in the lower - order version of our method ,
the expression for the total hamiltonian has the form @xmath140 & = \operatorname{\hat{t}r}{\hat{v}}^t m[{\hat{t}}(t ) ] + \sum_{k=1}^k \omega_k m[a_k^\dagger(t ) ] m[a_k(t ) ] \\ & \
+ \operatorname{\hat{t}r}\sum_{k=1}^k \left\ { \left ( m[a_k^\dagger(t ) ] { \hat{g}}_k + m[a_k(t ) ] { \hat{g}}_k^\dagger \right ) , m[{\hat{t}}(t ) ] \right\ } \ .
\end{split } \label{eq : h - roa - l}\ ] ] in the higher - order version , the last term acquires the form @xmath141 { \hat{g}}_k + m[{\hat{s}}_k(t ) ] { \hat{g}}_k^\dagger \right ) \ .\ ] ] a tedious but straightforward calculation shows that in both cases , the evolution equations from sec . [ sec : genver ] conserve @xmath142 $ ] .
thus , our method conserves the total energy . from the evolution equations or we instantly see that , since the trace of every commutator is zero , @xmath143 \equiv \sum_{m=1}^n m[t_{mm}(t ) ] \right ) = 0 \,.\ ] ] inserting @xmath110 $ ] instead of @xmath90 in eq .
, we obtain a trace - conserving expression for the reduced density matrix of the system , @xmath144\ , .
\label{eq : rho - heis - m}\ ] ] however , if we use it to calculate @xmath48 , its positivity is not guaranteed . to fix this problem
, we make use of the identity to derive @xmath145 and replace eq . with a different approximation @xmath146 ^ 2 \,.\ ] ] since @xmath147 ( and the same for the reduced representations ) , the above formula guarantees that @xmath48 is positive - semidefinite . on the other hand ,
the density matrix does not possess a conserved trace due to the fact that @xmath110 ^ 2 \neq m[{\hat{t}}(t ) { \hat{t}}(t)]$ ] .
thus , we normalise the density matrix to obtain @xmath148 ^ 2}{\sum_{n'=1}^n\operatorname{tr}_s \rho_s m[t_{n'n'}(t)]^2 } \,.\ ] ] for any type of bath , we can define the spectral density @xmath149 in the limit of an infinite number of modes , @xmath150 can be a continuous function .
one way to handle this situation is to discretise @xmath150 into a finite , but large number of modes . assuming a constant mode frequency spacing @xmath151
, we define a coupling constant for @xmath152 to be @xmath153 taking the square root ensures proper normalisation as @xmath154 . for a spectral density being a single lorentzian peak
, the method converges quite well already for @xmath155 modes per system basis state and @xmath156 , where @xmath157 is the half - width at half - maximum of the peak .
when each system basis state is coupled to its own independent bath with a continuous spectral density , a better method is to describe these baths in terms of collective mode excitations caused by the coupling with the system .
let us consider bath operators multiplied by their phase factors , @xmath158 , with dynamics given by @xmath159 .
employing the fact that taking the trace and integration over time commute , we obtain @xmath160 which we insert into the definition of @xmath119 , @xmath161 and then the above formula into eq . , @xmath162 + i \sum_{k=1}^k \left ( [ { \hat{g}}_k , { \hat{u}}_k^\dagger(t ) ] + [ { \hat{g}}_k^\dagger , { \hat{u}}_k(t ) ] \right ) \\ & \ + i \sum_{k=1}^k \left ( e^{i\omega_k t } a_k^\dagger(0 ) [ { \hat{g}}_k , { \hat{t}}(t ) ] + e^{-i\omega_k t } [ { \hat{g}}_k^\dagger , { \hat{t}}(t ) ] a_k(0 ) \right ) . \end{split}\ ] ] under the above assumptions , each bath mode @xmath37 is coupled to exactly one basis state @xmath163 , i.e. @xmath164 . hence , @xmath165 \\ & + t_{mn}(t ) \sum_{k=1}^k \left [ ( { \overline{g_{km } } } - { \overline{g_{kn } } } ) e^{-i\omega_k t } a_k(0 ) + \text{h.c . } \right]\ , , \end{split}\ ] ] where we have used the fact that for independent densities , @xmath166 only for @xmath167 . in the limit of infinite number of modes , using eq
. we obtain @xmath168 where @xmath169 is the bath correlation function . in this way
we obtain a closed system of differential - integral equations for @xmath170 , @xmath171 \end{split}\ ] ] where @xmath172 @xmath173 operators @xmath174 and @xmath175 satisfy canonical commutation relations for bosons , @xmath176 & = 0 \,,\\ [ \tilde{a}_m^\dagger(t ) , \tilde{a}_n(t ) ] & = - \frac{\delta_{mn}}{\sqrt{\kappa_m \kappa_n } } \lim_{k \to \infty } \sum_{k=1}^k { \overline{g_{kn } } } g_{km } = - \delta_{mn } \ , .
\end{split}\ ] ] they are _ pseudomode _ creation and annihilation operators , creating or destroying collective excitations in a single bath @xcite .
their dynamics is given by the equation @xmath177 using the proposed method we have reduced significantly the number of bath operators , from @xmath33 to @xmath1 ( for independent baths @xmath178 , while in many cases @xmath179 ) . however , numerical simulation of the differential - integral equation for the evolution of the reduced representation of @xmath175 is difficult . in the next section
we show that for a particular form of the spectral density function @xmath180 one can get rid of the explicit time integration at the cost of a moderate increase of the number of simulated bath operators .
continuous spectral densities composed of lorentzian peaks , @xmath181 are especially popular due to their analytical tractability . in this section , we will optimise our method for this type of the system - bath coupling .
the corresponding correlation function is @xmath182 .
hence , for @xmath183 , @xmath184 a continuous spectral density of the form is constructed from an infinite number of independent harmonic oscillator modes , with different modes contributing to each lorentzian peak . to derive the evolution equation for pseudomode bath operators we express them as sums of @xmath185 , where @xmath186 is constructed as @xmath187 and @xmath188 $ ]
is the set of indices of modes building the @xmath189-th peak , @xmath190 $ ] and @xmath191 .
this leads directly to @xmath192 = 0 $ ] , @xmath193 = 0 $ ] and @xmath194 = - \delta_{mn } \delta_{jj'}$ ] .
thus , @xmath195 and @xmath186 are pseudomode creation and annihilation operators corresponding to individual lorentzian peaks in bath spectral densities .
comparing @xmath186 with @xmath175 leads to @xmath196 differentiating over @xmath197 gives @xmath198 with the initial condition @xmath199 by splitting @xmath175 into a sum of @xmath186 , we have simplified the differential - integral evolution equation . in the reduced representation , @xmath200 = ( -i \omega_{mj } - \gamma_{mj } )
m[\tilde{a}_{mj}(t ) ] + \sqrt{\gamma_{mj } } m[t_{mm}(t)]\ ] ] with the initial condition @xmath201 = 0 $ ] . since @xmath202 \neq m[t_{mn}(t ) ] m[\tilde{a}_{m'j}(t)]$ ] , we evolve separately the reduced representation of operator products @xmath203 .
its adjoint equals @xmath204 .
the relevant commutators are @xmath205 = 0 $ ] and @xmath206 = - t_{nm}(t ) \delta_{m'm '' } \delta_{jj'}$ ] .
evolution equation of the system in the lorentzian bath acquires the form @xmath207 \\ & \
+ \sum_j \sqrt{\gamma_{nj } } \left [ s_{nmnj}^\dagger(t ) - s_{mnnj}(t ) \right ] \ , .
\end{split}\ ] ] hence , the operators @xmath208 themselves follow the evolution equation @xmath209 \tilde{a}_{m'j}(t ) \\ &
\quad + \sum_{j ' } \sqrt{\gamma_{nj } } \left [ s_{nmnj'}^\dagger(t ) - s_{mnnj'}(t ) \right ] \tilde{a}_{m'j}(t ) \\ & \quad + t_{mn}(t ) \lim_{k\to\infty } \sum_{k \in p^k_j } \frac{{\overline{g_{km ' } } } \omega_k e^{-i\omega_k t}}{\sqrt{\gamma_{m'j } } } a_k(0 ) \\ & \quad + ( -i \omega_{m'j } - \gamma_{m'j } ) s_{mnm'j}(t ) + \sqrt{\gamma_{m'j } } \delta_{nm ' } t_{mn}(t ) \end{split}\ ] ] with the initial condition @xmath210 evolution equations for the reduced repretentations of the above system and interaction operators , respectively , are @xmath211 & = i \sum_{m'=1}^n ( v_{m'm } m[t_{m'n}(t ) ] - v_{nm ' } m[t_{mm'}(t)])\\ & \ + \sum_j \sqrt{\gamma_{mj } } \left [ m[s_{mnmj}(t ) ] - m[s_{nmmj}^\dagger(t ) ] \right ] \\ & \ + \sum_j \sqrt{\gamma_{nj } } \left [ m[s_{nmnj}^\dagger(t ) ] - m[s_{mnnj}(t ) ] \right ] \end{split}\ ] ] and @xmath212 = i \sum_{m''=1}^n ( v_{m''m } m[s_{m''nm'j}(t ) ] - \\ & v_{nm '' } m[s_{mm''m'j}(t ) ] ) + \\ & \tfrac{1}{2 } \sum_{j ' } \sqrt{\gamma_{mj ' } } ( m[s_{mnmj'}(t ) ] - m[s_{nmmj'}^\dagger(t ) ] ) m [ \tilde{a}_{m'j}(t ) ] + \\ & \tfrac{1}{2 } m[s_{mnm'j}(t ) ] \sum_{j ' } ( \sqrt{\gamma_{mj ' } } m [ \tilde{a}_{mj'}(t ) ] - \sqrt{\gamma_{nj ' } } m[\tilde{a}_{nj'}(t ) ] ) \\ & + \tfrac{1}{2}\sum_{j ' } \sqrt{\gamma_{nj ' } } ( m[s_{nmnj'}^\dagger(t ) ] - m[s_{mnnj'}(t ) ] ) m [ \tilde{a}_{m'j}(t ) ] -\\ & \tfrac{1}{2}\sum_{j ' } ( \sqrt{\gamma_{mj ' } } m[\tilde{a}_{mj'}^\dagger(t ) ] - \sqrt{\gamma_{nj ' } } m[\tilde{a}_{nj'}^\dagger(t ) ] ) m[s_{mnm'j}(t ) ] \\ & + ( -i \omega_{m'j } - \gamma_{m'j } ) m[s_{mnm'j}(t ) ] + \sqrt{\gamma_{m'j } } \delta_{nm ' } m[t_{mn}(t ) ] \end{split}\ ] ] with initial condition @xmath213 = 0 $ ] .
we use the same operator ordering scheme as in sec .
[ sec : independent - baths ] .
higher - order lorentzian roa employs eqs . , and .
lower - order lorentzian roa describes the bath evolution with eq . .
to describe the system evolution , we represent @xmath214 $ ] as @xmath215,m[\tilde{a}_{m'j}(t)]\}$ ] , obtaining @xmath216 & = i \sum_{m'=1}^n ( v_{m'm } m[t_{m'n}(t ) ] - v_{nm ' } m[t_{mm'}(t)])\\ & \
+ \frac{1}{2 } \bigl\ { m[t_{mn}(t ) ] , \sum_j \bigl [ \sqrt{\gamma_{mj } } m[\tilde{a}_{mj}(t ) ] \\ & \qquad - \sqrt{\gamma_{nj } } m[\tilde{a}_{nj}(t ) ] - \text{h.c . } \bigr ] \bigr\ } \ , , \end{split}\ ] ] where the symmetrisation of the matrix product is justified by the same arguments as in sec .
[ sec : genver ] .
the lorentzian methods use much lower number of bath and interaction operators than the general ones , because they model the bath excitations as collective pseudomodes .
furthermore , thanks to the analytical integration of the spectral density , they automatically include the tails of the spectral density , which are cut off by the discrete method .
in this section we present an example application of our method to the description of a molecular aggregate interacting with a non - markovian quantum bath , and compare it with two other techniques : the pseudomode method @xcite ( pm ) and non - markovian quantum state diffusion @xcite within an approximation called the zero - order functional expansion @xcite ( nmqsd ) .
the pm method replaces each lorentzian peak in the spectral density by a pseudomode with a complex frequency , and models the dynamics of the original system and bath by simulating exactly , in the schrdinger picture , the system interacting with this pseudomode bath .
as the reduced density matrix of the system @xmath48 obtained in this way is exact , we use the pm method as our reference .
the downside of the pm method is that , since it requires an exact simulation of a quantum many - body system , its computational requirements increase exponentially with the number of bath spectral density peaks . the other method used for comparison , the non - markovian quantum state diffusion , uses a monte - carlo simulation to calculate @xmath217 .
its main advantage is the slow growth of the computational requirements with the system size . on the other hand
, this method does not allow for direct control of the total energy , unlike our method .
moreover , a numerically feasible realisation of the non - markovian quantum state diffusion requires a further approximation such as the abstract zero - order functional expansion ( zofe ) @xcite that we compare our results against .
in fact , we are not aware of any other practical , generally applicable implementation of nmqsd .
we model an exciton delocalised on a linear chain composed of @xmath218 sites coupled by the nearest - neighbour potential , @xmath219 .
each site interacts with a simple zero - temperature quantum bath with a unimodal lorentzian spectral density @xmath220 , where we set @xmath221 .
we consider four cases : [ cols="<,^,^",options="header " , ] the `` wide '' lorentzian peaks correspond to fast decreasing bath correlation function , while the `` narrow '' ones indicate long correlation times .
the coupling strength , `` strong '' or `` weak '' , determines the decoherence rate .
we simulate the reduced density matrix of the system initially in the state @xmath222^t$ ] , and compare the probabilities of finding the system in this state at later times , i.e. @xmath223 .
we use three variants of the roa method : low - order roa ( sec .
[ sec : genver ] ) , as well as low and high - order lorentzian roa ( sec . [ sec : lorentzian - spectral - densities ] ) ; the high - order roa has been skipped as being the least efficient .
the results are plotted in figs .
[ fig : weaknarrow][fig : strongwide ] . comparison with the exact pm method shows that the best results are obtained for the lorentzian variants of the method , which take into account the whole range of the lorentzian spectral density by analytical integration . at the same time nmqsd
has the tendency to converge too rapidly to a steady state solution , as compared to the exact pm method .
the figure captions contain a detailed analysis of the results . for further comparison with the nmqsd method
we calculate the transfer of the excitation on a ring aggregate in multimode lorentzian bath ( see ref.@xcite ) in fig.[fig : transport ] .
the results obtained using the low - order lorentzian roa method are characterised by a much slower decoherence rate , as shown in the inset .
this suggests that quantum coherence effects play a larger role in the dynamics of excitons in open systems than predicted in ref.@xcite .
we attribute the faster decoherence observed in the nmqsd method to the zofe approximation @xcite required to make the method numerically feasible .
it treats each path of the monte carlo simulation independently and thus increases artificially the amount of decoherence in the simulation causing the system to converge too rapidly to a steady state solution . comparing the computational efficiency of roa and nmqsd methods ( where we have averaged the transfer over 1000 realizations of the stochastic noise @xcite ) , for the considered system our method is more than 10 times faster .
calculated using both methods . ]
the presented reduced operator approximation is a simple , physically transparent and computationally efficient method of modelling open quantum systems .
it employs the heisenberg picture of the quantum dynamics , which allows us to focus on the system degrees of freedom in a natural and easy way .
our approach captures the decoherence of the system coupled with the bath , while conserving the total energy .
we have described different variants of the method : the low - order roa , the high - order roa ( including the interaction operators ) and their versions for lorentzian baths .
they have been applied to different systems ( coupled to different baths with different strengths ) .
comparison of our results with the exact pseudomode and the popular quantum state diffusion method favours the roa approach , while the performance of roa ( especially the low - order case ) is much higher than in the case of two other approaches .
furthermore , the roa simulations of the exciton transfer on a ring aggregate suggest that quantum coherence effects play a larger role in the dynamics of open quantum systems than predicted by nmqsd @xcite . the method has been derived for the simplest case of linear coupling between the system and the bath .
however , its general version can be easily extended to higher - order couplings .
its another advantage over methods which do not use the heisenberg picture ( such as those used in our comparisons ) is that a single simulation of the reduced system operators can be used to generate reduced density matrices for an arbitrary choice of initial system state .
let us consider a system - bath hamiltonian with @xmath1 system basis states and one bath mode per basis state ( @xmath224 ) , coupled only to this state ( @xmath225 ) . to further simplify the problem
, we assume that the basis states are degenerate and have zero energies when isolated from the bath ( @xmath226 ) . in this case
, one can solve analytically the heisenberg equations of motion for the bath operators and radically simplify the corresponding equations for the system operators .
it turns out that even in this simple case , the reduced operator matrix @xmath227 $ ] does not commute with @xmath111 $ ] .
we will exploit this to determine what is the best way of approximating @xmath228 $ ] by products @xmath227m[a_k(t)]$ ] and @xmath111m[t_{mn}(t)]$ ] , providing a justification for the choice of @xmath229 in sec .
[ sec : genver ] .
the analysis we perform here is similar in spirit to the one performed when analysing the stability of discretisation schemes for partial differential equations , where a detailed analysis of a very simple problem yields important insights into the behaviour of more complex ones .
we would like to stress that any numerically feasible method of modelling physical systems must be analysed from two points of view : physical foundations and numerical stability .
hence , the approximations one necessarily has to make may be motivated not only by physical principles , but also by numerical analysis .
substituting our assumptions into eq . , we obtain @xmath230 , leading to @xmath231 .
hence , we can solve eq .
explicitly , obtaining @xmath232 and , in the reduced representation , @xmath233 = - \frac{g_m}{\omega_m } ( 1 - e^{-i\omega_m t } ) m[t_{mm}(t ) ] \,.\ ] ] let us assume that at time @xmath197 , the relation @xmath234 m[t_{m'n'}(t ) ] = \delta_{nm ' } m[t_{mn'}(t ) ] \label{eq : mtmn - prod}\ ] ] is preserved . hence , @xmath235 m[a_k(t ) ] & = - \delta_{nk } \frac{g_n}{\omega_n } ( 1 - e^{-i\omega_n t } ) m[t_{mn}(t ) ] \ , , \\
m[a_k(t ) ] m[t_{mn}(t ) ] & = - \delta_{mk } \frac{g_m}{\omega_m } ( 1 - e^{-i\omega_m t } ) m[t_{mn}(t ) ] \,.\end{aligned}\ ] ] applying the above results to eq . , we obtain @xmath236 & = i ( 1 - \delta_{mn } ) [ \theta_l ( { \overline{\alpha_m(t ) } } - \alpha_n(t ) ) \\ & + ( 1-\theta_l ) ( \alpha_m(t ) - { \overline{\alpha_n(t ) } } ) ] m[t_{mn}(t ) ] \end{split}\ ] ] where @xmath237 . the heisenberg equations of motions preserve the identity @xmath238 .
given the fact that we use the products of @xmath227 $ ] matrices in calculating the reduced density matrix @xmath48 ( see sec .
[ sec : reduced - system - density - matrix ] ) , we would like the equations for motion for @xmath227 $ ] to preserve the identity @xmath227 = m[t_{mn'}(t ) ] m[t_{n'n}(t)]$ ] too .
one can easily verify that for almost all choices of @xmath239 , @xmath240 and @xmath241 this is the case for any value of @xmath242 $ ] .
however , in the case of @xmath243 and @xmath244 , we have @xmath245 m[t_{n'n}(t ) ] = i [ \theta_l ( { \overline{\alpha_m(t ) } } - \alpha_{n'}(t ) ) + \\ & ( 1-\theta_l ) ( \alpha_m(t ) - { \overline{\alpha_{n'}(t ) } } ) ] m[t_{mn}(t ) ] + i [ \theta_l ( { \overline{\alpha_{n'}(t ) } } - \alpha_{n}(t ) ) \\ & + ( 1-\theta_l ) ( \alpha_{n'}(t ) - { \overline{\alpha_{n}(t ) } } ) ] m[t_{mn}(t ) ] \ , .
\end{split}\ ] ] it is only for @xmath229 that the terms with @xmath246 cancel out and the right side is equal to @xmath247 $ ] .
therefore , only for @xmath229 evolution equations for @xmath227 $ ] preserve product identities for @xmath170 . extending the above result on the case of general system hamiltonian and general system - bath coupling , we use @xmath229 in eq . so that the time evolution of the product @xmath227 m[t_{nn'}(t)]$ ] is as close as possible to the time evolution of @xmath248 $ ]
heinz - peter breuer and francesco petruccione .
concepts and methods in the theory of open quantum systems . in fabio benatti and roberto floreanini , editors ,
_ irreversible quantum dynamics _ , volume 622 of _ lecture notes in physics _ , pages 6579
springer berlin heidelberg , 2003 .
g. panitchayangkoon , d.v .
voronine , d. abramavicius , j.r .
caram , n.h . c. lewis , s. mukamel , and g.s .
direct evidence of quantum transport in photosynthetic light - harvesting complexes .
, 108(52):2090820912 , 2011 .
simone montangero mohan sarovar tommaso calarco martin b. plenio k. birgitta whaley stephan hoyer , filippo caruso .
realistic and verifiable coherent control of excitonic states in a light harvesting complex .
jan roden , alexander eisfeld , wolfgang wolff , and walter t. strunz .
influence of complex exciton - phonon coupling on optical absorption and energy transfer of quantum aggregates .
, 103:058301 , jul 2009 . | we present the reduced operator approximation : a simple , physically transparent and computationally efficient method of modelling open quantum systems .
it employs the heisenberg picture of the quantum dynamics , which allows us to focus on the system degrees of freedom in a natural and easy way .
we describe different variants of the method , low- and high - order ( including the interaction operators ) , defining them for either general quantum harmonic oscillators baths or specialising them for independent baths with lorentzian spectral densities .
the method captures the decoherence of the system interacting with the bath , while conserving the total energy .
the wide range of the method applicability is presented on the example of systems coupled to different baths with different strengths , and compared with the exact pseudomode and the popular quantum state diffusion approach .
our results suggest that quantum coherence effects persist in open quantum systems for much longer times than previously thought . |
since gravitational wave interferometers will sense the passage of a gravitational wave as a difference in the light path between mirrors in the two arms , every kind of excitation is a noise source .
cooling the mirrors of these detectors , by measuring and controlling with a feedback loop the principal fundamental motions that may be thermally excited , may offer a means to reduce thermal noise . with this motivation in mind ,
we investigated models that generate frequency spectra with a tail @xmath0 for @xmath1 .
this dependence is found in many completely different physical systems , suggesting a possible underlying simple mechanism .
the typical physical parameters are different for each of them , and especially the characteristic microscopic quantities .
but many classes of phenomena are characterized by the same value of @xmath2 .
indeed the responsive behaviour of every system is classified according to the parameters of the phenomenological equations by which the main characteristics of its evolution and interactions are described .
moreover the contribution of the microscopic degrees of freedom is absorbed in the definition of the parameters so that it is possible to describe the motion of a system in terms of a phenomenologial equation for the macroscopic degrees of freedom .
the microscopic degrees of freedom motions are integrated over as in averages and their effect in the dynamics of the system is contained in the physical constants .
for example in high energy physics the masses are corrected for those particles which are not isolated but interacting with other ones , and since measurements imply interactions , the measured values are the renormalized values instead of the bare values .
similarly applications of those principles may be found in low energy physics @xcite .
a typical example is the up - conversion of resonant frequencies , related with a non - linear perturbation term that affects both the bare value of the resonant frequency and its width @xcite . in this paper
a very simple model is studied whose basic characteristic is the switching between equilibrium configurations across energy barriers .
there are no other requirements and this generality is the main feature of this flexible model that can be easily adapted for a lot of different physical phenomena .
the principal interest has been the investigation of the influence on the spectral function and particularly on the typical low frequency tail of a damping force .
for those materials characterized by a quality factor that may be increased by decreasing the amount of impurities , a reasonable picture is that the motion of those degrees of freedom may generate a fluctuation - dissipation process across the sites in the crystal . in other words
the impurities are supposed to move following a random walk .
the energy of those impurities is related with their temperature ; in cryogenic conditions they hardly can pass through the potential barrier and change their location .
the probability of a jump from one site to another depends upon @xmath3 .
the system is characterized by its statistical properties and the temperature is defined as a stationary thermodynamical quantity .
a simple random walk model may be considered to represent the switching between the equilibrium positions for a two well potential function @xmath4 it has two minima for @xmath5 pointed out in fig.[fig : aa ] .
increasing the constant @xmath6 implies a smaller distance between the two wells . from the taylor expansion
the following expression is achieved @xmath7\ ] ] for @xmath8 near the two minima .
the constant @xmath9 has been introduced so that @xmath10 has the physically right dimension of frequency . in the phase space @xmath11
whose solutions represent the trajectories that depend on the initial conditions .
the problem may be treated perturbatively near the two minima . using the linear approximation for small oscillations a stochastic force
may be introduced that satisfies the fluctuation - dissipation theorem . in other words , the deterministic part of this force and the random component
are related in such a way that the exchanged energies are balanced .
the resulting equation is @xmath12 near the two minima @xmath13 .
the constant @xmath14 is so large that underdamped conditions are always satisfied . in order to take into account the ratio of the thermal equilibrium energy over the local maxima in the potential function
, a scale factor may be used in order to recover the equation @xmath15 where @xmath16 is dimensionless . as a consequence , @xmath6 , that had
the same dimensionality as @xmath17 , has become dimensionless .
it is now easy to see that flat wells and high temperature are equivalent and the constant @xmath18 sums up the two effects .
all the comments referred to fig.[fig : aa ] are valid for @xmath19 as well .
more intutively the scaling of the x - axis by squeezing or stretching gives the potential function a different shape corresponding to making the wells nearer or farther , deeper or flatter .
following an expansion of the equation of motion near @xmath20 @xmath21\end{aligned}\ ] ] where the origin of @xmath22 has been fixed in @xmath23 .
it may be noted that @xmath22 appears also on the right side of the equation and this means that this definition is implicit .
using a perturbative expansion gives @xmath24\end{aligned}\ ] ] where @xmath25 represents the solution in the zero order approximation .
if @xmath25 is considered and the two point correlation function is calculated , the associated integral over frequencies is @xmath26 for each of the two linearly approximated solutions near the minima .
the function @xmath27 is called power spectral density . taking into consideration higher order terms for @xmath22 gives corrections to @xmath27 .
for the property @xmath28 the term representing the first correction of @xmath27 is proportional to @xmath19 .
it is @xmath29+\ldots]\end{aligned}\ ] ] where @xmath30 .
all the terms may be easily derived using the feynman diagram technique . in this paper
indeed the analogies between a wave equation and a heat equation are used and the solution @xmath31 is represented and expanded in a graphic way . the analogy with the feynman diagram techniques
is defined by associating a line with the green s function @xmath32 and a cross with the force @xmath33 . when two crosses are combined together
a @xmath34 arises .
the iterative method of getting the terms of the expansion consists of substituting for the solution in the implicit definition at any order , the expression corresponding to the lower one in @xmath19 .
when @xmath35 the power spectral density may be calculated for each one of the two linear approximations @xmath36 integrating over all frequencies @xmath27 gives @xmath37 for each of the two zero order expansions .
this is due to the fluctuation - dissipation theorem that links the dissipative force to the constant in front of @xmath38 in the motion equation .
the feynman diagrams are shown in fig.[fig : ac ] with the graphic representation of the solution .
when two @xmath39 are connected , the result is a tadpole type diagram .
indeed the mean value of @xmath40 does not depend on @xmath41 if the conditions are stationary . in this context
it should be reminded that the fluctuation - dissipation mechanism has been extended to the perturbed case .
this generalization is to be intended as a limit of the methods and it is based on the physical assumption that the results are slightly modified as it is assumed in field theory techniques .
the final result is @xmath42}\left[\frac{(\omega^2+{4\omega_{0}^{2}\over q^2 } ) ( \omega^2+{\omega_{0}^{2}\over q^2})-64\omega_{0}^{4}-{24\omega_{0}^{4}\over q^2}}{\omega^2 - 8\omega_{0}^{2}+{\omega_{0}^{2}\over q^2}}+ \right .
\\ & & \left .
\frac{{32\omega_{0}^{4}\over q^2}(\omega^2-\omega_{0}^{2})+{\omega^2 \omega_{0}^{2}\over q^2}({4\omega_{0}^{2}\over q^2}-3\omega^2)+\omega^2(10 \omega^2\omega_{0}^{2}-16\omega_{0}^{4}-\omega^4)}{(\omega^2 - 2\omega_{0}^{2})^2 + { \omega^2\omega_{0}^{2}\over q^2}}\left.\right]\right\}\end{aligned}\ ] ] where the first line represents the graphs ( 0),(2 ) and ( 3 ) in fig.[fig : ac ] and the corresponding contribution modifies the definitions of the parameters but does not introduce a different dependence on time .
this means that the typical shape of the power spectral density is not changed excepted some small corrections of its parameters .
the characteristics of the spectrum are indeed changed by the other contributions involving convolution integrals @xcite .
these terms correspond to the graphs ( 1 ) and ( 4 ) that introduce a new feature in the shape of the spectrum .
indeed , they modify the power spectral density at low frequency for @xmath43 . in this range
the spectrum decreases with increasing @xmath44 .
the spectral density changes its shape when the value of various parameters is modified , but there is still a tail for that interval falling as @xmath45 for @xmath46 .
moreover a new feature accounted for in the formula above is a new peak that is produced at a frequency which is approximately twice the value of the main peak frequency .
all these analytical results are obtained using a perturbative method that assumes that the system is never very far from the equilibrium conditions .
if the system is allowed to go back and forth between the two minima of the potential function , an alternative approach is needed in order to simulate the dynamics of the system .
a simulation code has been carried out and used to analyse different interesting cases .
the physical picture is that the impurity has some chance of oscillating between two equilibrium positions acquiring energy from thermal excitations .
it may be pushed into one direction or the other isotropically and the representation of those fluctuations is simply given by the stochastic term .
if some kicks are in the same direction , energy is gained to overcome the potential barrier .
this model reproduces the competition between thermal excitation and maxima between wells in the potential function .
if the temperature is high the passage through the barrier has more probabilities to occur as it might be expected since @xmath3 is a statistical definition of energy . at this point
there is no straightforward way for keeping the damping term in the non - linear regime , apart from making physical guesses about the type of damping process .
choosing other forms would mean different spectral densities since the dynamics would be different .
the relation between the random force and the damping term that is defined by the fluctuation - dissipation theorem in the linear approximation , has been used in the stochastic equation outside the two regions near the minima as well .
it provides a solution whose distribution in stationary conditions satisfies the boltzmann one . in general the langevin equation
implemented in the simulation code corresponds to a fokker - planck equation for the probability distribution that has the boltzmann distribution as its stationary solution .
this is one of the motivations to keeping the same expressions for the damping force and the random term in the non - linear case .
the other motivation is that the nature of the damping effect is often defined by the properties of the heat - bath rather than on the mechanical characteristics of the physical system in contact with it .
another point worthy of noting is that the natural scale for the x - axis is related with the magnitude of the fluctuating forces .
this is why the variable @xmath16 has been used throughout our analyses .
the analogy existing between the schredinger equation and the fokker - planck equation can be exploited and an algorithm may be written based on this similarity . in particular the drift term and the diffusion term in the fokker - planck equation ,
play the same role of the potential function and kinetic term in the hamiltonian operator . using the same rules for path - integral calculations ,
the evolution operator may be factored in such a way that successive steps are performed which make the system fluctuate according to the diffusion term or make a step according to the deterministic term .
depending upon the required precision of the stepping in the simulated states of the system , a particular number of intermediate steps is needed to make the system evolve from its state to the successive one .
this technique is simply an application of the baker - campbell - hausdorff formula to a stochastic phenomenon @xcite .
recall that large values of @xmath19 imply high @xmath3 or flat wells . in this high energy regime ,
the dynamics at low frequency are such that there is a constant trend as in harmonic oscillators damped by a viscous force . in fig.[fig : ad ] the spectral density is shown for @xmath47 .
no tail is appreciated at low frequency . if the value of @xmath19 is decreased the shape of the potential function is deeply characterized by two well defined minima and this affects the dynamics at low frequency where the trend of the power spectrum @xmath27 is not constant . on the contrary
there is a slope at low frequency that in fig.[fig : ae ] is quite visible and fitted with @xmath48 for the specific case of @xmath49 . indeed according to the tests i made using different values for the parameter @xmath19 the low frequency tail has the form @xmath0 with @xmath46 .
the dynamical model we constructed shows a low frequency tail in its spectrum .
how is this system affected by a frequency independent damping force ? will the low frequency tail be attenuated by this external force ? this system is a good candidate for investigation as the main fundamental modes are easily detected and suppressed by actively feeding back to the resonant motion .
the damping we will use is an external viscous force proportional to @xmath50 .
our aim is to find out whether an external damping mechanism can suppress @xmath27 at low @xmath51 .
some problems are to be faced in numerical studies .
the numerical fluctuations in simulations are the statistical errors that affect the mean value .
they are reduced by averaging over many sets of data in order to reduce the deviations the statistical fluctuations are especially large at low frequency so that if we are interested in obtaining a smooth curve with small error bars , many runs must be averaged over . in fig.[fig :
af ] the spectral density is shown when the damping is applied .
only one of the peaks is well defined and indeed it becomes sharper since the damping force limits the motion to small oscillations around one of the minima of the potential function .
the situation at low frequency is more interesting .
the values at @xmath52 are not affected . in the range below the peak
the low frequency tail is changed as @xmath53 .
a similar trend may be obtained by increasing the depth of the wells so that the motion is mostly confined to oscillations inside one of the wells .
many models have been proposed which involve a tail @xmath54 , as it is typical of electronic circuits noise . for mirrors the value of @xmath2 for a spectrum @xmath0 is not certain .
in fact , making experimental measurements of the fluctuations at low frequency is difficult even for state - of - the - art optics reasearch .
experimental data are mostly based upon resonance measurements and spectral curves are extrapolated for very low frequencies . for lack of off - resonance measurements linear equations
have been studied with a complex elastic force @xcite .
the random - walk problem studied in this work focuses on only one property that is the existance of more than one equilibrium configuration .
a similar model is used to reproduce the switching of polymeric units between two positions which are energetically equivalent .
the simulation program has also been modified in order to investigate what may happen if there are more than two minima for the potential function .
instead of the two wells , a periodic potential has been studied giving the same results . in fig.[fig : ag ] the spectra obtained for two values of @xmath19 are shown .
the fluctuation - dissipation theorem is extended to the general case although it could only be applied in the linear approximation near one of the minima of the potential function .
also @xmath8 may be scaled in order to have @xmath22 in the potential function @xmath55 from the physical point of view a potential barrier is needed to represent the limits of the periodic structure that otherwise would be infinitely large , and this is not realistic . in this case
a normalized stationary solution of the fokker - planck equation is the corresponding boltzmann distribution which takes into account the modified potential function . in other words what is needed
is a differentiable term that provides @xmath56 for both the limits @xmath57 @xcite .
we assumed that the size of the periodic structure is so large that no border condition affects the model .
this unidimensional model is characterized by a tail at low frequency in its power spectral density that is not modified even when a frequency independent damping force is applied .
the data are obtained from a time domain simulation .
we have considered both a simple case for a two wells potential function and the more general case of a periodic frame with oscillations between the minima .
the fluctuating motion of the degree of freedom and the associated dissipation are a model to produce a typical low frequency tail @xmath0 in the power spectral density @xmath27 .
the slope of such low frequency tail can be tuned by varying the constants ; it depends on the competition between the temperature and the depth of the wells .
thanks are due to director barry c. barish and deputy director gary h. sanders of ligo laboratories for generously giving me the opportunity of going on with my studies and dr .
riccardo desalvo for mediating my approach to ligo project .
99 shang - keng ma _ modern theory of critical phenomena _ a.b.p .
1976 jean zinn - justin _ quantum field theory and critical phenomena _
oxford university press 1996 g.curci and e.dambrosio _ physica a _ 273 n.3 - 4 ( 1999 ) 329 - 351 a.h.nayfeh _ nonlinear oscillations _ john wiley & sons ( 1979 ) j.guckenheimer , p.holmes _ nonlinear oscillations , dynamical systems and bifurcations of vector fields _ springer verlag ( 1983 ) mikio nakami _ stochastic quantization _ springer verlag ( 1991 ) m.beccaria , g.curci _ phys
d _ 49 ( 1994 ) 2590 - 2596 m.beccaria , g.curci , e.dambrosio _ phys .
d. _ 58 ( 1998 )
i d 065009 p.saulson _ phys .
d _ 40 ( 1990 ) 2437 - 2445 a.d.gillespie _ thermal noise in the initial ligo + interferometer
_ ph.d.thesis ( 1995 ) caltech f.bondu , p.hello , j.y.vinet p _ physics letters a _ 246 ( 1998 ) 227 - 236 p.saulson _ interferometric gravitational wave detectors _ world scientific ( 1994 ) v.b.braginsky , v.p.mitrofanov , v.i.panov _ systems with small dissipation _ university of chicago press ( 1985 ) g.parisi , n.sourlas _ nucl .
_ 206b ( 1982 ) 321 - 331 | this study investigates the modifications of the thermal spectrum , at low frequency , induced by an external damping on a system in heat contact with internal fluctuating impurities .
those impurities can move among locations and their oscillations are associated with a loss function depending on the model .
the fluctuation properties of the system are provided by a potential function shaped by wells , in such a way that jumps between the stationary positions are allowed .
the power spectral density associated with this dissipation mechanism shows a @xmath0 tail .
the interest of this problem is that many systems are characterized by a typical @xmath0 spectral tail at low frequency .
the model presented in this article is based on a threshold type behaviour and its generality allows applications in several fields .
the effects of an external force , introduced to produce damping , are studied by using both analytical techniques and numerical simulations .
the results obtained with the present model show that no reduction of the power spectral density is appreciable below the main peak of the spectral density .
+ pacs numbers : 05.40-a , 05.10-a |
wormholes are hypothetical objects which connect two distant parts of the same spacetime or two different spacetimes by a throat - like object which has the minimum radius of the spacetime .
although a wormhole solution first entered the physics literature in 1916 @xcite , the concept was first considered seriously in 1935 by einstein and rosen @xcite which was later called einstein - rosen bridge but the word wormhole was first time coined by wheeler @xcite in 1957 .
a more interesting analysis of wormholes was performed by morris and thorne in 1988 and they presented a new kind of wormhole ( traversable wormhole ) for the first time @xcite .
it was known from before that matter we need to support such a geometry violates the weak and strong energy conditions near the throat @xcite .
morris and thorne reconsidered these conditions for a traversable wormhole @xcite . since the matter that supports this geometry does nt satisfy the common energy conditions they called it exotic. an example of exotic matter is matter with negative energy density @xcite .
another property of these wormholes is the possibility of transforming them into time machines for backward time traveling @xcite and thereby , perhaps for causality violation by closed timelike curves .
teo @xcite found that the null energy condition ( @xmath1 ) is violated by stationary , axially symmetric , traversable wormholes but there can be classes of geodesics which do not cross energy condition violating regions . another case of evolving wormholes may alter this situation@xcite .
it is known that we can have violations of weak energy condition ( @xmath2 ) due to some quantum mechanical effects ( such as casimir effect ) @xcite .
if we search for this effect in the history of evolving cosmos , we can find such a situation at the quantum cosmological era when quantum gravity is dominant .
following the inflation theory by a. guth @xcite it has been supposed that non - trivial topological objects such as microscopic wormholes may have been formed during that era and then enlarged to macroscopic objects with expansion of the universe @xcite .
although most studies have focused on four dimensions , with the advent of string theory that demands higher dimensional spacetimes , it is natural to examine the possibility of wormholes beyond the ordinary four dimensions .
euclidean wormholes in string theory have recently been studied @xcite as solutions of supergravities . with this motivation
, we are going to choose the generalized friedmann - robertson - walker ( @xmath3 ) metric in @xmath0-dimensions and investigate exact lorentzian wormhole geometries with spherically symmetry .
this paper is organized in the following manner : in sec .
[ field ] we present the ansatz metric and the resulting solutions in 4-dimensions . in section [ ndim ] ,
we extend the solutions to ( n+1)-dimensions . in section [ energy ]
, we investigate the corresponding energy - momentum tensor and determine the exoticity parameter .
the last section is devoted to conclusions and closing remarks .
it is very common today that we start studying cosmology with the so - called cosmological principle. it states that the universe at large scale is homogeneous and isotropic . with this assumption ,
we find out that the metric we need to demonstrate such a spacetime in 4-dimensions is as follows @xmath4 , \label{fe1}\ ] ] which is known as the robertson - walker ( rw ) metric @xcite .
the coordinate system @xmath5 used here is the so - called co - moving coordinate system. @xmath6 is the scale factor and the only dynamical parameter to determine . @xmath7 correspond to spatially flat , closed and open spacetimes respectively .
this metric contains a high degree of symmetry which is demonstrated by its six killing vectors . for our aim
, we need to generalize this metric to @xmath0-dimensions and reduce its symmetry .
we break its homogeneity by replacing @xmath8 with @xmath9 .
this metric is still isotropic about @xmath10 but not necessarily homogeneous .
we therefore write our ansatz metric as @xmath11 , \label{fe2}\ ] ] where @xmath12 is an unknown function .
it is clear that the robertson - walker metric is a special case of this metric . with this ansatz metric
, we look at our equations .
we start with @xmath13 and then extend the solutions to arbitrary @xmath14 . in the first step we write our ansatz metric for 5-dimensions ( n=4 ) @xmath15 . \label{fe201}\ ] ] the non - vanishing components of the einstein tensor for our ansatz metric read @xmath16 @xmath17 @xmath18 @xmath19 in which dot is derivative with respect to @xmath20 , while prime is derivative with respect to @xmath21 .
such a geometry is supported by an anisotropic , diagonal energy - momentum tensor : @xmath22 , \label{fe7}\ ] ] @xmath23 , \label{fe8}\ ] ] @xmath24.\label{fe9}\ ] ] in these relations @xmath25 and @xmath26 are the transverse and radial pressures , respectively . here , we assume the following equation of state @xmath27 where @xmath28 depends on @xmath14 the dimension of the space .
this equation reduces to the vacuum equation of state @xmath29 when @xmath30 . using equations ( [ fe7]-[fe9 ] ) and relation ( [ fe10 ] ) in 5-dimensions ( n=4 )
, we obtain : @xmath31 + \\\nonumber & & \frac{1}{1+\gamma}\left [ \left[-3(\frac{r\ddot r+\dot r^2}{r^2})-3\frac{a}{r^2r^2(1+a)}\right ] + \gamma \left[-3(\frac{r\ddot r+\dot r^2}{r^2})-\frac{ra^{\prime}+a(1+a)}{r^2r^2(1+a)^2}\right]\right]=0.\label{fe11}\end{aligned}\ ] ] fortunately , this equation can be separated into radial and temporal equations .
@xmath32 this equation can be easily solved and has the following solutions : @xmath33 @xmath34 the energy - momentum tensor needed to support this geometry has the following components : @xmath35 @xmath36 and @xmath37
the solutions we obtained in 5-dimensional spacetime can be easily extended to @xmath0-dimensions . the resulting solutions in ( n+1)-dimensions read @xmath38 @xmath39 following these solutions
, the energy - momentum tensor which supports this structure in ( n+1)-dimensions will be : @xmath40 @xmath41 @xmath42 as a check for the correctness of these solutions one can see that they reduce to the solutions of @xcite for n=3 with proper definition for the integration constants . now let us have a look at radial behavior of the solutions . in the paper by morris and thorne @xcite ,
the metric of the wormhole is written in the form : @xmath43 in which @xmath44 is the shape function and the throat radius satisfies @xmath45 .
if the equation @xmath46 has any root @xmath47 and simultaneously @xmath48 for @xmath49 then we will have a wormhole and @xmath47 gives the throat radius of the wormhole . in our solutions , with choosing @xmath50 the condition for the existence of wormholes will be @xmath51 and @xmath52 where @xmath53 and @xmath54 are constants related to the integration constants : @xmath55 solving equation ( [ fend7 ] ) analytically is not possible except for special values of n , as presented in @xcite for n=3 .
then for investigating the properties of the solutions , we plot @xmath56 against @xmath21 in figure ( [ i1]-[i2 ] ) .
we classify the solutions in the following manner : in the case ( a ) , @xmath57 and @xmath58 , we see from fig.[i1 ] that we have a lower limit on @xmath21 which corresponds to the throat radius of the wormhole and we see also that there is no upper limit on @xmath21 which reminds us that we have an open spacetime . for the choice ( b ) , @xmath59 and @xmath58
, we have lower and upper limits on @xmath21 .
the lower limit corresponds to the throat radius of the wormhole and the upper limit signifies a closed spacetime . in the cases ( a ) and ( b ) , the kretschman scalar blows up at @xmath10 but , this point is not included in our physical spacetime with proper signature .
case ( c ) , @xmath60 and @xmath61 represents a naked singularity in a closed cosmological background , because the kretschman scalar blows up at the origin ( @xmath10 ) .
case ( d ) , @xmath62 and @xmath63 leads to @xmath64 , which corresponds to the @xmath65 @xmath3 metric . case ( e ) , @xmath66 and @xmath67 , @xmath9 becomes negative and the metric s signature is not of cosmological interest . and finally , ( f ) , @xmath68 and @xmath69 , @xmath9 is regular everywhere and represent a naked singularity , again but in an open universe .
we can also have a discussion on the cases which include @xmath70 . for case @xmath71 , from ( [ fend2 ] )
we obtain @xmath72 this equation with @xmath73 and @xmath58 , leads to a wormhole centered , open universe which is illustrated in fig.[i2 ] part ( g ) .
it is worth having a look at the ricci scalar ( @xmath74 ) . in
( n+1)-dimensional spacetime @xmath74 is : @xmath75 we can see that choosing @xmath76 or @xmath50 , constant curvature spacetime is retrieved . in the case
@xmath76 , we have a maximally symmetric de sitter spacetime .
we expect this result because , we see that the constant part of the ricci scalar reminds us a maximally symmetric spacetime curvature . the wormholes discussed in this paper are traversable .
we present two reasons here .
the first reasoning is based on the redshift of a signal emitted at the comoving coordinate @xmath77 and received by a distant observer . using the metric ( [ fe201 ] ) and for a radial beam ,
we obtain @xmath78 using this relation for two signals separated by @xmath79 in time when emitted ( and @xmath80 when detected ) , we obtain @xmath81 in which @xmath82 is the scale factor at the time of observation , and @xmath83 is the scale factor at the time of emission .
this leads to the exactly same relation as the cosmological redshift relation which shows that the wormhole does not introduce extra ( local ) redshift .
light signals , therefore can travel to the both sides of the throat and there is no horizon .
the second is based on the geodesic equation , which -for the metric ( [ fe201])- leads to @xmath84 and @xmath85 in which @xmath86 is an affine parameter along the geodesic .
the first equation has the following first integral @xmath87 since the proper distant element is @xmath88 , we see that there is no radial turning point and any particle can move in either radial directions at any point near to the wormhole , which clearly shows that the wormhole is traversable .
let us have a brief discussion on the observational consequences of these solutions .
we can have a variety of geodesics in a wormhole spacetime @xcite .
geodesics can pass through the throat right to the other part of the spacetime ( the parallel universe ) , or be deflected back to the same universe . as pointed out by cramer et .
@xcite , it should be possible -in principle- to detect wormholes via the gravitational lensing they cause .
this , however , depends on the wormhole being stable , which is not addressed in the present paper .
the @xmath0-dimensional version of diagonal elements of the energy- momentum tensor are were given in equations ( [ fend3]-[fend5 ] ) .
some points are interesting to mention about the energy - momentum tensor .
since we are looking for spherical structures in a cosmological background , @xmath25 , @xmath26 and @xmath89 should become almost @xmath21-independent at large @xmath21 .
in particular , for suitable values of @xmath28 , the solutions approach @xmath90 which correspond to dark energy ( cosmological constant ) .
the second point is the conservation equation of the energy - momentum tensor : @xmath91 which leads to @xmath92 our solutions satisfy this equation , as expected .
let us now investigate the exoticity parameter . for an isotropic fluid ,
the energy - momentum tensor is in the form @xmath93 and the exoticity parameter is defined according to @xmath94 where @xmath95 corresponds to the exotic matter and @xmath96 to the non - exotic matter . since we have an anisotropic medium , the energy - momentum tensor has the form @xmath97 .
we therefore take the average pressure @xmath98 taking this into account , we adopt the following , more general definition for @xmath99 : @xmath100 and modify ( [ e7 ] ) according to @xmath101 from ( [ fe10 ] ) we see that the modified exoticity parameter is everywhere zero and for all cases , which shows that we are on the border line between exotic and non - exotic matter according to the criterion
. the weak energy condition @xmath102 requires @xmath103 for every nonspacelike @xmath104 which leads to @xcite @xmath105 these equations and relations ( [ fend3]-[fend5 ] ) lead to @xmath106 @xmath107 @xmath108 these relations show that , with the choice @xmath109 , we ca nt have any wormhole without violating @xmath2 , or at least the border line at which we have the equality relation in ( [ e13 ] ) and ( [ e14 ] ) . in the case @xmath110 the relations ( [ e13 ] ) and ( [ e14 ] ) are both simultaneously satisfied and we have to investigate the relation ( [ e12 ] ) in order to see wether @xmath2 is satisfied .
it is interesting to note that for particular ranges of constants we have wormhole solutions which satisfy @xmath2 throughout the spacetime . as an example of these solutions
one can see case ( g ) in fig.[i2 ] .
we introduced time - dependent wormholes in an @xmath0-dimensional expanding cosmological background .
we presented an ansatz metric and a linear relation between the diagonal elements of the energy - momentum tensor . with these assumptions , we separated and solved the field equations .
solutions were classified into different categories with distinct geometries for the central object .
we distinguished two classes of solutions which represented lorentzian wormholes with expanding throat in closed and open universes .
we also found two classes of solutions that contain an intrinsic singularity in open and closed universes .
the other interesting solution was the familiar maximally symmetric de sitter spacetimes .
the ricci scalar was calculated and the constant curvature class of solutions were discussed .
the issue of the traversability of the wormhole solutions was considerd .
we investigated the energy - momentum tensor required to support these solutions .
our solutions led to an energy - momentum tensor which approaches the cosmological constant case far from the wormhole for @xmath111 .
finally , we investigated the exoticity parameter .
the exoticity parameter for the linear equation of state is zero everywhere which is border line between exotic and non - exotic matter .
we also checked the weak energy condition and found that the weak energy condition in the choice @xmath112 is violated except for the border line case .
for the case @xmath70 , @xmath2 is satisfied for suitable values of constants . | we discuss @xmath0-dimensional dynamical wormholes in an evolving cosmological background with a throat expanding with time .
these solutions are examined in the general relativity framework . a linear relation between diagonal elements of an anisotropic energy - momentum tensor
is used to obtain the solutions .
the energy - momentum tensor elements approach the vacuum case when we are far from the central object for one class of solutions . finally , we discuss the energy - momentum tensor which supports this geometry , taking into account the energy conditions . |
the intergalactic medium ( igm ) is thought to be photo - ionized by the uv radiation emitted by galaxies and quasars .
models of quasar absorption systems generally assume that the gas is exposed only to the mean background radiation that pervades the igm .
however , given that this background is produced by discrete sources , it can not be strictly uniform . the effect of fluctuations in the uv background on the statistics of the forest of weak hi ly@xmath0 absorption lines , seen in the spectra of distant quasars , has been the subject of a large number of papers ( e.g. , zuo 1992 ; fardal & shull 1993 ; croft et al . 1999 , 2002 ;
gnedin & hamilton 2002 ; linder et al .
2003 ; meiksin & white 2003 , 2004 ; croft 2004 ; mcdonald et al .
these theoretical studies found that because most of the ly@xmath0 lines arise in the low - density igm , far away from sources of ionizing radiation , global statistics of the forest are insensitive to fluctuations in the uv background .
however , fluctuations in the ionization rate may become detectable in the ly@xmath0 forest at at @xmath2 ( e.g. , meiksin & white 2003 , 2004 ) and fluctuations in the ionization rate have already been detected at @xmath3 from comparisons of the and ly@xmath0 forests ( e.g. , kriss et al.2001 ; smette et al .
2002 ; shull et al . 2004 ) .
in contrast to the ly@xmath0 forest , many metal and high column density ly@xmath0 absorbers are thought to arise in the extended halos of galaxies ( bahcall & spitzer 1969 ) , or at least in regions close to galaxies .
it is therefore not obvious that neglecting the contributions of local sources of ionizing radiation , which is currently common practice , is justified for such absorption systems .
indeed , it has been argued on the basis of metal - line column density ratios that there is evidence that some absorbers are exposed to a radiation field that is much softer than the general background , as would be expected if local stellar radiation were important ( e.g. , giroux & shull 1997 ; boksenberg , sargent , & rauch 2003 ) .
here we show that if we assume that a certain class of absorbers resides in the extended halos of a population of sources of uv radiation ( such as galaxies ) , then _ one can estimate the typical flux from local sources to which the absorbers are exposed using only the luminosity density of the sources and the rate of incidence ( i.e. , number per unit redshift ) of the absorbers .
_ we do not need to know the relation between the cross - section for absorption and the luminosity of the source .
we derive analytic formulas to estimate the moments of the local flux , as well as of the impact parameter and the luminosity of the central source , and apply them to galaxies and quasars .
we find that uv radiation from local galaxies may well be important for absorbers rarer than lyman limit ( ll ) systems ( in agreement with the recent work of miralda - escud 2005 ) and is likely far in excess of the background for absorbers as rare as damped ly@xmath0 ( dla ) systems .
we therefore conclude that the results from studies that employed ionization models of @xmath4 absorbers may need to be revised .
this paper is organized as follows . in [ sec : assumptions ] we state and discuss the approximations and assumptions which we use to estimate the local flux .
section [ sec : meanflux ] contains the derivation of the mean flux , which we show to be close to the median for a wide range of models in [ sec : medians ] .
section [ sec : comp ] presents two complementary methods to compare the local flux to the background flux and contains some general conclusions . using more restrictive assumptions , we derive expressions for the moments of the cross - section weighted flux , impact parameter , and luminosity in
[ sec : moments ] . in [ sec : gals ] and [ sec : quasars ] we compute the contributions of galaxies and quasars to the local ionization rate and we compare these with the background for various types of absorbers in [ sec : comparison ] , where we also estimate the global escape fraction for h ionizing photons .
finally , we summarize our main conclusions in [ sec : conclusions ] .
it is clearly beyond our means to compute the distribution of fluxes seen by a class of absorbers from first principles .
such a calculation would involve modeling radiative transfer in a characteristic volume of the universe , which would require specification of the full phase space distribution of elements and sources as well as of the spectra emitted by the sources .
we would then still need to decide what we mean by _ the _ flux seen by an absorber as the absorbing gas clouds have a finite size .
it is therefore necessary to make a number of simplifying assumptions .
we will present and discuss these assumptions in [ sec : assumptions ] and estimate the mean flux in [ sec : meanflux ] . in [ sec : medians ] we will show that the median flux , which we can only compute by making additional assumptions , is generally close to the mean . consider a certain class of absorption systems with an observed rate of incidence @xmath5 . in practice
, we will define a class of absorbers by specifying an ion and a minimum column density @xmath6 . to enable us to compute the flux from local sources ( e.g. , galaxies ) to which the absorbers are exposed
, we will make the following simplifying assumptions and approximations regarding absorbers with @xmath7 : 1 .
all absorbers reside in spherical halos which are each centered around a single source .
the flux seen by the absorber is dominated by its central source .
the probability that a sightline with impact parameter @xmath8 relative to a source with luminosity @xmath9 intersects an absorber residing in the halo around the source is @xmath10 for @xmath11 and zero otherwise [ i.e. , the absorbing halo has a finite radius @xmath12 .
all of the gas contributing to the absorption resides at a distance @xmath13 from the central source .
the central source is point - like .
the product @xmath14 , where @xmath15 is the fraction of the emitted radiation that is able to propagate to the absorber , is independent of the luminosity of the source .
we will now discuss each of these assumptions in turn .
the assumption that the absorbers are ( or reside in ) spherical halos centered on single sources will clearly break down for absorbers with gas densities around the cosmic mean , as these are typically distributed along sheets and filaments .
however , it is probably a reasonable approximation for gas with densities @xmath16 times the cosmic mean and/or for heavy elements whose distributions are likely concentrated towards galaxies .
the second assumption ( radiation from sources in nearby halos is negligible ) is conservative .
we will show in [ sec : gals ] that it is likely to be a good approximation if the sources are galaxies and if the flux from the central source exceeds the background .
however , if the sources are strongly clustered , then radiation from nearby halos could provide a non - negligible contribution to the ionizing flux seen by the absorbers .
assumption three ( @xmath17 for @xmath18 ) is probably reasonable provided that we define a class by its cumulative rate of incidence @xmath19 .
since @xmath19 must decrease with increasing @xmath20 , the assumption implies that larger columns typically arise in sightlines with smaller impact parameters and hence that @xmath21 is a decreasing function of @xmath22 .
although it is possible to fabricate density distributions that would violate this assumption , it can not be violated systematically because that would mean that the cosmic density of the ion in question , which is proportional to the integral @xmath23 , would diverge at large n. moreover , the assumption that column density typically increases with decreasing impact parameter is supported by a large number of observational studies of the relation between absorption systems and galaxies ( e.g. , lanzetta & bowen 1990 ; bergeron & boisse 1991 ; lanzetta et al.1995 ; chen et al .
1998 , 2001a , 2001b ; bowen , pettini , & blades 2002 ; penton , stocke , & shull 2002 ; adelberger et al . 2003 ) .
one might naively think that the assumption that column density increases with decreasing impact parameter would break down in the region within which the ionizing flux is dominated by the central source ( i.e. , the proximity zone ) .
this would be a problem , as this is one of the regimes of interest here .
however , the arguments given above indicate that this can not generally be the case . before discussing our remaining assumptions
, we will show that this is not a paradox . although the neutral hydrogen column density in a sightline passing close to a source can be small if the proximity zone is very large ( as can be the case for bright quasars ) , the column density is unlikely to increase with the impact parameter for any given source . to see this , consider that in photo - ionization equilibrium the ionization balance is determined by the ratio of the ionizing flux to the gas density , which is independent of the radius if the flux is dominated by the central source and if the gas density profile is isothermal , which should be close to the situation of interest here .
in general , for an optically thin gas in photo - ionization equilibrium that is irradiated by a central source we have @xmath24 if @xmath25 , where @xmath26 is the photo - ionization rate for neutral hydrogen . now consider a spherical cloud with density profile @xmath27 , where @xmath28 is the number density of the species of interest .
the column density in a sightline with impact parameter @xmath8 is @xmath29 for @xmath30 the dominant contribution to the column density comes from @xmath31 and @xmath32 . for example , for a singular isothermal sphere ( @xmath33 )
we have @xmath34 , which is the scaling derived by schaye ( 2001a , equations 3 and 4 ) based on more general arguments .
hence , for reasonable density profiles the column density will increase with decreasing impact parameter .
there are two reasons why we would in general expect this trend to be even stronger for metals than for .
first , since metals are produced in stars , it is natural to assume that their abundances relative to hydrogen are decreasing functions of @xmath35 .
second , the ionization balance of many of the metals of interest will asymptote to a constant as more stellar radiation is added , because ( ordinary ) stars essentially do not emit above 4 rydberg ( ryd ) and metals can not be fully photo - ionized without these energetic photons assumption four is also conservative : assuming that all the absorbing gas resides in a thin shell with radius @xmath35 minimizes the flux from the central source .
in fact , for a typical absorber it is probably a reasonable approximation to assume that all absorbing gas is at the maximum possible radius @xmath35 .
we already showed that for reasonable density profiles almost all the absorption takes place in the densest part of the gas cloud that is intersected by the sightline , at radii slightly greater than @xmath8 .
the median impact parameter of random sightlines through a halo of radius @xmath35 is @xmath36 , slightly smaller than @xmath35 .
hence , the column densities are typically dominated by gas at a distance @xmath37 from the source .
we therefore expect that the assumption that all of the gas that contributes to the column density resides at a radius @xmath35 will , for a typical absorber , result in an underestimate of the flux seen by the absorber by only a factor of a few .
note that because of assumption four , the flux seen by absorbers in some finite column density interval @xmath38 is identical to that seen by absorbers with @xmath39 .
assumption five ( point - like sources ) will be a good approximation for sightlines with @xmath40 , where @xmath41 is the radius of the source .
for smaller impact parameters it will , however , lead to a large overestimate of the flux .
hence , our model should not be applied to absorbers with @xmath42 .
finally , assumption six ( @xmath14 is independent of @xmath9 ) can be justified only by our ignorance .
given that we are still a long way off from being able to compute this factor from first principles , it would be hard to justify any particular dependence on @xmath9 .
assuming it is constant minimizes the number of free parameters .
the fraction of photons that is able to propagate to the absorber , @xmath15 , should not be confused with the conventional escape fraction @xmath43 , which is the fraction of emitted photons that is able to escape the entire halo surrounding the source .
they are different in general because the absorbers under consideration may in fact determine the global escape fraction @xmath43 .
thus , we have @xmath44 in general with @xmath45 asymptoting to @xmath43 for sufficiently small @xmath20 and to unity for sufficiently large @xmath20 ( i.e. , for column densities comparable to those surrounding the sources ) .
the covering factor @xmath46 ( @xmath47 ) is unlikely to be much smaller than unity given that absorption line systems typically contain multiple components and that sightlines which pass near to galaxies nearly always show absorption by metals and hydrogen ( e.g. , bergeron & boisse 1991 ; lanzetta et al .
1995 ; adelberger et al .
2003 ) .
we are now in a position to estimate the flux from the central source to which a typical absorber is exposed .
we will compute the mean flux @xmath48 by averaging the flux seen by an absorber over all source luminosities , weighted by the total cross - section for absorption . under the assumptions discussed above , a source of luminosity @xmath9
provides a ( proper ) cross - section for absorption of @xmath49 and the flux seen by the absorber is @xmath50 the average number of absorption systems per unit redshift and luminosity along a random sightline is @xmath51 where @xmath52 is the comoving luminosity function of the sources ( i.e. , the number of sources per unit comoving volume and per unit luminosity ) , @xmath53 is the speed of light and @xmath54 is the hubble parameter . the mean flux from the central source to which a class of absorbers with rate of incidence @xmath55 is exposed is thus @xmath56^{-1 } \int { d^2{\cal n } \over dz dl } f dl \\ & = & \left [ { d{\cal n } \over dz}(z , n_{\rm min})\right ] ^{-1}{c ( 1+z)^2 \over 4 h(z ) } f_{{\rm esc},n } f_{\rm cov } \int l\phi(z , l ) dl .
\label{eq : barf}\end{aligned}\ ] ] the important thing to note is that the result is independent of @xmath35 .
thus , _ we do not need to specify the function @xmath57 in order to compute the mean flux @xmath48_. equation ( [ eq : barf ] ) is a key result of this paper , it provides a means to estimate the flux seen by a class of absorbers from its observed rate of incidence and the observed luminosity function of sources .
the mean flux is proportional to the luminosity density @xmath58 . if the sources are galaxies , then this has the convenient consequence that @xmath48 is approximately proportional to the comoving star formation density because most of the uv radiation is emitted by young stars .
if the luminosity function is a schechter ( 1976 ) function , @xmath59 then equation ( [ eq : barf ] ) becomes @xmath60^{-1}{c ( 1+z)^2 \over 4 h } f_{{\rm esc},n } f_{\rm cov } k \phi_\ast l_\ast \gamma(2+\alpha ) , \label{eq : barf_sf}\ ] ] where @xmath61 is the gamma function and @xmath62 if @xmath63 is given per magnitude and @xmath64 if @xmath63 is given per @xmath65 ( unless specified otherwise , we will hereafter assume @xmath64 and omit @xmath66 from the equations ) . if , from some reason , the absorbers only reside in halos around objects with @xmath67 , then the integral should of course be computed using these limits . for physically sensible luminosity functions ( i.e. , those that yield a finite luminosity density ) , @xmath68 and @xmath48 will thus be insensitive to the limits as long as @xmath69 .
to assess the importance of local sources , we need to compare the mean flux from the central source to the background radiation field .
if the intensity of the background radiation is known , then we can directly compare it to our estimate of the mean flux [ equation ( [ eq : barf ] ) ] and infer the critical rate of incidence below which local radiation dominates : @xmath70 where @xmath71 is the background flux . for sufficiently large @xmath20 ( i.e. , small @xmath72 ) @xmath73 and
the critical rate of incidence can be determined relatively accurately .
but if @xmath20 is much smaller than the column in front of the sources , then @xmath74 and the critical rate becomes proportional to the highly uncertain global escape fraction .
on the other hand , if we assume that the population of sources under consideration dominates the background , that the proper mean free path of the photons @xmath75 is known , and that @xmath76 ( such that cosmological redshift and evolution may be neglected ) , then the background flux is given by @xmath77 substituting this into equation ( [ eq : dndzcrit1 ] ) , we see that the critical rate of incidence becomes independent of both the source luminosity function and the intensity of the background : @xmath78 where @xmath79 is the redshift interval corresponding to the mean free path . for sufficiently large @xmath20 , @xmath73 and
the critical rate of incidence is inversely proportional to the highly uncertain global escape fraction @xmath80 .
but for sufficiently small @xmath20 , @xmath81 and the critical rate of incidence becomes independent of the escape fraction .
hence , for column densities similar to those in front of typical sources , using equation ( [ eq : dndzcrit1 ] ) will probably give a more accurate estimate , whereas using equation ( [ eq : dndzcrit2 ] ) will likely yield the most accurate estimate for lower column densities .
in general , we can measure the global escape fraction @xmath43 by comparing the two estimates , which comes down to comparing equation ( [ eq : fbg ] ) with the measured intensity of the background .
of course , such a measurement is subject to the uncertainties in the luminosity density and in @xmath82 . moreover , if other sources contribute to the background , then the inferred escape fraction should be interpreted as an upper limit .
equation ( [ eq : dndzcrit2 ] ) corresponds with equation ( 4 ) of miralda - escud ( 2005 ) if we neglect evolution of the rate of incidence on timescales @xmath83 , as we have done .
however , miralda - escud ( 2005 ) implicitly assumed @xmath84 and @xmath85 .
the last assumption may well break down for column densities similar to those in front of the sources , such as damped ly@xmath0 ( dla ) systems .
the mean free path depends on the frequency of the photons . for energies between 1 and 4 rydbergs ( i.e. , between the ionization potentials of and ) , the mean free path depends mainly on the column density distribution of absorbers with column densities similar to those of ll systems ( @xmath86 )
unfortunately , this is exactly where the distribution is most uncertain , because the absorption lines are on the flat part of the curve of growth which makes it difficult to measure accurate column densities .
following miralda - escud ( 2003 ) , if we assume that the column density distribution is a power - law @xmath87 with slope @xmath88 , as appears to be a reasonable fit at lower column densities ( e.g. , kim et al .
2002 ) , then it can be shown that @xmath89 , where @xmath90 is the mean spacing between ll systems .
this implies that @xmath91 and hence that the critical rate of incidence is proportional to that of ll systems : @xmath92 which is to be considered an order of magnitude estimate because it makes strong assumptions about the shape of the column density distribution .
note that this equation is less general as the previous ones because it applies only to photons with energy above ( but near ) 1 ryd .
if we assume that @xmath93 , which is probably reasonable for @xmath94 , then we have @xmath95 , in agreement with miralda - escud ( 2005 ) .
thus , we conclude that _ for hydrogen , local ionizing radiation is generally unimportant for @xmath96 . for lyman limit systems
local radiation is important , provided that the sources associated with these systems dominate the uv background _ ( we expect this to be the case if the background is dominated by stars , as opposed to quasars ) .
the rate of incidence of dlas is much smaller than that of lyman limit systems ( about a factor 10 smaller at @xmath97 ; storrie - lombardi et al . 1994 ; storrie - lombardi & wolfe 2000 ) which implies that _ local sources dominate the ionization rate in dlas if the same population also dominates the uv background_. note that @xmath48 may exceed @xmath71 by a very large factor if @xmath98 , as may well be the case for dlas .
although we can compute the mean flux without specifying the function @xmath57 , the same is not true for a number of other interesting quantities , such as the variance of the flux and the cross - section weighted mean luminosity and impact parameter .
it is , however , possible to obtain analytic solutions for these and other quantities if we make the following additional assumptions : 1 .
the luminosity function is a schechter function . 2 .
the function @xmath35 is a power - law of @xmath9 . if @xmath99 ( assumption 8) , then the normalization factor @xmath100 is determined by the observed rate of incidence : @xmath101 which follows from integrating equation ( [ eq : dn / dz ] ) , and the flux ( eq . [ [ eq : f ] ] ) is given by @xmath102 where @xmath103 under these assumptions we can also solve numerically for the median luminosity and thus for the median flux and impact parameter .
but before doing so in [ sec : medians ] , let us first consider what a reasonable value of @xmath104 might be .
it is difficult to predict what the dependence of @xmath35 on @xmath9 should be , which is why it is so convenient that @xmath48 is independent of this unknown function . in particular
, it could well be different for different elements and for different @xmath20 .
however , a reasonable guess might be that @xmath35 scales with the virial radius , @xmath105 , which yields @xmath106 $ ] if @xmath107 .
on the other hand , since @xmath108 ( see [ sec : assumptions ] ) which becomes @xmath109 for a singular isothermal profile , it may be that @xmath110 ( because @xmath111 is constant ) and that @xmath112 $ ] is thus a better guess .
semi - analytic models of galaxy formation ( e.g. , benson et al .
2000 ) predict that mass - to - light ratios reach a minimum around @xmath113 , but that the dependence of @xmath114 on @xmath115 is weak .
for example , @xmath116 , in which case we would expect @xmath117 , is a good fit to the predictions of benson et al .
( 2000 ) ( who modeled the @xmath118-band at @xmath119 for a @xmath120cdm model and a mass range @xmath121 ) . assuming that @xmath57 is indeed a power - law , it is possible to measure @xmath104 by imaging fields containing quasars .
chen et al.(2001a , 2001b ) have done just this for the @xmath118-band ( the bluer bands are the most relevant here since we are interested in ionizing radiation ) at @xmath122 .
they found @xmath123 for and @xmath124 for @xmath125 , in good agreement with our naive estimates .
note that @xmath126 is special as it gives a flux that is independent of luminosity : @xmath127 .
hence , if @xmath104 is indeed close to 0.5 , as both observations and ( handwavy ) theoretical arguments seem to suggest , then this has the important implication that _ absorbers belonging to a given class [ defined by their rate of incidence @xmath128 , are nearly all exposed to approximately the same flux_. armed with the fitting functions ( [ eq : lf ] ) and ( [ eq : r ] ) , it is easy to derive a number of characteristics of the absorbers . substituting these functions into equation ( [ eq : barf ] ) , we obtain the following expression for the mean flux , @xmath129 the dependence of @xmath48 on @xmath104 is only apparent , because @xmath130 . as was shown in [ sec : meanflux ] , the mean flux is independent of @xmath57 ( and thus @xmath104 ) when expressed in physical units . more generally , it is not difficult to show that the @xmath131-th moment of the flux is given by , @xmath132 \over \gamma(1+\alpha+2\beta)}.\ ] ] note that @xmath133 if @xmath126 .
similarly to the moments of the flux , we can derive the moments of the impact parameter @xmath134 \over \gamma(1+\alpha+2\beta)},\ ] ] and the luminosity @xmath135 recall that all these moments are weighted by the cross - section for absorption .
putting in numbers , for @xmath136 and @xmath137 ( -1.2 ) we obtain @xmath138 ( @xmath139 ) and @xmath140 ( @xmath141 ) .
figure [ fig : meanmed ] shows @xmath142 ( _ left _ ) , @xmath143 ( _ middle _ ) , and @xmath144 ( _ right _ ) , all as a function of @xmath104 and for two values of @xmath0 as indicated in the figure . because @xmath145 is a monotonic function of @xmath9 ( which is a consequence of our assumptions 4 ( [ sec : assumptions ] ) and 8) , the median flux @xmath146 , and it can be seen from equation ( [ eq : dn / dz ] ) that the median luminosity @xmath147 can be obtained by solving the following equation @xmath148 it is easy to see from equation ( [ eq : f_pl ] ) that for @xmath126 we have @xmath149 because in this case @xmath145 is independent of @xmath9 .
the left - hand panel of fig .
[ fig : meanmed ] shows how the median and the mean flux compare for other values of @xmath104 and for two values of the slope of the faint - end of the luminosity function : @xmath150 and @xmath151 .
the results shown in this figure are insensitive to @xmath152 and @xmath153 provided that @xmath154 .
the mean flux is typically greater than the median , but they differ by less than a factor of two for @xmath155 if @xmath150 ( @xmath156 if @xmath157 ) . note that the power - law model for @xmath57 becomes unphysical for low @xmath104 because @xmath158 for @xmath159 .
the median impact parameters are shown in the middle panel of fig .
[ fig : meanmed ] , again for two values of @xmath0 as indicated in the figure . as was the case for the flux ,
the medians are typically close to the means .
for the mean flux to be an order of magnitude greater than the median , we would require @xmath160 .
such a large value would be in conflict with the observations of chen et al .
( 2001a , 2001b ) and would probably imply an extremely low value of @xmath161 ( @xmath162 for @xmath163 $ ] ) .
hence , although the median flux can not be computed without specifying @xmath57 , it is within a factor of a few of the mean for plausible models .
this gives us confidence that equation ( [ eq : barf ] ) gives a reasonable estimate of the typical flux seen by the absorbers .
thus , we expect that in general @xmath164 .
the two types of sources that we will consider are galaxies and quasars .
we will compute the characteristic flux at 1 ryd as well as the ionization rate for two relatively well measured luminosity functions : lyman - break galaxies ( lbgs ) at @xmath97 ( steidel et al .
1999a ) and quasars at @xmath165 ( croom et al . 2004 ) .
both of these luminosity functions were derived using samples selected in the rest - frame ultraviolet ( uv ) .
hence , they do not include highly obscured sources , which may provide a substantial contribution to the total rates of star formation and accretion onto black holes in the universe .
however , since such systems would likely also be absent from surveys of quasar absorbers , which typically target quasars that are exceptionally bright in the rest - frame uv , these luminosity functions are exactly what is needed for our purposes . by combining their ground - based data with data from the hubble deep field ( hdf ) , steidel et al .
( 1999a ) find that , down to at least @xmath166 , the @xmath1 luminosity function is well fit by a schechter function with parameters @xmath167 , @xmath168 ( ab magnitude ) , and @xmath169 [ for for @xmath170 .
we changed the units from per magnitude to per @xmath171 by dividing by @xmath172 and we converted to @xmath173 using the effective volumes listed in their table 3 . ]
@xmath174 and @xmath175 .
we note that because they relied on hdf data for @xmath176 , which covers only a relatively small field , the faint - end slope may be subject to systematic errors due to cosmic variance .
@xmath177 at @xmath178 yields @xmath179 at rest - frame @xmath180 .
starburst99 ( version 4.0 , leitherer et al .
1999 ) predicts @xmath181 for a continuous starburst ( the result converges within 10 myr ) with a salpeter initial mass function with upper and lower mass cutoffs of @xmath182 and @xmath183 respectively , and a metallicity @xmath184 [ @xmath185 for @xmath186 .
using this value we obtain @xmath187 .
equation ( [ eq : barf ] ) then gives @xmath188using the same starburst99 model for the spectral shape , we find that this corresponds to a ionization rate of @xmath189where @xmath190 is the average fraction of hydrogen ionizing photons that escapes to the absorber , weighted by the cross - section for ionization . neglecting evolution in the luminosity density
, the values for @xmath165 would be 0.90 times those for @xmath97 .
it is interesting to estimate the typical impact parameter of absorbers around lbgs . assuming the power - law radius - luminosity relation ( eq .
[ [ eq : r ] ] ) , equation ( [ eq : dn / dz_pl ] ) yields , @xmath191^{-1/2 } , \label{eq : rstar}\ ] ] where we assumed @xmath192 .
using @xmath126 gives @xmath193 . to put this in context ,
damped ly@xmath0 systems ( dlas , @xmath194 ) have an observed rate of incidence @xmath195 at @xmath1 ( storrie - lombardi & wolfe 2000 ) .
hence , dlas are likely to have impact parameters of @xmath196 kpc ( see also fynbo et al.1999 ; haehnelt , steinmetz , & rauch 2000 ) . comparing this with the half - light ratios of bright ( @xmath197 ) lbgs , which are typically about 2 kpc ( 0.20.3 arcsec ; giavalisco steidel , & macchetto 1996 ) we see that the approximation of point - like sources is likely to be reasonable .
however , we do expect our model to overestimate the local flux for absorbers that are much rarer than dlas .
we can also use equation ( [ eq : rstar ] ) to verify the validity of our assumption that the flux from nearby sources is negligible compared to that from the central source .
since galaxies with @xmath198 dominate the cosmic luminosity density , we would expect this assumption to be a good approximation if @xmath199 if the galaxies were randomly distributed . from equation ( [ eq : rstar ] )
we can see that this would imply @xmath200 .
however , bright lbgs are observed to cluster : their inferred comoving correlation length is @xmath201 for a two - point correlation function of the form @xmath202 ( steidel et al.1999b ) .
hence , the expected number of galaxies within a distance @xmath100 of a @xmath113-galaxy is about @xmath203^{-1.8}$ ] higher than for a random distribution and we require @xmath204 for the flux from nearby galaxies to be negligible .
we will see in [ sec : comparison ] that for @xmath205 this condition is just satisfied if the flux ( at around 1 ryd ) from the central galaxy exceeds the background .
however , given the approximate nature of this estimate , it is possible that we have underestimated the mean local flux by a factor of a few for classes of absorbers for which the flux from the central galaxy is not large compared with the background .
there is one class of absorbers for which the intensity of the local uv radiation has been measured : dlas with detectable absorption by @xmath206 .
wolfe et al . (
2004 ) have measured the uv flux at 1500 ( rest - frame ) for the 23 absorbers in their sample of 45 @xmath2074 dlas for which they were able to obtain a positive detection of @xmath206 absorption . by equating inferred cooling rates to grain photo - electric heating rates ( given the measured dust - to - gas ratios ) , they find that their @xmath206 absorbers are typically exposed to a flux that is within a factor of a few of @xmath208 .
our starburst99 spectral template predicts @xmath209 , which yields @xmath210 for @xmath211 ( @xmath212 , where 0.20 is the rate of incidence for @xmath213 dlas measured by storrie - lombardi & wolfe 2000 ) .
thus , _ for the one class of absorbers for which the local uv flux has been measured , the prediction of our model is in excellent agreement with the observations_. measurements of the local flux based on @xmath214 lines confirm that the uv radiation field in dlas far exceeds the background ( e.g. , hirashita & ferrara 2005 , srianand et al .
2005 ) . the good agreement between our prediction for the flux to which dlas are exposed , which relied on the assumption that dlas arise in the halos of lbgs , and the measurement of wolfe et al .
( 2004 ) suggests that dlas and lbgs may be drawn from the same population of galaxies , as has already been suggested on other grounds by schaye ( 2001b ) and mller et al .
if this is indeed the case , then we can read of the predictions of our model for the impact parameters and luminosities from figure [ fig : meanmed ] .
for @xmath215 and @xmath97 we predict a median impact parameter @xmath216 arcsec ) ( see eq . [ [ eq : rstar ] ] ) and a median luminosity @xmath217 . using the double power - law fit to the luminosity function of quasars in the 2df qso redshift survey ( croom et al .
2004 ) , we find @xmath218 and @xmath219 for @xmath220 and @xmath165 .
assuming @xmath221 , a power - law spectrum @xmath222 , where @xmath223 is the frequency at the hydrogen lyman limit , and a spectral index @xmath224 ( telfer et al .
2002 ) this gives @xmath225 and thus a mean flux @xmath226for a power - law spectrum the ionization rate is @xmath227 .
hence , if there exists a population of absorbers which all reside in halos around qsos , then such absorbers will on average be exposed to a local radiation field with an ionization rate @xmath228 since both @xmath48 and the intensity of the uv background ( uvb ) radiation are proportional to the luminosity density of the sources , the relative contributions of the central galaxies and quasars to @xmath48 are identical to their relative contributions to the uvb . however , although every quasar probably resides in a galaxy , only a few percent of galaxies host an active quasar ( e.g. , steidel et al.2002 ) .
therefore , the mean flux from local quasars may not be very relevant to quasar absorption studies .
studies employing non - parametric statistics ( such as the median and other percentiles ) and studies using small samples of absorbers , are unlikely to be significantly affected by local quasar radiation , unless the absorbing species requires the presence of a hard radiation source for its existence .
it should , however , be noted that regardless of how small the fraction of time that a quasar is active , local radiation from quasars would be important if they are dormant for periods comparable to or smaller than the timescale for the establishment of ionization equilibrium .
although this timescale is typically very short for hydrogen ( @xmath229 ) , it can be much longer for species with higher ionization potentials . in the remainder of this paper we will ignore radiation from local quasars . how does the flux from local sources compare to that from the extragalactic background ? in section [ sec : comp ] we found that , relative to the background , the local flux is given by @xmath230 where the critical rate of incidence can be computed using either equation ( [ eq : dndzcrit1 ] ) or ( [ eq : dndzcrit2 ] ) .
the former method requires a measurement of the luminosity density , the background flux , and the escape fraction to the absorber ( @xmath15 ) , while the latter requires measurements of the mean free path and the ratio @xmath231 , and also requires one to assume that the sources associated with the absorbers dominate the background . as the absorber column density tends to the column in front of the sources , @xmath15 asymptotes to unity , while for much lower absorber column densities it asymptotes to the global escape fraction @xmath80 . hence , provided the mean free path is known and that the sources associated with the absorbers dominate the background , the second method is probably more reliable for all but the highest column densities .
if the luminosity density , the background , and the mean free path are all known , then we can measure the global escape fraction @xmath80 by comparing the two methods .
we will now apply both methods to hydrogen ionizing photons . for other ions
the calculation is entirely analogous to that for , except that a model of the spectral shape of the uvb is required because the rates of photo - ionization of heavier elements by the background have not been measured . for ions with ionization potentials @xmath232
we expect the critical rates of incidence to be similar to those of , but for ions with @xmath233 , such as , the critical rates should be much smaller because stars ( with the exception of pop .
iii ) emit very little radiation at these energies .
note , however , that even though local stellar radiation is unlikely to be important above 4 ryd , ionization models of absorbers ( as well as other ions with ionization potentials above 4 ryd ) generally also use constraints from other species whose ionization balance is affected by the addition of radiation below 4 ryd .
first , we will compute the critical rate of incidence using method 1 ( i.e. , equation [ eq : dndzcrit1 ] ) . for the intensity of the uv background we use @xmath234 at @xmath235 .
these measurements were taken from , @xmath236 ; spergel et al .
2003 ) using the relation @xmath237 ( e.g. , rauch et al .
dav & tripp ( 2001 ) ( @xmath119 ) , scott et al .
( 2002 ) ( @xmath238 ) , and bolton et al .
( 2005 ) ( @xmath2394 ) . for redshift 3
we have already computed the mean ionization rate due to local sources ( equation [ eq : gamma_lbg ] ) .
we can scale this result to other redshifts if we assume that the luminosity density of ionizing radiation is proportional to the cosmological star formation density , for which we take @xmath240 and @xmath241 ( e.g. , heavens et al . 2004 and references therein ) .
we then find that @xmath242 second , we compute the critical rate of incidence using method 2 , which reduces to equation ( [ eq : dndzcrit3 ] ) for , if we neglect the variation of the mean free path with frequency and if we assume that the column density distribution is of the form @xmath243 around the lyman limit . making this assumption and using @xmath244 for @xmath245 ( proux et al .
2005 ) , we obtain @xmath246 we stress that the ratio of @xmath247-factors could be greater than unity for column densities similar to those shielding the sources . comparing equations ( [ eq : resulting_dndz1 ] ) and ( [ eq : resulting_dndz2 ] ) , we find @xmath248 these measurements of the global escape fractions result from the demand that the luminosity density due to lbgs gives rise to the measured uv background . because the measurements of the luminosity density , the mean free path , and the intensity of the uv background are all uncertain by a factor of a few or more , these measurements should be taken as order of magnitude estimates only .
thus , we conclude that a global escape fraction @xmath249 is consistent with all the measurements .
if quasars dominate the uvb then the escape fraction should be smaller .
steidel , pettini , & adelberger ( 2001 ) measured @xmath250 from their composite lbg spectrum ( after correction for intervening absorption ) whereas our spectral template predicts an intrinsic value of @xmath251 , implying that @xmath252 .
although this may be consistent with equation ( [ eq : fesc ] ) within the errors ( which are very poorly constrained ) there is some tension between the two .
however , since the galaxies studied by steidel et al .
( 2001 ) were unusually blue , it is important to keep in mind that their measurements may not be representative for the population as a whole .
if @xmath253 , then local ionizing radiation becomes important for for ll systems .
absorbers with @xmath254 have rates of incidence much lower than ll systems ( e.g. , proux et al .
2005 ) and are thus dominated by local radiation . since estimates of the local uv flux have generally not been available , the implication is that published results from ionization models of such systems may be significantly in error . note that for @xmath255 the absorber columns are probably comparable to those in front of the sources in which case @xmath15 may be much greater than @xmath43 , further boosting the local flux .
comparison of the rates of incidence of metal line absorbers with the critical rates indicates that uv radiation from local galaxies may not be negligible for strong metal systems , such as those containing detectable absorption by ions in low ionization stages .
for example , using the data from boksenberg et al .
( 2003 ) , who decomposed a large number of high - quality quasar absorption spectra into voigt profiles , we find3.5 .
we only used data with observed wavelength redwards of the quasar s
ly@xmath0 emission line and with a redshift more than @xmath256 bluewards of the redshift of the quasar . ]
@xmath257 for @xmath258@xmath259 and @xmath260 for @xmath258@xmath261 , both at @xmath1 .
these observed rates of incidence are likely to be overestimates of the true rates since many of the quasars in the sample of boksenberg et al.(2003 ) were originally selected to contain dlas .
indeed , more than half of their systems with @xmath262 and nearly half of their systems with @xmath263 are associated with dlas .
other frequently studied metal line systems with low rates of incidence include systems ( @xmath264 for systems with an equivalent width @xmath265 at @xmath266 ; steidel & sargent 1992 ) .
it is difficult to predict exactly how the physical properties inferred from ionization models would change if local radiation had been taken into account , because the effects will depend on the ions involved as well as on the rate of incidence of the absorbers .
however , if the column density of is used as a constraint , then the addition of local radiation will generally tend to increase the inferred density and to decrease the inferred size of the absorber .
the effects on the abundances of heavy elements are more difficult to predict and will likely be different for different elements .
we have constructed an analytic model to estimate the characteristic flux from local sources of radiation to which quasar absorption systems are exposed . since
many studies have shown that fluctuations in the uv background are unimportant for the low - column density ly@xmath0 forest ( at least at @xmath267 ) , we focused on the rarer systems with higher columns and/or detectable absorption by heavy elements .
the most important assumptions we made are that the absorption arises in a roughly spherical gas cloud centered on a source of ionizing radiation , that radiation from other nearby sources is negligible , and that the probability that a sightline intersects a total column density @xmath268 is zero beyond some radius @xmath35 ( which may depend on the luminosity @xmath9 and @xmath20 )
. the first assumption will break down for the low column density ly@xmath0 forest , which is thought to arise in a web of sheet - like and filamentary structures , but is likely to be reasonable for the rarer systems with higher columns and/or detectable absorption by heavy elements .
the second assumption is conservative .
we argued it is likely to be a reasonable approximation for those absorbers for which we predict the flux from the central galaxy to exceed the background ( around 1 ryd ) .
the third assumption implies that higher columns arise in sightlines with smaller impact parameters , which we argued must typically be the case .
we showed that the mean flux from the central galaxy is proportional to the luminosity density of the sources and inversely proportional to the rate of incidence @xmath19 of the absorbers , but that it does _ not _ depend on the function @xmath13 . assuming a power - law dependence @xmath269 , we derived analytical expressions for the cross - section weighted moments of the flux , of the impact parameter , and of the luminosity .
in addition , we computed the corresponding medians numerically .
we found that for reasonable values of @xmath104 , the distribution of fluxes is narrow and the mean ( which is independent of @xmath104 ) is of the same order of magnitude as the median .
for the special case of @xmath136 all absorbers belonging to a class with a given rate of incidence are exposed to the same flux .
interestingly , both observations and naive theoretical arguments suggest that @xmath104 is close to this value .
this implies that there exists a characteristic flux to which absorbers with @xmath270 are exposed , which can be estimated using equation ( [ eq : barf ] ) .
we applied our model to two relatively well - studied sources of ionizing radiation : lyman - break galaxies ( lbgs ) at @xmath1 and quasars at @xmath271 .
we argued that galaxies are more relevant to quasar absorption systems because quasars are thought to be dormant for most of their lifetimes .
our predictions are in excellent agreement with the observations for the one class of absorbers for which the flux has been measured : dlas at @xmath1 with detectable absorption by @xmath206 , since our calculation assumed that the absorbers are centered on lbgs , this agreement suggests that dlas and lbgs may be drawn from the same underlying population of galaxies .
we predicted ( for @xmath215 ) that dlas at @xmath1 typically have impact parameters of order 10 kpc and that the median luminosity of their lbg counterparts is about @xmath272 .
predictions for classes of absorbers with other rates of incidence can easily be obtained from the expressions derived in [ sec : method ] and [ sec : powerlaw ] .
we used two different methods to compare the mean , local flux to the background .
the first takes the luminosity density and the escape fraction to the absorber as inputs , and requires knowledge of the background intensity .
the second takes the mean free path for photons and the ratio of the escape fraction to the absorber and the global escape fraction as inputs , and relies on the assumption that the sources associated with the absorbers dominate the background .
we found that consistency between the two methods , which comes down to the requirement that the observed sources make up the observed background , requires a global escape fraction of order 10 percent . for absorbers as rare or rarer than lyman limit systems ( which have @xmath273 at @xmath97 ) ,
local h ionizing radiation becomes important .
( sub-)dla systems are likely dominated by local radiation .
studies that have modeled the ionization balance of such systems assuming that they are exposed to the background only , may therefore have produced spurious results .
it is a pleasure to thank jordi miralda - escud for enlightening discussions which greatly improved this work and anthony aguirre for a careful reading of the manuscript .
i also thank the institute for computational cosmology in durham and the institute of astronomy in cambridge , where part of this work was completed .
finally , i gratefully acknowledge support from the w. m. keck foundation , from nsf grant phy-0070928 , and from marie curie excellence grant mext - ct-2004 - 014112 . | a generic assumption of ionization models of quasar absorption systems is that radiation from local sources is negligible compared with the cosmological background .
we test this assumption and find that it is unlikely to hold for absorbers as rare as lyman limit systems . assuming that the absorption systems are gas clouds centered on sources of radiation , we derive analytic estimates for the cross - section weighted moments of the flux seen by the absorbers , of the impact parameter , and of the luminosity of the central source .
in addition , we compute the corresponding medians numerically . for the one class of absorbers for which the flux has been measured : damped ly@xmath0 systems at @xmath1 , our prediction is in excellent agreement with the observations if we assume that the absorption arises in clouds centered on lyman - break galaxies . finally , we show that if lyman - break galaxies dominate the uv background at redshift 3 , then consistency between observations of the uv background , the uv luminosity density from galaxies , and the number density of lyman limit systems requires escape fractions of order 10 percent . |
with the development of laser cooling and evaporative cooling techniques , it is easy to obtain ultracold atomic clouds at the temperature order of 1@xmath0k @xcite .
this kind of ultralow temperature atomic cloud is suitable as a medium in atom optics studies meystre , pr240 - 143 .
atom - optical lens is one of the fundamental atom - optical elements , which can focus , collimate , image and transmit the atom beam .
therefore it has been implemented in many ways , such as , atom lithography , atom interferometry and atom interference . up to now
two kinds of atomic lens , based on magnetic or far - detuned optical fields , have been developed .
focusing has been achieved through magnetic lenses prl67 - 2439 , prl87 - 030401 , pra65 - 031601,apb2007 .
atom - optical lens based on magnetic fields are advantageous for coherent atom - optic research owing to their extremely high optical quality .
the focusing dynamics have also been investigated with magnetic atomic lens ; specifically , the isotropic three - dimensional ( 3d ) focusing of atoms with a single - impulse magnetic lens @xcite .
however , it is difficult to build flexible optical systems because magnetic atom - optical elements have a large scale .
laser beams are often used to build atom - optical lens because of their interaction with atoms @xcite .
they have a small scale and is flexible to realize the combination of atomic lenses .
for example , atom - optical lenses can be achieved through red - detuned gaussian beams or blue - detuned doughnut beams @xcite .
there are other ways to achieve an atom - optical lens , such as by the use of radiation - pressure force @xcite , near - field light joa8 - 153 , pra77 - 013601 , and far - detuned and resonant standing wave fields @xcite .
a focused gaussian laser beam with red detuning also can be used as an atom - optical lens .
early experiments demonstrated this kind of atomic beams focusing using the dipole force @xcite .
the focusing dynamics of a trapped bec interacting with laser pulses have also been investigated @xcite .
aberration - free atom - optical lenses based on the optical dipole force have also been proposed @xcite .
heating and cooling effects caused by an adiabatic compression and expansion when the cold atomic cloud is close to or far from the focus of the gaussian beam , respectively , have likewise been found @xcite . in previous work
we have numerically and experimentally studied the ultracold atomic cloud and bec wave packet passing through focused red - detuned gaussian laser beam@xcite .
the focusing , advancement and one - dimensional collimation effects have been observed , which agreed with the numerical simulation results .
specially , the one - dimensional collimation effect of wave cloud is interesting because it is very useful in many important aspects required low divergence sources , such as atom lithography , atom interferometry , atomic fountain cloud , atomic physics collision experiments , ultra high resolution optical spectrum , quantum frequency standard and atom interference .
one method of producing such highly collimated beams is using one - dimensional cooling technologyap1984 .
another one - dimensional collimation method is also realized when atomic beam through atomic lens based on magnetic field@xcite .
our method , however , has the advantage of small scale and flexible adjustment of the parameters of the atomic lens . in this work the object - image relation and the collimation of the atom - optical lens created by the far red - detuned gaussian beams are studied theoretically and experimental .
this work is helpful to further understand the properties of atom - optical lens and one - dimensional collimation effect of atomic cloud without cooling .
the paper is organized as follows . in sec .
ii we first investigate the optical properties , such as object - image relation , focal length of atomic lens induced by far red - detuned gaussian laser beam using particle tracing method when the atom is with uniform velocity and under gravity field , respectively . in sec .
iii the collimation effect is analytically studied , and then the numerical simulation is used to verify the valid of the analytical results . in sec .
iv the experimentally studying of the focusing , imaging and collimation effect is also presented and discussed .
finally we conclude the work .
in figure 1 , we consider a cold atomic cloud with initial temperature @xmath1 located at the position @xmath2 .
when the atomic cloud with initial translational velocity @xmath3 propagates along @xmath4 axes to a far red - detuned focused gaussian laser beam , the atomic cloud undergoes enough free expansion that the position of every atoms in the cloud is correlated with its velocity distribution .
the potential induced by the laser beam has the following form : @xmath5where @xmath6 . here
@xmath7 is determined by the intensity in the center of the gaussian beam , and @xmath8 is the waist width of the gaussian beam .
@xmath9 presents the detuning between the laser frequency and the transition frequency of the atom , and @xmath10 represents the decay rate of the atomic excited state owing to spontaneous radiation . when the detuning is red , the potential is negative and presents an attractive force when the atoms passing through it .
the red - detuned gaussian laser beam can therefore serve as a focal cylindrical atomic lens .
now we will investigate the optical properties of the atomic lens , such as object - image relation , focal length and aberration , by solving the motion equation of atoms : @xmath11 where @xmath12 is the atom mass . because nonlinear coupling eq .
( [ r ] ) is difficult to solve , without loss of physics , we assume that the incident kinetic energy @xmath13 , so the velocity along @xmath4 direction is almost not changed in the process , i.e. , @xmath14 .
now , the potential can be expressed as @xmath15 substituting eq .
( [ v1 ] ) into equation ( [ r ] ) one obtains [ c - e ] @xmath16where @xmath17 eq.([c - e ] ) shows that the motion of atoms in different directions is decoupled .
the motion of atoms in central attractive force now reduces to the motion of the atom under the time dependent force along @xmath18 direction . owing to
no interaction happened along @xmath19 direction in our model , hereafter transverse direction only means @xmath18 direction . from fig .
[ focusscheme1 ] one observes that the deflection angle of atom passing through laser beam relates with the change of velocity of the atom along @xmath18 direction .
the momentum change along @xmath18 direction can be found with the law of impulse - momentum @xmath20 we assume the shift of the atom in the effective interaction range @xmath21 ( see fig .
[ focusscheme1 ] ) along @xmath18 direction is almost not changed , so the integral of eq .
( [ my ] ) gives @xmath22 where @xmath23 is distance of the atom trajectory from the @xmath4 axial .
considering @xmath24 , the deflection angle @xmath25 , and @xmath26 under paraxial approximation , i.e. , @xmath27 , the incident and exit angles , respectively , are [ thetaoi ] @xmath28and eq .
( [ theta0 ] ) , keeping to first order , can be rewritten as @xmath29 according to @xmath30 , one gets @xmath31 if we define the focal length as @xmath32 eq .
( [ deflectionangle ] ) becomes @xmath33 which is just the gaussian formula of thin - lens in light - ray optics .
we refer eq .
( [ g - f ] ) as gaussian formula of atomic optics . in this part we will consider an atom released from the position @xmath34 with a vanishing initial vertical velocity and then falls through the focused gaussian laser beam under the gravity . according to the law of conservation of energy , at the height equal to the center of laser beam , the vertical kinetic energy of the atom is@xmath35 in gravity field case , therefore
, the focal length of the atomic lens is written as @xmath36 assuming the atom reaches the maximal transverse distance @xmath23 after time @xmath37 and then flies back and crosses the @xmath4 axial again after time @xmath38 , the following relation holds @xmath39 where @xmath40 and @xmath41 are the initial and transverse velocities after the atom passing through the laser beam , respectively . likewise , keeping to first order , eq .
( [ deltavy ] ) can be rewritten as @xmath42 combining eqs .
( [ t ] ) and ( [ vyg0 ] ) one finds the time @xmath38 expressed as @xmath43 the falling distance after the atom flies back and reaches the @xmath4-axial again is @xmath44 defining effective image distance @xmath45 and effective object distance @xmath46 , eq .
( [ hi ] ) can be rewritten as @xmath47 which is similar with eq .
( [ g - f ] ) and called the object - image relation of atomic lens in the gravity . obviously , in the gravity case , the object and image distance is not the real falling distance .
if one wants to experimentally observe the minimal focused spot of the atomic cloud by absorption image technology , the probe laser beam should be located at @xmath48 and record the signal after time @xmath49 .
collimation of atomic beam is very useful for nanolithography , atom interferometry and gravity - gradiometer , and cold collision studies . now
the well collimated atomic beam can be obtained by nozzles , transverse cooling , atomic lens . in this section , however , we theoretically study the collimation effect of the atomic cloud passing through the gaussian beam . to begin with ,
we investigate the collimation effect when the atomic cloud with the initial temperature @xmath50has a uniform initial center velocity without gravity .
this is a good approximation when the initial center velocity of the atomic cloud is large enough .
different from the case of transverse cooling , atomic lens can not only collimate , but focus the atomic cloud , so in this paper we use the velocity spread of the atomic cloud to measure the collimation instead of the angular divergence . ) and ( [ tz ] ) .
the symbol @xmath51 and @xmath52 represent the corresponding simulation results . in the simulation
the initial radial and velocity of the atomic cloud respectively are @xmath53 m , and @xmath54ms@xmath55 .
the parameters for the optical potential respectively are @xmath56j , @xmath57m.,width=307 ] based on the geometric relation in fig . 1 , one has @xmath58 combining eqs .
( [ deltavy ] ) and ( [ hlo ] ) , the velocity @xmath59 of the atom deflected by the laser beam along @xmath18 direction is given by the following expression : @xmath60 , \label{vy}\ ] ] where @xmath61 .
then we find the root - mean - square ( rms ) velocity along @xmath18 direction of the atomic cloud passing through the laser beam @xmath62 ^{2 } \notag \\ & \times \exp \left ( -\frac{mv_{y_{0}}^{2}}{2k_{b}t_{0}}\right ) \,\mathrm{d}% v_{y_{0 } } \notag \\ & = \left ( \frac{1}{\left ( 2\alpha + 1\right ) ^{3/2}}\frac{l_{o}^{2}}{f^{2}}-2% \frac{1}{\left ( \alpha + 1\right ) ^{3/2}}\frac{l_{o}}{f}+1\right ) \left ( v_{0}^{rms}\right ) ^{2 } , \label{ty}\end{aligned}\ ] ] where @xmath63 with boltzmann constant @xmath64 , @xmath65 with @xmath66 the size of the atomic cloud when it reaches the laser beam center , and @xmath67 the initial rms velocity of the atomic cloud .
one can see that the parameter @xmath68 can denotes the degree of paraxial approximation .
4 shows the transverse rms velocity ( thin solid lines ) as function of the object distance .
we find that the transverse rms velocity first decreases , then increases and finally decreases and asymptotically tends to initial one with increasing the object distance .
obviously , there exists a region in which the transverse rms velocity is smaller than the initial one .
this phenomenon can be interpreted that , when the object distance is very short , the size of the atomic cloud @xmath69 is far smaller than the waist of laser beam @xmath8 , i.e. , @xmath70 , therefore the paraxial approximation is well satisfied . under this condition
the laser beam can be viewed as an ideal lens , and eq .
( [ ty ] ) is reduced to @xmath71 .
so the velocity spread first decreases and then increases with increasing the object distance .
when the object distance continuously increases , @xmath72 , or the paraxial approximation is not well satisfied , the aberration effect becomes significant .
the deflection angle of hot atom is less than that of those cool atoms .
the total effect is that , involving the aberration , the transverse rms velocity is smaller than the ideal one . if the object distance continuously increases , the size of the atomic cloud @xmath73 will be much greater than that of the laser beam and less atoms deflected by the atomic lens , and the final transverse rms velocity tends asymptotically to the initial rms velocity .
this also can be seen from eq .
( [ ty ] ) , if @xmath74 , @xmath75 . from fig .
4 one can also observe that the minimum transverse rms velocity is about @xmath76 for @xmath77 while the one is about @xmath78 for @xmath79 .
this tells us that the lower initial rms velocity is helpful to raise the collimation efficiency , which implies that atomic lens is suitable for collimating ultracold atomic cloud rather than hot atomic cloud . to verify the analytic result
, we use the direct simulation monte carlo ( dsmc ) approach @xcite to simulate the process . in the simulation the @xmath80rb atomic cloud with initial temperature @xmath1 and the center velocity @xmath81 along @xmath4 direction is placed at position @xmath82 , after the cloud passing through the laser beam we then measure the transverse velocity spread of the atomic cloud . from fig . 4
one notes that the analytical results agree well with the simulation ones ( represented by the square symbol ) . in the following text
, we will discuss the vertical velocity spread of the atomic cloud . according to the law of conservation of energy , in the dipole conservative potential , the change in the kinetic energy of the atom along @xmath83 direction should be equal to that along @xmath18 direction @xmath84 here
we define the variance of velocity along @xmath4 direction @xmath85 .
( [ delta_ey ] ) becomes @xmath86 \exp \left ( -\frac{% mv_{y_{0}}^{2}}{2k_{b}t_{0}}\right ) . \label{deltavz}\end{aligned}\ ] ] with the previous assumption @xmath87 , eq .
( [ deltavz ] ) can be written as @xmath88 \exp \left ( -\frac{mv_{y_{0}}^{2}}{2k_{b}t_{0}}\right ) .
\label{vz}\ ] ] finally , one can find the vertical rms velocity of the atomic cloud @xmath89 obviously , under the assumption @xmath24 , the vertical rms velocity is almost no change after the atomic cloud passing through the laser beam .
we can explain this phenomenon by the fact that , compared with vertical kinetic energy , the energy transferred from transverse direction to vertical direction is so small that the corresponding vertical velocity and its width is almost not changed .
the simulation result ( shown by diamond symbol in fig .
3 . ) confirms the analytical prediction .
m , respectively .
the dash - dotted line denotes the vertical velocity spread of the atomic cloud .
the symbol @xmath90 , @xmath52 , @xmath91 and @xmath92 are corresponding simulation results . in the simulation
the initial temperatures , radial and velocity of the atomic cloud respectively are @xmath93k , @xmath53 m , the initial height of the atomic cloud is @xmath94 mm .
the parameter for the optical potential respectively are @xmath95ghz.,width=288 ] next we will consider the collimation effect when an atomic cloud falls freely with a vanishing initial center velocity under the gravity . to obtain the transverse rms velocity of the atomic cloud , we should first find the transverse velocity of it . using the same method in previous part to find eq .
( [ vy ] ) one gets @xmath96 , \label{vyg}\ ] ] where @xmath97 .
so the transverse rms velocity of the atomic cloud in the gravity is @xmath98 where @xmath99 and @xmath100 .
( [ tyg ] ) is similar with that in the uniform velocity case . in the gravity case , @xmath101 ,
which is not dependent on the initial height of the atomic cloud but proportional to the strength of the laser beam .
fig . 5 plots @xmath102 against @xmath7 with different waists of the laser beam .
we see that the transverse rms velocity first decreases and then increases with increasing rabi frequency , which is similar with that in the uniform velocity case . additionally , from fig . 4 one can notice that large waist of the laser beam is helpful to obtain better collimation effect .
it is readily understandable that the waist of the laser beam is larger , the ratio of the size of atomic cloud to the width of the laser beam is smaller , and laser beam is more like an ideal lens , the collimating efficiency is therefore improved .
the corresponding simulation results are well consistent with analytical ones .
represents the reference one .
the flying time @xmath103 is 7ms before the atomic cloud entering the laser beam .
the initial temperatures of the atomic clouds are ( a ) 190nk and ( b ) 370nk respectively .
another parameters are @xmath104j , @xmath105.,width=307 ] j , @xmath106 , @xmath107.,width=288 ] our experimental setup consists of two magneto - optical traps pra80 - 033411 . the atomic cloud is firstly captured in the up - mot and then transferred into the second ultrahigh - vacuum mot(uhv - mot ) .
after the atomic number in the uhv - mot is stable , we did optical molasses and then loaded atoms into a quadrupole - ioffe - configuration ( quic ) trap .
evaporative cooling of the atoms was performed by rf - induced spin flips .
we swept the rf frequency from 25mhz to a value of around 1.6mhz over a period of 28s .
atomic clouds with various temperatures from about 1@xmath0k to below the phase transition point were obtained by setting different rf frequencies .
the cold atomic cloud then ballistically expanded after the magnetic trap was switched off .
the direction of propagation of the focused gaussian beam and the probing beam was parallel to the long axis of the quic trap .
therefore , the cold atomic clouds symmetrically distributed in the probing plane before traversing the gaussian beam . we acquired the distribution of atomic clouds from absorption images , as shown in fig .
the rayleigh length was about 8.5 mm .
since the rayleigh length was far greater than the scale of the atomic cloud while it was passing the light beam , we could approximate that the laser beam provided a two - dimensional gaussian potential .
when evaporative cooling was finished , we released the atomic cloud from the quic trap and turned on the gaussian beam .
after the atomic cloud had passed through it , we turned off the gaussian beam and waited for several milliseconds .
all information regarding the atomic cloud were derived from the absorption images .
5 is a typical experimental result about the focusing of the atomic cloud passing through the laser beam . from absorption images and
cross section of optical depth we can see that the atomic clouds are focused by the laser beam compared to the reference one . obviously , the focusing just happened on the transverse direction , while the vertical direction remains unchanged . owing to the finite action length of the laser beam , just part of the atomic cloud is focused
. with increasing the initial temperature , the focused part is increasing .
we fit the cross curves by an asymmetric bi - gaussian to derive the full - width at half - maximum ( fwhm ) of the width of the focused part in the atomic clouds along transverse direction . under the experimental parameters , the imaging time @xmath108ms , the flying time @xmath109ms ,
according to the previous analysis , the fwhm widths of the focused part should be about 8@xmath0 m .
however , the experimental results are both of them are about 13@xmath0 m and greater than the theoretical prediction , which is mainly caused by the aberration of the atom - optical lens @xcite .
imaging is one of important properties of atom - optical lens .
if an atomic cloud with very lower initial temperature passes through the laser beam , according to eq .
( [ t ] ) , the fwhm width of the focused atomic cloud will be @xmath110 obviously , when @xmath111 , the fwhm is inversely proportional to the flying time , which is plotted by solid line in fig .
the experimental result is well consistent with the analytical prediction .
j , @xmath105.,width=288 ] we also experimentally study the collimation effect of atomic lens . the collimation effect can be measured by the final velocity spread of the atomic cloud after it passing through the gaussian laser beam . as discussed in ref .
@xcite , when the initial temperature of the atomic cloud is high , only the part with lower transverse velocity is focused , other part not interacts with the laser beam .
there are focused and unfocused parts in the atomic cloud after it passing through the laser beam . by measuring the rms widths of the focused and unfocused parts in the atomic cloud
, one can determined the total velocity spread and then measure the total collimation effect .
the transverse rms velocities of focused part and unfocused part are @xmath112 and @xmath113 where @xmath114and @xmath114 are fwhm width of focused and unfocused parts in the atomic cloud .
so the total transverse rms velocity of the atomic cloud is @xmath115 where @xmath116 , @xmath117 and @xmath118 are the number of focused , unfocused and total atoms , respectively .
7 shows the experimental result of the total transverse rms velocity with dependence of the initial rms velocity of the atomic cloud . obviously , the total transverse rms velocity is smaller than the initial one , that is to say , the atom - optical lens efficiently collimate cold atomic clouds . with the parameters experimentally available ,
the collimation effect is not very well when the the initial velocity spread ( temperature ) is lower .
we study the optical properties of the atomic lens induced by the far red - detuned laser beam and collimation effect of atomic cloud passing through the atomic lens , with uniform translational velocity and with vanishing initial translational velocity in gravity , respectively . based on the atom deflection in dipole conservative potential , the object - image relation and the focal length of the atomic lens
is theoretically presented .
and then , the collimation effect of the atom - optical lens is theoretically studied . with the transverse rms velocity , we find that collimation effect of the atomic cloud with lower initial rms velocity is better than that with a high initial rms velocity .
furthermore , theoretically speaking , the collimation by atomic lens can reach smaller divergence angle than that by transverse cooling technology due to the collimation of atomic lens without the cooling limit .
therefore the atomic lens is suitable for collimating the ultracold atomic cloud or beam .
moreover , atomic lens created by the laser beam has the advantages of small scale and flexible adjustment
. it will play important role in many fields , such as integrated atom optics , atomic fountain clock .
finally we have experimentally demonstrated focusing , imaging and collimation of atomic cloud passing through the gaussian laser beam , which are well consistent with the theoretical prediction .
this work is supported by the national natural science foundation of china under grant nos .
10828408 , 10588402 , 10804115 and 10974211 , national fundamental research program of china under grant nos . 2011cb921604 and 2011cb921504 | we have demonstrated an atom - optical lens , with the advantage of a small scale and flexible adjustment of the parameters , created by far red - detuned gaussian laser beam perpendicular to the propagation direction of the cold atomic cloud .
this work presents the focusing and collimation effect of a cold atomic cloud , with uniform translational velocity and with vanishing initial translational velocity in gravity respectively , by the atom - optical lens .
the object - image relation , focal length and the collimation effect of the atom - optical lens have been analyzed both theoretically and experimentally . |
it has long been hoped that x - ray binaries and active galactic nuclei would prove to be good laboratories for testing general relativity .
the two most promising lines of attack for disentangling relativistic effects from the physics of accretion disks and of radiative transfer are high resolution spectroscopy of emission lines ( see e.g. reynolds & nowak 2003 for a review ) and studies of high frequency quasi - periodic oscillations ( see e.g. van der klis 2004 for a review ) .
recently , high frequency quasi - periodic oscillations ( hfqpos ) have been found in low mass x - ray binaries for which the black hole masses are reasonably well measured . in one case in particular , a frequency has been identified which is too large to be a keplerian orbit around a schwarzschild black hole ( strohmayer 2001 ) .
furthermore , a pattern has begun emerging where these high frequency qpos are found in pairs with 2:3 ratios of frequencies .
this was first pointed out by abramowicz & kluzniak ( 2001 ) , based on combining observational results from strohmayer ( 2001 ) and remillard et al .
more recent work from miller et al .
( 2001 ) , remillard et al .
( 2003 ) , remillard et al .
( 2004 ) and homan et al .
( 2004 ) has lent more weight to the idea that 2:3 frequency ratios are quite common in these systems .
a variety of theoretical models have been developed to explain these qpos , most requiring a spinning black hole , but often requiring rather different values of the spin ( compare , e.g. abramowicz & kluzniak 2001 ; rezzolla et al .
2003 ; li & narayan 2004 ) .
therefore , there is still much `` astrophysics '' ( i.e. physics of disk structure and stability and physics of radiative transfer ) that must be understood before the fundamental physics can be probed in these systems , but there is strong cause for optimism that these systems really will ultimately tell us something profound about spinning black holes .
a key first step to disentangling the `` astrophysics '' is , of course , to develop models which not only match the important frequencies , but also include radiation mechanisms such that the observed x - rays would actually be modulated at that frequency .
such has been done recently for a particular realization of the case of the parametric resonance model of abramowicz & kluzniak ( 2001 ) , by considering the possibility of hot spots that form at the resonant radii in the accretion disk ( schnittman & bertschinger 2004 ; schnittman 2004 ) .
these authors have found that different sets of model parameters can produce the same fourier power density spectrum with dramatically different qualitative appearances to the light curves . in this letter
, we will show that higher order variability statistics , particularly the bispectrum , can break this degeneracy .
one key way to distinguish between different mechanisms which produce the same power spectra from qualitatively different light curves is to study the non - linearity of the variability .
linear variability is that in which the phases at the different frequencies in the fourier spectrum of a time series are uncorrelated with one another , while time series with non - linear variability show fourier spectra with correlations between the phases at different frequencies .
some particularly useful tools for studying non - linearity are the bispectrum and the closely related bicoherence .
the bispectrum computed from a time series broken into @xmath0 segments is defined as : @xmath1 where @xmath2 is the frequency @xmath3 component of the discrete fourier transform of the @xmath4-th time series ( e.g. mendel 1991 ; fackrell 1996 and references within ) .
it is a complex quantity that measures the strength of the phase coupling of different fourier frequencies in a light curve and has a phase of its own which is the sum of the phases at the two lower frequencies minus the phase at the highest frequency .
its value is unaffected by additive gaussian noise , although its variance will increase for a noisy signal . a related quantity , the bicoherence is the vector magnitude of the bispectrum , normalised to lie between 0 and 1 . defined analogously to the cross - coherence function ( e.g. nowak & vaughan 1996 ) , it is the vector sum of a series of bispectrum measurements divided by the sum of the magnitudes of the individual measurements .
if the biphase ( the phase of the bispectrum ) remains constant over time , then the bicoherence will have a value of unity , while if the phase is random , then the bicoherence will approach zero in the limit of an infinite number of measurements .
mathematically , the bicoherence @xmath5 is defined as : @xmath6 this quantity s value is affected by gaussian noise , but it can be considerably more useful than the bispectrum itself for determining whether two signals are coupled non - linearly . in an astronomical time
series analysis context , it has been previously applied to the broad components in the power spectra of cygnus x-1 and gx 339 - 4 , in both cases finding non - linear variability through the presence of non - zero bicoherences over a wide range of frequencies ( maccarone & coppi 2002 ) . since that work ,
we have become aware of a correction which is , in principle , important for studying aperiodic variability with the bicoherence , namely that the maximum value of the bicoherence is suppressed by smearing of many frequencies into a single bin in the discrete fourier transform .
this suppression can not be calculated in a straightforward way ( see e.g. greb & rusbridge 1988 ) .
however , we also note that since comparisons in maccarone & coppi ( 2002 ) were made only with model calculations made with the same time binning as the real data , these effects , whatever they may be , are the same for the real data and the simulated data , and hence the conclusions of that paper are not affected substantially .
there exist , at the time of this paper s writing , at least four basic concepts for producing the high frequency quasi - periodic oscillations seen from accreting black holes . in historical order , these are diskoseismology ( e.g. okazaki , kato & fukue 1987 ) , relativistic coordinate frequencies ( e.g. stella & vietri 1999 ) , rayleigh - taylor instabilities ( e.g. titarchuk 2002 , 2003 ; li & narayan 2004 ) and oscillating tori ( rezzolla et al .
we will briefly summarize the properties of the models in a different order in an effort to smooth the flow of the paper .
the first model proposed was based on diskoseismology - the excitation of various trapped modes in the inner region of a keplerian accretion disk in a relativistic potential ( e.g. okazaki , kato & fukue 1987 ; nowak et al .
this model seems not to be directly applicable to the data , at least for the cases where small integer ratios of frequencies exist ; it would require considerable fine tuning in the different mass and spin values for the black holes to produce routinely a 2:3 frequency ratio .
chen & taam ( 1995 ) showed that a slim disk ( see e.g. abramowicz et al .
1988 ) can produce oscillations at a frequency very close to the radial epicyclic frequency of the accretion disk , sometimes with power at twice this frequency ; it also appears that there may be some excess power at 3/2 of the epicyclic frequency in one of the simulations presented in chen & taam ( 1995 ) see figure 5 of that paper but it is likely that a longer hydrodynamic simulation would be needed to confirm this . in a similar vein
is the oscillating torus model of rezzolla et al .
this model also applies calculations of the frequencies of @xmath7-modes ( i.e. sound waves ) of zanotti , rezzolla & font ( 2003 ) , but in a geometrically thick , pressure supported torus ( as expected at high accretion rates like those where the hfqpos are seen - de villiers , krolik & hawley 2003 ; kato 2004 ) , rather than in a geometrically thin , keplerian accretion disk . in this case
, the different overtones are found to be approximately in a series of integer ratios , starting from 2 , so the model is compatible with existing data on high frequency qpos in black holes .
the model most recently applied to high frequency qpos from black hole candidates is that of rayleigh - taylor instabilities , although the same basic idea had previously been applied to qpos from accreting neutron stars ( titarchuk 2002 , 2003 ) . in this picture ,
non - axisymmetric structures can grow unstably at the magnetospheric radius ( presumed to exist also for black holes , as their accretion disks can become magnetically dominated ) with frequencies of integer ratios of the angular frequency at that radius , though the lowest mode will be stable for low gas pressures ( li & narayan 2004 ) .
after the first indications that small integer ratios between hfqpo frequencies were likely , it was noted by abramowicz & kluzniak ( 2001 ) that if the relativistic coordinate frequencies determined the frequencies of the quasi - periodic oscillations ( see e.g. stella & vietri 1999 ) then resonances between these different frequencies ( e.g. vertical and radial epicyclic frequencies ) might occur at locations in the accretion disk where these frequencies have small integer ratios .
more recently , schnittman & bertschinger ( 2004 ) performed ray tracing calculations of the light curve of sheared hot spots produced with a 1:3 radial to azimuthal epicyclic frequency ratio and found good agreement with the observed locations of the power spectrum peaks and their relative amplitudes , while bursa et al .
( 2004 ) have shown that it is also possible to produce these frequencies from radial and vertical oscillations of a torus located such that the radial epicyclic frequency is 2/3 the orbital frequency .
schnittman ( 2004 ) extended the work by considering the effects of multiple hot spots under different conditions in order to broaden the _ periodic _ oscillations computed in schnittman & bertschinger ( 2004 ) into the _ quasi_-periodic oscillations that are actually observed . in this paper , we present bicoherence calculations only for this last model for the simple reason that this is the only model for which simulated light curves are currently available .
as simulated light curves for other models become available , we will consider their higher order statistical properties as well .
we now apply the bicoherence to the simulated data . we consider two different model calculations from schnittman ( 2004 ) which give nearly identical power spectra . in each case , the quasi - periodic oscillations are produced by a 1:3 resonance between the radial epicyclic frequency and the orbital frequency .
the parameters have been chosen such that the orbital frequency is 285 hz , and the radial epicyclic frequency is 95 hz .
this corresponds to a black hole mass of 10 @xmath8 and a spin @xmath9 , with the resonance occuring at a radius of 4.89@xmath10 in geometrized units where @xmath11 ; the black hole mass and qpo frequencies compare reasonably well to those observed for xte j 1550 - 564 ( see e.g. miller et al .
2001 ; remillard et al .
the disk inclination is also fixed to be 70 degrees ( where 90 degrees is an edge - on disk ) ; this parameter does not affect the frequencies observed , but can affect the amplitudes of the qpos in the context of the model we are considering here ( schnittman & bertschinger 2004 ) . in each case
we compute 1000 seconds of simulated data with a binning timescale of the lightcurve of 0.1 msec .
we then compute fourier transforms by breaking the data into 2440 segments of 4096 data points , making use of 999.424 seconds of the simulated data .
in the first case , short lived hot spots exist with their orbits all at a single radius , being continually created and destroyed with a characteristic lifetime of 4 orbits . in the second case ,
long - lived ( lifetimes of 100 msec , or about 30 orbits ) hot spots are distributed over a range of radii ( @xmath12 ) . in both cases ,
the hot spots are on orbits with eccentricities of 0.1 . for each model
, the variability appears quasi - periodic , rather than truly periodic , but for different reasons . in the first case
, the creation and destruction of hot spots on short timescales leads to a phase jitter in the light curves . these random offsets in phase broaden the observed periodicity . in the second case
, the power spectrum is truly showing that there are many periodicities in the system , with coherent phases .
the bicoherence easily detects this difference , as can be seen from figure [ sim ] . in case 1
, the bicoherence shows `` circular '' peaks at various combinations of frequencies where there is power at @xmath13 , @xmath14 and @xmath15 in the contour plot , essentially delta function peaks convolved with two - dimensional lorenzians due to the random phase broadening . in case 2 ,
the bicoherence shows thin elongated peaks , oriented in a variety of directions depending on @xmath13 and @xmath14 .
the reason for this difference is straightforward . in the first case ,
all hot spots have the same geodesic frequencies , so during a hot spot s lifetime , it is phase locked to all the other hot spots , giving a collection of delta function peaks at the coordinate frequencies .
the random phase jitter will broaden the @xmath16functions into qpos , with a similar lorentzian width as described in schnittman ( 2004 ) .
the hot spots being created and destroyed in the middle of a fourier transform window will thus create leakage in the power of the qpo to frequencies near the central frequency , but there will be a phase relation between the power in these frequencies and the phase in the central frequency .
the phase jitter should thus provide a broadening in the bicoherence similar to that in the power spectrum .
we note that the peaks do appear to be somewhat elongated , with the direction of elongation such that the sums of the two smaller frequencies equal the centroid frequencies of the highest frequency qpo in the triplet .
this is likely because the centroid is the only truly physical frequency in this case , but a rigorous proof of this point is beyond the scope of this paper .
in the second case , where there are many frequencies in the power spectrum due to hot spots found over a range of radii , there will be phase coherence between the different harmonics of each individual hot spot , but not with the hot spots at slightly different frequencies .
there will thus be bicoherence between the various harmonic frequencies found at any individual radius , but not between frequencies found at different radii
. this second case could be especially interesting .
we have calculated analytically the ratios expected between different harmonics frequencies if the radius at which the hot spot occurs is allowed to vary , and have plotted them in figure [ analytical ] .
if in real data , similar tracks are seen , then , in the context of this model , they would give the relationships between the different relativistic frequencies .
since these tracks trace the coordinate frequencies as a function of radial distance from the black hole , they may be used to make precise measurements of the black hole s mass and spin , plus the central radius of the perturbations . since the observed qpos are rather narrow , so this method would probably be of use only with high signal - to - noise data .
to consider whether this observational test is really feasible , we have performed simulations with the rms amplitude of the oscillations reduced to realistic levels and with poisson noise added .
we consider two count rate regimes - one similar to that detected by _
rxte _ for the typical x - ray transients at about 10 kpc , which is about 10,000 counts per second , and another which would be expected from the same source , but with a 30 m@xmath17 detector . in each case
, we allow 6% of the counts to come from the variable component and to have , intrinsically , count rates given by the simulated light curves of schnittman ( 2004 ) , and the remaining 94% of the counts to come from a constant component .
we then simulate observed numbers of counts in 100 microsecond segments as poisson deviates ( press et al .
1992 ) of the model count rates .
we then compute the bicoherence as above , but with 2440 segments of 4096 data points , for a total of 999.424 seconds of simulated data .
for the _ rxte _ count rates , we find that the bicoherence plots show only noise and only the strongest peak in the power spectrum is clearly significant in a 1000 second simulated observation , while marginal detections exist for the qpos at two - thirds of and twice this frequency ( @xmath18 and @xmath19 ) .
this is as expected based on real data , which generally requires exposure times much longer than 1000 seconds to detect these qpos .
longer simulated light curves have not been computed at this time due to computational constraints . however , since the signal - to - noise in the bicoherence should be substantially worse than the signal - to - noise in the power spectrum , bicoherence measurements should be possible only when a peak in the power spectrum is considerably stronger than the poisson level . for the count rates expected from a 30 m@xmath17 detector , we find that even within 1000 seconds , several of the higher ( i.e. @xmath20 ) harmonics are observable in the power spectrum and show the clear elongation in the bicoherence plot for case 2 , indicating that proposed missions should be capable of making use of the bicoherence for studying hfqpos .
a few very weak peaks are seen in the bicoherence in case 1 even in 1000 seconds .
the simulated bicoherences for a 30 @xmath21 detector are plotted in figure [ xeus ] .
we note that these simulations are a bit over - simplified , in that we have not included the lower frequency qpos and low frequency band - limited noise that are typically observed in conjunction with the hfqpos , but that these variability components should not significantly affect the phase coupling of the high frequency qpos .
we also note that it might be possible to make use of the bicoherence even with _ rxte _ if a more nearby x - ray transient goes into outburst , but that in such a case , the deadtime effects we have neglected here might become important .
we have shown here that the bicoherence can be useful in breaking the degeneracies between different qpo models which produce the same fourier power spectrum . in particular , we have shown that in the context of a resonance model for the high frequency quasi - periodic oscillations seen from accreting black holes , the bispectrum is capable of distinguishing between broadening due to phase jitter caused by the rapid creation and destruction of hot spots and broadening due to an intrinsic distribution of physical frequencies in a broad range around a central value . in future work
, we will examine the properties of the bicoherence for other models for these qpos , such as oscillations in a pressure supported torus ( rezzolla et al . 2003 ) .
we note though that this diagnostic is likely to be useful only with new instrumentation ( or the fortuitous outburst of a bright x - ray transient within about 3 kpc of the sun ) . at the typical 10 kpc distances of x - ray transients
, @xmath22 is capable of detecting these qpos generally only with rather long intergrations and careful selection of the photon energy bands to maximize the signal - to - noise , but proposed timing missions with considerably larger collections areas , like xtra ( e.g. strder et al .
2004 ) and the relativistic astrophysics explorer ( e.g. kaaret 2002 ) should have considerably greater potential for making use of these statistics .
tm wishes to thank marek abramowicz , wlodek kluzniak , shin yoshida and olindo zanotti for stimulating discussions re - motivating his interest in x - ray variability ; phil uttley and simon vaughan for discussions of statistical properties of the bispectrum ; mariano mendez and marc klein wolt for discussions of observational properties of hfqpos ; ron elsner for providing some useful background information on the bispectrum ; and luciano rezzolla and michiel van der klis for comments on the manuscript as well as additional useful discussions .
jds would like to thank edmund bertschinger and ron remillard for many useful discussions , and would like to acknowledge support from nasa grant nag5 - 13306 .
abramowicz , m.a . ,
czerny , b. , lasota , j.p .
& szuszkiewicz , e. , 1988 , apj , 332 , 646 abramowicz , m.a .
& kluzniak , w. , 2001 , a&a , 374l , 19 bursa , m. , abramowicz , m.a . , karas , v. & kluzniak , w. , 2004 , astro - ph/0406586 chen , x. & taam , r.e .
, 1995 , apj , 441 , 354 de villiers , j.p . ,
krolik , j.h . & hawley , j.f . , 2003 ,
apj , 599 , 1238 fackrell , j 1996 , ph.d .
thesis , university of edinburgh greb , u. & rusbridge , m.g .
, 1988 , plasma physics and controlled fusion , 30 , 537 homan , j. , miller , j. m. , wijnands , r. , van der klis , m. , belloni , t.,steeghs , d.,lewin , w. h. g. , 2004 , apj , submitted ( astro - ph/0406334 ) kaaret , p. , 2002 , cospar , 1240 kato , y. , 2004 , pasj , in press li , l .- x .
& narayan , r. , 2004 , apjl , 601 , 414 maccarone , t.j .
& coppi , p.s . , 2002 , mnras , 336 , 817 mendel , j. 1991 , proc ieee , 79 , 278 miller , j.m . , wijnands , r. , homan , j. , belloni , t. , pooley , d. , corbel , s. , kouveliotou , c. , van der klis , m. & lewin , w.h.g . , 2001 , apj , 563 , 928 nowak , m.a
& vaughan , b.a . , 1996 ,
mnras , 280 , 227 nowak , m.a . ,
wagoner , r.v . ,
begelman , m.c .
& lehr , d.e . , 1997 ,
apjl , 477 , 91 okazaki , a.t . , kato , s. & fukue , j. , 1987 , pasj , 39 , 457 press , w.h . ,
flannery , b.p .
, teukolsky , s.a . & vetterling , w.t . , 1992 , _ numerical recipes in fortran _ , cambridge university press remillard , r.a . ,
morgan , e.h .
, mcclintock , j.e . ,
bailyn , c.d .
& orosz , j.a . , 1999 , apj , 522 , 397 remillard , r.a .
, et al . , 2002 , apj , 580 , 1030 remillard , r.a . ,
muno , m.p . ,
mcclintock , j.e .
& orosz , j.a . , 2003 , head , 35 , 3003 remillard , r.a . ,
mcclintock , j.e . ,
orosz , j.a . & levine , a.m. , 2004 , apj , submitted ( astro - ph/0407025 ) reynolds , c.s .
& nowak , m.a . , 2003 , phr , 377 , 389 rezzolla , l. , yoshida , si . , maccarone , t.j . ,
zannotti , o. , 2003 , mnras , 344l , 37 schnittman , j.d . & bertschinger , e. , 2004 , apj , 606 , 1098 schnittman , j.d . , 2004 ,
apj , submitted ( astro - ph/0407179 ) stella , l. & vietri , m. , 1999 , phrvl , 82 , 17 strohmayer , t.e . , 2001 , apjl , 552 , 49 strder , l. , barret , d. , fiorini , c . , kendziorra , e. & lechner , p. , 2004 ,
spie , 5165 , 19 titarchuk , l. , 2002 , apjl , 578 , 71 titarchuk , l. , 2003 , apj , 591 , 354 zanotti , o. , rezzolla , l. & font , j.a . , 2003 ,
mnras , 341 , 832 | we discuss the use of the bicoherence - a measure of the phase coupling of oscillations at different frequencies - as a diagnostic between different models for high frequency quasi - periodic oscillations from galactic black hole candidates .
we show that this statistic is capable of finding qualitative distinctions between different hot spot models which produce nearly identical fourier power density spectra .
finally , we show that proposed new timing missions should detect enough counts to make real use of this statistic .
[ firstpage ] epsf methods : data analysis , methods : statistical , binaries : close , black hole physics , x - rays : binaries , stars : oscillations |
quantum entanglement@xcite is the most important notion in quantum technology ( qt ) and quantum information theory ( qit ) .
as shown for last two decades it plays a crucial role in quantum teleportation@xcite , superdense coding@xcite , quantum cloning@xcite , and quantum cryptography@xcite .
it is also quantum entanglement , which makes the quantum computer outperform the classical one@xcite .
thus , in order to develop qt and qit it is essential to understand how to quantify and how to characterize the multipartite entanglement .
since the quantum entanglement is a non - local property of given multipartite quantum state , it should be invariant under the local unitary ( lu ) transformations , i.e. the unitary operations acted independently on each of the subsystems .
if @xmath2 and @xmath3 are in the same category in lu , one state can be obtained with certainty from the other one by means of local operations assisted classical communication ( locc)@xcite .
this implies that @xmath2 and @xmath3 can be used , respectively , to implement the same task of qit with equal probability of successful performance of the task .
however , the classification of entanglement through lu generates infinite equivalence classes even in the simplest bipartite systems . in order to escape this difficulty the classification through stochastic local operations and classical communication ( slocc )
was suggested in ref.@xcite .
if @xmath2 and @xmath3 are in the same slocc class , one state can be converted into the other state with nonzero probability by means of locc .
this fact implies that @xmath2 and @xmath3 can be used , respectively , to implement the same task of qit although the probability of success for this task is different .
mathematically , if two @xmath4-party states @xmath2 and @xmath3 are in the same slocc class , they are related to each other by @xmath5 with @xmath6 being arbitrary invertible local operators .
however , it is more useful to restrict ourselves to slocc transformation where all @xmath6 belong to sl(@xmath7 , @xmath8 ) , the group of @xmath9 complex matrices having determinant equal to @xmath10 .
the slocc classification was first examined in the three - qubit pure - state system@xcite .
it was shown that the whole system consists of six inequivalent slocc classes , i.e. , fully separable ( s ) , three bi - separable ( b ) , w , and greenberger - horne - zeilinger ( ghz ) classes .
moreover , it is possible to know which class an arbitrary state @xmath2 belongs by computing the residual entanglement@xcite and concurrences@xcite for its partially reduced states .
similarly , the entanglement of whole three - qubit mixed states consists of s , b , w , and ghz types@xcite .
it was shown that these classes satisfy a linear hierarchy s @xmath11 b @xmath11 w @xmath11 ghz .
generally , a given qit task requires a particular type of entanglement .
in addition , the effect of environment generally converts the pure state prepared for the qit task into the mixed state .
therefore , it is important to distinguish the entanglement of mixtures to perform the qit task successfully .
however , it is notoriously difficult problem to know which type of entanglement is contained in the given multipartite mixed state . even for three - qubit state
it is very difficult problem because analytical computation of the residual entanglement for arbitrary mixed states is generally impossible so far .
recently , classification of the entanglement classes for three - qubit mixed states has been significantly progressed . in ref.@xcite
the ghz symmetry was examined in three - qubit system .
this is a symmetry that ghz states @xmath12 have up to the global phase and is expressed as a symmetry under ( i ) qubit permutations , ( ii ) simultaneous flips , ( iii ) qubit rotations about the @xmath13-axis .
the whole ghz - symmetric states can be parametrized by two real parameters , _ say _ @xmath14 and @xmath15 .
authors in ref .
@xcite succeeded in classifying the entanglement of the ghz - symmetric states completely .
this complete classification makes it possible to compute the three - tangle analytically for the whole ghz - symmetric states@xcite and to construct the class - specific optimal witnesses@xcite .
it also makes it possible to obtain lower bound of three - tangle for arbitrary @xmath16-qubit mixed state@xcite .
more recently , the slocc classification of the extended ghz - symmetric states was discussed@xcite .
extended ghz symmetry is the ghz symmetry without qubit permutation symmetry .
it is larger symmetry group than usual ghz symmetry group , and is parametrized by four real parameters .
the purpose of this paper is to extend the analysis of ref.@xcite to four - qubit system .
four - qubit ghz states is an unique maximally entangled @xmath16-qubit state up to lu , @xmath17 is not unique maximally entangled state . in @xmath18-qubit system
there are two more additional maximally entangled states @xmath19 and @xmath20@xcite . ] ( or @xmath18-cat states@xcite , in honor of schrdinger s cat ) are defined as @xmath21 like a @xmath16-qubit ghz symmetry we define a @xmath18-qubit ghz symmetry as a symmetry which @xmath22 have up to the global phase .
straightforward generalization , which is ( i ) qubit permutations , ( ii ) simultaneous flips ( i.e. , application of @xmath23 ) , ( iii ) qubit rotations about the @xmath13-axis of the form @xmath24 is obviously a symmetry of @xmath22 .
thus , we will call this symmetry as @xmath18-qubit ghz symmetry . as will be shown later the @xmath18-qubit ghz - symmetric states are represented by three real parameters while @xmath16-qubit states contain only two .
thus , it is more difficult to analyze the entanglement classification in @xmath18-qubit ghz - symmetric case than that in @xmath16-qubit case . furthermore , if number of qubit increases , we need real parameters more and more to represent the ghz - symmetric states .
therefore , classification of the entanglement for the ghz - symmetric states becomes a formidable task in the higher - qubit system . in this reason it is advisable to restrict the ghz - symmetry to reduce the number of real parameters
this can be achieved by modifying ( ii ) into ( ii ) simultaneous and any pair flips without changing ( i ) and ( iii ) . in four qubit - system
this modification can be stated as an invariance under the application of @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 , @xmath30 , and @xmath23 . the simplest pure state which has the modified symmetry ( ii ) is @xmath31 it is easy to show that @xmath32 is symmetric under the flips of @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , or @xmath39 parties .
obviously , @xmath22 do not have this modified symmetry .
of course , the states , which have this modified symmetry , are also ghz - symmetric .
therefore , we call this modified symmetry as restricted ghz ( rghz ) symmetry given in eq .
( [ modified-2 ] ) is not rghz - symmetric because it is not symmetric under the qubit rotation about the @xmath13-axis although it is symmetric under the modified ( ii ) .
in fact , there is no pure rghz - symmetric state . ] .
as will be shown , the rghz - symmetric states are represented by two real parameters like the @xmath16-qubit case .
this paper is organized as follows . in section
ii the general forms of the ghz- and rghz - symmetric states are derived , respectively .
it is shown that while the ghz - symmetric states are represented by three real parameters , the rghz - symmetric states are represented by two real parameters . in section iii we classify the entanglement of the rghz - symmetric states .
it is shown that entanglement of the rghz - symmetric states is either @xmath0 or @xmath1 . in section
iv a brief conclusion is given .
in this section we will derive the general forms of the ghz - symmetric and rghz - symmetric states and compare them with each other .
it is not difficult to show that the general form of the ghz - symmetric states is @xmath40 \\
\nonumber & & + \mbox{diag } \left(\tilde{\alpha}_1 , \tilde{\alpha}_2 , \tilde{\alpha}_2 , \tilde{\alpha}_3 , \tilde{\alpha}_2 , \tilde{\alpha}_3 , \tilde{\alpha}_3 , \tilde{\alpha}_2 , \tilde{\alpha}_2 , \tilde{\alpha}_3 , \tilde{\alpha}_3 , \tilde{\alpha}_2 , \tilde{\alpha}_3 , \tilde{\alpha}_2 , \tilde{\alpha}_2 , \tilde{\alpha}_1 \right)\end{aligned}\ ] ] where @xmath41 , @xmath42 , @xmath43 and @xmath44 are real numbers satisfying @xmath45 .
unlike the three - qubit case , @xmath46 is represented by three real parameters .
now , we define following two real parameters @xmath47 , @xmath48 , as @xmath49 \\
\nonumber & & \tilde{z } = { \cal n}_2 \left [ ( \sqrt{10 } + 3 ) \tilde{\alpha}_1 - \tilde{\alpha}_2 \right]\end{aligned}\ ] ] where @xmath50 then , it is straightforward to show that the hilbert - schmidt metric of @xmath46 is equal to the euclidean metric , i.e. , @xmath51 = ( \tilde{x } - \tilde{x}')^2 + ( \tilde{y } - \tilde{y}')^2 + ( \tilde{z } - \tilde{z}')^2\ ] ] where @xmath52 .
the three real parameters @xmath41 , @xmath47 , and @xmath48 can be represented as @xmath53 \nonumber \\ & & \tilde{y } = \frac{{\cal n}_1}{2 } \bigg[\hspace{.01 cm } _
+ { \langle \mbox{ghz}_4 \lvert } \rho_4^{\mbox{\scriptsize{ghz } } } { \lvert \mbox{ghz}_4 \rangle}_+ + _
-{\langle \mbox{ghz}_4 \lvert } \rho_4^{\mbox{\scriptsize{ghz } } } { \lvert \mbox{ghz}_4 \rangle}_- \nonumber \\ & & \hspace{7.0 cm } + 2 ( \sqrt{10 } + 3 ) { \langle \phi \lvert } \rho_4^{\mbox{\scriptsize{ghz } } } { \lvert \phi \rangle}\bigg ] \\
\nonumber & & \tilde{z } = \frac{{\cal n}_2}{2 } \bigg[\hspace{.01cm}(\sqrt{10 } + 3 ) _ + { \langle \mbox{ghz}_4 \lvert } \rho_4^{\mbox{\scriptsize{ghz } } } { \lvert \mbox{ghz}_4 \rangle}_+ \\
\nonumber & & \hspace{3.0 cm } + ( \sqrt{10 } + 3 ) _
-{\langle \mbox{ghz}_4 \lvert } \rho_4^{\mbox{\scriptsize{ghz } } } { \lvert \mbox{ghz}_4 \rangle}_- - 2 { \langle \phi \lvert } \rho_4^{\mbox{\scriptsize{ghz } } } { \lvert \phi \rangle}\bigg ] \end{aligned}\ ] ] where @xmath54 is either @xmath55 or @xmath56
. -qubit ghz state @xmath22 .
the surface of triangle in the tetrahedron is the place where the rghz - symmetric states reside .
( b ) each point in triangle is correspondent to the rghz - symmetric state .
this triangle is equivalent to the triangle in fig .
thus , rghz symmetric states have very small portion and are of zero measure in the whole set of the ghz - symmetric states.,title="fig:",height=226 ] -qubit ghz state @xmath22 .
the surface of triangle in the tetrahedron is the place where the rghz - symmetric states reside .
( b ) each point in triangle is correspondent to the rghz - symmetric state .
this triangle is equivalent to the triangle in fig .
thus , rghz symmetric states have very small portion and are of zero measure in the whole set of the ghz - symmetric states.,title="fig:",height=226 ] in order for @xmath46 to be a physical state the parameters should be restricted to @xmath57 and @xmath58 this physical conditions imply that any ghz - symmetric physical state is represented as a point inside a tetrahedron shown in fig .
1(a ) . in this figure
two black dots represent @xmath22 , respectively .
it is worthwhile noting that the sign of @xmath14 does not change the character of entanglement because @xmath59 , where @xmath60 .
it is straightforward to show that the general form of rghz - symmetric states is @xmath61 \\
\nonumber & & + \mbox{diag } \left(\alpha_1 , \alpha_2 , \alpha_2 , \alpha_1 , \alpha_2 , \alpha_1 , \alpha_1 , \alpha_2 , \alpha_2 , \alpha_1 , \alpha_1 , \alpha_2 , \alpha_1 , \alpha_2 , \alpha_2 , \alpha_1 \right)\end{aligned}\ ] ] with @xmath62 and @xmath63 .
the parameters @xmath14 and @xmath15 are chosen such that the euclidean metric in the @xmath64 plane coincides with the hilbert - schmidt metric @xmath65 again .
the parameters can be represented as @xmath66 \\
\nonumber & & y = \sqrt{2 } \left[_+{\langle \mbox{ghz}_4 \lvert } \rho_4^{\mbox{\scriptsize{rghz } } } { \lvert \mbox{ghz}_4 \rangle}_+ +
_ -{\langle \mbox{ghz}_4 \lvert } \rho_4^{\mbox{\scriptsize{rghz } } } { \lvert \mbox{ghz}_4 \rangle}_- - \frac{1}{8}\right].\end{aligned}\ ] ] it is also worthwhile noting that the sign of @xmath14 does not change the entanglement because @xmath67 .
this is evident from the fact that the rghz - symmetric state is also ghz - symmetric .
since @xmath68 is a quantum state , it should be a positive operator , which restricts the parameters as latexmath:[\[\label{restriction-3 } y \geq \pm 2 \sqrt{2 } x - \frac{\sqrt{2}}{8 } \hspace{2.0 cm } represented as a point in a triangle depicted in fig .
it is easy to show that @xmath46 is rghz - symmetric if and only if @xmath70 , @xmath71 , and @xmath72 or equivalently @xmath73 \hspace{1 cm } \tilde{z } = { \cal n}_2 \left[\frac{\sqrt{10 } + 2}{16 } + \frac{\sqrt{10 } + 4}{2\sqrt{2 } } y \right].\ ] ] using this relation it is possible to know where the rghz - symmetric states reside in the tetrahedron in fig .
in this figure the red triangle is equivalent one of fig .
thus , the states on this triangle are rghz - symmetric . from fig .
1(a ) one can realize that the rghz - symmetric states have very small portion and are of zero measure in the entire set of ghz - symmetric states .
the slocc classification of the @xmath18-qubit pure - state system was first discussed in @xcite by making use of the jordan block structure of some complex symmetric matrix .
subsequently , same issue was explored in several more papers using different approaches@xcite . unlike , however , two- and three - qubit cases
, the results of ref.@xcite seem to be contradictory to each other .
this means that still our understanding on the @xmath18-qubit entanglement is incomplete . in this paper
we adopt the results of @xcite , where there are following nine inequivalent slocc classes ; @xmath74 where @xmath75 , @xmath76 , @xmath77 , and @xmath78 are complex parameters with nonnegative real part . in eq .
( [ slocc - ver ] ) @xmath1 is special in a sense that its all local states are completely mixed . in other words , @xmath1 is a set of normal states@xcite . .
in the polygon abcd states of @xmath0 reside .
theorem 2 implies that there is no one - qubit tensor product three - qubit entangled states in the ghzl - symmetric states .
this fact implies that the rghz symmetry exclude @xmath79 , @xmath80 , and @xmath81 .
theorem 3 implies that there are states of @xmath1 outside the polygon abcd.,height=377 ] in this subsection we examine a question where the states of @xmath0 reside in the triangle in fig .
before proceeding further , it is important to note that there is a correspondence between four - qubit pure states and rghz - symmetric states .
let @xmath2 be a four - qubit pure state .
then , the corresponding rghz - symmetric state @xmath82 can be written as @xmath83 where the integral is understood to cover the entire rghz symmetry group , i.e. , unitaries @xmath84 in eq .
( [ four-1 ] ) and averaging over the discrete symmetries .
for example , if @xmath85 , @xmath82 becomes eq .
( [ general - form ] ) with @xmath86 \\
\nonumber & & \alpha_1 \equiv \frac{1}{16 } + \frac{y}{2\sqrt{2 } } = \frac{1}{8 }
\bigg[|\psi_{0000}|^2 + |\psi_{1111}|^2 + |\psi_{0011}|^2 + |\psi_{0101}|^2 \\
\nonumber & & \hspace{4.0 cm } + |\psi_{0110}|^2 + |\psi_{1001}|^2 + |\psi_{1010}|^2 + |\psi_{1100}|^2 \bigg].\end{aligned}\ ] ] now , we are ready to discuss the main issue of this subsection . * theorem 1 . * _ the rghz - symmetric states of @xmath0-class reside in the polygon abcd in fig .
* proof . *
first we note that when @xmath87 , @xmath0 reduces to the fully separable state @xmath88 .
since lu is a particular case of slocc , this fact implies that all fully separable states are in the @xmath0 .
let @xmath89 , where @xmath90 then , it is easy to derive the parameters @xmath14 and @xmath15 of @xmath91 easily using eq .
( [ 4q-3 ] ) .
our method for proof is as follows .
applying the lagrange multiplier method we maximize @xmath14 with given @xmath15 .
then , it is possible to derive a boundary @xmath92 in the @xmath64 plane .
if a region inside the boundary is convex , this is the region where the @xmath0-class states reside . if it is not convex , we have to choose the convex hull of it for the residential region . from a symmetry
it is evident that the maximum of @xmath14 occurs when @xmath93 .
then the constraint of @xmath15 yields @xmath94 , which gives @xmath95 since the sign of @xmath96 does not change the entanglement class , the region represented by green color in fig . 2 is derived .
since it is not convex , we have to choose a convex hull , which is a polygon abcd in fig .
2 . this completes the proof .
although we start with fully separable state , this does not guarantee that all states in the polygon abcd are fully separable because @xmath0 has @xmath18-way entangled states as well as fully separable states .
the only fact we can assert is that all @xmath0-class rghz - symmetric states reside in the polygon abcd . in this subsection
we will show that the rghz symmetry excludes all slocc classes except @xmath1 .
* theorem 2 . *
_ there is no one - qubit product ghz state in the rghz - symmetric states . _
* let @xmath98 , where @xmath99 then , it is easy to compute @xmath14 and @xmath15 of @xmath100 using eq .
( [ 4q-3 ] ) .
now , we want to maximize @xmath14 with given @xmath15 and @xmath101 . from symmetry of the lagrange multiplier equations it is evident
that the maximum of @xmath14 occurs when @xmath102 and @xmath103 .
then , we define @xmath104 , where @xmath105 and @xmath106 are lagrange multiplier constants , and @xmath107.\end{aligned}\ ] ] now , we want to maximize @xmath14 under the constraints @xmath108 .
first , we solve the two constraints , whose solutions are @xmath109 where @xmath110 and @xmath111 . from @xmath112 one
can express the lagrange multiplier constants as @xmath113 combining eqs .
( [ 4ghz-3 ] ) , ( [ 4ghz-4 ] ) , and @xmath114 , we obtain @xmath115 where @xmath116
. then , the maximum of @xmath14 with given @xmath15 becomes @xmath117 using eq .
( [ 4ghz-5 ] ) and performing long and tedious calculation , one can show that the right - hand side of eq .
( [ 4ghz-6 ] ) reduces to @xmath118 , which results in the identical equation with eq .
( [ 4sep-2 ] ) .
this implies that there is no one - qubit product three - qubit ghz state in the rghz - symmetric states .
this completes the proof . from this theorem
one can conclude that there is no @xmath80 in the rghz - symmetric states , because this class involves one - qubit product ghz - state .
since it is well - known that the three - qubit states consist of fully separable ( s ) , bi - separable ( b ) , w , and ghz states , and they satisfy a linear hierarchy s @xmath11 b @xmath11 w @xmath11 ghz , theorem 2 also implies that there is no one - qubit product w state in the rghz - symmetric states .
thus , rghz symmetry excludes @xmath79 too because this class contains one - qubit product w state when @xmath119 .
this theorem also implies that there is no one - qubit product one - qubit product b state , which excludes @xmath81 .
similarly , one can exclude all classes except @xmath1-class for each slocc class .
next , we generate @xmath120 random numbers and identify them with @xmath121 .
then , using a mapping ( [ 4q-3 ] ) one can compute @xmath14 and @xmath15 for pure state @xmath122 . repeating this procedure over and over and collecting all @xmath123 data
, one can deduce numerically the residential region of this class .
the numerical calculation shows that the residence of all slocc class except @xmath1 is confined in the polygon abcd of fig .
2 . ] . now
, we want to discuss the entanglement classes of remaining rghz - symmetric states . in order to conjecture the classes quickly ,
let us consider the following double bi - separable state @xmath124 such a state belongs to @xmath1 with @xmath125 or @xmath126 . then , eq .
( [ 4q-3 ] ) shows that the parameters of @xmath127 are @xmath128 and @xmath129 , which correspond to the right - upper corner of the triangle in fig .
2 . since mixing can result only in the same or a lower entanglement class , the entanglement class of this corner state should be @xmath1 or its sub - classes .
however , the sub - class of this state should be a class , where fully separable states belong , and those states are confined in @xmath130 . therefore , the corner should be @xmath1 .
this fact strongly suggests that all remaining states in fig .
2 are @xmath1 .
the following theorem shows that our conjecture is correct .
* theorem 3 .
* _ all remaining rghz - symmetric states in fig
. 2 are @xmath1-class . _
* let @xmath131 , where @xmath132 is given in eq . ( [ 4ghz-1 ] ) .
then , it is easy to compute the parameters @xmath14 and @xmath15 of @xmath127 using eq .
( [ 4q-3 ] ) .
similar to the previous theorems we want to maximize @xmath14 with given @xmath15 . from a symmetry of lagrange multiplier equations
it is evident that the maximum of @xmath14 occurs when @xmath133 for later convenience we define @xmath134 , @xmath135 , @xmath136 , @xmath137 , @xmath138 , and @xmath139 . in order to apply the lagrange multiplier method we define @xmath104
, where @xmath140 the constraints @xmath141 and @xmath142 come from @xmath143 and eq .
( [ 4q-3 ] ) , respectively .
now , we have eight equations @xmath144 , @xmath145 , and @xmath108 . analyzing those equations
, one can show that the maximum of @xmath14 occurs when @xmath146 and @xmath147 .
then , the constraint @xmath142 implies @xmath148 which corresponds to the right side of the triangle in fig .
this fact implies that the whole rghz - symmetric states are @xmath1 or its sub - class .
since @xmath0 are confined in the polygon abcd and the remaining classes except @xmath1 are already excluded , the states outside the polygon abcd should be @xmath1-class , which completes the proof .
although we start with a double bi - separable state , this fact does not implies that all states outside the polygon are double bi - separable because @xmath1 contains @xmath18-way entangled states as well as double bi - separable states .
the only fact we can say is that all states outside the polygon abcd are @xmath1-class .
in this paper the ghz and rghz symmetries in four - qubit system are examined .
it is shown that the whole rghz - symmetric states involve only two slocc classes , @xmath0 and @xmath1 .
following ref .
@xcite we can use our result to construct the optimal witness @xmath149 , which can detect the @xmath1-class optimally from a set of @xmath0 plus @xmath1 states .
as remarked earlier if we choose ghz symmetry , the symmetric states are represented by three real parameters as eq .
( [ 4q-1 ] ) shows .
probably , these symmetric states involve more kinds of the four - qubit slocc classes . the slocc classification of eq .
( [ 4q-1 ] ) will be explored in the future .
another interesting extension of present paper is to generalize our analysis to any @xmath150-qubit system .
then , our modification of symmetry should be changed into ` any one - pair , two - pair , @xmath151 , and @xmath4-pair flips ' .
this would drastically reduce the number of free parameters in the set of symmetric states .
this strongly restricted symmetry may shed light on the slocc classification of the multipartite states .
99 r. horodecki , p. horodecki , m. horodecki , and k. horodecki , _ quantum entanglement _ , rev .
* 81 * ( 2009 ) 865 [ quant - ph/0702225 ] and references therein . c. h. bennett , g. brassard , c. crepeau , r. jozsa , a. peres and w. k. wootters , _ teleporting an unknown quantum state via dual classical and einstein - podolsky - rosen channles _ ,
phys.rev . lett .
* 70 * ( 1993 ) 1895 . c. h. bennett and s. j. wiesner , _ communication via one- and two - particle operators on einstein - podolsky - rosen states _ ,
* 69 * ( 1992 ) 2881 .
v. scarani , s. lblisdir , n. gisin and a. acin , _ quantum cloning _
* 77 * ( 2005 ) 1225 [ quant - ph/0511088 ] and references therein .
a. k. ekert , _ quantum cryptography based on bell s theorem _ ,
lett . * 67 * ( 1991 ) 661 . m. a. nielsen and i. l. chuang , _ quantum computation and quantum information _
( cambridge university press , cambridge , england , 2000 ) .
g. vidal , _ efficient classical simulation of slightly entangled quantum computations _ , phys .
* 91 * ( 2003 ) 147902 [ quant - ph/0301063 ] . c. h. bennett , s. popescu , d. rohrlich , j. a. smolin , and a. v. thapliyal , _ exact and asymptotic measures of multipartite pure - state entanglement _ ,
phys . rev . * a 63 * ( 2000 ) 012307 [ quant - ph/9908073 ] .
g. vidal , entanglement monotones , j. mod . opt .
* 47 * ( 2000 ) 355 [ quant - ph/9807077 ] .
w. dr , g. vidal and j. i. cirac , _ three qubits can be entangled in two inequivalent ways _ ,
a * 62 * ( 2000 ) 062314 . v. coffman , j. kundu and w. k. wootters , _
distributed entanglement _ ,
phys . rev . * a 61 * ( 2000 ) 052306 [ quant - ph/9907047 ] .
w. k. wootters , _ entanglement of formation of an arbitrary state of two qubits _ , phys .
lett . * 80 * ( 1998 ) 2245 [ quant - ph/9709029 ] .
a. acn , d. bru , m. lewenstein , and a. sanpera , _ classification of mixed three - qubit states _ ,
* 87 * ( 2001 ) 040401 .
r. lohmayer , a. osterloh , j. siewert and a. uhlmann , _ entangled three - qubit states without concurrence and three - tangle _ , phys .
* 97 * ( 2006 ) 260502 [ quant - ph/0606071 ] ; c. eltschka , a. osterloh , j. siewert and a. uhlmann , _ three - tangle for mixtures of generalized ghz and generalized w states _ , new j. phys .
* 10 * ( 2008 ) 043014 [ arxiv:0711.4477 ( quant - ph ) ] ; e. jung , m. r. hwang , d. k. park and j. w. son , _ three - tangle for rank-@xmath16 mixed states : mixture of greenberger - horne - zeilinger , w and flipped w states _ , phys
. rev . * a 79 * ( 2009 ) 024306 [ arxiv:0810.5403 ( quant - ph ) ] ; e. jung , d. k. park , and j. w. son , _ three - tangle does not properly quantify tripartite entanglement for greenberger - horne - zeilinger - type state _
, phys . rev . * a 80 * ( 2009 ) 010301(r ) [ arxiv:0901.2620 ( quant - ph ) ] ; e. jung , m. r. hwang , d. k. park , and s. tamaryan , _ three - party entanglement in tripartite teleportation scheme through noisy channels _ , quant .
* 10 * ( 2010 ) 0377 [ arxiv:0904.2807 ( quant - ph ) ] . c. eltschka and j. siewert , _ entanglement of three - qubit greenberger - horne - zeilinger - symmetric states _ ,
* 108 * ( 2012 ) 020502 [ arxiv:1304.6095 ( quant - ph ) ] .
j. siewert and c. eltschka , _ quantifying tripartite entanglement of three - qubit generalized werner states _
* 108 * ( 2012 ) 230502 .
c. eltschka and j. siewert , _ optimal witnesses for three - qubit entanglement from greenberger - horne - zeilinger symmetry _ ,
* 13 * ( 2013 ) 0210 ( 2013 ) [ arxiv:1204.5451 ( quant - ph ) ] . c. eltschka and j. siewert ,
_ practical method to obtain a lower bound to the three - tangle _ , phys .
rev . * a 89 * ( 2014 ) 022312 [ arxiv:1310.8311 ( quant - ph ) ] .
eylee jung and daekil park , _ entanglement classication of relaxed greenberger - horne - zeilinger - symmetric states _ ,
* 14 * ( 2014 ) 0937 [ arxiv:1303.3712 ( quant - ph ) ] .
a. osterloh and j. siewert , _ entanglement monotones and maximally entangled states in multipartite qubit systems _ , quant .
comput . * 4 * ( 2006 ) 0531 [ quant - ph/0506073 ] .
f. verstraete , j. dehaene , b. de moor , and h. verschelde , _ four qubits can be entangled in nine different ways _ , phys .
rev . * a 65 * ( 2002 ) 052112 [ quant - ph/0109033 ] .
l. lamata , j. len , d. salgado , and e. solano , _ inductive entanglement of four qubits under stochastic local operations and classical communication _ ,
phys . rev . * a 75 * ( 2007 ) 022318 [ quant - ph/0603243 ] ; y. cao and a. m. wang , _ discussion of the entanglement classification of a 4-qubit pure state _ , eur .
j. * d 44 * ( 2007 ) 159 ; o. chterental and d. z. djokovc , in _ linear algebra research advances _ , edited by g.
d. ling ( nova science publishers , inc . ,
hauppauge , ny , 2007 ) , chap .
133 - 167 ; d. li , x. li , h. huang , and x. li , _ slocc classification for nine families of four - qubits _ ,
* 9 * ( 2009 ) 0778 [ arxiv:0712.1876 ( quant - ph ) ] ; s. j. akhtarshenas and m. g. ghahi , _ entangled graphs : a classification of four - qubit entanglement _ , arxiv:1003.2762 ( quant - ph ) ; l. borsten , d. dahanayake , m. j. duff , a. marrani , and w. rubens , _ four - qubit entanglement classification from string theory _ , phys . rev . lett .
* 105 * ( 2010 ) 100507 [ arxiv:1005.4915 ( hep - th ) ]
. f. verstraete , j. dehaene , and d. de moor , _ normal forms and entanglement measures for multipartite quantum states _ ,
phys . rev . * a 68 * ( 2003 ) 012103 [ quant - ph/0105090 ] . | similar to the three - qubit greenberger - horne - zeilinger ( ghz ) symmetry we explore the four - qubit ghz symmetry group and its subgroup called restricted ghz symmetry group . while the set of symmetric states under the whole group transformation is represented by three real parameters , the set of symmetric states under the subgroup transformation is represented by two real parameters . after comparing the symmetric states for whole and subgroup ,
the entanglement is examined for the latter set .
it is shown that the set has only two slocc classes , @xmath0 and @xmath1 .
extension to the multi - qubit system is briefly discussed . |
cavity polaritons are bound light - matter states that appear due to the strong coupling of 2d excitons and photons in semiconductor microcavities.@xcite the optically driven system of polaritons behaves like a weakly imperfect bose gas , which results in a spectacular row of polariton collective phenomena such as multistability,@xcite parametric scattering,@xcite pattern formation,@xcite self - organization effects,@xcite dynamical bose - einstein condensation.@xcite polariton multistability attracts much attention as a potential candidate for very fast picosecond range optical switching on a micron size scale.@xcite moreover , the sharp transitions in intra - cavity field , stemming from the multistability , can proceed concurrently with inter - mode parametric scattering , which determines interesting ways of polariton self - organization under pumping near the inflection point of the lower polariton branch ( the `` magic angle'').@xcite the polariton bistability has recently been studied in the `` scalar '' approximation neglecting the spin degrees of freedom.@xcite because of the mutual interaction of polaritons , their energy effectively depends on the intra - cavity field magnitude .
the repulsion between excitons leads to the blueshift of polariton energy .
when the pump is itself blue - detuned from the polariton resonance , a strong positive feedback loop between the resonance energy and the field amplitude is created in a certain range of system parameters , resulting in sharp jumps in the intra - cavity field and , hence , in the cavity transmission .
consequently , the transmission signal can exhibit a prominent hysteresis in the dependence on continuous - wave ( cw ) pump intensity.@xcite the response of the optically excited polariton system becomes more complicated with allowance made for exciton spin degrees of freedom ( that corresponds to arbitrary optical polarization of the intra - cavity field ) . in the general case
the system has up to four stable states under a given cw pump , whereas the actual state of the system is determined by the history of the excitation process.@xcite in critical points , where a number and/or stability of stationary solutions change , the system can exhibit sharp jumps in both the amplitude and polarization of the intra - cavity field . under a spatially inhomogeneous ( e.g. , gaussian shaped ) cw excitation
the system can also exhibit a nontrivial spatial distribution of polarization of the luminescence signal , like the `` spin ring '' patterns,@xcite due to the same underlying phenomena .
the strong multistability effect predicted in ref . has recently been observed experimentally,@xcite including the sharp jumps in the cavity transmission under a smooth variation of the pump polarization degree .
the spin ring patterns under gaussian shaped excitation have also been reported.@xcite all these experiments were carried out under the cw excitation .
thus , one of the still remaining questions is the dynamical peculiarities of the transitions in a multistable system , which could only be traced using the time - resolved techniques . particularly , the characteristic switching times between different stability branches call for the experimental study , for they can have a crucial impact upon the practical implementations .
another question concerns the theoretical approach allowing to describe the observed multistability effect in microcavities .
traditionally , the multistability is considered in terms of the self - acting classical fields corresponding to macro - occupied coherent polariton modes which appear under a coherent resonant excitation.@xcite although such an approach is supposed to be sufficient to describe the bistability in a circularly polarized system,@xcite it gives wrong predictions for the general case of elliptically polarized excitation.@xcite most probably , the incoherent states of the exciton reservoir which are inevitably excited in optical experiments have a substantial impact on the decay rates and energies of polaritons . for instance , the nonlinear decay of polaritons with different circular polarizations was taken into consideration in order to explain the experimental data in ref . .
further , the nonequilibrium transitions reported in ref . can only be reproduced in calculations taking into account the reservoir induced shifts of the polariton energy . in the present work we report the experimental study of nonequilibrium transitions in the multistable cavity polariton system .
unlike in recent refs .
, we have studied the optically polarized system under a pulsed nanosecond - long excitation , which allowed us to trace the time - resolved dynamics of the intra - cavity field .
the shifts in polariton energy were reflected by temporal variations in the transmission energy spectrum .
thus , the employed technique is capable of visualizing the temporal correlations between the resonance energy and intensity of the intra - cavity field .
we discuss in detail the time dependence of the transmission signal polarization , below as well as above the threshold , for several polarizations of the pump beam .
the observed polarization behavior can not be reproduced within a semi - classical model based on the gross - pitaevskii equations considered in refs . .
the experimental results allowed us to develop a phenomenological model to describe a multistable polariton system with a proper regard to the exciton reservoir .
the reservoir excitons shift the polariton energy that , in turn , influences the thresholds of nonequilibrium transitions .
both the linear and nonlinear mechanisms of exciton scattering into the reservoir are found to be significant within a sub - nanosecond time scale . while the nonlinear decay of cross - circularly polarized excitons leads to the levelling of the @xmath0 and @xmath1 jump points ( which has also been found in refs . in the cw pump regime ) , the linear decay mechanism leads to the temporal delay of the jumps with respect to the pump intensity peak .
further , we apply the developed approach to simulate the parametric scattering under pumping at the magic angle .
as we show , this model allows to explain self - consistently the temporal dependences of polarizations of both the driven mode and the scattering signal which appears at the polariton branch bottom .
the paper is organized as follows . in sec .
[ sec : theory ] the available models of the multistability effect are considered and compared with each other in view of the recent experimental results of refs . .
[ experimental ] references the experimental setup . sec .
[ sec : experiment ] contains the experimental results and compares them with the calculations performed in the framework of the suggested model , for the cases of pumping at normal incidence ( sec .
[ sec : normal ] ) and at the magic angle ( sec .
[ sec : magic ] ) , with a brief reference to some of the still unresolved issues ( sec .
[ sec : discussion ] ) .
the results are summarized in conclusion ( sec .
[ sec : conclusions ] ) .
the multistability effects in microcavities were recently considered in the framework of gross - pitaevskii equations written for the macro - occupied polariton modes.@xcite these equations can also be written for the strongly coupled exciton and photon fields ( @xmath2 and @xmath3 respectively ) in the cavity active layer:@xcite @xmath4 ( in the @xmath5 basis ) . here
, @xmath6 stands for the incident electric field that is usually treated as a plane wave : @xmath7 ; @xmath8 and @xmath9 are the eigenfrequencies and decay rates of the intra - cavity photon and exciton modes ; @xmath10 and @xmath11 are the cavity response coefficients , and @xmath12 is exciton polarizability ( so that @xmath13 equals rabi splitting ) ; @xmath14 are the matrix elements of the interaction between excitons with same ( @xmath15 ) and opposite ( @xmath16 ) circular polarizations . for simplicity , eqs .
( [ ef_p][xp_m ] ) are written for only the driven polariton mode with zero quasi - momentum , although they can be easily generalized to the case of essentially many - mode system.@xcite in the latter case , the usage of the exciton - photon basis ( instead of the simplified polariton basis ) brings the advantage of taking account of the te / tm splitting effects as well as the angular dependence of the polariton interaction constants . as will be shown below , it is also convenient for taking account of the incoherent exciton reservoir .
the multistability conditions are defined by the relation between exciton - exciton interaction constants , @xmath15 and @xmath16 . according to the study of polariton spin quantum beats in a gaas cavity,@xcite the cross - circularly polarized excitons exhibit attraction ( @xmath17 ) that is weaker than the repulsive interaction of co - circular excitons ( @xmath18 and @xmath19 ) .
the fact that @xmath16 is negative is also confirmed by the rotation of polariton polarization direction during the process of parametric scattering.@xcite in the general case the processes of exciton - exciton interaction are mediated by biexciton and dark exciton states,@xcite hence the relation between the interaction constants of _ polaritons _ can be dependent on the exciton - photon detuning.@xcite more precisely , one should also consider the saturation of excitonic transitions as an additional nonlinear mechanism affecting the blueshift of polaritons under a sufficiently strong excitation.@xcite if @xmath18 and @xmath20 ( as assumed in refs . basing on the experimental estimations of ref . ) , then the @xmath5-components of the intra - cavity field are almost uncoupled . if @xmath17 , then the threshold intensity @xmath21 .
contrary to the above predictions , @xmath22 has been found to be slightly less than @xmath23 in the experiments using cw as well as time - resolved pumping techniques.@xcite@xmath24 was observed under normal incidence cw excitation in the 3@xmath25 m wide cavity mesa @xcite and ( under the same cw pump conditions ) in ordinary ( spatially uniform ) microcavity structure.@xcite in the latter case , the negative sign of @xmath16 in a studied sample was proven by the observed rotation of the scattering signal polarization . in the case of weakly
circularly polarized pump , the jumps in both @xmath0 and @xmath1 modes were found to occur simultaneously@xcite in the sense of cw excitation conditions with `` infinitesimally slow '' variations of pump intensity . in view of the gross - pitaevskii equations ( [ ef_p])([xp_m ] ) , this looks like a manifestation of a large repulsion between polaritons with opposite pseudospins . at the same time , in a case
when the pump energy is less than a half of biexciton energy , one expects an attractive interaction between excitons with different pseudospins,@xcite so they can not provide blueshift for each other ; it agrees with the observed indications of the negative sign of @xmath16.]in order to explain it , two theoretical approaches have been proposed .
the first one@xcite presumes a repulsive interaction of polaritons with opposite pseudospins ( @xmath5 polarizations ) in the framework of gross - pitaevskii equations of type ( [ ef_p])([xp_m ] ) .
the second approach@xcite considers the additional blueshift resultant from an exciton reservoir , which leads to @xmath26 without the necessity for a repulsive interaction between cross - circularly polarized excitons .
( a somewhat similar model has previously been proposed in ref .
to describe the interaction of a `` spinless '' polariton condensate with a reservoir . ) in the present work we explore the second model accounting for the transitions of the optically driven excitons into the incoherent ( reservoir ) state in which the overall pseudospin is relaxed , so that the reservoir provides equal blueshifts for both polarization components of the coherent state.@xcite those transitions are introduced phenomenologically as non - radiative decays of bright excitons accorded with the rate of occupation of the optically inactive reservoir .
the model for the intra - cavity electric field @xmath3 and exciton polarization @xmath2 coupled with the integral population of the reservoir @xmath27 can be written as @xmath28 \mathcal{p}_+ + a \mathcal{e}_+ , \label{xp_p_r } \\
i \dot{\mathcal{p}}_- & = \bigl [ \omega_\mathrm{x } + v_1 |\mathcal{p}_-|^2 + v_2 |\mathcal{p}_+|^2 + ( v_1 + v_2 ) \ , \mathcal{n}/2 - { } \nonumber\\ & i \left ( \gamma_\mathrm{x } + \gamma_\mathrm{xr } + v_\mathrm{r } |\mathcal{p}_+|^2 \right ) \bigr ] \mathcal{p}_- + a \mathcal{e}_- , \label{xp_m_r }
\\ \dot{\mathcal{n}}~ { } & = -\gamma_\mathrm{r } \mathcal{n } + 2 \gamma_\mathrm{xr } \left ( |\mathcal{p}_+|^2 + |\mathcal{p}_-|^2 \right ) + { } \nonumber\\ & 4 v_\mathrm{r } |\mathcal{p}_+|^2 |\mathcal{p}_-|^2 . \label{res}\end{aligned}\ ] ] this model is a generalization of eqs .
( [ ef_p])([xp_m ] ) . here , @xmath29 is an additional decay rate of excitons that corresponds to the linear mechanism of light absorption , providing the term @xmath30 that provides an additional occupation of the reservoir ( @xmath31 stands for the reservoir own decay rate . and
@xmath32 ) has been considered in ref . .
in such a case , the reservoir induced blueshifts are determined mainly by the ratio between reservoir s decay rate and the coupling constant @xmath33 .
for instance , the calculations performed with different @xmath34 but constant @xmath35 and a sufficiently slow variation of the pump power lead to exactly same results for the cases of circular and linear pump polarizations . in this connection , the experiments carried out with time resolution using the nanosecond long pump pulses ( described in sec . [
sec : experiment ] ) are capable of clarifying the characteristic lifetimes of the reservoir states , as well as the relation between linear and nonlinear mechanisms of polaritons scattering into the incoherent exciton reservoir .
for instance , the former mechanism is expected to play a role irrespectively of pump polarization , affecting the dynamics of polariton energy and the time moments when the nonequilibrium transitions occur . ]
microscopically , the nonlinear absorption of cross - circularly polarized excitons is resultant from the scattering of a pair of bright excitons with opposite spins ( @xmath36 and @xmath37 ) into dark excitons ( @xmath38 and @xmath39 ) , which is closely related to biexciton creation ( see ref . for details ) .
we do not consider the reverse transitions of incoherent ( reservoir ) excitons into the driven mode ; this approximation is valid as long as the occupation of the reservoir states is small enough .
under resonant pumping at @xmath40 the energy mismatch of the pump with respect to free exciton makes the filling of reservoir quite ineffective , but it is still possible due to energy level broadening and non - zero temperature .
the reservoir can accumulate excitation due to its comparatively long lifetime ( see sec .
[ sec : experiment ] ) .
equations ( [ ef_p_r])([res ] ) can easily be generalized to the case of many - mode system , with exactly the same nonlinear interaction terms ( which are `` local '' in the real space ) , and a linear part being in the @xmath41-space the same as in ref . .
in the many - mode calculations represented below in figs .
[ fig:2 ] , [ fig:3 ] , and [ fig:11][fig:13 ] , the cavity dispersion @xmath42 and response coefficients @xmath43 , @xmath44 for te and tm cavity modes are properly taken into account using the transfer matrix technique.@xcite ( thin solid line ) .
thick solid line shows the data calculated using eqs .
( [ ef_p_r])([res ] ) with @xmath45 , @xmath46 , and @xmath47 ( explicitly as in ref . ) .
dash - and - dot line represents the same dependence calculated using eqs .
( [ ef_p])([xp_m ] ) with @xmath45 ( this ratio was used in refs . ) .
dash line corresponds to the case of @xmath48 .
the calculated data is time - averaged for each temporal point over 100 ps intervals , in order to eliminate the fast transitional effects which are not observable under the cw pump conditions .
( b - d ) the corresponding time dependences of the intra - cavity field intensity in @xmath5 polarization components . ] ) ([res ] ) with @xmath49 , @xmath50 , and @xmath47 ( explicitly as in ref . ) .
the pump dcp is @xmath51 ( as in fig .
[ fig:1 ] ) .
panels _ a e _ show the spatial distribution of intensity of the intra - cavity field in @xmath0 and @xmath1 polarization components ( solid and dashed lines , respectively ) , for several values of pump power @xmath52 ; the pump distribution profile is shown by dotted lines . @xmath53
roughly corresponds to the threshold excitation density .
panels _ d
f _ represent the corresponding spatial distribution of the intra - cavity field dcp . ] and @xmath54 .
all other parameters coincide with those of fig .
[ fig:2 ] . unlike fig .
[ fig:2 ] , almost no ring structure appears in the cavity transmission dcp even at @xmath55 . ] the two models , eqs .
( [ ef_p])([xp_m ] ) and ( [ ef_p_r])([res ] ) , can give similar predictions for the cases of circular and linear polarizations of excitation .
in particular , the key experimental effect waiting for explanation that the threshold of linearly polarized pump is smaller ( @xmath56)can be reproduced within the eqs .
( [ ef_p])([xp_m ] ) with @xmath57 .
nonetheless , the transitional dynamics under the intermediate pump polarizations @xmath58 are completely different in the two systems , see an example in fig .
[ fig:1 ] . if @xmath17 and @xmath59 ( fig .
[ fig:1]_a _ , solid line , and fig .
[ fig:1]_b _ ) , the cavity transmission polarization increases with pump in the sub - threshold area , due to the dominant blueshift of the leading polarization component ( @xmath60 ) .
on reaching the threshold , the leading component jumps up to the high - energy state , which causes a step - like increase in reservoir population and the blueshift of @xmath1 mode .
consequently , the minor mode also enters the above - threshold area , and the degree of circular polarization ( dcp ) of transmission ( @xmath61 ) decreases down to the values smaller than the pump dcp . the same effect results in formation of the spin ring patterns under the gaussian shaped excitation@xcite in a close vicinity of the threshold pump power ( fig .
[ fig:2 ] ) . as a matter of fact ,
the ring shaped pattern of dcp indicates that both polarization components reach the high - energy state at the center of the pump spot where , therefore , the transmission dcp is minimal .
the spatial distribution of @xmath61 ( fig .
[ fig:2]_d f _ ) exhibits a well pronounced minimum at @xmath62 already when the threshold is surpassed by 1.5 times ( fig .
[ fig:2]_e _ ) , in agreement with the experimental observations ( see fig.3 in ref . ) .
this markedly differs from a prediction of the conventional theory based on gross - pitaevskii equations@xcite with @xmath63 ( see fig . [ fig:3 ] ) which does not exhibit the rings even at @xmath64 for @xmath65 .
the temporal dynamics for @xmath66 is shown in fig .
[ fig:1]_a _ ( dash - and - dot line ) and in fig .
[ fig:1]_c_. in the opposite case , when @xmath67 and the system is `` purely coherent '' ( fig .
[ fig:1]_a _ , dash line , and fig .
[ fig:1]_d _ ) , which also corresponds to the right relation @xmath68 , the polarization dynamics is completely different .
first , the transmission dcp decreases in the sub - threshold area , since an increase in the `` leading '' @xmath0 component provides a larger blueshift for @xmath1 , so that the difference in @xmath5 intensities weakens with increasing pump . on reaching the threshold , @xmath61 shows a minor step - like increase followed by the very weak changes with further increasing pump power .
no prerequisites for the spin ring formation are satisfied in this case .
thus , the observed phenomena the disparity of thresholds , @xmath56 , together with the field polarization dynamics cannot be reproduced within eqs . ( [ ef_p])([xp_m ] ) using only two constants @xmath14 of exciton - exciton interaction even if a single polariton mode at @xmath69 is considered .
the experiments revealing both the energy ( blueshift ) and polarization dynamics of the driven polariton mode under the nanosecond - long excitation pulses are described below in sec .
[ sec : experiment ] .
the microcavity structure grown by a metal - organic vapor - phase epitaxy technique has top ( bottom ) bragg reflectors composed of 17 ( 20 ) @xmath70 al@xmath71ga@xmath72as / alas layers .
the @xmath73 gaas cavity contains six @xmath74 nm thick in@xmath75ga@xmath76as / gaas quantum wells .
the rabi splitting is about @xmath77mev .
a gradual variation of the active layer thickness along the sample provides a change in the photon mode energy @xmath78 and , accordingly , in the detuning @xmath79 between the exciton @xmath80 and photon @xmath81 mode energies .
experiments are carried out in several regions of the same sample with @xmath82 ( sec .
[ sec : normal ] ) and @xmath83 ( sec .
[ sec : magic ] ) .
the sample is placed into the optical cryostat with controlled temperature . to excite the cavity
, we use a pulsed ti : sapphire laser producing picosecond long pulses at a repetition rate of 5 khz . prior to coming into the cryostat , the pulses pass trough a long multi - mode optical fiber and then through a monochromator . after the fiber , the pulses have a duration of about @xmath84ns , and after the monochromator they have spectral full width at half maximum ( fwhm ) of @xmath85mev .
the excitation is performed either along the normal to the cavity plane ( @xmath86 ) or at the magic angle ( @xmath87 ) slightly above the lower polariton branch resonance , with various pump polarizations ( circular , linear , or elliptical ) .
the main axis of pump polarization is directed along the @xmath88 axis of the structure .
the pump beam is focused onto the spot with a diameter of @xmath89 m .
the kinetics of the cavity transmission signal @xmath90 is detected at @xmath91k by the streak camera with spectral , angular , and time resolution of @xmath92mev , @xmath93 , and @xmath94ps , respectively , in various polarizations .
the temporal dependences of the signal are averaged over @xmath95 pulses , for we have to collect the signal in order to eliminate noises .
the information on the intra - cavity field kinetics is obtained from the cavity transmission measurements.@xcite the active region of the cavity is separated from a detector by the bragg mirror that does not introduce any nonlinearity and/or spectral selectivity .
thus , the intensity of the transmission signal , @xmath96 , is proportional to the squared magnitude of the intra - cavity electric field @xmath97 , and the first momentum @xmath98 at @xmath99 reflects the time dependence of the average ( effective ) energy of excited mode @xmath100 . for different pump powers @xmath52 ( _ b d _ and _ f h _ ) .
the pump is polarized circularly .
the normalized temporal shape of the excitation is shown at the left side ( white solid line ) ; the weighted mean value of the pump energy is shown by the vertical line . left and right panels represent the experimental and calculated data , respectively .
the color scale is logarithmic . ]
figures [ fig:4]_b d _ show the measured time dependences of transmission signal @xmath101 under the excitation with circularly polarized ( @xmath0 ) light .
the spectra are recorded in the same @xmath0 polarization ; the fraction of signal in the opposite polarization ( @xmath1 ) does not exceed a few percent . at low peak excitation density @xmath102kw/@xmath103
the maximum of the signal is found at the low energy side of the excitation , near the lower polariton eigenfrequency .
although the blueshift in polariton energy remains relatively small , the time dependence of the signal differs markedly from that of the pump pulse .
this indicates the onset of nonlinear processes at the red edge of pump spectrum .
in particular , the maximum of the signal is delayed for a few hundred picoseconds with respect to the pump peak .
nonlinearities in both the magnitude and spectral position of the signal increase with pump power : as the peak energy of the signal shifts up to the pump energy @xmath104 , ( i ) the signal intensity shows the superlinear ( threshold like ) increase and ( ii ) the time when the signal reaches its maximum value shifts backward to the onset of excitation pulse .
the observed correlation between the energy blueshift and intensity of the signal is expected in the framework of the multistability model@xcite eqs .
( [ ef_p])([xp_m ] ) , however this model fails to explain the process of instability development .
figures [ fig:4]_e h _ show that the observed time and spectral dependences can be qualitatively reproduced in simulations using eqs . with appropriate time and energy shapes of the excitation ( see appendix ) .
the chosen parameters of microcavity coincide with those used in the experiment ; the constants of interaction with reservoir are considered below . and @xmath105 , resp . ) for different pump polarizations @xmath106 ( circular , elliptical , and linear polarizations ) .
the top panels show the time shape of the exciting pulse . left and right panels represent the experimental and calculated data , respectively . in the experiment , @xmath107 . ] ) . left and right panels represent experimental and calculated data , respectively . in the experiment , @xmath107 . ]
components in the range of polarization instability ( experiment ) . weighted mean values of the energy @xmath108 are indicated by arrows . ]
the temporal dependence of transmission under the excitation with linearly , elliptically and circularly polarized pulses ( dcps 0 , 0.2 , 0.6 , and 1 ) at @xmath109kw/@xmath103 is shown in fig .
[ fig:5 ] . the temporal delay of the dominating @xmath0 and @xmath110 polarization components of the signal with respect to the pump peak is seen to be in the range of 150200ps , which is much larger than the lifetime of polaritons with @xmath111 .
the `` anomalous '' increase in the intra - cavity field observed in the range of decreasing pump power ( after the pump peak is already passed ) can be explained by the blueshift in effective polariton energy . in turn , this blueshift could only be provided by the long - lived polariton states accumulated during the pulse .
such an effect corresponds to the `` linear '' process of polariton scattering into the exciton reservoir , which is determined by the ratio between @xmath34 and @xmath29 in eqs .
( [ xp_p_r])([res ] ) .
the experimental results are qualitatively reproduced in numerical simulations with @xmath112 , @xmath113 , @xmath114mev , and @xmath115mev .
these parameters are chosen to meet the following conditions : ( i ) the overall occupation of the reservoir is comparable to that of the driven ( optically active ) polariton mode , so that @xmath56 , and ( ii ) the temporal peak of the signal can be delayed by hundreds of picoseconds with respect to the excitation peak even for @xmath116 .
note the long decay times of the reservoir states , @xmath117ps .
so long times are characteristic of excitons localized due to fluctuations of quantum well potential and/or free excitons with large lateral wave numbers .
figures [ fig:5]_b _ and [ fig:5]_e _ reveal the transmission signal to be retaining both circular and linear pump polarizations ; in these cases only a weak depolarization of the transmitted light is observed .
retaining of the circular polarization in agreement with the angular momentum conservation law is expected .
measurements performed under linearly polarized pump in a wide range of @xmath118kw/@xmath103 ( not shown ) have revealed the magnitude of transmitted light depolarization to be within the limits of 1217% , almost independently of the pump power . on the other hand , under the elliptically polarized excitation ( @xmath119 )
the signal dcp does depend on the pump intensity due to the anisotropy in polariton - polariton interaction.@xcite accordingly , the ratio between both circularly and linearly polarized components of the signal intensity ( @xmath5 and @xmath105 , respectively ) varies with time ( fig . [ fig:5]_c , d _ ) .
the corresponding temporal dependences of the degrees of circular and linear polarizations of the signal are shown in fig .
[ fig:6 ] . from comparison of figs .
[ fig:5 ] and [ fig:6 ] , it is seen that the reaching of maximum signal intensity at @xmath120 is accompanied by a decrease in the signal dcp , in much the same way as the cw driven system behaves ( fig .
[ fig:1 ] ) , though the latter system exhibits the expectedly sharper transitions .
on the other hand , our experimental technique allows us to visualize the energies of intra - cavity field in addition to their intensities .
the information on the energies is extracted from the transmission spectra as shown in fig .
[ fig:7 ] since @xmath121 .
the energies of intra - cavity field in @xmath0 and @xmath1 polarizations demonstrate a well pronounced blueshift with increasing polariton density and a well pronounced splitting .
in particular it is seen in fig .
[ fig:7 ] that in the range of maximum polarization the @xmath0 component of the field is markedly ( by 0.3 mev ) blue - shifted with respect to @xmath1 . the basis in which shown degree of linear polarization ( dlp ) was measured is related to the pump polarization axis .
we observe no linear polarization in the basis rotated by @xmath122 with respect to the former one .
thus , the sum of squared stokes parameters per signal intensity is less than 1 , which means that the signal is partially depolarized ; this is because the signal polarization components are measured as averaged quantities .
, under excitation with elliptically polarized pulses with @xmath123 .
grey dotted line corresponds to the normalized temporal shape of excitation . left and right panels represent experimental and calculated data , respectively . ] of the transmission energy in @xmath0 ( thick lines ) and @xmath1 ( thin lines ) polarization components ( _ d , h _ ) , for several values of the pump power @xmath52 .
the pump dcp is @xmath124 .
the mean value of pump energy , which exhibits minor variations in a close vicinity of @xmath125 , is shown by solid thin lines in panels ( _ d , h _ ) . left and right panels represent experimental and calculated data , respectively . ] components of the signal with respect to the pump vs. pump power ( as defined by eq .
( [ eq : delay ] ) ) , for two pump polarizations , @xmath126 ( panels _ a , c _ ) and @xmath127 ( panels _ b , d _ ) . left and right sides of the figure represent experimental and calculated data , respectively . ] in order to clarify the physics underlying the signal polarization transitions , let us consider the series of measurements performed under elliptically polarized pulses with @xmath123 and various pump intensities .
[ fig:8 ] represents measured ( left side ) and calculated ( right side ) time dependences of the transmission signal in @xmath5 polarization components . from these series
, the pump power dependence of the temporal delay between the signal and pump peaks is evidenced .
the certain conditions for a switch in signal polarization depend on both the pump intensity and its temporal shape . in case of the very low pump densities ,
the intra - cavity field is insufficient for multistability to reveal ; on the other hand , if the pump density is increasing quickly enough then a jump into a high - energy state happens with no delay related to the long - lived reservoir , since most of the intra - cavity field is concentrated in the driven ( optically active ) polariton state . except the case of maximum @xmath52 ,
the dynamics shown in fig .
[ fig:8 ] lies in the intermediate range of pump powers which all are insufficient to pull a polariton state by itself up to the threshold magnitude . instead of that
, the non - equilibrium transitions proceed through the mediation of excitation accumulated by the reservoir . in such a case
the development of instability in the leading polarization component ( @xmath0 ) always precedes the reservoir induced switch in signal polarization .
figure [ fig:9 ] , which also shows a series for two peak pump intensities at fixed @xmath123 , allows us to visualize the temporal correlations between the energy ( in both @xmath5 polarization components ) and polarization degree of the transmission signal . the overall process of the instability development is accompanied by blueshifts in @xmath5 energies . at the first stage ( which starts the earlier , the higher is pump density @xmath52 ) , the leading polarization component ( @xmath0 ) enters the region of positive feedback between its effective resonance energy @xmath128 and the @xmath0 field amplitude , so the difference in @xmath5 energies grows along with the dcp of transmission signal ( fig .
[ fig:9]_b , d _ ) . accordingly , a well pronounced decrease in the dlp of transmission is observed at the same time interval ( fig .
[ fig:9]_c _ ) . at approximately the half way through the energy growth
( see panel _ d _ ) , the integral occupation of reservoir becomes sufficient to provide the shift in the minor component ( @xmath1 ) , which leads to decrease in both the energy difference @xmath129 and dcp , and to the corresponding increase in dlp .
all these effects are qualitatively reproduced in calculations shown in fig .
[ fig:9]_e h_. thus , the observed dynamics confirm the effect of reservoir , though the energy broadening makes the transitions much smoother compared to the cw driven system ( fig .
[ fig:1 ] ) .
figure [ fig:10 ] shows the time delay of the signal with respect to the excitation pulse in a wider range of pump powers @xmath52 ; it summarizes the consideration of the threshold effects in the cavity transmission .
the value of the delay @xmath130 is determined as the difference between the moments the signal and the pump reach their half - maxima for the first time : @xmath131 at small @xmath52 the response is linear , and the delays @xmath130 are expectedly small . at threshold ,
the superlinear growth of the signal is observed with @xmath130 of about 150 to 200 picoseconds , i.e.at the back front of excitation .
thus , this transition is induced by the reservoir as explained above . with further increasing @xmath52 ,
the optically active fraction of excitons grows , which lowers the system s intertness , and the delay of response decreases .
note , that the minor ( @xmath1 ) polarization component exhibits the superlinear growth at the same _ peak _ power @xmath52 as @xmath0 , and the relative delay of @xmath0 and @xmath1 components , @xmath132 , remains nearly constant above the threshold .
this is an extra evidence of the fact that the jump in the minor component is provided by the reservoir even at high @xmath52 .
let us now consider the polarization properties of the signal of parametric scattering generally referred to as cavity `` optic parametric oscillator '' ( opo ) signal@xcite which appears at the branch bottom ( @xmath111 ) if pumping is at the magic angle ( @xmath134 ) .
there are two points to discuss here .
first , in the general case of elliptically polarized excitation the signal is expected to reflect variations in the dcp of the driven mode that , in turn , are shown to be strongly affected by the excitonic reservoir .
second , if the pump is polarized linearly , the experiments show the @xmath135 rotation of the signal polarization axis;@xcite thus , we have to check if eqs .
( [ ef_p_r])([res ] ) can reproduce this effect in spite of the reservoir induced interaction between the cross - circularly polarized excitons .
the experiments are carried out at negative detuning between the photon and exciton ( @xmath83 ) in order to have the pump frequency close to that employed for the normal incidence pumping considered in sec [ sec : normal ] .
the cavity and excitation parameters used in the simulations coincide with the experimental ones , whereas the nonlinear interaction constants as well as reservoir characteristics are the same as used in the previous section .
( _ d , j _ ) in two polarization components @xmath0 and @xmath1 ( thick and thin lines , resp . ) , opo signal intensity ( _ e , k _ ) and dcp ( _ f , l _ ) , for several values of the pump power @xmath52 . straight lines in panels _
c , i _ indicate the mean value of pump energy . left and right panels represent experimental and calculated data , respectively . ]
polarized linearly along the tm cavity mode ( @xmath136 ) , for some different values of the pump power @xmath52 . left and right panels represent the experimental and calculated data , respectively .
the temporal shape of excitation is indicated in panels _ a , c _ as a guide for eye . ] . left and right panels represent experimental and calculated data , respectively .
the temporal shapes of excitation are indicated in panels _ a , c _ as a guide for eye . in calculation , the te / tm splitting of the driven polariton mode equals to @xmath137mev . ]
figure [ fig:11 ] represents measured and calculated dynamics of intra - cavity field at @xmath138 ( which corresponds to the cavity transmission ) and @xmath111 ( opo signal ) under elliptically polarized excitation with @xmath139 and different pump powers @xmath52 .
when @xmath52 exceeds the critical value ( in the experiment @xmath140 ) , a well pronounced superlinear growth of the cavity transmission ( fig .
[ fig:11]_b _ ) is observed at @xmath141 . near the threshold
, a doubling of the pump power @xmath52 leads to an order - of - magnitude growth of the opo signal intensity ( fig .
[ fig:11]_e _ ) .
the observed nonlinearities indicate the onset of parametric instability , being , however , by far not as sharp as under cw excitation.@xcite the dynamics of the driven mode is similar to that under the pumping at normal incidence . the growth of the field amplitude , which continues with increasing pump density ( fig .
[ fig:11]_b _ ) , expectedly leads to the growth of transmission dcp ( fig .
[ fig:11]_c _ ) due to the dominantly repulsive exciton - exciton interaction . at @xmath142ns
the transmission starts to behave strongly nonlinearly with respect to the driving field , and the peak of transmission becomes significantly delayed with respect to the peak of pump pulse .
the latter indicates that the transition is assisted by the reservoir induced blueshift .
accordingly , the jump in transmission dcp is immediately followed by dropping back as soon as the system enters the above - threshold region where the energies of @xmath5 components of intra - cavity field ( fig . [ fig:11]_d _ ) become partially levelled due to the reservoir filling .
the domination of the leading polarization component ( @xmath0 ) is strongly enhanced in the opo signal ( fig .
[ fig:11 ] ) whose dcp reaches 0.85 at @xmath143ns . the subsequent lowering of dcp of the driven mode forces the opo signal , too ,
to partially loss its polarization .
however , the signal restores a high dcp during the further evolution accompanied by the decrease in transmission intensity .
the calculations performed on the basis of eqs .
( [ ef_p_r])([res ] ) , which are presented in the right panels of fig .
[ fig:11 ] , reproduce the observed threshold - like behavior of intra - cavity field along with its polarization properties . however , a quantitative agreement is not achieved , partially due to a complicated spectral shape of the pump pulses ( see fig . [ fig:4 ] ) and due to intrinsic limitations of the suggested theoretical model ( see discussion in sec . [
sec : discussion ] ) .
let us turn to the case of exactly linear polarization of the excitation at @xmath144 ( figs .
[ fig:12 ] and [ fig:13 ] ) . in this case
the opo signal depends on whether the pump excites a pure cavity state
one of te and tm modes which are the eigenstates of 2d photon in the empty cavity or a mixture of both the te and tm components . fig .
[ fig:12 ] represents the dynamics of opo signal under the pump polarized along the tm cavity mode . in the range of high intensity of the signal ,
its degree of linear polarization ( dlp ) reaches 90% .
the polarization axis is @xmath135 rotated with respect to the pump ( what corresponds to negative dlp values ) , which is well reproduced in the calculations . on the other hand ,
if the pump is polarized in between the te and tm directions ( @xmath145 $ ] , fig . [ fig:13 ] )
, the signal dlp does not exceed 30% in the range of a strong signal ( fig .
[ fig:13]_e _ ) .
the shown dlp values were measured ( and calculated ) in the basis of directions parallel and orthogonal to pump polarization ; there is no marked dlp in the te - tm basis as well .
this effect is explained by the lifted degeneracy ( te - tm splitting ) of cavity modes which leads to the variation with time of the @xmath5 phase shift .
accordingly , it causes the misbalance of @xmath0 and @xmath1 components of intra - cavity field , i.e. a reduced dlp and non - zero dcp of the driven mode ( fig .
[ fig:13]_a , b _ ) , which is seen in both measured and calculated data . due to permanent modification of the direction of signal polarization , the eventual opo signal appears to be significantly depolarized ( the sum of squared dcp and dlp is markedly less than 1 , fig .
[ fig:13]_d , e _ ) .
thus , the dynamics of the opo signal confirms the effect of reservoir . under elliptic excitation ,
the @xmath5 polarization components tend to be levelled by the reservoir . by itself
, this effect could be assigned to a repulsion between cross - circularly polarized excitons ( @xmath146 ) . on the other hand ,
the axis of signal polarization is found inverted with respect to the pump , which restricts the sign of @xmath16 to be negative irrespectively of the reservoir induced blueshifts . since we have reproduced these effects within eqs .
( [ ef_p_r])([res ] ) , it is proven to provide a self - consistent description of the intra - cavity field polarization dynamics . we have shown that the semi - phenomenological eqs .
( [ ef_p_r])([res ] ) explain the observed polarization dynamics in a good qualitative agreement with the experimental data .
let us now consider the limitations of the suggested approach .
its most important shortcoming is the absence of a feedback between the driven polariton modes and the exciton reservoir , as long as the latter is introduced using the integral occupation number @xmath27 , neglecting the actual microscopic distribution of reservoir states . as a result , eqs .
( [ ef_p_r])([res ] ) do not describe , e.g. the establishing of equilibrium between coherent and incoherent exciton `` phases '' which would normally take place in a real system ; this process implies ( i ) the reverse transitions from the reservoir into the macro - occupied mode ( similar to bose - einstein condensation ) and ( ii ) dependence of the scattering rates ( @xmath29 and @xmath33 ) on the polariton density .
these shortcomings lead to several visible discrepancies between the measured and calculated data .
in particular , the superlinear drop in transmission intensity observed at the back front of exciting pulses , which is evident from comparison of left and right panels of fig .
[ fig:8 ] , is not reproduced by the present model .
note as well that the nonlinear scattering rate @xmath33 should depend on the energy of biexciton resonance which mediates the transitions of optically driven polaritons into the reservoir.@xcite accordingly , the actual relation between the multistability thresholds @xmath23 and @xmath22 would depend on the pump frequency as well as on the frequency detuning between photon and exciton modes . as a result
, we expect eqs .
( [ ef_p_r])([res ] ) to be qualitatively valid in the case of resonant excitation of the optically active cavity states . on the other hand , in the cases of comparably large pump wave numbers and/or non - resonant excitation conditions ,
when a much larger contribution of reservoir excitons is expected , the model is likely to become unsatisfactory .
in the present work we have studied the non - equilibrium transitions in a multistable system of cavity polaritons under resonant nanosecond - long excitation . using the spectrally broadened pulses , we have visualized the temporal correlations between the effective resonant energy , intensity , and optical polarization of the intra - cavity field which all undergo the strong changes on reaching the threshold pump power . in the vicinity of the threshold , the dynamics of such system
is strongly affected by the long - lived exciton reservoir ( excited due to polariton scattering ) which influences both the characteristic times of instability development and output signal polarization .
the temporal behavior of the intra - cavity field is found to be not described in the conventional model based on the gross - pitaevskii equations written for purely coherent macro - occupied polariton states .
most importantly , the observed phenomena can not be explained even qualitatively within a model with only the two exciton - exciton interaction constants ( @xmath14 ) allowed for .
to explain the experiments , we have proposed the model for the macro - occupied polariton states coupled with an exciton reservoir . in spite of some limitations , this model provides a self - consistent description of the observed intra - cavity field dynamics under both pulse and continuous wave excitation conditions .
authors thank m.s.skolnick for rendered samples , d.n.krizhanovskii for fruitful discussions , and a.v.larionov for assistance in the experiment .
this work was supported by russian foundation for basic research and the russian academy of sciences .
in attempt to reproduce the experimental pump spectrum , which is characterized by fwhm of about 0.8mev and , hence , has a lowered coherence , the modeled pump is ` broadened ' by means of the partial randomization of phases of different spectral harmonics .
namely , the pump amplitude is defined as follows : @xmath147 here , * @xmath148 are real - valued amplitudes of the @xmath5 polarization components ; * @xmath149 defines the direction of the polarization axis ; * @xmath150 is the central pump frequency ; * @xmath151 ; * @xmath152 defines the width of the pump spectrum , @xmath153 ; * @xmath154 is a set of random uniformly distributed numbers within the interval @xmath155 .
the spectral fwhm of the pump defined such a way approximately coincides with experimental one , and its coherence time is about 5 picoseconds . under these conditions
the system is inevitably stochastic .
hence , for each set of parameters eqs .
( [ ef_p_r])([res ] ) are solved a number of times ( 20100 ) with different @xmath156 until the averaged evolution of the signal intensity becomes independent on the number of realizations .
this corresponds to the collection of numerous signals , which allows us to reduce noises .
the degrees of polarization are then calculated using the _ averaged _ intensities . as a result ,
the transmission can be partially depolarized ( under elliptically polarized excitation ) due to small random differences in the individual pulses in much the same way as the measured signal .
33ifxundefined [ 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 link:\doibase 10.1103/physrevlett.69.3314 [ * * , ( ) ] @noop _ _ ( , , ) link:\doibase 10.1103/physreva.69.023809 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.236401 [ * * , ( ) ] \doibase http://dx.doi.org/10.1038/nmat2787 [ * * , ( ) ] , link:\doibase 10.1103/physrevb.63.041303 [ * * , ( ) ] link:\doibase 10.1103/physrevb.63.193305 [ * * , ( ) ] http://stacks.iop.org/0295-5075/67/i=6/a=997 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.126402 [ * * , ( ) ] \doibase http://dx.doi.org/10.1038/nphys1668 [ * * , ( ) ] , link:\doibase 10.1103/physrevb.77.115336 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.136401 [ * * , ( ) ] \doibase http://dx.doi.org/10.1038/nature05131 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.100.116401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.016402 [ * * , ( ) ] link:\doibase 10.1134/s1063776110050146 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.216402 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.216403 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.035328 [ * * , ( ) ] link:\doibase 10.1103/physrevb.72.075317 [ * * , ( ) ] link:\doibase 10.1103/physrevb.73.073303 [ * * , ( ) ] link:\doibase 10.1103/physrevb.61.2863 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.045319 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.245324 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.075301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.79.1341 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.116402 [ * * , ( ) ] @noop * * ( ) link:\doibase 10.1103/physrevlett.99.140402 [ * * , ( ) ] link:\doibase 10.1103/physrevb.66.045102 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.165308 [ * * , ( ) ] | the polarization dependence of nonequilibrium transitions in a multistable cavity - polariton system is studied under a nanosecond long resonant optical excitation at the normal and magic angle incidences with various polarizations of the pump beam .
the temporal correlations between the frequency , intensity , and optical polarization of the intra - cavity field , which all undergo sharp threshold - like changes due to the spin dependent interaction of cavity polaritons , are visualized .
the observed dynamics can not be reproduced within the conventional semi - classical model based on the gross - pitaevskii equations . to explain the observed phenomena , it is necessary to take into account the unpolarized exciton reservoir which brings on additional blueshift of bright excitons , equal in the @xmath0 and @xmath1 polarization components .
this model explains the effect of polarization instability under both pulsed and continuous wave resonant excitation conditions , consistently with the spin ring pattern formation that has recently been observed under gaussian shaped excitation . |
mesoscopic models provide coarse - grained descriptions of the dynamics of systems that neglect certain details at microscopic scales while retaining essential dynamical features at mesoscopic and macroscopic scales .
consequently , a convenient way to study of the dynamics of complex systems over a large range of interesting space and time scales is through the use of such models . in physical and biological systems we often encounter situations where mean field descriptions of reactions break down and molecular fluctuations play an important role in determining the character of the system s dynamics .
such effects are especially relevant for reactions taking place in nano - scale domains or biochemical reactions at the cellular level
. fluctuations also play a role in far - from - equilibrium systems near bifurcation points or when the system behaves chaotically since the system is especially susceptible to perturbations in such regimes .
@xcite mesoscopic models are able to capture the influence of such molecular fluctuations on the dynamics .
mesoscopic models are also useful for simulating the dynamics of macroscopic systems because they often provide stable particle - based simulation schemes and can be implemented in complex geometries . in this article
we consider a generalization of a mesoscopic multi - particle collision ( mpc ) ( or stochastic rotation ) model @xcite to a pattern - forming chemically reacting system .
we show how the multi - particle collision rule can be generalized to a multi - component system to yield different diffusion coefficients for the chemical species .
differences in diffusion coefficients can give rise to chemical instabilities which can not occur if the diffusion coefficients of all species are equal .
reactions are incorporated , also at a mesoscopic level , by combining a birth - death description of reactive events with multi - particle collisions .
the mesoscopic dynamics preserves all the basic conservation laws of the system and leads to the macroscopic evolution laws on long distance and time scales .
to illustrate the scheme , the reactive mpc dynamics is used to investigate the evolution and structure of a cubic autocatalytic front .
the cubic autoatalytic reaction is @xmath0 , where the autocatalyst @xmath1 consumes the fuel @xmath2 .
if one considers a two - dimensional rectangular domain ( or a thin rectangular slab in three dimensions ) with @xmath1 in left portion and @xmath2 in the right portion , a reaction front will propagate from left to right .
while the simulations presented in this paper are for cubic autocatalytic fronts , the manner in which the diffusion process is modelled to yield different diffusion coefficients for different chemical species and the way reactions are incorporated in the model presage extensions of the theory and applications to more complex far - from - equilibrium reactive systems .
the paper is organized as follows : in sec .
[ sec : mesomodel ] we sketch the basic elements of the multi - particle collision model and present its generalization to reactive systems where the chemical species can have different diffusion coefficients . section [ sec : front ] describes the simulation of cubic autocatalytic fronts and compares the results of the mesoscopic simulations with the predictions of reaction - diffusion equations .
the conclusions of the paper are given in sec .
[ sec : conc ] .
in multi - particle collision dynamics a system containing @xmath3 particles with continuous positions @xmath4 and velocities @xmath5 evolves through a sequence of free streaming and collision steps @xcite .
the collisions among the particles take place in the following way : the system is divided into cells and at time intervals @xmath6 each cell labelled by @xmath7 is assigned at random a rotation operator @xmath8 from some suitable set of rotation operators .
the center of mass velocity @xmath9 of the particles in cell @xmath7 is computed and the post - collision velocity @xmath10 of particle @xmath11 in the cell is determined by rotating its velocity , relative to the cell center of mass velocity , and adding the center of mass velocity to the result of this rotation : @xmath12 the velocity of every particle in cell @xmath7 is rotated by the same rotation operator but the rotation operator varies from cell to cell .
the dynamics then consists free streaming interspersed by these multi - particle collision events .
it has been shown that this dynamics conserves mass , momentum and energy and thus leads to the full set of navier - stokes equations on long distance and time scales @xcite .
the method has been applied to the study of a variety of systems @xcite including hydrodynamic flows @xcite , colloids @xcite , polymers @xcite , brownian motion @xcite and simple diffusion - influenced reaction dynamics @xcite . we present a generalization of this model that allows the dynamics of reaction - diffusion systems to be investigated .
this generalization entails several extensions of the mpc model .
in particular , a multi - component version of the mpc model @xcite must be constructed that accounts for reactions among the chemical species and allows for the possibility that the diffusion coefficients of the species differ .
a multi - component mpc dynamics that provides a simple way to control the diffusion coefficients of different chemical species can be constructed as follows .
suppose we have @xmath13 species labelled by an index @xmath14 . instead of applying the mpc operator to all particles in a cell
, we assume that multi - particle collision operators act to change the velocities of a fraction of the particles of species @xmath14 in a cell for @xmath15 .
more specifically , in each cell @xmath7 each particle of species @xmath14 is chosen with probability @xmath16 .
if @xmath17 is the velocity of a chosen particle @xmath11 of species @xmath14 and @xmath18 is the center of mass velocity of all chosen particles , the post - collision velocities of those particles that undergo collision are given by @xmath19 the post - collision velocities of the particles that do not take part in the multi - particle collision do not change .
the diffusion coefficients @xmath20 are functions of @xmath21 , which can be tuned to change the values of the diffusion coefficients . in order to investigate the range over which the diffusion coefficients can vary , we consider the self diffusion coefficient of a single species @xmath2 and change both the mean particle density @xmath22 and the fraction @xmath23 of particles that participate in the multi - particle collisions .
figure [ fig : d - vs - na ] plots @xmath24 , determined from the slope of the mean square displacement versus time , as a function density @xmath22 for different values of @xmath23 .
> from these results one sees that the self diffusion coefficient can be varied by about a factor of five by changing the values of @xmath23 at a fixed density .
the self diffusion coefficient for @xmath25 can be estimated in the boltzmann approximation where correlations are neglected .
the discrete - time green - kubo expression for the diffusion coefficient is @xcite @xmath26 where , without loss of generality , we have set @xmath27 .
taking into account the collision rule where , on average , a fraction @xmath23 of the particles undergo multi - particle collisions and fraction @xmath28 do not , we have @xmath29 where @xmath30 is the post - collision value of the velocity at time @xmath27 . assuming that higher order collision terms can be expressed in terms of the first collision so the series is geometric , we obtain @xmath31\ ; , \label{eq : diffusioncoeficient}\end{aligned}\ ] ] where @xmath32 was computed in ref . @xcite .
the comparison in fig .
[ fig : d - vs - na ] shows that this analytical expression ( solid lines ) accurately describes the simulation data .
the mesoscopic dynamics must also be generalized to allow for chemical reactions among the species .
our earlier study of diffusion - influenced reactions @xcite was restricted to a simple @xmath33 reaction that occurs when the @xmath2 or @xmath1 particles collide with catalytic spheres @xmath34 .
since we are now interested in reactions that occur among the mesoscopic particles , we instead use a birth - death stochastic law to describe the reactive events @xcite . here we restrict our considerations to the cubic autocatalytic reaction @xmath35 . independently in each cell we assume the reaction takes place with probability @xmath36 , where @xmath37 is the number of molecules of species @xmath14 in a cell .
the reactive dynamics in a cell is described by the markov chain , @xcite @xmath38 where @xmath39 , @xmath40 is the probability that there are @xmath41 particles in the cell at time @xmath42 and the transition matrix @xmath43 is given by @xmath44 the ( discrete time ) rate of change of the mean density of species @xmath14 , @xmath45 , is @xmath46 we assume that the mpc non - reactive collisions are sufficiently effective to maintain a local equilibrium poissonian distribution in the cells so that @xmath47 where the local poisson distribution is @xmath48 . if we insert the local poissonian approximation for @xmath40 in the right hand side of eq .
( [ eq : meanfield ] ) for @xmath49 we obtain the discrete - time mean - field rate law , @xmath50 a similar equation can be derived for species @xmath1 .
thus , the mass action rate law will describe the dynamics provided diffusion is sufficiently rapid compared to reaction so that a local poissonian distribution of particles is maintained during the evolution of the reactive system . in this limit the discrete - time rate law will closely approximate the continuous - time mass action rate law . after the reaction step , the particles free stream using the post - collision values of the velocities , taking into account the boundary conditions of the system .
once all the particles have been moved , the time advances one unit and the multi - particle collision and reaction steps are applied again .
this mesoscopic dynamics conserves the total mass , momentum and energy of the system .
in this section we show that the mesoscopic mpc model can be used simulate the dynamics of cubic autocatalytic fronts on macroscopic scales where comparisons with the predictions of reaction - diffusion equations can be made .
cubic autocatalytic fronts have been studied often in the context of a coupled pair of reaction - diffusion equations for the @xmath2 and @xmath1 species .
@xcite the particular focus of many of these studies was on the transverse front instability that occurs when the diffusion coefficient of the fuel is sufficiently larger than that of the autocatalyst : at a critical value of the diffusion coefficient ratio an instability will develop and the planar front will become nonplanar and exhibit complex dynamics . our investigations will be confined to a simpler case of a binary mixture undergoing the cubic autocatalytic reaction .
for such a reacting mixture the relevant macroscopic field variables are the total mass density @xmath51 , the local concentration @xmath52 , the center of mass velocity @xmath53 and the energy density @xmath54 . for the isothermal cubic autocatalytic reaction with no net fluid flow so that @xmath55 , and taking equal masses for the @xmath2 and @xmath1 species , the macroscopic equation for the number density of @xmath2 is @xcite , @xmath56 where @xmath57 is the mutual diffusion coefficient .
the equation for @xmath58 is not independent and follows from number conservation , @xmath59 .
the simulations of the reaction front using the mpc model were carried out in a rectangular prism with length @xmath60 along @xmath61 , width @xmath62 along @xmath63 and height @xmath64 units along @xmath65 .
the system was open along its length @xmath61 , periodic boundary conditions were imposed in the @xmath63-direction , and bounce - back reflection boundary conditions were imposed on the top and bottom of the prism along @xmath65 . in order to initiate a chemical front , @xmath2 particles were distributed uniformly in the right side of the prism , ( @xmath66 , @xmath67 ) , while @xmath1 particles were uniformly distributed in left side of the prism ( @xmath68 , @xmath69 ) .
the velocities were chosen from a maxwell - boltzmann distribution with reduced temperature @xmath70 .
starting from this initial condition a reaction front will develop as the autocatalyst @xmath1 consumes the fuel @xmath2 in the reaction .
the front will move with velocity @xmath71 and it is convenient to study the front dynamics in a frame moving with velocity @xmath71 . propagating fronts
are depicted in fig .
[ frontpic1 ] , which shows the concentration field at a given time instant .
the upper two panels plot the front for two values of the reaction rate constant @xmath72 and @xmath73 .
we see that for @xmath74 the front profile is much thicker than that for @xmath75 .
this dependence is in accord with predictions based on a reaction - diffusion description of the front as can be seen from the analysis given below .
the structure of these planar fronts can investigated quantitatively by studying the front dynamics in a frame moving with the front velocity , @xmath76 , and averaging the concentration profile over the width ( along y ) of the front , @xmath77 .
figure [ front ] plots @xmath78 for the two values of @xmath72 used in fig .
[ frontpic1 ] .
> from this figure we see that a well - defined propagating reaction front is obtained and the width of the front decreases as the reaction rate increases relative to the diffusion rate . the front shape and velocity can be determined from the reaction - diffusion equation . for a planar front propagating along the @xmath61-direction , in a frame moving with the front velocity ,
the reaction - diffusion equation ( [ eq : rd ] ) is , @xmath79 the front profile can be found analytically from the solution of this equation and is given by @xcite @xmath80 where the front speed is @xmath81 .
the profile for species @xmath1 can be found from the conservation condition @xmath59 .
figure [ front ] compares this analytical prediction with the simulation results of the mpc reaction - diffusion dynamics .
for @xmath73 the mutual diffusion coefficient @xmath57 is given by eq .
( [ eq : diffusioncoeficient ] ) .
there is good agreement between the simulation and analytical values for small @xmath72 where the conditions for the validity of the mean field approximation are satisfied . for larger @xmath72 values , such as @xmath75 in the right panel of the figure
we see that there are deviations from the mean field result . for this value of @xmath72 , the reaction is fast and there is a breakdown of the local poissonian equilibrium and a reaction - diffusion description is not applicable .
a similar breakdown is observed for very small @xmath72 , for example for @xmath82 , due to the fact that very few reactive events occur in the reaction zone of the front and fluctuations are important .
the front velocity was determined from the simulation data as a function of @xmath72 . in fig .
[ fig : c - vs - k ] we plot the front velocity @xmath71 versus @xmath72 and compare the simulation results with the prediction @xmath81 .
the front velocity agrees with the simulation results for @xmath83 , although the front profile deviates slightly from the predicted value for somewhat smaller values of @xmath72 ( @xmath84 ) .
more microscopic aspects of the front structure and dynamics that are captured by the mpc model are illustrated in the lower two panels of fig .
[ frontpic1 ] .
these figures plot snapshots of the front for @xmath74 , the same value of @xmath72 as in the top left panel of the figure , but for two different pairs of @xmath16 values , @xmath85 and @xmath86 .
comparison of the lower panels of the figures , and also with the upper left panel , shows that the structures of the interfacial zones are different . in the mpc dynamics employed here ,
the diffusion of the species depends on their density and @xmath16 .
since the density of the species changes significantly in the interfacial zone , it is likely that a concentration - dependent mutual diffusion coefficient is required to describe this structure .
the generalizations of the multi - particle collision model described here , and its extensions , allow one to study a variety of phenomena at the mesoscopic level .
in particular , the ability to simulate the dynamics of multi - component systems whose diffusion coefficients can be different means that diffusion - driven instabilities , such as the transverse cubic autocatalytic front instability considered in this paper , can be investigated . since the mesoscopic mpc model preserves the basic conservation laws in the system , to study such instabilities requires the presence of a third solvent species so that there are two independent diffusion coefficients in the system . the method could also be used to study reactive and non - reactive binary fluid
_ flows _ which also show interesting instabilities where fluctuations play a role near the onset of instabilities .
the cubic autocatalytic reaction is simply one example of a much broader class of reaction - diffusion systems that can be studied using reactive versions of the mesoscopic multi - particle collision dynamics . in particular
, more general reaction - diffusion dynamics in specific geometries relevant for the materials science and biological applications may be carried out .
the presence of flows can also be treated easily in this context . while we have focused primarily on parameter domains
where mean field approximations are largely applicable , one of the most interesting applications of the methodology introduced in this paper is to systems on mesoscales where particle numbers are small so that fluctuations play a crucial role in the dynamics and system geometry is important .
acknowledgements : this work was supported in part by a grant from the natural sciences and engineering research council of canada and in part by the grant c-1279 - 0402-b from consejo de desarrollo cientfico humanstico y tecnolgico of universidad de los andes .
t. ihle , d. m. kroll , phys .
e * 63 * , 020201 ( 2001 ) ; a. lamura , g. gompper , t. ihle , d. m. kroll , europhys .
lett . , * 56 * , 768 ( 2001 ) ; a. lamura , g. gompper , t. ihle , d. m. kroll : europhys .
lett . , * 56 * , 319 ( 2001 ) .
y. hashimoto , y. chen , h. ohashi , comp .
* 129 * , 56 ( 2000 ) ; y. inoue , y. chen , h. ohashi , colloids and surfaces a , * 201 * , 297 ( 2002 ) ; t. sakai , y. chen , h. ohashi , phys . rev .
e , * 65 * , 031503 ( 2002 ) . | a mesoscopic multi - particle collision model for fluid dynamics is generalized to incorporate the chemical reactions among species that may diffuse at different rates .
this generalization provides a means to simulate reaction - diffusion dynamics of complex reactive systems .
the method is illustrated by a study of cubic autocatalytic fronts .
the mesoscopic scheme is able to reproduce the results of reaction - diffusion descriptions under conditions where the mean field equations are valid .
the model is also able to incorporate the effects of molecular fluctuations on the reactive dynamics . |
the discovery of a family of circumstellar disks seen as dark silhouettes in projection against the orion nebula using the _ hubble space telescope _ provided strong confirmation of the disk paradigm of star formation ( odell 1993 ; odell & wen 1994 ; mccaughrean & odell 1996 [ mo96 ] ) .
the disks range in diameter from 501000au , and thus the @xmath250au ( 0.1 arcsec at 450pc ) resolution of the hst observations was sufficient to examine their structure directly at optical wavelengths .
an important finding was that the radial surface density profiles appear to be abruptly truncated at some outer radius , perhaps due to external effects from the surrounding region and dense cluster ( mo96 ) , and more detailed examination of this transition zone should lead to a greater understanding of the evolution of disks in harsh environments .
the discovery images were obtained over a relatively narrow wavelength range ( 50076585 ) , and further insight should be possible through hst observations at shorter and longer wavelengths . in the blue / near - uv ( @xmath220004000 ) ,
the spatial resolution approaches @xmath215au , while increased dust opacity at these wavelengths should also allow more tenuous structures to be traced to larger radii .
conversely , the considerable _ reduction _ in dust opacity at near - ir wavelengths should allow us to trace structures to smaller radii , albeit with commensurately poorer spatial resolution .
consequently , we are conducting follow - up hst studies from the near - uv to near - ir ( 0.32.5 ) , and in the present paper , we report preliminary near - ir observations using nicmos of one silhouette disk , orion 114 - 426 .
the largest of the sample at @xmath21000au diameter , this disk is seen near edge - on , and while the central star is not directly visible at optical wavelengths , its presence is betrayed by two polar nebulosities believed to be illuminated by it .
a comprehensive general observer program ( mccaughrean : go7367 ) studying the orion silhouette disks with nicmos , stis , and wfpc2 is being carried out during hst cycle 7 .
early release observations using nicmos were subsequently proposed by the instrument development team ( erickson : sm2/ero7114 ) for scientific verification and media use . due to this overlap , the ero data were reduced and analysed collaboratively , resulting in studies of 114 - 426 ( presented here ) and of the 182 - 413/183 - 419 field ( chen 1998 ) .
nicmos observations of the 114 - 426 field were obtained on 19 april 1997 during the servicing mission orbital verification following installation in the hst .
images were taken through broad - band , narrow - band , and polarimetric filters between 1 and 2.1 as summarized in table 1 .
data reduction combined standard ground - based near - ir imaging techniques with parts of the nicmos calibration pipeline .
multiple read - outs combined with multiple positions on the sky were used to reject cosmic - ray events ; electronic offsets were removed with on - orbit dark images ; quantum efficiency variations were removed with flat fields taken on - orbit where possible , otherwise from ground tests .
finally , mosaics were made registering the multiple images using stars or hst pointing information .
detailed photometric calibration was not attempted , but ground - based near - ir magnitudes for stars in the field were used to calibrate within @xmath3 . despite integration times significantly shorter than those planned for the go program
, important preliminary results were nevertheless obtained from the narrow - band imaging against the bright pa@xmath0 background at 1.87 , broad - band imaging at 1.1 and 1.6 , and the polarization imaging at 2.0 .
the three polarizer position images were combined to form a 2 continuum image , but due to remaining uncertainties in the correct analysis techniques for nicmos polarimetry and incomplete on - orbit polarization calibration , the polarization results themselves are deferred to a future paper .
the remaining narrow - band images did not provide useful additional information and are not further discussed .
the highest s / n images of the silhouettes obtained by mo96 were through a narrow - band h@xmath0 ( @xmath46565 ) filter , admitting the full emission line flux from the bright orion nebula region , while minimizing continuum emission from the central stars , or in the case of 114 - 426 , its polar lobes .
the brightest near - ir counterpart is the pa@xmath0 line at 1.87 , which can not be detected from the ground due to atmospheric absorption . for typical region ionization parameters ( 10@xmath5k , 10@xmath5@xmath6 , case b ) and @xmath7@xmath2 foreground to the nebula
, the detected photon flux at pa@xmath0 should be @xmath260% of that at h@xmath0 : the brightest equivalent line available to ground - based observers ( br@xmath8 at 2.16 ) would be a further factor of ten fainter ( osterbrock 1989 ) .
the pa@xmath0 1.87 image of 114 - 426 is shown in figure 1 with the h@xmath0 ( @xmath46565 ) image from mo96 .
the s / n in the p@xmath0 image is poor ( @xmath95:1 ) since the integration time was short ( 288 sec ) , and the nic1 image scale of 0.0432 arcsec / pixel over - resolved the 0.19 arcsec fwhm diffraction - limited resolution of the telescope at 1.87 .
nevertheless , the silhouette is clearly detected , allowing a preliminary measurement of its size .
the data were binned by a factor of two to better match the appropriate pixel size ( 2 pixels per fwhm ) and then averaged across the minor axis .
the resulting 1d major axis profile had high enough s / n to show the two ends of the disk as sharp dips separated by 1.8 arcsec .
as discussed in detail by mo96 , the apparent size and structure of a silhouette disk is a convolution of its real form with the instrumental point spread function , and following mo96 , we adjusted the parameters of a model edge - on disk convolved with a model hst+nicmos psf calculated using the tinytim software ( krist & hook 1997 ) until the major axis length was reproduced .
the resulting best - fit model disk has a major axis size of @xmath10800au at 1.87 , @xmath220% less than the 1012au measured at 0.6 ( mo96 ) .
the same procedure used on the 2 continuum image ( section [ sec : continuum ] ) yielded the same result to within 5% .
finally , we verified the overall procedure by degrading the high s / n [ oiii ] image from mo96 to the same spatial resolution and s / n as the pa@xmath0 image , then performing the same fitting process , before retrieving the correct size for the disk at optical wavelengths .
the optical continuum image of 114 - 426 showed faint polar lobes , interpreted as reflection nebulae of tenuous dust above and below the plane of the disk , illuminated by the otherwise unseen central star ( mo96 ) .
similar reflection nebulae are seen above and below the plane of an edge - on disk in the hh30 system ( burrows 1996 ) .
the wavelength dependent morphology and polarization structure of the lobes in 114 - 426 should allow us to probe the underlying form of the disk , the geometry of polar cavities , the grain size , and scattering function .
the near - ir broad - band continuum images , along with the f547 m image from mo96 , are shown in figure 2 in grayscale and contour forms . as the wavelength increases , three effects are seen .
first , the initially fainter se polar lobe increases in brightness until it equals then outshines the intensity of the nw lobe .
the peak intensities in the lobes ( after background subtraction ) are in the ratios 7.3:1 , 2.2:1 , 1.2:1 , and 0.85:1 for the nw : se lobes , at 0.57 , 1.1 , 1.6 , and 2.0 respectively .
second , the nebulae move closer together , as the reduced extinction allows us to probe closer to the disk midplane : the separations of the peak pixels in the two lobes are at 0.64 , 0.43 , 0.40 , and 0.32 arcsec at 0.57 , 1.1 , 1.6 , and 2.0 respectively .
third , the nebulae appear to flatten from conical to slab - like .
these features can be compared to model disks and envelopes ( lazareff , pudritz , & monin 1990 ; whitney & hartmann 1992 [ wh92 ] , 1993 [ wh93 ] ; fischer , henning , & yorke 1996 [ fhy96 ] ) . the general broad fan shape and increasing flatness of the nebulae are best reproduced by model sh of fhy96 , a thin , massless disk ( in comparison to the central star ) in keplerian rotation , with an envelope and broad polar cavities .
models with thick , massive disks have more polar material and result in images with too much elongation perpendicular to the disk .
the same is true of models with just a narrow , cylindrical polar hole rather than a broader , so - called `` streamline '' cavity ( wh93 ) .
since the central star in 114 - 426 is not seen , the disk must lie within a few degrees of edge - on , as the thin disk of the sh model does not occult the central star unless this condition is met ( fhy96 ) .
this degree of alignment is also argued for on the grounds that the two lobes have nearly equal brightness in the near - ir .
the asymmetry between the two lobes in the optical continuum ( figure 2a ) can probably explained by asymmetries in the outer , more diffuse parts of the envelope , perhaps due to external effects in the region .
the disk appears @xmath220% smaller at 1.87 than at 0.6 , and thus we are clearly resolving structure in its outer parts . in order to understand the implications of these observations ,
we need to examine the theoretical expectations . for a thin , massless , keplerian disk
that is hydrostatically supported and vertically isothermal ( shakura & sunyaev 1973 ; pringle 1981 ; lazareff 1990 ) , the density @xmath11 as a function of radius , @xmath12 , and height above the midplane , @xmath13 , is : @xmath14 \label{eq : one } \ ] ] where @xmath15 is the midplane density at the outer disk radius , @xmath16 , and @xmath17 is the disk scale - height : @xmath18 where @xmath19 is the scale - height at @xmath16 .
the disk surface density is then : @xmath20 for typical orion nebula dust grains ( r=5 ; cardelli , clayton , & mathis 1989 ) , the extinction at 1.87 is one sixth of that at 0.6 , and thus achieving the same effective optical depth requires six times higher column density at the longer wavelength . for a face - on disk with unity optical depth at 0.6 at 506au radius
, eq.[eq : three ] shows that the equivalent optical depth at 1.87 would occur at 46au , the disk would appear much smaller in the near - ir .
for an edge - on disk , the calculation is harder : we are probing the midplane density of the disk , not the surface density , and the line - of - sight through the disk at a given `` impact parameter '' ( the distance off - center ) integrates over different densities at different radii . assuming the disk is truncated at some outer radius
, we can integrate the total column density through the midplane as a function of the impact parameter , @xmath21 .
we have calculated profiles for disks in which the midplane density scales as @xmath22 , with @xmath23 and @xmath24 ( figure 3 ) .
these values were chosen since simple analytical integrals exist in each case , but they also closely correspond to plausible disk models : @xmath25 would yield a surface density independent of radius ; @xmath26 yields the canonical @xmath27 ; @xmath28 yields @xmath29 , a commonly assumed density law for more massive circumstellar disks ( adams , shu , & lada 1988 ) .
rough power laws can be fit to the inner section of each curve in fig.3 , with the column density increasing as @xmath30 , and @xmath31 , and @xmath32 corresponding to @xmath23 and @xmath24 respectively . in the canonical case of @xmath33 , a disk with unity optical depth at 506au at 0.6
would have the same optical depth at 84au at 1.87 .
thus , even though an edge - on disk will shrink less with wavelength than a face - on one , it is clear that for 114 - 426 , the observed decrease in size of only 20% ( @xmath34dex ) for an increase in density of six ( 0.78dex ) would require @xmath35 , and thus @xmath36 and @xmath37 , quite inconsistent with conventional understanding of disk structure .
the exception comes near the edge of a disk , where the integrated column density rises sharply as the path length increases most rapidly .
indeed , fig.3 shows that in the @xmath38 case , the density _ does _ increase by the required factor of six within 80% of the outer radius , as observed .
thus , the observed shift between the optical edge at 506au and the infrared edge at 400au could still be consistent with the midplane density following the canonical power law for a thin , massless disk , _ as long as _
the disk is physically truncated somewhere near the optical edge . as discussed by mo96
, it seems plausible that the disk may be truncated due to physical processes present in the orion nebula region and/or trapezium cluster , and it may be possible to ascertain which particular process is responsible from our future hst observations , which will trace the radial structure of 114 - 426 over a full order of magnitude in limiting column density . the mass of the 114 - 426 disk remains unknown : in particular , does it exceed the @xmath20.01 of the minimum mass solar nebula ( mmsn ) , is it massive enough to form a planetary system similar to our own ?
odell & wen ( 1994 ) and mo96 discuss how the mass of a silhouette disk can be roughly estimated , noting that the minimum intensity seen towards it is typically @xmath210% of that of the background region . assuming this to be background emission attenuated by the disk , a line - of - sight column density
is calculated for each point in the silhouette , and summing over the whole disk area , a total mass is estimated . however , virtually all of this `` disk light '' is an artifact , as the the background emission is blended into the disk by the instrumental psf ( mo96 ) .
thus the usefulness of the `` transmission technique '' is diminished , but it at least provides an absolute _ lower - limit _ mass , an important counterpoint to the results from millimeter interferometry which are , as yet , generally provide only _ upper - limit _
masses for disks in the trapezium cluster ( mundy , looney , & lada 1995 ; lada 1996 ) .
mo96 used the transmission technique to estimate lower - limit masses for the six silhouette disks based on their optical images .
the masses were small , typically several orders of magnitude less than the mmsn .
the most massive was 114 - 426 at 0.002 , but we must report here that an error was made in calculating the area and thus mass of 114 - 426 in mo96 : its minimum mass should be revised downwards to @xmath39 .
( the masses for the other disks given in table 2 of mo96 remain unaffected by this error ) . in principle , the transmission technique could be applied to our new near - ir images of 114 - 426 , resulting in a significantly _ higher _ estimate of the minimum mass . since
the disk is only marginally smaller at 1.87 , the reduced near - ir extinction should lead to a higher mass , on the order of the factor of six in @xmath40 .
however , the present images are not good enough to make such estimates : the pa@xmath0 image has very poor s / n , making a reliable measure of the disk / background flux ratio impossible , and while the continuum images have much higher s / n , the silhouette is contaminated by the reflection nebulae , making it difficult to measure the underlying disk / background flux ratio and the true silhouette area .
such calculations must wait until higher - quality pa@xmath0 data are available . in the interim , alternative approaches to estimating the disk mass
can be taken .
the non - detection of the central star in the pol 2 continuum image can be used to set a lower limit on the total column density through the disk midplane , depending on the intrinsic brightness of the star .
the integrated flux from the two reflection nebulae is = , which clearly only represents a fraction of the flux we would see from the central star if it had no disk .
some estimate of this fraction can be made from wh92 and wh93 , who tabulate @xmath41 ( the total flux observed as a fraction of the flux that would be seen if the star had no disk or envelope ) for a variety of models . while @xmath41 ranges from 0.0110% depending on the disk and envelope structure , inclination angle , grain properties , and total opacity , @xmath21% is typical for models inferred for 114 - 426 ( near edge - on flared disk , wh92 models 3,4 with @xmath42 , perhaps with a tenuous envelope with broad polar cavities , wh93 models 58 )
. the inferred intrinsic brightness of the central star is then @xmath2 , corresponding to a 1myr old , @xmath21.5 star at 450pc ( dantona & mazzitelli 1994 ) . since the star is undetected in the 2 image to a 5@xmath43 point source detection limit of = , the implied line - of - sight extinction towards it is @xmath44 at 2 , or @xmath45 . assuming the disk is exactly edge - on
, the corresponding column density through the disk midplane to the central star is @xmath46@xmath47 ( assuming 1@xmath48@xmath47 ) .
the disk outer radius , @xmath49au , is known from the data of mo96 .
the same data can be used to determine @xmath19 , the scale - height at @xmath16 : we fit a 1d average minor axis profile across the [ oiii ] image of 114 - 426 ( fig.4 of mo96 ) with a gaussian characterized by eq.[eq : two ] , yielding @xmath50au .
finally , we assume the disk inner radius , @xmath51au : since the surface density increases towards the center with a power @xmath52 , our results are not too sensitive to this assumption .
integrating eq.[eq : one ] ( with @xmath53 ) through the midplane , we calculate the midplane density at the outer edge , @xmath54@xmath6 .
then integrating eq.[eq : three ] between @xmath16 and @xmath55 , the total number of particles ( hi+2h@xmath56 ) is @xmath57 , yielding a mass estimate of @xmath58 g or @xmath59 . although it is clear that very little mass is required to render the central star invisible , the disk mass is likely to be significantly greater . using the analytical solution for the total column density seen through the midplane of an edge - on disk in the @xmath33 case ( section [ sec : structure ] ) , and assuming @xmath60@xmath6 at an outer radius of 506au
, we obtain values @xmath61@xmath47 or @xmath62 just inside the edge .
yet the disk must have a significantly higher column density there to be _ seen _ as an edge : the results of mo96 show that the attenuation at the edge of the disk is at least 85% , equivalent to @xmath63 , in turn implying @xmath64@xmath6 .
this is a plausible value if ( for example ) the location of the disk edge is set by pressure balance with the surrounding region : assuming a temperature and density for the latter of 10@xmath5k and 10@xmath5@xmath6 and an outer disk temperature of 10100k , densities at the disk edge of @xmath65@xmath66@xmath6 would be predicted .
inserting @xmath64@xmath6 into eq.[eq : three ] along with our other derived disk parameters yields a disk mass of @xmath67 g or @xmath1 .
while still considerably less than the mmsn , this is a lower limit , and the disk might be more massive .
however , our present simple calculations based on lower - limit extinction measurements alone can not be pushed much further . burrows ( 1996 )
have shown for hh30 that more realistic mass estimates can be made by fitting the structure of the polar reflection nebulae , in particular the width of the dark lane of obscuration between them , and such an approach will be taken with 114 - 426 when our higher s / n and resolution cycle 7 hst data are available .
ultimately , the most meaningful estimate of disk mass must come from measurements of optically thin tracers at millimeter wavelengths : such observations have been obtained for 114 - 426 and will be discussed in a future paper .
we have presented preliminary nicmos images of the orion 114 - 426 circumstellar disk .
the disk appears to be roughly 20% smaller at 1.87 than at at 0.6 , which we attribute to a standard radial power - law density distribution inside a heavily truncated edge to the disk at @xmath2500au radius .
the general wavelength - dependent morphology of the polar lobes is consistent with models of a thin , massless disk surrounded by a tenuous envelope with broad polar cavities , and their flux was used as an indirect way of estimating the mass of the central star to be 1.5 .
the non - detection of the star at 2 was then used to estimate the minimum total column density through the disk midplane , and thus a minimum disk mass of @xmath59 , assuming that it follows the standard form for a massless disk in keplerian rotation .
the hard edge of the disk was used to improve the lower - limit estimate under similar assumptions , resulting in a mass @xmath1 .
we thank the university of arizona , ball aerospace , rockwell , nasa gsfc , and the stsci for developing , managing , and commissioning nicmos .
mjm thanks abi saha for brokering the agreement between the nicmos idt ero and cycle 6/7 go teams , chris skinner and eddie bergeron for data reduction help , and matthew bate , olaf fischer , bob odell , peter schilke , and hans zinnecker for useful discussions .
finally , we would like to thank the referee , karl stapelfeldt , for his insightful comments .
adams , f. c. , shu , f. h. , & lada , c. j. 1988 , , * 326 * , 865 burrows , c. j. 1996 , , * 473 * , 437 cardelli , j. a. , clayton , g. c. & mathis , j. s. 1989 , , * 345 * , 245 chen , h. , bally , j. , odell , c. r. , mccaughrean , m. j. , thompson , r. i. , rieke , m. j. , schneider , g. , & young , e. t. 1998 , , in press dantona , f. , & mazzitelli , i. 1994 , , * 90 * , 467 fischer , o. , henning , t. , & yorke , h. 1996 , , * 308 * , 863 ( fhy96 ) krist , j. , & hook , r. 1997 , tinytim user manual version 4.3 , http://scivax.stsci.edu/@xmath2krist/tinytim.html lada , e. a. , dutrey , a. , guilloteau , s. , & mundy , l. g. 1996 , _ baas _ , * 189 * , # 53.01 lazareff , b. , pudritz , r. e. , & monin , j .- l . 1990 , , * 358 * , 170 mccaughrean , m. j. , & odell , c. r. 1996 , , * 111 * , 1977 mundy , l. g. , looney , l. w. , & lada , e. a. 1995 , , * 452 * , l137 odell , c. r. , wen , z. , & hu , x. 1993 , , * 410 * , 696 odell , c. r. , & wen , z. 1994 , , * 436 * , 194 osterbrock , d. e. 1989 , _ astrophysics of gaseous nebulae and active galactic nuclei , _ ( mill valley : university science books ) , p84 pringle , j. e. 1981 , , * 19 * , 137 shakura , n. i. , & sunyaev , r. a. 1973 , , * 24 * , 337 whitney , b. a. , & hartmann , l. 1992 , , * 395 * , 529 ( wh92 ) whitney , b. a. , & hartmann , l. 1993 , , * 402 * , 605 ( wh93 ) | we present the first high - resolution near - infrared images of the edge - on silhouette circumstellar disk , orion 114 - 426 , made using nicmos on the _ hubble space telescope_. images taken against the bright nebular background of the ionized hydrogen pa@xmath0 line at 1.87 show the major axis of the disk to be approximately 20% smaller than at 0.6 , from which we deduce the structure of the edge of the disk .
continuum images of diffuse polar lobes above and below the plane of the disk show a morphology and evolution with wavelength consistent with predictions for reflection nebulae in a diffuse envelope with large polar cavities , surrounding a thin , massless , keplerian disk , centered on an otherwise hidden central star .
we make use of our observations and reasonable assumptions about the underlying disk structure to show that the disk mass is at least 10 and plausibly @xmath1 . 6.25 in 8.9 in subject headings : accretion , accretion disks
circumstellar matter ism : individual ( orion nebula ) stars : formation , pre - main sequence infrared : ism : continuum , lines and bands |
the main motivation behind this work lay in developing approximate analytical tools for dealing with the long range physics of quantum chromodynamics .
the large-@xmath0 approximation is an attractive candidate for this project .
initiated by t hooft , ref .
@xcite , it was immediately applied to solve large-@xmath1 1 + 1 dimensional qcd , refs .
@xcite , @xcite , @xcite , @xcite and led to extremely fruitful phenomenological insights and models , refs .
@xcite , @xcite , @xcite .
the factorization property of correlators of gauge invariant operators suggested the idea of the master field the gauge field configuration which dominates the path integral in the large-@xmath0 limit , cf .
@xcite , @xcite , @xcite .
knowledge of the master field should allow to calculate any gauge invariant observable as if it were a classical non - fluctuating object .
a concrete example of a master field was provided by the exact solution of a model which described the dynamics ( classical as well as quantum mechanical ) of a single hermitian @xmath2 matrix , ref .
it was found that in the large-@xmath3 limit the boltzmann integral ( in the classical case ) and the ground state properties ( in the quantum case ) of this model were completely determined by an ensemble of matrices with a frozen distribution of eigenvalues given by a solution of a certain integral equation of the mean field type .
the `` angular '' orientations of the matrices in this ensemble , i.e. the unitary matrices @xmath4 in the decomposition @xmath5 with @xmath6 were completely free , i.e. distributed with equal probability according to the unbiased @xmath7 group theoretical haar measure .
this work and , in its sequel , ref .
@xcite gave rise to an intense interest in solutions of large-@xmath3 matrix models and made contact with the study of such models in other fields of physics , most notably nuclear physics , cf .
@xcite and , more recently , two - dimensional quantum gravity , cf . ref .
@xcite and quantum chaos , cf . ref .
the general interest in large-@xmath3 matrix models has been behind a steady progress in understanding solutions of various versions and generalizations of such models as reported e.g. in refs .
@xcite , @xcite , @xcite , @xcite , @xcite .
a new direction in this activity opened up when the physics community became acquainted with recent advances in the mathematical literature , associated with the work of voiculescu , ref .
@xcite , on non - commutative probability theory , ref .
the statistical mechanics ( classical or quantum mechanical ) of matrix ( non - commuting ) variables or ( gauge ) fields is an example of probability theory of non - commutative variables . a concept which turned out to be especially useful for the study of large-@xmath3 matrix and gauge theories , cf .
@xcite and @xcite , was the introduction in ref .
@xcite of the so - called _ free random variables_. the simplest examples of such variables occur in independent large-@xmath3 matrix models , i.e. in the classical statistical mechanics of several @xmath8 hermitian matrices @xmath9 , with the probability distribution ( the boltzmann factor ) given as @xmath10 with @xmath11 and @xmath12 polynomial functions .
typical `` observables '' in such a model are correlators @xmath13\}\prod_{k=1}^{d } d\mu[m_k ] \ ; , \ ] ] with the integration measure @xmath14 \equiv \prod_{\gamma=1}^n d(m_k)_{\gamma\gamma } \prod_{\gamma > \nu = 1}^n d \mbox{re } ( m_k)_{\gamma\nu } \;d \mbox{im } ( m_k)_{\gamma\nu } \ ; , \label{intmes}\ ] ] and @xmath15 \exp[-n^2 v_i(m_i ) ] \ ; .
$ ] were the @xmath16 regular commuting variables , one would call them independent since @xmath17 for any selection @xmath18 . for the matrix model
this is obviously not the case .
the probability distribution depends only upon the eigenvalues @xmath19 of the @xmath20 .
the expectation of products of the @xmath20 on the other hand involves also the `` angular '' variables @xmath21 in the decomposition @xmath22 .
these variables however are `` free '' , meaning that they are weighted with the unbiased haar @xmath7 measure . the integrals over the @xmath21
can therefore be evaluated in a universal manner independent of what the potentials @xmath23 are .
the remaining integrals over the @xmath19 can then be evaluated in the limit @xmath24 by a saddle point approximation as in ref .
the remarkable fact is that this procedure can be formulated in terms of very general rules relating any correlator of independent matrices to a linear combination of products of various individual moments of these matrices , cf .
these rules will be reviewed in section [ tracfr ] below .
any non - commutative variables distributed in such a manner that they satisfy these rules are termed `` free random variables '' .
@xcite and @xcite provide a review of the techniques which were developed to deal with such things as sums and products of free random variables , to define a certain operator algebra in a suitably defined fock space , and further applications , cf .
also @xcite , @xcite , @xcite .
it has been possible to prove that a master field for any matrix model can be constructed on the basis of the operator algebra developed for free models , ref .
@xcite , @xcite .
however , so far no dynamical principle has been advanced in order to carry out this construction explicitely for interacting matrix and field theories .
this paper presents an attempt to make progress in this direction by using a variational method based on the inequality @xmath25 .
this approach is fairly standard in statistical physics .
the new element will be to use the free matrix models as the trial variational space .
the details of the development are outlined in section [ secone ] below .
given a certain set of variables in terms of which the action is formulated , one can give a complete solution of the variational problem in the space of free matrix models by defining what will be called the free partner of the exact action of the model .
subsequently one observes however that the available variational freedom is even wider since one can first transform to a new set of variables , @xmath26 . in the new @xmath27
it is again possible to define the free partner of the action and give the general variational solution .
thus the `` art part '' of the proposed variational method is the choice of the optimal @xmath27 or in other words of the optimal parameters of the transformation @xmath26 .
unfortunately it is hard to give estimates of the accuracy of this variational approximation . as usual , one must gain experience from examples and comparison with exact solutions . in section [ secexa ] ,
a number of such examples are considered starting with classical interacting matrices and then going to quantum models of first one and then two interacting quantum matrices . only linear transformations
@xmath28 are considered and the variational results using free random variables are compared with exact monte carlo calculations and , where available , with analytic results . in almost all models considered ,
e.g. in the models of two interacting matrices with the action @xmath29 and various values of @xmath30 , of many interacting matrices with @xmath31 and in the quantum models of one and two interacting matrices , impressively good agreement is found with exact results .
some models are also presented for which the method ( with the additional limitation to the linear @xmath32 transformations ) failed completely .
presumably nonlinear transformations could improve the agreement but
the authors have not investigated this .
in this paper , matrix models described by partition functions of the type @xmath33 ) \prod_{k=1}^d d\mu[m_k ] \label{matint}\ ] ] are considered , with @xmath34 , @xmath35 , representing hermitian @xmath36 dimensional matrices , in the limit @xmath37 . a factor
@xmath38 has been pulled out in the exponent and in return , all matrix traces appearing in actions @xmath39 will be accompanied by a factor @xmath40 . note that the term `` action '' is used very loosely here to denote essentially any weight appearing in the exponent of the boltzmann factor in a partition function . only actions
@xmath41 $ ] which are invariant under the global unitary `` rotations '' @xmath42 will be considered .
one can further distinguish between `` classical '' and `` quantum '' matrix models . by classical
one means the models for which the above integral is just the ordinary integral with the integration measure given by ( [ intmes ] ) whereas in quantum models this is a path integral with @xmath20 depending on ( imaginary ) time @xmath43 , @xmath44 , and the measure @xmath45 = { \cal n}\prod_{0<\tau<\beta}\prod_{\gamma=1}^n d(m_k(\tau))_{\gamma\gamma } \prod_{\gamma > \nu = 1}^n d \mbox{re } ( m_k(\tau))_{\gamma\nu } \;d \mbox{im } ( m_k(\tau))_{\gamma \nu } \ ; .\ ] ] here @xmath46 is the inverse temperature and @xmath47 is a normalization factor which in principle depends on @xmath48 , but will not be important in the following . in both the classical and quantum cases one deals with integrals over many matrix variables and the real distinction will lie in the form of the action . in the classical case the action ( the classical potential ) is a _
real function _ usually represented as a sum of monomials in powers of the @xmath20 , @xmath49 with real coefficients @xmath50 . in the quantum case
the action will be a sum of kinetic and potential terms @xmath51 ^ 2 + \sum_{0<\tau<\beta } \delta\tau v(m_1(\tau ) , \ldots , m_d(\tau ) ) \ ; , \ ] ] with @xmath52 as in ( [ clact ] ) . as was reviewed in the introduction , only a limited number of matrix models are amenable to analytical solution so that various methods of approximation are called for . here , the powers of the variational method , combined with the large-@xmath3 saddle point evaluation , will be explored .
as is common with statistical integrals , the variational approximation will be developed on the basis of the inequality , cf .
@xcite , @xmath53 which expresses the convexity property of the exponential function the average value of the exponential of a random variable @xmath54 is greater than or equal to the exponential of the average provided @xmath54 is real and the weights used in the averaging are positive .
if one can find an action @xmath55 $ ] which is simple enough to allow the calculation of integrals like @xmath56e^{-n^2 s_0[m]}\;\;\;\ ; { \rm and } \;\;\;\ ; \int \prod_k d\mu[m_k ] k(m_1 , \ldots , m_d)e^{-n^2 s_0[m ] } \;\ ] ] for some simple functions @xmath57 then one can approximate the integral ( [ matint ] ) as follows
. one can represent @xmath58e^{-n^2 s[m ] } = \int \prod_k d\mu[m_k ] e^{-n^2 ( s[m ] - s_0[m ] ) } e^{-n^2 s_0[m ] } = e^{-f_0 } \big < e^{-n^2 ( s - s_0 ) } \big>_{s_0 } \ ; .\ ] ] where @xmath59 means @xmath60 .
using the inequality ( [ ineq ] ) one obtains @xmath61 } \;,\ ] ] or @xmath62 if @xmath63 depends on adjustable parameters , one can find their optimal values by minimizing the right hand side of the above inequality .
the resulting @xmath63 can then be adopted as an approximation for @xmath39 with which one should calculate the relevant correlators . as always with variational methods ,
it is hard to assess the accuracy of the approximation . here
, it will be tested in different classes of matrix models , using the so - called free matrix models ( cf .
below ) as the variational space , by comparing ( where available ) with exact analytical results or with monte carlo simulations .
freeness is a concept recently developed in the mathematical literature associated with noncommuting random variables @xcite .
while freeness is abstractly defined without reference to a specific realization of the random variables , in the context discussed here , namely in relation to large-@xmath3 hermitian multi - matrix models , it essentially means that the weight @xmath64 according to which the matrices are distributed contains no bias with respect to the relative @xmath65 `` orientations '' of the matrices . in terms of the decomposition @xmath66 with diagonal @xmath67 and unitary @xmath68
this means that the action @xmath39 in the free models depends only on the @xmath67 .
consequently all the angular integrals in any given correlation function are weighted only with the group theoretical @xmath7 haar measure and give universal results independent of what the action is .
the remaining integrals over the @xmath67 in the large-@xmath3 limit are controlled by the saddle point of the effective action which includes the contribution from the vandermonde determinants for each @xmath69 , cf .
examples below . as a consequence of these properties
all correlation functions for free multi
matrix models are systematically accessible in terms of the _ moments of individual matrices _ through the recently developed techniques , as will be elaborated upon below , cf .
also refs .
@xcite , @xcite , @xcite .
a family of noncommuting random variables @xmath20 is called free if and only if @xmath70 where @xmath71 ( and @xmath72 is identified with @xmath73 ) .
note that the bracket notation introduced here following @xcite means taking the expectation value of the trace of the expression inside , divided by @xmath3 . in order to evaluate a general correlation function @xmath74
, one can use @xmath75 and multiply out the terms in the expectation value .
this recursively determines correlation functions of a certain order in terms of lower - order correlation functions until one arrives at an expression containing only _ the moments of the individual matrices_. for two free matrices @xmath76 and @xmath77 this algorithm yields , e.g. , @xmath78 building on this information , it has been possible , e.g. , to determine , @xcite , how the eigenvalue distributions of matrices which are free with respect to one another are additively and multiplicatively convoluted .
freeness in matrix models is a slightly more general concept than what is usually meant by the term `` independent matrix model '' , in which the potential governing the distribution of the matrices is a sum of terms for each individual matrix .
independent matrix models are certainly free ; however , so is e.g. a two - matrix model with potential @xmath79 more generally , the potential of a free matrix model can be any function of traces over functions of the individual matrices .
such more general free matrix models are almost as easy to solve as the independent ones .
this is best illustrated by an example .
consider the model described by ( [ freex ] ) .
after the transformation of variables @xmath80 , with diagonal @xmath19 and unitary @xmath21 , the potential @xmath81 depends only on the @xmath19 .
moreover , the transformation introduces the vandermonde jacobians which , when exponentiated , modify the potential into @xmath82 \ ; . \nonumber\end{aligned}\ ] ] in the large-@xmath3 limit , the distributions @xmath83 of the eigenvalues are `` frozen '' at a saddle point , so that the moments of the matrices have definite values @xmath84 this means that the saddle point eigenvalue distributions @xmath83 can be solved for as if they were controlled by the potentials @xmath85 for such quadratic potentials , one obtains the well - known semicircular distributions , cf .
@xcite,@xcite @xmath86 the constants @xmath87 and @xmath88 can now be determined by the self - consistency conditions @xmath89 including such self - consistency conditions on selected moments of the matrices is the only additional step needed compared with solving independent matrix models .
knowledge of the eigenvalue distributions @xmath90 is sufficient to calculate any correlation function since the latter can be reduced using eq .
( [ algo ] ) to a sum of products of individual moments , i.e. to the integrals @xmath91 .
it was argued e.g. in refs .
@xcite,@xcite that a free behaviour of the matrix variables representing physical degrees of freedom may capture correctly the main features of the dynamics in the large-@xmath3 limit of nonabelian gauge theories .
this hope and the wealth of available analytical techniques presented above render the space of free matrix models a good candidate for variational calculations .
the main idea of the present approach is to choose the optimal action @xmath63 in the variational principle ( [ minp ] ) from among all actions which leave the matrices in terms of which @xmath39 is given , free with respect to one another . in making a variational approximation based on free matrix models one must first decide which combinations of the original matrix variables @xmath20 will be assumed to be free , i.e. in which variables to formulate the exact action @xmath39 .
one should realize that e.g. free @xmath76 , @xmath77 do not imply free @xmath92 , @xmath93 nor vice versa ( more about this below ) .
the second step is to find the best trial action @xmath63 which is free in the variables one has settled for .
in this subsection , the solution of the second step is addressed .
in other words , consider @xmath94 to be _ free in terms of the original _ @xmath20 . under this limitation , it is possible to give a _ general recipe _ for finding the trial @xmath63 which minimizes the right hand side of ( [ minp ] ) .
for this it is useful to introduce the following algorithm . given a trace of any function of matrices @xmath20 , @xmath95 , write down its expectation value in terms of the moments of the individual matrices under the assumption that the matrices are free with respect to one another .
this is easily accomplished for any multinomial expression using eq .
( [ algo ] ) .
now rewrite the resulting expression with @xmath96 replaced by tr@xmath97 .
this defines what will be called the `` free partner '' @xmath98 of @xmath99 ( the `` liberated '' @xmath99 , so to speak ) .
for instance , the free partner of @xmath100 is @xmath101 cf .
( [ m1m2e ] ) .
evidently , the original function and its free partner have the same expectation values when evaluated in any free matrix model for the @xmath16 ( note that the expectation values of the products of traces appearing in the free partner factorize at large @xmath3 ) . in general , the two expectation values will of course differ . furthermore ,
if the free partner @xmath102 of any action @xmath39 is used as a trial action , it will generate a free matrix model ; hence the terminology . with the help of the above definitions , one can give the general solution of the minimization principle ( [ minp ] ) in the space of matrix models which are free in terms of the variables @xmath16 .
the crucial observation is that , when @xmath94 is to be chosen from among all free actions , the original action @xmath39 and its free partner @xmath103 lead to the same minimization problem , i.e. identical right hand sides of eq .
( [ minp ] ) , since @xmath104 by construction for any free @xmath63 , as mentioned above . however , the minimization problem for @xmath103 is trivial to solve : since @xmath103 itself describes a free matrix model , the best approximation to @xmath103 in the space of free actions is @xmath105 itself . since the approximation to @xmath39
is governed by the same minimization problem , @xmath105 is at the same time the best free approximation to @xmath39 .
this result provides a general solution of the minimization problem in the space of matrix models which are free _ in terms of the original variables _ @xmath16 .
note that the solution @xmath106 of the minimization problem defines a type of mean field approximation to the exact action @xmath39 .
interaction terms in @xmath39 which are sensitive to angular correlations between the different matrices are replaced by terms in which each matrix couples only to selected moments of the other matrices , i.e. to some mean properties independent of the angular orientations .
this is entirely analogous to the standard development of the hartree or hartree - fock mean field theories from a variational principle using the space of ( properly symmetrized or antisymmetrized ) product states .
there , in second - quantized language , the combinations in which the mean field enters the single - particle hamiltonian are determined by wick s theorem ; here , in complete analogy , the combinations in which the moments enter the free partner are determined by the free random variable axioms as encoded in eq .
( [ algo ] ) .
in fact , ( [ algo ] ) is nothing but wick s theorem generalized from the usual bosonic and fermionic cases to the case of objects obeying the cuntz algebra ( this is the algebra obeyed by the fock space operator representation of free variables , cf .
@xcite ) .
the idea of the present variational approach was stated as choosing the optimal action @xmath94 from among all matrix model actions which leave the matrices in terms of which @xmath39 is given , free with respect to one another . in this
there exists an additional freedom of choice which will now be exploited , namely the choice of matrix variables in terms of which @xmath39 is given , i.e. in terms of which one develops the mean field approximation by constructing the free partner @xmath106 . up to now
, only a fixed set of variables was considered the original @xmath16 . however ,
if one rewrites the action @xmath39 in terms of other variables , @xmath107 and again looks for the optimal free action , now in terms of the new variables @xmath108 , approximating @xmath109 , one may find a better approximation than the one found using the variables @xmath16 .
it should be emphasized that free @xmath108 in general do not imply free @xmath16 nor vice versa .
therefore , the accessible variational space using free matrix models is in fact much larger than was apparent in the last section . by allowing different sets of variables in which to formulate the problem to be approximated
, one can even include into the variation models in which the original matrices @xmath16 are not free with respect to one another .
to formalize this idea , it is necessary to reexamine the derivation of the variational principle ( [ minp ] ) . carrying out a variable transformation @xmath110
, one has @xmath111
e^{-n^2 \tilde{s } [ b_j ] + \ln j [ b_j ] } \ge \exp \left [ - ( f_0 + n^2 \langle \tilde{s } - ( \ln j)/n^2 -s_0 \rangle_{s_0 } ) \right]\ ] ] i.e. @xmath112 with the jacobian ( which must be non - vanishing to have a legitimate change of variables ) @xmath113 } { d\mu [ b_j ] } \right|\ ] ] note that this is the determinant of a @xmath114 matrix , where @xmath115 denotes the number of matrix variables .
now , one can use the theorem of the previous section with the immediate result that the optimal free choice of @xmath94 in the new variables is the free partner of @xmath116 .
thus , while for any given set of variables , the optimal @xmath94 is known , one may now try to further improve the approximation ( i.e. diminish the right hand side of ( [ linv ] ) ) by trying different sets of variables leading to different @xmath109 and @xmath117 . in general
, this is technically difficult , since @xmath118 is not a multinomial expression , for which the algorithm of finding the free partner is easiest to carry out .
therefore , in this paper the treatment will be specialized to linear transformations , @xmath119 then @xmath120 is independent of the matrix variables @xmath108 .
this greatly simplifies the treatment of ( [ linv ] ) ; the jacobian does not play any role in the determination of the optimal free action @xmath94 for given new variables @xmath108 and thus the best free approximation @xmath94 to the given @xmath109 is the free partner @xmath121 of @xmath109 . using furthermore that @xmath122 , one arrives at the residual minimization problem @xmath123 where @xmath124 is the free energy associated with @xmath121 .
it now remains to try different choices of @xmath125 , leading to different @xmath109 , @xmath121 , and finally @xmath124 , such as to minimize the right hand side of ( [ linvs ] ) . before actually carrying out the minimization procedure ,
it is convenient to clarify a technical point concerning the actual evaluation of @xmath126 given @xmath127 .
the best way to achieve this seems to be the following . by subtracting a constant @xmath128 , independent of @xmath129 , from both sides of ( [ linvs ] ) ,
one arrives at the equivalent problem of minimizing @xmath130 . for @xmath128
, one can choose a reference free energy , preferably of an independent matrix model with the same number of variables as @xmath39 and whose potential @xmath131 is a polynomial of the same degree as @xmath39 .
this specification is not necessary , but convenient from the point of view of keeping the expressions one deals with in practice as regular as possible . in many applications , @xmath132 is adequate
. then one can write @xmath133 e^{-n^2 \tilde{s}_f[b ] } + \ln \int \prod_j d\mu[b_j ] e^{-n^2 v_{ref } [ b ] } \nonumber\\ & = & \left .
- \ln \int \prod_j d\mu[b_j ] e^{-n^2 ( \alpha \tilde{s}_f[b ] + ( 1-\alpha ) v_{ref } [ b ] ) } \right|^{\alpha = 1 } _ { \alpha = 0 } \nonumber\\ & = & -\int_{0}^{1 } d\alpha \ , \frac{\partial}{\partial \alpha}\ln \int\prod_j d\mu[b_j ] e^{-n^2 \tilde{s}^{\prime}_f [ b,\alpha ] } \\ & = & n^2 \int_{0}^{1 } d\alpha \ , \langle \tilde{s}_f [ b ] - v_{ref } [ b ] \rangle_{\tilde{s}^{\prime}_f } \label{linmod}\end{aligned}\ ] ] where @xmath134 = \alpha \tilde{s}_f [ b ] + ( 1-\alpha ) v_{ref } [ b ] \label{dpridef}\ ] ] in practice , the matrix model described by @xmath135 is only insignificantly harder to solve than the one described by @xmath121 ; on the other hand , expectation values like the ones contained in ( [ linmod ] ) are very easy to evaluate and thus constitute a convenient way of evaluating the right hand side of the variational principle ( [ linvs ] ) .
thus , in applications , the best formulation of the variational problem seems to be to demand minimization of @xmath136 - v_{ref } [ b ] \rangle_{\tilde{s}^{\prime}_f } -(\ln j)/n^2 \label{modprin}\ ] ] as a function of the transformation matrix @xmath125 , where , to recapitulate , @xmath135 is given by ( [ dpridef ] ) and @xmath121 is the free partner of @xmath109 , which in turn arises from performing linear transformations on the original matrix variables @xmath16 , i.e. @xmath137 .
furthermore , @xmath120 , and @xmath138 $ ] is a conveniently chosen reference potential independent of @xmath125 generating an independent matrix model . as a last remark ,
if one has succeeded in minimizing ( [ modprin ] ) by varying the transformation @xmath125 , one will usually be interested in transforming back to the original variables ; e.g. , given the eigenvalue distributions of the matrices @xmath108 , one may wish to extract the eigenvalue distributions of the matrices @xmath16 . here
, the additive convolution techniques for free variables developed in @xcite find fruitful application , as will become clear in the examples to be treated further below .
the simplest type of interacting matrix model involves just two classical matrices ; this provides a first testing ground for the variational approach developed in the previous sections . without any deeper motivation ,
the actions @xmath139 and @xmath140 will be considered .
the variational approximation will be compared mainly with monte carlo experiments , although there also already exist exact analytical solutions to some simple models of this type , see e.g. @xcite in the case of @xmath141 .
the models considered in this section are invariant under the transformation @xmath142 , and throughout the treatment it will be assumed for simplicity that , at the large-@xmath3 saddle point , @xmath143 . of course , it constitutes no problem in principle to work without this assumption and to verify it from the solution ; this would just unnecessarily complicate the notation . in more complicated models
, it may become an interesting issue whether such a reflection symmetry can be broken spontaneously .
consider now the model described by @xmath144 .
this model will be treated here in some detail to exhibit the new techniques in practice .
later examples will receive a more cursory treatment . allowing for a general linear transformation of variables , @xmath145 and inserting into @xmath144 to arrive at @xmath146 , one obtains for the free partner of the latter @xmath147 with the notations @xmath148 here the reflection symmetry
@xmath149 has been assumed to carry over from @xmath150 .
choosing furthermore the reference action @xmath151 one has ( cf .
( [ dpridef ] ) ) @xmath152 treating the last term in analogy to ( [ freex ] ) in conjunction with ( [ freeea ] ) and ( [ freeeb ] ) , one has to solve one - matrix problems for @xmath153 and @xmath154 with the moments @xmath155 to be determined self - consistently .
the explicit solution of the one - matrix model with arbitrary symmetric quartic potential is listed in appendix [ appa ] ; the expressions for the second moments yield equations for the @xmath155 which can be solved numerically for given @xmath156 and @xmath125 . then @xmath157 can be evaluated , again using the expressions for the moments given in appendix [ appa ] .
finally , evaluating the integral over @xmath156 leads to the free energy , cf .
( [ modprin ] ) , with the additional jacobian piece @xmath158 . minimizing the free energy numerically as a function of the @xmath125 yields the following result , cf . also fig .
[ fig0 ] : for sufficiently small @xmath159 , the original variables @xmath76 and @xmath77 are the optimal ones for a free approximation , whereas for large @xmath30 , it becomes favorable to switch variables by substituting @xmath160 and @xmath161 .
this change in behavior is of course simply induced by the competition between the different terms in the action . for sufficiently low @xmath30 , the quartic part , which is already independent in the original variables @xmath76
, @xmath77 , dominates . for larger @xmath30 ,
it becomes favorable to accomodate the interaction term as much as possible .
the competition is essentially linear in the sense that one optimizes the treatment of either one or the other part of the action and the switch between the optimal choices of variables @xmath76 , @xmath77 or @xmath162 , @xmath163 happens suddenly at @xmath164 ; the optimal choice does not change continuously with @xmath30 .
the authors have considered the possibility that this transition in the variational approximation may signal a large-@xmath3 phase transition also in the exact solution of the model , visible e.g. as a discontinuity of the derivative of the exact free energy as a function of the coupling constant @xmath30
. however , the monte carlo results do not show such a transition the expectation of @xmath165 displays a smooth behaviour near @xmath166 .
thus , the transition in the variational calculation seems to merely reflect a crossover between two physically different regimes rather than a genuine large-@xmath3 phase transition .
a higher order phase transition can however not be ruled out on the basis of the data taken by the authors .
having ascertained the optimal choice of variables , one can proceed to give the corresponding approximations to the eigenvalue distributions . for @xmath167 , the original variables @xmath168 , @xmath77 are best
; the free partner of @xmath144 is @xmath169 using the formulae of appendix [ appa ] to solve these models , one obtains ( assuming the eigenvalue distributions of @xmath76 and @xmath77 to be identical ) the consistency conditions @xmath170\ ] ] solved by @xmath171 for @xmath172 and @xmath173 for @xmath174 ( the relevant solution can be picked out by using positivity of @xmath175 , etc . ) .
the corresponding eigenvalue distributions are given by ( cf .
appendix [ appa ] ) @xmath176 these distributions are plotted in figs . [ fig1 ] and [ fig2 ] , and compared with monte carlo results for @xmath177 matrices . in the case of @xmath172 , the dependence of the monte carlo results on the size of the matrices
is exhibited in fig .
[ fig3 ] . in this , as in all other cases tested , @xmath178 already seems to embody the large-@xmath3 asymptotic bulk behavior rather well , up to the characteristic oscillations in the eigenvalue density induced by the eigenvalue repulsion . in this respect ,
no deviation was observed from the well - known finite-@xmath3 phenomena observed in one - matrix models . for @xmath179 ,
on the other hand , one considers the best free approximation after the variable substitution @xmath160 and @xmath161 , i.e. @xmath180 here , the consistency condition becomes @xmath181 \label{s1fly}\ ] ] solved by @xmath182 .
( [ s1fjk ] ) and ( [ s1fly ] ) imply that the @xmath183 are both governed by the quartic potential @xmath184 , and the corresponding eigenvalue distribution is ( cf .
appendix [ appa ] ) @xmath185 in order to obtain the eigenvalue distributions of the original variables @xmath76 and @xmath77 from this , one must use the free additive convolution techniques derived in @xcite . in appendix
[ appb ] , the convolution of two identical distributions governed by an arbitrary symmetric quartic potential is discussed .
the resulting eigenvalue distribution @xmath186 is plotted in fig . [ fig4 ] along with the corresponding monte carlo results for @xmath178 . also , the comparison with the free approximation in the original variables @xmath168 , @xmath77 is plotted .
evidently , allowing the choice of free variables to vary leads to a vastly improved approximation in this case . in exactly the same vein as the case
@xmath179 one can treat the extreme case ( corresponding , up to a rescaling , to @xmath187 ) @xmath188 here , in the absence of any piece in the action which is free in the original variables and which even in the case of quite strong coupling @xmath179 ( see above ) kept the behavior of the model reasonably regular , the disagreement between the variational calculation and the exact result is rather catastrophic , cf .
this is not hard to understand ; in the case of the action @xmath189 , the variables @xmath76 , @xmath77 can use configurations such as @xmath190 where @xmath191 , @xmath192 are roughly @xmath193 matrices , to preserve a low value of @xmath189 while realizing very large eigenvalues ( together with a concentration of very small ones ) .
this is indeed what is seen in fig .
dominance of configurations such as ( [ redco ] ) implies a strong angular correlation between the matrices ; if one rotates the above matrices with respect to one another , e.g. such that the eigenvalues are arbitrarily reordered , @xmath189 will in general take a very large value .
such angular correlations are completely lost in the free partner @xmath194 and evidently even the best choice of linearly transformed variables @xmath195 and @xmath196 , though incorporating some angular correlations between @xmath76 and @xmath77 , can not do much to alleviate this problem .
it is however possible that a nonlinear change of variables exists for which the difficulty will be satisfactorily removed .
the authors did not explore this direction .
the dismal failure of the variational approximation in this case thus does not come as a complete surprise .
rather , it is gratifying to see how well the variational approximation still works in the previous , not quite as pathological , example described by @xmath144 with the strong coupling of @xmath197 .
one striking feature of the monte carlo eigenvalue distribution in the case of @xmath198 is the absence of the characteristic finite-@xmath3 oscillations observed in all other cases .
this can be understood as follows : usually , the potential confines the eigenvalues to a compact domain and , due to the eigenvalue repulsion , they tend to be equidistant
. this ( fluctuating ) lattice has a favored equilibrium position , leading to the oscillations in the eigenvalue density .
however , when the action allows the eigenvalues to spread over the entire real axis , the correlations induced by the eigenvalue repulsion become insignificant and the oscillations in the eigenvalue density disappear .
apart from plotting the eigenvalue distributions , which essentially contain the information about all the moments of the individual matrices , one can test mixed correlators for freeness properties . in particular , if the matrices @xmath76 and @xmath77 are free with respect to each other , then there are e.g. the following relations between correlators , following from the axioms of freeness : @xmath199 in the monte carlo calculations , both sides of these equations were sampled in order to give a measure for the deviation from freeness ( i.e. strength of angular correlations ) contained in the exact models . the results
are tabulated in table 1 for different values of @xmath30 in @xmath144 .
note that for @xmath189 , the moments of the exact distribution diverge due to the evidently too slow fall - off of the eigenvalue distribution for large eigenvalues , making such a comparison impossible . ' '' ''
@xmath30 & @xmath200 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 + -1 & .483 & .377 & .254 & .233 & .171 & .142 & .150 & .121 + 1 & .336 & .197 & .106 & .113 & .0339 & .0387 & .0272 & .0317 + 4 & .267 & .133 & .0585 & .0713 & .0116 & .0177 & .0085 & .0139 + table 1 : mixed correlators , eqs . ( [ mixmf ] ) , for the model @xmath208 compared to the predictions for free @xmath20 one observes how , for growing @xmath30 , the interaction term introduces stronger angular correlations , evident in the stronger deviations from free predictions .
now , for sufficiently large @xmath30 , e.g. @xmath179 , the variational principle asserts that a better approximation is obtained with a new set of mutually free variables @xmath183 . using this new set of free variables , of course also the predictions for mixed correlators , derived in ( [ mixmf ] ) for free @xmath209 , change .
this shall be illustrated here for the correlator @xmath210 .
inserting @xmath211 and @xmath212 , using the axioms of freeness on @xmath153 and @xmath154 together with @xmath213 , and then substituting back @xmath214 and @xmath215 , one obtains @xmath216 where in the last line , identical distributions for the @xmath183 were assumed . putting these relations together yields @xmath210 in terms of the individual moments , @xmath217 in the case @xmath179 , the right hand side , using the individual moments from the monte carlo experiment , takes the value .0617 , which indeed is closer to the exact value for @xmath218 than the prediction using free variables @xmath16 quoted in table 1 .
consider now the model described by @xmath219 after performing a general linear transformation as in ( [ glt ] ) , one arrives at the free partner @xmath220 evidently , from the point of view of the variational approximation , this model is slightly simpler than the one considered above . since the noninteracting part of @xmath141 is invariant under linear transformations of the variables and subsequent construction of the free partner , up to trivial variable rescalings ( note also that , as before , @xmath221 was used in deriving ( [ ts2f ] ) ) , the choice of variables will always be such as to best accomodate the interacting term , for any coupling @xmath30 . carrying out the calculation of the free energy in complete analogy to the calculation for @xmath144 ,
this turns out to be the choice @xmath211 , @xmath161 , corresponding to the free partner @xmath222 the corresponding consistency condition is @xmath223 \label{ts2foptc}\ ] ] from the resulting eigenvalue distributions of the matrices @xmath183 one again obtains the eigenvalue distributions of the original variables @xmath16 by additive convolution .
the result for @xmath224 , leading to @xmath225 in ( [ ts2foptc ] ) , is displayed in fig .
[ fig6 ] , compared with the monte carlo result for @xmath226 .
also , various correlators are tabulated in table 2 . ' '' ''
@xmath30 & @xmath227 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 + 2/5 & .53 & .56 & .30 & .28 & .37 & .32 & .28 & .24 + table 2 : mixed correlators , eqs .
( [ mixmf ] ) , for the model @xmath228 , compared to the predictions for free @xmath20 the exact eigenvalue distribution turns out to be surprisingly similar to the semicircle of radius @xmath229 obtained at @xmath230 .
this is well reproduced by the variational calculation despite a quite strong dependence on @xmath30 of the individual coefficients in @xmath231 .
the effects of the @xmath30-dependence of the quadratic and quartic coefficients nearly cancel in the complete potential , leading to a quite stable eigenvalue distribution .
turning to the mixed correlators in table 2 , one does not observe a strong deviation from free behavior of the @xmath16 . in this case
, there is almost no room for improvement by assuming free @xmath183 instead of @xmath16 , leading to equation ( [ mixbf ] ) for @xmath232 . indeed , one here obtains for the right hand side of ( [ mixbf ] ) the value @xmath233 , the same as @xmath203 in table 2 .
the reader may wonder why only the relatively small value @xmath224 was displayed in the comparison above .
in fact , the model described by @xmath141 becomes unstable for too large values of @xmath159 .
note that the product of two hermitian matrices @xmath234 is in general not hermitian ; therefore , the interaction term , proportional to @xmath235 , is not bounded from below even for positive @xmath30 .
strictly speaking , the model is unstable for all @xmath30 , but for sufficiently small @xmath30 , there is effectively a barrier posed by the independent part of the action preventing the system from spilling over to the region of large eigenvalues where the interaction will allow eigenvalues to grow without bound .
a glimpse of this instability ( for negative @xmath30 , namely setting in at @xmath236 ) was already given in @xcite will display a similar instability when the coupling @xmath30 becomes too negative .
this was not investigated further by the authors . ] . here , instead the instability at positive @xmath30 was investigated in slightly more detail .
according to monte carlo simulations at @xmath178 , @xmath237 , and @xmath238 , the critical coupling lies in the interval @xmath239 $ ] . on the other hand ,
the variational approximation in the best basis , described by the action ( [ ts2fopt ] ) , is stable up to @xmath240 . beyond this point
, there is no solution to the consistency condition ( [ ts2foptc ] ) .
the authors also checked that there is no two - cut solution to ( [ ts2fopt ] ) above @xmath240 ( note that ( [ ts2foptc ] ) is valid only for one - cut solutions ) .
thus , the variational approximation does qualitatively capture the phase structure of the model described by @xmath141 , albeit with a badly overestimated critical coupling . in this respect , the choice of variables @xmath241 and @xmath215 represents a drastic improvement over the original variables @xmath16 . in terms
of the latter , the free counterpart of @xmath141 is simply @xmath242 , which entirely misses the unstable phase of the model ( in fact , all the dependence on the coupling @xmath30 ) .
consider a model of @xmath115 classical matrices with an action of the type @xmath243 note the scaling of this action with @xmath115 , which is necessary to retain a meaningful balance between it and the eigenvalue repulsion originating in the haar measure of the @xmath16 . before proceeding , a comment is in order regarding the special form of ( [ ldtyp ] ) and the interpretation of @xmath115 as the number of space - time dimensions . in ( [ ldtyp ] ) , each matrix degree of freedom interacts with all others in the same way .
if one wishes to attach physical meaning to the matrices as degrees of freedom living in space - time , this occurs most naturally when all the degrees of freedom are attached to the same space - time point .
such a model may become relevant if one has managed to decouple the different space - time points of a @xmath115-dimensional large-@xmath3 field theory , e.g. as the result of an eguchi - kawai reduction ( for a review , see ref .
@xcite ) . now
, in view of the scaling of the individual terms in ( [ ldtyp ] ) , one might hope that the detailed angular correlations between pairs of matrices become unimportant at large @xmath115 and a free approximation becomes exact .
this argument can be made rigorous e.g. in the kazakov - migdal model , where the leading large-@xmath115 behavior is described by a free matrix model @xcite .
the proof however depends on the specific link variable structure of the kazakov - migdal model . in order to test this idea more generally
, the models described by @xmath244 were investigated in the present framework for different @xmath115 .
the reader is reminded that in the case of @xmath245 for @xmath246 , the free variational approximation failed miserably ( cf .
previous section ) . in analyzing the models ( [ ladmo ] ) in the free variational approximation ,
the authors did not treat different choices of free variables ; for simplicity , only the best free approximation in the original variables @xmath16 was considered . then the best free approximation is easily derived : the free counterparts of ( [ ladmo ] ) are @xmath247 and thus each variable @xmath16 is described by the action @xmath248 independent of @xmath115 , with @xmath249 and @xmath250 to be determined self - consistently .
one obtains @xmath251 and @xmath252 and therefore the eigenvalue distributions @xmath253 comparing this with the results of monte carlo experiments , cf . figs .
[ fig7 ] and [ fig8 ] , the correspondence is quite satisfying already at @xmath254 ( the simulations were carried out for @xmath178 , some checks for @xmath255 did not reveal any significant differences ) . also , comparing mixed correlators of two matrices ( without loss of generality , @xmath76 and @xmath77 ) with free predictions ( cf .
equations ( [ mixmf ] ) ) in the monte carlo experiments , one obtains good agreement for @xmath254 , cf .
table 3 . ' '' '' & @xmath227 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 + @xmath256 & .62 & 1.00 & .25 & .38 & .33 & 1.00 & .18 & .62 + @xmath257 & .51 & .53 & .25 & .26 & .25 & .28 & .18 & .21 + @xmath258 & .49 & .46 & .20 & .25 & .12 & .21 & .09 & .16 + @xmath259 & .46 & .37 & .21 & .22 & .13 & .13 & .11 & .11 + table 3 : mixed correlators , eq .
( [ mixmf ] ) , for the models ( [ ladmo ] ) , and predictions for free @xmath16 as mentioned above , this agreement does not come unexpected .
however , while in the cases considered here the monte carlo solution seemed to converge towards the free approximation at large @xmath115 , it seems doubtful that this should be true in general . on the one hand ,
a solution with angular correlations , say with angles restricted to a certain sector as compared with some fixed reference matrix , costs a weight @xmath260 in the partition sum ( there are @xmath261 angles , and the coefficient @xmath156 depends on the details of the angular distributions ) .
on the other hand , the interaction itself is also scaled to be of order @xmath261 .
therefore , there should be a genuine competition between the two terms , the former favoring freeness and the latter preventing it . for a wide class of potentials , the quantum mechanical ground state of a single large-@xmath3 matrix can be solved for exactly @xcite .
this is achieved by working directly with continuous ( euclidean ) time and mapping the problem to a noninteracting fermion problem .
if one discretizes time as in the original definition of the path integral , the problem can be interpreted in terms of many interacting classical matrices . in this form ,
the quantum mechanics of one matrix is hard to solve exactly ; on the other hand , it becomes amenable to the variational method dealt with here .
interest in such a discretized form has arisen in recent applications to quantum gravity , @xcite . for definiteness ,
the model described by @xmath262 will be considered here , where @xmath263 is the number of time slices , and @xmath264 is identified with @xmath265 , i.e. periodic boundary conditions are posited .
generically , there is a competition between two different interaction terms : the nearest - neighbor coupling and the potential term , which acts instantaneously . in order to best
accomodate the former in a free variational approximation , one will want to change matrix variables to a fourier basis , which decouples the kinetic part of the action .
on the other hand , the potential term will prefer the original variables .
thus , in the present treatment , not a general linear change of variables will be considered , but only the aforementioned two discrete choices ; the variational minimization principle will decide which choice is preferred .
the fourier basis is defined as @xmath266 with ( @xmath263 is for simplicity taken to be even ) @xmath267 here , the wavevectors are @xmath268 ; note that @xmath269 is orthogonal and therefore the jacobian term in the free energy ( [ modprin ] ) gives no contribution for the fourier choice of variables . for convenience in notation , the index @xmath270 labeling the variables @xmath108 in ( [ foutra ] ) , i.e. the columns in ( [ umat ] ) , will be taken to run from @xmath271 to @xmath272 . in the fourier basis ,
the kinetic part of the action ( [ sq1def ] ) is exactly diagonalized and coincides with its free partner @xmath273 on the other hand , the potential part in the new variables is @xmath274 assuming as before @xmath275 , the only terms which will give a contribution to the free partner are ones in which the four indices @xmath276 all take the same value or pairwise two different values .
thus , one has the free partner @xmath277\ ] ] by explicit calculation , @xmath278 is of order @xmath279 and therefore the quartic term can be dropped for a large number of time slices @xmath263 . on the other hand , again by explicit calculation , @xmath280 and therefore one finally has for the free partner of the action ( [ sq1def ] ) in the fourier variables @xmath183 , @xmath281 on the other hand , the free partner of ( [ sq1def ] ) in the original variables is @xmath282 again assuming @xmath143 . here , the potential term is treated exactly , whereas the nearest - neighbor coupling has been completely truncated .
the free models are now easily solved . for the purpose of calculating the free energies , the reference action @xmath283
will be used ; then one considers ( cf .
( [ dpridef ] ) ) @xmath284 \label{sfcorig}\ ] ] for the original variables and @xmath285 for the fourier variables . in the latter ,
introducing the abbreviation @xmath286 one has semicircular distributions for the matrix variables @xmath183 of radius squared @xmath287 implying second moments @xmath288 , which self - consistently determines @xmath289 : @xmath290 or , @xmath291 the free energy is then @xmath292 \nonumber\end{aligned}\ ] ] which must be calculated numerically ( remember that at every @xmath156 , @xmath289 is determined by the equation ( [ xq1eq ] ) ) . on the other hand , in the decoupled basis , one has the quartic action ( [ sfcorig ] ) , identical and independent for all the @xmath16 . using appendix [ appa ] , the resulting moments are @xmath293 \\
\langle m^4 \rangle & = & \frac{1}{216\alpha^{3 } g^3 } \left [ ( 1+\alpha ) ^4 + 18\alpha g ( 1+\alpha ) ^2 + 54\alpha^{2 } g^2 -(1+\alpha ) ( ( 1+\alpha ) ^2 + 12\alpha g)^{3/2 } \right]\end{aligned}\ ] ] and , inserting in the free energy , @xmath294\end{aligned}\ ] ] numerically , @xmath295 at @xmath296 . for smaller @xmath30 ,
the fourier basis provides the better approximation , for larger @xmath30 the original choice of variables does .
it has been argued in @xcite that matrix models such as the one considered here exhibit a kosterlitz - thouless transition at some value of the lattice spacing separating the two regimes where the matrix chain behaves more like a collection of independent sites and the one where it supports spin wave type of modes .
the variational approach presented here , while giving no details of the transition region , is capable of capturing these two different possible regimes .
it remains to give the eigenvalue distributions in the two regimes . in the original variables , described by the action @xmath297 , cf .
equation ( [ sforig ] ) , one immediately has , using appendix [ appa ] , @xmath298 for all the variables @xmath16 . in the fourier case , one must convolute the semicircular distributions of the @xmath183 to obtain the distributions of the @xmath16 .
it is well known @xcite that the semicircular distributions are closed under additive convolution and that the square radii add up to the square radius of the resulting semicircle .
i.e. , @xmath299 inserting ( [ rsbi ] ) for @xmath300 and ( [ umat ] ) , one obtains identical radii for all the @xmath16 , @xmath301 where in the last line it has been used that @xmath302 solves ( cf .
( [ xq1eq ] ) for @xmath300 ) @xmath303 the corresponding semicircular distribution for the @xmath16 is then @xmath304 the variational approximation derived above should now be compared to monte carlo experiments and the exact results known about this model . in figs .
[ fig9]-[fig11 ] monte carlo results for @xmath305 and @xmath306 are exhibited together with variational results at different couplings @xmath30 .
checks with @xmath255 and @xmath307 revealed no significant differences .
one indeed observes , as already borne out by the calculation of the free energies , that the fourier basis is preferable in a weak potential and the original basis is favored by a strong potential .
this is corroborated by the monte carlo evaluation of correlators between matrices at neighboring time slices , cf .
table 4 . ' '' '' @xmath30 & @xmath227 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 + .1 & .55 & .57 & .41 & .31 & .59 & .33 & .52 & .26 + 1 & .24 & .11 & .066 & .059 & .015 & .012 & .014 & .0093 + 10 & .098 & .017 & .0099 & .0097 & .00031 & .00029 & .00025 & .00024 + table 4 : mixed correlators , eqs .
( [ mixmf ] ) , between matrices @xmath16 at neighboring time slices and predictions for free @xmath16 at large coupling , neighboring matrices are to a very good approximation free , meaning the original decoupled basis gives a good description . by contrast ,
at low @xmath30 , there are significant angular correlations induced by the nearest - neighbor coupling .
these angular correlations are washed out by the fluctuations when one considers matrices far apart on the temporal lattice , cf .
there is virtually no deviation from free behavior between distant matrices even at low values of @xmath30 . ' '' ''
@xmath30 & @xmath227 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 + .1 & .55 & .57 & .31 & .31 & .35 & .33 & .28 & .26 + 1 & .24 & .11 & .059 & .059 & .012 & .012 & .0093 & .0093 + 10 & .098 & .017 & .0097 & .0097 & .00029 & .00029 & .00024 & .00024 + table 5 : mixed correlators , eqs .
( [ mixmf ] ) , between matrices @xmath16 separated by half of the total length of the temporal lattice , and predictions for free @xmath16 in order to make contact with physical ( still euclidean ) time , one should write the discretized action of the quantum one - matrix model as @xmath308 where @xmath309 is now the length of the circle in time direction . rescaling @xmath310 , one regains the form ( [ sq1def ] ) of the action , with the identification @xmath311 .
therefore , @xmath30 behaves like the third power of the lattice spacing . in the continuum limit
, the ground state distribution of the eigenvalues of the quantum mechanical one - matrix model has been found analytically in @xcite .
this can now be compared to the variational solution . according to the correspondence established above ,
the limit of vanishing lattice spacing corresponds to vanishing @xmath30 ; therefore the fourier basis is the preferable one .
the eigenvalue distribution for the rescaled variables @xmath312 is ( cf . equation ( [ q1brho ] ) ) @xmath313 and @xmath289 is determined by ( cf . equation ( [ q1bx ] ) ) @xmath314 which for large @xmath263 is solved by @xmath315 therefore @xmath316 note that there is no @xmath309-dependence left in the continuum limit . by contrast , the exact solution is @xcite @xmath317 determined by the normalization condition .
the two solutions are compared for @xmath318 in fig .
[ fig12 ] .
while the case of one quantum mechanical matrix discussed in the previous section is amenable to exact treatment @xcite due to its special form , an analogous solution is not possible once more than one quantum mechanical matrix is involved . on the other hand , the approximate variational approach , using discretized time ,
is easily generalized to the case of more than one matrix . here , a commutator - type interaction will be considered , as it is especially interesting from the point of view of applications to yang - mills theories : @xmath319 ) ^2 \right ) \label{sq2def}\ ] ] where the first index labels the two different matrix variables and the second one the different time slices .
again , only the original choice of variables will be compared to a fourier basis ( cf .
( [ umat ] ) ) using the variational criterion .
note that the action @xmath320 is invariant under simultaneous shifts of all the matrices by a multiple of the unit matrix .
this trivial freedom should be removed in order to obtain stable monte carlo results ; otherwise , the system just performs a random walk in the trace of the matrices .
thus , the @xmath321 here are constrained to be traceless , i.e. the condition @xmath322 , assumed to be realized dynamically in prior examples , is enforced by hand .
the free partner to ( [ sq2def ] ) in the original basis is @xmath323 ( using @xmath324 ) . on the other hand , in the fourier basis @xmath325 with @xmath269 as in ( [ umat ] ) ,
the kinetic part becomes @xmath326 the potential part on the other hand becomes @xmath327 ^ 2 = \frac{2g}{n } \mbox{tr } \
, \sum_{i } \sum_{jklm } u_{ij } u_{ik } u_{il } u_{im } ( b_{1,j } b_{1,k } b_{2,l } b_{2,m } - b_{1,j } b_{2,k } b_{1,l } b_{2,m } ) \ ] ] using again the assumption @xmath328 , the second term in the round brackets on the right hand side never gives a contribution to the free partner , whereas the first term only gives a contribution if @xmath329 and @xmath330 .
thus one has @xmath331 now one can again easily solve the free models .
for the purpose of calculating the free energies , the reference action @xmath332 is convenient ; then one has to solve the models @xmath333 for the original variables @xmath321 and @xmath334 for the fourier variables @xmath335 .
starting with ( [ q2fcor ] ) , abbreviating @xmath336 , one obtains identical semicircular distributions for all matrix variables with radius squared @xmath337 whereupon the consistency condition determining @xmath175 becomes @xmath338 solved by @xmath339 the free energy then is given by @xmath340 on the other hand , in the fourier case ( [ q2fcfo ] ) , abbreviating @xmath341 one notices that the variables @xmath342 and @xmath343 are controlled by exactly the same potential as the variables @xmath108 in the case of the quantum mechanical one - matrix model discussed in the previous section . in particular , one has the consistency condition @xmath344 and the second moments @xmath345 the free energy now is @xmath346\ ] ] numerical evaluation of the two free energies yields @xmath347 for all @xmath30 , i.e. the fourier basis is always preferable over the original basis . while this is to be expected at small @xmath30 , it is not necessarily an indication that the fourier basis represents a particularly good approximation for large @xmath30 as well : rather , working in the original basis still truncates the interaction term rather badly at every time slice ( cf . equation ( [ sq2fdef ] ) )
this is different from the one - matrix case , where assuming the original variables to be free merely implied decoupling variables at different time slices , which certainly should become exact for large coupling @xmath30 . in the two - matrix case , by contrast , there is still an additional truncation at every time slice which prevents convergence to the exact problem at large @xmath30 . in light of this
, it is not so surprising that the fourier basis is preferable for all @xmath30 . nevertheless , in figs .
[ fig13]-[fig15 ] , this basis appears to provide quite a satisfactory description even for @xmath348 ; the agreement with monte carlo results ( @xmath178 and @xmath305 were used in the latter ) is not much worse at @xmath348 than at @xmath349 . also in the correlators , cf .
tables 6 and 7 , one observes that the angular correlations involving the same matrix variable at different time slices are stronger than the angular correlations between the two different matrix variables at a fixed time , up to quite high @xmath30 . at @xmath348 ,
these two correlations seem to be roughly equally strong .
this again corroborates the result that using the fourier basis , which treats the nearest - neighbor coupling exactly , provides a good approximation for a large range of @xmath30 . ' '' '' @xmath30 & @xmath350 & @xmath351 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 + .1 & .86 & 1.5 & 1.2 & .75 & 5.6 & 2.3 & 5.1 & 1.7 + 1 & .38 & .29 & .19 & .14 & .16 & .084 & .13 & .063 + 10 & .18 & .068 & .038 & .033 & .0064 & .0047 & .005 & .0034 + table 6 : mixed correlators , eqs .
( [ mixmf ] ) , between matrices @xmath321 for one fixed @xmath352 at neighboring time slices @xmath353 and predictions for free @xmath321 ' '' '' @xmath30 & @xmath350 & @xmath351 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 + .1 & .86 & 1.5 & .66 & .75 & 1.7 & 2.3 & 1.3 & 1.7 + 1 & .38 & .29 & .12 & .14 & .059 & .084 & .043 & .063 + 10 & .18 & .068 & .027 & .033 & .0027 & .0047 & .002 & .0034 + table 7 : mixed correlators between the matrices @xmath354 and @xmath355 at a fixed time slice @xmath353 and predictions for free @xmath354 , @xmath355 .
ultimately , though , one would expect a local treatment , which assumes different time slices to be free with respect to one another , to provide better agreement at high @xmath30 if it manages to well approximate the commutator interaction .
this is precisely what is not achieved if the original variables are assumed to be free . on the contrary ,
a nontrivial mixing of the variables @xmath354 and @xmath355 at every time slice is necessary such as to better accomodate the commutator - type interaction in ( [ sq2def ] ) .
note that linear combinations of @xmath354 and @xmath355 will not effect any improvement , since the commutator is invariant under such linear transformations ( up to trivial rescalings of the variables ) .
it is disappointing that precisely this phenomenologically very interesting type of interaction term seems particularly resistent to free approximation combined with linear transformations . here
, it seems that using ( technically more complicated ) nonlinear variable transformations must be contemplated in order to capture the essential angular correlations .
note also that the commutator term in the action on its own does not confine the eigenvalue distributions of the involved variables to a compact support , even if one enforces tracelessness ; the eigenvalues can become arbitrarily large if the matrices are located in the regions of configuration space where they commute .
this pathology however seems to disappear ( according to monte carlo experiments carried out by the authors ) when more than two matrices are involved ( i.e. a higher - dimensional yang - mills type of action ) .
presumably there , the regions of configuration space where all the matrices commute are too small compared with the whole space such that entropy suppresses configurations with arbitrarily large eigenvalues .
in this work , a new variational approach to interacting large-@xmath3 multi - matrix models was developed .
the partition function was approximated using the variational principle @xmath356 with @xmath94 initially taken from the space of all matrix models which are free in the set of variables the original interacting problem is given in .
it turned out to be possible to give a general solution to this variational problem for a fixed set of matrix variables in terms of the concept of the `` free partner '' introduced by the authors .
the free partner defines a type of mean field approximation to the original action and is constructed using the axioms of freeness in a fashion analogous to the construction of more conventional mean field theories for fermionic or bosonic particles using wick s theorem .
the freeness axioms constitute the analog of wick s theorem for objects obeying the cuntz algebra , which is the algebra obeyed by the fock space operator representation of free random variables @xcite .
however , the variational approach presented here goes beyond simply acting as a device to derive a mean field theory . above
, the variational space was characterized as being the space of all matrix models which are free in the set of variables the original interacting problem is given in .
this implies that one can considerably enlarge the variational space by first allowing for a change of variables in the original problem and only then varying over all matrix models which are free in the variables one has settled for . in this way one
includes into the variation models which are not free in the original variables .
every set of variables defines a different mean field theory given by the corresponding free partner .
the variational principle not only allows to derive these mean field theories , but further allows to decide which out of a set of mean field theories provides the best approximation to the exact problem .
a number of examples were considered in order to assess the accuracy of the variational method ; the variational results were compared with monte carlo simulations as well as analytical results , where available . to begin with
, classical two - matrix models described by actions with at most quartic terms were investigated allowing for a general linear transformation of matrix variables .
impressive agreement was observed , except in the case of the action @xmath198 , eq . ( [ s1redef ] ) .
the authors did not attempt to improve the approximation by allowing for nonlinear variable transformations , which may relieve the discrepancy .
classical models with a large number of matrices were considered in which all matrices pairwise interact in the same way .
such models e.g. become relevant when one has managed to decouple the different space - time points of a large-@xmath3 matrix field theory in a large number of dimensions @xmath115 , e.g. in the framework of an eguchi - kawai reduction . among the models considered
was one which reduces to @xmath198 for @xmath246 .
it turned out that the quality of the variational approximation using only the original set of variables improves as @xmath115 rises .
the variational approximation was also considered for the problems of a single and two coupled matrix chains simulating path integrals for one and two quantum matrices .
two discrete choices of free variables , namely the original variables and a fourier basis , were considered . in the single chain case , as expected , for small ( large ) lattice spacing , the fourier ( original ) basis provided the better approximation . in the continuum limit of the one - matrix case ,
good agreement was achieved with the available analytical solution . for two matrix chains coupled via a commutator - type interaction
, the fourier basis provided a better approximation for all couplings ; it also compares quite favorably with the exact monte carlo results for a large range of coupling constants .
however , for very large coupling , one would expect a local basis , constructed such as to capture the angular correlations introduced by the commutator interaction , to ultimately be more appropriate . here , again it seemed that technically more involved nonlinear transformations must be contemplated in order to achieve further progress .
following @xcite , @xcite the one - cut eigenvalue distribution @xmath357 resulting from a hermitian large-@xmath3 one - matrix model in a quartic potential @xmath358 is most easily found by considering the resolvent , @xmath359 where @xmath360 is a polynomial .
@xmath360 , along with @xmath361 , which defines the support of @xmath357 , is fixed by comparing coefficients in the asymptotic behavior of @xmath362 , @xmath363 where @xmath364 is the @xmath352-th moment of the distribution @xmath357 .
one obtains @xmath365 then one can also immediately read off @xmath366 ( where @xmath367 ) , and by expanding @xmath362 in @xmath368 , @xmath369 \\
c_4 & = & \frac{1}{216 a_4 ^ 3 } \left [ a_2 ^ 4 + 18 a_2 ^ 2 a_4 + 54 a_4 ^ 2 - a_2 ( a_2 ^ 2 + 12 a_4 ) ^{3/2 } \right ] \end{aligned}\ ] ] when @xmath370 , these expressions simplify to @xmath371 i.e. the second moment of the semicircular distribution is one fourth of its square radius . in complete analogy , one can derive the two - cut solution in the quartic potential .
then , @xmath362 has the form @xmath372 from which one again obtains by comparing coefficients in the asymptotic expansion @xmath373 one can immediately read off the eigenvalue distribution , and the second moment e.g. becomes @xmath374
in order to additively convolute the eigenvalue distributions of two matrices which are free with respect to one another , i.e. to obtain @xmath375 given @xmath376 and @xmath377 , one should go to the corresponding r - transforms @xcite .
these are defined as follows : find the resolvents @xmath378 and @xmath379 as defined in appendix [ appa ] ; then the r - transforms are given by @xmath380 ( here @xmath381 denotes the inverse function , @xmath382 ) .
the r - transforms behave additively , i.e. @xmath383 . in this sense
they are the free analogs of the logarithm of the fourier transform for probability distributions of ordinary commuting random variables . in the case of matrices governed by a quartic potential @xmath384
, the resolvent has already been given explicitly in appendix [ appa ] ; denoting @xmath385 , one has that @xmath386 satisfies @xmath387 where @xmath361 was defined in appendix [ appa ] . isolating the square root term and squaring the resulting equation leads to a third order equation for @xmath388 ,
@xmath389 it is not necessary to solve this explicitly , since the case of interest here is the case of two identical distributions , i.e. identical r - transforms .
then one has @xmath390 implying @xmath391 however , the equation satisfied by @xmath388 is known , equation ( [ eqforz ] ) , which by inserting ( [ zitow ] ) leads to the equation satisfied by @xmath392 .
since one is actually interested in @xmath393 , one can solve directly for @xmath394 by rearranging ( [ eqforz ] ) in conjunction with ( [ zitow ] ) as a polynomial in @xmath394 , @xmath395 rather than obtaining a cumbersome analytical solution of ( [ eqfory ] ) , the relevant solution was tracked numerically .
this is best done in small steps in along the positive real @xmath396-axis starting from the known asymptotic behavior @xmath397 and using the positivity of the eigenvalue distribution @xmath398 .
the latter is essentially given by the imaginary part of @xmath394 , namely @xmath399 .
g.t hooft , nucl .
b 72 ( 1974 ) 461 .
g.t hooft , nucl .
b 75 ( 1974 ) 461 .
s.coleman , 1/n , in : aspects of symmetry ( cambridge university press , cambridge , 1985 ) .
c.g.callan , n.coote and d.j.gross , phys .
d 13 ( 1976 ) 1649 .
m.b.einhorn , phys .
d 14 ( 1976 ) 3451 .
e.witten , nucl .
b 160 ( 1979 ) 57 , nucl . phys .
b 223 ( 1983 ) 433 .
g.s.adkins , c.r.nappi and e.witten , nucl .
b 228 ( 1983 ) 552 . y.m.makeenko and a.a.migdal , phys .
b 88 ( 1979 ) 135 ; a.a.migdal , phys . rep .
102 ( 1983 ) 199 .
e.brzin , c.itzykson , g.parisi and j .- b.zuber , comm . math .
59 ( 1978 ) 35 .
c.itzykson and j.b.zuber , j. math .
21 ( 1980 ) 411 .
m.l.mehta , random matrices ( academic press , new york , 1991 ) . the large - n expansion in quantum field theory and statistical physics : from spin systems to 2-dimensional gravity , eds . e.brzin and s.r.wadia ( world scientific , singapore , 1993 ) . b.l.altshuler and b.d.simons , in : proceedings of les houches summer school , session lxi , 1994 , eds .
e.akkermans , g.montambaux , j .- l.pichard and j.zinn-justin .
j.ambjrn , j.jurkiewicz and yu.m.makeenko , phys .
b 251 ( 1990 ) 517 . e.brzin and a.zee , nucl .
b 402 [ fs ] ( 1993 ) 613 . e.brzin and a.zee , phys .
e 49 ( 1994 ) 2588 ; e.brzin , s.hikami and a.zee , phys .
e 51 ( 1995 ) 5442 .
v.a.kazakov , m.staudacher and t.wynter , preprints hep - th/9502132 , hep - th/9506174 , hep - th/9601069 .
v.a.kazakov and a.a.migdal , nucl.phys .
b 397 ( 1993 ) 214 .
d.v.voiculescu , k.j.dykema and a.nica , free random variables ( am . math .
soc . , providence , ri , 1992 ) .
i.m.singer , unpublished , and in : functional analysis on the eve of the 21st century , in honor of i.m.gelfand , eds . s.gindikin , j.lepowsky and r.l.wilson , progress in mathematics , vol . 131 ( birkhuser , boston , 1995 ) . r.gopakumar and d.j.gross , nucl .
b 451 ( 1995 ) 379 .
m.r.douglas , nucl .
b 41 ( proc . suppl . )
( 1995 ) 66 .
a.zee , nucl .
b 474 [ fs ] ( 1996 ) 726 . i.ya.arefeva and i.v.volovich , preprint hep - th/9510210 .
p.neu and r.speicher , j. phys . a 28 ( 1995 ) l79
; j. stat . phys .
80 ( 1995 ) 1279 .
r.p.feynman and a.r.hibbs , quantum mechanics and path integrals ( mcgraw - hill , new york , 1965 ) .
s.das , a.dhar , a.sengupta and s.wadia , mod .
lett . a 5 ( 1990 ) 1041 .
l.chekhov and c.kristjansen , preprint hep - th/9605013 .
s.r.das , rev . mod .
59 ( 1987 ) 235 .
y.makeenko , preprint hep - th/9512211 . d.j.gross and i.klebanov , nucl .
b 344 ( 1990 ) 475 . n.i.akhiezer and i.m.glazman , theory of linear operators in hilbert space , vol.i ( frederick ungar , new york , 1963 ) . | matrices are said to behave as free non - commuting random variables if the action which governs their dynamics constrains only their eigenvalues , i.e. depends on traces of powers of individual matrices . the authors use recently developed mathematical techniques in combination with a standard variational principle to formulate a new variational approach for matrix models .
approximate variational solutions of interacting large - n matrix models are found using the free random matrices as the variational space .
several classes of classical and quantum mechanical matrix models with different types of interactions are considered and the variational solutions compared with exact monte carlo and analytical results .
impressive agreement is found in a majority of cases . |
one of the most common ways to investigate the properties of a dynamical system is to study how it responds to controlled external perturbations .
the response of a system to a weak perturbing field is related to its equilibrium fluctuations by the celebrated fluctuation dissipation relation @xcite .
the response provides a direct measure of system dynamics and fluctuations . in a time - domain response measurement
one uses a series of impulsive perturbations ( fig .
[ fig1 ] ) and records some property of the system as a function of their time - delays .
impulsive perturbations make it possible to study the free dynamical evolution of the system during the time delays unmasked by the time profile of the perturbing field .
furthermore , the joint dependence on several time delays can be used to separate the contributions of different dynamical pathways . due to its dependence on multiple time delays
this method is termed multidimensional .
the response of a system is typically measured by the expectation value of some operator @xcite .
this is a linear functional of the system probability density ( or the density matrix for quantum systems ) .
multidimensional response have had considerable success in nonlinear spectroscopy , due to the ability to control and shape optical fields .
applications range from spin dynamics in nmr @xcite , vibrational dynamics of proteins in infrared systems and electronic energy transfer in photosynthetic complexes as probed by visible pulses @xcite .
these span a broad range of timescales from milliseconds to femtoseconds .
interestingly , there exist non - linear functionals of probability densities which have interesting physical interpretations .
one such quantity is the von - newman entropy @xmath0 .
a related quantum nonlinear measure called the concurrence serves as a measure of quantum entanglement @xcite .
the kullback - leibler distance ( kld ) or relative entropy , @xmath1 , which compares one probability distribution to another , is a nonlinear measure that had been found useful in many applications .
this paper aims at developing multidimensional measures based on the kld .
numerous applications of the kld differ in the probability distributions involved .
the ratio of the probability of a stochastic path and its reverse at a steady state has been connected to a change of entropy @xcite . for an externally driven system a similar quantity was found to be related to the work done on the system @xcite . as a result , the kld which compares the path distribution to a distribution of reversed paths is a measure of the lack of reversibility of a thermodynamical process . for distributions in phase - space ( as opposed to path - space ) , the kld between the density of a driven system and the density of a reversed process @xcite , or the distance between the driven density and the corresponding equilibrium density @xcite ( for the same value of parameters ) were shown to be bounded by dissipated work in the process .
the transfer of information through a stochastic resonance is quantified by the kld between the probability distributions with and without the external input @xcite .
the ability of neuronal networks to retain information about past events was characterized by the fisher - information @xcite , which is closely related to the kld between a distribution and the one obtained from it by a small perturbation .
we shall examine the response of a system to impulsive perturbations which drive it out of a stationary ( steady state or equilibrium ) state .
the kld between the distribution before and after the perturbation does not correspond to an entropy , or work .
however , since it compares the perturbed and unperturbed densities , it characterize how `` easy '' it is to drive the system away from its initial state .
in ordinary response theory , one compared the expectation values of some operator taken over the perturbed and unperturbed probability density .
this depends on the specific properties of the observed operator .
the kld is a more robust measure for the effect of the impulsive perturbation on the probability density . by expanding the kld in the perturbation strength
we obtain a hierarchy of kullback - leibler response functions ( klrf ) .
these differ qualitatively from the hierarchy of ordinary response functions ( orf ) , since they are nonlinear in the probability density .
the klrf serve as a new type of measures characterizing the dynamics and encoding different information than the orf .
for example , the second order klrf , which we connect to the fisher - information , is found to exhibit qualitatively different dependence on the time delays , depending on whether the system is perturbed out of a steady state or out of thermal equilibrium .
this is in contrast to the corresponding orf .
the fluctuation dissipation relation @xcite , which is linear in the density matrix , can also distinguish between systems driven out - of - equilibrium and out of a steady state .
the kld offer a different window into this aspect .
the structure of the paper is as follows . in sec .
[ formalsec ] we describe the multidimensional measures and present the two heirarchies of orf and klrf response functions .
these are then calculated using a formal perturbation theory in the coupling strength to the external perturbation . in sec .
[ overdampedsec ] we show that for systems undergoing overdamped stochastic dynamics the non - linear klrf are naturally described using a combination of the stochastic dynamics and its dual dynamics . in sec .
[ mastersec ] we extend the results of sec .
[ overdampedsec ] to discrete markovian systems with a finite number of states .
our results are discussed in sec .
[ discsec ] .
we consider a system initially at a stationary state ( either equilibrium or a steady state ) , which is perturbed by a series of short pulses , as depicted in fig .
[ fig1 ] . and
subjected to impulsive perturbations .
the @xmath2th pulse is centered at @xmath3 and its strength is denoted by @xmath4 .
@xmath5 are the time intervals between successive pulses .
[ fig1 ] ] the probability distribution describing the driven system at time @xmath6 , @xmath7 , depends parametrically on @xmath8 , the strength of the @xmath9th pulse , as well as on the time differences between the pulses @xmath10 with @xmath11 .
ordinary response theory focuses on the expectation value of some observable @xmath12,\ ] ] and its dependence on the parameters @xmath13 .
the lowest ordinary response functions ( orf ) are @xmath14 and so forth .
the time differences in eqs .
( [ defr2])-([defr4 ] ) can be expanded in terms of the time delays between the pulses , @xmath15 .
@xmath16 are used to investigate various properties of the unperturbed dynamics , such as the existence of excited modes , and the relaxation back to a steady state . here
, we focus on different , but closely related quantity . instead of studying an expectation value of an observable ,
we focus on a quantity that compares the perturbed and unperturbed probability distributions .
the kld , also known as the relative entropy , is defined as @xmath17 the kld vanishes when the two distributions are equal @xmath18 , and is positive otherwise , @xmath19 for @xmath20 @xcite .
note that the kld is not a true distance since @xmath21 and it does not satisfy the triangle inequality .
the kld measures the dissimilarity between two distributions .
it had found many applications in the field of information theory @xcite .
for instance , the mutual information between two random variables @xmath22 is @xmath23 , where @xmath24 is the joint distribution while @xmath25 ( @xmath26 ) is the marginal distribution of @xmath27 ( @xmath28 ) . in the present application
the kullback - leibler distance is a measure for the deviation of the system from its initial state . in a manner similar to the definition of the orf
, we define a klrf hierarchy by taking derivatives of the kld with respect to pulse strengths , and displaying them with respect to the time delays @xmath29 all the derivatives are calculated at @xmath30 , and we have used the relation @xmath31 . this is also true for all other @xmath32 derivatives in the following . to keep the notation simple
we will not state this explicitly .
higher order klrf are defined similarly .
it is important to note that the orf are linear in @xmath33 whereas the kld are nonlinear .
we thus expect the kld to carry qualitatively different information about the dynamics .
the second derivative ( [ second ] ) , known as the fisher memory ( or information ) matrix , plays an important role in information theory , since the cramer-rao inequality means that it is a measure of the minimum error in estimating the value of a parameter of a distribution @xcite .
the fisher information has been used recently to analyze the survival of information in stochastic networks @xcite .
conservation of probability implies that the first klrf vanishes @xmath34 the second derivative , the fisher memory matrix , is given by @xmath35 a straightforward calculation allows to recast the third order derivative of @xmath36 in terms of products of lower order derivatives @xmath37 in what follows the derivatives will be calculated perturbatively in @xmath38 .
it is important to note that the @xmath39th derivative of @xmath40 has contributions from interaction with at most @xmath39 pulses .
the contribution from the linear component , which interacts with @xmath39 pulses , has the same structure of the perturbation theory for observables ( which is also linear ) .
however , since @xmath40 is a non linear function of @xmath33 the @xmath39th derivative contains a non - linear contribution which is a product of lower order contributions for @xmath33 .
the klrf encode qualitatively new information about the system dynamics in comparison to the orf .
the time evolution of the probability distribution is given by @xmath41 this formal equation is quite general , and can describe either hamiltonian ( unitary ) or stochastic dynamics where the operator @xmath42 will accordingly be the liouville , or the fokker - planck operator . for a system subjected to a time dependent weak perturbation we can write @xmath43 where we assume that the unperturbed system is time independent and is intially in a steady state , @xmath44 , so that @xmath45 .
we consider an impulsive perturbation of the form @xmath46 where @xmath47 describes the action of a pulse on the probability distribution , and @xmath8 is the overall strength of the @xmath9th pulse .
using these definitions , the state of the system at time @xmath6 can be expanded as a power series in the number of interactions with the pulses @xmath48 the partial corrections for the density , @xmath49 , appearing in eq .
( [ pseries ] ) , contain all the information necessary for computing both the klrf and orf .
they are given by @xmath50 here @xmath51 $ ] is the free propagator of the unperturbed system .
conservation of probability requires that @xmath52 , which , in turn , means that @xmath53 .
( [ pseries])-([defs2 ] ) can be used to calculate the logarithmic derivatives , which then determine the klrfs .
we will only need the first two logarithmic derivatives , which are given by @xmath54 and @xmath55 to calculate the orf , we substitute eq .
( [ pseries ] ) in eq .
( [ avga ] ) , resulting in @xmath56 it is interesting to compare the expressions of the klrf the orf .
we calculate @xmath57 and @xmath16 for @xmath58 perturbing pulses . to leading order
, we find @xmath59 at the next order , we compare the fisher information to the second order orf , @xmath60 and @xmath61 the non - diagonal elements of @xmath62 depend on the two delay times .
expressions for the third order response functions are given in app .
[ thirdorderformal ] . both @xmath16 and @xmath57
depend on the same set of @xmath58 time intervals with some important differences .
@xmath63 vanishes , while the linear response @xmath64 does not .
@xmath65 and @xmath66 have a different structure : @xmath66 can be calculated from the second order correction to the density ( or @xmath67 ) while @xmath65 is determined from a product of @xmath68 s describing the first order interaction with different pulses .
this difference reflects the non - linear dependence of the klrf on @xmath33 , and also applies to higher orders .
a comment is now in order regarding our choice of the kld ( [ defkld ] ) .
we have chosen to use @xmath69 as the measure for the effect of the perturbations .
@xmath70 would have been equally suitable . however , as discussed in app .
[ hamiltonianapp ] , the leading order of both klds in the strength of the perturbation , i.e. their fisher informations , coincide . therefore all the following results pertaining to the fisher information would hold for either choice .
in the following we use the formal results of sec . [ formalsec ] to calculate the leading order orf and klrf for a system undergoing overdamped stochastic dynamics .
we show that the fisher information is related to a forward - backward stochastic process .
the backward part is driven by the @xmath44-dual process , which will be simply referred to as the dual in what follows .
the fisher information is found to exhibit qualitatively different properties for systems perturbed from equilibrium , or from a steady state .
we also use the eigenfunctions and eigenvalues of the dynamics to derive explicit expressions for several low order orf and klrf . in stochastic dynamics
the probability density plays the role of a reduced density matrix , which depends on a few collective coordinates . in this reduced description
the entropy @xmath71 typically increases with time .
this should be contrasted with a description which includes all the degrees of freedom , where the dynamics is unitary and the entropy does not change in time . for completeness unitary dynamics
is discussed in app .
[ hamiltonianapp ] .
the fisher information can be represented in terms of the dual stochastic dynamics .
this interesting property reflects its non - linear dependence on @xmath33 .
we examine a stochastic dynamics of several variables @xmath72 , given by @xmath73 here we use the ito stochastic calculus . ) .
both methods are equally viable as long as they are used in a consistent manner .
details can be found is ref .
the noise terms are assumed to be gaussian with @xmath74 with @xmath75 a symmetric positive definite matrix . while for many systems this matrix does not depend on the coordinate , @xmath76 , this assumption will not be used in what follows
. equation ( [ stochasticoriginal ] ) is equivalent to the fokker - planck equation @xmath77 in what follows we present the dual dynamics , which can be loosely thought as the time reversed dynamics : it have the same steady state , but with reversed steady state current .
we consider the current density @xmath78 the fokker - planck equation can be written in terms of the current , @xmath79 the steady - state is the solution of @xmath80 we write @xmath81 which defines @xmath82 . for systems at equilibrium @xmath82
is simply the potential .
however , this is not the case for general steady states .
the steady state current can be written as @xmath83 after some algebra , the generator of the stochastic dynamics can be written in terms of the steady state density and currents @xcite @xmath84 the dual dynamics is given by @xmath85 with @xmath86 a straightforward calculation gives @xmath87 with @xmath88 it is a simple matter to verify that the dual dynamics has the same steady state as the original one , but the steady state currents have opposite signs .
it can be simulated by integrating the ito stochastic equation @xmath89 the dual dynamics reverses the non conservative forces in the system .
this relates the joint probability to go from one place to another in the original dynamics to the joint probability of the reversed sequence of events in the dual dynamics @xcite @xmath90 the left hand side of eq .
( [ conddual ] ) is the joint steady state probability to first the system at @xmath27 , and at @xmath91 after a time @xmath92 .
the right hand side is the joint probability of the reversed sequence of events , but for a modified dynamics . when this modified dynamics is the dual these joint probabilities become equal .
we next turn to discuss the system s response to a series of impulsive perturbation .
we assume that the perturbation is of the form @xmath93 with @xmath94 as a potential field perturbing the system . using eq .
( [ defs1 ] ) we have @xmath95 where @xmath96 with the help of equation ( [ conddual ] ) , we obtain @xmath97 we now have all the tools needed to compare the orf and klrf for overdamped stochastic dynamics .
the leading order response function is given by @xmath98 while @xmath99 .
at the next order , we have @xmath100 and @xmath101 b({\bf x}_0 ) \rho_0 ( { \bf x}_0 ) .
\label{r2deriv}\ ] ] some insight into the structure of different response functions can be gained by representing them as ensemble averages over stochastic trajectories
. the first order response function can be simulated directly using stochastic trajectories of the original dynamics .
the appearance of a derivative of the conditional probability complicates the direct simulation of @xmath102 . it may be possible to circumvent this difficulty using the finite field method , where one combined simulations with and without a finite , but small perturbation @xcite .
@xmath103 can be simulated with trajectories which follow the original dynamics for time @xmath104 and then the dual dynamics for time @xmath105 .
systems at equilibrium are self dual , allowing to substitute @xmath106 in eq .
( [ q12usingdual ] ) . as a result
the fisher information only depends on a single time variable @xmath107 .
this is in contrast to systems which are perturbed from a nonequilibrium steady state , whose fisher information is a two dimensional function of @xmath92 and @xmath105 .
the fisher information is therefore qualitatively different for systems which are perturbed out of a steady state , or out of an equilibrium state . for the self - dual case
@xmath103 has the same structure as @xmath64 , up to a replacement of @xmath94 with @xmath108 .
however , in the general , non self - dual case , the structure of @xmath109 is manifestly different than that of @xmath64 and @xmath66 .
an alternative approach for the calculation of the fisher information , as well as higher order klrf , uses eigenfunction expansions for the density @xmath33 .
it will be sufficient to examine a simple one dimensional model where @xmath110 is a fokker - planck operator and the perturbation is given by eq .
( [ la ] ) . the propagation of the unperturbed system can be described in terms of the eigenvalues and eigenfunctions of @xmath110 .
the right eigenfunctions satisfy @xmath111 similarly , the left eigenfunctions , @xmath112 , satisfy @xmath113 .
it is assumed that the right and left eigenfunctions constitute a byorthogonal system , that is @xmath114 any probability density can be expanded in terms of right eigenfunctions , @xmath115 with @xmath116 we consider systems perturbed out of thermal equilibrium , with a probability density @xmath117 . in this case the left eigenvalues are simply related to the right eigenvalues . for variable @xmath27 which is even with respect to time reversal , this relation takes the form @xcite @xmath118 as a result , one can use only the left eigenfunctions , which in this case satisfy the orthogonality condition @xmath119 we note that @xmath120 and @xmath121 correspond to the equilibrium distribution .
our goal is to calculate the response of the system to a perturbation with a coordinate dependent operator @xmath94 .
several types of integrals appear repeatedly in the calculation , and it will be convenient to introduce an appropriate notation .
one comes from the need to decompose the probability distribution into eigenstates after each interaction with a pulse @xmath122 = -\int dx \frac{\partial q_m}{\partial x } \frac{\partial a}{\partial x } q_n(x ) \rho_0(x),\label{defb}\ ] ] where integration by parts was used in the second equality .
( we assume that @xmath123 falls of fast enough to eliminate boundary terms . )
the calculation of the response also involves an evaluation of the average of an observable , which , in the current setting , leads to integrals of the form @xmath124 it is straightforward to calculate @xmath68 and @xmath67 with the help of eq .
( [ defb ] ) .
we find @xmath125 and @xmath126 we now calculate the first few orf and klrf .
the first order response functions is @xmath127 similarly , the second order response function is given by @xmath128 @xmath66 should be compared to the klrf of the same order , namely the fisher information @xmath129},\ ] ] which is calculated with the help of eqs .
( [ fisher2 ] ) and ( [ orthleft ] ) .
again , the fisher information of systems perturbed out of equilibrium depends only on @xmath130 . for a system initially in a steady state the fisher information is @xmath131 - \lambda ( m ) t_2 } \int dx \rho_0^{-1 } ( x ) \rho_n ( x ) \rho_m ( x ) .
\label{geneigenq}\ ] ] however , since the relation eq .
( [ lrrelation ] ) does not hold in this case , the integral in eq .
( [ geneigenq ] ) does not vanish for @xmath132 , and the fisher information becomes a two dimensional function of @xmath92 and @xmath105 .
@xmath133 and @xmath134 are calculated in app .
[ thirdordereigen ] , where it is shown that @xmath133 is a three dimensional function of all its time variables .
this qualitatively different signature of systems initially at steady state vs equilibrium is unique to the fisher information , due to its quadratic dependence on @xmath135 . it does not apply to higher order quantities , such as @xmath136 .
the eigenfunctions for a simple example , of an harmonic oscillator with an exponential perturbation , are presented in app .
[ exampleosc ] .
the description of the klrf in terms of a combination of the regular stochastic dynamics and its dual holds also for markovian systems with a finite number of states .
below we derive a simple expression for the fisher information of a stochastic jump process in terms of the dual dynamics of the original process .
consider a system with a finite number of states , undergoing a markovian stochastic jump process described by the master equation @xmath137 @xmath138 is a vector of probabilities to find the system in its states and @xmath139 is the transition rate matrix @xcite .
its off diagonal elements are positive , @xmath140 for @xmath141 , and express the rate of transitions from state @xmath58 to @xmath9 , given that the system is at @xmath58 .
the diagonal elements satisfy @xmath142 .
we further assume that there exist a unique steady state , @xmath143 , satisfying @xmath144 and that at this steady state there is a non vanishing probability to find the system at each of the states @xmath9 , @xmath145 .
the master equation is one of the simplest models for irreversible stochastic dynamics .
below we briefly describe some of its relevant properties , such as the backward equation , and its dual dynamics .
one can define an evolution operator @xmath146 which satisfy the equation of motion @xmath147 with the initial condition @xmath148 . for this model
a dynamical variable @xmath149 is a vector with a value corresponding to each state of the system .
its expectation value is given by @xmath150 instead of calculating this average by propagating @xmath138 , one can define a time dependent dynamical variables , @xmath151 , such that @xmath152 .
this is the analogue of the scrdinger and heisenberg pictures in quantum mechanics .
the equation of motion for @xmath151 is easily shown to be @xmath153 this equation is known as the backward equation , which turns to be related to the dual dynamics .
( the backward equation is also written as a function of an _ initial _ rather than a final time . )
@xmath154 and @xmath155 have the same eigenvalues . however , the roles of the right and left eigenvectors are interchanged and @xmath151 decays to a uniform vector with at long times .
let us now define a diagonal matrix @xmath156 , using @xmath157 .
the dual evolution is then defined as @xmath158 the fact that @xmath159 is built from the steady state of @xmath154 , @xmath160 , guaranties that @xmath161 is a physically reasonable rate matrix , that is , that is satisfies @xmath162 .
the dual dynamics describes a physically allowed process which has the same steady state as the original process it was derived from .
however , at this steady state the dual currents have opposite signs compared to the steady state currents of the original dynamics .
a process is self - dual , that is , @xmath163 , if and only if it satisfies detailed balance .
self duality is therefore related to being in thermal equilibrium .
consider a system subjected to several impulsive perturbations @xmath164 where @xmath165 corresponds to some physical perturbation @xmath166 . here
the free evolution is given by @xmath167 similarly the evolution during an impulsive perturbation can be described by @xmath168 where @xmath169 is arbitrarily small .
we expand the exponent in eq .
( [ impulse ] ) , and collect all terms of the same order in @xmath170 .
the expression for the fisher information , following eq .
( [ fisher2 ] ) is @xmath171 this expression can be simplified by writing one of the propagators in terms of the dual process , using the relation @xmath172 after some algebraic manipulations , we find @xmath173 in eq .
( [ fisherdual ] ) we have defined @xmath174 .
@xmath175 is not the dual of @xmath165 since @xmath159 is not composed of the eigenvector of @xmath165 which corresponds to a vanishing eigenvalues .
( @xmath159 is composed of the eigenvalue of @xmath176 . )
equation ( [ fisherdual ] ) is the analogue of eq .
( [ q12usingdual ] ) for discrete systems .
it demonstrates that for this model the fisher information can be simulated using a combination of the ordinary process and its dual .
however one must also include a ( possibly artificial ) dual perturbation , @xmath175 , which can nevertheless be computed from the known physical one .
as before , equation ( [ fisherdual ] ) shows that the fisher information , for self - dual systems , is a function of @xmath130 . as a side remark ,
the model considered here has only even degrees of freedom under time reversal .
more general models may also include odd degrees of freedom , such as momenta . in that case
one may speculate that there would be anti - self - dual systems whose dual turns out to be the time reversed dynamics .
this would lead to a fisher information depending on @xmath92 alone , as is the case for unitary dynamics .
in this work we have studied a system driven from its steady state by a sequence of impulsive perturbations .
we have defined a new set of measures for the response of the system to the perturbation , the klrf , which are given by the series expansion of the kullback - leibler distance between the perturbed and unperturbed probability distributions . at each order
the klrf and orf depend on the same time differences between the pulses . however , there are important differences stemming from the nonlinear dependence of the klrf on @xmath33
. the expression for the klrf , for instance eqs . ( [ fisher2 ] ) and ( [ q123 ] ) reveal quantities which can be simulated using several trajectories which end at the same point .
we have shown that a simpler , but equivalent description exists .
it uses the dual dynamics which allows to `` run some of the trajectories backward '' .
this description is especially appealing for the fisher information .
instead of viewing the fisher information as composed over sum of pair of trajectories joined at their end point one can view it as an average of contributions of a single forward - backward trajectory .
another difference between the klrf and the orf has to do with the appearance of derivatives of conditional probabilities , which for _ deterministic _ systems would correspond to groups of very close trajectories .
these first appear in @xmath67 , see for instance eqs .
( [ r2deriv ] ) for @xmath66 .
the nonlinear character of the klrf means that such terms appear in comparatively higher order of the perturbation theory .
for example , @xmath67 contributes to @xmath66 but not to the fisher information .
instead it first contributes to @xmath133 .
we have demonstrated that the fisher information behaves in a qualitatively different way depending on whether the system is perturbed from equilibrium or from an out - of - equilibrium steady state .
for the classically stochastic systems considered here we have seen that in the former case @xmath177 .
that is , the fisher information is a one dimensional function of the two time delays .
this qualitative difference results from the self - duality of equilibrium dynamics , which is another expression of the principle of detailed balance . @xmath178 does not show a similar reduction of dimension [ see eq .
( [ eigenq3 ] ) ] .
this property is special to the fisher information .
we have focused on the properties of the fisher information for overdamped stochastic dynamics .
do other types of systems exhibit similar behavior ? in app .
[ hamiltonianapp ] we consider deterministic hamiltonian systems . in that case
the unitary dynamics results in a fisher information which depends only on the time @xmath92 .
it will be of interest to study how ( not overdamped ) stochastic systems bridge between the overdamped and unitary limits .
we showed that the klrf can serve as a useful measure characterizing the system s dynamics .
they encode information which differs from the information encoded in the orf .
this is demonstrated by the ability of the fisher information to distinguish between systems perturbed out of equilibrium or out of a non - equilibrium steady state .
we expect other useful properties of the klrf to be revealed by further studies .
one of us ( s. m. ) wishes to thank eran mukamel for most useful discussions .
the support of the national science foundation ( grant no . che-0745892 ) and the national institutes of health ( grant no .
gm-59230 ) is gratefully acknowledged .
for completeness , in this appendix we discuss the application of nonlinear response theory to deterministic hamiltonian systems .
we use simple examples to clarify the relation between the response of a system and the kullback - leibler distance .
the hamiltonian of the system is assumed to be of the form @xmath179 , where @xmath166 is the perturbation .
we start with some general comments stemming from the fact that the dynamics in phase space is an incompressible flow .
the state of a classical system is described by its phase space density @xmath33 .
it is important to note that this is the full probability density , which includes information on all the degrees of freedom .
let us denote the propagator of the classical trajectories by @xmath180 , so that @xmath181 where @xmath182 denotes a phase space point ( all coordinates and momenta ) .
similarly , the propagator for the probability distribution is denoted by @xmath183 .
liouville theorem tells us that phase space volumes do not change in time . as a result
, there is a simple relation between the phase space density at different times , @xmath184 the density is simply transported with the dynamics in phase space . the density at @xmath185 at time @xmath6 is equal to the density at @xmath182 at time @xmath186 .
this property of the phase space dynamics have an interesting consequence .
any integral whose integrand depends locally on @xmath33 alone is time independent , since the values of the integrand are just transported around by the dynamics .
an interesting example is the entropy function @xmath187 let us change the integration variable to @xmath188 , which is just the phase space point which would flow to @xmath182 after a time @xmath6 .
liouville theorem assures us that the jacobian of the transformation is unity and therefore @xmath189 where we have used eq .
( [ transport ] ) .
it is clear that this entropy does not depend on time .
( [ constent ] ) is a result of the unitary evolution in phase space .
it connects with all the intricate problems related to the emergence of macroscopic irreversibility out of microscopic reversible dynamics .
this deep problem is beyond the scope of the current paper .
such problems are circumvented when one uses a reduced probability density , whose dynamics is irreversible to begin with .
unitary dynamics lead to an interesting result for the reverse kullback - leibler distance @xmath190 we have seen that the first term is a constant of the motion . assuming a hamiltonian system intially in equilibrium @xmath191 .
we have @xmath192 where @xmath193 is the free energy of the initial equilibrium state .
@xmath194 is therefore linear in @xmath33 , and thus it is equivalent to a calculation of response functions !
the general comments above raise two interesting points .
first , the fact that terms such as @xmath195 are constant means that their expansion in powers of pulse strengths @xmath8 turns out to have only the constant term , all other terms in the expansion must vanish .
the second point of interest has to do with the relation between the kullback - leibler distance and its reverse .
generally , @xmath196 .
we are interested in systems pertubed by a series of pulses and compare the initial and final distributions . in that case we can use unitarity to change variables from the phase space points at the final time to the points at the initial time which are connected to it by the dynamics .
this gives @xmath197 here @xmath198 , that is , @xmath199 is the density that would evolve to the equilibrium density under the influence of the pulses .
it is not equal to @xmath200 , which tells us that the distance and its reverse are not equal .
while in general the distance and its reverse are not equal , when the distance between the distributions is small , its easy to show that their leading order expansion in @xmath201 is the same . loosely speaking @xmath202
we can deduce that the fisher information , which is determined by this leading order , could be obtained from a calculation of response for systems with unitary evolution .
( since it could also be obtained from the reverse distance . )
our last general point is also related to unitarity , but has to do with the fisher information . according to eq .
( [ fisher2 ] ) the fisher information is built from two partial densities .
( more accurately , density differences . )
these evolve with respect to the same hamiltonian @xmath203 between interactions with the pulses . as a result , the integral in eq .
( [ fisher2 ] ) do not depend on @xmath105 , for the same reason that caused the entropy @xmath204 to be time independent .
we find that @xmath205 as a result of the unitarity of the phase space dynamics .
similar independence on the final time interval would also appear for higher order terms in the expansion of the kullback - leibler distance . the simplest system that can serve as an example is a single harmonic oscillator @xmath206 we take the perturbation to be @xmath207 for this system it is trivial to solve for the free evolution @xmath208 this relation can be inverted , expressing @xmath209 in terms of @xmath210 @xmath211 eq . ( [ reverseosc ] ) will be useful when one propagates probability distributions in time .
the equilibrium distribution of the harmonic oscillator is a gaussian @xmath212 we also note that @xmath213 is the linear order correction for the density just after interaction with one pulse . to calculate @xmath68 we need to propagate eq .
( [ tempdrho ] ) . with the help of eqs .
( [ transport ] ) and ( [ reverseosc ] ) we find @xmath214 } \rho_0 ( q , p).\ ] ] in the derivation we have used the fact that @xmath44 is invariant under the evolution with respect to @xmath203 . to calculate @xmath67 we operate with @xmath215 on @xmath68 , and then propagate the resulting correction for the density for a time interval @xmath105 .
a straightforward calculation leads to @xmath216 + \frac{1}{q_0 m \omega } \sin \omega
t_1 \right ) \right\ } \\ & \times & \exp \left[-\frac{q}{q_0 } \left ( \cos \omega t_2 + \cos \omega ( t_1+t_2)\right ) + \frac{p}{m \omega q_0}\left(\sin \omega t_2 + \sin \omega ( t_1+t_2 ) \right)\right ] \rho_0 \nonumber\end{aligned}\ ] ] we now turn to calculate the first two orf , using eqs .
( [ formalr1 ] ) and ( [ formalr2 ] ) .
the calculation of @xmath64 is cumbersome but straightforward .
@xmath217.\ ] ] the calculation of @xmath66 is more involved , and we only include the final result @xmath218 \\
\times \exp \left [ \frac{1}{m \beta \omega^2 q_0 ^ 2 } \left\ { \frac{3}{2 } + \cos \omega t_1 + \cos \omega t_2 + \cos \omega ( t_1+t_2)\right\}\right].\end{gathered}\ ] ] we would like to compare these response functions to the klrf , and in particular to the fisher information , which can be calculated using eq .
( [ fisher2 ] ) .
the calculation can be simplified by using the classical coordinates right after the second interaction with the pulse as integration variables . due to unitarity
, the fisher information only depends on the time difference @xmath92 , see also the discussion in the previous subsection .
we get @xmath219 \exp \left [ \frac{1}{\beta m \omega^2
q_0 ^ 2 } \left ( 1+\cos \omega t_1\right)\right].\ ] ] this expression seems similar to the first order response function .
one can indeed show that they are related by @xmath220 it will be interesting to check whether this expression could be generalized to any hamiltonian system ( with unitary dynamics ) .
in the main text we have calculated the orf and klrf to second order . in the appendix we present the third order quantities , @xmath134 , and @xmath133 .
following the calculations performed in sec .
[ formalsec ] , the third order response functions are given by @xmath221 and @xmath222 \right .
\\ \left . -2
\rho_0^{-2}(x ) { \cal s}^{(1)}(x;t_1+t_2+t_3 ) { \cal s}^{(1)}(x ; t_2+t_3 ) { \cal s}^{(1)}(x ; t_3 ) \right\},\label{q123}\end{gathered}\ ] ] where @xmath223 here we present expressions for the third order orf and klrf in term of the eigenvalues and eigenfunctions of the stochastic dynamics . at high orders the nonlinear character of the contributions to the kullback - leibler distance result in integrals with products of several eigenfunctions . here
we will only need the one with three eigenfunctions @xmath224 the third order orf is given by @xmath225 this orf should be compared to the third order klrf , which is calculated using eq .
( [ q123 ] ) .
we find @xmath226 } \right .
+ e^{-\lambda(n ) \left [ t_1 + t_2 \right ] } e^{-\lambda ( m ) \left [ t_2 + 2 t_3\right ] } + e^{-\lambda ( n ) t_1 } e^{-\lambda ( m ) \left [ t_2 + 2 t_3\right ] } \right\ } \\ - 2 \sum_{nml } { \cal b}_{n0 } { \cal b}_{m0 } { \cal b}_{l0 } { \cal j}_{nml } e^{-\lambda ( n ) \left [ t_1+t_2 + t_3\right ] } e^{-\lambda ( m ) \left [ t_2 + t_3\right ] } e^{-\lambda ( l ) t_3}. \label{eigenq3}\end{gathered}\ ] ] the expression for @xmath133 , presented in eq .
( [ eigenq3 ] ) , has three terms in which the orthogonality condition ( [ orthleft ] ) has been used , pointing to a reduction of dimension in the time dependence of this specific term . however , the time combinations in these terms are all different . in addition , the fourth term in eq .
( [ eigenq3 ] ) clearly depends on all its time variables .
we conclude that in contract to the fisher information , the higher order klrf depend on all their time variables .
the reduction of dimension is therefore specific for the fisher information .
it results from the fact that it is built out of a single product of two density corrections .
in this appendix we consider a simple example of an overdamped harmonic oscillator with a potential @xmath227 , with a perturbing potential @xmath228 . for this system it is possible to write explicit expressions for the eigenvalues and eigenfunctions of the fokker - planck operator , as well as to perform several of the integral , defined in sec .
[ overdampedsec ] . in this case @xmath229 \\
\hat{\cal l}_a \rho & = & - \frac{\alpha}{q_0 \gamma } \frac{\partial}{\partial q } \left [ e^{-q / q_0 } \rho \right].\end{aligned}\ ] ] this model has been studied extensively @xcite .
the equilibrium density is given by @xmath230 the left eigenfunctions , and the eigenvalues , of this model are @xmath231 here , @xmath232 are the hermit polynomials
. for instance , @xmath235 one can substitute eq .
( [ hn2 ] ) for the hermit polynomial , and use integration by parts .
this leads to @xmath236 this is a gaussian integral which is easily evaluated to give @xmath237 one can also easily calculate the integrals @xmath238 by the same technique .
one finds @xmath239 comparing eqs .
( [ osccalc ] ) and ( [ osccalb0 ] ) , we see that for this model @xmath240 as a result , there is a simple relation between the fisher information and the first order response function , @xmath241 this relation is a special result for this model , and is not expected to hold for other systems .
one can also obtain explicit results for @xmath242 with non vanishing indices .
however , the calculation and the result are quite combersome , and are omitted .
we were not able to calculate @xmath243 explicitly . | by subjecting a dynamical system to a series of short pulses and varying several time delays we can obtain multidimensional characteristic measures of the system .
multidimensional kullback - leibler response function ( klrf ) , which are based on the kullback - leibler distance between the initial and final states , are defined .
we compare the klrf , which are nonlinear in the probability density , with ordinary response functions ( orf ) obtained from the expectation value of a dynamical variable , which are linear .
we show that the klrf encode different level of information regarding the system s dynamics . for overdamped stochastic dynamics
two dimensional klrf shows a qualitatively different variation with the time delays between pulses , depending on whether the system is initially in a steady state , or in thermal equilibrium . |
for significant quantum effects to appear in the transport properties of a conductor , it must be shorter than the distance over which an electron propagates without losing its quantum coherence , which for usual metals at low temperatures ( < 1k ) is on the order of microns .
the exploration of this quantum transport regime started in the early 1980s with the observation of aharonov - bohm interferences in submicron metallic rings @xcite .
a very powerful point of view , put forward by r. landauer @xcite , describes a transport experiment across a coherent conductor as the scattering of the quantum electronic waves injected by the contact probes . in other words ,
the conductor is viewed as an electronic waveguide , whose modes , called `` conduction channels '' , are each characterized by a transmission probability . for example , the total conductance is given by the famous landauer formula : @xmath0 , where @xmath1 is the number of channels , the @xmath2 are the individual transmission probabilities , and @xmath3 the conductance quantum .
in fact , in systems with non - interacting quasiparticles all transport quantities can be expressed in terms of the transmission probability set @xmath4 , which is viewed in this framework as the personal identity number ( pin code ) of the conductor @xcite . for most systems ,
the number of conduction channels is very large , and the description of transport can only be based on a distribution function for the transmission eigenvalues . however , the number of channels can be significantly reduced if the lateral dimensions of a conductor become comparable to the fermi wavelength of the electrons .
this was demonstrated in 1988 using constrictions tailored in the two - dimensional electron gas of a semiconductor heterostructure , with a width adjustable on the scale of the fermi wavelength ( @xmath5 ) by means of an electrostatic gate @xcite . in these `` quantum point contacts '' , as the constriction width is continuously decreased , a descending staircase conductance curve is observed , beautifully illustrating the closing of the channels one after another . since
then much activity has been devoted worldwide to the investigation of transport through a broad range of coherent conductors ( diffusive and ballistic conductors , quantum dots , nanowires , individual molecules like carbon nanotubes ) connected to reservoirs in the normal or in the superconducting state @xcite . among the various systems investigated ,
atomic - size contacts @xcite between three - dimensional metallic electrodes play an important role .
these contacts have been accessible since the early years of stm @xcite , and more stable ones were achieved at the beginning of the 1990s by the group of jan van ruitenbeek at leiden using the break - junction technique @xcite .
since their characteristic transverse dimensions are of the order of the fermi wavelength ( typically @xmath6 ) , atomic contacts accommodate only a small number of channels and behave as quantum point contacts even at room temperature .
an interesting difference with respect to quantum point contacts in 2d electron systems is that superconductivity can be built - in with the proper choice of material .
the discovery by our group in 1997 that the pin - code could be accurately measured for every contact @xcite paved the way to a new generation of quantum transport experiments in which the measured transport quantities could be compared to the theoretical predictions with no adjustable parameters . as an example , we describe here our experiments on the supercurrent flowing through atomic - size contacts .
they probe the main concepts of the `` mesoscopic '' theory of the josephson effect , _
i.e. _ the theory in terms of the transmission probability of the channels , that we sketch in section 3 .
before discussing previous and on - going experiments in sections 4 and 5 respectively , we describe the microfabrication technique that has made possible the experiments and the procedure we use to determine the pin - code of atomic - size contacts in section 2 .
in order to produce atomic contacts we developed the microfabricated break - junction ( mbj ) technique @xcite . using electron beam lithography and reactive ion etching , a metallic bridge clamped to an elastic substrate is suspended over a few microns between two anchors ( see left panel of fig .
[ fig : mbj ] ) .
the bridge has a constriction at the center with a diameter of approximately 100 nm . using the mechanism shown in the center panel of fig .
[ fig : mbj ] the substrate is first bent until the bridge breaks at the constriction .
this is performed under cryogenic vacuum so that there is no contamination of the two resulting electrodes which are then gently brought back into contact .
afterward , the contact can be repeatedly made and broken at will .
of the pushing rod results in a @xmath7 change of the distance between the electrodes of the bridge . *
right : * conductance of a bridge , in units of the conductance quantum , as a function of displacement when pulling the electrodes apart . the origin of the displacement was set arbitrarily at the point where the bridge actually breaks and the conductance starts decreasing exponentially signaling the tunnel regime .
measurements were performed at 50 mk under a magnetic field of 200 mt to suppress superconductivity in the aluminum films .
, title="fig : " ] of the pushing rod results in a @xmath7 change of the distance between the electrodes of the bridge .
* right : * conductance of a bridge , in units of the conductance quantum , as a function of displacement when pulling the electrodes apart . the origin of the displacement was set arbitrarily at the point where the bridge actually breaks and the conductance starts decreasing exponentially signaling the tunnel regime .
measurements were performed at 50 mk under a magnetic field of 200 mt to suppress superconductivity in the aluminum films .
, title="fig : " ] of the pushing rod results in a @xmath7 change of the distance between the electrodes of the bridge . *
right : * conductance of a bridge , in units of the conductance quantum , as a function of displacement when pulling the electrodes apart . the origin of the displacement was set arbitrarily at the point where the bridge actually breaks and the conductance starts decreasing exponentially signaling the tunnel regime .
measurements were performed at 50 mk under a magnetic field of 200 mt to suppress superconductivity in the aluminum films .
, title="fig : " ] the right panel of fig .
[ fig : mbj ] displays a typical conductance evolution observed while opening an aluminum mbj .
the conductance evolves through a series of slanted plateaus and sharp steps .
the conductance on the last plateau before full opening and the height of the previous steps is of the order of the conductance quantum @xmath8 .
although similar staircase patterns are observed every time the experiment is performed , the horizontal extension ( of the order of a fraction of nanometer ) and the vertical position of the plateaus are not always the same for subsequent compression - extension cycles .
these general features have been observed by many groups for a variety of metals under different experimental conditions ( temperature , technique used to produce atomic size contacts , rate of compression and extension ) .
the typical conductance on the last plateau , the typical lengths of the plateaus , and the behavior within the plateaus are characteristic of each material @xcite .
the succession of plateaus and conductance jumps is directly related to the dynamics of the atomic configuration of the contact .
combined stm - afm experiments that measure the force between the tip and the surface simultaneously with the conductance have directly shown that on a plateau the atomic configuration is only elastically deformed while a conductance jump results from an abrupt reconfiguration of the atoms at the contact accompanied by stress relief @xcite .
molecular dynamics simulations @xcite confirm this interpretation of the staircase pattern .
these were the first clues that the smallest contacts are indeed made of a single atom . in the case of gold ,
one - atom contacts and atomic chains @xcite have even been observed directly with high resolution transmission electron microscopes @xcite . as compared to other techniques , microfabricated break junctions present several major advantages essential to the realization of the experiments presented here .
first , atomic - size contacts fabricated this way are extremely stable and can be maintained in the same configuration for weeks .
second , with a given suspended bridge many different contacts can be created in - situ , allowing exploration of the physics of interest for a broad range of transmission coefficients .
finally , the flexibility offered by microfabrication techniques makes possible embedding the contacts into on - chip circuits .
characteristics of an atomic contact .
* open symbols * : experimental data .
* full red line * : best fit resulting from the sum of three single channel characteristics * ( dashed lines ) * calculated using the theory of mar and the transmission values indicated on the right edge.[fig : fitmar ] ] our first major result was devising a method to determine not only the number of channels accommodated by a contact but also its pin code with good accuracy @xcite .
this was achieved by measuring the current - voltage characteristic @xmath9 of the contact in the superconducting state ( see fig . [
fig : fitmar])[multiblock footnote omitted ] . at exactly zero voltage a finite current , _
i.e. _ a supercurrent , can flow . above some bias current
, the system switches to a finite voltage state , _
i.e. _ the current becomes dissipative__. _ _ the voltage and transmission dependence of the dissipative current through a weak link was explained for the first time in the early 1980s in terms of multiple andreev reflections ( mar ) @xcite .
a full quantum theory of mar for a single channel of arbitrary transmission was achieved in the mid 90s @xcite .
three numerically calculated current - voltage characteristics resulting from these mar processes are shown in fig .
[ fig : fitmar ] .
these elementary curves are highly non - linear below twice the superconducting gap @xmath10 and present current steps at voltages @xmath11 , with @xmath12 an integer , which mark the onset of mar of different orders .
the n - th order process involves the transfer of @xmath12 electrons , and in a given channel its intensity varies as the n - th power of the transmission .
consequently , the @xmath13 characteristic depends on all powers of every transmission coefficient and carries sufficient information to reconstruct the pin code .
the determination of the pin code of any atomic - size contact is achieved by decomposing the measured total current into a series of such elementary characteristics , each of them corresponding to a well defined transmission probability .
the individual transmission probabilities of the channels are adjusted so as to get the best fit of the measured current - voltage characteristic @xcite .
this automatically yields the number of channels having a non - negligible transmission .
the uncertainty on the transmissions obtained by this fitting procedure depends on the uncertainty in the measurement of both voltage and current , on the transmission of the channels participating in the contact , and on the environmental impedance .
typically , for channels with a large transmission the relative uncertainty achieved on @xmath2 is of the order of @xmath14 for channels having low transmissions @xmath15 , the procedure fails to disentangle the contributions of the different channels , and yields a larger error bar ( see chapter 1 of @xcite for details ) . using this method we showed that for one - atom contacts the number of conduction channels is directly related to the number of valence orbitals of the metal @xcite . for aluminum and lead , which have p - electrons at the fermi level
, three channels were found to contribute to the conductance of the smallest contacts . for niobium , a transition metal with s- and d - electrons , five conduction channels were found . in the case of gold , with one s - electron , and for which superconductivity was induced through the proximity effect by intimate contact with a thick aluminum layer , we found that only one conduction channel contributes to the conductance on the lowest conductance plateau @xcite .
in 1962 , josephson predicted that a dc current could flow at zero bias voltage between two superconducting electrodes coupled by a tunnel barrier @xcite .
this supercurrent results from the coherent transfer of cooper pairs and is driven by the phase difference @xmath16 between the superconducting order parameters of the two electrodes .
the phase difference is related to the voltage difference @xmath17 between the electrodes by the josephson relation @xmath18 , with @xmath19 the flux quantum . since its prediction
, the josephson current has been observed in a great variety of systems involving superconducting electrodes connected by a `` weak link '' .
a weak link can be an insulating layer , as originally proposed by josephson , or any quantum coherent conductor , such as a short normal ( diffusive or ballistic ) metallic wire , a point contact , a carbon nanotube , or a graphene layer @xcite .
a great deal of theoretical activity has been devoted to relate the maximum ( or critical ) supercurrent @xmath20 of a weak link to its normal state properties .
a unifying theoretical description providing this relation for an arbitrary structure in terms of its transmission pin emerged in the 1990s and is based on the concept of andreev bound states @xcite . due to the existence of a gap in the quasiparticle density of states in a superconductor ,
an electron impinging from a normal metal into a superconducting electrode with an energy @xmath21 can not enter it as such , and should in principle be reflected back .
however , if there is no barrier at the interface it will be `` andreev reflected '' as a hole at the same energy and moving in the opposite direction , adding a cooper pair to the superconductor @xcite .
the formation of andreev bound states is illustrated for a single channel in fig .
[ fig : andreevloopbal ] @xcite . and
the wiggly lines represent an andreev reflection in which an electron ( hole ) is reflected as a hole ( electron ) acquiring the local superconducting phase .
the upper ( lower ) loop corresponds to the transfer of cooper pairs to the right ( left ) . when the transmission probability of the channel is below unity , in addition to the andreev reflection processes , normal reflection processes ( dashed lines ) connects electron ( hole ) states traveling in opposite directions.[fig : andreevloopbal ] ] in a perfectly transmitting channel , a right - moving electron is andreev reflected at the interface with the right superconducting electrode into a left - moving hole , and acquires an energy dependent phase shift @xmath23 where @xmath22 is the phase of the local superconducting order parameter . in turn
, this left - moving hole is andreev reflected as a right - moving electron at the left electrode , removing a cooper pair from it .
these successive reflections interfere constructively , like in a fabry - prot interferometer , when the global phase shift acquired along one round - trip is an integer multiple of @xmath24
. of course , a similar process occurs for left - moving electrons reflected as right - moving holes , and eventually two resonant quasiparticle states @xmath25 appear in the channel region .
their phase - dependent energies lie within the superconducting energy gap and are symmetric with respect to the fermi level . in a single channel of transmission @xmath26 ( solid curves ) as a function of the superconducting phase difference @xmath27 across it .
the dashed curves represent those in a reflectionless channel @xmath25 . *
right : * supercurrent carried by the upper @xmath28 and lower @xmath29 andreev states calculated for @xmath30 ( full lines ) .
the dashed curves are the currents for the reflectionless channel.[fig : andreev],title="fig : " ] in a single channel of transmission @xmath26 ( solid curves ) as a function of the superconducting phase difference @xmath27 across it .
the dashed curves represent those in a reflectionless channel @xmath25 . *
right : * supercurrent carried by the upper @xmath28 and lower @xmath29 andreev states calculated for @xmath30 ( full lines ) .
the dashed curves are the currents for the reflectionless channel.[fig : andreev],title="fig : " ] if the channel is not perfectly transmitting ( @xmath31 ) , a right - moving electron can also be simply reflected as a left - moving electron .
the existence of a finite reflection probability thus couples the two reflectionless states .
the two resulting states , denoted @xmath32 , are called the `` andreev states '' , and have energies @xmath33 shown in fig .
[ fig : andreev ] , where : @xmath34 let us point out that in the tunnel limit @xmath35 this expression correctly yields the @xmath36 dependence of the coupling energy predicted by josephson for a tunnel junction .
the transition frequency reaches its minimum @xmath37 at @xmath38 . because the number of transferred pairs and the phase difference are conjugated variables , these states carry supercurrents given by : @xmath39 and shown in the right panel of fig . [
fig : andreev ] . since at a given phase
the two andreev bound states carry the same current but in opposite directions , the net supercurrent results from the imbalance of their populations . at zero temperature ,
only the lower energy state is occupied and the maximum ( or critical ) supercurrent is given by@xmath40 which is not simply proportional to @xmath41 and thus neither to the normal state conductance . in aluminum , where the superconducting gap @xmath10 is typically of the order of @xmath42 , this current reaches at most @xmath43 for a single perfectly transmitting channel , and the transition frequency is in the microwave range . ) .
_ _ * _ ( b ) _ * _ _ excitation spectrum : besides the usual continuum of excited states above the energy gap @xmath10 , that extends all across the structure , there is at the constriction a discrete andreev spin - degenerate doublet with energy @xmath44 above the ground state , where quasiparticles can get trapped . *
_ ( c ) _ * four possible configurations of the andreev doublet at the constriction : in the global ground configuration , labeled @xmath45 , the doublet is empty ( it corresponds to just the ground andreev state @xmath29 of fig . [
fig : andreev ] being occupied ) ; in the two odd configurations , each with definite spin @xmath46 and labeled @xmath47 and @xmath48 , one quasiparticle is added to the contact ( they correspond to the andreev states of fig .
[ fig : andreev ] being both occupied or both empty ) @xcite ; the last configuration is a spin - singlet double excitation , labeled @xmath49 ( it corresponds to just the excited andreev state @xmath28 of fig .
[ fig : andreev ] being occupied ) .
* _ ( d ) _ * total energy of the four configurations vs phase .
the energy of the ground configuration @xmath45 is that of the lowest andreev level @xmath50 .
the two odd configurations , @xmath47 and @xmath48 , have zero energy .
the double excitation configuration @xmath49 , has the energy of the excited andreev state @xmath51 . ]
it is important to note that the pair of andreev bound states leads locally to an excitation spectrum ( fig.[fig : excitation spectrum ] ) containing , at an energy @xmath52 above the ground state , a discrete spin - degenerate doublet @xcite .
there are four possible occupation configurations for the system . in the global ground state configuration ,
the total energy contains a phase dependent term @xmath53 arising from the lowest lying localized andreev state , which is occupied by a spin - singlet pair of electrons and gives rise to a supercurrent @xmath54 with @xmath55 then , there are two `` odd '' configurations ( spin 1/2 ) with a single excitation of the doublet at @xmath56 , _ _ corresponding to a quasiparticle trapped in the one - channel constriction . in this case
the global energy is zero , _ _ phase independent , and the total supercurrent is zero .
finally , there is another spin - singlet configuration with a double excitation which carries a supercurrent @xmath57 exactly opposite to the one in the ground configuration .
hence the supercurrent through the constriction is a probe of the excitation configuration of the system .
when the structure is voltage - biased at @xmath58 , the phase varies linearly with time at a speed @xmath18 .
when the latter is low enough , one can assume that the andreev levels move adiabatically within the superconducting gap @xmath10 .
as the motion is periodic , there is no energy transfer to the system on average and a purely ac current flows .
this corresponds to the second striking prediction of josephson , the ac josephson effect . for larger voltages , and
therefore larger speeds , non - adiabatic transitions ( landau - zener type ) arise between the andreev levels , giving rise to the onset of the dissipative mar current discussed in section 2.2 @xcite . the onset of this dissipative current depends on the channel transmission .
for an arbitrary quantum coherent conductor the phase - driven supercurrent is governed by the occupation numbers of the andreev states in all channels . at very low temperature ,
when only the lower energy andreev state of each channel is occupied , the current - phase relation is : @xmath59 this is one of the fundamental predictions of the mesoscopic theory of the josephson effect that we have probed in our experiments with atomic contacts . in a first series of experiments carried out on current - biased contacts @xcite , we measured the average supercurrent @xmath60 at which the system switches to the dissipative branch . in this case
, the phase is not an externally tunable parameter , but a dynamical variable moving in a potential landscape fixed by the total josephson coupling energy and submitted to random fluctuations imposed by the biasing circuit . as a consequence
the switching current is always smaller than the critical current @xmath61 however , because we embedded the contact in a properly designed on - chip circuit , dissipation and thus fluctuations were under control .
most of the experimental results could be well understood by considering just the contribution of the ground andreev state of each channel , however , these experiments were actually only an indirect test of the prediction of eq .
( [ eq : curr - phase n ] ) , since the phase was not an external parameter that could be swept over its entire range .
a thorough test of the theory actually requires measuring , for a given atomic contact , both its current - phase relation and its @xmath13 characteristic ( to determine its pin code ) .
these two measurements require contradictory biasing conditions . for the former
, the atomic contact must be phase - biased , which requires inserting it in a small superconducting loop threaded by a magnetic flux . for the latter
, one needs to voltage - bias the same atomic contact , which can not be achieved if it is shunted by the superconducting loop .
obviously , one needs a superconducting reversible switch in order to toggle between the two biasing conditions without disturbing the atomic contact . in the setup presented in fig .
[ fig : setupac1 ] a josephson junction acts as such a switch @xcite .
the atomic contact and the josephson junction are embedded in a small superconducting loop , forming a device that was coined the `` atomic - squid '' .
this ancillary junction not only allows both biasing configurations but is also used to measure the loop current .
the critical current of the josephson junction is chosen to be much larger than the one of a typical aluminum one - atom contact ( @xmath62 ) so that the squid essentially behaves like a slightly perturbed josephson junction .
forms an atomic - squid .
an on - chip capacitor @xmath63 lowers the plasma frequency of the junction to @xmath64 and the associated resistance @xmath65 damps the dynamics of the phase .
the squid is biased through a resistor @xmath66 by a current @xmath67 , and the voltage @xmath68 across it is monitored to detect switching . *
right : * sem micrographs of sample # 1 at different scales . in the upper - right panel ,
the brighter pads are gold electrodes , while the darker part constituting the loop is made out of aluminum .
the whole structure is deposited on top of a polyimide layer on a metallic substrate .
the gold electrodes form the shunting capacitor @xmath69 through the metallic substrate , and also act as quasiparticle traps ( see @xcite for details ) .
the junction ( lower - left panel ) is fabricated using the technique of double - angle evaporation through a suspended mask .
this also leads on the suspended bridge side to an extraneous metallic bridge , as seen on the bottom - right panel.[fig : setupac1],title="fig : " ] forms an atomic - squid .
an on - chip capacitor @xmath63 lowers the plasma frequency of the junction to @xmath64 and the associated resistance @xmath65 damps the dynamics of the phase .
the squid is biased through a resistor @xmath66 by a current @xmath67 , and the voltage @xmath68 across it is monitored to detect switching .
* right : * sem micrographs of sample # 1 at different scales . in the upper - right panel ,
the brighter pads are gold electrodes , while the darker part constituting the loop is made out of aluminum .
the whole structure is deposited on top of a polyimide layer on a metallic substrate .
the gold electrodes form the shunting capacitor @xmath69 through the metallic substrate , and also act as quasiparticle traps ( see @xcite for details ) .
the junction ( lower - left panel ) is fabricated using the technique of double - angle evaporation through a suspended mask .
this also leads on the suspended bridge side to an extraneous metallic bridge , as seen on the bottom - right panel.[fig : setupac1],title="fig : " ] the @xmath13 characteristic of the atomic contact , denoted @xmath70 , is obtained from that of the squid @xmath71 . in principle , in the region below the superconducting gap @xmath72 , the dc current flowing through the junction is expected to be zero . in practice however , a sizable current is observed experimentally in this region in the characteristic @xmath73 of the junction alone , which can be measured when the metallic bridge forming the atomic contact is fully open .
assuming that @xmath74 is not affected by the contact , @xmath70 is then obtained by the subtraction:@xmath75 which is then fitted using the mar theory to obtain the transmissions .
the superconducting loop allows imposing a phase difference across the atomic contact by applying an external magnetic flux @xmath76 .
if the loop is sufficiently small so that the screening flux can be neglected of the loop , typically @xmath77 is chosen to be negligible as compared to the inductance of both the josephson junction @xmath78 and the atomic contact @xmath79 . in this way
, the phase - drop takes place essentially across these two last elements . ] , the phase differences @xmath80 ( across the tunnel junction ) and @xmath27 ( across the atomic - size contact ) are linked through:@xmath81 where @xmath82 is the reduced flux threading the loop @xcite . at zero temperature ,
the large josephson junction would switch out of its zero - voltage state at a phase difference @xmath83 , when it carries exactly its critical current @xmath20 .
assuming that the contributions of the junction and of the atomic contact can be separated , the critical current @xmath84 of the squid is the sum of the critical current @xmath20 of the junction and of the flux - dependent critical current @xmath85 of the atomic contact:@xmath86 measuring the flux dependence of the critical current of the squid would thus be a direct way to probe the current - phase relation of the atomic contact .
however , as mentioned before , in practice the critical current of a small josephson device is not accessible . due to fluctuations the system switches stochasticaly to a dissipative state before reaching this maximum current , and one simply measures the mean switching current of the squid .
( a ) for an atomic squid obtained with sample * # 1 * as a function of the reduced bias current @xmath87 , for two values of the external flux . *
right : * switching probability ( color - coded ) as a function of both the reduced bias current and flux .
the two vertical lines correspond to the two curves plotted in the left panel .
the full black line corresponds to the modulation curve @xmath88 calculated using the full pin of the contact , taking into account the contribution of just the andreev ground state in each channel .
the pin code of the atomic contact was @xmath89 * * * * + * ( b ) left : * switching probability @xmath90 for an atomic - squid obtained with sample * # 2 * as a function of the reduced bias current @xmath87 , for two values of the external flux .
the upper curve is similar to those obtained in sample * # 1 * but the lower curve presents a clear intermediate plateau .
the pin code of the atomic contact was @xmath91 * right : * switching probability ( color - coded ) as a function of both the reduced bias current and flux .
the two vertical lines correspond to the two curves plotted in the left panel .
the full line corresponds to the modulation curve @xmath88 calculated taking into account the contribution of the andreev ground state of all the channels .
the dashed line corresponds to the modulation curve calculated for @xmath92 , which excludes the contribution of the most transmitted channel.[fig : modpswit],title="fig : " ] ( b ) for an atomic squid obtained with sample * # 1 * as a function of the reduced bias current @xmath87 , for two values of the external flux . *
right : * switching probability ( color - coded ) as a function of both the reduced bias current and flux .
the two vertical lines correspond to the two curves plotted in the left panel .
the full black line corresponds to the modulation curve @xmath88 calculated using the full pin of the contact , taking into account the contribution of just the andreev ground state in each channel .
the pin code of the atomic contact was @xmath89 * * * * + * ( b ) left : * switching probability @xmath90 for an atomic - squid obtained with sample * # 2 * as a function of the reduced bias current @xmath87 , for two values of the external flux .
the upper curve is similar to those obtained in sample * # 1 * but the lower curve presents a clear intermediate plateau .
the pin code of the atomic contact was @xmath91 * right : * switching probability ( color - coded ) as a function of both the reduced bias current and flux .
the two vertical lines correspond to the two curves plotted in the left panel .
the full line corresponds to the modulation curve @xmath88 calculated taking into account the contribution of the andreev ground state of all the channels .
the dashed line corresponds to the modulation curve calculated for @xmath92 , which excludes the contribution of the most transmitted channel.[fig : modpswit],title="fig : " ] the mean switching current is determined accurately from the switching probability @xmath90 .
this is measured using a train of bias - current pulses of a given height and counting the number of pulses for which a voltage appears across the squid .
we plot in the right panel of fig .
[ fig : modpswit](a ) the switching probability as a function of both the reduced bias current @xmath93 and the reduced flux @xmath82 in the loop , for a particular atomic - contact obtained with sample # 1 the left panel of the same figure shows two examples of @xmath94 for two different values of flux .
one observes that the switching probability evolves very rapidly from 0 to 1 in a narrow range of bias current which depends on the applied flux . in the following ,
we denote @xmath95 the reduced bias current leading to an switching probability @xmath96 , and refer to @xmath88 as the squid `` modulation curve '' . in order to calculate the expected modulation curve one considers the phase dynamics in the full josephson potential of the squid in the presence of thermal fluctuations @xcite .
the potential is dominated by the josephson energy of the junction but contains a contribution which consists of the sum of the ground sate andreev energies of every channel of the atomic contact . as shown in the right panel of fig .
[ fig : modpswit](a ) this model works very well . in conclusion for all the contacts that we investigated using sample # 1
the current - phase relation measured at low temperatures is in quantitative agreement with the mesoscopic description of the josephson effect , considering that only the ground andreev state is occupied in each channel .
the next step in order to achieve a comprehensive test of the theory would be to probe the excited andreev states for example by directly performing spectroscopy of the transitions between andreev levels .
however , during our first attempts to do so , we observed unexpected but important phenomena which shed light on the microscopic nature of the system .
the experiments presented in the previous sections were performed with atomic - squids obtained on sample # 1 , which was fabricated on top of a metallic substrate and in which the on - chip lines connecting the squid to the outside world were made out of gold , a dissipative metal at low temperature . on the contrary in our first design to attempt microwave andreev spectroscopy , in order to minimize dissipation which is expected to limit the life - time of the excited andreev levels @xcite and
thus broaden the spectral lines , we decided to fabricate the samples on kapton substrates with all on - chip electrodes made out of aluminum .
apart from this change of material sample # 2 was almost identical to the previous one and in particular the critical current of the junction was @xmath97 , less than @xmath98 smaller than before .
however , as we will discuss now , this change of material had strong consequences on the overall behavior of the system .
period ) each voltage pulse corresponds to a switching event . *
left * : at a flux value for which the @xmath99 curve is normal .
* right * : similar measurement , at a flux value for which @xmath99 displays a step , showing intermittency attributed to the dynamics of single quasiparticle trapping and untrapping in the andreev doublet.[fig : bunching ] ] the right panel of fig .
[ fig : modpswit](b ) shows the measured switching probability as a function of both the flux and the reduced bias current for a squid with an atomic contact of transmissions @xmath100 in sample#2 . as compared to the previous sample ( see fig .
[ fig : modpswit](a ) ) the switching probability @xmath99 does not vary sharply from 0 to 1 for all values of the flux . in some flux ranges , it displays a step as a function of the bias current , as shown for @xmath101 in the left panel of the same figure .
these features appear only in a region bounded by the expected modulation curve @xmath88 and by a second modulation curve calculated without the contribution of the highest transmitted channel .
in other words , in this region the highest transmitted channel does not seem to always contribute to the total current through the contact .
this is exactly what is expected if the two andreev states of this channel are either both empty or both occupied , an occupation configuration achieved when a single quasiparticle is trapped in the andreev doublet as discussed in section 3.2 . since the andreev states have energies below the superconducting gap the quasiparticles are easily trapped .
the fact that this effect only appears on samples having no large normal electrodes able to efficiently remove the quasiparticles strongly supports this hypothesis .
moreover , if trapping happens in a flux region for which the critical current gets enhanced by the double occupancy , subsequent current pulses of the same height would be unable to make the system switch .
it is only after the quasiparticle gets untrapped that switching occurs again .
this results in intermittency of switching as quasiparticles trap and untrap , as shown in fig .
[ fig : bunching ] .
recently , we have measured the full dynamics of trapping and found that the lifetime of trapped quasiparticles can be quite long , on the order of @xmath102 @xcite .
as shown in the previous section , up to now only the properties of the andreev ground state have been probed in detail . performing spectroscopy of the transition between two andreev states
is clearly the next important step which we are presently pursuing along two different lines .
the andreev frequency ( magenta dashed line ) is for a channel of @xmath103 and the coupling energy is @xmath104 note the anti - crossing signaling the hybridization of the two quantum systems.,title="fig : " ] the andreev frequency ( magenta dashed line ) is for a channel of @xmath103 and the coupling energy is @xmath104 note the anti - crossing signaling the hybridization of the two quantum systems.,title="fig : " ] the idea here is to couple an atomic squid to the electromagnetic field of a coplanar waveguide resonator ( left panel of fig .
[ fig : cavity ] ) . by varying the flux threading the squid loop the andreev transition frequency can be brought into resonance with one of the resonator s modes .
this will result in the hybridization between the andreev levels and the modes of the cavity ( right panel of fig .
[ fig : cavity ] ) .
the goal of the experiment would be to detect this hybridization , a strategy that has been implemented successfully for superconducting qubits @xcite .
the required technical developments are underway .
cooper pairs can tunnel only if their energy drop @xmath105 is absorbed as photons in the modes of its electromagnetic environment . *
right : * measured current ( color - coded ) through the junction as a function of the voltage @xmath106 across it ( vertical axis ) and of the magnetic flux threading the squid ( horizontal axis ) .
the flux dependent lines correspond to the excitation of the plasma mode of the squid .
an example of the junction @xmath107 ( for a flux @xmath108 ) is shown in the rightmost panel.,title="fig : " ] cooper pairs can tunnel only if their energy drop @xmath105 is absorbed as photons in the modes of its electromagnetic environment . *
right : * measured current ( color - coded ) through the junction as a function of the voltage @xmath106 across it ( vertical axis ) and of the magnetic flux threading the squid ( horizontal axis ) .
the flux dependent lines correspond to the excitation of the plasma mode of the squid .
an example of the junction @xmath107 ( for a flux @xmath108 ) is shown in the rightmost panel.,title="fig : " ] the second strategy that we are implementing to perform spectroscopy avoids these challenges by using an on - chip microwave source and detector . usually there is no current through a josephson junction voltage biased below twice the superconducting gap . however
, current peaks can develop in its sub - gap _ @xmath9 _ curve at voltages @xmath109 if the energy @xmath105 lost by a cooper pair tunneling across the barrier can be absorbed as photons by the modes @xmath110 of the electromagnetic environment of the junction , @xcite . by capacitively coupling an ancillary josephson junction to the atomic - squid ( see fig . [
fig : icpt ] ) we expect to detect in the _ i - v _ of the junction the andreev mode of the atomic contact . for the moment we have simply tested this idea on a conventional two - junction squid and detected , as shown in the right panel of fig .
[ fig : icpt ] , its plasma resonance corresponding to the oscillations of the lc circuit formed by the josephson inductance of the junctions and their intrinsic capacitance .
atomic - size contacts between metallic electrodes are now routinely obtained by a variety of techniques . because the number of conduction channels they accomodate can be adjusted in - situ , and their corresponding transmission coefficients are amenable to precise measurement , these structures are model quantum point contacts which can be considered as a test - bed for mesoscopic physics . besides the experiments described here
, we have also tested quantitatively the predictions for other transport phenomena : in all cases the agreement between theory and experiment is remarkable and the ensemble of results strongly supports the scattering theory of quantum transport . in the superconducting state
atomic contacts are the simplest possible weak - links and perfectly illustrate the key role played by the andreev bound states in the josephson effect .
the andreev states in a short single channel josephson structure constitute a two - level system @xcite that has been proposed as the basis for a new class of superconducting qubits @xcite .
what is particularly interesting and novel is that in contrast with all other superconducting qubits based on josephson junction circuits @xcite an andreev qubit is a microscopic two level system more like spin qubits in semiconducting quantum dots .
it can be viewed as a superconducting `` quantum dot '' possibly allowing manipulation of the spin degree of freedom of a single quasiparticle @xcite . as reviewed in this work we have already characterized andreev states by measuring the current they carry but the coherence properties of andreev doublets are a key issue still to be addressed .
the relaxation time of the excited state and the dephasing time of a quantum superposition of the two states have to be measured , understood and if possible controlled .
* acknowledgments*. we gratefully acknowledge the expert technical support of pief orfila and pascal senat , the input and assistance of other members of the quantronics group ( michel devoret , patrice bertet and denis vion ) and the crucial contributions of the graduate students and postdocs who participated in this research effort in the past : elke scheer , ronald cron , peter vom stein , martin chauvin , maria luisa della rocca , benjamin huard , maciej zgirski and quentin le masne .
we have enjoyed along the years fruitful collaborations with nicols agrat , juan carlos cuevas , alfredo levy yeyati , alvaro martn rodero , gabino rubio and jan van ruitenbeek , and greatly benefited from discussions with joachim ankerhold , sebastin bergeret , hermann grabert , john martinis and vitaly shumeiko . | we review our experiments on the electronic transport properties of atomic contacts between metallic electrodes , in particular superconducting ones . despite ignorance of the exact atomic configuration ,
these ultimate quantum point contacts can be manipulated and well characterized in - situ .
they allow performing fundamental tests of the scattering theory of quantum transport .
in particular , we discuss the case of the josephson effect .
atomic contacts , superconductivity , quantum transport 74.45.+c , 74.50.+r , 73.23.-b |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.